Novel strategies for the safe chemical degradation of organic peroxide explosives: A mechanistic investigation

Mark Stephen Bali Captain, Australian Army

A thesis in fulfilment of the requirements for the degree of

Doctor of Philosophy

School of Physical, Environmental and Mathematical Sciences

The University College, UNSW Canberra

November 2014

i “The nation that will insist on drawing a broad line of demarcation between the fighting man and the thinking man is liable to find its fighting done by fools and its thinking done by cowards.”

Sir William Francis Butler

ii iii Table of Contents

CHAPTER 1. Introduction ...... 1 1.1 Terrorism and explosives ...... 1

1.2 Organic Peroxides ...... 3

1.2.1 Organic peroxides as explosives ...... 4

1.2.2 Organic peroxide incidents ...... 8

1.2.3 Unique Hazards ...... 9

1.2.4 Neutralising peroxide explosives ...... 12

1.3 The chemistry of TATP – A literature review...... 12

1.3.1 Acidic Degradation ...... 13

1.3.2 Thermolysis ...... 18

1.3.3 Metals and metal ions ...... 20

1.3.4 Other reactions ...... 22

1.3.5 Theoretical possiblities ...... 24

1.4 Other peroxide chemistry of relevance...... 24

1.4.1 Baeyer-Villiger reactions ...... 24

1.4.2 Redox and electrochemistry of cyclic peroxides ...... 27

1.4.3 Lewis acid reactions with organic peroxides...... 27

1.5 Related peroxides ...... 28

1.6 Conclusion ...... 32

1.7 This Study ...... 35

1.8 References ...... 36

CHAPTER 2. Tripentanone triperoxide as a model for TATP degradation studies 41 2.1 Introduction ...... 41

2.2 Research Goals ...... 42

iv 2.3 Experimental ...... 43

2.3.1 Instruments ...... 43

2.3.2 Sensitiveness testing ...... 45

2.3.3 Synthesis ...... 45

2.3.4 Acid degradation studies ...... 47

2.3.5 Thermal degradation studies ...... 47

2.3.6 Explosive Detection Dog field trial ...... 47

2.4 Results ...... 49

2.4.1 Pentanone peroxides and their characterisation ...... 49

2.4.2 Relative Safety ...... 54

2.4.3 Acidic Degradation ...... 56

2.4.4 Thermal Degradation ...... 60

2.4.5 Scent characteristics ...... 61

2.5 Conclusion ...... 65

2.6 Acknowledgements ...... 66

2.7 References ...... 66

CHAPTER 3. Redox and electrochemistry of cyclic peroxides ...... 69 3.1 Introduction ...... 69

3.2 Research Goals ...... 70

3.3 Experimental ...... 70

3.3.1 Materials ...... 70

3.3.2 Instruments ...... 71

3.3.3 Redox degradation studies ...... 72

3.3.4 Electrochemistry ...... 72

3.4 Results and Discussion ...... 73

3.4.1 Redox reactions ...... 73

3.4.2 Electrochemistry ...... 76

3.5 Conclusion ...... 79

v 3.6 References ...... 81

CHAPTER 4. The reaction of metal-based Lewis acids with TPTP ...... 82 4.1 Introduction ...... 82

4.2 Research Goals ...... 84

4.3 Experimental ...... 84

4.3.1 Materials ...... 84

4.3.2 Instruments ...... 85

4.3.3 Methods ...... 86

4.3.3.1 Preparation and determination of TiCl4 stock solution ...... 86

4.3.3.2 TPTP-TiCl4 degradation experiments ...... 86

4.3.3.3 Preparation and determination of TiCl4L2 stock solutions ...... 90

4.3.3.4 Degradation of TPTP by TiCl4L2 ...... 90

4.3.3.5 Degradation of TPTP by SbCl3 ...... 91

4.4 Results and discussion ...... 92

4.4.1 Titanium tetrachloride ...... 92

4.4.1.1 The reaction of TPTP with a large excess of TiCl4 ...... 92

4.4.1.2 1:0.5 reaction of TPTP with TiCl4 ...... 93

4.4.1.3 Characterisation of solid product from Ti-limited reactions ...... 100

4.4.1.4 IR Studies ...... 103

4.4.1.5 The effect of reagent ratio ...... 108

4.4.1.6 The effect of ligands ...... 114

4.4.1.7 Diffusion effects ...... 119

4.4.1.8 Inter- vs. Intra-molecular rearrangement ...... 120

4.4.2 Proposed Mechanism ...... 121

4.4.3 Antimony trichloride ...... 127

4.5 Conclusion ...... 130

4.6 Acknowledgements ...... 131

4.7 References ...... 131

vi CHAPTER 5. Generalisation of the TiCl4 reaction with other organic peroxides134 5.1 Introduction ...... 134

5.2 Research Goals ...... 136

5.3 Experimental ...... 137

5.3.1 Materials ...... 137

5.3.2 Instruments ...... 137

5.3.3 Methods ...... 138

5.3.3.1 Preparation and determination of TiCl4 stock solution ...... 138

5.3.3.2 TiCl4 degradation experiments ...... 138

5.4 Results and discussion ...... 139

5.4.1 The reaction of acetone peroxides with TiCl4 ...... 140

5.4.1.1 The reaction of TATP with 3-12 equivalents of TiCl4 ...... 141

5.4.1.2 The reaction of TATP with 0.5-2 equivalents of TiCl4 by FTIR ...... 144

1 5.4.1.3 The reaction of TATP with 1-3 equivalents of TiCl4 by H NMR ...... 146

1 5.4.1.4 The reaction of DADP with 1-2 equivalents of TiCl4 by H NMR...... 154

5.4.1.5 The formation of chloromethane and acetate ...... 156

5.4.1.6 Acetone peroxide overview ...... 160

5.4.2 The reaction of MEKP with TiCl4 ...... 163

5.4.3 The reaction of HMTD with TiCl4 ...... 169

5.5 Conclusion ...... 170

5.6 References ...... 172

CHAPTER 6. A catalytic degradation system ...... 174 6.1 Introduction ...... 174

6.1.1 An acid-mediated catalytic system...... 174

6.1.2 Baeyer-Villiger degradation of peroxides...... 175

6.1.3 Catalysis of the Baeyer Villiger reaction...... 177

6.2 Research Goals ...... 179

6.3 Experimental ...... 179

vii 6.3.1 Materials ...... 180

6.3.2 Instruments ...... 180

6.3.3 Methods ...... 180

6.3.3.1 Synthesis of Pt2Cl2(bis(diphenylphosphino)propane) ...... 181

6.3.3.2 Synthesis of [Pt2(bis(diphenylphosphino)propane)2(μ-OH)2.(BF4)2] (Pt-DPPP) 181

6.3.3.3 Synthesis of Sn(II)-montmorillonite complex (Sn-MMT) ...... 182

6.3.3.4 Catalytic oxidation of cyclohexanone by HOOH ...... 182

6.3.3.5 Catalytic oxidation of cyclohexanone by DPPDHP ...... 183

6.3.3.6 NMR studies of TPTP acidic degradation in the presence of Pt-DPPP ... 183

6.4 Results and discussion ...... 184

6.4.1 Catalytic oxidation with DPPDHP ...... 184

6.4.2 Catalytic oxidation via in situ TPTP degradation...... 187

6.4.3 Alternate catalysts for BV oxidation...... 191

6.4.3.1 Modification of the Pt-DPPP system ...... 192

6.4.3.2 Acid- and water-resistant BV catalysts ...... 195

6.4.3.3 Alternative BV catalysts ...... 196

6.5 Conclusion ...... 198

6.6 References ...... 198

CHAPTER 7. Conclusion and Recommendations ...... 201 7.1 Key Findings ...... 201

7.1.1 Insights into acidic degradation ...... 201

7.1.2 Lewis acid-mediated rearrangement ...... 203

7.1.3 Other findings of note ...... 205

7.1.4 Summary ...... 206

7.2 Recommendations ...... 206

7.3 References ...... 207

viii List of Figures

Figure 1.1- Common peroxide molecular structures...... 5

Figure 1.2 - Crystalline organic peroxide...... 6

Figure 1.3 - Structures of acetone- and pentanone-based peroxides...... 30

Figure 2.1- - Pentanone and Acetone peroxides in this study...... 42

Figure 2.2 - Organic peroxides derived from acid-catalysed reaction of 3-pentanone with

hydrogen peroxide...... 49

1 Figure 2.3 - H NMR spectra of acetone and pentanone cyclic peroxides in CDCl3...... 50

1 Figure 2.4 - H NMR of DPPDHP in CDCl3...... 50

Figure 2.5 - T1 Relaxation times for TPTP and DPDP...... 52

Figure 2.6 – COSY spectrum of TPTP/DPDP Mixture...... 53

Figure 2.7 - The IR spectra of TPTP and DPDP...... 54

Figure 2.8 - Degradation of TPTP by HCl tracked by quantitative GCMS...... 57

Figure 2.9 - Scent detection of organic peroxides by Explosive Detection Dogs...... 62

Figure 3.1 - Reaction of Mo2O5(OH) with various peroxides...... 76

Figure 3.2 - Cyclic voltammograms of DADP, TPTP and TATP in MeCN...... 77

Figure 3.3 - Cyclic voltammagrams of tert-butyl hydroperoxide, TPTP and DPPDHP in MeCN.

...... 77

Figure 4.1 - Anhydrous NMR sample sealing...... 87

1 Figure 4.2 - The effect of temperature on the H NMR spectrum of 3P 1:3 TiCl4...... 95

Figure 4.3 - Rate of TPTP degradation by a half mole equivalent of TiCl4...... 96

1 Figure 4.4 - H NMR spectra of the reaction of TPTP with 0.5 equiv TiCl4 over time...... 97

1 Figure 4.5 - H NMR (1D & COSY) of 0.5:1 TiCl4-TPTP reaction chlorinated products...... 99

Figure 4.6 - FTIR of precipitate from 0.5:1 TiCl4-TPTP reaction...... 101

Figure 4.7 - FTIR of TiCl4:L complexes and free L...... 105

ix Figure 4.8 - FTIR of 1:0.5/1:1/1:2 TPTP-TiCl4 reactions...... 106

Figure 4.9 - Reaction of TPTP and TiCl4 and varying concentrations and ratios...... 109

Figure 4.10 – Time required for 98% degradation of TPTP at different TiCl4 ratios...... 110

Figure 4.11 - Degradation extrapolated to t=0 at various TiCl4 ratios...... 111

Figure 4.12 - Comparison between 1:0.5 and 2:1 TPTP-TiCl4 reactions...... 112

Figure 4.13 - Reaction of TPTP with TiCl4L2 vs. TiCl4...... 115

Figure 4.14 - 2C3P production plotted against TPTP degradation...... 118

Figure 4.15- The progression of 1:1 reaction of SbCl3 and TPTP at 50°C...... 128

Figure 4.16 - Reaction of SbCl3 and TPTP monitored by direct GCMS sampling...... 129

Figure 5.1 - Structures of organic peroxides studies in Chapter 5...... 136

Figure 5.2 - 1H NMR of TATP and DADP...... 141

1 Figure 5.3 – H NMR spectra of the reaction of 1:6 TATP and TiCl4...... 142

Figure 5.4 - Ratios of products to ester in TATP/TPTP reaction with increasing TiCl4...... 142

Figure 5.5 - FTIR spectra of TPTP and TATP precipitates after reaction with TiCl4...... 145

Figure 5.6 - Reaction of 1:1 and 1:2 TATP to TiCl4 over time...... 148

Figure 5.7 - Change in relative amounts of key analytes over time for 1:1 and 1:2 TATP/TiCl4

reaction...... 149

Figure 5.8 - Comparison of TATP/TiCl4 1:1 reaction with predicted 1H NMR spectrum for

chloro-peroxy species...... 153

1 Figure 5.9 - H NMR spectra of the reaction of 1:1 and 1:2 DADP with TiCl4...... 154

Figure 5.10 - Relative abundance of key analytes in the 1:1 and 1:2 reaction of TATP with

TiCl4...... 155

1 Figure 5.11 - H NMR of possible acetato signal over time in the 1:1 reaction of TiCl4 and

TATP...... 158

Figure 5.12 - Schematic of the localised generation of CM in the 1:1 reaction of TATP and

TiCl4 prepared by the free-seal-thaw method...... 162

Figure 5.13 - Chromatogram and mass spectrum of the MEKP mixture used in this study. .... 164

x Figure 5.14 - 1H NMR of MEKP mixture used in this study...... 165

1 Figure 5.15 - H NMR of the reaction of MEKP with TiCl4 at 1:1-3 ratios...... 166

Figure 6.1 - 1H NMR of Pt-DPPP...... 184

Figure 6.2 - Oxidation of CyHexO to CLO by HOOH catalysed by Pt-DPPP...... 185

Figure 6.3 – Catalysed and uncatalysed oxidation of CyHexO by HOOH and DPPDHP

monitored by GCMS...... 186

Figure 6.4 - Reaction of TPTP and MSA observed by 1H NMR...... 188

Figure 6.5 - Reaction of MSA, TPTP and Pt-DPPP observed by 1H NMR...... 189

xi List of Schemes

Scheme 1.A - Ionic mechanism for the stepwise degradation of TATP in acidic conditions. .... 14

Scheme 1.B - General scheme for thermal degradation of TATP...... 19

Scheme 1.C - The general scheme of the Baeyer-Villiger oxidation...... 25

Scheme 1.D - Elucidation of the Criegee Intermediate...... 26

Scheme 1.E – The formation of cyclic ketone-peroxide dimers...... 31

Scheme 2.A - Baeyer-Villiger reactions schemes...... 58

Scheme 4.A - Proposed mechanism for R1TPTP and R2TPTP...... 123

Scheme 4.B - Proposed mechanism R3P - the α-chlorination of 3P by TiCl3(OCl)...... 125

Scheme 5.A - Proposed mechanism for degradation of TATP by TiCl4 and TiCl4Lx...... 152

Scheme 5.B- Proposed mechanism for the oxidation of acetone by TiCl3(OCl) forming acetato

ligands...... 156

Scheme 5.C - Proposed mechanism for the oxidation of acetone by TiCl3(OCl) forming acetato

ligands...... 157

Scheme 5.D - Possible resonance mechanisms of the proposed Ti-acetato moiety providing

shielding effects to methyl protons...... 159

Scheme 5.E - Overall scheme of the 1:1 reaction of TATP and TiCl4...... 161

Scheme 6.A - The application of BV catalysis for safe reduction of hydroperoxide

intermediates...... 176

Scheme 6.B - General mechanisms of Baeyer-Villiger catalysis...... 177

Scheme 6.C - Proposed mechanism for the reaction of MSA nd Pt-DPPP...... 190

Scheme 6.D - The role of water in the acidic degradation of cyclic peroxides...... 192

xii List of Tables

Table 1.1 - Comparisons of explosive classifications...... 3

Table 2.1 - Sensitiveness Data for TPTP and TATP...... 55

Table 2.2 - Ester formation in TPTP acidic degradation in the presence of substrate ketones. .. 59

Table 2.3 - Thermolysis products of TPTP and TATP...... 60

Table 3.1 - Redox reaction results...... 74

Table 4.1 - 1H NMR chemical shifts of EP and 3P at given Ti ratio...... 94

Table 4.2 - Organic products from 0.5:1 TiCl4-TPTP reaction (9 days)...... 100

Table 4.3 - Elemental analysis of precipitate from 0.5:1 TiCl4-TPTP reaction...... 102

Table 4.4 - IR data for reaction participants and mixtures...... 104

xiii List of Abbreviations

1D 1 Dimensional 2D 2 Dimensional 2C3P 2-chloro-3-pentanone 3P 3-pentanone AcMe acetone AcO- acetate ANFO Ammonium Nitrate Fuel Oil AO Active Oxygen BAM Bundesanstalt für Materialprüfung BP Boiling Point BV Baeyer-Villiger CA chloroacetone CM chloromethane CLO ε-captrolactone COSY Correlation Spectroscopy CyHexO cyclohexanone DADP diacetone diperoxide DCE dichloroethane DCM dichloromethane DMDO dimethyl dioxirane DPDP dipentanone diperoxide DPPDHP dipentanoneperoxy dihydroperoxide DPPP (diphenylphosphino)propane DSTO Defence Science and Technology Organisation EA ethyl acetate EDD Explosive Detection Dog EDTA ethylenediaminetetraacetic acid EOD Explosive Ordnance Disposal EP ethyl propanoate ESD Electrostatic Discharge FTIR Fourier Transform Infra Red GCMS Gas Chromatography - Mass Spectroscopy HMB Hexamethylbenzene HMTD hexamethylene triperoxide diamine HOOH hydrogen peroxide HPOM Hydrogen Peroxide – Organic Matter IED Improvised Explosive Device IMS Ion Mobility Spectroscopy xiv MeCN Acetonitrile MeOAc Methyl acetate MEKP methyl ethyl ketone peroxide MeOAc methyl acetate MMT montmorillonite-K10 MP melting point MSA methanesulfonic acid MTO methylrheniumtrioxide MW molecular weight NIST National Institute for Standards and Technology NMR Nuclear Magnetic Resonance OPE Organic Peroxide Explosives PETN pentaerythritol tetranitrate PhAc acetophenone

Pt-DPPP [Pt2(bis(diphenylphosphino)propane)2(μ-OH)2.(BF4)2] Research Department Explosive, cyclonite, 1,3,5-trinitroperhydro-1,3,5- RDX triazine ROOH organic hydroperoxide RT room temperature TATP triacetone triperoxide THF tetrahydrofuran TMS tetramethylsilane TNT 2,4,6-trinitrotoluene TPTP tripentanone triperoxide UNSW University of New South Wales WA Western Australia WWI World War I

xv List of Publications and Presentations

The work in this thesis has been presented or published in part in the following papers/presentations:

Bali, M. S., Armitt, D., Wallace, L., & Day, A. I. (2011). Rapid chemical neutralisation of TATP. Paper presented at the PARARI: The Australian Explosive Ordnance Symposium, Brisbane.

Bali, M. S., Armitt, D., Wallace, L., & Day, A. I. (2013). Towards the Catalytic Degradation of Organic Peroxide Explosives. Poster presented at the International Symposium for the Analysis and Detection of Explosives (ISADE), The Hague.

Bali, M. S., Armitt, D., Wallace, L., & Day, A. I. (2013). Facile rearrangement of cyclic peroxides – an approach to safe degradation. Poster presented at the RACI NSW Branch Organic One-Day Symposium, Canberra.

Bali, M. S., Armitt, D., Wallace, L., & Day, A. I. (2013). Towards the Catalytic Degradation of Organic Peroxide Explosives. Poster and oral presentation at the RACI Research and Development Topics Conference, Canberra.

Bali, M. S., Armitt, D., Wallace, L., & Day, A. I. (2014). Cyclic pentanone peroxide: sensitiveness and suitability as a model for TATP degradation studies. Journal of Forensic Sciences.

xvi Acknowledgements

A page is insufficient to thank all of those who have assisted in the production of this thesis; so I will use two. My first thanks must go to my supervisors Anthony, Lynne and Dave. Each of you has left a mark on this thesis, and on my development as a chemist. Lynne and Anthony, your open and frank approach as supervisors has been priceless. I have not taken lightly your willingness to tackle this project, which rests outside your usual research interests, and the hours you have invested in working through unfamiliar problems with me. I also want to thank you for your unwavering support on the many occasions where others doubted my ability to prevail over personal challenges or work requirements. Not once did I feel that either of you were unsure of my ability to complete this thesis, which helped make the come back after every adversity that little bit easier. Also, a big thank you to the team at PEMS: Barry, Julie, Deb, Di, Annabelle, Nadia and Tessa, the workshop lads, Colin, fellow postrgrads… the list goes on! To Dave, I owe you my sincere thanks for your part in setting up this project in the first place. I would never have had a chance to do a PhD without your willingness to trust in my ability to tackle the peroxide neutralisation problem way back in 2008. The project you suggested and your strong recommendation built the case for the scholarship that started this journey. Thank you for giving me the flexibility to explore, ample advice, a few hard questions at just the right moments, many hours of editing, and more than a few hours of preparing samples. Thanks also the rest of the DSTO team – Phil Davies, Craig Wall, Mark Smith, Mark Fitzgerald, and many others – for help in synthesis, specialist sensitiveness testing and advice. I also owe my thanks to a number of military colleagues who continue to champion the cause for the place of academic learning within the Australian Defence Force. AIRMSHL Angus Houston (Rtd), LTCOL Dave Evans, BRIG Wayne Budd, BRIG Phil Winter (Rtd), COL Brennan, LTCOL Daunt , LTCOL Davidson, MAJ Fish, Dr Albert Pallazzo, Mr Dale Cooper, to name a few. I hope that this body of work in some ways justifies your support. I also hope that news of my research raises interest in unlocking the potential of the Australian Defence Force Academy to be an institution that capitalises on the synergy of academic research and military service.

xvii To my dear friends and office mates - Mick, Lubna and Haiqiang – I’ll always remember these past few years for many good laughs, far too many coffees, and for your support when things were not going well. I plan to celebrate with each of you when you make it through your projects, and have one more completely unnecessary coffee to celebrate. Jase, Matt, Christie, Deb L., Michelle, – Your support through these years for Deb and I has been an absolute blessing. I wouldn’t have got here without you all. To those who gave up valuable hours to edit and polish this document, my thanks knows no bounds! To my family – dearest Barbara and Dad, and all my Wearmouth family - I hope you can forgive my distance and distraction because of the time spent on this work. I will do my best to make it up by being the best son, brother, uncle, son-in-law and brother-in-law that I can possible be. I hope I was still able to be there when you needed me most. Most importantly, to my beloved wife Deborah: This is our thesis. Not just because your edits are on every page, but also because your blessing was on every minute I spent in the lab, at a conference, or in the office late at night. During most of the time this thesis was prepared we were chasing at least two burning ambitions. Your support and belief has helped me achieve one: my lifelong ambition of becoming a scientist. I look forward to conquering the second with you, knowing that we will achieve the same success.

xviii Preface

It is perhaps not normal for a dissertation to describe the author’s background; it is also, however, not altogether normal for a chemist researching a military problem to also be a military officer. As the author’s background provides significant relevance to the discussion of field-application, this short biography is provided for context.

Captain Mark Bali is an officer of the Royal Australian Engineer Corps in the Australian Army. He graduated from the Australian Defence Force Academy with a Bachelor of Science with Honours Class One in Chemistry in 2005. Since his graduation from the Royal Military College, Duntroon, and the School of Military Engineering, Captain Bali has actively sought to apply his technical chemistry knowledge through his career as a combat engineer. He has served the majority of his career in specialist units working with Explosives Disposal, High Risk Search, and Chemical Biological Radiological and Nuclear Defence (CBRND). Captain Bali has also deployed as a Troop Commander to Afghanistan, where he was largely responsible for the counter-IED and forensic evidence management activities of his task force. This operational experience combined with his academic background led to the award of the Chief of Defence Force Fellowship in 2010, which sponsored one year of research to find a method to rapidly and safely chemically neutralise organic peroxide explosives (OPEs). Captain Bali continued this research as a PhD program in parallel with his career whilst working part time as a capability development officer in the Australian Defence Force’s Counter IED Task Force. The outcomes of this research are presented in this thesis.

xix CHAPTER 1. Introduction

1.1 Terrorism and explosives

The gradual proliferation of explosives knowledge, largely via the internet, has provided terrorists with an unprecedented ability to commit large scale acts of terrorism.

In response, significant effort has been invested in research aimed at detecting explosives, preventing their illicit manufacture, and finding safer ways to dispose of them. As this thesis has a strong focus on working toward a field-relevant outcome, the broader context of explosives neutralisation will be set by providing a general introduction to explosives and the way in which they are utilised by terrorists.

An explosive is a semi-stable chemical or mix of chemicals that is able to sustain a rapid chemical reaction without the participation of external reactants such as oxygen.

Such rapid reactions typically result in a liquid or solid being converted into large amounts of gas and heat in a very short space of time, resulting in a blast wave as the evolved gas rushes out from the explosion in order to equilibrate its pressure with the surrounding atmosphere. It is this blast wave that causes mass damage to nearby objects through thermal burns, shock force and kinetic transfer.

To enable a meaningful discussion of explosives, it is essential to understand the major classes and their relative differences. Compounds such as flammable liquids and those found in fireworks are known as incendiaries. Incendiaries are not explosives as their rate of burn is not fast enough to generate a shock wave under normal conditions.

Between incendiaries and high explosives are the ‘tertiary explosives’. Ammonium nitrate, a common fertiliser, is the best known of this class of materials and has been

1 used extensively in Afghanistan by insurgents. Whilst quite stable in normal circumstances, the addition of fuel such as diesel to make Ammonium Nitrate-Fuel Oil

(ANFO) results in a mixture capable of generating a blast wave. These compositions still require a ‘boost’ from a charge of more sensitive and higher powered explosives in order to deliver a reliable detonation, and are generally not used in isolation from other explosives.

The next major class is the secondary explosives, including cyclonite (RDX) and trinitrotoluene (TNT). These explosives can propagate an explosive shock wave with only a small initiating charge or a large detonator. Secondary explosives are relatively insensitive to heat, kinetic shock, electrostatic discharge and friction making them safe enough in routine handling, moulding and transport. Some secondary explosives, such as pentaerythritol tetranitrate (PETN) are slightly more sensitive and have a very high detonation velocity. These explosives are often used as ‘booster charges’ to provide the necessary explosive force to initiate detonation in tertiary explosives. Their stable characteristic means that on their own, secondary explosives are not able to be reliably detonated without an external explosive force.

For reliable initiation, a highly sensitive explosive is needed to translate an external force (electrical, chemical, thermal or mechanical) into a shockwave able to trigger the detonation of the secondary explosive which generally constitutes the bulk material. This class of materials, known as primary explosives, have a detonation wave analogous to secondary explosives but are much more sensitive to friction and heat than secondary explosives. This characteristic makes them essential to the function of a detonator, a device that is used to trigger the initiation of an bomb. A commonly used example is lead azide in detonators, for which the relative sensitiveness data is provided in Table 1.1. Other examples include the organic peroxides hexamethylene triperoxide

2 diamine (HMTD) and triacetone triperoxide (TATP), which are discussed in detail later.

Their sensitivity means that a flame, hot electrical filament or mechanical striker can provide enough energy to trigger an explosive chain reaction. These primary explosives are critical resources for the manufacture of Improvised Explosive Devices (IED), or home-made bombs. Primary explosives can be the determining factor between a pop with a small fire, and a devastating blast.

Table 1.1 - Comparisons of explosive classifications. Maximum Friction Volume of Classification detonation sensitivityb gas per kg velocitya (N downward explosivec (m.s-1) force on test (L) device) ammonium nitrate / fuel oil <36001 >3532 9702 Tertiary (ANFO) trinitrotoluene (TNT) 69002 >3532 7302 Secondary pentaerythritol tetranitrate 84002 602 8232 Secondary (PETN) (Booster) lead azide 4500-53002 0.1-12 3082 Primary hexamethylene triperoxide 45002 <0.12 10002 Primary diamine (HMTD) triacetone triperoxide ~53003 <0.12 Not reliably Primary (TATP) reported This table highlights the differences in characteristics of each of the key types of explosive (references indicated for values). aMaximum detonation velocity measures are an indicator of the impact of the shockwave. Pressed samples with higher density deliver a higher velocity. bFriction sensitivity indicates the pressure exerted on two sliding surfaces between which the tested explosive will detonate upon movement. cThe volume of gas defines the mechanical work capacity of the explosive.

1.2 Organic Peroxides

The term peroxide refers to a molecule within which two oxygen atoms are bonded together by a single bond. The simplest form of peroxide is hydrogen peroxide

(Figure 1.1(a)), which is commonly used in dilute solution as a bleaching agent and disinfectant. Organic peroxides build upon the basic O-O structure but also include carbon.4 An example of an organic peroxide in common industrial use is di-tert-butyl peroxide (Figure 1.1(b)), which is a catalyst in the production of certain plastics.5 This compound is usually sold and transported in solution rather than its pure form to reduce

3 its explosive character. The peroxide O-O single bond is particularly weak, a characteristic which leads to the unstable nature of peroxides. For instance, dimethyl peroxide has a bond dissociation energy of only 146 kJ/mol compared to the 377 kJ/mol required to break the C-C bond of ethane.6

1.2.1 Organic peroxides as explosives

Whilst peroxides are inherently unstable, not all organic peroxides can be used as explosives. Some organic peroxides are regarded as explosives in their own right, and are able to sustain a powerful detonation without added fuel. At this point, an important distinction needs to be made between organic peroxides and explosive mixtures containing hydrogen peroxide. Hydrogen Peroxide – Organic Matter (HPOM) mixes are more akin to ANFO. In a HPOM, the oxidiser is hydrogen peroxide, and the fuel can be almost any finely ground organic material, although performance will vary based on the fuel used. A HPOM is not an organic peroxide as there are no covalent bonds between the peroxide unit and organic moieties. HPOMs are significantly less friction sensitive than organic peroxide explosives (OPEs), and usually less powerful. The most commonly encountered OPEs include TATP, HMTD and methyl ethyl ketone peroxide

(MEKP) (Figure 1c, 1e and 1f respectively). The title MEKP in fact refers to a family of cyclic and open-chain organic peroxide oligomers formed in the reaction of methyl ethyl ketone and hydrogen peroxide.7 The monomer and open-chain dimer dihydroperoxides are commonly used in industry as polymerisation catalysts.8

4 a b c d e f

Figure 1.1- Common peroxide molecular structures. A selection of peroxides, from the commercially used hydrogen peroxide (a) and tertiary butyl peroxide (b) to organic peroxide explosives triacetone triperoxide (TATP) (c) and the closely related diacetone diperoxide (DADP) (d), methyl ethyl ketone peroxide (open chain dimer) (e) and HMTD (f) are all considered primary explosives. Structures produced using HYPERCHEM software package.

To understand the power of these explosives, it is valuable to compare them to a standard military grade explosive such as TNT. A simple test of explosive power is the

‘lead block test’ or Trauzl test. In this protocol, a standard 10 g charge is inserted into a hole in a block of lead, detonated, then the volume of the resulting cavity measured. The cavity size is dependent on the amount of gas liberated by the explosive, the heat of explosion and the velocity of detonation. In the lead block test, TNT provides a value of

300 cm3. TATP provides a volume of 250 cm3, or 83% of TNT.2 HMTD proves to be even more powerful that TNT, at 330 cm3, or 110% of TNT.2 It is also important to note that whilst TNT has a higher velocity of detonation, HMTD evolves up to 1000 L of gas for each kg of explosive, significantly higher than TNT with 730 L per kg.2 The amount of gas is particularly important when the mechanical work output of the explosive is of concern, such as destruction in a confined space. The destructive power of an explosive is indicated by a combination of its maximum detonation velocity and the volume of gas generated.

A notable characteristic of OPEs, and one of the reasons they see no use in commercial or military spheres, is their very high sensitivity to mechanical impact, friction, heat and electrostatic discharge. TATP, HMTD and MEKP are all extremely sensitive to such initiation stimuli. The data presented previously in Table 1.1 illustrates 5 how OPEs compare against other common explosives in sensitivity. TATP and the related diacetone diperoxide (DADP) have a further hazardous characteristic of subliming at room temperature and re-depositing as crystals on nearby surfaces.3 This makes the storage of OPEs in containers very dangerous, further increasing the practical sensitiveness of the explosive. For example, the action of grinding a crystal trapped in a screw-top thread can trigger a detonation. Electrostatic discharge (ESD) is also a major hazard. The smallest of static shocks, even less than 0.0056 J, can cause TATP to detonate9 (the human body can accumulate up to 0.02 J in the right conditions).10-11

These characteristics make OPEs extremely dangerous to handle in any significant quantity. Interestingly, these same characteristics are also very attractive features for a potential terrorist seeking an easily accessible primary explosive.

Figure 1.2 - Crystalline organic peroxide. TATP and HMTD both form white crystalline solids in their raw form, very similar to this peroxide used as an analogue in our laboratory work (discussed later). Crystal size can range from large single crystals to fine powders depending on synthetic procedure and age.

Most OPEs are synthesised by very straightforward procedures. The synthesis of

TATP was first reported in 1895, when acetone was reacted with a cold solution of hydrogen peroxide and sulfuric acid.12 Whilst the reaction conditions have been improved since this first study, the basic process of acid-catalysed synthesis at low

6 temperatures remains unchanged. No technical equipment is required, and the raw materials of TATP are acetone (a common industrial solvent), acid (hydrochloric or sulfuric being most common) and hydrogen peroxide of concentration >3% (w/w).

These items are all available at hardware stores or supermarkets, and a well-organised, motivated group could acquire considerable quantities without raising alarm.

The synthesis of TATP is straightforward as long as some basic precautions are taken with temperature and rate of addition of reagents.13-15 Purification is simple as the raw ingredients and most impurities are highly volatile or can be washed off with water, leaving pure TATP. The major impurity, DADP, is generally not removed as it is also a primary explosive, however if necessary it can be easily removed by recrystallisation.

Syntheses of HMTD and MEKP are similarly straightforward, and the raw ingredients for these OPEs are almost as readily available (hexamine hiking stove tablets for

HMTD, and the industrial solvent methyl ethyl ketone for MEKP).

As noted earlier, primary explosives are crucial to the construction of an

Improvised Explosive Device (IED or home-made bomb) that can be relied upon to detonate properly. A lack of primary explosives (together with a generally poor understanding of explosives) was thought to be part of the reason the Times Square

Bomber of 2009 failed to initiate his car-load of gas cylinders and fertiliser.16 The initiation method he used resulted in a fire rather than an explosion as there was no material sensitive enough to detonate fully.

OPEs, whilst extremely dangerous to handle, occupy a unique niche in the explosive world by being both accessible and sensitive. TATP’s sensitiveness makes it a prime candidate for converting a source of heat or spark (i.e. electrical current) into a detonation wave capable of initiating a larger bulk explosive such as ANFO.

Furthermore, the sheer volumes of TATP’s raw ingredients used in industry make

7 tracking the small quantities required for TATP synthesis virtually impossible. This unfortunate intersection of explosive properties and ready access make organic peroxides a key threat to national security, underscoring the need for a thorough understanding of their chemical properties.

1.2.2 Organic peroxide incidents

It is not uncommon in modern Western democracies, such as Australia, to discover

OPEs in bombing incidents. The most recent example is the discovery of three kilograms of TATP in the Western Australian (WA) suburbs of Australind and Bunbury in 2013.17 This incident highlights the ability of individuals to synthesise large quantities of OPE despite the risks of accidental detonation. The packages of TATP discovered in Australind were capable of significant damage, yet were only brought to police attention when a fisherman accidentally found an abandoned package of TATP.

Unfortunately, the use of TATP is not limited to home experimenters. Aside from consistent use in the Israel-Palestinian conflict, organic peroxides have seen increased use in major international terrorism cases. One of the most prominent cases was the

‘shoe bomber’ Richard Reid, a self-admitted member of Al Qaeda. Reid attempted to detonate a bomb containing the high explosive PETN with a TATP priming charge whilst on Flight 63 from Paris to Miami in December 2001. The attempt failed due to a damp fuse, but had the potential to bring down the passenger airliner.18 It was later discovered that the shoe bomb was not the only one, with a second, identical bomb found in the possession of Gloucester resident, Saajid Badat.19

The shoe-bomb case highlights the importance to terrorists of TATP as an accessible primary explosive. Although PETN is a relatively sensitive secondary explosive, it is not sensitive enough to be reliably detonated by a simple mechanism such as a firing pin or flame. In this instance, TATP was to be the trigger to initiate the 8 powerful PETN main charge. The device probably only failed due to Reid being delayed for a day by suspicious airport security. During this time the fuse absorbed moisture making it difficult to ignite, giving passengers on the plane sufficient time to restrain Reid before he could ignite the bomb.

1.2.3 Unique Hazards

First responders faced with organic peroxides must also confront a unique set of hazards in any device or container that includes organic peroxides. The first challenge is to identify that organic peroxides are actually contained in an IED. It is unlikely that

Explosive Ordnance Disposal (EOD) operators could identify if an IED contains a

TATP detonator as the detonator itself is rarely accessible, but this is not generally of immediate concern. Of greater concern is the use of TATP as a main charge. Whilst such a design may be dismissed as too dangerous, the Morrocan attacks of 2003 provide at least one historical example which shows that personal risk is often not a major concern to terrorists. This incident, which claimed 45 lives including the bombers’, was reportedly conducted with TATP as the bulk charge.9 The recent large amounts of

TATP found in WA also indicate the willingness of bomb-makers to accept the risks of handling large amounts of organic peroxides in their pursuit of building a powerful device.

In order to deal with a main charge consisting of OPEs, an EOD technician would first need to detect and identify the explosive. As their chemistry is so different to conventional nitrogen based explosives, conventional chemical kits (the use of chemicals that are added to a test sample) will not identify organic peroxides. Some newer kits (such as the DropEx Plus kit) have additional steps that will detect peroxides, however they are not able to distinguish between OPEs, HPOMs or innocuous hydrogen peroxide solutions. Advanced explosive vapour detectors based on ion 9 mobility spectroscopy (IMS) are able to distinguish between some OPEs. Explosive

Detection Dogs (EDDs) have proven to be very receptive to the scent of TATP and can quickly be trained to indicate its presence.20 However the inherent danger of storing and handling organic peroxides has meant most EDDs are not being trained on organic peroxide scents. Furthermore EDDs cannot indicate which explosive they have located, only that they have found one of the scents they have been trained on. A complete discussion on the detection of OPEs is beyond the scope of this work.

Leaving aside the difficulties in determining if an IED contains an explosive fill of organic peroxides, dealing with the sensitivity of such an IED can test conventional

EOD methods and equipment. The safest method of dealing with any IED, regardless of explosive fill, is to blow it up in place using a counter charge or by triggering the IED remotely. This removes the explosive hazard by triggering it at a time when the area is cleared of people who may be injured by the blast. Whilst this is the preferred method of disposing of IEDs, it is not always appropriate due to the positioning of the device next to sensitive installations (flammable stores, high value infrastructure, etc). The use of blow-in-place procedures also destroys much of the forensic evidence on the IED which may be crucial in identifying the bomber and the source of the explosives. In such situations, disablement or ‘rendering safe’ (colloquially ‘defusing’) is the preferred option.

There are a number of ways to render safe, however the key principle is to separate the components of the IED in a way that prevents the device from detonating as designed. One common method is the use of kinetic disruptors in which very high speed jets of water are shot at the IED to separate strategic compents; e.g. isolating the power source from the initiator. The aim is to deactivate the arming circuit in a manner that does not permit the firing circuit to function. Whilst this is a very effective method

10 under ordinary conditions, it carries an unreasonable risk of triggering an organic peroxide main charge as the high kinetic energy transfer is liable to directly detonate an impact sensitive explosive fill. For this reason, kinetic disruption is not a recommended method of rendering safe organic peroxide IEDs.

Dissolving the explosive fill is one method that is utilised by EOD teams in dealing with larger quantities of organic peroxides. Diesel fuel is one solvent which is often recommended in standard operating procedures, as it has a higher ignition temperature than normal fuel and is generally readily available. Whilst feasible, this method has not undergone rigorous testing. Aside from practical concerns over how to contain, store and move the litres of explosive contaminated fuel, the dissolution process may be endothermic or exothermic.

There are concerns that organic peroxides may generate sufficient localised temperature differences to risk a detonation or fire. This is especially of concern as the most likely dissolution scenario would see the application of solvent onto solid explosive, which could result in high localised solution concentrations. The acid impurities likely to be present in homemade organic peroxide explosives may amplify this risk, as the solvation of acids is often an exothermic process. Despite these uncertainties, dissolution remains one of the few viable methods to deal with such a situation, and continues to be recommended in Standard Operating Procedures for bomb squads around the world.

Dissolving organic peroxides also leaves the issue of final disposal. Incineration is usually suggested, but there is no known study that can identify how concentrated a solution can be before a detonation might occur.

The lack of a robust, remote means of disablement coupled with a lack of a proven neutralisation method is a significant hazard for the safety of EOD personnel

11 when dealing with an OPE IED. It is also a risk to mission success where the aim is to remove the threat without the use of blow-in-place procedures.

1.2.4 Neutralising peroxide explosives

As discussed, existing render safe techniques such as disruption carry undue risk of detonation, and dissolution (despite extensive use) has a number of unquantified risks and is logistically unwieldy. A potential solution to this problem would be a means of rapid and safe chemical neutralisation. Practically, this would ideally take the form of a homogeneous liquid that could be safely sprayed directly onto solid or liquid peroxide explosive and lead to its gentle degradation to safe products within a few hours. Whist such a method would not be able to address all conceivable deployments of peroxide explosives, it would give first responders a great deal more flexibility in dealing with the threat than is currently available to them.

1.3 The chemistry of TATP – A literature review

It is counterintuitive that such an ‘unstable’ explosive compound would be difficult to degrade, but the few studies that have looked at neutralisation of TATP have found a surprising level of chemical stability. Typical methods for destruction of organic peroxides, such as reaction with acidified potassium iodide solution, produce little destruction.21 Most studies have concentrated on the only properly understood mechanisms, these being the action of strong acids22-24 and thermolysis.25-28 These and other methods will be discussed in turn to examine what is understood about the mechanisms, as well as cover some of the lesser described methods. The limitations of applying these methods in the field environment will also be addressed.

12 1.3.1 Acidic Degradation

Arguably the best understood method of neutralising cyclic organic peroxides is the use of acid-catalysed degradation. This reaction is essentially the reverse of the synthesis (Scheme 1.A). Accordingly, the key end products in the case of TATP are its raw materials - acetone and hydrogen peroxide.22

A challenge in elucidating the acidic mechanism is the huge variability in results when experiments are carried out under different conditions. In an early study, Furuya and Ogata studied the degradation of TATP in the presence of acetic acid and sulfuric acid.23 Under these conditions, complete degradation of TATP was found to yield 3 equivalents each of acetone and hydrogen peroxide. The hydrogen peroxide was detected as peroxyacetic acid under these conditions due to the action of hydrogen peroxide on the acetic acid. In a more recent experiment, Tsaplev used Fe(II) (the

Fenton reagent) to spectroscopically quantify the kinetics of TATP degradation by sulfuric and hydrochloric acids.29 In these conditions HOOH (or ROOH) oxidised the reagent to Fe(III). In both these experiments, the reagent used to test for active oxygen

(AO) (ie Fe(II) or acetic acid) was highly reactive, consuming the species as it was being formed. The analysis method affected the rate of reaction and drove the degradation of TATP to completion. Together, these two studies confirm that AO species (either intermediate hydroperoxides or hydrogen peroxide itself) and acetone are produced in the breakdown of TATP in acidic media.

Armitt et al. used headspace GC-MS analysis to measure the degradation products of solid samples of TATP which were exposed to HCl and H2SO4 vapours in sealed containers.22 The HCl sample generated significant quantities of acetone and chlorinated derivatives thereof. The latter were postulated to be due to HCl and AO species forming a hypochlorite species, which then chlorinates the ketones in the α-position. As was the

13 case in the Furuya and Tsaplev experiments,23 29 it seems that the consumption of hydroperoxides by chloride oxidation drove the HCl reaction to completion. In contrast, the H2SO4 sample showed much slower degradation, and large amounts of DADP were produced with acetone only a minor product. Drawing on past observations and their own experiments, Armitt et al. proposed that the reaction proceeded via a stepwise, ionic mechanism as described in Scheme 1.A.

Scheme 1.A - Ionic mechanism for the stepwise degradation of TATP in acidic conditions. This mechanism is reproduced from the work of Armitt et al.22

This mechanism is supported by the findings of other studies. The Fe(II) study of

Tsaplev described previously found that the rate of AO formation for TATP increased at higher concentrations of acid.29 Furthermore, this effect was more pronounced for sulfuric acid than for hydrochloric. Although not discussed in Tsaplev’s report, this difference could be attributed to chloride competing with Fe(II) to be oxidised by AO, thereby slowing the overall Fe(II) oxidation rate – an effect which would not occur with

H2SO4. This reaction of chloride with hydroperoxide intermediates can also explain

14 why Armitt et al. did not find significant quantities of DADP in the HCl degradation of

TATP. Chloride oxidation was suggested by Armitt et al. to explain the presence of chloroacetones, probably derived from the action of hypochlorite (known to form with

HCl and ROOH) on the acetone products.22 The chloride reaction may also occur with the open-chain alcohol-hydroperoxide dimer required to form DADP, thus limiting the formation of the cyclic dimer. This finding matches studies of the synthesis of TATP

(which relies on the same mechanism), which report that if HCl is used as the catalyst,

DADP is not observed.30 A different study did report the formation of DADP in the

HCl-catalysed synthesis of TATP, but in this case a much lower concentration of HCl was used,31 potentially reducing the impact of Cl-/ROOH side-reactions. Even in this last example, the relative abundance of DADP was much less for HCl catalysed samples than those using H2SO4. It seems the presence of the oxidisable chloride counter-ion has a significant effect on the overall acidic reactions of TATP.

The further degradation of DADP by acid has also been studied. Tsaplev found that DADP’s rate of degradation was not affected significantly by [H+] regardless of the acid’s counter-ion, but did depend on [Fe(II)].29 Tsaplev suggested that DADP’s protonation did not necessarily lead to ring opening, which can more clearly be interpreted as the equilibrium between DADP and its open chain hydroperoxide lying heavily toward the cyclic dimer. It was clear that Fe(II) was required under these conditions for the efficient ring-opening of DADP. This again supports Scheme 1.A, where under acid catalysis and in the absence of an AO scavenger (Fe(II), Cl-, etc), the formation of DADP is catalysed. As DADP is also a highly energetic cyclic peroxide and the aim of degradation processes are to remove this hazard, it can be concluded that acid catalysis without an AO scavenger does not provide an effective process.

15 A critical observation on acidic degradation of cyclic peroxides was made by

Oxley, who found that small (5 mg) samples of TATP (and also DADP and HMTD) could be destroyed by the direct action of concentrated acids within 15 mins.24 This was accompanied by a strong warning that scale-up of this reaction was dangerous as the reaction was believed to be exothermic, and would cause detonation of the sample.

Further investigations by Oxley’s group found that the equilibrium between the formation and degradation of TATP by acid catalysis was extremely sensitive to the relative concentrations of acid, acetone, peroxide and water.14 Whilst this work was largely aimed at formation of TATP, it provided further evidence of an ionic, stepwise mechanism with acid and water playing vital roles, consistent with Scheme 1.A.

As this thesis was being finalised, Oxley published their group’s latest experiments in applying acidic degradation to solid peroxides.32 The general protocol involved wetting TATP (or HMTD) with ~2 equivalents (by weight) of aqueous alcohol, then gradually adding ~1 equiv of acid, whilst keeping the solution temperature at or below 70°C. This reaction was successfully applied to a field-trial-sized sample of

460 g (using isopropanol and 36% HCl), The reaction was visible as vigorous self- agitation of the solution at the surface of the solid peroxide, and neutralisation of visible

TATP was complete within 30 mins. Analysis of smaller samples with different acids and solvents showed that the products depended on the nature of the acidic anion,

22 mirroring the work of Armitt. In the case of H2SO4, DADP was again a major product, however using HCl, chlorinated acetones comprised 70-97% of the product distribution.

This was again attributed to the in situ formation of hypochlorous acid (HOCl). This study also showed a difference in reaction rate, with 36% HCl degrading TATP almost three times faster than 65% H2SO4.

16 A publication from the TATP detection literature provides another illustration of the synergy of acid degradation in the presence of an oxidisable substrate. The focus of Xie and Cheng’s study was a method for the detection of TATP by chronoamperometry,33 however the reaction used to derivatise the peroxide sample is in itself an interesting degradation method. The peroxide sample was reacted in a MeCN solution of 12 M acetic acid, 0.4 M HCl and 6 mM KBr at 55°C. It is likely the small amount of HCl was necessary to trigger effective acid catalysis, as other studies have found acetic acid too weak an acid to perform this role.22 Under Xie and Cheng’s conditions, it was claimed that degradation of 0.1 mM solutions of TATP to form bromoacetones was effectively instantaneous as observed by GCMS (i.e. within mins), or one hour at room temperature.33 These products are almost certainly produced by an analogous route to the chloroacetones found in HCl-catalysed degradation.22, 32 Whilst the focus of this study was not on degradation, it further illustrates how an AO scavenger (in this case, Br-) is able to effect fast acid-degradation of organic peroxides.

To briefly summarise the work on TATP’s acidic degradation, it is clear that this type of reaction is effectively a rebalancing of the equilibrium to favor the starting materials (acetone and peroxide). As such, without stringent control of the conditions it is possible to continue to generate dangerous organic peroxides such as DADP through this process. Such control of conditions is hard to achieve in a field setting, and requires a trained chemist. Disregarding these practicalities for a moment, the most effective methods reported to date in terms of reaction rate are those that interrupt this equilibrium by consuming AO species as they are formed. The reaction with the hydroxy-hydroperoxide precursors of DADP seems critical, as the cyclic dimer is quite resistant to acid-catalysed ring-opening. An optimised acid-catalysed degradation scheme must include a suitable species which can be readily oxidised by

17 hydroperoxides, thus inhibiting the formation of DADP. Using HCl or adding other halogen anions achieves this aim, but unfortunately produces halogenated acetones.

These compounds have seen use as war gases during World War I, and are known to be toxic lachrymators.34-35 Thus the acid degradation mechanism remains unfavorable as a field solution in its current state of development.

1.3.2 Thermolysis

Though thermolysis is generally not considered as a field-relevant method of neutralising peroxide explosives*, its very distinct reaction mechanism makes it important to discuss in the context of seeking new methods of neutralisation. A great deal of our understanding of organic peroxide thermal degradation is derived from work focused on radical polymerisation initiators, where even inherently unstable cyclic peroxides have been proposed for the production of polystyrenes.36

TATP’s lack of thermal stability is attributed to the weak O-O bond which has been estimated at approximately 30.5 kcal/mol.37 Whilst theoretical calculations have suggested that C-O bonds in DADP are only slightly more stable than the O-O bonds at certain stages of the degradation,38 O-O scission is generally considered the primary degradation route for both oligomers.

A unimolecular, radical reaction was indicated by the work of Eyler et al studying the thermolysis of TATP in toluene solution (an effective radical trap).27 The presence of a high proportion of bibenzyl in the product mixture was an indication that the reaction was acting by a radical mechanism, leading to the coupling of two toluene molecules. Further elucidation of the mechanism was provided by Oxley, who investigated the thermolyses of TATP in the gas and condensed phase, as well as in the

* The heating of peroxides (or solutions thereof) is neither practical nor safe in an explosives disposal scenario. 18 presence of hydrogen-donating solvents.25 Oxley found results consistent with first- order kinetics across a 75°C temperature range (150 - 225°C) for the majority of the reaction course, suggesting that secondary reactions were not an important factor. A linear Arrhenius plot further suggested that the same rate-determining step was at work across the whole temperature range. The primary reaction products were acetone, CO2, methyl acetate and acetic acid. Interestingly, no DADP was observed in partially degraded TATP. The radical mechanism in Scheme 1.B was used to rationalise the results.

Scheme 1.B - General scheme for thermal degradation of TATP. Oxley proposed this scheme for rationalising the observations of TATP thermal degradation25: 1 – catalysed in gas-phase (a – 230°C; b – 150°C) 2 – catalysed in condensed-phase or hydrogen donating solvents 3- DADP is not found to form under the conditions studied, despite its greater thermal stability.

The presence of ester and absence of DADP were the two critical observations in this work. It was proposed that rapid β-scission of the C-O bonds adjacent to the radicals yielded two acetone molecules, leaving a diperoxyacetone diradical intermediate. In aprotic solvents, the diperoxy diradical was thought to rapidly collapse to O2 and dimethyl dioxirane (Scheme 1.B(1)). Formation of methyl acetate was highest in the lower temperature reaction, which was taken to indicate that it formed via an intramolecular re-arrangement of dimethyl dioxirane, a known reaction (Scheme

19 1.B(1b)).39 At higher temperatures, this di-radical species was considered too unstable to undergo this rearrangement, and instead decomposed to CO2 and methyl radicals

(Scheme 1.B(1a)), evidenced in the results by more products attributable to methyl radical reactions. In protic solvents this highly active species was thought to be somewhat stabilised, abstracting hydrogen from the solvent to form water and a further molecule of acetone (Scheme 1.B(2)). The absence of any trace of DADP in the samples

(despite it being much more stable than TATP under thermolysis) meant that a stepwise reaction similar to acid degradation (Scheme 1.B(3)) was ruled out.

1.3.3 Metals and metal ions

In the first paper dealing directly with the chemical neutralisation of TATP,

Bellamy chose the logical route of reacting the cyclic peroxide with an easily oxidisable

21 metal salt, in this case SnCl2. A ~0.35 M solution was found to be completely degraded by a double molar equivalent of SnCl2 in refluxing ethanol/toluene solution over approximately 1 hour. In this work, no attempt was made to analyse the products nor was there any discussion of the possible mechanism. As this work made no mention of anhydrous conditions, it seems likely that the reaction could have occurred due to the hydrolysis of SnCl2 leading to the formation of HCl. This then makes it unclear if the

Sn(II) ion alone facilitated the reaction with TATP, or if the acid-catalysed ring opening of the TATP produced reactive hydroperoxides which might then have oxidised Sn(II) to (presumably) SnIII. The fate of the inorganic salts was not investigated in this study. It is difficult to make any conclusion on this from the limited observations provided in the paper, as the work was targeted at providing a suitable method for laboratory destruction rather than a mechanistic investigation.

Bellamy’s study was used by Oxley as a starting point in her first key study on the field-relevant degradation of peroxides.24 As was highlighted in section 1.3.1, a 20 reasonable outcome of the degradation of TATP is the oxidation of a substrate. In

Oxley’s study, a broad range of metal salts were assayed in aqueous and neat THF or ethanol solution. In some cases the base metal was included in the reaction mixture and acid was also added to some of the assays. Complete destruction within 24 hours was achieved by ZnSO4 or CuCl2 in the presence of either copper or zinc metal

(respectively), or other metal salts, but the salts were not very effective on their own.

The requirement for both metal and ion to be present appears to indicate reductive couples being harnessed in these reactions. Neither organic nor inorganic products from these degradations were reported, so it is difficult to draw conclusion on the mechanism of neutralisation from this study.

40-41 ZnSO4 and CuCl2 both generate acidic conditions in aqueous solution, again making it possible that the reaction with the cyclic peroxide is actually still catalysed by

+ H , although the use of THF and ethanolic solvents in this study may reduce proton availability. Importantly, large (at least 3x) excesses of reagent to cyclic peroxide were needed for effective neutralisation, indicating that one of the reagents is acting as an oxidisable substrate. In the absence of information on the fate of the organic and inorganic reagents, again making impossible to draw further conclusions on how the metal salts are affecting the reaction of TATP.

The use of zero-valent metals is also found in a study by Fidler et al. which describes the use of mechanically alloyed bimetal ‘catalysts’ (sic) being used to neutralise TATP in methanol/water solution.42 The magnesium/PdC particles were found to degrade TATP, and whilst a rate constant of 1.3 x 10-3 Lg-1min-1 was provided, no information was given on the initial concentration of TATP. Once again, a large excess of metal was used, with 0.25 g of bimetal added to 5 mL of what would likely be quite a dilute solution considering that TATP would be only sparingly soluble in the 4:1

21 methanol/water used in the experiment. In this study acetone was reported as the major product, although the ratio of peroxide to ketone product was not given. No mention is given to the pH of the reaction media, so it is not possible to comment of the role of acid in this process.

In summary, the literature shows that metals are able to influence the rate of degradation of TATP, most likely via some form of redox reaction. It is not clear, however if the metals or metal ions themselves are able to open the peroxide ring structure, although it would seem in the bimetals study that this could possibly be the case. Thus whilst they have shown potential as reagents for the destruction of TATP and other organic peroxides, no overall mechanism has been suggested or can be extrapolated from the results published to date.

1.3.4 Other reactions

Nanoparticulate molybdenum hydrogen bronze Mo2O5(OH) has also been reported to successfully neutralise TATP.43 The material has been shown to be a convenient storage medium for reactive hydrogen,44 and as such it was suggested that these bronzes could readily reduce reactive hydroperoxides. It can be applied as a dark blue nanoparticulate solution, which is ideal for applying to explosive in a field setting.

The paper reports a fast reaction, with a colour change from dark blue to pale yellow, providing a simultaneous detection method. The following proposed reaction is given:

2 Mo2O5(OH) + ROOH → 4 MoO3 + H2O + ROH

The study does not describe the process of generating ROOH from the cyclic peroxide, but it is likely (on the basis of the broader literature) that the unopened peroxide ring would be resistant to reaction with the molybdenum bronze. The bronze was synthesised by the reduction of MoO3 by n-butanol in the presence of HCl. The liquid nanoparticulate reagent was then yielded as the filtrate when the crude 22 Mo2O5(OH) solid was washed with fresh butanol. This dark blue suspension was reserved and used for the TATP neutralisation reaction. It is highly likely that butanol suspensions synthesised by this method contains trace acid which can facilitate ring- opening to the reactive hydroperoxides ROOH. The hydroperoxides can then be reduced by the suspended Mo2O5(OH) via the suggested reduction reaction.

This paper suggested that the butanol suspension of Mo2O5(OH) could be used for in situ TATP degradation, yet this application would be limited as the proposed reaction required a stoichiometric amount of reagent. The reported weight fraction of suspended

Mo2O5(OH) in butanol was 2.61%, thus each gram of active reagent Mo2O5(OH) (3.4 mmol) would weigh 38.3 g, inclusive of solvent. If it is assumed that the reaction proceeds to completion at stoichiometric equivalence, the neutralisation of 100 g of

TATP (0.45 mol) would require over 5 kg of reagent solution. If the recent Western

Australian discovery of 3 kg of TATP is used as a potential field example, this would require at least 150 kg of the butanol suspension. If sufficient degradation is afforded by only degrading one peroxide unit of each TATP, this would still require 50kg of reagent. Indeed, these are optimistic figures that assume every available unit of molybdenum reagent is active, not just those at the surface of the nanoparticles.

Applying this volume of reagent to a compact package of solid explosive would be impractical. This difficulty appears to be reflected in the patents generated from this work. Despite the claims in the original paper that this reagent had application in bulk explosive neutralisation, it has only been patented for the detection of peroxides (and chlorates).45 It has also found commercial application as an inhibitor/colorimetric test for trace peroxide formation in organic solvents,46 where the stoichiometric nature of the reaction is not such a limitation.

23 1.3.5 Theoretical possiblities

A possible alternate direction to effect the degradation of TATP was indicated by the theoretical investigations of Dubnikova et al. on TATP binding with metal ions in the vapour phase. In silico calculations on the binding affinities of a panel of metal cations indicated the potential for a crown ether-like binding between the cyclic peroxide ring structure and the cation, which was suggested as a possible selective detection method in the vapour phase.47 The results indicated that Cu+, Li+, Cd2+, Zn2+ and In3+ would form progressively stronger complexes with the TATP peroxide ring.

Interestingly from a degradation perspective, the models indicated that Sb3+, Sc3+ and

Ti4+ bound with such affinity that it was energetically favourable to break the ring structure to form 3 acetone-peroxy moieties. A follow-on study investigated the possibilities such metal ions might have in degradation, and suggested that the cations might be useful as catalysts in the degradation of peroxide explosives.48 Both of these works were purely theoretical studies based on gas-phase interactions of the peroxides with the metal/metalloid ions, and it remains to be determined how these findings relate to experimental observations.

1.4 Other peroxide chemistry of relevance

1.4.1 Baeyer-Villiger reactions

A possible means of degrading organic peroxides is by harnessing organic reactions in which they or their acidic degradation products (hydroperoxides) are able to be used at higher rates, thereby driving the equilibrium to completion. An example of this approach can be seen in the acidic degradation reaction where overall destruction of

TATP was much faster when an oxidisable substrate such as Cl- or Fe2+ was present.22-

23, 29 A well-studied reaction of certain hydroperoxides is the Baeyer-Villiger (BV)

24 oxidation (or rearrangement). BV oxidations involve a ketone being reacted with a strong oxidant such as trifluoroperoxyacetic acid, which causes insertion of an oxygen atom to yield an ester (Scheme 1.C).

Scheme 1.C - The general scheme of the Baeyer-Villiger oxidation.

The BV reaction was first identified by Adolf Baeyer and Victor Villiger in 1899, when they oxidised menthone to its corresponding lactone (cyclic ester).49-50 In these first reactions, Caro’s acid (peroxymonosulfuric acid) was used as the oxidant, however other oxidants including perbenzoic acid, metachloroperbenzoic acid (MCPBA) and trifluoroperacetic acid can be utilised to perform the same reaction. Whilst the reaction was identified at the start of the 20th century, it was not until over 50 years later that a coherent understanding of the mechanism was reached.

It was an elegant experiment designed by Doering and Dorfman that provided the vital insight into the mechanism that facilitates the rearrangement.51 At that time there were three competing theories: the formation of a carbonyl oxide intermediate, the formation of a dioxirane intermediate, or the so-called Criegee intermediate. As depicted in Scheme 1.D, each of these possibilities would deliver a different result when

O18-labelled benzophenone was reacted with perbenzoic acid. Upon completion of the oxidation, the ester was reduced with lithium aluminum hydride to phenol and benzyl alcohol. It was found that the benzyl alcohol contained 93% of the labeled oxygen with the phenol containing none, thus confirming the labeled oxygen was retained in the carbonyl, illustrating that the Criegee intermediate provided the best explanation of the reaction behaviour.

25 Scheme 1.D - Elucidation of the Criegee Intermediate. The spheres identify the labeled oxygen atoms. The wavy line indicates where the ester was reductively cleaved by LiAlH to the respective aromatic alcohols.

Generally, Baeyer-Villiger oxidations are carried out under acidic conditions.

Under these conditions, it is thought that the protonation of the ketone carbonyl enhances the nucleophilic addition of the hydroperoxidic oxygen, although there is significant conjecture whether this is a stepwise (ionic) or concerted (neutral) process.52

Thus, the acid catalyst drives the reaction by ‘activating’ the ketone substrate. In addition to Brønsted acids, some Lewis acids are also capable of catalyzing Baeyer-

Villiger oxidations. The example found in the literature with the highest activity was a

Sn-Zeolite complex reported by Corma.53 It has been suggested that certain platinum- based catalysts are able to activate both the substrate ketone and the hydroperoxide oxidant by simultaneously coordinating to both.54 This peroxy-platinum quasi-cyclic intermediate is thought to enhance the rate of reaction by increasing the nucleophilicity

26 of the hydroperoxide, and also facilitates the loss of OH from the Criegee intermediate, which is usually a rate-determining barrier due to HO- being a poor leaving group.

1.4.2 Redox and electrochemistry of cyclic peroxides

Electrochemistry is the branch of chemistry that deals with the interface between chemical processes and their electrical currents and potentials.55 Electrochemical studies are an excellent way to understand electron transfer reactions, and can be readily applied to the determination of reaction mechanisms.56-58 To date, most electrochemical studies of TATP and related peroxides have focused on detection of its breakdown products. Of these studies, one monitors bromine formation during bromide-based degradation of TATP,33 whilst all others focus on the detection of hydrogen peroxide after the photochemical36, 59 or acidic decomposition60-61 of TATP. Only one of these studies makes any reference to the electrochemical behaviour of the cyclic peroxide itself. Without other studies to compare against, it is difficult to assess the reliability of this isolated report. This leaves an information gap that limits our ability to properly understand the breakdown mechanism of these unique compounds. Since organic peroxides are generally classed as oxidising agents, and much of their chemistry might be considered redox-based, information from electrochemical studies might inform the design of a more effective reaction for the destruction of cyclic peroxides.

1.4.3 Lewis acid reactions with organic peroxides.

Lewis acids are a class of compounds that can easily accept an electron pair, making them ideal for reacting with peroxides due to their affinity for the lone pairs of the peroxidic oxygens. There are few examples of the reaction of Lewis acids with organic peroxides. Diacyl peroxides are known to rearrange to their respective esters plus CO2 in a reaction catalysed by antimony pentachloride via an ionic mechanism

27 rather than a radical mechanism. Dialkyl peroxides, a similar class of compound to

TATP, are known to be catalytically decomposed by Fe2+ in aqueous solution producing largely ketone products.4 This reaction may proceed by the same underlying mechanism that led to Tsaplev’s observation that the presence of Fe(II) increased the rate of DADP degradation in acidic solution.29 Copper salts have also been shown to be active in catalytic degradation of dialkyl peroxides,4 although how this relates to the activity of copper salts in Oxley’s study24 is not clear. There has been no experimental study which has shown the mechanism of the reaction of Lewis acids with cyclic ketone peroxides such as TATP. Interestingly, many of the metal ions found to break down TATP’s cyclic structure in the theoretical studies of Dubnikova47-48 (discussed in detail in section 1.3.5) can form strong Lewis acid compounds (eg. TiCl4, SbCl3), but no study of the reaction of these compounds with TATP or other organic peroxides has been published.

1.5 Related peroxides

As has been discussed, the hazards of friction, heat, impact and electrostatic sensitivity make working with organic peroxides a hazardous proposal. The particular tendency of TATP to sublime at room temperature increases these risks further and make TATP generally difficult to store. This same behaviour also places limitations on laboratory techniques that can be used (eg, vacuum), and can make quantitative work difficult, particularly in anhydrous conditions. In many laboratories, it is considered an unacceptable risk to synthesise TATP and handle it in the quantities that would be required for certain analytical work. The concept of working with a safer model peroxide, closely related in structure and chemistry to TATP but with reduced hazards, was an extremely attractive approach for a number of reasons:

28 1. The ability to handle solid samples without losses through sublimation,

permitting flexibility in experimental design;

2. Comparing the reactions of related peroxides would allow an exploration

of structure-activity rules, shedding light on reaction mechanisms;

3. Slightly different substituent groups may make it easier to isolate related

products such as intermediate degradation products for greater flexibility

in understanding reaction mechanisms; and

4. An improvement in laboratory safety.

Despite these possibilities, the use of model peroxides has not seen application in the literature of organic peroxide degradation. TATP and DADP are often considered side by side under the same conditions,21-22, 24, 29 however the two compounds share the same monomers and intermediates. Comparative mechanistic deductions are limited in this case as DADP itself can be a product of TATP degradation. It is probable that the imperative of generating a field-relevant outcome has driven a focus directly on the behaviour of the threat molecules of interest, rather than on the general chemistry of the geminal-peroxide moiety that imparts this particular explosive hazard. The use of related peroxide structures to help elucidate field-relevant OPE degradation reaction mechanisms is not known.

Fortunately, cyclic gem-peroxides (of which TATP is but one example) are not only studied in the context of their explosive properties, but also as radical catalysts for the synthesis of polymers such as polystyrene. In fact TATP itself has been studied with this application in mind.62 The polymerisation industry’s need for strong understanding of reaction kinetics in order to control the degree of polymerisation has resulted in some structure-activity studies, although these studies are usually only carried out under the thermal conditions used to initiate such polymerisation reactions.

29 Figure 1.3 - Structures of acetone- and pentanone-based peroxides.

TATP’s pentanone-based homologue (tripentanone triperoxide (TPTP)) has been investigated for this application.36 As shown in Figure 1.3, the peroxide ring structures of TPTP and TATP are identical, with the addition of a methylene to each of the alkyl arms being the only difference. The degradation of TPTP under thermal conditions was found to proceed by a similar unimolecular radical mechanism to that found in TATP.25,

36 TPTP was found to adopt a twisted boat-chair confirmation similar to TATP in crystallographic studies, and it was also reported that TPTP was up to 50 times ‘more stable’ than TATP,63 although it was not clear what metric was being used to describe stability by the author (possibly thermal stability).

In contrast to the good body of work comparing TATP and TPTP from the thermal degradation perspective, very little is known about how other types of degradation reactions are affected by changing substituents on the molecule. Some minor points regarding TPTP’s acidic degradation can be inferred from the earliest comprehensive report of its synthesis by Milas et al.64 They made a short note of altered product distributions between syntheses catalysed by HCl and H2SO4, noting that a particular hydroxyl-hydroperoxy open chain dimer was not detected in the presence of

HCl. This species was hypothesised to be a precursor to the cyclic dimer dipentanone diperoxide (DPDP) as illustrated in Scheme 1.F, a conclusion also reached by more recent studies on the mechanism of DADP formation.14, 22 Whilst Milas did not explicitly comment on the effect of HCl on the formation of DPDP, the absence of its

30 likely precursor (the hydroxyl-hydroperoxy open chain dimer) links to the reduced presence of DADP in the degradation of TATP under equivalent conditions.22 This almost passing observation from Milas provides some indication that the acidic degradation of TATP and TPTP may be quite similar, or at least proceed by the same general mechanism.

Scheme 1.E – The formation of cyclic ketone-peroxide dimers. Similarities in observations in the synthesis of TPTP (R=ethyl)64 and TATP (R=methyl)14, 22 suggest a similar route to the cyclic dimer.

The similarities presented thus far and the reported ‘stability’ of TPTP would suggest that it is a suitable model compound for TATP. This should be taken with some caution however, as the presence of the methyl vs. ethyl groups adjacent to the geminal peroxide may have a large effect on the kinds of chemistry that can take place. For instance, primary carbocations are known to be significantly less favourable as intermediates than secondary carbocations. Thus any mechanism that requires a primary carbocation in the transition state (for example, alkyl migration) may be more accessible for TPTP but not TATP. This does not discount TPTP as a useful model – indeed the potential for different chemistry may facilitate the identification of reactions which are energetically unfavorable for TATP normally, but could be accessed through catalysis or modification of the reaction conditions. Thus a comparative study has the potential to add new concepts to facilitate the degradation of cyclic peroxides.

31 1.6 Conclusion

Whilst the degradation of cyclic peroxides has been studied since as early as the

1960s,23 the aim of providing first responders with a means of removing the explosive hazard that these materials pose was first stated explicitly by Bellamy in 1999.21 In the intervening years, a number of research groups have extended our understanding of two mechanisms in particular – thermal and acidic degradation. It is valuable to consider the progress on these two general methods and relate them to the field environment.

Thermal degradation mechanisms of cyclic peroxides have been studied in the context of their application as radical catalysts, yet the relevance to in situ neutralisation is limited. Most thermolysis studies take place under extreme temperatures (>150°C) and pressures which would pose an unacceptable risk of detonation in a solid explosive sample. In fact, it has been proposed on the basis of modeling studies that the detonation of TATP begins through the same homolytic cleavage of the peroxide bond which occurs in thermolyis.65 This similarity highlights the inherent risk in using heat to degrade organic peroxides. Thermolysis can only safely be conducted in solution, thus a first responder would be first required to dissolve the solid material, which in itself can be practically challenging to perform safely. Once dissolved, the advantage afforded by any further treatment via in situ thermolysis is questionable, as the sensitivity of peroxide solutions is already greatly reduced to most stimuli. It is much easier to remove the solution off site to a safe area and use a less complicated method such as burning off the dilute solution. For this reason, it is unlikely that an in situ neutralisation method could rely solely on thermolysis.

Having considered the practicalities of thermolysis, it is apparent that an ideal field neutralisation method should be applicable directly to the explosive peroxide, without requiring dissolution, and should affect neutralisation in situ. This would imply

32 a chemical approach is needed. Of the common organic peroxides, MEKP has only been studied from a thermolysis perspective,66 and HMTD has received limited attention in the literature.24 Only TATP has been investigated to an extent that sheds significant light on its mechanism of chemical degradation.

It is now well understood that acids catalyse the ring opening and ring closing of

TATP based on a complex equilibrium involving water, acid, ketones and intermediate hydroperoxides.14 In situations where the acidic counterion does not participate (ie

H2SO4), the key product is DADP. Clearly this does not achieve the outcome of removing the peroxide threat, as DADP is equally hazardous. Where the acidic counterion can react with (and is usually oxidised by) intermediate hydroperoxides, electrophilic species (such as hypochlorite in the case of HCl) are often generated.

These species can interrupt DADP formation by reacting with intermediate AO species.

Whilst this alleviates the peroxide threat, care must be taken that the side-reactions do not generate a secondary threat; for example the chloride ion in HCl leads to the formation of chlorinated ketones, which are known lachrymators and are highly toxic.

The findings to date indicate that H+ in itself is not an effective neutralisation method as the primary product is explosive, whilst in the case of HCl the outcome is undesirable (toxic products). Despite these limitations, there is scope to explore acid- based methods; for example the acid counter-ion might be specifically selected to give end products that are both safe and non-toxic. This potential solution might more accurately be described as ‘acid-catalysed reduction’ of the cyclic peroxide, involving the oxidation of a substrate.

Other groups have suggested novel solutions such as the use of metal salts and nanoparticulate metals to chemically degrading the peroxide hazard. On the whole, experimental investigation of the reaction mechanisms operating in such procedures has

33 not been published, nor detail given of the products formed by these reactions. The degradation is purely articulated in terms of decreasing the concentration of organic peroxide. Whilst the lack of detail is understandable in the event that some of these research groups are protecting commercial interests, it makes consideration of their field applicability extremely difficult. The only discussions of mechanism come from theoretical investigations involving vapour phase reagents, which are somewhat removed from reactions in solution with oxidisable counter-ions. Without a detailed understanding of the mechanism underlying these reactions, it is difficult to design a catalytic system for the neutralisation of these materials.

In considering the acid-catalysed reduction24 and the nano-particulate/metal ion reduction43 of cyclic peroxides, one common factor is they are both stoichiometric reactions. This has significant implications for field use, as such a method would require at least a molar equivalent of reagent. Depending on molecular weight, the amount required may be in the order of or greater than the weight of explosive material itself, not including any additional solvent that might be required to facilitate the reaction. The transport of such bulky reagents and the process of applying this quantity of material to the peroxide could become problematic. Hence the use of a stoichiometric reagent system, unless extremely dense and of low molecular weight, may be an unfeasible field neutralisation proposition.

If a chemical neutralisation method requires minimum reagent mass, then a catalytic approach seems the most appropriate solution. Such a catalyst may function in a number of possible ways:

1. It could itself trigger the ring-opening and subsequent degradation of the

cyclic peroxide to form the initial starting materials (thus retaining ROOH or

HOOH as the oxidising species);

34 2. It may catalyse the oxidation of a low-molecular weight substrate by AO

species formed via the acidic ring-opening of the cyclic peroxide. This

substrate could also act as the solvent, thereby reducing the overall bulk and

weight of material required to effect neutralisation.

One paper claimed TATP’s chemical neutralisation could be ‘catalysed’ by using alloyed bimetals,67 but did not provide sufficient detail to substantiate the claim. The identification of acetone as the major product in this case indicates either HOOH was also generated, or another stoichiometric (and possibly catalysed) oxidation reaction took place. The latter is more likely considering the known susceptibility of metals to oxidation by AO species. In any case, it is not clear if the initial reaction with TATP was in fact catalysed by the metal. Whilst catalysts have been identified to enhance the rate of thermal degradation of TATP,68-69 the rate increase is rather modest and the barriers to the field application of thermolysis are unchanged. Consequently, these catalysts are not considered suitable for the suggested in situ degradation method.

1.7 This Study

The overall goal of this study is to identify new methods by which the degradation of organic peroxides, in particular TATP, can be achieved, and to add to the understanding of the mechanisms underlying the known methods. Whilst there are limitations to the field application of both thermal and acidic methods to date, to strike out on a completely new path without comparison to this existing work would be counter-productive. The preceding analysis of the literature has highlighted a number of knowledge gaps where potential new methods might be identified:

1. Studies of closely related cyclic peroxides under equivalent chemical conditions

might reveal common mechanisms or reactive intermediates which are not

35 energetically favourable or directly observable in the reaction conditions studied

to date. This is particularly pertinent considering the very high energetic barriers

which can be encountered in reactions occurring adjacent to methyl moieties, as

seen with TATP. Thus, it is prudent to revisit some existing work, especially

acidic degradation, utilising a suitable pair of related peroxides, such as TPTP

and TATP.

2. Whilst it has been noted that TATP is quite resistant to chemical reduction, this

has not been explored electrochemically. Such a study may well provide further

detail on the basis of the chemical stability.

3. Whilst the ability of metals to catalyse the degradation of peroxides are not new,

only studies of thermolysis have demonstrated true catalytic behaviour with

cyclic peroxides. Furthermore any detailed conclusions regarding possible

mechanisms of catalytic cyclic peroxide degradation have so far been confined

to theoretical studies. The suggested ability of particular metal ions to directly

trigger the degradation of TATP has yet to be explored experimentally and may

provide some novel chemistry.

The desired catalytic approach is best sought via a thorough understanding of the reaction mechanisms. By broadening our investigations to discern the mechanisms driving existing and novel reactions, it is possible to evaluate which methods are most likely to achieve the target of safe, fast and practical chemical neutralisation of organic peroxides explosives.

1.8 References

1. Miyake, A.; Takahara, K.; Ogawa, T.; Ogata, Y.; Wada, Y.; Arai, H., Journal of Loss Prevention in the Process Industries 2001, 14 (6), 533-538.

36 2. Meyer, R.; Köhler, J., Explosives. 4th, rev. and extended ed.; VCH: New York, N.Y., USA, 1993; p 457. 3. Kuzmin, V. V.; Solov'ev, M. Y.; Tuzkov, Y. B.; Kozak, G. D., Central European Journal of Energetic Materials 2008, 5 (3-4), 77-85. 4. Swern, D., Organic Peroxides. Wiley-Interscience: New York, 1970. 5. Mageli, O. L.; Kolczyns.Jr, Industrial and Engineering Chemistry 1966, 58 (3), 25-32. 6. Brown, W., Organic Chemistry. Saunders College Publishing: Orlando, 1995. 7. Milas, N. A.; Golubovic, A., J. Am. Chem. Soc. 1959, 81 (21), 5824-5826. 8. Yeh, P. Y.; Shu, C. M.; Duh, Y. S., Ind. Eng. Chem. Res. 2003, 42 (1), 1-5. 9. Matyas, R.; Pachman, J., Primary Explosives. Springer-Verlag: Heidelberg, 2012; p 338. 10. Wilson, N., Journal of Electrostatics 1977, 4 (1), 67-84. 11. Talawar, M. B.; Agrawal, A. P.; Anniyappan, M.; Wani, D. S.; Bansode, M. K.; Gore, G. M., J. Hazard. Mater. 2006, 137 (2), 1074-1078. 12. Wolffenstein, R., Chem. Ber. 1895, 28, 2265-9. 13. Matyas, R.; Pachman, J., Propellants Explosives Pyrotechnics 2010, 35 (1), 31- 37. 14. Oxley, J. C.; Smith, J. L.; Bowden, P. R.; Rettinger, R. C., Propellants Explosives Pyrotechnics 2013, 38 (2), 244-254. 15. Peterson, G. R.; Bassett, W. P.; Weeks, B. L.; Hope-Weeks, L. J., Cryst. Growth Des. 2013, 13 (6), 2307-2311. 16. West, B.; Stewart, S. Uncomfortable Truths and the Times Square Attack, Stratfor, webpage: http://www.stratfor.com/weekly/20100505_uncomfortable_truths_times_square_attack (accessed 11/02/2014). 17. Author_Unknown, Australian Broadcasting Commission: ABC News, 2013. 18. Rai, M., 7/7 : the London bombings, Islam and the Iraq War. Pluto Press: London ; Ann Arbor, MI, 2006; p 196. 19. Beckman, J., Comparative legal approaches to homeland security and anti- terrorism. Ashgate Publishing: Burlington, VT, 2007; p 185. 20. Oxley, J. C.; Smith, J. L.; Moran, J.; Nelson, K.; Utley, W. E., Training dogs to detect triacetone triperoxide (TATP), in Sensors, and Command, Control,

37 Communications, and Intelligence(C31) Technologies for Homeland Security and Homeland Defense III, Pts 1 & 2, Orlando FL, 2004, Carapezza, E., Ed. Proceedings of SPIE, 2004, 5403, pp 349-353. 21. Bellamy, A. J., Journal of Forensic Sciences 1999, 44 (3), 603-608. 22. Armitt, D.; Zimmermann, P.; Ellis-Steinborner, S., Rapid Commun. Mass Spectrom. 2008, 22 (7), 950-958. 23. Furuya, Y.; Ogata, Y., Bull. Chem. Soc. Jpn. 1963, 36 (4), 419-422. 24. Oxley, J. C.; Smith, J. L.; Huang, J. R.; Luo, W., Journal of Forensic Sciences 2009, 54 (5), 1029-1033. 25. Oxley, J. C.; Smith, J. L.; Chen, H., Propellants Explosives Pyrotechnics 2002, 27 (4), 209-216. 26. Matyas, R.; Pachman, J., Science and Technology of Energetic Materials 2007, 68 (4), 111-116. 27. Eyler, G. N.; Mateo, C. M.; Alvarez, E. E.; Canizo, A. I., J. Org. Chem. 2000, 65 (8), 2319-2321. 28. Eyler, G. N.; Canizo, A. I.; Alvarez, E. E., Afinidad 2007, 64 (530), 538-542. 29. Tsaplev, Y. B., Kinetics and Catalysis 2012, 53 (5), 521-524. 30. Matyas, R.; Pachman, J.; Ang, H. G., Propellants Explosives Pyrotechnics 2009, 34 (6), 484-488. 31. Fitzgerald, M.; Bilusich, D., Journal of Forensic Sciences 2011, 56 (5), 1143- 1149. 32. Oxley, J. S., J. L.; Brady, J. E.; Steinkamp, L., Propellants Explosives Pyrotechnics 2014, 39 (2), 289-298. 33. Xie, Y. Q.; Cheng, I. F., Microchem. J. 2010, 94 (2), 166-170. 34. Heller, C. E. Chemical Warfare in World War I: The American Experience, 1917- 1918 Leavenworth Papers [Online], 1984, p. 15. http://www.worldwar1.com/dbc/pdf/chemwarfare.pdf. 35. Hazardous Substances Database, National Library of Medicine, http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen?HSDB, accessed 30/10/2013. 36. Canizo, A. I., Trends in Organic Chemistry 2006, 11, 55-64. 37. Golubev, V. K., Effect of charged and excited states on the decomposition mechanism of several peroxides molecules, in New Trends in Research of Energetic Materials, Czech Republic, 2012, pp 147-172.

38 38. Hong, S. G.; Li, Y. H., Theochem-Journal of Molecular Structure 1996, 363 (3), 339-342. 39. Murray, R. W., Chemical Reviews 1989, 89 (5), 1187-1201. 40. Scientific, F. MSDS - Copper (II) Chloride dihydrate, Iowa State University, Online multimedia: http://avogadro.chem.iastate.edu/msds/cucl2-2h2o.htm. 41. Wang, W.; Dreisinger, D. B., Metallurgical and Materials Transactions B- Process Metallurgy and Materials Processing Science 1998, 29 (6), 1157-1166. 42. Fidler, R.; Geiger, C. L.; Clausen, C. A.; Sigman, M. E., Abstracts of Papers of the American Chemical Society 2008, 235, 1053. 43. Apblett, A. W.; Kiran, B. P.; Malka, S.; Materer, N. F.; Piquette, A., Ceram. Trans. 2006, 172, 29-35. 44. Apblett, A. W.; Kiran, B. P.; Oden, K., Chlorinated Solvent and DNAPL Remediation 2003, 837, 154-164. 45. Apblett, A. W.; Materer, N. F. Nanometric ink for detection of explosives. WO2011041006 A3, 2011. 46. Apblett, A. W.; Materer, N. F. Colorimetric reagent for prevention of peroxide formation in solvents. PCT/US2011/023723, 2011. 47. Dubnikova, F.; Kosloff, R.; Zeiri, Y.; Karpas, Z., J. Phys. Chem. A 2002, 106 (19), 4951-4956. 48. Dubnikova, F.; Kosloff, R.; Oxley, J. C.; Smith, J. L.; Zeiri, Y., J. Phys. Chem. A 2011, 115 (38), 10565-10575. 49. Baeyer, A.; Villiger, V., Berichte Der Deutschen Chemischen Gesellschaft 1899, 32 (3), 3625-3633. 50. Baeyer, A.; Villiger, V., Berichte Der Deutschen Chemischen Gesellschaft 1900, 33, 124-126. 51. Doering, W. V.; Dorfman, E., J. Am. Chem. Soc. 1953, 75 (22), 5595-5598. 52. Alvarez-Idaboy, J. R.; Reyes, L.; Mora-Diez, N., Org. Biomol. Chem. 2007, 5 (22), 3682-3689. 53. Corma, A.; Nemeth, L. T.; Renz, M.; Valencia, S., Nature 2001, 412 (6845), 423- 425. 54. Gavagnin, R.; Cataldo, M.; Pinna, F.; Strukul, G., Organometallics 1998, 17 (4), 661-667.

39 55. Nagy, Z. Electrochemistry Introduction, The University of North Carolina, webpage: http://www.scribd.com/doc/191245419/Electrochemistry-Introduction-FR (accessed 3/08/2010). 56. Bard, A. J.; Faulkner, L. R., Electrochemical method: fundamentals and applications. Wiley: New York ; Chichester, 1980; p xviii,718p. 57. Braide, O.; Helfrick, J. C.; Bottomley, L. A., Abstracts of Papers of the American Chemical Society 2006, 231, -. 58. Gallardo, I.; Vila, N., J. Org. Chem. 2010, 75 (3), 680-689. 59. Schulte-Ladbeck, R.; Karst, U., Chromatographia 2003, 57, S61-S65. 60. Munoz, R. A. A.; Lu, D. L.; Cagan, A.; Wang, J., Analyst 2007, 132 (6), 560-565. 61. Laine, D. F.; Roske, C. W.; Cheng, I. F., Anal. Chim. Acta 2008, 608 (1), 56-60. 62. Morales, G.; Eyler, G. N.; Cerna, J. R.; Canizo, A. I., Molecules 2000, 5 (3), 549- 550. 63. Cerna, J.; Bernes, S.; Canizo, A.; Eyler, N., Acta Crystallographica Section C- Crystal Structure Communications 2009, 65, O562-O564. 64. Milas, N. A.; Golubovic, A., J. Am. Chem. Soc. 1959, 81 (13), 3361-3364. 65. van Duin, A. C. T.; Zeiri, Y.; Dubnikova, F.; Kosloff, R.; Goddard, W. A., J. Am. Chem. Soc. 2005, 127 (31), 11053-11062. 66. Kirillov, A. I., Zh. Org. Khim. 1965, 7 (1), 1226. 67. Fidler, F. L., T.; Carvalho-Knighton, K.; Geiger, C.L.; Sigman, M.E.; Clausen, C. A., Degradation of TNT, RDX, and TATP using Microscale Mechanically Alloyed Bimetals. In Environmental Applications of Nanoscale and Microscale Reactive Metal Particles, Geiger, C. L., et al, Ed. American Chemical Society: Washington DC, 2009; pp 117-134. 68. Tasca, J.; Lavat, A.; Nesprias, K.; Barreto, G.; Alvarez, E.; Eyler, N.; Canizo, A., Journal of Molecular Catalysis A-Chemical 2012, 363, 166-170. 69. Zareba, M.; Legiec, M.; Sanecka, B.; Sobczak, J.; Hojniak, M.; Wolowiec, S., Journal of Molecular Catalysis A -Chemical 2006, 248 (1-2), 144-147.

40 CHAPTER 2. Tripentanone triperoxide as a

model for TATP degradation studies*

TATP’s inherent sensitivity to shock, heat and friction is a significant hazard in its own right, but is considerably magnified by the tendency of solid TATP to sublime and redeposit on nearby surfaces. This makes long-term storage of samples a difficult proposition. At an early stage of this research project, it was identified that a safer analogue compound would allow more flexibility in experimental design, as well as reduce overall risk whilst conducting initial experiments on cyclic peroxide degradation.

2.1 Introduction

As discussed in section 1.5, tripentanone triperoxide (TPTP) was identified as a suitable candidate for these purposes for the following reasons:

1. Literature suggested TPTP was up to 20-50 times more ‘stable’ than

TATP.1† It should be noted that this study did not provide an explanation on

what measurement of stability was used in this statement, however it seems

thermal stability in solution was the basis of this statement.

2. The limited literature on the behaviour of this peroxide in HCl media

suggested a similar acidic degradation,2 including inhibited formation of the

dimer in the presence of acids with oxidisable conjugate bases.

* This chapter has been published in part as: Bali, M. S.; Armitt, D.; Wallace, L.; Day, A. I., Journal of Forensic Sciences 2014, 59 (4), 936-942. † It was noted during review that the referenced publication had a typographical error, and that TPTP was in fact less stable that TATP in solution. Whilst this might be the case, TPTP was chosen on the basis that this reference asserted it might be more stable in the solid phase, thus this reference has been retained. 41 Figure 2.1- - Pentanone and Acetone peroxides in this study. TPTP – tripentanone triperoxide. DPDP – dipentanone diperoxide, DPPDHP – dipentanoneperoxy dihydroperoxide, TATP – triacetone triperoxide, DADP – diacetone diperoxide.

TPTP and DPDP (Figure 2.1) have been researched as radical initiators for styrene polymerisation,3 and DPDP investigated as a potential anti-malarial drug,4-5 but no reference could be found in which their sensitiveness to intiating stimuli was quantified. The longer ethyl moieties on TPTP would facilitate better van der Waals interactions, hence it could be reasonably expected that TPTP would have a lower vapour pressure than TATP, potentially reducing the sublimation hazard. The ethyl arms would also be expected to produce slightly different chemical reactivity. This could expose mechanisms or intermediates which are not detectable or are not energetically possible for TATP.

Comparative studies between TPTP and TATP have thus far been limited to thermal degradation kinetics.6-7 It was hoped that by extending these studies to chemical reactions, it might be possilbe identify new neutralisation reaction whilst simultaneously achieving our intent of reducing risks in experimental work.

2.2 Research Goals

The aim of this part of the research was to establish the suitability of TPTP as a reliable and safer model compound on which to base our exploration of new methods for the neutralisation of organic peroxides. To establish this, the following key goals were pursued:

42 1. Reproduce and, if necessary, modify literature procedures for the synthesis,

purification and analysis of TPTP to suit the requirements of the study;

2. Extend the characterisation of TPTP, particularly regarding its sensitiveness

to initiating stimuli and tendency for hazardous sublimation;

3. Conduct experiments on the thermal degradation of TPTP to compare

against the literature, and conduct a more thorough investigation of the

reaction products for mechanism elucidation;

4. Conduct novel experiments on the acidic degradation of TPTP to compare

against the established behaviour of TATP under similar conditions,8-10 and

compare their relative mechanisms of degradation;

5. Explore the potential of TPTP as a ‘pseudo-scent’ in canine scent training,

an application where TATP’s volatility similarly poses significant hazards.

2.3 Experimental

CAUTION! The organic peroxides detailed in this study have the characteristics of primary explosives with extreme levels of friction, heat and impact sensitiveness. All procedures must be carried out by properly qualified and equipped personnel taking all relevant precautions.

2.3.1 Instruments

A Shimadzu QP 2010 Ultra fitted with a SGE SolGel Wax column (1.0 µm film, I.D.

0.25 mm, 30 m) was used for all reported GCMS methods. All analyses (except thermal degradation in toluene) of peroxides and degradation products utilised an initial temperature of 40°C for 1 min, 20°C/min ramp to 100°C, held for 3 mins, followed by

30°C/min finishing at 190°C and held for 2 mins. Injector temperature was 100°C, interface 200°C, and linear column flow at 2 mL/min of He gas. Split ratio was set at

43 100. Concentrations were quantified using external standards. Standard concentration curves used four data points, and were not corrected to intersect at the origin. GCMS standards were of analytical quality except ethyl propanoate which was synthesised by the acid catalysed esterifcation of propanoic acid and ethanol, purified by distillation and its purity confirmed by NMR. Aqueous acidic samples were neutralised over

CaCO3 before analysis. Non-acidic mixtures free of non-volatiles were sampled directly.

For the thermal degradation of TPTP in toluene, products were analysed by

GCMS using an initial temperature of 40°C for 1 min, 20°C/min ramp to 100°C, held for 3 mins, followed by 30°C/min finishing at 278°C and held for 5 mins. Injector temperature was 285°C, interface 270°C, and linear column flow at 2 mL/min of He gas. Split ratio was set at 100. Bibenzyl, ethyl propanoate and 3-pentanone were purified by published procedures,11 and calibrated against 0.005 M naphthalene in acetonitrile as an internal standard. Approximate concentrations of other products were estimated qualitatively from their peak area against the internal standard, and were identified by matching against the GCMS software’s NIST mass spectral library.12

[TPTP] was determined separately by using the previously described lower temperature method, due to degradation in the high temperature method.

NMR spectra were recorded on a Varian Unityplus-400 spectrometer. All NMR experiments were conducted at 25°C unless otherwise stated. 1H NMR spectra were referenced to TMS (0 ppm) at 25°C using the residual 1H signal of the respective deuterated solvent. 1D spectra were recorded with between 16 and 1024 transients. T1 experiments were conducted using the Varian VNMR software using inversion recovery method and relaxation delay of 20 s between increments. An initial pulse (calibrated to

180°) was followed by an incremental delay between 1.25 ms and 20.48 s then a 90° observe pulse. T1 was obtained through exponential data analysis by VNMR software.

44 IR Spectra were recorded on a Shimadzu IRPrestige-21 spectrometer. Samples were prepared in KBr pellets and analysed between 400-3500 cm-1 with 16 scans.

2.3.2 Sensitiveness testing

Explosive sensitiveness testing was carried out at Defence Science and

Technology Organisation (DSTO; Edinburgh, South Australia). The Rotter impact test was utilised for impact sensitiveness testing. A 30 mg test sample in a brass cap was inverted over a steel anvil and connected to a manometer, and a 5 kg weight was dropped from varying heights. An initiation was assumed if gas evolution was >1 mL.

The standard BAM Friction Test method was used for friction sensitiveness testing.

Electrostatic discharge sensitiveness was tested on a custom made DSTO apparatus.

The sample was placed in 5 individual holes of a plastic disk over a copper base. A copper disc was placed over each hole, and the electrode lowered to touch the disc. The machine was armed and the charge released 5 times for each hole. No initiation is defined as 25 shots at a particular charge without reaction. Test energies of 0.045, 0.45 and 4.5 J were used for the test. Temperature of initiation was measured by a 200 mg sample placed in a glass tube and heated at 5°C/min until reaction was indicated by smoke, flame or explosion.

2.3.3 Synthesis

TPTP and DPDP were synthesised using a procedure based on the method of

Cerna et al.1 A typical synthesis was conducted as follows: 14 g of 70% w/w aqueous

H2SO4 was added to 6.34 g (56 mmol) of 30 % w/w aqueous H2O2, and the mixture was then cooled to -10°C. 4.3 g (50 mmol) of 3-pentanone was added dropwise over 1 h with continued stirring, and the temperature maintained for 3 days. The mixture was extracted with 100 mL of 40°C BP petroleum ether, and the organic layer washed 3

45 times with 50 mL saturated aqueous ammonium chloride, 3 times with 50 mL distilled water then dried overnight over Na2SO4. The solvent was removed under vacuum to yield a crude mixture of DPDP, TPTP and mixed hydroperoxides.

Separation of these products was achieved using flash chromatography (2.5 cm dia., 20 cm length of 200 mesh silica gel, eluted with petroleum ether BP 60-70°C).

TPTP was eluted first followed closely by DPDP (DPDP could only be isolated with a minimum of ~20% TPTP impurity) and finally the polar, 1,1'-[dioxybis(1- ethylpropylidene)]dihydroperoxide (3). 3 was eluted in high purity, and was

1 characterised by NMR. H NMR (CDCl3): δ 0.95 (t, J=8.2Hz, 12H), 1.71 (m, 8H), 9.60

(s, 2H) 13C NMR: δ 8.0, 22.1, 115.4. The NMR assignment including 2D spectra correlates with the structure proposed by Milas.2 Repeated flash chromatography of the

DPDP-rich fractions via column chromatography yielded very small amounts of

1 relatively pure DPDP as a colourless oil MP ~ -8°C. H NMR (CDCl3): δ 0.93 (t,

J=7.6Hz, 12H), 1.62 (br, 4H), 2.21 (br, 4H); 13C NMR: δ 8.0, 22.1, 115.6. IR (cm-1)

2978, 2947, 2886, 146, 1385, 1278, 1155, 1130, 966, 922, 861, 759, 615, 558.

The TPTP-rich fractions were concentrated under vacuum and dissolved in minimal warm methanol. A drop of water was added, and overnight refrigeration yielded a white needle-like precipitate which was dried in vacuo at room temperature overnight to yield TPTP (825 mg, 2.7 mmol, 5% yield). Anal. Calcd. for C15H30O6: C,

58.80; H, 9.87; N, 0.0. Found: C, 58.98; H, 9.70; N, 0.00. M.P. 60°C. 1H NMR

13 (CDCl3): δ 0.90 (t, J=7.6Hz, 12H), 1.43 (m, 4H), 1.80 (m, 4H). C NMR (CDCl3): δ

8.0, 22.3, 111.6. IR (cm-1) 2978, 2947, 2878, 1458, 1342, 1273, 1246, 1023, 1157,

1134, 1064, 1041, 1010, 972, 926, 748, 594, 555.

46 2.3.4 Acid degradation studies

To test the reactivity of TPTP, its reaction under acid catalysed conditions was compared with that of TATP. Aqueous HCl degradations were conducted in methanol.

A 30-40 mg sample of TPTP was dissolved in 10 mL of methanol in a volumetric flask and a 100 µL aliquot of 3 M HCl added. A 1 mL sample was taken at 10 mins, and quenched over solid Na2CO3 or CaCO3. This sample was filtered then analysed by

GCMS. Further samples were taken at 40 and 70 mins or as dictated by reaction rate.

2.3.5 Thermal degradation studies

Thermal degradation assays were conducted using the method reported by Eyler et al.13 Aliquots of 1 mL 20 mM solution of TPTP in toluene in vacuum sealed ampoules were heated in a paraffin wax bath at 173°C (±1°C) for 16 h. The solutions were diluted to 10 mL with acetonitrile, then analysed by GCMS.

Sublimation was assessed by taking a sample of ~300 mg of TPTP in a 20 mL

Wheaton vial and drying in vacuo overnight, weighing, then leaving open to the air for

1 month at ~18°C. The sample was weighed at weekly intervals.

2.3.6 Explosive Detection Dog field trial

Explosive Detection Dog (EDD) field trials were carried out at the EDD Cell,

School of Military Engineering, Moorebank, New South Wales. Three trained

Australian Army EDDs and their handlers were recruited for this trial. Two of the dogs were newly trained to detect conventional military explosives and one more experienced dog had been briefly trained to indicate the presence of TATP a number of years ago.

None of the dogs were considered ‘current’ in being able to detect peroxide explosives.

The tests were conducted in abandoned buildings in large rooms with remnant furniture.

47 The University of New South Wales Ethics Secretariat approved the use of dogs for the purpose of this trial, approval ID 10/146.

Eppendorf microcentrifuge tubes with clip-seal tops were oven-dried for 2 days at

70°C to minimise any trace volatiles. TPTP was dried in vacuo at room temperature overnight to remove other volatiles. 5 mg of TATP or 25 mg of TPTP was weighed into each tube and sealed. The larger TPTP sample was used to compensate somewhat toward its lower vapour pressure, but this was not calculated quantitatively. Samples were stored over dry ice until required. During all serials conducted, empty vials which had undergone the same preparation and storage were used as controls.

The dogs were ‘imprinted’ on TPTP by cueing them on to a visible sample (in metal cages) with their reward toy on top of the sample. After about three such serials, the reward was only given once the dog ‘indicated’ (sat) near the sample. Once indication was given without verbal or other prompting, the training was reinforced with hidden samples in single-blind tests, with reward only provided on correct indication of

TPTP. The hidden samples did not utilise the metal cages, with the open vials placed in inaccessible locations such as behind books or furniture, eliminating any possible visual stimulus or extraneous odours. The hidden samples were exposed once to each dog

(with the order of dogs being different each time) before being moved to a fresh room.

Experiments to assess the ability of the dogs to distinguish between TATP and

TPTP were carried out by using hidden samples of TATP and observing the EDD’s reaction. Again, the samples were moved after each dog had a single attempt to locate it.

The handler directed the dog around the room, but did not attempt to cue the dog onto the hidden target if it was missed. The final serial was conducted with five individual

TATP samples (25 mg total) in a single hide (equivalent to the TPTP sample mass).

48 2.4 Results

2.4.1 Pentanone peroxides and their characterisation

The synthesis of the pentanone peroxides has been described previously.2, 7 Milas provided the most thorough description of the synthesis, and utilised fractional crystallisation combined with paper chromatography, IR spectroscopy, molecular weight measurements and elemental analysis to separate and identify the peroxides shown in Figure 2.2 from the crude mixture.2

Figure 2.2 - Organic peroxides derived from acid-catalysed reaction of 3-pentanone with hydrogen peroxide. Adapted from the work of Milas,2 the labels in italics are those used in the referenced work. PP1 - 3,3-dihydroperoxypentane; PP2 – 3-hydroxy-3'-hydroperoxy-3,3'-dipentyl peroxide; PP3 (DPPDHP) - 3,3'-dihydroperoxy-3,3'-dipentyl peroxide; PP4 - 1,1,4,4,7,7-hexaethyl-1,4- diperoxy-l,7-dihydroperoxide; PP5 - 1,1,4,4,7,7,10,10-octaethyl-1,4,7-triperoxy-1,10 dihydroperoxide; PP6 (TPTP) - 1,1,4,4,7,7-hexaethyl-1,4,7-cyclononatriperoxane.

In order to provide useful quantities of pure material for our studies, purification was performed by silica gel column chromatography. TPTP was successfully separated from the crude mixture. DPDP and DPPDHP were also isolated by this means, although

DPDP could not be completely freed of TPTP, and would not crystallise in our hands by the method described by Milas.2 The 1H NMR spectra of TPTP and DPDP are given in

Figure 2.3 and compared against those of the acetone peroxides TATP and DADP. The

49 1H NMR spectrum of DPPDHP is given in Figure 2.4. NMR experiments for TATP and

DADP were conducted on samples provided by DSTO. The trace acetone in the TATP samples is attributed to degradation caused by trace HCl present in the dichloromethane solution in which the sample was provided.

1 Figure 2.3 - H NMR spectra of acetone and pentanone cyclic peroxides in CDCl3. Structures of monomer units and peak assignments for TPTP, DPDP, TATP and DADP are shown. ▲ = acetone, ●= water in chloroform, ■ = trace TPTP.

1 Figure 2.4 - H NMR of DPPDHP in CDCl3. Structures and peak assignments shown, alkyl arms are equivalent.

The overall product distribution varied greatly depending on temperature, reagent concentration and duration of synthesis; however this was not studied in detail.

50 Characterisation by 1H NMR was consistent with the literature for DPDP,5 and the spectra for TPTP and DPPDHP are reported for the first time. As the 1H NMR spectrum of TPTP has not been reported previously, an independent means of confirming the proton assignment was needed. Being related cyclic oligomers, TPTP and DPDP cannot be conclusively distinguished from one another by their 1H NMR spectra alone, as the integral ratios of the dimer and trimer are identical. Elemental analysis is similarly unhelpful as the elemental distributions are also identical. Mass spectroscopy was not a solution as it rarely provides molecular ions for cyclic peroxides using the ionisation methods available to us (EI), and the fragments seen are usually common to both oligomers. Assignment based on coupling data from 1D or 2D 1H NMR techniques requires knowledge of the conformation of the ring structure via an independent means, so an attempt was made to find such a link in the literature. The crystal structure of

TPTP has been reported,1 but the same paper only provided a melting point as further characterisation. Studies of DPDP report 1H NMR data,5, 14 but only exaltone molecular weight measurements were used to confirm the assignment. In neither of these cases was the same sample subjected to both NMR and a conclusive structural method such as x-ray crystallography, making any confirmation of the NMR assessment information secondary in nature,* rather than direct.

To confirm that the spectra were correctly assigned, an inversion recovery experiment was conducted to investigate the T1 values of the oligomers. T1, or spin- lattice relaxation time, is a term used to describe the time taken for nuclear spins to recover their thermal equilibrium distributions after being aligned by an external magnetic field. There is a well-established relationship between molecule size,

*i.e. relying on methods like melting point which do not in themselves provide any structural information, and can be affected by solvent crystallisation. 51 tumbling rate and T1 (the spin-lattice relaxation time), where smaller molecules exhibit longer T1 relaxation under identical conditions due to faster tumbling rates.15 In our experiment, a mixture of DPDP and TPTP was used to ensure completely identical conditions, eliminating a possible source of error. As shown in Figure 2.5, it was found that the signals assigned as DPDP had an average T1 of ~1.6 s, whilst those assigned as

TPTP were significantly shorter at ~0.9 s. The shorter T1 for TPTP is consistent with its larger molecular size, providing validation that the assignments of TPTP and DPDP are correct.

Figure 2.5 - T1 Relaxation times for TPTP and DPDP. T1 relaxation times shown in seconds.

The splitting exhibited in the methylene signals of TPTP is due to geminal splitting between the methylene protons, rather than an axial/equatorial effect. This can be clearly seen in comparing the COSY of a mixture of the two molecules (Figure 2.6).

The strong cross-coupling between the methylene signals at 1.43 ppm and 1.80 ppm

(Figure 2.6, solid red lines) confirm a geminal effect. These cross-peaks do not appear for the methylenes of DPDP (Figure 2.6, dashed blue lines), indicating that the methylene splitting in this case is from axial/equatorial effects in the same way as observed for DADP, and reported by Peña.16 This suggests that the trimeric peroxide rings of the acetone and pentanone peroxides share similar solution structures, despite the bulkier ethyl groups of the pentanone family.

52 Figure 2.6 – COSY spectrum of TPTP/DPDP Mixture. The solid red lines highlight the geminal coupling between the TPTP methylenes, not seen in the (dashed blue) DPDP interactions.

The broad methylene peaks in the DPDP spectrum resolved into quartets at temperatures below 15°C, as has been previously described.5 TPTP was further characterised by elemental analysis, and its IR spectrum matched well with published results.2 Comparison of the IR spectrum of TPTP and DPDP (Figure 2.7) reveals a general blue shift as the ring size increases, which mirrors the reported theoretical and experimental behaviour of the acetone homologues.17 The blue shift is particularly apparent in the ~500-800 cm-1 region which contained the peaks assigned as ring deformation in the previous reference.

53 Figure 2.7 - The IR spectra of TPTP and DPDP. TPTP spectrum obtained as a pressed KBr disc. DPDP spectrum obtained using a drop of DPDP between two KBR windows.

2.4.2 Relative Safety

Having successfully purified and characterised our target analogue, sensitiveness testing was carried out to identify if TPTP was less susceptible to initiating stimuli than TATP.

A combination of industry standard and custom-designed equipment was used to test friction, impact electrostatic and thermal stability. The results are summarised in Table

2.1.

54 Table 2.1 - Sensitiveness Data for TPTP and TATP.

Initiating Rotter Impact Ignition BAM Friction Electrostatic Stimulus (Figure of Temperature (N of pressure) (J) (unit) Insensitiveness) (°C) Initiation at 4.5 153 TPTP <10* <5* but not 0.45 violent reaction Sublimes before TATP <10* <5* 0.023* reaction Figures notated with * are below sensitivity threshold of instrument, and may be significantly lower.

It is important to note that some of the measurements reported here are below the sensitivity limitations of the equipment used, so it is not possible to determine if TPTP is in fact more or less friction or impact sensitive than TATP. Nevertheless, TPTP is a very sensitive primary explosive in terms of friction and impact sensitiveness, and thus all handling must still be conducted with extreme care. This is illustrated by the ‘violent reaction’ of 200 mg of TPTP in the ignition temperature experiment which resulted in the thick-walled glass test tube being shattered to a fine dust. The friction and heat risks can be controlled by synthesising only small quantities, removing heat sources, use of

Teflon coated joints and spatulas, and storing samples as solutions till required.

It is favourable that electrostatic sensitiveness of TPTP is significantly less than that of TATP, as this is a difficult stimulus to eliminate during handling without the use of specialised equipment, facilities and clothing. Utilising anti-static or conductive footware alone is of little use if the flooring material itself is not conductive. It has been estimated that the human body can discharge up to ~0.02 J of electrostatic energy during normal activity,18-19 which the results in Table 2.1 suggest is enough to initiate

TATP. In contrast, no initiation was observed for TPTP even at 0.45 J, a full order of magnitude higher than the energy able to be accumulated on the body. Whilst this does not discount other sources of electrostatic buildup (on equipment etc), it at least

55 mitigates one hazard of handling the explosive for essential laboratory tasks such as weighing and purification processes.

Another significant factor is that TPTP exhibits extremely slow sublimation at room temperature. Less than 1 mg weight loss was noted in a sample (previously dried in vacuo) that was left open to the atmosphere at ambient temperature for 30 days.

Whilst such an assay is not strictly quantitative,* it can confidently be said that < 1% by weight was lost over the 30 day period. As a result, TPTP was routinely dried in vacuo

(~1 mbar) at room temperature for 12 h with no sign of peroxide accumulation in the pre-pump cold trap, indicating negligible sublimation under these conditions.

Throughout the three years that this compound was studied, no evidence of the hazardous re-deposition which characterises TATP was noted. TPTP samples stored at ambient conditions remained as loose crystalline powders. No significant degradation impurity was noted in these older samples when checked by NMR.

2.4.3 Acidic Degradation

As the broader aim of this study is focused on an evaluation of methods of safely degrading cyclic peroxides, it was important to confirm the suitability of TPTP as a model compound in this context. The degradation of TATP by mineral acid and thermal processes are well-studied mechanisms discussed in sections 1.3.1 and 1.3.2

(respectively) of this thesis, providing suitable benchmarks for this purpose.

The reaction of TPTP with HCl yielded 3-pentanone (3P) as the primary product with ethyl propanoate (EP) as the next most abundant species, particularly in the early stages of the reaction. Figure 2.8 shows the relative abundance of reaction species over time using GCMS to track progress.

* Loss of trace co-crystallised solvent, or gain of adsorbed moisture from the atmosphere may change the mass of the sample, rather than sublimation alone. 56 Figure 2.8 - Degradation of TPTP by HCl tracked by quantitative GCMS. Proportion of key reaction products presented as mole fraction per mole of TPTP being degraded.

The predominance of the ketone 3P as the primary product correlates well with the acidic degradation of TATP. Furuya et al. detected 3 equivalents of acetone for each mole of TATP when degraded by acid.20 Furuya’s study did not, however, show any existence of ester products. Other work published by Armitt et al. also did not report methyl acetate in the acidic degradation of TATP (See Chapter 1.3.1).8 The generation of ester products in the early stages of TPTP’s acidic degradation may be explained by a side-reaction in which ketones are subjected to Baeyer-Villiger (BV) oxidations. This process is distinct from those previously described in acid catalysed degradation of any cyclic peroxide molecule.8-10, 20-21

As discussed in Section 1.5 of this thesis, a conventional BV oxidation involves a ketone reacting with a hydroperoxide, which inserts an oxygen to yield an ester

(Scheme 2.A(A)). The BV oxidation of 3P cannot be effected by HOOH even under acid catalysis, as secondary ketones usually require a extremely strong oxidant such as trifluoroperoxyacetic acid to react by this mechanism.22 In the degradation of cyclic

57 peroxides, intermediate species such as gem-peroxyhydroperoxides or gem- dihydroperoxides (which are likely breakdown products of cyclic peroxides8-9, 21) could act in place of trifluoroperoxyacetic acid to oxidise 3P to EP (Scheme 2.A(B)). This hypothesis is supported by the observation that the rate of ester formation is highest early in the reaction, when such hydroperoxides would be at their highest concentration relative to ketones.

Scheme 2.A - Baeyer-Villiger reactions schemes. (A) The general scheme of a Baeyer-Villiger rearrangement reaction where ROOH is a strong organic hydroperoxide. (B) The side-reaction where 3P is oxidised to EP by gem-peroxides formed in the acidic degradation of TPTP.

EP formation was reduced or eliminated when a substrate was added which was more readily subjected to BV rearrangements as an ‘oxidation trap’ (Table 2.2). While their concentrations were not determined quantitatively, ester products of 3-methyl-2- butanone and acetophenone were observed in the GC chromatograms of these reaction mixtures. This is a strong indication that ester formation occurs via an intermolecular mechanism. If EP was being formed directly from the stepwise acid-catalysed breakdown of the open-chain hydroperoxides, these added substrates would not be expected to affect the proportion of ester produced with such a clear relationship to cationic stability.

58 Table 2.2 - Ester formation in TPTP acidic degradation in the presence of substrate ketones. Substrate ketone present during TPTP acidic degradation (% conc)*

none added 3-methyl-2-butanone acetophenone

3-pentanone >86 >94% 100%

Ethyl propanoate <13 <6% none detected

*Figures are representative across a number of repetitions, percentages refer to total 3- pentanone and ethyl propanoate concentration at 1hr reaction time. Substrate added in molar equivalence to peroxide.

In BV rearrangements migratory aptitude is ordered as H>C(R3)>Ph>CH(R2)>

22 CH2R>CH3, in keeping with cationic stability of the migratory group. Methyl acetate is not a notable reaction product in the acidic breakdown of TATP8, 20-21 which is consistent with the known poor migratory properties of methyl groups.22 This argument is further supported by the results of TPTP’s acidic degradation in the presence of 3- methyl-2-butanone, which yielded oxygen insertion on the secondary (iso-propyl) side of the ketone rather than at the methyl. Hence, it is apparent that the predominant breakdown mechanism for TPTP mirrors that proposed for TATP by Armitt et al.,8 while a side reaction of the major product 3P with intermediate hydroperoxides leads to the formation of EP. Our investigations show that whilst there are some differences in reaction products between TPTP and TATP, the fundamental mechanism in acidic conditions appears to be the same, giving the ketone product in both cases. When coupled with slightly safer handling, this makes TPTP a suitable analogue for reactions that mimic the ionic, stepwise, acidic degradation mechanism.

The identification of BV mechanisms operating in the acidic degradation of cyclic peroxides might also provide a useful insight into the fate of trace impurities in the precursor ketone used for the synthesis of homemade cyclic peroxides. The hydroperoxides formed in the degradation of cyclic peroxides are also present in the

59 synthesis. Thus, BV chemistry should also apply for the synthesis of peroxides.

Commercial acetone has been found to usually contain traces of ketones such diacetone alcohol23 which is just as susceptible to BV rearrangements as the oxygen trap used in this study, 3-methyl-2-butanone. Hence the presence of trace esters in crude cyclic peroxide mixtures could be linked to characteristic ketones during synthesis, potentially providing an indicator of the source of precursor materials.

2.4.4 Thermal Degradation

Thermal degradation of TPTP was also investigated and compared against previously published studies concerning TATP. It has been shown that the toluene solvent acted as a radical trap when TATP was thermally degraded via a radical mechanism,13 leading to the coupling of two toluene molecules to form bibenzyl.

Bibenzyl was also present after the degradation of TPTP under identical conditions

(Table 2.3), indicating a similar radical mechanism is responsible for the thermal degradation of TPTP.

Table 2.3 - Thermolysis products of TPTP and TATP. Product TPTP TATP bibenzyl 1.21 1.54 13 ketone (3-pentanone / acetone) 0.14 0.8513 ester (ethyl propanoate / methyl acetate) 0.42 ~0.26* ethyl benzene ~0.2* 0.072 propyl benzene / 1-ethyl-2-methylbenzene ~0.3* NR† Moles of product per mole of cyclic peroxide in toluene solution at 173°C. * - Determined qualitatively by using library matches and comparing peak area with an internal standard. †NR – Not reported

Interestingly, thermal degradation showed a marked decrease in the concentration of ketone produced by TPTP compared to the acetone produced by TATP in the referenced work. This was balanced in part by the production of the ester EP. Previous thermal degradation studies on TATP found methyl acetate present as a degradation

60 product in ratios of ~20% of the final acetone concentrations.6 The larger proportion of ester product present in the degradation of TPTP indicates that the intramolecular rearrangements may be more favourable than for TATP. This mechanism is consistent with the greater radical stability associated with the α-methylenes in TPTP, whereas

TATP has methyl groups making such rearrangements less favoured. Thus in the case of thermolysis, the variation in the proportions of the different products appears to reflect the stability of intermediates in the radical breakdown mechanism, while the fundamental mechanism is the same. This conclusion is supported by the comparable concentrations of bibenzyl observed for TPTP and TATP, indicating that the reaction initially proceeds by a broadly similar radical mechanism in both cases, such as that proposed by Eyler et al.13

2.4.5 Scent characteristics

Noting the potential for a small but significant improvement in safety, the utility of TPTP as a substitute for TATP in EDD scent training was investigated. EDDs have proven very receptive to imprinting on the scent of TATP,24 however difficulties in handling, storing and disposing of TATP due to its inherent volatility and sensitiveness make training too risky for most handlers. Attempts to embed TATP in safe matrices to reduce its sensitiveness have not met with great success,25 particularly due to concerns in the EDD handler community relating to altered scent profile. Considering the structural similarity shared by TATP and TPTP, a trial was conducted to investigate if

EDDs could distinguish between the two compounds. If the two compounds could not be distinguished by the dogs, then potentially TPTP could be used as a TATP substitute in EDD training.

Three Australian Army EDDs were recruited to conduct the pilot field trial. Using

Standard Operating Procedures (SOPs), the dogs were trained to indicate the presence 61 of the explosive samples which were contained in Eppendorf sample tubes (Fig 2.9). At first visible samples were used, but within a few serials the samples were hidden under opaque cones and later behind furniture to ensure that the only stimulus for the EDDs was the scent. To ensure that the dogs were not cued by the scent of the sample containers, empty tubes which had undergone analogous preparation and storage were placed separately from the explosive sample as controls in each serial. The dogs showed no interest in these controls.

Figure 2.9 - Scent detection of organic peroxides by Explosive Detection Dogs. An Australian Army EDD indicating the presence of a hidden sample of TPTP during a confirmatory serial. TATP did not elicit any response from EDDs trained to detect TPTP.

It was found that one day of training (6-13 exposures per dog) was sufficient for the EDDs to recognise and reliably indicate the presence of hidden TPTP samples. The following day each dog was presented with three hidden samples in a large (~80m2) room. The order of the dogs that searched the room was varied between each serial, and each serial was in a fresh room. Each dog successfully detected all three hidden TPTP samples with 100% success. After 3 hours (to allow the scent to dissipate), the same test

62 was run utilising samples of TATP. It was found two dogs could not find any scent of interest at all, indicating 0% success. The third dog displayed weak, indecisive indications in two of the three TATP serials, but the handlers were of the opinion that the EDD was attempting to ‘seek reward’ by cueing onto the strong scent of TATP.

They based their conclusions on the fact that the animal would circle the whole room looking for a scent before coming back to the hidden TATP sample, looking back at the handler and sitting near the hidden sample. As such, these indications were judged to be questionable at best, although it is possible that to this particular animal the scent was

‘similar’ to that of TPTP

Whilst the trial was limited in size and scope, it was well controlled for the dogs being cued by patterns, visible cues, human odours and container scents. The strong negative result clearly suggests that the two compounds exhibit a different scent to the canine olfactory senses. Whilst disappointing in terms of our initial aim, the ability to distinguish such closely related compounds underscores the high level of discretion inherent in the EDD’s sense of smell. A more rigorous protocol would be required to definitively confirm this result, including a double blind format to avoid possible cueing by handlers’ body language, controls for possible impurities and a larger panel of animals.

The result also points to a need for canine scent training to incorporate as many threat compounds as possible; using compounds that are merely similar or of a particular ‘class’ of compound is not effective and may result in false negatives on live tasks. In this case, it seems that the EDDs are not responding to a peroxide scent, or even a cyclic peroxide scent. It seems that the receptors or combination of receptors being used in the EDD nose are more subtle in their selectivity than such obvious characteristics. This conclusion is supported in the fragrance and odour literature, in

63 which no clear relationship between odour and structure has reached widespread acceptance.26-27

A conclusion of equal importance is that one can expect that dogs trained on

TATP are unlikely to identify TPTP, which is still a viable (if unlikely, based on precursor availability) explosive material. The result also indicates that commercial pseudo-scents that are based on ‘de-sensitising’ or even removing the energetic components of explosive compositions28 need extensive analysis and field testing to be certain that the original energetic material is not a significant odiferous component in the authentic explosive scent picture. This is plausible in situations where the explosive composition is based on an energetic material that has a relatively low vapour pressure, such as RDX,28 but with TATP’s high volatility making it a major part of the scent

‘bouquet’, modifying its chemistry is likely to lead to major changes in scent.

In this instance, TPTP’s odor was distinguishable from TATP, from a canine perspective. The ethyl arms have ‘desensitised’ TPTP compared to TATP at least in terms of electrostatics, which could theoretically make it a useful pseudo-scent, but in the process its odor characteristics have also changed. Whilst the TPTP/TATP comparison reported here is only one example, it demonstrates that even minor changes to the structure of energetic materials with high vapour pressures to effect desensitisation can change the odour to a point where it is no longer useful as a pseudo- scent. This small trial has yielded a useful contribution to the scientific community’s relatively poor understanding of the canine sense of smell in the context of explosives detection.25

64 2.5 Conclusion

The presented results show that TPTP has potential as an effective analogue for

TATP in degradation experiments. The TPTP analogue, whilst still demonstrating primary explosive sensitiveness, mitigates two of the key hazards involved in handling

TATP in laboratory processes and storage: sublimation (with ensuing hazardous re- deposition), and electrostatic discharge susceptibility. It has proved to be an effective model compound by displaying very similar reactivity to its acetone homologue TATP during our investigations. Indications are that the primary mechanisms of TPTP’s acidic and thermal degradation are broadly the same as those seen for TATP, but consideration needs to be given to secondary rearrangement reactions and products facilitated by

TPTP’s ethyl moieties. TPTP is also a useful comparative tool in evaluating the mechanism at work in TATP’s degradation, and has revealed nuances in the side- reactions of cyclic peroxide degradation, which have potential application as a forensic approach to trace the fate of trace impurities in precursor chemicals. Given these findings, coupled with the small but significant improvement in safety, TPTP may be considered as an alternative to TATP in some experiments by other groups studying this class of compounds, particularly where the primary focus is degradation. Whilst our exploration of TPTP’s suitability as a training aid for EDD detection of TATP did not provide the answer hoped for, it did give an insight into the remarkable ability of EDDs to distinguish between two closely related compounds, and provided a valuable illustration of the limitations of pseudo-scents in high vapour pressure energetic materials.

This is the first time the basic sensitiveness characteristics of TPTP have been reported, and the extreme friction and impact sensitiveness results provide a valuable tempering of previous assertions that TPTP is significantly more ‘stable’ than TATP.1

65 Whilst this may still prove to be the case when tested on more sensitive instrumentation, in a practical sense both peroxides deserve the utmost respect during all handling procedures.

2.6 Acknowledgements

This work was completed with the support of the Australian Defence Force’s

Chief of Defence Force Scholarship. We thank Craig Wall and Mark Fitzgerald of the

Defence Science and Technology Organisation for conducting the sensitiveness testing.

2.7 References

1. Cerna, J.; Bernes, S.; Canizo, A.; Eyler, N., Acta Crystallographica Section C- Crystal Structure Communications 2009, 65, O562-O564. 2. Milas, N. A.; Golubovic, A., J. Am. Chem. Soc. 1959, 81 (13), 3361-3364. 3. Morales, G.; Eyler, G. N.; Cerna, J. R.; Canizo, A. I., Molecules 2000, 5 (3), 549- 550. 4. Posner, G. H.; O'Dowd, H.; Ploypradith, P.; Cumming, J. N.; Xie, S.; Shapiro, T. A., J. Med. Chem. 1998, 41 (12), 2164-7. 5. Zmitek, K.; Stavber, S.; Zupan, M.; Bonnet-Delpon, D.; Charneau, S.; Grellier, P.; Iskra, J., Bioorg. Med. Chem. 2006, 14 (23), 7790-7795. 6. Oxley, J. C.; Smith, J. L.; Chen, H., Propellants Explosives Pyrotechnics 2002, 27 (4), 209-216. 7. Canizo, A. I., Trends in Organic Chemistry 2006, 11, 55-64. 8. Armitt, D.; Zimmermann, P.; Ellis-Steinborner, S., Rapid Commun. Mass Spectrom. 2008, 22 (7), 950-958. 9. Oxley, J. S., J. L.; Brady, J. E.; Steinkamp, L., Propellants Explosives Pyrotechnics 2014, 39 (2), 289-298. 10. Oxley, J. C.; Smith, J. L.; Steinkamp, L.; Zhang, G., Propellants Explosives Pyrotechnics 2013, 38 (6), 841-851. 11. Armarego, W. L. F.; Chai, C. L. L., Purification of laboratory chemicals. 6th ed.; Butterworth-Heinemann: Oxford, 2009; p xvi, 743 p.

66 12. NIST Mass Spectral Library, NIST 05, National Institute of Standards and Technology, Maryland, US.

13. Eyler, G. N.; Mateo, C. M.; Alvarez, E. E.; Canizo, A. I., J. Org. Chem. 2000, 65 (8), 2319-2321. 14. Brune, H. A. W., K.; Hetz, W., Tetrahedron 1971, 27 (15), 3629-44. 15. Sanders, J. K. M.; Mersh, J. D., Prog. Nucl. Magn. Reson. Spectrosc. 1982, 15, 353-400. 16. Peña, A. J.; Pacheco-Londoño, L.; Figueroa, J.; Rivera-Montalvo, L. A.; Román- Velazquez, F. R.; Hernández-Rivera, S. P., Characterization and differentiation of high energy cyclic organic peroxides by GC/FT-IR, GC-MS, FT-IR and Raman Microscopy, in Sensors and Command, Control, Communications, and Intelligence (C31) Technologies for Homeland Security and Homeland Defense IV, Pts 1 and 2, Orlando FL, 2005, Carapezza, E., Ed. Proceedings of SPIE, 5778, pp 347-358. 17. Oxley, J.; Smith, J.; Brady, J.; Dubnikova, F.; Kosloff, R.; Zeiri, L.; Zeiri, Y., Appl. Spectrosc. 2008, 62 (8), 906-915. 18. Wilson, N., Journal of Electrostatics 1977, 4 (1), 67-84. 19. Talawar, M. B.; Agrawal, A. P.; Anniyappan, M.; Wani, D. S.; Bansode, M. K.; Gore, G. M., J. Hazard. Mater. 2006, 137 (2), 1074-1078. 20. Furuya, Y.; Ogata, Y., Bull. Chem. Soc. Jpn. 1963, 36 (4), 419-422. 21. Tsaplev, Y. B., Kinetics and Catalysis 2012, 53 (5), 521-524. 22. March, J., Advanced organic chemistry: reactions, mechanisms, and structure. McGraw-Hill: New York, 1968; p ix, 1098. 23. Wahl, J. H.; Bolz, C. D.; Wahl, K. L., Lc Gc Europe 2010, 23 (4), 18-27. 24. Oxley, J. C.; Smith, J. L.; Moran, J.; Nelson, K.; Utley, W. E., Training dogs to detect triacetone triperoxide (TATP), in Sensors, and Command, Control, Communications, and Intelligence(C31) Technologies for Homeland Security and Homeland Defense III, Pts 1 & 2, Orlando FL, 2004, Carapezza, E., Ed. Proceedings of SPIE, 2004, 5403, pp 349-353. 25. Lorenzo, N.; Wan, T.; Harper, R.; Hsu, Y.-L.; Chow, M.; Rose, S.; Furton, K., Anal. Bioanal. Chem. 2003, 376 (8), 1212-1224. 26. Frater, G.; Bajgrowicz, J. A.; Kraft, P., Tetrahedron 1998, 54 (27), 7633-7703. 27. Turin, L., J. Theor. Biol. 2002, 216 (3), 367-385.

67 28. Adebimpe, D. B.; Zgol, M. A.; Wright, G. R. Method of producing energetically- inert pseudoscents of explosive materials, and compositions thereof. US20070221087 A1, 2010.

68 CHAPTER 3. Redox and electrochemistry of

cyclic peroxides

3.1 Introduction

The discussion in Section 1.3.3 of the introduction shows that certain metal salts such as Cu(II) and Zn(II) are able to degrade TATP at moderate rates.1 At first glance, this seems to be an eminently predictable reaction – that an oxidant (in this case TATP) would be degraded by oxidising a susceptible substrate. Yet whilst there are examples of such (suspected) redox reactions, there are also many observations that indicated

TATP is relatively resistant to chemical reduction. The iodide ion, for example, is often used as a titrant for the determination of hydroperoxides which oxidise the former to iodine.2 Ferrous sulfate is also highly reactive with peroxides of the form ROOR such as diethyl peroxides.3 Accordingly these reagents have already been tested by studies

4 1, 4 seeking a convenient neutralisation of TATP. Both KI and FeSO4 have been found to be very poor at degrading TATP on their own, indicating this class of cyclic gem- peroxides is highly resistant to chemical reduction.

Electrochemistry can be a useful tool to measure the relative susceptibility of compounds to chemical reduction. In the case of TATP, very little information has been published about its electrochemical behaviour. A few works have used electrochemical response as a means for the detection of TATP and other organic peroxides, however all these studies utilised some form of degradation (either acid or photolysis) to yield

ROOH or HOOH or another derivatised analyte.5-10 Only one of these reports

69 commented on the direct reduction potential of TATP, however their stated value of

-1.0 V (vs Ag/Ag+) seems very low compared to the reported reduction potential of

~-2.5V for di-tert-butylperoxide in the same solvent11 (albeit with different supporting electrolytes). It seems possible that the -1.0 V figure is incorrect considering the reported difficulty in chemically reducing this peroxide, and is worthy of re- investigation.

3.2 Research Goals

With the aim of this thesis being to identify new methods of degradation, it was considered important to revisit this aspect of cyclic peroxide chemistry with a short redox study. The aims of this part of the research were to:

1. Evaluate a range of organic and inorganic reducing agents in a simple assay

to extend work done by previous studies, and

2. Conduct an electrochemical investigation of the redox potentials of some

example organic peroxides in an effort to relate this to their susceptibility to

chemical reduction.

3.3 Experimental

CAUTION! The organic peroxides detailed in this study have the characteristics of primary explosives with extreme levels of friction, heat and impact sensitiveness. All procedures must be carried out by properly qualified and equipped personnel taking all relevant precautions.

3.3.1 Materials

TPTP was synthesised and purified as described in section 2.3.4 of this thesis.

TATP used in this work was sourced from Dr David Armitt (DSTO) as a 1 mg/mL methanol solution. HPLC grade solvents were used in the preparation of samples for 70 GCMS analysis; all other solvents were of analytical grade. GCMS standard solutions were of prepared using analytical grade analytes with the exception of ethyl propanoate which was prepared as described in Section 2.3.1.

Reagents for redox assays were lab grade and sourced as small samples from Dr

David Armitt (DSTO). Mo2O5(OH) was synthesised according to the procedure of

Apblett12 and the nanoparticle suspension was drawn from the butanol washings of the solid filtrate. The nanoparticles were extracted by addition of 10 times excess toluene to the butanol suspension followed by centrifugation. The fine blue solid was washed with toluene then dried in vacuo and used for the degradation experiments.

13 Cu(I)Cl.Neocuproine2 was synthesised by the method of Pallenberg. Cu(II).Salen was synthesised by the method of Velusamy.14 Montmorillonite K10 (MMT).Ammonia was synthesised by stirring 1 g of MMT with 20 mL of conc. aqueous ammonia, filtering, washing till the washings were neutral pH, then drying for 4h @ 50°C in vacuo.

3.3.2 Instruments

GCMS analysis of redox assays was carried out under the conditions in Section

2.3.1, except that the column used was a SGE BPX5 column (0.25 µm film, I.D. 0.25 mm, 15 m).

Cyclic voltammetry was conducted using an eDAQ EA161 potentiostat operated via an eDAQ ED401 e-corder. A 1 mm glassy carbon in PEEK working electrode was utilised with a platinum wire counter electrode and Ag/AgCl reference electrode inserted into a 3 mL reaction vial. A sweep rate of 250 mV/s was used for cyclic peroxides, with 100 mV/s used for organic hydroperoxide samples.

71 3.3.3 Redox degradation studies

A variety of potential reducing agents were tested for reactivity against TPTP and

TATP. 3 mL of a 1 mg/mL solution of the substrate peroxide in MeOH was placed in a

20 mL Wheaton vial with a magnetic stir bar and screw cap. The reducing agent was added (3-5 equivalents) and the timing started. 1 mL samples were taken at 15 mins and

2 hours and placed in a fresh vial together with 1 mL hexane and 2 mL H2O, stirred, and then allowed to separate. The aqueous layer was discarded and the hexane layer dried over Na2SO4 before analysis by GCMS. To confirm that this extraction method was quantitative, control extractions with known concentrations of peroxide were conducted.

3.3.4 Electrochemistry

0.2-0.3 mg of peroxide was dissolved in 3 mL of 0.1 M TBABF4 MeCN solution.

Samples were purged of O2 by bubbling with argon gas for 1 min prior to conducting experiments. Ferrocene was used to check the reference electrode at the beginning and end of each set of experiments; this oxidation occurs at +0.55 V vs. Ag/AgCl in MeCN.

Analyte concentrations varied between 1-5 mM.

TATP and DADP were stored as MeOH solutions. For electrochemical analysis, the calculated volume of this solution was added to an empty reaction vial, and the solvent removed by a stream of N2. Weighing the resulting product was inaccurate due to possible residual solvent/water as well as uncertainties due to the tendency of DADP and TATP to sublime at RT. Thus concentrations were only broadly controlled for these two analytes.

72 3.4 Results and Discussion

3.4.1 Redox reactions

To extend the work of previous studies that indicate TATP’s resistance to chemical reduction, a broad sample of multiple types of reducing agents was assessed against the model peroxide TPTP. Table 3.1 shows the results of these assays. The assays were conducted at room temperature as the aim was to identify reagents that did not require additional heat to react with TPTP. Samples were taken at 15 mins and 8 hours and the organic extract of these samples was analysed by quantitative GC/MS. A negative result was recorded for samples where there was no significant change in

[TPTP] between the two analyses and the [TPTP] in a control solution.

Despite using large excesses of the reducing agents, none of the listed compounds reacted to a significant degree with TPTP at room temperature. Such a broad resistance to chemical reduction under the described conditions is remarkable. It is possible that some reactions might have occurred at higher temperatures, however the use of heat is not acceptable for the intended field application, and as such elevated temperatures were not investigated.

73 Table 3.1 - Redox reaction results.

Reaction Compound Degradation 1 Molybdenum hydrogen bronze N

2 Cu(I)Cl.Neocuproine2 N 3 Cu(II)Salen N

4 Sc2(SO4)3 N 5 FeCl2 N 6 CuCl2 N 7 MnO2 N 8 MnCl2 N 9 Zn (metal powder) N

10 Co(NO3)2 N 11 Li2B4O7 N 12 Ethanox 330 N 13 Antioxidant 425 N 14 Montmorillonite K10 N 15 Montmorillonite K10.Ammonia N

16 TiO2 N 17 ZnCl2 N 18 ZnS N

19 K3Fe(CN)6 N 20 Na2SO3 N 21 Na2S2O3 N 22 p-aminophenol N 23 diethylenetriamine N 24 triphenylphosphine N 25 hydroquinone N 26 p-tolylhydrazine hydrochloride N 27 hydroxylamine hydrochloride] N 28 thiourea N 29 ferrocene N

30 B(OH)3 N 31 p-quinone dioxime N 32 semicarbazide hydrochloride N 33 trimethyl phosphite N

Interestingly, a number of compounds appear in this list that have been reported as active degradation agents for TATP (albeit under different conditions), yet did not show noticeable activity in this assay. An example is molybdenum hydrogen bronze

(Mo2O5(OH)), which was included in this study due to the reported ability to rapidly

74 reduce TATP.12 It is important to note that the nanoparticles used in this study had been precipitated from the butanol, washed and dried prior to use in the degradation study.

As discussed earlier (section 1.3.4), it is deduced from the reported experimental procedure that the referenced study utilised acidified butanol solutions of Mo2O5(OH) nanoparticles for peroxide neutralisation, although this is not explicitly stated in the reference. The inability of Mo2O5(OH) to degrade TPTP in the absence of its butanol mother liquor is consistent with our deduction. To confirm the presence of acid in the nanoparticulate butanol mother liquor synthesised via the literature procedure,12 3 mL of the Mo2O5(OH)-butanol suspension was extracted with 50 mL of water. The aqueous sample was found to have a pH of ~3.

To further test the relative stability of cyclic peroxides, a simple rate of reaction assay was conducted. 0.7 mL each of 30% aqueous HOOH, 70% aqueous di-tert- butylperoxide and dilute (30 mM) methanol solutions of DPPDHP and TPTP were reacted with 0.3 mL of Mo2O5(OH)-butanol suspension. The characteristic blue colour of the Mo2O5(OH) was visually monitored over time as illustrated in Figure 3.1.

Hydrogen peroxide reacted much faster in eliminating the blue colour than tert- butylperoxide despite being half the concentration of the latter, indicating that HOOH was the most reactive form of peroxide tested. Interestingly DPPDHP reacted more rapidly than tert-butyl peroxide despite being orders of magnitude less concentrated, suggesting it is also very reactive. In contrast, TPTP was very slow to bleach the

Mo2O5(OH) colour, although it appears some reaction did occur over 8 h. Whilst not conclusive, these observations suggest that in order to achieve the reported rapid degradation, TATP first reacts with acid to form more reactive ROOH and HOOH species which then further react with Mo2O5(OH). The observation that TPTP reacts very slowly with Mo2O5(OH)-butanol is consistent with this hypothesis.

75 Figure 3.1 - Reaction of Mo2O5(OH) with various peroxides.

ZnCl2 has also been reported to be an active degrading agent for TATP, however the reported assay was again conducted in acidified solution.1 All other identified examples of redox reactions also involved acidic conditions (either stated or deduced) to initiate the degradation of TATP.4, 15-17 It seems our observations are in keeping with the broader evidence in the literature that the cyclic gem-peroxide molecule is quite resistant to direct chemical reduction without first being degraded to an ROOH or

HOOH by acid.

3.4.2 Electrochemistry

In order to directly compare the reduction potential of cyclic peroxides and the observed resistance to reduction, a study of the electrochemistry of cyclic peroxides was conducted. Dilute acetonitrile solutions of cyclic peroxides were subjected to cyclic voltammetry and compared with tert-butylhydroperoxide. DPPDHP was also included

76 to indicate the relative reduction potential for a compound very similar to the product of Chart Title acid-catalysed ring opening of TATP. The results are shown in Figure 3.2.

DADP TPTP TATP

-3 -2.5 -2 -1.5 -1 -0.5 0

Potential (V vs Ag/AgCl) Figure 3.2 - Cyclic voltammograms of DADP, TPTP and TATP in MeCN.

Figure 3.3 - Cyclic voltammagrams of tert-butyl hydroperoxide, TPTP and DPPDHP in MeCN.

Tert-butyl hydroperoxide’s observed reduction potential of -2.16 V vs Ag/AgCl matches the literature value of -2.10 V to -2.20 V reported by Vasudevan,11 validating our method. The very negative reduction potentials for the cyclic peroxides, ranging from –2 V to -2.6 V, confirm these compounds are indeed quite difficult to reduce electrochemically, consistent with their resistance to chemical reduction. By contrast, the reduction of the oxidant perchlorate to chlorate under aqueous conditions occurs at

+1.23 V,18 with the positive value indicating a spontaneous reaction.

77 The reduction potentials for cyclic peroxides observed in this study seem more consistent with other organic peroxides than the weak signal at -1.0 V reported by

Parajuli as the reduction of TATP.10 The similar potentials across the related cyclic peroxides measured in the present study also add reliability to the new values. It is possible that Parajuli’s measurement (conducted in aqueous MeCN), is the correct potential in protic conditions,* however considering that the reported solvent window only extended to -1.75 V and the weak reduction signal, it seems more likely that TATP reduction did not occur under Parajuli’s conditions.

The relative susceptibilities of cyclic peroxides and hydroperoxides to electroreduction is also reflected in these results, with DPPDHP being reduced at approximately half the potential of the cyclic peroxides. DPPDHP was also easier to reduce than tert-butylhydroperoxide. Both of these observations are consistent with the relative rates of reaction (chemical reduction) seen in the Mo2O5(OH) redox assay. The relative amplitude of the reduction waves were not meaningful due to analyte concentration not being precisely known (see experimental for explanation).

The fact that TPTP displayed the most negative reduction potential is interesting, and is likely to be a reflection of the fact that the ethyl arms of the molecule are able to donate more electron density to the electronegative peroxide ring which might add stability to the weak peroxide bonds. Accordingly, the molecule is somewhat less susceptible to electroreduction. The slightly lower potentials of TATP and DADP may be due to the fact that their samples were stored in MeOH solution and were deposited in the analysis vessel by blowing off the solvent under a stream of N2 gas. It is likely that this process left traces of water or methanol in the sample which would have

* The reduction potentials of organic molecules can be highly dependent on the presence of a protic solvent. 78 resulted in a slightly more protic environment, and may have facilitated earlier reduction than the drier TPTP sample. This appears to be the case especially for DADP where the solvent front was shifted positively by ~0.1 V compared to the TATP and

TPTP experiments.

As the reduction potential is a means for establishing the susceptibility of a substance to chemical reduction, it appears that at least by this measure TPTP may be more difficult to reduce than TATP. From this perspective, our methodology of screening against the pentanone peroxide means that any reducing displaying activity should also be active against TATP. It is possible that some compounds may have displayed slight activity against TATP but not with TPTP. This high benchmark suits our purpose in this instance as our goal is to seek fast reactions. The lack of any significant chemical reduction of TPTP in our assay indicates none of the reagents tested are suited to this purpose.

3.5 Conclusion

Whilst many studies have opted for a TATP degradation scheme that utilises chemical reduction, our results combined with a critical analysis of the literature indicate that it is unlikely that the cyclic gem-peroxide family of molecules is directly susceptible to such chemistry. A wide range of organic and inorganic reducing agents were assayed in non-aqueous conditions that were less prone to generating free acid, and in this case no significant peroxide degradation could be detected. Some of the compounds which had shown activity in other studies were found inactive under the described experimental conditions. This may indicate that in previous studies the reduction was in fact occurring with ring-opened hydroperoxides as generated by acid catalysed ring-opening. These hydroperoxides appear to be much more reactive, as

79 demonstrated by our tests on Mo2O5(OH). Whilst this ‘acid-redox’ scheme still achieves the aim of degradation, it is important to explicitly note that acid-catalysed ring-opening appears to be a vital first step in known redox degradation mechanisms. With this knowledge, degradation schemes can be designed to utilise this approach, rather than expecting a direct reaction between TATP and the reducing agent.

This work has reported the reduction potentials for some key organic peroxides, with the cyclic and ether peroxides exhibiting much more negative potentials than hydroperoxides. This is the first study to report the electrochemical behaviour of DADP and the pentanone peroxides. Although TATP’s reduction potential has been reported previously,10 the consistency of the presented values across multiple cyclic peroxides and with other electrochemical literature suggests the previous value may have been erroneous. The large reduction potentials further demonstrate the inherent resistance of cyclic gem-peroxides to direct reduction. As such, chemical reduction alone is unlikely to be the answer in degrading cyclic peroxide explosives. Acid-catalysed ring opening or an equivalent method of ring opening to yield reactive hydroperoxide species or hydrogen peroxide seems a useful if not necessary step to achieve an effective redox degradation method.

A degradation scheme based on an acid-redox scheme will only ever achieve stoichiometric degradation of peroxides. As an ideal degradation system would be best achieved by finding a catalytic means of degrading these organic peroxides, it is important to identify new reactions of organic peroxides and understanding the mechanisms by which they work. The next two chapters present the identification of new degradation reaction which has potential catalytic application.

80 3.6 References

1. Oxley, J. C.; Smith, J. L.; Huang, J. R.; Luo, W., Journal of Forensic Sciences 2009, 54 (5), 1029-1033. 2. Vogel, A. I.; Jeffery, G. H., Vogel's textbook of quantitative chemical analysis. 5th ed.; Longman Scientific & Technical: Harlow, 1989; p xxix, 877 p. 3. Swern, D., Organic Peroxides. Wiley-Interscience: New York, 1970. 4. Bellamy, A. J., Journal of Forensic Sciences 1999, 44 (3), 603-608. 5. Lu, D. L.; Cagan, A.; Munoz, R. A. A.; Tangkuaram, T.; Wang, J., Analyst 2006, 131 (12), 1279-1281. 6. Laine, D. F.; Roske, C. W.; Cheng, I. F., Anal. Chim. Acta 2008, 608 (1), 56-60. 7. Laine, D. F.; Cheng, I. F., Microchem. J. 2009, 91 (1), 125-128. 8. Schulte-Ladbeck, R.; Karst, U., Chromatographia 2003, 57, S61-S65. 9. Xie, Y. Q.; Cheng, I. F., Microchem. J. 2010, 94 (2), 166-170. 10. Parajuli, S.; Miao, W. J., Anal. Chem. 2013, 85 (16), 8008-8015. 11. Vasudevan, D., Bull. Electrochem. 2000, 16 (6), 277-279. 12. Apblett, A. W.; Kiran, B. P.; Malka, S.; Materer, N. F.; Piquette, A., Ceram. Trans. 2006, 172, 29-35. 13. Pallenberg, A. J.; Koenig, K. S.; Barnhart, D. M., Inorg. Chem. 1995, 34 (11), 2833-2840. 14. Velusamy, S.; Punniyamurthy, T., Eur. J. Org. Chem. 2003, (20), 3913-3915. 15. Tsaplev, Y. B., Kinetics and Catalysis 2012, 53 (5), 521-524. 16. Furuya, Y.; Ogata, Y., Bull. Chem. Soc. Jpn. 1963, 36 (4), 419-422. 17. Matyas, R., Chemical decomposition of triacetone triperoxide and hexamethylenetriperoxidediamide, in New Trends is Research of Energetic Materials, Czech Republic, 2003, pp 241-247. 18. Aylward, G. H.; Findlay, T. J. V., SI chemical data. 5th ed.; Wiley: Milton, Qld., 2002; p xiv, 202 p.

81 CHAPTER 4. The reaction of metal-based

Lewis acids with TPTP

4.1 Introduction

Since the first investigation of TATP degradation in 1963,1 and significant effort since 1999,2-6 acidic degradation has been the primary route by which safe destruction of TATP has been sought. Much progress has been made in understanding the acid- catalysed reaction, making it a relatively well understood mechanism.4-8 The mechanisms of thermal and photocatalytic degradation have also been studied, but these methods are not considered useful for field or laboratory neutralisation due to a range of safety and practicality limitations. A number of researchers have utilised metal-based compounds to effect neutralisation although little investigation of the mechanism was reported.3, 6, 9-10 The review of these studies in Chapter 1 and the subsequent discussion in Chapter 3 suggest that these methods may also rely on acid catalysis for the initial breakdown of TATP. Thus it appears that the published methods for TATP degradation rely, at least in part, on the acid-catalysed breakdown of the cyclic peroxide structure.

An acid-redox system could become an effective solution for the field neutralisation of Organic Peroxide Explosives (OPEs), however the scheme’s “Achilles heel” is the need for a stoichiometric substrate for reduction. Effective selection of substrate could minimise the resulting reagent bulk, but each molecule of TATP would require at least three molar equivalents of substrate (reducing agent) to ensure complete eradication of the peroxide threat. Whilst this approach might be suitable for small quantities of OPE, large volumes of reagent would be impractical for larger samples of

82 OPE, especially when packaged in a container of limited volume. An ideal means of circumventing this issue is to devise a catalytic means of degradation that eliminates the peroxide groups without requiring a stoichiometric amount of substrate. A good understanding of a reaction mechanism is necessary to ‘design’ a catalytic system, making the identification and characterisation of new OPE degradation reactions vital to achieving the desired endstate of catalytic degradation.

A promising alternative technique for TATP is suggested by Dubnikova’s in silico studies which suggested that metal ions such as Ti4+, Sb3+ and Sc3+ have a strong affinity to bind with TATP in the gas phase, and predicted these ions woud trigger the complete destruction of the molecule.11-12 At this stage, no published report appears to have followed up this theoretical work with experimental results, so it is unclear how this proposed reactivity translates from model to reality. In the search for novel methods of neutralising TATP, it is important to investigate the activity of these metal species and assess their effectiveness as degradation agents. It is possible that the modeled reactivity of the free metal ions may translate to reactions with compounds containing the relevant ion in solution.

The study described in this chapter found that both titanium tetrachloride and antimony trichloride react with cyclic peroxides via a novel rearrangement mechanism, forming ester and ketone products. These investigations have largely utilised the model compound tripentanone triperoxide (TPTP) which has been previously demonstrated as an effective model for TATP in degradation reactions (Chapter 2).8 This approach was taken due TPTP’s to safer handling characteristics and ready access to data on its reaction behaviour.

83 4.2 Research Goals

The aim of this section of the research was to understand if strong Lewis acids might be able to mediate a different mechanism of degradation to that of Brønsted acids upon reaction with the model peroxide TPTP. To establish this, the following key goals were pursued:

1. To investigate the reaction of TPTP with TiCl4 and SbCl3 under controlled

conditions; and

2. Work toward an understanding of the reaction mechanism.

4.3 Experimental

CAUTION! The organic peroxides detailed in this study have the characteristics of primary explosives with extreme levels of friction, heat and impact sensitiveness. All procedures must be carried out by properly qualified and equipped personnel taking all relevant precautions.

4.3.1 Materials

All chemicals and solvents were of analytical grade unless otherwise noted.

CDCl3 (99.9% D) was dried by distillation from P2O5. TPTP, DPDP and DPPDHP were synthesised and purified as published previously.8 GC standards were made with analytical purity ethyl acetate and 3-pentanone with capillary grade dichloromethane

(DCM), and TPTP was purified by sublimation in vacuo. TiCl4 was double distilled

13 according to literature procedure from copper turnings under dry N2 gas, before being weighed and dissolved in a measured volume of CH2Cl2 (GC grade, freshly distilled from CaH2 under N2). TiCl4 solutions were used and stored under dry N2 atmosphere, and discarded after 2 weeks.

84 4.3.2 Instruments

A Shimadzu QP 2010 Ultra fitted with a SGE SolGel Wax column (1.0 µm film,

I.D. 0.25 mm, 30 m) was used for GCMS analyses. The method utilised an initial column temperature of 40°C for 1 min, 20°C/min ramp to 100°C, held for 3 mins, followed by 30°C/min finishing at 190°C and held for 2 mins. Injector temperature was

130°C, interface 200°C, and linear column flow at 2 mL/min of He gas. Split ratio was set at 40. Concentrations were quantified using external standards.

1H NMR spectra were recorded on a Varian Unityplus-400 spectrometer. All 1H

NMR experiments were conducted at 25°C unless otherwise stated. 1H NMR spectra were referenced to TMS (0 ppm) at 25°C using the residual 1H signal of the respective deuterated solvent. 1D spectra were recorded with between 1 and 16 transients with a 6 second relaxation delay. COSY experiments were conducted with 1 scan per increment,

512 t1 increments and 2 s relaxation delay. Integration data was processed and graphed using Graphpad Prism.

IR Spectra were recorded on a Shimadzu IRPrestige-21 spectrometer. Samples were typically analysed between 350-3500 cm-1 with 16 scans. DCM solutions were analysed in a 0.25 mm pathlength demountable liquid cell with KBr windows which was oven dried and stored over phosphorus pentoxide. The cell, filled with dry DCM, was used for the background measurement. The cell was rinsed with dry DCM and purged with dry argon between samples. The liquid film spectrum of DPDP was collected from a drop of sample between two KBr discs.

UV/Vis spectra were recorded on a Cary 50 spectrometer. Antimony tartrate standard solutions for SbIII calibration curves were analysed within 2 h of preparation.

85 4.3.3 Methods

All weighing and handling was conducted in a dry N2 atmosphere using a glovebox or standard Schlenk techniques, with flame dried glassware (cooled under N2 flow) and oven-dried syringes. Any trace of moisture in the presence of TiCl4 results in immediate cloudiness – any experiments where this was observed during preparation were re-prepared.

4.3.3.1 Preparation and determination of TiCl4 stock solution

The concentration of TiCl4 in DCM solution (estimated initially from weight added of TiCl4 and DCM added and known densities) was confirmed by utilising a modified published method.14-15 0.1 M aqueous EDTA was standardised against 0.01 M aqueous CaCO3 primary standard, and 0.1 M aqueous ZnSO4 was standardised against the EDTA solution. A sample of the TiCl4 solution was added to 5 mL of 3 M

H2SO4(aq) and heated to drive off DCM. After cooling, 3 drops of 30% H2O2 were added to stabilise Ti in solution followed by a 10 mL aliquot of EDTA solution and 3.5 g solid hexamethylenetetraamine as buffer. The mixture was adjusted to pH 5.5 with concentrated aqueous ammonia and diluted to 80 mL with water. 4 drops of xylenol orange indicator were added and the remaining free EDTA was back-titrated with

ZnSO4 solution to a lasting red-orange endpoint. The concentration derived from this titration ([Ti]) was used for all further calculations.

4.3.3.2 TPTP-TiCl4 degradation experiments

TPTP was measured into a dry 5 mL volumetric flask, sealed and diluted with dry

CDCl3 to make a 0.05 M solution. A 1.4 M stock solution of TiCl4 in dry DCM was diluted in dry CDCl3 such that 100 μL would provide the desired mole ratio when added to 500 μL of the analyte stock solution (final TiCl4 solutions ranged from 0.12 M to 0.74

86 M). 500 μL of TPTP solution was placed in a dry 5 mm NMR tube and vacuum degassed. 100 μL TiCl4/CDCl3 was added (forming an immediate yellow/green clear solution at the mixing interface) and the mixture immediately frozen in liquid N2. The tube was flame-sealed under high vacuum and stored frozen till analysis. The sample was thawed and shaken well immediately prior to inserting into the NMR spectrometer for measurement.

Figure 4.1 - Anhydrous NMR sample sealing. The distillation head shown is connected to a N2/vacuum manifold to allow N2 purging and flame sealing (under vacuum) of the frozen sample which is immersed in liquid N2 (dewar visible at the bottom of the image). Reagent solutions are injected through the septum using syringes. The same apparatus was used for non-frozen samples, with the tube capped immediately under the head with strong N2 flow to minimise air ingress to the tube, and shaken immediately by hand.

In other experiments [TiCl4] was kept constant, with the volume used to control reagent ratios. Aliquots were added to 500 μL of TPTP/CDCl3 solution in NMR tubes.

The samples were exempted from the freeze-seal-thaw method and were capped, sealed and shaken within a few seconds of removing from the N2 manifold, and the stopwatch started. These samples were analysed immediately by 1H NMR.

87 The residual solvent signal (previously quantified against a solid aliquot of hexamethylbenzene by 1H NMR) was used as an internal standard for quantifying integration in these NMR reactions. Single transient spectra with minimum 20 s relaxation time were used to ensure accurate integration. Negative integrals were corrected to zero. Integrals that were significantly off the trend for all analytes at a given timestep were excluded. As EP and 3P peaks shifted upfield due to complexation with

Ti, as discussed in Section 4.4.4.1, some error in integration was unavoidable.

Accordingly, the resulting data was corrected for the expected total number of moles of organic product (based on TPTP added) by using the following formula:

where n(t) = the number of moles of a species (TPTP, EP, 3P or 2-chloro-3-pentanone

(2C3P)) at time t. The known number of moles of TPTP added at the start of the reaction provides . TPTP Equivalent (t) represents the number of moles of unreacted and degraded TPTP present as measurable products at time t and is calculated as:

where n(t) = the measured number of moles of a particular reagent (2C3P = 2-chloro-3- pentanone). Using this approach, the overall integral at each time interval is corrected to reflect the overall number of moles of organic product expected from the initial TPTP.

This assumes that TPTP, 3P, EP and 2C3P are the only significant organic products, which is the case in almost all reactions studied. This approach was not applied in reactions where other chlorinated products became significant.

Certain samples were quenched prior to 1H NMR or GCMS analysis. In these cases, the sample was prepared as above only in a GCMS vial with a N2 atmosphere.

88 To quench the reaction, the entire reaction sample was transferred to a small vial containing Na2CO3 and a drop of water, agitated and allowed to settle for ~5 mins. The supernatant was transferred over Na2SO4 to dry for 10 mins, then syringe filtered and analysed by the relevant method.

A sub-stoichiometric TiCl4 reaction was scaled up in order to isolate the precipitate for further analysis. TPTP (135 mg, 0.44 mmol) was weighed into a dry 10 mm NMR tube. 3 mL of dry CDCl3 was added and the contents vacuum degassed.

0.87 M TiCl4 DCM solution (0.25 mL, 0.22 mmol) was added, and the mixture immediately frozen and vacuum sealed. On thawing, the reaction was monitored by 1H

NMR. A pale yellow precipitate formed within an hour and continued to form over the following week. After 9 days, the tube was opened under dry N2, the bulk of the supernatant solution removed by syringe, and the sample dried in vacuo. The resulting dry sample was not weighed as quantitative transfer was not possible. A small sample was ground with KBr in air and immediately analysed by FTIR. For 1H NMR, a drop of

D2O was added to a small sample resulting in an immediate white suspension which was diluted with MeCN-d3. Microanalysis of the precipitate was conducted by the

Australian National University microanalysis laboratory.

For FTIR assays, aliquots of TiCl4/DCM stock solution were added directly to DCM solutions of TPTP, EP and 3P (50 mM) in dry, sealed vials fitted with septa and

N2 supply for anhydrous sampling. The samples were injected into the liquid IR cell through septa and the spectrum taken immediately. Reactions with substoichiometric aliquots of TiCl4 were allowed to proceed in the vial at ambient temperature (~20°C) and aliquots were sampled and observed at 30 min intervals.

89 4.3.3.3 Preparation and determination of TiCl4L2 stock solutions

EP and 3P were dried over molecular sieves for 4 h under N2. In a dry vial, 10 mL of 0.41 M TiCl4 (4.1 mmol) in DCM was cooled to 0°C under N2 with a magnetic stir bar. 0.87 mL (8.2 mmol) 3P or 0.96 mL (8.2 mmol) EP was added dropwise (strong exothermic reaction) to the stirred TiCl4 solution giving immediate yellow colour. A yellow-orange complex precipitated in the early stages of the addition. This precipitate began to redissolve as the second half of the aliquot was added, resulting in a clear orange solution once the sample warmed to RT.

The residual CHCl3 solvent signal of a batch of dry CDCl3 was quantified by integration against a solid aliquot of hexamethylbenzene. The decision to use the residual solvent signal as the internal standard was taken due to the difficulty in identifying another suitable compound that was inert to the highly reactive conditions, and did not overlap the other signals of interest. 50 μL of TiCl4L2 was dissolved in 500

1 μL of the same stock of dry CDCl3 and [L] determined by H NMR integration. The sample was also subjected to the [Ti] determination method described in section 4.3.3.1, both methods giving [TiCl4L2] = 0.40 M ± 0.02 M.

4.3.3.4 Degradation of TPTP by TiCl4L2

24.5 mg TPTP (0.08 mmol) was weighed into a 2 mL dry volumetric flask and made up to the mark with CDCl3. The residual solvent concentration was pre-quantified against a solid sample of hexamethylbenzene, and used as an internal standard. 500 μL of this 0.04 M solution was added to a 5 mm NMR tube, the required amount of 0.4 M

TiCl4L2 solution added under a dry N2 atmosphere and the mixture immediately frozen in liquid N2. The tube was flame sealed under high vacuum. The sample was thawed and well shaken immediately prior to inserting into the NMR spectrometer for measurement. Other samples were exempted from the freeze-vacuum-seal method and 90 were capped, sealed and shaken immediately upon removing from the N2 manifold.

These samples were analysed as soon as practicable (after inserting, shimming, etc) by

1H NMR.

4.3.3.5 Degradation of TPTP by SbCl3

In initial experiments, 10 mM solutions of TPTP were added to solid aliquots of

SbCl3 in dry conditions, and the mixture sampled directly by GCMS. It was found that this method resulted in very fast TPTP degradation within the injection liner of the

GCMS. The following method was used as a deliberate study of the reaction.

TPTP (61.2 mg, 0.2 mmol) and SbCl3 (45.6 mg, 0.2 mmol) were separately weighed into two dry 10 mL volumetric flasks and sealed, then diluted with dry DCM to make 10 mM solutions. 1 mL of each sample was added to a dry ampoule with a manifold attached. Upon the addition of the solutions, the sample was frozen in liquid

N2, vacuum degassed and flame sealed under vacuum. 14 samples were prepared and stored in liq N2 (two were reserved as t0 controls). To commence the reaction, 12 samples were quickly thawed in warm water with agitation, then placed in a thermostated bath at 50°C and the stopwatch started. At 5, 10 and 20 mins, 4 samples were removed and immediately frozen in liquid N2. One by one, the samples were thawed and the ampoule opened. 500 μL was used for SbIII determination, and the remaining sample quenched over CaCO3 with a drop of water. The sample was syringe filtered, then 500 μL was placed in a 2 mL volumetric flask along with 20 μL of 0.2 M ethyl acetate in MeCN as internal standard and made up to the mark with MeCN.

Samples were analysed by GCMS.

In parallel, the reserved 500 μL reaction sample was used for spectrophotometric determination of SbIII using the method of Christopher et al.16 Control samples of

III known [Sb ] in DCM were used to confirm that the direct addition of DCM reaction

91 samples to the buffer did not affect the accuracy of the described method. The absorption at 560 nm was recorded and compared against standard calibration samples.

The t0 samples were lost due to a sampling error.

4.4 Results and discussion

4.4.1 Titanium tetrachloride

The reactions of TiCl4 with TPTP proved to be highly dependent on sample preparation methods, concentration of the reagents and the stringent absence of moisture. Many variables and multiple simultaneous reactions made for a complex system which was investigated by controlling selected reaction conditions over a number of experiments. In order to present the results in a coherent fashion, the results of the excess and sub-stoichiometric TiCl4 reactions will be described first as these conditions describe the two primary reactions most clearly. This will be followed by a discussion of the 1:0.5-2* regime, which proved to be a complex combination of the excess and sub-stoichiometric reactions. Finally the differences in rate, products and other observations under these conditions will be used to indicate key aspects of the mechanism of this reaction.

4.4.1.1 The reaction of TPTP with a large excess of TiCl4

The reaction of TPTP with excess TiCl4 (1:>2) at room temperature in dry CDCl3 resulted in the formation of a pale yellow/green solution immediately upon addition. 1H

NMR showed the key products were Ti-bound ethyl propanoate (EP) and 3-pentanone

(3P), with the proton signals strongly shifted downfield from those of the free ligands.

These shifts are discussed in the context of the substoichiometric reaction in Section

* In all further discussion (unless indicated otherwise), ratios are given as peroxide:Ti, where Ti is the relevant Ti species . 92 4.4.1.2. The addition of pure 3P and EP (either individually or as a mixture) directly to a solution of TiCl4 resulted in the same immediate colour change and NMR shift, which appear to be indicative of the formation of a Ti-(O=C) complex. The reaction was extremely fast, as evidenced by 1H NMR or FTIR spectra of the TPTP reaction mixtures taken ~2 mins which showed no residual TPTP.

Further reactions were conducted in vacuum-sealed NMR tubes to confirm the reaction products in the strict absence of moisture and oxygen. The initial ratio of products was extremely consistent between repetitions, with two EP and one 3P molecules formed for each molecule of TPTP reacted:

TPTP + TiCl4 → TiA + 2(EP) + (3P) (R1TPTP)

where TiA represents a new (unknown) titanium species. The nature of TiA is discussed later in this chapter.

4.4.1.2 1:0.5 reaction of TPTP with TiCl4

In reactions with sub-stoichiometric quantities of TiCl4, the reaction seemed to occur in two general phases. The initial very fast reaction (R1TPTP) was accompanied by a slower reaction where remaining TPTP continued to degrade to completion over ~2-4 hours* (at 0.05 M initial [TPTP]) forming further EP and 3P. A pale yellow precipitate formed after ~1 h, continuing over several days.

* Adding exact stoichiometric ratios of the dilute anhydrous TiCl4 solutions to NMR samples was extremely difficult due to minute exposure to atmospheric moisture during μL-syringe transfers. Additionally, the chlorination products produced concurrently to the degradation reaction overlap TPTP's NMR signals, making precise integration of the final stages of reaction 2 impossible. As such, the 2 h time of reaction was estimated by analysing the % remaining TPTP during the first 60 min over a number of reactions, and extrapolating to a plateau at [TPTP] = 0. } 93 Strong deshielding effects were again visible by 1H NMR, consistent with the association of EP and 3P products with the Ti species present post-reaction under these anhydrous conditions. Deshielding of a similar magnitude was found when EP and 3P were added to TiCl4 under identical conditions. TPTP signals did not exhibit any notable downfield shift, suggesting negligible coordination of this species with Ti.

Table 4.1 shows the chemical shifts of the reaction products at various ratios of TPTP to

Ti, which governs the post-reaction L:Ti ratio (where Ti signifies all present Ti species, and L = EP or 3P). The Ti-bound ligands were immediately released upon moisture entering the solution, accompanied by a white precipitate which was reasonably assumed to be Ti hydroxide and oxide products. The same reaction with water occured for TiCl4:L complexes that had not reacted with peroxide. This method of adding water

17 to recover ligands from TiCl4 complexes is established in the literature.

Table 4.1 - 1H NMR chemical shifts of EP and 3P at given Ti ratio.

TPTP:Ti ratio (calc’d. L:Ti ratio) L:Ti ratio

Free 0.5 (0.33a) 1 (0.33) 3 (1) 1b a 2.37 +0.10 +0.33 +0.60 +0.56 b 0.99 +0.08 +0.18 +0.30 +0.31 a 4.08 +0.13 +0.29 +0.49 +0.48 b 2.26 +0.14 +0.40 +0.63 +0.60 c 1.21 +0.06 +0.12 +0.22 +0.22 d 1.09 +0.07 +0.14 +0.21 +0.21 L:Ti refers to ligand to Ti ratio, which in the case of TPTP reaction mixtures is calculated as Ti:(EP+3P). Shifts of bound ligands are given in ppm as a shift from free ligand. a Value at close to time=0 when 3 ligands are formed per titanium centre. This ratio changes over time as TiA continues to slowly degrade TPTP. b Free ligand added directly to TiCl4 in absence of TPTP.

The size of the shift is dynamically linked with the amount of Ti species present, with ∆δ increasing to a maximum at 1:1 Ti to L ratio, indicating fast intermolecular exchange between bound and free L and the formation of a strongly associated 1:1

TiCl4:L complex. This behaviour is consistent with that observed for other TiCl4-ketone

94 adducts.17-18 Variable temperature experiments showed that at higher temperature the chemical shifts moved slightly closer to those of the free ligand in both TiCl4:L and

TiA:L samples, which is again consistent with very fast intermolecular exchange (see

Figure 4.2).

1 Figure 4.2 - The effect of temperature on the H NMR spectrum of 3P 1:3 TiCl4.

The observation of a precipitate suggested that TiA was still reactive toward TPTP via this slower degradation reaction, a conclusion supported by elemental analaysis and

FTIR (discussed later). This second TPTP degradation reaction was sufficiently slow to allow its progression to be monitored by 1H NMR. The gradual decrease in the integral of TPTP after the initial fast reaction is depicted in Figure 4.3. When loss of TPTP over time, as the mole percentage of total organic species visible by 1H NMR, was fitted to a

* single phase exponential decay, the y-intercept reflected that at t0 for the reaction of

TPTP with 0.5 equivalent of TiCl4, TPTP represented ~40% of all organic species. This

*Exponential decay was used as ln([TPTP]) over time was highly linear, indicating pseudo-first order kinetics. 95 is roughly in proportion with the amount of added TiCl4, indicating that the initial (1:1) reaction of TiCl4 with TPTP is much faster than the reaction of TiA with TPTP. As a result, only the slower rate of the TiA-TPTP reaction was measureable after the 5-minute time delay imposed by NMR sample preparation.

5 0

4 0

3 0

2 0 T P T P (% )

1 0

0 0 2 0 4 0 6 0 T im e (m in ) Figure 4.3 - Rate of TPTP degradation by a half mole equivalent of TiCl4. The percentage of the total 1H integration (sum of EP/3P/TPTP) which is accounted for by TPTP is shown. Calculated by two independent signals (1.8 ppm and 0.9 ppm, mean and SD shown). Line represents a single phase exponential decay fit.

1 The H NMR spectra of TiCl4:L complexes were sharp at ambient temperatures, whereas TPTP degradation mixtures exhibited broadening of TiA:L signals that was most evident at these sub-stoichiometric initial ratios of TiCl4 (Figure 4.4). Dynamic behaviour is known to occur in dissolved titanium complexes,19 and can arise from various phenomena including ligand exchange and aggregation of titanium species to form dimers and higher oligomers. Fluxional behaviour is common in pentacoordinate metal centres20 and has also been reported for hexacoordinate titanium centres.21 In the present case, the slight broadening of peaks observed in reactions of TPTP could be caused by a somewhat slower rate of ligand exchange with the TiA centre than with the

TiCl4 centre, or by exchange between different Ti species, or another mechanism altogether. The very broad peaks initially observed in the Ti-limited reactions sharpen and move upfield as the reaction between TiA and TPTP continues, presumably 96 consuming TiA. Spectra obtained over only a single transient also exhibit broadening, so the effect is not caused by an averaging of the peak shifts as the reaction progresses (as seen in Figure 4.4). Resolving the bound and free states by cooling the sample to slow the rate of exchange could not be achieved as lower temperatures precipitated the analytes. This limitation did not impact the overall aim of the present investigation, and hence was not pursued further.

1 Figure 4.4 - H NMR spectra of the reaction of TPTP with 0.5 equiv TiCl4 over time. ■- ethyl propanoate,▲- 3-pentanone, ●- TPTP. The unlabled peaks are chlorinated organic products due to side reactions and are assigned in later spectra.

Figure 4.4 also shows the gradual appearance of a number of chlorinated products, such as the characteristic doublet of 2-chloro-3-pentanone (2C3P) at ~1.6 ppm. It is important to note that over similar time periods these chlorination products were only detected in much smaller quantities when excess TiCl4 was reacted with TPTP, and at trace levels where TiCl4 was reacted with EP and 3P. These observations combined

97 with the relative rates of the >2:1 and 0.5:1 reactions are indicative of a two-stage general reaction of TiCl4 with TPTP:

2(TPTP) + TiCl4 → TPTP + TiA + 2(EP) + (3P) (R1TPTP)

TiA + TPTP → 2(EP) + (3P) + Cl* + TiB (R2TPTP)

where Cl* is indicative of an active oxidative chlorinating agent and TiA and TiB refer to the first and second Ti products respectively. It is also possible that TiB is in fact the active chlorinating agent itself. In total, one mole of TiCl4 is able to degrade two equivalents of TPTP within 2 h under the conditions described. The multiple simultaneous chlorination reactions and overlapping peaks make the exact stoichiometry of R2TPTP difficult to follow, but the qualitative product ratios suggest that the 1:2 ratio of 3P:EP is maintained in R2TPTP, with both of these products then being consumed by side-reactions. A detailed discussion of the chlorination products can be found in section 4.4.1.5.

The chlorination products were identified by comparison against NMR spectra from the literature of probable rational products, and further validated by structural information from COSY experiments (Figure 4.5) conducted on the reaction mixtures.

Analysing the 1H NMR integration of all identifiable organic solution products after chlorination reactions had stabilised (9 days) showed a relative increase in ketone (or related chloro-species) in the total product distribution. Assuming R1TPTP proceeds as described above, the actual proportion of EP observed after chlorination is ~25% below that required for a 2:1 ratio with 3P (Table 4.2). Possible explanations for this might include breakdown of EP via chlorination reactions, or increased production of ketone in R2TPTP. Based on the results presented thus far, it was not possible to conclusively

98 determine the primary cause of the change in product ratio. This effect is discussed further in the next chapter where it is more prominent.

1 Figure 4.5 - H NMR (1D & COSY) of 0.5:1 TiCl4-TPTP reaction chlorinated products. Spectra taken 9 days after initial reaction. Compound name and reference: A- chloroethanol,22 B- 2-chloropropropanoyl chloride,23 C- 2-chloro-3-pentanone,24 D- 3-pentanone,25 E- chloroethane,25 F- 3-pentanone.25 Crosspeak between 3.0 ppm and 1.2 ppm was not able to be assigned.

Table 4.2 also shows that of the 6 units of organic substrate produced by the degradation of 2 units of TPTP, the identified chlorinated organics account for 3 units of chlorine atoms from TiCl4, with ~2 of these being related to oxidative chlorination

99 processes (mainly 2C3P as well as 2-chloropropanoyl chloride). Other chlorination products are possibly linked to HCl or radical processes.

Table 4.2 - Organic products from 0.5:1 TiCl4-TPTP reaction (9 days). Product mol % Sum 1 ethyl propanoate 44 2 1-chloroethanol 5 57 3 chloroethane 2 4 2-chloropropanoyl chloride 6 5 3-pentanone 11 6 3-chloro-3-pentanol 6 43 7 2-chloro-3-pentanone 26 Total mol % of chlorine 51a Sum reflects the total amount of organic product which are derivatives of EP or 3P. Accuracy of figures is slightly affected by overlapping peaks in some regions of the spectra. Italicised species contain two chlorines.

4.4.1.3 Characterisation of solid product from Ti-limited reactions

The precipitate from R2TPTP which had been allowed to proceed for 9 days was isolated under N2, dried and examined as a solid in KBr discs. Figure 4.6 shows the resulting spectrum.

The strong broad peak at 3250-3400 cm-1 and broad absorption from 400-900 cm-1 are strongly indicative of a propanoate salt with the overall spectrum being very similar to that of potassium propanoate.25 Once dried, the precipitate appeared reasonably stable to small exposures of atmospheric moisture. It was noted that long drying in vacuo led to the reduction of the organic scent associated with the sample, which would then return within minutes of exposure to air. The IR displayed some minor reduction in complexity after vacuum treatment. This appears to indicate that the precipitate retains a degree of moisture sensitivity whereby the organic moieties are released from the Ti centre as it is hydrolysed.

100

TPTP reaction. TPTP - 4 FTIR of precipitate from 0.5:1 TiCl 0.5:1 from of precipitate FTIR

-

6 . 4 Figure

101 1 H NMR analysis of a sample of R2TPTP precipitate which had been hydrolysed using D2O showed that the bulk of the remaining organic component was propanoic acid (>80 mol%), an assignment confirmed by comparison with authentic samples. A complex mix of chlorinated and non-chlorinated alcohols, acids and acid chlorides were also present in minor quantities (these products were not exhaustively elucidated). Thus it can be concluded that the precipitate from R2TPTP is largely compromised of a Ti- propanoate species. The propanoate may be derived from the breakdown of EP.

The isolated R2TPTP precipitate (before significant exposure to the atmosphere) was analysed for C, H, N, Ti and Cl. Table 4.3 shows the results compared against a possible Ti propanoate species indicated by the stoichiometry of the reaction. A second

(Ti-only) analysis on a smaller sample from a separate reaction showed excellent reproducibility of the Ti content.

Table 4.3 - Elemental analysis of precipitate from 0.5:1 TiCl4-TPTP reaction.

Measured C H Ti Cl O a R2TPTP Precipitate 12.00 2.76 25.9 20.2 39.11 Calculated C H Ti Cl O (C3H5O2)2 Ti3H5O8Cl3 13.62 2.86 27.14 20.10 36.28

N was not detected in the analysis. a - The balance required to make 100% was used to estimate O content, a reasonable assumption given that no other elements were present during the reaction and subsequent handling.

Since the NMR spectrum of the hydrolysed precipitate showed a number of organic products, this precipitate may not be a homogeouous product, or may be a complex polymer. Accordingly, this elemental analysis is not intended to provide a definitive proof of characterisation. Instead, the result confirms an oxygen-rich content with a clear 1:1:4.5 Ti-Cl-O ratio in the precipitate. The 1:1 Ti:Cl ratio agrees well with the earlier chlorinated organics analysis, as it accounts for the remaining chlorine of

TiCl4. The calculated O balance of 4.5 accounts for the two oxygen atoms in propanoate, as well as the two oxygen atoms “missing” from the two units of degraded

102 TPTP (i.e. not accounted for in the 2(EP):1(3P) product distribution). The slightly higher oxygen ratio may indicate further hydrolysis or oxo-bridges, however this remains speculative. Further experimental work is necessary to completely characterise this product.

The precipitate formed in the end-stages of R2TPTP may resemble sol-gel products derived from the reaction of Ti-alkoxides and acetic acid, esters and water. These reactions have been shown to form highly ordered products such as hexameric ring structures.26-27 A specific, ordered Ti polymeric or crystalline propanoate-oxido- chlorido species may also drive the formation of the precipitate in the TiCl4-TPTP reaction, explaining the observed reproducibility in the elemental analysis results for the solid. The ratio of ~2 propanoate to ~3 Ti indicated in the elemental analysis supports the presence of such a crystalline product.

4.4.1.4 IR Studies

Data from FTIR spectroscopy experiments were used to support the conclusions drawn from 1H NMR experiments. The reactions were directly observed under anhydrous conditions by injecting samples into pre-dried liquid cells. These experiments were also conducted at higher (0.05 M) concentration conditions to allow comparison with the 1H NMR data presented up to this point. The spectra of key reaction participants were taken as well to allow comparative assignment of the reaction spectra. In the case of the limited TiCl4 reactions, a sequence of spectra was taken over time to monitor the progression of R2TPTP. Dipentanoneperoxydihydroperoxide

(DPPDHP) and the cyclic dimer dipentanone diperoxide (DPDP), which bear a strong resemblance to the intermediates present in the acidic breakdown of TPTP,8, 28 were also analysed for comparison. Table 4.4 summarises the key data discussed, and the spectra of the TiCl4 reactions are provided in Figures 4.7 and 4.8.

103 Table 4.4 - IR data for reaction participants and mixtures. Reference samples TPTP 595(w), 927(s), 951(w), 973(s), 1133(s), 1158(s), 1464(s), 2978(s) 486(w), 669(s), 928(s), 957(s), 1043(m), 1144(s), 1157(s), 1275(m) 1352(m), DPDP 1464(s). 570(s), 615(s), 921(s), 966(s), 1155(m), 1130(m), 1386(s), 1463(m), DPPDHP 2978(m), 3415(m) EP 806 (w), 860(w), 1196(s), 1346(w), 1728(s) 3P 812(w), 1122(w), 1460(w), 1713(s) TiCl4-EP 405(s),455(s),495(s) , 802(w), 858(w), 1330(m),1616(s) 397(m), 457(s), 496(s), 813(w), 860 (w), 1097(w), 1381(w), 1460(w), TiCl -3P 4 1651(s) TiCl4 496(s) 500-800(br, s), 880 (m), 1078 (m),1299 (m), 1412(m), 1556(s), 1647(m), K propanoate25 3400(br, s) Reaction mixtures

(TiA /TiCl4) mix 374(s), 389(m), 452(w), 804(m), 860(m), 1026(w), 1087(m), 1196(w), 2:1 TiCl4:TPTP 1331(m), 1377(m), 1462(w), 1628(s) (TiA/TiB) mix 0.5:1 TiCl4:TPTP 366(m), 376(s), 397(w), 418(w), 808 (w), 860(w), 924(m), 951 (w), 976(m), at t=10min 1032(w), 1134(m), 1157(m), 1196(s), 1346(w), 1373(w), 1462(m) , 1730(s) Precip a 0.5:1 TiCl4:TPTP 739(w), 883(w), 1156(w), 1302(w), 1436(m), 1522(m), 1624(m), 1700(w), after 4 days 2362(w), 3258(br, vs), 3375(br, vs) aPrecipitate was measured as a KBr disc, all other spectra were in DCM solution. Data for K propanoate is from the literature as referenced. Numbers in bold are the key stretches discussed in the text.

104 stretches. stretches.

- ion, Ti low field reg field low

–

with EP and 3P are at 1:1 molar ratio. Blue box, carbonyl region. Red Box Box Red region. carbonyl box, ratio. Blue molar at 1:1 3P are and EP with

4 4 4 :L complexes and free L. free and complexes :L 4 TiCl TiCl - - 3P 3P EP EP FTIR of TiCl FTIR

-

7 . 4 Figure of TiCl Reactions = DCM. Solvent

105

intervals.

reactions.

4 TiCl - 5 min 35 min 65 min 95 min

– – – – 4 TiCl 1:1 1:0.5 1:2 1:0.5 1:0.5 1:0.5 FTIR of 1:0.5/1:1/1:2 TPTP of 1:0.5/1:1/1:2 FTIR

-

8 . 4 Figure min 30 in tracked is time. 1:0.5 over change do not & 1:2 1:1 = DCN. Solvent

106 Upon 1:1 reaction of TiCl4 with EP or 3P (analogous to a 3:1 TiCl4/TPTP

-1 -1 degradation reaction), the carbonyl stretches were red-shifted by -112 cm and -62 cm respectively (blue box, Figure 4.7), consistent with the bonding of the C=O moiety to the Ti centre. For EP the C-O stretch was also shifted, in the opposite direction (+135 cm-1). A clear pattern of strong absorption appeared in the low frequency region where the Ti-Cl stretches would be expected (red box, Figure 4.7).

A similarly red-shifted carbonyl peak was seen for the products (EP and 3P) in

TPTP reactions where excess TiCl4 was used (blue box, Figure 4.8). Peaks for both free and bound carbonyl could be seen at 1:1 ratio of TPTP to TiCl4 (and to a lesser extent in the 1:2 reaction), consistent with the 1H NMR experiments that showed resonances intermediate between free and bound ligands under the same conditions. In reactions of

TPTP with 0.5 equivalent of TiCl4, carbonyl peaks were observed at frequencies corresponding to free EP and 3P. This observation together with the significant broadening effect and very small downfield shift seen by 1H NMR for the equivalent reaction sample suggest weaker binding of the ligands with TiA compared to TiCl4, and/or altered molecular geometry.

The clear, strong peaks evident in the low frequency region of TiCl4:L complexes

(red box, Figure 4.7) were not present in the reaction spectra (red box, Figure 4.8), supporting the formation of new, distinct Ti species upon reaction with TPTP. The

-1 broad absorption band at ~350-500 cm in the 1:2 TPTP to TiCl4 reaction, where some

TiCl4 is expected to have remained unreacted, may indicate extremely fast exchange of ligands between TiA and any TiCl4 that has remained unreacted.

The complex mix of products makes the detection of the possible intermediates

DPDP and DPPDHP difficult. The relatively uncluttered region below 700 cm-1 provides the best opportunity to avoid the absorption of 3P and EP, however DPDP’s

107 most characteristic stretch in this region is 669 cm-1 which is masked by the solvent.

Most of DPDPs other signals are quite close to TPTP, making it difficult to identify which homologue is responsible for the signals. Thus FTIR is unsuitable for confirming the absence of DPDP as indicated by our 1H NMR results. DPPDHP might be assigned to the small peak at ~580 cm-1, but the presence of this peak in every spectrum in Figure

4.8 (including unreacted TiCl4) seems to suggest this absorption may instead be caused by deposition of a hydrolysed Ti species in the cell due to trace moisture. The absence of DPPDHP’s other major stretch at 615 cm-1 suggests that this species is not present in any appreciable quantity in these reactions. The lack of intermediate peroxy species correlates well with the 1H NMR results, although it is impossible to be conclusive due to extensive overlap in the spectra of the likely species.

4.4.1.5 The effect of reagent ratio

The very fast rate of TPTP degradation seen in the presence of excess TiCl4 became highly variable as the ratio of TiCl4 was reduced, indicating a complex overlap between the fast R1TPTP and the slower R2TPTP. Precise control of the molar ratios became extremely important in these reactions, which was made difficult due to the extreme reactivity of TiCl4. Small variations in the way the reagents were added to the reaction also impacted on the products and rate of reaction. As a result, it is problematic to attempt to draw quantitative conclusions from an individual set of conditions. It is, however, possible to outline the qualitative trends as each of these factors is controlled.

To investigate the effect of reagent ratios, 0.02 - 0.08 M TPTP in CDCl3 was

1 reacted with varying equivalents of TiCl4 and monitored by H NMR over time. Species were quantified using internal standard (IS) integration, and corrected for the expected moles of organic products based on the known aliquot of TPTP added to the reaction

(see Section 4.3.3.2). The abundance of key species (TPTP, EP, 3P and 2C3P) is plotted

108 in Figure 4.9. As R2TPTP has shown pseudo-first order kinetics for TPTP, the exponential decay model has been fitted to the data. The EP and 3P data sets also provided satisfactory fits to the first-order model. TPTP was heavily degraded in the

0.5:1 sample (Figure 4.9(f)) by the time the first spectrum was taken, providing insufficient data to allow robust fitting of the model.

a. TPTP 1:0.5 TiCl4 c. TPTP 1:1.5 TiCl4 e. TPTP 2:1 TiCl4 4 0 4 0 8 0

3 0 3 0 6 0 EP EP EP 3 P 3 P 3 P m o l m o l 2 0 2TPTP 0 m o l TPTP4 0 TPTP    2 C 3 P 2 C 3 P 2 C 3 P 1 0 1 0 2 0

0 0 0 0 1 0 2 0 3 0 0 1 0 2 0 3 0 0 1 0 2 0 3 0 m in s m in s m in s

b. TPTP 1:1 TiCl f. TPTP 0.5:1 TiCl 4 d. TPTP 1:2 TiCl4 4 4 0 4 0 2 0

3 0 3 0 1 5 EP EP EP 3 P 3 P 3 P

m o l 2 0 m o l 1 0 m o l 2 0TPTP TPTP TPTP    2 C 3 P 2 C 3 P 2 C 3 P 1 0 1 0 5

0 0 0 0 1 0 2 0 3 0 0 2 4 6 8 0 2 4 6 m in s m in s m in s

Figure 4.9 - Reaction of TPTP and TiCl4 and varying concentrations and ratios. (a-e) solid line is based on a single phase exponential decay fitted using GraphPad PRISM software. (e) resolved to a single phase exponential decay fit, but result was ambiguous. (f) would not resolve to a rational result via single phase exponential decay fit, no fit shown. Blue circle on y axis indicates actual nTPTP at t0.

By examining the 1:X reaction series (Figure 4.9(a-d), X=0.5-2), a clear trend is apparent where the rate of degradation increases with higher ratio of TiCl4. This can be clearly observed in Fig 4.10 which shows the time required to achieve 98% degradation of the TPTP sample.

109 2 0 0

1 5 0

1 0 0 m in s

5 0

0 0.0 0.5 1.0 1.5 2.0 2.5

1 :X (T iC l4 ra tio )

Figure 4.10 – Time required for 98% degradation of TPTP at different TiCl4 ratios. Figures are based on the models shown in Fig 4.9, but constrained to plateau at [TPTP] = 0. Data represents time at the extrapolated [TPTP] = 0.4 µmol.

The data to this point has shown a large difference in the rate of reaction for

R1TPTP and R2TPTP. It can therefore reasonably be assumed that the initial fast degradation was completed in less than 5 mins, before the first data point.* Thus, the trends being observed in these plots are largely reflective of the slower step R2TPTP. The validity of the fitted exponential decay curves is attested by R-values averaging 0.98 across all species, and consistently higher values found for the fit of TPTP’s data. Good agreement between extrapolated t0 values for TPTP degraded compared with EP and 3P produced also supports this rate approximation, with the t0 values broadly reflecting the expected ratio of 1:2:1. The extrapolated t0 values for TPTP can therefore be used to approximate the proportion of the reaction that took place at faster kinetics (i.e.

R1TPTP): 1:0.5 – 33%, 1:1 – 52%, 1:1.5 – 60%, and 1:2 – 81%.

This series demonstrates quite a linear trend as illustrated in Figure 4.11.

Extrapolating forward indicates that R1TPTP would dominate completely at a ratio of

~1:2.7. Considering the probable effect of R1TPTP not proceeding to completion due to

* As described in section 4.4.1.1, it took approximately 2 mins to prepare a sample and acquire the first NMR spectrum. 5 mins was used as the first time point in these kinetic studies to allow proper adjustment of settings for accurate integration. 110 poor ability to mix in the NMR tube (i.e. diffusion limited), this result is quite close to

1:3. This is consistent with our experimental observation that at 1:2 ratio, traces of

TPTP are still detectable at 5 mins, however reactions at 1:3 had no detectable TPTP within 2 mins of reaction.

100 0 80

60 y = 30.4x + 18.5 R² = 0.9749 40

20 % TPTP degradation TPTPdegradation % at t

0 0 1 2 3

TiCl4 Ratio

Figure 4.11 - Degradation extrapolated to t=0 at various TiCl4 ratios. Extrapolated from the fitted single phase exponential decay curves in Figure 4.9.

The 1:3 ratio matches the number of peroxide bonds in each molecule of TPTP as well as the number of organic products produced by each molecule of TPTP when degraded by TiCl4. This may indicate one of two possibilities:

a) Upon reaction with a single peroxide moiety within TPTP, the resulting

TiA species is less reactive than TiCl4. This may be due to an oxygen-

containing ligand suggested by the oxygen missing from the organic

products, a possibility supported by the high inferred O content in the

microanalysis results of the precipitate formed during the slower reaction.

b) Upon liberation of sufficient ketone and ester to allow coordination of

these ligands with Ti species (either TiCl4 or TiA) before reaction with

TPTP (ie, below 1:3), the rate of reaction is significantly slowed.

111 For a) to hold true, we would expect that other, intermediate peroxy species would

1 be detectable by H NMR, considering the slower rate of reaction in R2TPTP. As no intermediate peroxy species (such as DPPDHP) are seen in any spectra where TPTP is undergoing R2TPTP, b) would seem the more plausible explanation. It is not possible from these experiments to conclusively determine the cause of the switch to R2TPTP.

a . 1 : 0 .5 b . 2 : 1 3 0 6 0

TPTP 2 C 3 P 2 0 4 0 3 P EP m o l m o l   1 0 2 0

0 0 0 1 0 2 0 3 0 0 1 0 2 0 3 0 m in s m in s

Figure 4.12 - Comparison between 1:0.5 and 2:1 TPTP-TiCl4 reactions. Dotted lines represent fitting to full datasets, solid lines represent fitting for first 15 mins only. Blue circle on y axis indicates actual nTPTP at t0.

It is also relevant to compare the 1:0.5 and 2:1 reactions (Figure 4.12) which demonstrates the effect of concentration at equivalent reagent ratios. There was little change in the overall TPTP degradation at 5 mins between the samples (1:0.5 - 43% vs.

2:1 - 42%), suggesting that R1TPTP is not affected by concentration to an extent that we can measure. Fitting the reaction progression (R2TPTP) using the exponential decay model used thus far worked well for the 1:0.5 reaction (Figure 4.12(a), dotted lines), but gave an ambiguous fit* for the 2:1 reaction data (Figure 4.12(b), dotted lines). To explore the reason for this poor fit, only the first three data points were fitted to the model (solid lines), as in this region the data visually appeared to be more consistent with exponential decay. The 2:1 data (Figure 4.12(b)) reveals a decided increase of reaction rate after a 15 min ‘induction period’, illustrated by the increased production of

EP (red ovals) and 3P (blue ovals) over the pre-15 min trend, significant acceleration of

* R2 was satisfactory (0.99), but 95% confidence intervals were very wide. 112 TPTP degradation (red arrow) and a spike in 2C3P production (green ovals). Treating the 0.5:1 data in the same manner reveals a similar but much less marked increase in overall reaction rate, suggesting a reaction phenomenon is responsible for this effect, rather than an artifact of the data analysis.

This ‘flattening’ of the fitted exponential decay curve as the reaction proceeds, particularly at higher concentrations, appears to reflect the increasing impact of the chlorination of 3P on the overall degradation of TPTP. Such interlinked reactions cannot be accommodated by fitting to a simple first-order model, implying that the overall rate of R2TPTP is affected by the chlorination of 3P. The sharp rate increase post

15 mins in the 2:1 reaction is accompanied by 6 times more chlorinated pentanone at 30 mins (relative to starting TPTP) than the lower concentration reaction. This observation suggests that the chlorination of 3P may be able to induce higher rates of TPTP degradation, possibly by re-generating a more active Ti species. The fact that chlorination is most significant after 50-60% TPTP is degraded strengthens the link between TiA and the chlorination process, as in this phase of the reaction the bulk of the

Ti is in the form TiA according to the proposed overall reaction scheme.

It should be noted that the samples in these reactions were not prepared using the freeze-seal-thaw method, but were rather capped and agitated by hand, and analysed immediately. The impact of diffusion and mixing was investigated by returning to the freeze-seal-thaw method of sample preparation. Two 1:1 reactions at 0.04 M were prepared using the freeze-seal-thaw technique or the cap-agitate method. Total degradation at 5 mins was calculated by comparing the integrations of EP and TPTP.

This method was chosen as the peaks were well resolved, and [EP] was less affected by chlorination than 3P. Furthermore, calculating concentrations using the residual solvent

IS method was not accurate in this instance, as solvent boil-off during the freeze-seal-

113 thaw process did not occur in the non-frozen sample, affecting the relative IS concentrations. As EP is much less volatile than CDCl3 and DCM, using the relative integration of TPTP and EP is a more accurate method to directly compare the reactions. The sample subjected to the freeze-seal-thaw process showed ~58% TPTP degradation, whilst the non-frozen sample had showed ~48% degradation. The strong mixing that occurs via bubbling as the sample thaws under vacuum has a measurable effect on amount of degradation taking place by the faster R1TPTP kinetics. The impact this might have on the fast, diffusion limited reaction is considered in the next section together with the related impact of the EP/3P ligands on reaction rate.

4.4.1.6 The effect of ligands

In order to understand how the binding of the ketone and ester to Ti affected the degradation reaction, TiCl4 was pre-reacted with the ligands EP and 3P prior to reaction with TPTP. The ability of TiCl4 to accommodate two such coordinated carbonyl- compounds is established in the literature,29-30 including in solution phase,31 so a 1:2

TiCl4 to L complex was used in our degradations to ensure the Ti centre was ‘saturated’.

During the synthesis of the TiCl4:L2 compounds, the capacity for two ligands to coordinate seemed to be reflected by a yellow/orange precipitate forming during the first half of the ligand addition (presumed to be TiCl4L complex). This precipitate re- dissolved during the second half of the addition, possibly due to a higher solubility of

TiCl4L2, forming a deep orange solution. The addition of the ligands was highly exothermic as was expected based on previous calorimetric studies of Ti-O=C complexes.31 Determinations of [L] and [Ti] by 1H NMR* and colorimetric back- titration were consistent with a 1:2 Ti to L complex in solution.

* Downfield shifts of bound ligands was broadly comparable to those reported in Table 4.1. 114 Individual 1:1 reactions of TiCl4(EP)2 and TiCl4(3P)2 with TPTP were conducted in NMR tubes at 0.04 M using the freeze-seal-thaw method to ensure complete anhydrous conditions and thorough mixing. The reactions were tracked by 1H NMR for

1 h, however pre-20 min spectra failed to provide reliable integration data. This may have been due to the high initial [L] coupled with the rapid change in [L] in organic analytes in this phase of the reaction. The problem was not encountered in post-20 min data. The post-20 minute data was compared with TiCl4 degradation experiments, although data for these experiments was only available for the first 30 mins. As both samples were prepared in the same manner, the residual solvent IS was used to quantify the products, however in this instance correction of overall organic products to match the known amount of TPTP added was not possible due to the additional ligands added with TiCl4L2. Figure 4.13 provides a comparison of reaction progression over 40 mins using 20-60 min data, which provided a consistent set of integrals for exponential decay fitting.

a . T P T P b . E P /3 P 2 0 T iC l4 (EP) 2 - T P T P T iC l4 (EP) 2 - 3 P 3 0 T iC l4 (3 P ) 2 - T P T P T iC l4 (3 P ) 2 - E P 1 5 T iC l4 - T P T P T iC l4 - E P 2 0 T iC l4 - 3 P m o l m o l 1 0  

1 0 5

0 0 0 1 0 2 0 3 0 4 0 0 1 0 2 0 3 0 4 0 m in u te s m in u te s c . 2 C 3 P 1 0 T iC l4 (3 P ) 2 - 2 C l3 P

8 T iC l4 (EP) 2 - 2 C 3 P

T iC l4 - 2 C 3 P 6 m o l  4

2

0 0 1 0 2 0 3 0 4 0 m in u te s Figure 4.13 - Reaction of TPTP with TiCl4L2 vs. TiCl4. Absolute molar abundances of analytes from the reactions are displayed in three separate graphs for clarity only. For the TiCl4L2 reactions, only the product that was not pre-reacted with TiCl4 is shown, as the very high concentration the prereacted L did not allow accurate integration. Blue circle on y axis indicates actual nTPTP at t0.

115 Some striking differences emerge between the reactions of the two TiCl4L2 species. Firstly, the TiCl4(3P)2 reaction appears to display greater initial degradation

(R1TPTP) than the TiCl4(EP)2 reaction. This may reflect the stronger Ti binding of EP as suggested by the twofold larger carbonyl red-shift seen in the IR results (Section

4.4.1.3). The reaction course for TiCl4(3P)2 bears a very close resemblance to the TiCl4-

TPTP 1:1 reaction although with a slight increase in overall TPTP degradation rate

(Figure 4.13(a)). Taken in isolation, the reliability of this observed rate enhancement would be questionable; however, as will be discussed later, this result is supported by the analysis in Figure 4.12.

The t0 values for nTPTP once again provide an indication into the extent to which the faster R1TPTP kinetics is applicable with each reagent. There is little difference between TiCl4(3P)2 and TiCl4, with both showing nTPTP values of ~9 μmol at t0, indicating that 3P has no measurable impact on R1TPTP. The reaction with TiCl4(EP)2

2 * extrapolates to nTPTP = ~13 μmol at t0 with an extremely good fit of R =0.9999. This suggests that ~1/3 of the degradation took place via faster kinetics (compared to ~1/2 with TiCl4(3P)2 and TiCl4). It seems that whilst EP does have a notable effect on the rate of TPTP degradation, its presence does not appear to be the only cause for the switch from R1TPTP to R2TPTP kinetics. If EP coordination was the primary cause of rate reduction, the TiCl4(EP)2 nTPTP t0 value would reasonably be expected to be closer to the actual initial nTPTP value of 20 μmol.

Previous experiments indicated that the exhaustion of R2TPTP is evidenced by precipitation in sub-stoichiometric reactions i.e. as the last 10-20% of TPTP is being degraded in the 1:0.5 TiCl4 reaction. The absence of precipitation in the 1:1 TiCl4(L)2

* Note that 5 datapoints (20, 30, 40, 50 and 60 min) were used to generate these decay fits, not just the 3 points shown. 116 reactions indicates that neither reaction was mediated exclusively via R2TPTP. It is likely that the rate inhibition effect of EP occurs equally across both R1TPTP and R2TPTP in line with its increasing concentration.

Returning to the possible reasons for loss of fast R1TPTP kinetics proposed in section 4.4.1.5, these results indicate that the presence of ligands is not the sole reason for the switch to R2TPTP. Adding excess EP does slow R1TPTP to some extent, but not enough to account for the full rate difference. Thus the alternate explanation is also responsible for the slower rate of R2TPTP; ie. an oxygen-containing ligand being bound to Ti after the reaction with TPTP (accounting for the missing organic oxygen). This oxygen-containing ligand also appears to be the primary cause for the precipitation of the Ti-propanoate species. This deduction becomes a foundation of the mechanism described in section 4.4.2.

The second major difference is the large increase in 2C3P production in the

TiCl4(3P)2 reaction compared to TiCl4 or TiCl4(EP)2 at equivalent reaction time. Whilst this is logical considering the much higher initial relative concentration of 3P in this reaction, the accompanying acceleration of TPTP degradation once again provides a subtle link between overall degradation rate and chlorination of 3P.* To ensure this is a real effect, the relative 2C3P abundance in Figure 4.13(c) must be considered in the context of the significant rate difference between the TiCl4(3P)2 and TiCl4(EP)2 reactions (Figure 4.13(a)). To correct for this difference, production of 2C3P was plotted against the progress of TPTP degradation. Figure 4.14 shows that both TiCl4L2 complexes result in significant increase in 2C3P compared to reactions using only TiCl4 at the same point in the overall TPTP degradation process.

* It is possible that this rate enhancement is merely a ‘dilution’ of EP’s inhibitory effect, however the strong binding of the ester (together with other experimental support for chlorination-related enhancement) makes this less likely. 117 Chlorinated products vs TPTP Degradation 12

10

8

6 TiCl4 mol 2C3P

μ 4 TiCl4(3P)2 TiCl4(EP)2 2

0 60.0% 70.0% 80.0% 90.0% 100.0% % TPTP degradation Figure 4.14 - 2C3P production plotted against TPTP degradation. Symbols for TPTP-X 1:1 reaction where X= ● – TiCl4 ■ – TiCl4(3P)2 ▲ – TiCl4(EP)2. All reactions conducted using freeze-seal-thaw reactions at ~0.04 M.

The fact that TiCl4 and TiCl4(3P)2 have similar TPTP degradation profiles in

Figure 4.13 adds confidence that the increase in 2C3P in the latter reaction (shown in

Figure 4.14) represents a real increase in relative chlorination activity. The fact that

TiCl4(EP)2 degradation is somewhat slower (shown in Figure 4.13) introduces the possibility that the increase of chlorination shown in Figure 4.14 over the TiCl4(3P)2 reaction must be treated with caution. This may merely reflect that the chlorination reaction has had more time to proceed at a given percentage of degraded TPTP, rather than necessarily indicating that chlorination is more prevalent in the TiCl4(EP)2 reaction.

This set of experiments provides the following key deductions:

a) 3P appears to exert little effect on the initial rate of reaction between TPTP

and TiCl4.

b) EP slows of the rate of R1TPTP, and may explain a portion of the change from

fast to slow kinetics in the early part of the reaction.

c) The chlorination of 3P is likely to be a separate process from the TPTP

degradation itself, as it is affected by overall [3P] and time (as suggested by

Figure 4.14 analysis).

118 d) It appears the comparatively facile chlorination of 3P provides a small but

reproducible enhancement in the overall TPTP degradation rate.

4.4.1.7 Diffusion effects

The complex relationships between reagent ratio and concentrations make it difficult to pinpoint the exact cause for the slowdown of TPTP degradation, and to explain how the degradation of two TPTP molecules occurs with just one equivalent of TiCl4. Yet the complexity of the results in itself provides part of the probable answer. The static

(i.e. unstirred) NMR tube reactions studied here mean that if R1TPTP is very fast and proceeds directly to R2TPTP, then diffusion will limit the proportion of the substrate that can undergo R1TPTP. This blurs the line between the two reactions, and denies access to clear stoichiometric ratios that might clarify the exact conditions under which the change occurs. These effects were evidenced in our experiment during the injection of the reagents into the NMR tube (before agitation or freeze-seal-thaw), when partitions could be seen where the reaction had proceeded (evidenced by the characteristic yellow colour associated with TiCl4-carbonyl complexes) and where the solution remained unreacted (colourless). This diffusion limitation may explain the slight decrease in

R1TPTP degradation in manually shaken samples compared to those prepared by freeze- seal-thaw methods. The latter undergo much more vigorous mixing and accordingly faster initial degradation.

As each unit of TPTP degrades, the local concentrations of EP and 3P increase.

EP slows the rate of TPTP degradation, an effect that increases together with [EP] as

R1TPTP proceeds. Whilst our results suggest that EP is partly responsible for the slowing of the reaction rate, a number of observations indicate that there are other factors at play. Most critically, the oxygen atom missing from the organic products post- degradation of TPTP appears to be bound to Ti (supported indicated by elemental

119 analysis). This oxy-ligand reduces the reactivity of the Ti species, and thereby the rate of TPTP degradation. Finally, rate deceleration is compounded by the decrease in Ti concentration as the propanoate precipitate is formed.

4.4.1.8 Inter- vs. Intra-molecular rearrangement

In Chapter 2, it was shown that the formation of esters in the degradation of TPTP in aqueous methanol and HCl was due to the intermolecular reaction of hydroperoxides and ketone products (Section 2.4.3). Acetophenone was used in Chapter 2 as an

‘oxidation trap’, as it is known to undergo BV oxidation.32 To determine if the same mechanism is responsible for ester formation in the TiCl4 degradation of TPTP, a similar assay was conducted by adding the acetophenone ‘oxidation trap’ to the 1:1

1 TPTP:TiCl4 degradation. This experiment was conducted in the same manner as the H

NMR experiments, except a GCMS vial was used rather than an NMR tube.

When TPTP was decomposed under acidic conditions, the addition of acetophenone resulted in a dramatic reduction in the proportion of EP formed. In contrast, the TiCl4-mediated degradation of TPTP showed no change in the relative peak areas for EP and 3P by GCMS in the presence of acetophenone. Furthermore, no oxidation products of acetophenone were detectable by GCMS. This experiment provides strong evidence that the degradation of TPTP by TiCl4 proceeds by a different mechanism to that of acidic degradation. Together with the lack of intermediate breakdown products (DPDP or DPPDHP), it seems likely that the formation of esters in this instance is due to an intra-molecular rearrangement rather than the action of hydroperoxides on ketones via the Baeyer-Villiger rearrangement.

120 4.4.2 Proposed Mechanism

Within the limitations of the methods utilised to study this multi-step reaction, there are some clear observations which can be used to propose a mechanism:

a) TiCl4 is able to mediate an extremely fast degradation of TPTP, but it is rapidly

converted to a different, less active Ti species (TiA);

b) The presence of multiple Ti species throughout the course of a 1:0.5 degradation

is supported by IR observations and elemental analysis;

c) The loss of only one organic oxygen per TPTP degraded, the fact that complete

degradation is possible from 1:0.5-1:2, and that fast kinetics dominates well

before ratios of 1:3 are reached, all indicate that a single molecule of TiCl4 can

degrade a complete molecule of TPTP via R1TPTP, not just a single peroxide

bond.

d) A chlorinating species is generated that is only slightly active with excess TiCl4

but becomes much more prevalent when TiCl4 is added in sub-stoichiometric

quantities.

The very fast initial reaction (R1TPTP) between TiCl4 (or TiCl4L2) and TPTP is illustrated by complete degradation in ~5 mins at room temperature at a 1:2 excess of

TiCl4. At lower TPTP:TiCl4 ratios, this reaction is followed by a slower reaction

(R2TPTP) producing the same primary products (EP and 3P) in approximately the same ratios before the onset of chlorination.* The second reaction is linked to the oxidative chlorination (primarily of 3P to 2C3P) either directly by a species of Ti, or via the liberation of a chlorinating species.

This chlorination reaction is most evident in reactions where TiCl4 is used in sub- stoichiometric amounts (1:0.5). Under these conditions, for every half equivalent of

* This cannot be conclusively proven at this stage. Chapter 5 re-examines this issue in detail. 121 TiCl4 is used, one of the chlorine atoms is accounted for in the precipitate and two by organic oxidative chlorination products (almost entirely 2C3P). The remaining chlorine is likely to be accounted for in the non-quantifiable mixture of minor chlorination products, and probable inorganic species (eg. hypochlorous acid) not elucidated experimentally by our methods. The time delay between the slowdown in rate and generation of stoichiometrically relevant quantities of 2C3P suggests that a highly active, oxidative chlorination agent is generated during R2TPTP. Between TPTP:Ti ratios of 1:0.5 and 1:2, a complex transition between R1TPTP and R2TPTP occurs. This may be partly explained by R2TPTP commencing before TiCl4 has been completely exhausted, due to R1TPTP being diffusion limited.

Cyclic peroxide breakdown intermediates have been shown to be peroxy/hydroperoxy oligomers or dimers under acidic conditions.4-5 No such products

1 were observed by H NMR or FTIR in either R1TPTP or R2TPTP. R1TPTP is so fast that intermediates would not be expected to be visible, but the fact that intermediate peroxy compounds were still not detected with the slower overall rate of R2TPTP makes it likely that both reactions involve a similar concerted mechanism. The overall rate of the reaction seems to depend on the relative reactivity of the Ti species present toward

TPTP. The extremely reproducible ratio of ketone to ester products (especially in

R1TPTP) also supports a concerted mechanism. The absence of ethyl radical products and the clean rearrangement process suggests radical processes are unlikely to be responsible for the central mechanism. Whilst an ultra-fast, ionic stepwise mechanism similar to acid degradation cannot be ruled out, all indications at this point suggest that this reaction is fundamentally different to the broadly accepted mechanisms of acidic and thermal degradation. Thus, this reaction is a completely novel addition to the known chemistry of cyclic peroxides.

122 The mechanism in Scheme 4.A provides an interpretation of the central reaction consistent with the experimental results. Whilst complexation of EP was shown to exert some effect on degradation rate, this is expected to affect both R1TPTP and R2TPTP and therefore does not appear to be the key distinguisher between the two reactions. For this reason and for clarity, EP/3P-binding was omitted in this scheme, as were the chlorination reactions. Both of these processes add a further layer of complexity to the main reaction occurring in these experiments, but appear to be secondary to the central

Ti-mediated TPTP degradation process.

R1TPTP

R2TPTP

Scheme 4.A - Proposed mechanism for R1TPTP and R2TPTP. - The candidates for each reaction and TiA/TiB are discussed in the text. Free Cl is shown for clarity only i.e. steps may be concerted.

123 The suggestion of an OCl ligand on a titanium complex is a novel one, with such ligands only sparsely reported in the literature (an extensive search only found two references,33-34 with neither providing experimental confirmation of the species). An alternate possibility considered was the production of Cl2 via reaction R1TPTP(b), however this would suggest chlorinated organic products forming in similar quantities throughout R1TPTP and R2TPTP and thus was not supported by the experimental observations. The relative absence of chlorination in the excess TiCl4 reaction supports the concept of TiA acting as a reservoir for the chlorinating agent OCl. Also, the

R1TPTP(b) mechanism would imply the formation of TiOCl2 which should exhibit strong absorption at ~820 cm-1 according to previous studies,35 yet no such absorption is observed in the FTIR studies of either R1TPTP or R2TPTP. It is therefore suggested that

R1TPTP is more likely to proceed via (a) than (b). Direct spectroscopic evidence for an

OCl ligand is troublesome, especially as there is no such moiety described in spectroscopic literature for comparison. A potential reference source might be the O-Cl stretch of hypochlorous acid36 at 739 cm-1, however this is within the absorption band of the solvent DCM and is masked in our spectra. A Ti-bound hypochlorite ligand is also likely to have significantly different absorption to hypochlorous acid making a direct match unlikely, thus changing the reaction solvent to observe this region was not pursued. The presence of a significant IR absorption around ~750 cm-1 region in the final precipitate could be argued to be related to O-Cl, however this is far from conclusive. A separate, dedicated study would be needed to confirm this moiety, and was beyond the scope of this work.

Whilst no definitive experimental evidence confirmed the presence of an OCl ligand, the stringent anhydrous conditions make it difficult to suggest alternative explanations that account for the fate of the missing organic oxygen and the observed

124 chlorination processes. Formation of free hypochlorous acid would require the loss of an organic proton, which is not accounted for in our 1H NMR observations until the first chlorination of 3P. Hypochlorous acid cannot be used to explain the fate of oxygen in

R1TPTP, where very little chlorination occurs. As the ratio of TiCl4 is reduced, however, the chlorination of 3P becomes prevalent. This reaction would liberate a proton by the substitution reaction shown in Scheme 4.B (reaction R3P), yielding hypochlorous acid.

Scheme 4.B - Proposed mechanism R3P - the α-chlorination of 3P by TiCl3(OCl).

R3P relies on 3P forming a titanium enolate, a highly favourable process due to the oxophilicity of titanium. The instantaneous, highly exothermic binding of 3P to TiCl4 observed in Section 4.4.1.6 demonstrates this affinity. A similar mechanism has been used to explain an analogous α-chlorination of ketones using urea-hydrogen peroxide with tetrachloridosilane, where a silyl-OCl species was postulated to be the active chlorinating agent.34 Hypochlorous acid could then readily chlorinate at the α-carbon position, breaking down the now-unstable Ti-adduct to leave a hydroxy ligand. The latter species could immediately form Ti-O-Ti bridges (a significant driving force), releasing HCl or HOCl in the process, depending on the particular Ti species involved.

This gradual release of free acids could then effect the ring-opening of TPTP to more reactive hydroperoxide species, potentially explaining the slight rate enhancement seen in the presence of high [3P] (Figure 4.13).

125 The ‘induction period’ observed most prominently in the 1:2 experiments* might also be explained by the autocatalytic release of organic protons (shown in scheme 4.B as Ti-OH) as R3P progresses, leading to the cascading generation of further HOCl. Once this free HOCl is released, R3P can also occur with TiCl4-bound 3P, increasing the [3P] susceptible to R3P. Thus the mechanism in Scheme 4.B not only accounts for the formation of stoichiometrically significant levels of 2C3P, but also the ~15 min delay in

2C3P formation and potentially the slight increase in overall reaction rate seen with

TiCl4(3P)2 vs. TiCl4. TiCl4 is far less likely to effect a R3P-type reaction due to the chlorido ligands being δ-, in contrast to the opposite δ+ polarisation on chlorine in the hypochlorido ligand.

R3P addresses the primary chlorination process of 3P to 2C3P, but the reduced chlorination observed in excess TiCl4 reactions must also be accommodated by this mechanism. The chlorination mechanism described should be equally available after

R1TPTP, however under excess TiCl4 conditions, it is likely that a lower proportion of 3P would be bound to TiCl3(OCl) due to competition from the presumably more oxophilic

TiCl4 to form TiCl4L1-2, thus explaining the reduced prevalence of R3P under these conditions.

At this stage the chlorination process becomes very complicated, with the probable generation of HOCl, HCl, and potentially Cl2O leading to the multiple chlorination products seen. Whilst it is not feasible to fully describe this complex inter- related web of reactions, this scenario does explain the profusion of chlorinated organic products seen in the latter stages of reactions with low TiCl4 ratios.

* Induction period refers to the ~15 minute delay before siginificant formation of 2C3P is observed, concomitant with a slight increase in overall TPTP degradation rate. Discussed in 4.4.1.5 (esp. Figure 4.11) and 4.4.1.6 (esp. Figure 4.12). 126 The proposed mechanism rationalises the described observations in a cohesive sense, but further work is needed to test the mechanism and fully understand this novel reaction. In particular, there are numerous possible branches to R2TPTP which require further study. Whilst supported by the overall reaction observation, the Ti-OCl moiety

* suggested in TiA and TiB have yet to be characterised conclusively. The ability for a number of different Ti compounds to facilitate the rearrangement of cyclic peroxides indicates the potential for a future project to design a catalytic system based on these reactions.

4.4.3 Antimony trichloride

The theoretical work by Dubnikova et al. suggested that the Sb3+ ion would be similarly active to Ti4+ in facilitating the ring-opening of TATP.11 During initial experiments undertaken as part of this investigation Lewis acids, samples taken directly from anhydrous reactions of SbCl3 and TPTP were analysed by GCMS. Whilst this gave impressive results (near-complete degradation at within 10 mins even at 50% mol equivalent Sb) it was soon identified that this reaction was actually taking place at the

GCMS injection liner at 130°C. Considering the absence of broad or tailing peaks on the chromatogram, it also appeared that the reaction was complete before the sample entered the column for separation. For this to be the case the reaction must have proceeded to completion in the few seconds for which the sample was in the injection port. These experiments, whilst not an ideal example of GCMS sampling, provided important results as will be discussed below. Modification of the sample preparation to quench SbCl3 and extract the organic analytes prior to analysis revealed a considerably slower reaction, indicating the reaction is sluggish at room temperature.

* Indeed, the simplistic assignment of TiA and TiB may in fact describe a range of different Ti products, and may not be able to isolated as such. 127 The 1:1 reaction was conducted at 50°C under anhydrous conditions in an attempt to measure degradation on a manageable timescale. The results in Figure 4.15 show complete degradation within 20 mins. As with the direct injection results, EP was formed in slight excess to 3P, but without the consistent 2:1 ratio seen in the TiCl4 reaction. Colorimetric speciation16 of Sb in the same samples indicated that SbIII was being consumed in the reaction, which is presumably an indication of oxidative processes taking place during the reaction with TPTP. Although attempts to quantify

SbV by literature methods37 were not successful, the fact that the loss of SbIII did not correspond with either the moles of EP/3P produced, nor the moles of TPTP consumed suggests that this presumed oxidation process is not directly linked to the central degradation mechanism.

8 [S b III]

Ethyl Propionate 6 3-Pentanone

TPTP 4 m M

2

0 0 5 1 0 1 5 2 0 2 5 T im e

Figure 4.15- The progression of 1:1 reaction of SbCl3 and TPTP at 50°C. EP, 3P and TPTP results from GCMS, [SbIII] via colorimetric assay. Error bars reflect SD from mean value over 4 repetitions.

The SbCl3 reaction resulted in a higher ratio of ketone to ester products compared

III to the TiCl4 reaction. A likely explanation of this may be that the oxidation of Sb consumed the oxygen required for ester formation, hence leading to a higher proportion of ketone. The reaction which produced the results most similar to that of TiCl4 degradation (2/3 ester, 1/3 ketone) was the 10% SbCl3 direct GCMS injection (Figure

128 4.16). At injection port temperatures, it is possible the formed Sb compounds are effectively acting as a catalyst for either a mechanism similar to that proposed for TiCl4, or potentially a radical thermal mechanism. It can be hypothesised that at this ratio the impact of the oxidation of SbIII to SbIV or SbV (with the accompanying increased ketone ratio) is minimised.

Figure 4.16 - Reaction of SbCl3 and TPTP monitored by direct GCMS sampling.

As a result of our experiments with SbCl3, the previously described studies with

1 TiCl4 utilised H NMR instead of GCMS to permit more direct and quantitative observation of the reaction mixture. The TPTP/SbCl3 reaction proved to be a difficult mechanism to analyse, largely as the potential for oxidation of the metal centre simultaneous to a TiCl4-like degradation considerably complicated quantitative analysis of the reaction products. This, together with the significantly slower rate of reaction,

1 meant that the SbCl3 reaction was not re-examined by H NMR with the same level of rigour as TiCl4.

129 4.5 Conclusion

The reactions of TiCl4 and SbCl3 described in this chapter present a new direction in the search for a rapid means of degrading cyclic peroxides. Although the anhydrous conditions required for these reactions makes their use for neutralisation of cyclic peroxides in the field (or even the laboratory) limited, this does not diminish their significance. The effective degradation of cyclic peroxide solutions by a double molar equivalent of TiCl4 within 5 minutes at room temperature proceeds by a novel mechanism that warrants further investigation.

A number of key characteristics of the reaction present useful advances in terms of peroxide degradation. Most importantly, the reaction products cannot re-form peroxides. The sequestering of peroxide oxygen atoms into the organic products (esters) is a significant step forward from acidic degradation, which in itself is a reversal of the synthesis equilibrium for these peroxides. The analysis in Chapter 3 showed that most if not all methods described in the literature for neutralising cyclic peroxides rely on

Brønsted acid-catalysed ring opening in the first instance. The use of Lewis acids opens a whole new range of possibilities due to the ability of these metal compounds to effectively induce a degree of ‘self-oxidation’, where 2 peroxidic oxygens are retained within the alkyl moieties of TPTP in a stable ester form. Unlike conventional acid- catalysis which triggers an equilibrium between various peroxides, TiCl4 activates an intramolecular rearrangement to break down peroxide bonds, negating the explosive threat. TiCl4’s extreme sensitivity toward water and its exothermic binding with reaction products make it an unlikely field neutralisation agent, yet the observation that

TiA continues to react with TPTP demonstrates that other, more stable Ti complexes may still maintain reactivity toward cyclic peroxides. Such complexes may be more suitable to be cycled back to an active state, providing the tantalising possibility of

130 creating a catalytic reaction system in which a Ti compound facilitates the degradation of multiple molar excesses of cyclic peroxide. This potential is discussed further in

Chapter 6.

Finally, the evidence of a highly active chlorinating species which mediates facile chlorination side-reactions on carbonyls at room temperature makes this newly identified reaction of more general interest in chemical synthesis development. The use of organic peroxides rather than the hydrogen peroxide (as used in the cited tetrachloridosilane reaction34) avoids in situ generation of hydrogen chloride, thereby possibly avoiding unwanted side-reactions. The isolation and/or further characterisation of the reaction intermediates described in this exploratory work has the potential to yield useful synthetic methodologies.

4.6 Acknowledgements

This research was financially supported by the Commonwealth of Australia through the National Security Science and Technology Centre within the Defence

Science and Technology Organisation. This support does not represent an endorsement of the contents or conclusions of the research. This research was also supported by the

Australian Defence Force’s Chief of Defence Force Fellowship.

4.7 References

1. Furuya, Y.; Ogata, Y., Bull. Chem. Soc. Jpn. 1963, 36 (4), 419-422. 2. Bellamy, A. J., Journal of Forensic Sciences 1999, 44 (3), 603-608. 3. Oxley, J. C.; Smith, J. L.; Huang, J. R.; Luo, W., Journal of Forensic Sciences 2009, 54 (5), 1029-1033. 4. Armitt, D.; Zimmermann, P.; Ellis-Steinborner, S., Rapid Commun. Mass Spectrom. 2008, 22 (7), 950-958. 5. Oxley, J. S., J. L.; Brady, J. E.; Steinkamp, L., Propellants Explosives Pyrotechnics 2014, 39 (2), 289-298.

131 6. Tsaplev, Y. B., Kinetics and Catalysis 2012, 53 (5), 521-524. 7. Pachman, J.; Matyas, R., Forensic Science International 2011, 207 (1-3), 212- 214. 8. Bali, M. S.; Armitt, D.; Wallace, L.; Day, A. I., Journal of Forensic Sciences 2014, 59 (4), 936-942. 9. Fidler, F. L., T.; Carvalho-Knighton, K.; Geiger, C.L.; Sigman, M.E.; Clausen, C. A., Degradation of TNT, RDX, and TATP using Microscale Mechanically Alloyed Bimetals. In Environmental Applications of Nanoscale and Microscale Reactive Metal Particles, Geiger, C. L., et al, Ed. American Chemical Society: Washington DC, 2009; pp 117-134. 10. Apblett, A. W.; Kiran, B. P.; Malka, S.; Materer, N. F.; Piquette, A., Ceram. Trans. 2006, 172, 29-35. 11. Dubnikova, F.; Kosloff, R.; Zeiri, Y.; Karpas, Z., J. Phys. Chem. A 2002, 106 (19), 4951-4956. 12. Dubnikova, F.; Kosloff, R.; Oxley, J. C.; Smith, J. L.; Zeiri, Y., J. Phys. Chem. A 2011, 115 (38), 10565-10575. 13. Armarego, W. L. F.; Chai, C. L. L., Purification of laboratory chemicals. 6th ed.; Butterworth-Heinemann: Oxford, 2009; p xvi, 743 p. 14. Sreekumar, N. V.; Bhat, N. G.; Narayana, B.; Nazareth, R. A.; Hegde, P.; Manjunatha, B. R., Microchim. Acta 2003, 141 (1-2), 29-33. 15. Perik, M. M. A.; Oranje, P. J. D., Anal. Chim. Acta 1974, 73 (2), 402-404. 16. Christopher, D. H.; West, T. S., Talanta 1966, 13 (3), 507-513. 17. Bose, A. K.; Srinivas.Pr; Trainor, G., J. Am. Chem. Soc. 1974, 96 (11), 3670- 3671. 18. Corcoran, R. C.; Ma, J. N., J. Am. Chem. Soc. 1992, 114 (12), 4536-4542. 19. Hwang, T. Y.; Cho, J. Y.; Jiang, M. K.; Gau, H. M., Inorg. Chim. Acta 2000, 303 (2), 190-198. 20. Westmoreland, T. D., Inorg. Chim. Acta 2008, 361 (4), 1187-1191. 21. Fay, R. C.; Lindmark, A. F., J. Am. Chem. Soc. 1983, 105 (8), 2118-2127. 22. No reference found, reasonable match with calculated spectra (Scifinder Scholar, ChemBioDraw ChemNMR) 23. Cocivera, M.; Effio, A., J. Org. Chem. 1980, 45 (3), 415-420.

132 24. Marigo, M.; Bachmann, S.; Halland, N.; Braunton, A.; Jorgensen, K. A., Angewandte Chemie-International Edition 2004, 43 (41), 5507-5510. 25. AIST Spectral Database for Organic Compounds, National Institute of Advanced Industrial Science and Technology http://sdbs.db.aist.go.jp/sdbs/cgi-bin/cre_index.cgi, accessed 10/02/2014. 26. Doeuff, S.; Dromzee, Y.; Taulelle, F.; Sanchez, C., Inorg. Chem. 1989, 28 (25), 4439-4445. 27. Birnie, D. P.; Bendzko, N. J., Mater. Chem. Phys. 1999, 59 (1), 26-35. 28. Oxley, J. C.; Smith, J. L.; Bowden, P. R.; Rettinger, R. C., Propellants Explosives Pyrotechnics 2013, 38 (2), 244-254. 29. Weber, R.; Susz, B. P., Helv. Chim. Acta 1967, 50 (8), 2226-&. 30. Marcos, B. A., J.; Devin, C., Journal Chemical Research (S) 1977, (7), 166-7. 31. Cavallo, L.; Del Piero, S.; Ducere, J. M.; Fedele, R.; Melchior, A.; Morini, G.; Piemontesi, F.; Tolazzi, M., J. Phys. Chem. C 2007, 111 (11), 4412-4419. 32. Friess, S. L., J. Am. Chem. Soc. 1949, 71 (1), 14-15. 33. Cavallo, L.; Jacobsen, H., Inorg. Chem. 2004, 43 (6), 2175-2182. 34. El-Ahl, A. A. S.; Elbeheery, A. H.; Amer, F. A., Synth. Commun. 2011, 41 (10), 1508-1513. 35. Raoot, S.; Desikan, N. R.; Shekhar, R.; Arunachalam, J., Appl. Spectrosc. 2000, 54 (9), 1412-1415. 36. Hedberg, K.; Badger, R. M., J. Chem. Phys. 1951, 19 (4), 508-509. 37. Abbaspour, A.; Najafi, M.; Kamyabi, M. A., Anal. Chim. Acta 2004, 505 (2), 301- 305.

133 CHAPTER 5. Generalisation of the TiCl4

reaction with other organic peroxides

5.1 Introduction

In Chapter 4, a new mechanism of degradation for cyclic organic peroxides has been identified where Lewis acids have been shown to activate the rapid degradation of tripentanone triperoxide (TPTP). The identified rearrangement mechanism provides some notable advantages over existing approaches by limiting the need for oxidisable substrates and achieving fast degradation. There are limitations to practical application of this method, as the reagents studied require anhydrous conditions. Hence further research is necessary to capitalise on the strengths of this newly identified mechanism to neutralise cyclic peroxides, and find solutions to the current weaknesses.

A critical first step in this research is to identify how this reaction with our model compound TPTP translates to the actual organic peroxides utilised as home-made explosives. The investigations in Chapter 2 showed that differences in reactivity in acid- catalysed degradation can arise between even the closely related TATP and TPTP, hence it is important to confirm the applicability of our findings more broadly amongst organic peroxides, especially those most commonly utilised by terrorists due to precursor availability.

This comparative study also provides an opportunity to test the mechanism that was developed in Chapter 4 by observing the structure-activity relationships. This can be achieved with the convenient series of DADP, TATP, MEKP and TPTP – methyl- methyl (dimer and trimer), methyl-ethyl (trimer and hydroperoxides), and ethyl-ethyl

134 (trimer) (Figure 5.1). TATP presents the most difficult target, as methyls are difficult substrates for many types of chemistry. Baeyer-Villiger reactions with methyl migrating groups are practically unknown. Considering Baeyer-Villiger type rearrangement was identified in the degradation of TPTP by Lewis acids, the analysis of the TATP-TiCl4 reaction may provide some enlightening contrasts.

The presence of hydroperoxides alongside cyclic trimers in MEKP mixtures provides another valuable difference which may provide information about the reaction mechanism. The synthesis of MEKP results in complex mixtures of cyclic trimer and a large array of open-chain hydroperoxides.1 The unsymmetrical monomer further splits the cyclic compound into syn/anti forms, and the hydroperoxide oligomers into a myriad of stereoisomers. This array of products with very similar solvation properties makes it difficult to isolate one form of MEKP by simple crystallisation (the usual means of purifying TATP, TPTP, and even HMTD).1,* Whilst separation of oligomers

(cyclic trimer from hydroperoxides, open-chain dimer from open-chain trimer, etc) has been achieved using chromatography,1 separating oligomeric stereoisomers can be extremely difficult.

It is important to note at this point that whilst the purification challenges of MEKP make it a nuisance for chemists, these issues do not deter a would-be terrorist from utilising a crude mixture. All the isomers of MEKP are extremely sensitive and explode with high brisance,1-2 and hence are ‘useful’ products from this perspective. The clear, oily liquid of the crude mixture is dense, easily portable, and potentially less likely to arouse suspicion than solid materials. The threat posed by these materials is reflected in post-9/11 crackdowns on the carriage of Liquids, Aerosols and Gels (LAGs) onboard

* Milas did have some success crystallising the dimeric dihydroperoxide at -70C from pentane, which then needed further treatment by sublimation under vacuum to remove traces of the cyclic trimer. Compared to the ready crystallisation of TATP at room temperature from a variety of solvents, MEKP requires a much more careful approach and access to low temperature apparatus or dry ice. 135 aircraft. A general chemical degradation system for OPEs must be able to accommodate such mixtures.

HMTD was also assessed to identify if reactivity is maintained with its radically different N-based structure. The molecular structures of the peroxides considered in this chapter are provided in Figure 5.1.

Figure 5.1 - Structures of organic peroxides studies in Chapter 5. 5.2 Research Goals

The aim of this part of the research was to understand how the novel Lewis acid- mediate rearrangement reaction translates from TPTP to threat-relevant organic peroxides. To establish this, the following key goals were pursued:

1. To confirm if TiCl4 (as a model Lewis acid) reacts more broadly with organic

peroxides;

2. To investigate the reaction of TiCl4 with TATP, MEKP and HMTD under the

same controlled conditions used for TPTP; and

3. Use the observations to build on the mechanism proposed in Chapter 4.

136 5.3 Experimental

CAUTION! The organic peroxides detailed in this study have the characteristics of primary explosives with extreme levels of friction, heat and impact sensitiveness. All procedures must be carried out by properly qualified and equipped personnel taking all relevant precautions.

5.3.1 Materials

All chemicals and solvents were of analytical grade unless otherwise noted.

CDCl3 (99.9% D, Cambridge Isotopes) was dried by distillation from P2O5 under N2.

KBr was FT-IR grade from Sigma-Aldrich. TATP, MEKP and HMTD were supplied as

0.02 M DCM solutions by Dr Phil Davies and Dr David Armitt of DSTO Edinburgh, with the purity of these samples checked by GCMS and NMR before use. TiCl4 was

3 double distilled from copper turnings according to literature procedure under dry N2 gas, before being weighed and dissolved in a measured volume of DCM (GC Grade,

Sigma-Aldich, freshly distilled from CaH2 under N2). TiCl4 solutions were used and stored under dry N2 atmosphere, and discarded after 2 weeks.

5.3.2 Instruments

A Shimadzu QP 2010 Ultra fitted with a SGE SolGel Wax column (1.0 µm film,

I.D. 0.25 mm, 30 m) was used for GCMS analyses. The method utilised an initial column temperature of 40°C for 1 min, 20°C/min ramp to 100°C, held for 3 mins, followed by 30°C/min finishing at 190°C and held for 2 mins. Injector temperature was

130°C, interface 200°C, and linear column flow at 2 mL/min of He gas. Split ratio was set at 40. Concentrations were quantified using external standards.

NMR spectra were recorded on a Varian Unityplus-400 spectrometer. All NMR experiments were conducted at 25°C unless otherwise stated. 1H NMR spectra were

137 referenced to tetramethylsilane (0 ppm) at 25°C using the residual 1H signal of the respective deuterated solvent. 1D spectra were recorded with between 1 and 16 transients with a 6 second relaxation delay. COSY experiments were conducted with 1 scans per increment, 512 t1 increments and 2 s relaxation delay. Integration data was processed and graphed using Graphpad Prism.4

IR Spectra were recorded on a Shimadzu IRPrestige-21 spectrometer. Samples were analysed between 350-6000 cm-1 with 16 scans at 4 cm-1 resolution in a KBr disc.

5.3.3 Methods

All weighing and handling was conducted in a dry N2 atmosphere using a glovebox or standard Schlenk techniques with flame dried glassware (cooled under N2 flow) and oven dried syringes unless otherwise stated. Any trace of moisture in the presence of TiCl4 results in immediate cloudiness – any experiments where this was observed during preparation were re-prepared.

5.3.3.1 Preparation and determination of TiCl4 stock solution

The estimated concentration of TiCl4 DCM solutions (from weight/volume calculations) was confirmed by a modification of published titration methods5-6 as described in section 4.3.3.1.

5.3.3.2 TiCl4 degradation experiments

Analyte (TATP, MEKP or HMTD) was measured into a dry, sealed 5 mL volumetric flask, and solvent removed dried under N2 flow at RT, then diluted with dry

CDCl3 to make a 0.04 M solution. [Peroxide] was quantified by standardising a 500 μL aliquot of solution against a solid sample of hexamethylbenzene. In the case of MEKP, which was a mixture, an average MW reflective of the distribution of 80% cyclic trimer

20% monomeric hydroperoxide was utilised for the purposes of calculations.

138 Appropriate volumes of 0.4 M TiCl4 in dry DCM stock solution was added to a flame dried NMR tube under N2 to achieve the required reagent ratios. To this was added 500

μL of TPTP/CDCl3 solution in NMR tubes, and the mixture immediately frozen in liquid N2. Reactions of TATP with equivalent or sub-stoichiometric ratios of TiCl4 had to be allowed to react for ~10 seconds with shaking whilst open to N2 atmosphere before freezing to avoid NMR tubes cracking upon thawing due to pressure generated by gas formation in the reaction. Once the sample was frozen, the tube was flame sealed under high vacuum and stored frozen till analysis. The sample was thawed and shaken well and the stopwatch started. These samples were analysed as soon as practicable

(allowing for loading/shimming) by 1H NMR.

For these reactions, the residual solvent signal (previously quantified against a solid aliquot of hexamethylbenzene by 1H NMR) was used as an internal standard for quantification of integration, although in many cases accurate quantitation was not possible due to volatile or unidentified products.

5.4 Results and discussion

Of the Lewis acids considered in Chapter 4, TiCl4 was selected for investigation with other organic peroxides. This was based on the higher activity toward TPTP

3+ compared to SbCl3, and the fact that Sb complexes are susceptible to oxidation of the metal centre which can result in competing reactions. The investigations conducted with these organic peroxides were not conducted to the same level of detail as the TPTP reaction described in Chapter 4 as our intent was to extend the detailed work conducted on TPTP, rather than repeat it with new analytes. The limitations of working with small amounts of diluted peroxide samples from DSTO due to safety and security limitations also restricted our experimental options.

139 Despite these issues, comparison against TPTP’s reaction led to some further useful insights into the Lewis acid-cyclic peroxide reaction mechanism. Definitive spectral evidence for Ti-OCl species was not found in Chapter 4, however the proposed mechanism relies on the critical role of such a moiety to rationalise the observed organic reaction. To test the proposed mechanism, it has been applied to the observations of the other ketone-peroxides. The reliance on Ti-OCl species in the following discussion is not intended to presuppose its conclusive identification, but rather to test the suggested mechanism’s ability to describe the reaction with these related peroxides.

5.4.1 The reaction of acetone peroxides with TiCl4

Of the organic peroxide explosives considered in this chapter, TATP bears the most similarity to TPTP and thus presents a suitable place to begin investigating the broader applicability of the Lewis acid reaction. The 1H NMR spectrum of TATP and its dimer homologue diacetone diperoxide (DADP) is included in Figure 5.2. Common features of TPTP and TATP include having symmetrical alkyl groups, three gem- peroxide moieties, similar solution structure, and comparable underlying mechanisms of acidic and thermal degradation (discussed in detail in chapter 2). Straightforward purification allows the effective removal of dimer and hydroperoxide impurities, thereby avoiding any complications in analysis caused by competing reactions in degradation experiments. Critically, the change from ethyl to methyl arms has the potential to affect the means by which the reaction proceeds, and may offer useful mechanistic clues.

140 Figure 5.2 - 1H NMR of TATP and DADP. Spectra taken in CDCl3 at 25°C. *- water in CDCl3.

5.4.1.1 The reaction of TATP with 3-12 equivalents of TiCl4

Initial reactions were conducted with 3-12 times molar excess of TiCl4 in order to minimise the complicated competing reactions seen in the Ti-limited TPTP reactions.

Using the capped and shaken NMR tube procedure, the immediate yellow colour was again visible upon mixing of the reagents. Analysis by 1H NMR within 3 mins of mixing confirmed the degradation was complete in the time taken to load and shim the sample. Comparison of spectra taken before and after quenching with D2O, and addition of authentic standard samples, confirmed that the principal products were the analogous ketone and ester expected based on the TPTP reaction; Ti-bound methyl acetate

(MeOAc) and acetone (AcMe). The Ti-bound ligands subjected to downfield shifts as shown in Figure 5.3. These deshielding effects are comparable with the maximum seen for 3-pentanone (3P) and ethyl propanoate (EP) in the degradation of TPTP, albeit

~0.05 ppm less in magnitude. The size of this shift varied at most 0.03 ppm between the

1:3, 1:6 and 1:12 reaction, which should result in Ti to L ratios of 1:1-4 respectively

(where L = AcMe or MeOAc). This indicates that 1:1 Ti:L complexes dominate even in the presence of excess Ti centres. The chemical shifts of free AcMe and MeOAc after quenching match their respective literature values to within 0.03 ppm,7 confirming the products.

141 1 Figure 5.3 – H NMR spectra of the reaction of 1:6 TATP and TiCl4. The spectra show after reaction (Bound) and after quenching with D2O (Quenched). The minor peak at ~3.00 ppm is assigned as chloromethane.

Ratio X:ester with increasing TiCl4

1 .5 X = AcMe (TATP Reaction) X = CM (TATP Reaction) X = 3P (TPTP Reaction) 1 .0

0 .5 ratio X:ester

0 .0 0 3 6 9 1 2

ra tio T iC l4 per TATP/TPTP

Figure 5.4 - Ratios of products to ester in TATP/TPTP reaction with increasing TiCl4. CM = Chloromethane, the TPTP reaction equivalent (chloroethane, not shown) was only seen in the 1:0.5 reaction in trace levels. The dashed line shows the equivalent ratio of ketone to ester in the TPTP reaction. Solid lines show linear trends fitted for to the data. Error bar is indicative of SD, only visible at this scale on 1:3 reactions which showed greater variability.

142 The most significant difference to TPTP in the reaction products is a notable increase in ketone relative to ester. The ratio varied slightly depending on the amount of

TiCl4 added, with increasing excess of TiCl4 producing ratios closer to that seen in

TPTP (Figure 5.4).

The general trend towards 2:1 ester to ketone at higher TiCl4 ratios is of note as it suggests that at very high relative TiCl4 concentration, the initial TATP reaction may in fact produce the same 2:1 ester to ketone ratio seen with TPTP. This suggests the following analogous initial reaction:

TATP + TiCl4 → 2 MeOAc + AcMe + TiCl3(OCl) R1TATP

The fact that this ester to ketone ratio is only approached and never fully achieved at the ratios studied may suggests a second reaction (R2TATP), which produces largely acetone unlike the equivalent R2TPTP which continues to produce ester in significant quantities. This is explored in more detail in the next section.

Other key observations include a small peak at 2.98 ppm which is assigned as chloromethane (CM) based on previously reported spectra in the same solvent.7 No authentic sample was available to confirm this assignment. This peak consistently appears at a ratio of ~0.1 to MeOAc along with a very broad peak of similar magnitude at 0.88 - 0.92 ppm regardless of the TiCl4 ratio used (this second species is discussed later). The presence of CM is analogous to chloroethane seen in the final stages of

8 TPTP degradation with 0.5 equivalent of TiCl4. As CM boils at -24°C, consideration was given to quantifying the volume of gas being produced, but to gain a measurable volume would have required significant quantities of TATP. As the solubility of CM in

143 CDCl3 under these pressurised, room temperature conditions has not been quantified, no quantitative link can be made between the changing ketone/ester ratio and [CM].

5.4.1.2 The reaction of TATP with 0.5-2 equivalents of TiCl4 by FTIR

The R2TATP reaction was investigated in more detail by reducing the TATP-TiCl4 ratio below 1:3. The immediately visible impact of the reduced TiCl4 ratios was the precipitation of an off-white solid in the 1:0.5, 1:1 and 1:2 reactions. This precipitate was visible within seconds for the 1:0.5 sample, and took ~5 min and ~30 min to be visible for the 1:1 and 1:2 samples respectively. In all cases, the precipitate continued to gradually build as the reaction progressed. In general, precipitation occurred at much higher initial relative TiCl4 concentrations than seen in the TPTP experiments, where it was only seen in sub-equivalent TiCl4 reactions. The solid from the 1:0.5 reaction was isolated, dried and analysed by FT-IR. Comparison of this data with the Ti-propanoate isolated from the TPTP reaction (Figure 5.5) shows many similar features including strong absorption at ~3300 cm-1, ~1620 cm-1 and ~350-900 cm-1, consistent with this

TATP product being the equivalent acetate. The carbonyl stretch at 1615 cm-1 is significantly red-shifted from the 1714 cm-1 reported for free acetic acid.7 This shift is broadly in-line with other Ti-induced carbonyl shifts seen in Section 4.4.1.4, and correlates well with literature values for other acetates such as LiOAc7 which has its equivalent stretch at 1582 cm-1. A potential O-Cl stretch is visible at 760 cm-1, but remains a speculative assignment.

144 Figure 5.5 - FTIR spectra of TPTP and TATP precipitates after reaction with TiCl4.

* Adding a drop of D2O to the precipitate of the 1:1 reaction and suspending in

1 MeCN-d3 showed only traces of acetic acid by H NMR, whereas propanoic acid was the dominant product in the equivalent TPTP experiment in Section 4.4.1.4. Instead, a

* A significant proportion of precipitate did not dissolve. 145 strong singlet at 2.54 ppm was the major signal apart from the solvent, for which the most plausible assignment (given the sample’s preparation) is a Ti-AcO species. The

~0.58 ppm downfield shift from the reported methyl signal of free acetic acid7 is in line with maximum downfield shift exerted by TiCl4 on 3P and EP (0.56 and 0.60 ppm respectively). The presence of only trace amounts of free acetic acid might indicate that the Ti-AcO bond is stronger than the propanoate equivalent, possibly making it comparatively unreactive toward water.

5.4.1.3 The reaction of TATP with 1-3 equivalents of TiCl4 by 1H NMR

Some key observations made in the conduct of the sub-1:3 experiments must be highlighted before discussing the spectral results themselves. Gas production in these reactions significantly increased as the relative amount of Ti was reduced, an observation which manifested in 1:1 reactions by the constant shattering of the flame- sealed NMR samples upon thawing of the sample. This was avoided by allowing the sample to vent on the N2 manifold for 10-20 seconds after mixing the reagents, but prior to freezing in liquid N2. It appears the NMR tube was being shattered due to excessive internal pressures generated largely at the very early stages of the reaction post-thaw.*

Consequently a significant portion of organic material was being lost through gas generation, making quantitative comparisons between the samples questionable.

The largely methyl-based organic products hindered conclusive assignment of species from 1H NMR data alone due to the lack of coupling information. Alternative methods to confirm the assignments were difficult to apply due to the requirement for anhydrous conditions. Solution FTIR, used for the analysis of the TPTP reaction, was

* At least 5 consecutive samples shattered in this manner before the reason was identified and the method changed, after which no further samples were lost. All tubes were of the same batch, and apart from delaying the freezing process slightly no other part of the method was changed. The cracked tubes were most likely caused by internal pressure buildup (possibly localised in small pockets during thawing), rather than any glass defect or handling error. 146 considered unlikely to be of use due to the lack of suitable reference spectra to compare against for some of the unusual analytes encountered in this chemistry, a problem experienced with the TPTP reactions. The process of quenching to allow analysis by

GCMS was considered, but the likelihood of unleashing a multitude of hydration- related reactions meant that this approach was judged unhelpful, and would potentially raise more questions than it would answer. Pleasingly, the tentative assignments using the NMR shift data available gave a consistent account of the reaction progression, which in itself adds a degree of confidence in their accuracy.

Figure 5.6 illustrates the progression of 1:1 and 1:2 reactions; the 1:0.5 reaction is discussed separately. As the rate of reaction varied between the samples, the time intervals in Figure 5.6 were selected to highlight key species and trends. Methyl acetate

(MeOAc) and acetone (AcMe) were still the principal products, although ketone was now in significant excess to ester, a major change from all TiCl4/peroxide reactions discussed to this point. Another important feature is the appearance of broad singlets at

1.35 and 1.80 ppm (marked D), which match the chemical shift of an authentic sample of DADP. These signals were not detected at the higher TiCl4 ratios, and DPDP was not detected in any TPTP-Lewis acid degradation experiments. These peaks each have 2 sharper singlets slightly upfield (most visible in 1:1) which become relevant and are discussed later. Chloroacetone (CA) is also generated as evidenced by singlets at 4.08 and 2.30 ppm, although in quite low quantities.

147 1:2

1:1

Figure 5.6 - Reaction of 1:1 and 1:2 TATP to TiCl4 over time. Labels and references: ■ – chloroacetone (CA),7 ○ – acetone (AcMe),7 ◘ - Chloromethane (CM),7 □ - methyl acetate (MeOAc),7 D - DADP9 and T – TATP.9 * denotes Ti-bound species with associated downfield shifts.

The appearance of a peak possibly consistent with free acetic acid7 at ~1.90 ppm seems odd, considering that any available AcO- would be expected to coordinate to Ti.

The presence of AcO- does, however, align well with the AcO-based precipitate indicated by IR and the presence of chloromethane, which together account for the fate of a TATP acetone-peroxide unit. Acetic acid’s OH peak (not shown in Figure 5.6) is usually found at 11.4 ppm, but is not seen in these spectra. Finally, it should also be noted that the peaks of AcMe and MeOAc were significantly broadened. This phenomenon is consistent with the TPTP reaction and was considered to be due to the

148 continuously changing ligands available to the Ti centres as the degradation progresses, resulting in dynamic behaviour averaged over the NMR timescale.

A qualitative way to view the reaction progression is provided by plotting the absolute peak areas (raw integral divided by protons per signal) of the principle non- volatile products. In these reactions, AcMe has become the primary product (Figure

5.7), continuing the trend toward ketone products from the 1:12 to 1:3 reactions.

1 : 2 1 : 1

2 .0 2 .0 TATP M e O A C M e O A c 1 .5 1 .5 A c M e A c M e TATP DADP 1 .0 1DADP .0 chloroacetone

0 .5 0 .5 Integral (arbitrary units) Integral (arbitrary units) 0 .0 0 .0 0 2 0 4 0 6 0 0 2 0 4 0 6 0 m in u te s m in u te s

Figure 5.7 - Change in relative amounts of key analytes over time for 1:1 and 1:2 TATP/TiCl4 reaction. Qualititative data drawn from absolute 1H NMR integrals divided by the number of protons represented by each signal. Lines do not signify a trend, and merely connect the data points for clarity.

Considering the high TiCl4 reactions trended toward the ideal 2:1 ester to ketone ratio of the TPTP reaction, it seems reasonable to suggest that one TiCl4 molecule can degrade a TATP molecule via a similar concerted rearrangement as R1TPTP. This then rationalises the observed trend toward ketone products if R1TATP is considered as an extremely fast and therefore diffusion-limited reaction i.e. TiCl4 in the mixed zone is exhausted before all TATP has reacted. Although increasing the ratio of TiCl4 does counter the diffusion-limitation to an extent, the remaining TPTP in the mixed zone undergoes a second reaction - (R2TATP).

Extending the mechanism proposed in Chapter 4, TiCl4 should react rapidly with

TATP via R1TATP to form TiCl4Lx (L = MeOAc, AcMe; x = 1-2), TiCl3(OCl) and

’ TiCl3(OCl)Lx. These Ti compounds are the equivalent of ‘TiA in Chapter 4, and

149 continue to react further with TATP. Whilst R2TPTP continued to produce ester in the

1:0.5-1:2 TPTP reactions (refer to Figs 4.9 & 4.10), R2TATP appeared to produce largely ketone at equivelant ratios. In the TPTP reactions, the two-stage reaction was largely identified by the difference in their reaction rates. The percentage of initial degradation taking place by R1TPTP (nTPTP at t0, approximated by extrapolating the measurable

R2TPTP kinetics to time=0) progressively increased with increasing initial TiCl4 ratio

(Figure 4.10). In the same way, the AcMe to MeOAc ratio for the TATP reaction trends toward 1:2 with increasing initial TiCl4 ratio (Figure 5.4). Reaction rate was the primary differentiator between the reactions R1/R2TPTP, whilst the R1/R2TATP difference was manifested by a distinct change in major products, as well as a rate difference from rapid to slow.

The increasing dominance of acetone (from R2TATP) in situations where the TiCl4 ratio drops below 3 suggests that TiCl4 can either only degrade a single acetone- peroxide monomer, and/or that the binding of the formed ketone/ester products form a less reactive species. The trend in Figure 5.4 suggests that well in excess of 12 times excess of TiCl4 is needed for return to complete R1TATP dominance (rather than a 3x excess being needed for R1TPTP dominance), indicating that both effects may be important. Each reaction with TPTP may deactivate four Ti centers through the coordination of L (where L = MeOAc or AcMe), or due to a chlorido ligand being oxidised to hypochlorido.

In Scheme 5.A, R1TATP proceeds as expected from our observations of TPTP, with the highlighted Baeyer-Villiger type migrations (*, in red) being driven by the strong Lewis acidity of TiCl4. This R1TPTP-like concerted mechanism is indicated by the trend toward 2:1 ester to ketone ratio in the presence of large TiCl4 excess.

Considering that methyl groups are known to make poor migratory groups for Baeyer-

150 Villiger rearrangements, TiCl4(L)x (x = 1-2) and TiCl3(OCl) may lack the necessary

Lewis acidity to drive the concerted mechanism required to form ester products in

R1TATP. Instead, the carbocation which had before driven rearrangement, now either reforms TATP (thus slowing the reaction), or undergoes a stepwise degradation generating a single ketone, DADP and a Ti-OCl moiety (Scheme 5.A, R2TATP). This forms a key distinction between the proposed TATP and TPTP mechanisms, as only

R1TATP generates ester products, whilst both R1TPTP and R2TPTP are observed to do so.

In this way R2TATP accounts for the change in product distribution as TiCl4 excess is reduced. The stepwise process can repeat via R2DADP(a) to produce dimethyl dioxirane

(DMDO), acetone and Ti-OCl. R2DMDO completes the degradation to acetone and Ti-

OCl. Whilst minor peaks are visible by 1H NMR around the region where DMDO is reported (1.67 ppm),10 it is highly unlikely that DMDO is observable in these reactions as such an unstable intermediate would be fleeting in the presence of such reactive Ti species.

151 Scheme 5.A - Proposed mechanism for degradation of TATP by TiCl4 and TiCl4Lx. L=MeOAc or AcMe, X = Cl or OCl. Red arrows marked with * involve migration of methyls. Note that free Cl- is shown in these mechanisms for clarity i.e. the steps may be concerted.

The carbocation intermediate in R2DADP could possibly generate a Ti-bound chlorinated peroxy species (R2TATP-b). This latter product is unknown in the literature, however as shown in Figure 5.8, the estimated 1H NMR spectrum is a very good match for the small peaks flanking DADP in the 1:1 and 1:2 TATP degradations, and more

152 prominent still in the 1:0.5 reaction. Whilst the NMR prediction software does not account for Ti-induced shifts, it should be noted that by this stage of the reaction, the downfield shifts are quite modest (~0.15 ppm for AcMe), presumably due to the increasing [OCl] from the stepwise degradation reducing the electron withdrawing nature of the Ti centres. Whilst this final assignment is speculative, the proposed stepwise R2TATP/DADP/DMDO mechanisms account for a number of supporting observations - not only the increased production of DADP and acetone in the 1:<3 reactions, but also for these previously unassigned peaks.

Figure 5.8 - Comparison of TATP/TiCl4 1:1 reaction with predicted 1H NMR spectrum for chloro-peroxy species. Upper spectrum highlights relevant peaks in the reaction mixture with shifts, integrals shown above x-axis. Labels ○ - AcMe,7 □ - MeOAc,7 D - DADP9 and T – TATP.9 * denotes Ti-bound species with associated downfield shifts. Lower spectrum shows predicted NMR spectrum (ChemBioDraw Ultra 12.011) of the organic moiety, shifts denoted on structure in ppm.

The mechanism for the stepwise degradation of TATP can be explained by the formation of Ti-OCl also suggested for the concerted rearrangement reaction. This process may be part of the driving force of the overall reaction of TiCl4 with cyclic

153 peroxides. Further experiments to study the reaction of TATP with TiCl4L2 (analogous to Section 4.4.1.6) may provide further insight into the role of ketone and ester in the change from R1TATP to R2TATP. These experiments, however, were outside the scope of the immediate investigation which focused on the generalisation of the novel rearrangement reaction to threat OPEs.

5.4.1.4 The reaction of DADP with 1-2 equivalents of TiCl4 by 1H NMR

Having accounted for the reduced ester ratio and the appearance of DADP, the role of DADP in these mechanisms was further investigated. DADP is only observed when the Ti ratio falls below 1:3, suggesting TiCl4 (in the absence of ketone or ester ligands) can still readily react with the dimer. To test this, 1:1 and 1:2 degradation reactions were also conducted with DADP. Figure 5.9 provides representative 1H NMR spectra of the reaction progression, whilst Figure 5.10 plots the overall reaction course.

1 Figure 5.9 - H NMR spectra of the reaction of 1:1 and 1:2 DADP with TiCl4. Labels and references: ○ - AcMe,7 ◘ - CM,7 □ - MeOAc,7 D – DADP,9 T - TATP9 and ? - unassigned. * denotes Ti-bound species with associated downfield shifts.

154 1 : 1 1 : 2

M e O A c M e O A c 6 0 6 0 DADP DADP A c e to n e A c M e

4 0 4 0

2 0 2 0 (arbitrary units (arbitrary units) Relative integral Relative integral

0 0 0 2 4 6 8 1 0 0 2 4 6 8 1 0 m in s m in s Figure 5.10 - Relative abundance of key analytes in the 1:1 and 1:2 reaction of TATP with TiCl4. Initial starting integral is quantitatively calculated, as described in experimental. Lines show single phase exponential decay fitting.

Once again the proportions of ketone and ester qualitatively point to a fast initial degradation via the hitherto demonstrated intra-molecular Baeyer-Villiger mechanism, which at the earliest time point gives the expected 1:1 ratio. Our observations appear to commence just as the re-arrangement reaction ceases to occur (2 mins) at around ~20% and 50% TATP degradation for the 1:1 and 1:2 reactions respectively. From this point,

AcMe becomes the primary reaction product, with only slight increases in MeOAc which quickly plateau. It again seems that the production of sufficient ligands to bind most Ti centres marks the cessation of rearrangement reactions. In the TPTP-TiCl4L2 reactions (section 4.4.1.6), it was found that ligand binding was not the primary cause of the shift to R2TPTP kinetics. Conversely, the observation in Figure 5.10 that the 1:2 reaction switches to R2DADP at ~50% DADP degradation suggests that in the acetone peroxide case, OCl and ester and perhaps even acetone binding are all sufficiently deactivating toward TiCl4 to drive a switch to R2 processes, with the associated change in reaction products. Reaction 1.1 illustrates how rapid exchange of L between Ti centres may deactivate multiple centres, even if there is not sufficient ligand to bind every Ti centre:

4TiCl4 + 2 DADP → DADP + TiCl3(OCl) + 3TiCl4 + MeOAc + AcMe (1.1) Rapid exchange

155 The degradation of DADP is rationalised with scheme 5.B. The equimolar distribution of AcMe and MeOAc produced in R1DADP match the results from the early stages of degradation, when free TiCl4 dominates. As Ti is deactivated by the generated

OCl, ketone and ester, the slower, stepwise production of acetone via R2DADP and

R2DMDO explains the dominance of ketone products. DMDO again is not likely to be observable by its expected singlet at 1.67 ppm due to its inherent reactivity.

Scheme 5.B- Proposed mechanism for the oxidation of acetone by TiCl3(OCl) forming acetato ligands. Coordinated ester or ketone shown as L for clarity. Free Cl- shown for clarity of mechanisms only.

5.4.1.5 The formation of chloromethane and acetate

Ti-AcO and CM form from the action of TiCl3(OCl) on AcMe, via the R1AcMe mechanism illustrated in Scheme 5.C.

156 R1AcMe

Scheme 5.C - Proposed mechanism for the oxidation of acetone by TiCl3(OCl) forming acetato ligands.

Scheme 5.C is based on the estimation that enolisation is less likely to occur in the

acetone-peroxide reactions due to the weaker oxophilicity of TiCl3(OCl) and the lower

relative susceptibility to enolisation of AcMe compared to 3P. Unlike Scheme 5.C., this

mechanism would only become prominent when [Ti-OCl] and [AcMe] is high,

conditions seen in the latter stages of reactions with low ratios of added TiCl4. Small

amounts of ethyl chloride seen in the TPTP reaction show this mechanism may also be

accessible in the pentanone regime, but to a lesser extent. Equally, TATP and DADP

degradation still produce CA via the same enolisation process described in section 4.4.2

for the chlorination of 3P, but to a reduced degree. In the presence of large excesses of

TiCl4, this reaction would be limited in extent as acetone would bind preferentially to

the more oxophilic TiCl4.

Considering the prominence of CM, its associated Ti-AcO species should be

visible as a significant peak via 1H NMR. The presence of acetate is also indicated by

the FTIR results. The only unassigned peak of a similar order of magnitude is the broad,

upfield peak around ~1ppm in the degradation assays. A methyl-titanium moiety was

considered, but the 1H NMR chemical shift does not match the literature value of 1.41

12 - ppm for MeTiCl3, and is mechanistically difficult to arrive at. AcO may provide a

more robust assignment for this peak. A detailed view of this peak from the TATP 1:1

reaction is provided in Figure 5.11. Similar peaks are observable in all TATP and

DADP reactions including samples where a large excess of TiCl4 was added, which

157 could be explained by the R1AcMe reaction occurring due to incomplete mixing in the early stages of the reaction.

1 Figure 5.11 - H NMR of possible acetato signal over time in the 1:1 reaction of TiCl4 and TATP. T – TATP, D – DADP, CDAPHP- chloroperoxy species (see Figure 5.8).

The broadness in Figure 5.11 indicates dynamic behaviour, possibly consistent with Ti interactions. So far, the binding of L (C=O species) to Ti has produced deshielding effects on the nearby protons, making the ~1 ppm upfield shift from the

7 usual 1.96 ppm of free acetic acid in CDCl3 appear inconsistent. In this case, however, the observation that the peak migrates toward the expected shift for the free acid over the course of the reaction (as [Ti] is reduced due to precipitation) suggests a Ti-induced screening effect may also be possible. The broadening that accompanied this shift suggests the conformation represented by the 0.9 ppm peak was in very fast exchange with another, less shielded species. The equilibrium between these species moved toward the downfield species as the reaction progressed. This could be explained

158 through a decentralisation of the carbonyl bond through resonance of acetate and a Ti centre, or two Ti centres, which access the AcO- anion illustrated in Scheme 5.D.

Scheme 5.D - Possible resonance mechanisms of the proposed Ti-acetato moiety providing shielding effects to methyl protons.

The resonance stabilisation of the intermediate structure may explain how the shielded anion could account for a large proportion of the species observed on the 1H

NMR timescale, hence the large upfield shift. The proposed screening effect due to the stabilisation of the AcO anion is a known phenomenon. Dillon found a 0.19 ppm upfield shift in D2O solutions of sodium acetate when compared to the free acid resonance.13 A thorough search of the literature did not locate any description of such an extreme upfield shift for AcO, which is not surprising considering the very unique chemistry taking place under the described conditions. Such a large screening effect may suggest the formation of a stable complex, which may be a significant thermodynamic driving force in the formation of AcO- and CM from acetone via

- R3TATP/DADP/DMDO or R1AcMe. The strong attraction between AcO and positively charged Ti complexes suggested in our explanation for strong shielding effects may also drive the early precipitation of the Ti product.

159 Further indirect support for the presence of AcO- is provided by the FTIR evidence for this species in the precipitate, although the precipitate is not likely to share the resonance-stabilised structures described in Scheme 5.E. It was shown that after

1 exposure to D2O, the precipitate shows a strong H NMR signal in the region where one would expect to find a Ti-bound AcO-, though in this case with a large 0.58 ppm downfield shift from the free acid (Section 5.4.1.2). Under these conditions, the Ti centres would be rapidly hydrolysed which would disrupt the AcO- ion stabilisation.

Thus the downfield shift seen in the quenched NMR sample can be reconciled with the upfield shift seen in the anhydrous sample, as in the former the usual de-shielding effect seen with Ti-bound alkoxides is restored, although the suggested resistance of hydrolysis remains unconfirmed.

5.4.1.6 Acetone peroxide overview

Having provided assignments for the key species observed in our FTIR and NMR observations, and explained their presence through plausible mechanisms, a cohesive overview of the way in which acetone peroxides reaction with TiCl4 can now be proposed. The proposed reactions provide the building blocks to explain the complex interplay of multiple peroxide and titanium species, further complicated by ever increasing concentrations of multiple ligands which themselves change in relative concentration. To test the validity of the overall scheme, it is useful to assess how it addresses some of the more qualitative features of the overall reaction

The production of CM is a suitable test case for this purpose. CM was observed in every TATP and DADP reaction conducted in this study, regardless of the initial concentration of TiCl4 used. Although exact quantification of CM was not possible due to its volatility, samples with lower concentrations of TiCl4 seemed to generate a larger signal, and the 1:1 TATP/TiCl4 reaction appeared to produce sufficient pressure to crack

160 NMR tubes unless allowed to vent prior to freezing and sealing. Scheme 5.E shows the idealised, stepwise progression of this reaction at the described ratios. A four molecule reaction is used as a model to enable a discussion of how diffusion may affect the outcome in real reactions.

Scheme 5.E - Overall scheme of the 1:1 reaction of TATP and TiCl4. Bold text indicates dominant processes/products at each stage. Ti species are arranged with most reactive species on top (reactivity estimated based on qualitative observations). Organic species are arranged roughly from most abundant on top, and ratios indicated where relevant. Stages are indicated for guiding discussion only, many of these stages overlap.

In scheme 5.E, it is the oxophilicity of the available Ti species which dictates organic reactions available in each of the general stages of degradation. Stage 1 of the reaction involves the rapid generation of 2 MeOAc and 1 AcMe molecule via the

R1TATP rearrangement reaction producing only one Ti-OCl moiety in the process. The

BV rearrangement mechanism allows the oxygen to be retained in the organic compounds, rather than remain with Ti. A consistent but small amount of CM and AcO are also formed via R1AcMe due to the lack of efficient mixing. As TiCl4 is more oxophilic than TiCl3(OCl), most of the available ligands will bind preferentially to the former, limiting the extent of R1AcMe. The presence of ligands means that in Stage 2,

161 * the R2TATP/DADP/DMDO mechanisms dominate. Each TATP degraded in Stage 2 produces 3 Ti-OCl groups, quickly eliminating any remaining TiCl4L1-2. With all titanium centres now bearing at least one hypochlorite,† Stage 3 sees the dominance of

R1AcMe, which results in the production of CM gas and a Ti-O-Cl-AcMe precipitate.

The copious gas production to which was attributed the shattering of the NMR tubes in the 1:1 reaction can now be explained by considering how diffusion effects would be magnified in the thawing of a frozen sample. In the experiments, the reagents were injected rapidly to provide a level of initial mixing, and then immediately immersed in liquid N2. Figure 5.12 shows a schematic illustrating the samples frozen predominatly in Stage 1 (Figure 5.12-A).

Figure 5.12 - Schematic of the localised generation of CM in the 1:1 reaction of TATP and TiCl4 prepared by the free-seal-thaw method. A- Sample frozen in liquid N2. B- general and in-detail view of localised, self-heating reaction spots moving quickly to Phase 3. C- localised buildup of CM gas results in NMR tube being shattered whilst melting.

Upon thawing, areas at the glass will melt first, forming localised pockets of reaction continuing via R1TATP at the melt boundary (Figure 5.12-B) which generates

* Our experiments do not show if TiCl4(AcMe) can mediate R1TATP or R1DADP, but the rapid switch to R2DADP indicated in the DADP experiments suggest that it cannot. Conducting a range of TiCl4L2 (L= MeOAc or AcMe) degradation experiments with TATP and DADP would confirm this experimentally, however this small detail does not greatly affect the overall scheme and was not pursued at this time. † It is also unclear if TiCl3(OCl) can degrade TATP or DADP. Based on the observation that 1:1 reaction does not completely degrade TATP before precipitation, it would appear this reaction is not fast if it proceeds at all, or precipitation prevents it from occouring. 162 14 significant heat due to the binding of ligands to as-yet unreacted TiCl4. In these small, confined pockets of localised heating, Stage 2 can be expected to proceed quickly, rapidly generating the high ratios of acetone TiCl3(OCl) needed to trigger Stage 3, whilst further melting creates a Stage 1 “boundary layer” which continues to generate further heat, MeOAc and AcMe from R1TATP. Finally, the accumulated pressure of trapped CM gas generated by R1AcMe results in the shattering NMR tubes whilst much of the sample is still frozen (Figure 5.12-C). The Ti(OCl)-to-acetone ratio is particularly high in the 1:1 reaction due to the large proportion of degradation taking place via

R2TATP/DADP/R1DMDO, and hence the buildup of CM gas would be much higher than other reactions which did not exhibit this effect. Thus the proposed overall reaction scheme can be used not only to describe observations at the molecular level, but also at the macro level.

The defined reactions proposed here are not occurring in isolation, and in reality a complex equilibrium would exist between these mechanisms. The fast exchange of the

Ti-bound ligands means the transition from R1TATP to R2TATP would be blurred. The description of each of these mechanisms in isolation should be viewed as a necessary simplification to explain the highly interlinked reactions at work in the described experiments.

5.4.2 The reaction of MEKP with TiCl4

TATP’s methyl side-arms created some differences in the overall degradation with TiCl4 when compared against TPTP. These differences in reaction behaviour and products appear to be derived from the resistance of the methyl groups to Baeyer-

Villiger migration. The MEKP family of peroxides was next examined, which has both ethyl and methyl arms, bringing the expectation that its behaviour might be similar to

TPTP.

163 Reflective of the complexity seen in crude MEKP, the sample of MEKP (provided by DSTO) utilised in this pilot study was a mixture of cyclic peroxide and hydroperoxide. The GCMS chromatogram (Figure 5.13) showed twinning in the hydroperoxide peak (A) which could be taken to indicate D/L vs. meso isomers of the dihydroperoxide dimer, although the sample was thought to contain mainly cyclic trimer with a small fraction of dihydroperoxide monomer based on previous analysis undertaken at DSTO.15 A small leading peak to the main cyclic trimer peak (B) was thought to be the anti/syn- and pure syn-isomers respectively. The stock solution was assessed as approximately 80% cyclic trimer / 20% hydroperoxy monomer based on qualitative peak area, and the usual dominance of these species in the synthetic procedures used to make the provided sample of MEKP.15 Given the weak peroxide bonds and the EI ionisation used in our MS experiments, it is not surprising that no molecular ion could be observed, making definitive identification impossible using

GCMS. The same conclusion was also reached by the authors of the only other known detection study of MEKP.2

Figure 5.13 - Chromatogram and mass spectrum of the MEKP mixture used in this study. A- MEKP open chain dihydroperoxide monomer (2,2-Dihydroperoxybutane). B- MEKP cyclic trimer (1,4,7-trimethyl-1,4,7-triethyl-1,4,7-cyclononatriperoxane). Mass spectrum shown as counts vs. m/z.

164 Confirming this distribution by 1H NMR was no less challenging due the extreme complexity of the spectrum, again reflective of the mixture of isomers. The 1H NMR spectrum of the mixture (Figure 5.14) showed considerable overlap between isomers, particularly in the resonances of the ethyl arm protons (Figure 5.14, b and c). As a result, only very general assignment of peaks was possible. The relative peak size of the hydroperoxide protons (~9.7 ppm) compared to the alkyl peaks is broadly in line with the assignment of the majority species being the cyclic trimer.

Figure 5.14 - 1H NMR of MEKP mixture used in this study. General assignment is shown on major peaks, however assignment of individual isomers not possible due to overlapping peaks. Inset shows hydroperoxide peaks with magnified intensity for clarity. Minor peaks are severely overlapped, but are considered to reflect mainly the hydroperoxide monomer, and minor traces of other hydroperoxy polymers also indicated by GCMS.

As it was expected that MEKP would in general behave more like TPTP, experiments were conducted with 1:3 – 1:1 ratios of MEKP to TiCl4. An estimated average MW based on the approximate distribution of isomers in the mixture was used to calculate the concentration of the MEKP mixture, and MEKP to Ti ratio was based on [(C2H5)C(CH3)] rather than [O-O]. Due to the uncertainty inherent in dealing with this mixture, the control of molar ratio was less accurate than for the TATP and TPTP experiments. Accordingly, a less quantitative approach has been taken in analyzing the reaction products. Instead, a qualitative indication of how MEKP reacts with TiCl4 was sought.

165 1 Figure 5.15 - H NMR of the reaction of MEKP with TiCl4 at 1:1-3 ratios. Spectra taken at 5 mins after addition. □ - ethyl acetate (EA),7 ○ – methylethylketone (MEK),7 ◘ - chloroethane (CE),7 D – cyclic dimethylethylketone diperoxide (DMEKDP). * denotes Ti- bound species with associated downfield shifts. ■ – possible Ti-acetato, based on TATP discussion. Inset compares DMEKDP and DPDP (DPDP sample contains trace hydroperoxide impurities).

Figure 5.15 shows the spectra of these reactions 5 mins after mixing the reagents.

The 3:1 and 2:1 reaction exhibit no residual peroxide, with the major species peaks corresponding to the Ti-bound ethyl acetate (EA) and methyl ethyl ketone (MEK). The peaks assigned as the EA and MEK methylenes (4.52 ppm and 2.91 ppm respectively) are shifted ~0.41 ppm and ~0.48 ppm downfield of the analogous signals in free EA and

MEK. This shift is comparable to equivalent shifts for the TPTP reaction products. The

1:1 reaction exhibits broad peaks which, when compared against DPDP (Figure 5.15, inset), suggests that the cyclic dimer of dimethylethylketone diperoxide (DMEKDP) may be formed when Ti is slightly limited. The broad signals do not match perfectly to a report of DMEKDP’s 1H NMR spectrum,16 however this study utilised a different

166 solvent and was from the very early days of NMR development, which might have led to the divergent result. The formation of the cyclic dimer is reminiscent of DADP forming in the degradation of TATP once Ti-L complexes are formed. The suspected

DMEKDP signal continues to degrade over the following ~40 mins forming largely

MEK (not shown).

The presence of EA shows that the Baeyer-Villiger type rearrangement is again occurring with MEKP, as was expected. The ester:ketone ratio was quite constant across the reactions, consistently providing 1.1 excess of ester, falling slightly to equivalence in the 1:1 reaction. This deviation from the 2:1 ratio seen in the TPTP reaction is unexpected considering both contain an ethyl moiety. Part of the explanation for the higher proportion of ketone may be that the hydroperoxide in the sample produces only MEK. Assuming a 20% dimer hydroperoxide impurity producing all

MEK, and the cyclic trimer producing the 1:2 MEK to EA ratio, a ~1:1.7 excess of ester would still be expected. This suggests other factors may also be responsible for the higher than expected proportion of ketone products. Further experiments with purified cyclic trimer would be needed to confirm if this is the case.

The signals consistent with chloroethane and the proposed resonance-shielded acetato (Figure 5.15, ◘ and ■ respectively) in the early stages of the degradation reaction is a feature more similar to the TATP degradation than TPTP. Again it is unclear if this is related to higher levels of Ti-OCl due to hydroperoxides, potentially activating an oxacationic mechanism analogous to R3TATP.

It is possible that the mixture of isomers in the cyclic trimer may also interrupt the rearrangement reaction, as the Baeyer-Villiger mechanism is known to be stereo- specific.17 In situations where the methyl arm is situated in the migratory postion,

TATP-like mechanisms will dominate, whilst TPTP-like mechanisms may be available

167 where the ethyl falls in the migratory position. This effect could explain the higher than expected MEK ratio, as a proportion of the trimer may be undergoing stepwise, (purely ketone-producing) degradation rather than the ester-producing rearrangement reaction.

This can only be confirmed by comparing the reactions of the all syn cyclic trimer with the syn-anti cyclic trimer. The separation of these isomers has not been reported in the literature, and was beyond the scope of this investigation.

Some minor peaks in the reaction spectra remained unassigned, and could not be related to the species found in the reactions of TATP, DADP or TPTP. The sharp singlet

3.72 ppm is a very close match to the reported shift of dimethyl carbonate at 3.79 ppm,7 and a slight shielding may be argued on the same grounds as that for acetato.* The oxygen rich structure of carbonate suggest a carbonate species could be sourced from the hyrdoperoxide dimer, but as no mechanistic route could be found, this signal is still considered as unassigned.

The uncertainties in the exact species present in our stock solution meant only general comparisons could be made with the other ketone-based cyclic peroxides studied. Compared to the TiCl4/TPTP reaction rate, MEKP seems to react as fast or faster, as evidenced by the 1:1 reaction being >80% degraded at 5 min. This is perhaps reflective of the ease of rearrangement of the ethyl arms and/or the higher reactivity of the hydroperoxide chains compared to cyclic peroxides. Yet in terms of the species seen, MEKP resembles TATP, with indications of cyclic dimers, chloroalkanes, greater production of ketone and the possible AcO- species produced early in the reaction. It seems that the presence of the shared aceto moeity in the MEKP and TATP regimes is the key in this instance. The proposed strong interaction between Ti and AcO may drive

* Dimethyl carbonate has less alkyl character than MEK and EA, suggesting it would be a less competitive ligand for Ti. It’s Ti-induced resonance is likely to be much more complex than acetate, and as such the effect would be quite small in this case 168 part of the reaction in favour of this product, however further investigation is needed to confirm this proposed interaction.

The aim of this section was to identify if the general reactivity of Lewis acids

(specifically TiCl4) was maintained with mixtures of MEKP. The lack of peroxide signals in our 1:2 reaction spectrum at 5 minutes shows that the complexity of the mixture does not hinder fast, complete degradation. The presence of EA confirms the novel rearrangement mechanism is still active in the presence of hydroperoxides. The action of TiCl4 is equally effective at degrading pure cyclic peroxides and MEKP mixtures.

5.4.3 The reaction of HMTD with TiCl4

The tertiary amine moieties and unusual cage-like structure of HMTD brings an entirely different kind of chemistry to this study. As so few comparisons can be drawn to the alkyl peroxides considered thus far in this thesis, this study was limited to assessing the general reactivity of TiCl4 as a representative Lewis acid with HMTD.

The HMTD sample provided by DSTO showed a single peak in its 1H NMR spectrum at 4.79 ppm in CDCl3. This differs slightly from the 4 singlets reported by three previous studies due to helical isomers, however all of these studies had been conducted in DMSO.18-20 All three studies found the peaks centered around 4.70 ppm which is broadly comparable with our observations. The difference may reflect that unlike spectra recorded in DMSO, the inter-conversion between helical isomers is not visible on the NMR timescale in CDCl3.

A threefold excess of TiCl4 to HMTD in dry CDCl3 resulted in the immediate formation of a yellow suspension. Left sealed in a GCMS vial with no additional protection from the atmosphere, this suspension was unchanged for over 1 month at ambient conditions. Attempts to isolate this product by removing the solvent under a

169 stream of N2 resulted in the formation of an intractable brown gum unsuitable for IR analysis.

Analysis of this sample by 1H NMR within 5 mins of mixing showed no remaining HMTD, and a few broad singlets at 1.88, 4.66, 7.86 and 8.09 ppm. Given the broad nature of these peaks, and the immediate precipitation observed, these signals may indicate the formation of formato and chloroacetato ligands (through the addition of chlorinated solvent to formate), which could form resonance structures with Ti similar to those proposed for AcO- in the section 5.4.1.5. This is plausible as the isolated methylenes would be susceptible to oxidation by Ti-OCl, should it form under these conditions. The sample did not contain enough dissolved product to permit 13C NMR, so no further data was available.

Due to the difficulty in isolating the precipitate, the lack of structural information available by NMR, and the lack of related homologues for comparison, it is not clear how the mechanism of HMTD reaction with TiCl4 relates to the other OPEs studied.

HMTD demonstrated high reactivity with TiCl4, however further studies would be required to elucidate the products being formed. A useful approach to extend the knowledge of the TPTP reaction to HMTD might be to use 3P bound to TiCl4 to ‘trap’ hypochlorite as 2-chloro-3-pentanone. Should this indeed occur, the experiment would also serve to strengthen the argument in support of OCl formation. The presence of N in

HMTD may facilitate a completely different set of reactions, which are beyond the scope of this study. 15N NMR may be a useful tool for any future investigation which aims to properly characterise the products of this reaction.

5.5 Conclusion

The work presented in this chapter shows that the strong reactivity of TiCl4 and

TPTP is maintained across all the studied peroxides. Although HMTD’s mechanism 170 could not be investigated in detail, it seems that all the ketone-based peroxides share the same fundamental mechanism of reaction. Whilst the data were not able to provide quantifiable reaction rates, all the studied peroxides except DADP were completely degraded by a 3-fold excess of TiCl4 within 5 minutes.

In Chapter 4, it was shown that strong Lewis acids, particularly TiCl4, are able to mediate degradation of cyclic peroxides via a novel intra-molecular Baeyer-Villiger type rearrangement mechanism. This mechanism has also been found active for acetone peroxides which demonstrated that, remarkably, the rearrangement of methyls is also possible. Methyls are extremely poor migrating groups in Baeyer-Villiger reactions, making R1TATP/DADP highly unusual reactions. The requirement for an extremely oxophilic species to drive methyl migration is reflected in TATP’s rapid switch to a stepwise, ketone-producing reaction once ligands become available for coordination with Ti.

It appears that MEKP, including its hydroperoxide forms, can also be degraded by

Lewis acids. The mechanism could not be elucidated to the same level of detail as

TPTP, TATP and DADP due to the complexity of the MEKP mixture used in these experiments. Further degradation studies with purified MEKP hydroperoxides or cyclic trimers would be needed to understand to what extent intra-molecular rearrangement is occouring in this case. Cyclic MEKP in particular may yield an interesting insight into the stereo-aspects of this novel rearrangement, as it is possible that the isomeric distribution of the cyclic trimer could have an impact of the ratio of ester to ketone.

Fundamentally, however, rapid degradation has been demonstrated in these complex mixtures, extending the scope of Lewis acids as peroxide degradation reagents.

Finally, it has also been shown that HMTD is also able to be degraded by Lewis acids. Whilst the nature of the products remains unknown, this initial result indicates

171 that Lewis acids may indeed be able to provide a broad-spectrum approach to degradation of organic peroxides.

5.6 References

1. Milas, N. A.; Golubovic, A., J. Am. Chem. Soc. 1959, 81 (21), 5824-5826. 2. Delcourt, E.; Lefebvre, M. H., New Trends in Research of Energetic Materials, University of Pardubice: Czech Republic, 2012; pp 107-114. 3. Armarego, W. L. F.; Chai, C. L. L., Purification of laboratory chemicals. 6th ed.; Butterworth-Heinemann: Oxford, 2009; p xvi, 743 p. 4. Muzyka, A.; Bazna, A.; Semiletko, Y.; Yemelyanov, A. Prism, 6.01; Graphpad Software: San Diego, CA, 2012. 5. Sreekumar, N. V.; Bhat, N. G.; Narayana, B.; Nazareth, R. A.; Hegde, P.; Manjunatha, B. R., Microchim. Acta 2003, 141 (1-2), 29-33. 6. Perik, M. M. A.; Oranje, P. J. D., Anal. Chim. Acta 1974, 73 (2), 402-404. 7. AIST Spectral Database for Organic Compounds, National Institute of Advanced Industrial Science and Technology http://sdbs.db.aist.go.jp/sdbs/cgi-bin/cre_index.cgi, accessed 10/02/2014. 8. Haynes, W. M. "Physical Constants of Organic Compounds" in CRC Handbook of Chemistry and Physics, 94th Edition, CRC Press/Taylor and Francis, accessed Jan 2014. 9. Peña, A. J.; Pacheco-Londoño, L.; Figueroa, J.; Rivera-Montalvo, L. A.; Román- Velazquez, F. R.; Hernández-Rivera, S. P., Characterization and differentiation of high energy cyclic organic peroxides by GC/FT-IR, GC-MS, FT-IR and Raman Microscopy, in Sensors and Command, Control, Communications, and Intelligence (C31) Technologies for Homeland Security and Homeland Defense IV, Pts 1 and 2, Orlando FL, 2005, Carapezza, E., Ed. Proceedings of SPIE, 5778, pp 347-358. 10. Murray, R. W., Chemical Reviews 1989, 89 (5), 1187-1201. 11. ChemBioDrawUltra, 12.0.2.1076; PerkinElmer: Cambridge, MA, 2010. 12. Berger, S.; Bock, W.; Frenking, G.; Jonas, V.; Muller, F., J. Am. Chem. Soc. 1995, 117 (13), 3820-3829. 13. Dillon, K. B.; Harrison, M. R.; Rossotti, F. J. C., J. Magn. Reson. 1980, 39 (3), 499-508.

172 14. Cavallo, L.; Del Piero, S.; Ducere, J. M.; Fedele, R.; Melchior, A.; Morini, G.; Piemontesi, F.; Tolazzi, M., J. Phys. Chem. C 2007, 111 (11), 4412-4419. 15. Smith, M. E.; Wall, C.; Fitzgerald, M., Propellants Explosives Pyrotechnics 2012, 37 (3), 282-287. 16. Brune, H. A. W., K.; Hetz, W., Tetrahedron 1971, 27 (15), 3629-44. 17. Mislow, K.; Siegel, M., J. Am. Chem. Soc. 1952, 74 (4), 1060-1061. 18. Sulzle, D.; Klaeboe, P., Acta Chemica Scandinavica Series A-Physical and Inorganic Chemistry 1988, 42 (3), 165-170. 19. Guo, C. L.; Persons, J.; Harbison, G. S., Magn. Reson. Chem. 2006, 44 (9), 832- 837. 20. Edward, J. T.; Chubb, F. L.; Gilson, D. F. R.; Hynes, R. C.; Sauriol, F.; Wiesenthal, A., Can. J. Chem. 1999, 77 (5-6), 1057-1065.

173 CHAPTER 6. A catalytic degradation system

6.1 Introduction

A key theme of this thesis has been the goal of seeking catalytic methods for the degradation of organic peroxides. The potential advantages of such a system from an application perspective are significant: low logistic burden, fast degradation, controlled rate of reaction, and (with appropriate reaction selection) safe products. In this chapter, the design of such catalytic systems for the degradation of TATP (and preferably other organic peroxides) is investigated.

Catalysis is best approached by first having a clear understanding of the reaction to be accelerated. As the acid-catalysed reaction constitutes the bulk of the known literature on cyclic peroxide degradation, this discussion and investigation of a possible catalytic system begins with this approach. The mechanistic links to the novel Lewis acid degradation described in Chapters 4 and 5 are discussed in detail in the conclusion of this chapter.

6.1.1 An acid-mediated catalytic system

The literature has so far provided only one proven catalyst capable of opening the relatively inert peroxide ring structure, that being acid. Yet acid in itself cannot degrade the peroxide moiety; it can only facilitate its transition between organic substrates. As discussed in detail in Chapters 1 & 2, it has been shown that in the absence of other reagents susceptible to oxidation, acid degradation of TATP forms the equally dangerous DADP.1 Acid-catalysed ring opening does, however provide an effective means to overcome the initial hurdle of the peroxidic ring’s general inertness via the 174 formation of the more reactive hydroperoxides. If oxidisable substrates are also present, they can then degrade the hydroperoxide intermediates inherent to the acid equilibrium to generate non-peroxide products. Whilst there are examples that appear to use this approach in the literature (discussed in detail in Chapter 1), the best examples still retain weaknesses such as toxic products,2 slow reaction rates3 or the requirement for unwieldy quantities of reducing agent.4

6.1.2 Baeyer-Villiger degradation of peroxides.

In Chapter 2 of this work, it was shown that the hydroperoxides generated in the acidic breakdown of TPTP were active Baeyer-Villiger (BV) oxidants capable of reacting with linear ketones (3-pentanone, 3P) to produce esters (ethyl propanoate, EP)

(Reaction 6.1). That this reaction proceeds to a measurable extent at room temperature is remarkable, as linear ketones are not particularly susceptible to Bayer-Villiger rearrangement.5 The reaction was assisted by the presence of H+, which is known to catalyse BV reactions.5

+ 3P + ROOH H EP + ROH (6.1)

Additionally, the gem-peroxyhydroperoxide intermediates appear to have comparable reactivity to peroxytrifluoroacetic acid, one of the few known oxidants that can efficiently effect this reaction on primary R groups, though at much higher temperatures.6 With esters being relatively inert products (the most significant hazard is usually flammability), finding ways to further catalyse the BV side-reaction (beyond the extant acid catalysis) presents an attractive avenue of generating safe degradation products from the gem-peroxyhydroperoxide intermediates produced by the action of acid on cyclic peroxides (Scheme 6.A(A)). The reduction in [ROOH] would then disrupt the formation of the cyclic dimer usually formed under acid conditions (Scheme

6.A(B)). In the case of TPTP, an oxidisable substrate (3P) is conveniently yielded from

175 the acid-degradation of the parent cyclic peroxide. Practically, this means that only a solvent, BV catalyst and a small amount of acid would need to be applied, with no requirement for a stoichiometric reagent.

Scheme 6.A - The application of BV catalysis for safe reduction of hydroperoxide intermediates. Acid catalyses the ring-opening of the cyclic trimer to trigger an equilibrium of hydroperoxide polymers, ketones and hydrogen peroxides. Hydroperoxides where n = 2-3 are most crucial in disrupting the formation of the thermodynamically catalysed cyclic dimer via B, however all hydroperoxides and even HOOH may be active via A given a suitable substrate and BV catalyst.

Whilst the catalysis of BV reactions appears to be a straightforward solution to the degradation of TPTP, there are still challenges to its application to the actual threat materials. Such gem-peroxyhydroperoxides are not strong enough to oxidise acetone, which is why methyl acetate is not generally a reported product of acid degradation of

TATP. This is expected as methyl is the most resistant group to this form of rearrangement.5 It is possible, however, that in the presence of a suitable BV catalyst, activation of the ketone or hydroperoxide may allow the reaction to proceed. Another solution to the resistance of methyl groups to BV reactions would be to use a solvent that itself is readily susceptible to these reactions ie. a tertiary cyclic ketone. Whilst this presents a trade-off in terms of the larger bulk of catalyst plus substrate which would need to be applied to the OPE to effect degradation, facile BV oxidation would significantly enhance the overall rate of degradation. Furthermore, it may also make the

176 system more applicable across all explosive peroxides, providing they are able to generate hydroperoxides of sufficient oxidising strength in the presence of a ring- opening acid catalyst.

6.1.3 Catalysis of the Baeyer Villiger reaction.

As the BV reactions are a potential route to safer degradation products, the existing literature on their catalysis may provide methods to overcome the challenges in applying this mechanism to cyclic peroxides. The industrial importance of the BV reaction has driven research to overcome issues such as regulatory restrictions on conventional peracetic acid oxidants, and the broader move toward greener processes.

Much of this has focused on switching to a more atom-efficient oxidant such as hydrogen peroxide. Multiple approaches to catalysis have been proposed, with a review by ten Brink7 dividing them into 5 broad categories illustrated in Scheme 6.B:

1. Electrophilic activation of the substrate (ketone),

2. Electrophilic activation of the Criegee intermediate,

3. Nucleophilic activation of the Criegee intermediate,

4. Nucleophilic activation of hydrogen peroxide, and

5. Electrophilic activation of hydrogen peroxide.

Scheme 6.B - General mechanisms of Baeyer-Villiger catalysis. Adapted from the review of ten Brink.7 The Criegee intermediate shown to illustrate mechanism 2 & 3 is described in detail in section 1.4.1.

Whilst hydrogen peroxide is generated by the acidic degradation of cyclic peroxides,1, 8 organic hydroperoxides would be much more prevalent, particularly in the 177 early stages of the reaction. Additionally, the ability of gem-peroxyhydroperoxides to oxidise 3-pentanone (3P) to ethyl propanoate (EP) suggests they are much stronger oxidants than hydrogen peroxide. Thus it makes sense to select a catalytic method that also accommodates such hydroperoxide intermediates. Approaches 3, 4 and 5 were not considered optimal as the presence of alkyl moieties on the intended oxidants might reduce their effectiveness. As methods that rely on the activation of the ketone substrate (Scheme 6.B(1)) have been shown to provide versatile activity9 including with linear ketones,7 this group formed the focus of our study.

Activation of the ketone in BV reactions is traditionally achieved by the use of a

Brønsted acid, with protons activating the formation of an electrophilic carbon which facilitates attack by the hydroperoxide (Scheme 6.2-A).5 The formation of EP in the presence of sulfuric acid and TPTP (section 2.4.3) is a perfect example of such conditions. The same reaction’s low yield may be related to the availability of the

ROOH oxidant by formation of dimeric peroxides, which are a thermodynamic product under acid catalysis of ketone-peroxides.1, 7, 10 Lewis acids coordinate strongly with ketones, as was seen in Chapter 4 and 5, thus are ideal for the activation of BV substrates.

Reviews of BV catalysis have found that the Pt-based catalysts pioneered by Strukul gave the broadest activity of those catalysts which function by ketone-activation.7, 9, 11-12

The catalysts themselves have evolved from the initially reported mononuclear compound [(P-P)Pt(CF3)(CH2Cl2)]ClO4 (where P-P = a variety of diphosphine

12 ligands) to the more recent dinuclear [(P-P)Pt(μ-OH)]2 (BF4)2 complexes. The latter complexes were highly active in catalyzing the oxidation of cyclic ketones with up to

50% conversion in 3h at room temperature, and were also active in oxidising acyclic ketones.13 As similar Pt catalysts have been used as oxidation catalysts with tert-

178 butylhydroperoxide,14 it was anticipated that BV activity might be maintained with the hydroperoxide degradation products of cyclic peroxides. Accordingly, this class of catalysts was chosen to conduct a pilot investigation of the feasibility of catalyzing BV reactions as a means to safely reduce cyclic organic peroxides, via their hydroperoxide intermediates.

6.2 Research Goals

The aim of this part of the research was to determine if the catalysis of BV reactions may be a suitable method to reduce cyclic peroxides into safe, stable products.

To establish this, the following key goals were pursued:

1. To confirm if Lewis acids can catalyse the oxidation of a model substrate

using gem-peroxyhydroperoxide oxidants;

2. To investigate a method to generate gem-peroxyhydroperoxide oxidants from

TPTP in the presence of the catalyst and substrate; and

3. To comment on the feasibility of this approach as a nuetralisation method for

organic peroxides, and provide suggestions for future investigations.

6.3 Experimental

CAUTION! The organic peroxides detailed in this study have the characteristics of primary explosives with extreme levels of friction, heat and impact sensitiveness. All procedures must be carried out by properly qualified and equipped personnel taking all relevant precautions.

179 6.3.1 Materials

All chemicals and solvents were of analytical grade unless otherwise noted.

CDCl3 (99.9% D) was dried by distillation from P2O5. TPTP and DPPDHP were synthesised and purified as described in Chapter 2.

6.3.2 Instruments

A Shimadzu QP 2010 Ultra fitted with a SGE SolGel Wax column (1.0 µm film,

I.D. 0.25 mm, 30 m) was used for GCMS analyses. The method utilised an initial column temperature of 40°C for 1 min, 20°C/min ramp to 100°C, held for 3 mins, followed by 30°C/min finishing at 190°C and held for 2 mins. Injector temperature was

130°C, interface 200°C, and linear column flow at 2 mL/min of He gas. Split ratio was set at 40. Peak area was used as a qualitative measure of relative concentrations of species analysed.

NMR spectra were recorded on a Varian Unityplus-400 spectrometer. All NMR experiments were conducted at 25°C unless otherwise stated. 1H NMR spectra were referenced to TMS (0 ppm) at 25°C using the residual 1H signal of the respective deuterated solvent. 1D spectra were recorded with between 1 and 16 transients with a 6 second relaxation delay. COSY experiments were conducted with 1 scans per increment, 512 t1 increments and 2 s relaxation delay. Integration data was processed and graphed using Graphpad Prism.

6.3.3 Methods

DCM and C2H2Cl2 (DCE) used in synthesis was freshly distilled from CaH2 under dry N2. Cyclohexanone (CyHexO) was purified by passage through a column of neutral alumina, and sparged with N2 for 1 h.

180 6.3.3.1 Synthesis of Pt2Cl2(bis(diphenylphosphino)propane)

15 The title compound was synthesised using the procedure of Brown. 402 mg PtCl2(1,5- cyclooctadiene) (1.1 mmol) was dissolved in 2 mL of DCM. To this solution was added

448 mg (1.1 mmol) of (diphenylphosphino)propane (DPPP) in 5 mL DCM. A further

6 mL DCM was added, and the mixture stirred for 2 days under N2 at RT. The mixture was filtered, washed with 3 x 1 mL ice-cold DCM and the solid product was dried in vacuo for 4h @ 150°C. Yield 582 mg, 80%. Product was used directly for next step without characterisation.

6.3.3.2 Synthesis of [Pt2(bis(diphenylphosphino)propane)2(μ-OH)2.(BF4)2]

(Pt-DPPP)

The title compound was synthesised using the procedure of Li16 with a modified workup. 390 mg (2.0 mmol) of AgBF4 was suspended in 8 mL of 10:1 (v/v) methanol- water mixture, and stirred at RT. To this suspension 390 mg (1.0 mmol) of Pt2Cl2-

(bis(diphenylphosphino)propane) was added with a further 26 mL of aqueous methanol.

The mixture was stirred at RT for 2 h, then dried by rotary evaporation followed by high vacuum. The solid was dissolved in dry DCM, filtered to remove AgCl, and the volume reduced by half by rotary evaporation. 2 mL of Et2O was added and the solution refrigerated overnight to yield a fine white precipitate which was collected by filtration, washed with cold Et2O, and dried in vacuo at 70°C overnight. Yield: 402 mg (0.286

1 mmol), 57% by Pt. H NMR (400 MHz, DMSO-d6): 7.3-7.8 (m, 40H, phenyl), 2.9 (br,

8H, PCH2CH2CH2P), 1.96 (s, 2H, OH), 1.80 (br, 4H, PCH2CH2CH2P). This assignment differs from that of the reference16 which (based on the expected downfield shift induced by the phosphino moieties) appears to assign the methylene protons incorrectly.

181 6.3.3.3 Synthesis of Sn(II)-montmorillonite complex (Sn-MMT)

Sn-MMT was synthesised by the method reported by Lei,17 which was subsequently modified. ~30 g Montmorillonite K-10 (Sigma-Aldrich) was suspended in

50 mL distilled water, stirred for 4 h, and allowed to settle for 30 mins. The supernatant was removed, and the sediment discarded. The supernatant was centrifuged and the solid dried overnight at 50°C, ground in a mortar, and passed through a 75 μm sieve. 1.4 g of this material was suspended in 50 mL of water and 5 drops of 1 M HCl (this was later modified to suspending in 50 mL of 4 M HCl to avoid the immediate hydrolysis of

SnCl2). 2 g of SnCl2 was added with vigorous stirring, and stirred at RT overnight. The sample was centrifuged, the liquid decanted and the MMT washed with copious distilled water until the washings did not react with silver nitrate. (In our modified procedure, fresh 4 M HCl and SnCl2 were added, and the process repeated.) The sample was dried in the oven, then in vacuo at 100°C overnight. The sample was used directly without further analysis.

6.3.3.4 Catalytic oxidation of cyclohexanone by HOOH

24 mg of Pt-DPPP was weighed into a RB flask, sealed and purged with N2.

167 mg (176 μL, 1.7 mmol) of CyHexO and 3.7 mL of DCE was added, and the solution stirred for 15 min in a thermostated 25°C oil bath to allow the temperature to equilibrate. 175 μL 30% HOOH (~1.7 mmol) was added and the time started. 1 mL samples were taken at required times, quenched over NaSO3 and CaCO3, passed through a silica plug to remove Pt (washed with 1 mL DCM), then analysed by GCMS.

Sn-MMT was assessed in an analogous way, using 20 mg of catalyst.

182 6.3.3.5 Catalytic oxidation of cyclohexanone by DPPDHP

12 mg (0.017 mmol) of Pt-DPPP was weighed into a RB flask and purged with

N2. 83 mg (88 μL, 0.85 mmol) of CyHexO and 1 mL of DCE was added, and the solution stirred for 15 min in a thermostated 25°C oil bath to allow the temperature to equilibrate. 100 mg DPPDHP (0.42 mmol) in 1 mL DCE was also equilibrated in the bath in a separate vial, then added by syringe and the stopwatch started. 1 mL samples were taken at required times, quenched over NaSO3 and CaCO3, passed through a silica plug to remove Pt (washed with 1 mL DCM), then analysed by GCMS. A control experiment was run in an identical fashion, except Pt-DPPP was omitted. Sn-MMT was assessed in an analogous way, using 10 mg of catalyst.

6.3.3.6 NMR studies of TPTP acidic degradation in the presence of Pt-DPPP

18.5 mg (0.013 mmol) Pt-DPPP was weighed into a 5 mL volumetric flask and dissolved in ~2 mL CDCl3. 11.9 μL (11.6 mg, 17.6 mmol) of acetophenone (PhAc) was added, and the solution made up to the mark with CDCl3 (2.6 mM Pt-DPPP, 19.3 mM

PhAc). Separately, 17.8 μL (26.3 mg, 0.27 mmol) of methanesulfonic acid (MSA) was dissolved in 5 mL CDCl3 to make a 55 mM solution. 100 μL of this solution was further diluted in 2 mL of CDCl3 to make a 2.7 mM solution of MSA. 2.2 mg (7.2 μmol) TPTP was weighed into a dry NMR tube and dissolved in 500 uL of the PhAc/Pt-DPPP solution (9.6 μmol and 1.3 μmol respectively), and the 1H NMR spectrum taken.

Aliquots of the 2.7 mM or 55 mM MSA solution were added, and the reaction monitored by 1H NMR. Separate experiments were conducted with both TPTP/MSA mixtures and Pt-DPPP/MSA mixtures to observe their behaviour in isolation.

183 6.4 Results and discussion

Pt-DPPP was synthesised to test the feasibility of catalysed BV reactions as a method of generating safe oxidation products from the acid-redox degradation of organic peroxides. Catalytic reactions were first conducted with DPPDHP as a model of the likely degradation intermediates to confirm if the catalysts were active with gem- peroxyhydroperoxide instead of hydrogen peroxide. Further experiments were then conducted to examine the feasibility of generating the gem-peroxyhydroperoxide by in situ degradation of TPTP.

6.4.1 Catalytic oxidation with DPPDHP

Pt-DPPP was successfully synthesised by literature procedures.16 Figure 6.1 shows the assigned 1H NMR spectrum of the pure product. It should be noted that our assignment of the methylenes differs from that of Li16 however considering the clear integral ratios and expected downfield shift of the the α-methylene in similar bis(diphenylphosphino) ligands (eg. 1,4-bis(diphenylphosphino)butane18) it seems that the earlier reporting may have been erroneous.

Figure 6.1 - 1H NMR of Pt-DPPP. Solvent = DMSO-d6 Integrals below. Major unassigned peaks at 2.5 and 3.3 ppm are residual DMSO and water.

184 As the bulk of the literature on this class of catalysts for BV oxidation focus on the use of HOOH as the oxidant, the initial aim was to identify if they were also able to mediate oxidation with organic hydroperoxides. DPPDHP is a good model for the gem- peroxyhydroperoxides expected as intermediates in the degradation of TPTP. As

DPPDHP was readily available, it was utilised as the oxidant for our initial catalytic oxidation experiments. The substrate (CyHexO) was chosen as it readily oxidised to ε- caprolactone (CLO), and has been used in other literature studies with this class of catalyst.12

The activity of the synthesised catalyst was confirmed by repeating a literature method12 of oxidising CyHexO by a molar equivalent of HOOH with a ~2.5% molar loading of Pt-DPPP catalyst at room temperature. Care was taken to exclude atmospheric oxygen as a possible alternate oxidant. Samples taken at 30 min, 60 min, 2 h and 14h showed a steady increase in peak area for CLO, and a decline for cyclohexanone (Figure 6.2). No oxidation was observed in the absence of Pt-DPPP

(Figure 6.3). Having demonstrated that the synthesised catalyst was working as expected in accordance with literature, the activity was tested with DPPDHP in place of

HOOH as the terminal oxidant.

1 0 0 C yH e xO 9 5 CLO 9 0

8 5

2 0

1 5 % peak area 1 0

5

0 0 200 400 600 800 1000 T im e (m in ) Figure 6.2 - Oxidation of CyHexO to CLO by HOOH catalysed by Pt-DPPP. Qualitative proportion of species is indicated by % peak area as monitored by GCMS. Dashed line does not show trend, only joins the data to lead the eye. 185 Figure 6.3 – Catalysed and uncatalysed oxidation of CyHexO by HOOH and DPPDHP monitored by GCMS. All reactions conducted at 25°C in DCE. Y-axis is peak counts, not shown as only qualitative peak area was used.

Identical reaction conditions were used for the DPPDHP experiments, again with one molar equivalent of oxidant. In this instance, only one sample was taken at 14 h to compare total activity with HOOH. Using the qualitative peak area measurement, ~8% of CyHexO had been oxidised to CLO by DPPDHP, compared to 17% for HOOH

(Figure 6.3). This result suggested the catalyst was approximately half as efficient when the gem-peroxyhydroperoxide was used instead of HOOH. A catalyst-free control experiment was conducted to confirm that the reaction was indeed being catalysed by

Pt-DPPP rather than the inherent reactivity of DPPDHP. As with HOOH, no CLO was detectable after 14 h using the same GCMS method, indicating the reaction only proceeds via the action of the catalyst.

186 6.4.2 Catalytic oxidation via in situ TPTP degradation.

The ability of DPPDHP to oxidise CyHexO in the presence of a known BV catalyst strengthens the findings of Chapter 2 which indicated these gem- peroxyhydroperoxides are active BV oxidants. The next step was to understand if these catalysts could have an impact on the overall degradation of TPTP. Initial experiments confirmed that TPTP was not able to act as an oxidant with Pt-DPPP, as no CyHexO oxidation was detectable after 24 h. The fate of TPTP and Pt-DPPP was not determined in these initial tests. This was an unsurprising result considering the general chemical inertness of cyclic peroxides seen thus far, and the soft Lewis acid character of these catalysts. As strong Brønsted acids are known to catalyse the degradation of cyclic peroxides to generate the active hydroperoxide oxidants, attempts were made to conduct the oxidation of CyHexO by in situ acid degradation of TPTP in the presence of Pt-

DPPP. HCl, p-toluenesulfonic acid, and H2SO4 were all trialled in 0.1%-2% loadings

(with reference to oxidant), but no evidence of CyHexO oxidation was detected by

GCMS.

To understand why oxidation did not proceed via in situ TPTP degradation,

NMR was used instead of GCMS to allow direct monitoring of the reaction. The acid catalyst was changed to methanesulfonic acid (MSA) due to the methanesulfonate anion being visible on 1H NMR, which could provide further information on the reaction processes. The ability of MSA to degrade TPTP was confirmed in CDCl3 solution, where a molar equivalent of the acid converted the cyclic peroxide almost quantitatively to DPDP within minutes. DPDP continued to degrade at a slower rate to form 3- pentanone as the major product (Figure 6.4). This behaviour is similar to cyclic

1, 8, 10 peroxide degradation by H2SO4, which suggested that hydroperoxides and/or

HOOH might also be formed in this process.

187 Figure 6.4 - Reaction of TPTP and MSA observed by 1H NMR. ▲-TPTP, ●-DPDP, ■-MSA, ◄-3P. Spectra of DPDP and 3P included as reference. The slight downfield shift of 3P in the presence of MSA is likely to be caused by shifting of the keto-enol equilibrium by the acidic conditions.

A number of unassigned smaller peaks may indicate the products of other oxidative processes triggered by these peroxides, however as the species could not be elucidated the presence of ROOH could not be shown conclusively. Whilst not conclusive, the appearance of an extra singlet for MSA might indicate the methanesulfonate anion can react with carbocation intermediates to trap out ROOH in a similar reaction as that observed in the TiCl4 degradation of TATP (section 5.4.1.3).

DPPDHP was not seen by 1H NMR in the course of this reaction. The NMR evidence established that MSA could degrade TPTP, though the lack of solid evidence for ROOH posed a problem for the intended catalytic reaction. A NMR-tube catalytic oxidation experiment was designed to determine if the presence of Pt-DPPP would ‘trap out’ any fleeting hydroperoxides. In this case, acetophenone (PhAc) was selected as the substrate as its methyl singlet at 2.6 ppm provided a convenient means to track reaction

188 progression without overlapping peaks from other species. TPTP was added in two-fold excess to PhAc (i.e. a 6:1 peroxide to ketone ratio) to maximise opportunity to detect any BV product. Catalyst loading was increased to 0.2 mol equivalent (with respect to

TPTP) to make the Pt-DPPP signals readily detectable, and to increase the rate of any oxidative reaction. MSA was added at first in 0.002 mol equivalent, as it was only required to initiate the degradation of TPTP.

Figure 6.5 - Reaction of MSA, TPTP and Pt-DPPP observed by 1H NMR. Solvent=CDCl3. □ – acetophenone (PhAc), ■ – TPTP, ○ – trace water. Pt-DPPP’s spectrum is given in D. Relative concentration of reagents shown on left. Additionally, all reactions contained 0.5 equivalent of PhAc (with respect to TPTP).

Figure 6.5(A) shows the sample before the addition of acid, with the Pt-DPPP methylene signals clearly visible. Figure 6.5(B) shows that 10 mins after the addition of

MSA, TPTP remains unreacted, but the presence of acid is confirmed by the broadening of the residual water signal (indicated, blue arrow). Although much less acid had been added in this sample than the previously described TPTP-MSA degradation, the complete absence of TPTP degradation products such as DPDP or 3P (even after a further 1 h elapsed) was unexpected, and at first attributed to the reduced [MSA].

189 Attempts to initiate TPTP degradation by addition of further MSA resulted instead in the degradation of Pt-DPPP, seen in Figure 6.5(C) as the shift in the methyl signals

(indicated, red arrows). Not shown in the figure is the loss of the bridging –OH signal which is found at -0.90 ppm in CDCl3, indicating a breakdown of the dimeric structure of the Pt-DPPP compound.

Reaction of the hydroxy bridges together with the broadening and complete absence of the water signal in Figure 6.5(B) and (C) respectively indicates that Pt-

DPPP is reacting with MSA, presumably breaking the bridging hydroxy ligands to form aqua ligands. This reaction absorbs the free protons required for catalyzing the opening of the peroxide ring, protecting TPTP from any detectable acid degradation. Scheme

6.C shows a potential mechanism for this process, which may involve the coordination of methanesulfonate. The strong affinity of Pt for aqua ligands is a well known phenomenon which has been widely researched with respect to its importance in the anti-cancer mechanism of certain platinum compounds.19

Scheme 6.C - Proposed mechanism for the reaction of MSA nd Pt-DPPP. It is likely this process would also occur for the second OH-bridge to form a symmetrical structure. The Pt-methanesulfonate ligand was not confirmed spectroscopically.

This degradation of Pt-DPPP explains the lack of BV reactions taking place in the presence of TPTP and acid. Multiple studies have concluded that the formation of a hydroxy-bridged dimeric structure is central to the catalytic activity of compounds such as Pt-DPPP by driving the release of the oxidised substrate and regenerating the active

190 dimeric catalyst.11, 13, 20-21 MSA appears to be breaking down the dimeric structure, (it is likely that HCl and H2SO4 also triggered a similar process) explaining why the experiments aimed at in situ generation of ROOH did not replicate the BV products seen with DPPDHP. Whilst DPPDHP is clearly an active oxidant for these BV catalysts, in situ generation of the ROOH via the described acidic degradation of cyclic peroxides is not compatible with this particular class of catalyst.

6.4.3 Alternate catalysts for BV oxidation.

The Pt-DPPP experiments show that it is possible to oxidise substrates using hydroperoxides similar to the breakdown products of TPTP, however the catalyst is not able to tolerate the acid required to generate these compounds in situ. A review of our results and the literature was conducted to assess strategies to overcome this limitation to the acid sensitivity of Pt-DPPP. Three approaches were identified:

1. Modify the Pt-DPPP BV catalyst or acid catalyst to avoid BV catalyst

degradation;

2. Substitute Pt-DPPP for an acid- and water-resistant catalyst proven to

catalyse BV reactions; or

3. Substitute Pt-DPPP for an acid- and water-resistant metal complex that

might reasonably be expected to catalyse BV reactions based on the

known mechanisms of activation.

Only the second of these strategies was able to be assessed to any extent in the laboratory; however the other approaches will be outlined as potential future avenues for investigation.

191 6.4.3.1 Modification of the Pt-DPPP system

As Pt-DPPP has shown good activity at room temperature in oxidising substrates using DPPDHP, it is worth considering possible strategies to allow the effective use of this class of catalyst. Whilst the acid-sensitivity of the catalyst is a critical issue, it is also important to examine the differences between the stable compound DPPDHP and the intermediate ROOH species being generated from TPTP by acid. Building on the observations in this thesis and previous studies,10, 22-24 Scheme 6.D illustrates the generation of dimeric peroxides and ROOH from TPTP under acid catalyis. The NMR observations of MSA’s reaction with TPTP (Figure 6.5) confirm that in acidic media with very low [H2O], TPTP rapidly decomposes to almost exclusively DPDP via R1H+, with ketone being liberated much more slowly via reactions R2H+, R3H+ and R4H+.

Multiple studies have found the formation of ketone and HOOH are concomitant in acidic degradation.1, 8, 10

Scheme 6.D - The role of water in the acidic degradation of cyclic peroxides.

The very fast rate of R1H+ suggested by the complete conversion of TPTP to

DPDP within 2 mins (Figure 6.D) is consistent with the findings of Oxley’s study of the effect of water on the formation and degradation of TATP.22 Oxley reported that TATP was degraded by sulfuric acid at a rate 3 orders of magnitude faster in dry MeCN than

192 in MeCN with 10% H2O, but DADP was the exclusive product. Furthermore, in the aqueous reaction, DADP was no longer the dominant product, highlighting the crucial role of water in DADP degradation. Oxley found that only in the presence of water does degradation proceed all the way to ketone. All of these observations support the acidic degradation mechanism proposed by earlier by Armitt, where water is also central.10

The formation of ROOH or HOOH are both reliant on the presence of water via

R3H+ and R4H+ respectively. Thus, in very dry, aprotic conditions, ROOH exists largely as a highly unstable carbocation intermediate (Scheme 6.D(A)), If water is present, it

‘traps out’ the carbocation A, generating B via R3H+. B can then react further, producing ketone and HOOH in R4H+. This then links with the numerous observations that HCl degradation of TATP avoids the formation of DADP and results in faster

1-2, 10, 22 - degradation. As Cl , like H2O, would be reactive toward the carbocation A,

R3HCl (Scheme 6.D) would also assist in shifting the equilibrium of R2H+ to the right.

This would increase the overall reaction rate, enhancing the formation of the required

ROOH and HOOH products. Thus the emphasis of Scheme 6.D on the carbocation provides further explanation for the rate enhancement of HCl beyond a purely [H2O] effect as proposed by Oxley.2

This explains why the BV oxidation of CyHexO was successful when DPPDHP was used as oxidant, but not when the oxidant was being generated in situ by the acidic degradation of TPTP. In hindsight, even if the catalyst had been stable in the acidic conditions, it is probable that the reaction conditions would still have required modification to ensure sufficient ROOH was available for oxidation to occur. As Pt-

DPPP was shown to be more active with HOOH as oxidant than ROOH, conditions that promote complete degradation of cyclic peroxides to HOOH would be optimal. The addition of water to the system would therefore be a useful alteration to the reaction

193 conditions, although the amount required is not clear. HCl would also be useful to

- increase the overall rate of reaction, however Cl reacts with the desired HOOH forming

HOCl. The presence of HOCl is thought to be responsible for the chlorination of acetone to highly toxic chloroacetones,2, 10 making HCl unsuitable. Other reagents (eg. methanol) may be able to perform a similar rate-enhancing role to Cl- by reacting with the carbocation without forming the undesireable toxic products.

The issue of Pt-DPPP’s apparent degradation in the presence of MSA and other acids remains a problem. MSA was selected for the experiments for the convenience of being visible by 1H NMR, however even the weakly-donating methanesulfonate counter ion appears able to coordinate to Pt. In re-examining the literature on Pt-based BV catalysis, it was found that perchloric acid had been used in early studies to examine the effect of acidic conditions on the catalyst. As HClO4 did not appear to adversely impact the function of the catalyst,13 it would seem that it may be a suitable replacement acid to facilitate in situ generation of active oxidants. It has been shown to be active in triggering the degradation of TATP, albeit in a highly vigorous reaction.2 As only trace amounts of acid are required to trigger degradation of the cyclic peroxide it is unlikely this would cause any significant concern in this case.

After careful consideration of the observations made in this pilot trial and examining the relevant literature, the following suggestions are provided for future investigations of a cyclic peroxide degradation system based on Pt-catalysed BV oxidation:

1. Use of a low (~1%) loading of perchloric acid and medium loading of a Pt

catalyst (~5%) to minimise possible impacts of the acid on the catalyst;

2. Addition of water, although the effect on the Pt catalyst will need to be tested

and the [H2O] adjusted accordingly; and

194 3. Use of a solvent system which provides a BV-oxidisable moiety, and a

nucleophilic moeity able to attack a carbocation without forming toxic

products.

6.4.3.2 Acid- and water-resistant BV catalysts

A second possible approach involves replacing Pt-DPPP with a more acid- resistant catalyst, and one that can also tolerate the presence of water (considering the need for water to generate HOOH). The preference was to seek a catalyst that operates by a similar ketone-activation mechanism as Pt-DPPP, as this study has confirmed this approach is compatible with both HOOH and ROOH as the active oxidant. The next most active catalysts after Strukul’s Pt catalysts are based on the incorporation of Lewis acid ions (usually Sn2+ or Ti4+) into a zeolite matrix. An example of this class of catalyst is the Sn-montmorillonite (Sn-MMT) complex reported by Lei as simple to prepare and demonstrating extremely good activity.17 Indeed Lei reported quantitative conversion of

4-methyl-2-pentanone to the ester, suggesting the compound was a good candidate to evaluate the zeolite class. Attempts to synthesise this complex suggested that there may have been detail missing in the reported method, as the very dilute HCl in which the

SnCl2 was reportedly dissolved to impregnate the MMT immediately hydrolysed the bulk of the SnCl2 to insoluble hydroxides. The method was modified for a second batch, which utilised 4 M HCl instead of the ~0.01 M in the published method, which resulted in less hydrolysis. Both samples were assessed using the same conditions as Pt-

DPPP, but no active oxidation products could be detected at all. Previously reported reactions utilising this catalyst are usually conducted at 90°C,17 thus the reaction may have been sluggish at room temperature, or potentially a crucial step is missing in the reported synthetic method.17 In either case, the observed performance was not

195 encouraging. Considering that the most active of the zeolite-based Lewis-acid catalysts* reported by Corma required heating to ~90°C for efficient oxidation of CyHexO.25 and bearing in mind the intended application of neutralising peroxides in the field, catalysts with such low activity at room temperature were not considered good solutions in the search for an acid-tolerant system.

Other BV catalysts show a similar requirement for heat. ReCl4(MeCN)2 is one example of a number of rhenium catalysts that have demonstrated the ability to catalyse the oxidation of a range of ketones with good conversion rate, including acyclic ketones with ~70% conversion; however the reactions require 6 h at 70°C to achieve this.26

Whilst the heating needed in these studies is again undesirable, other characteristics are more favourable. ReCl4(MeCN)2 has been reported as tolerant of water in solution at room temperature,27 and could be expected to be reasonably resistant against acids HX as long as the acidic counter ion X- is a poor ligand (perchloric acid may again be a good choice). These characteristics suggest that ReCl4(MeCN)2 and other Re catalysts

(including methylrheniumtrioxide (MTO), which is also active in catalysing BV oxidations28) are worthy of investigation as more robust replacements for Pt-DPPP.

6.4.3.3 Alternative BV catalysts

Considering the preceding discussion (section 6.1.3) regarding the mechanisms of

BV catalysis, an alternative approach was to identify a compound not previously known as a BV catalyst that could reasonably be expected to activate BV oxidations by one of the five mechanisms described in Scheme 6.A. In the review where these mechanism were proposed, ten Brink noted that catalysts for Diels-Alder reactions function via a similar ketone activation method (although the substrates in this case are α,β-

* This compound could not be assessed as a furnace was not available for the essential calcining process required in the synthesis of this catalyst. 196 unsaturated ketones) and thus may be applicable to BV catalysis.7 This is supported by the observation that MTO (described in the previous section) not only catalyses BV oxidations,28-29 but is also an active Diels-Alder catalyst that displays water and acid tolerance.30 Such Diels-Alder catalysts are worthy of investigation as catalysts for BV oxidation with in situ acidic cyclic peroxide degradation.

Considering the extreme oxophilicity of TiCl4 demonstrated by the results of

IV Chapter 4 and 5, it is not surprising to find compounds of Ti have been explored as ketone-activating catalysts in Diels-Alder chemistry. Indeed, TiCl4 itself is used as a catalyst in Diels-Alder reactions, however its extreme reactivity presents numerous limitations to its application, and rules out these simple Lewis acid catalysts for in situ degradation/BV catalysis. Odenkirk presents an excellent discussion on these limitations,31 and also presents a clear methodology for designing an ideal catalyst to overcome these issues. In the same paper, Odenkirk reported Ru-salen (salen= N,N'- ethylenebis(salicylimine)) based Diels-Alder catalysts which are air-stable, active at

25°C in the presence of acetone* and chlorinated solvents, and tolerant of water in the reaction with only slight reduction in activity. Ti-based catalysts have also been designed in which water-sensitivity is overcome through the use of caged, multidentate ligands, that are also active in Diels-Alder catalysis.32 How this Diels-Alder activity translates to BV catalysis can only be determined experimentally and remains to be seen, but the indications are that Diels-Alder chemistry may be a useful source of catalysts to assess for BV activity in the presence of acidic cyclic peroxide degradation.

* It is important that the catalyst does not bind so strongly to reaction products that it is de-activated. As TATP, DADP and MEKP form ketones in their acidic degradation, the fact that Ru-salen is not deactivated by acetone is an encouraging observation. 197 6.5 Conclusion

The application of catalysis in cyclic organic peroxide degradation to this point has focused on H+, with the best example of this approach being limited by the hazard of the highly toxic, volatile product chloroacetone.2 This preliminary investigation of

BV rearrangement is the first true attempt to introduce a non-H+ catalyst to degrade cyclic peroxides to safe, non-toxic products. A possible exception to this statement is the study of thermal/radical degradation catalysts33-34 which were discussed in section

1.6, however these studies were aimed at synthetic process development, and are not directly applicable to field degradation. These experiments using Pt-DPPP showed that gem-peroxyhydroperoxides can be active oxidants, suggesting that BV catalysts can be used to irreversibly turn peroxides into esters given a suitable substrate at room temperature. The choice of catalyst in this study meant that ROOH could not be generated in situ from TPTP due to the preferential reaction of MSA with Pt-DPPP in the first instance, and also due to the relatively dry reaction conditions making the formation of ROOH unfavorable. A number of strategies have been proposed to overcome these problems, providing a range of avenues by which to pursue a catalytic, field-relevant and safe method to degrade OPEs.

6.6 References

1. Tsaplev, Y. B., Kinetics and Catalysis 2012, 53 (5), 521-524. 2. Oxley, J. S., J. L.; Brady, J. E.; Steinkamp, L., Propellants Explosives Pyrotechnics 2014, 39 (2), 289-298. 3. Oxley, J. C.; Smith, J. L.; Huang, J. R.; Luo, W., Journal of Forensic Sciences 2009, 54 (5), 1029-1033. 4. Apblett, A. W.; Kiran, B. P.; Malka, S.; Materer, N. F.; Piquette, A., Ceram. Trans. 2006, 172, 29-35.

198 5. March, J., Advanced organic chemistry: reactions, mechanisms, and structure. McGraw-Hill: New York, 1968; p ix, 1098. 6. Emmons, W. D.; Lucas, G. B., J. Am. Chem. Soc. 1955, 77 (8), 2287-2288. 7. ten Brink, G. J.; Arends, I. W. C. E.; Sheldon, R. A., Chemical Reviews 2004, 104 (9), 4105-4123. 8. Furuya, Y.; Ogata, Y., Bull. Chem. Soc. Jpn. 1963, 36 (4), 419-422. 9. Uyanik, M.; Ishihara, K., ACS Catalysis 2013, 3 (4), 513-520. 10. Armitt, D.; Zimmermann, P.; Ellis-Steinborner, S., Rapid Commun. Mass Spectrom. 2008, 22 (7), 950-958. 11. Michelin, R. A.; Sgarbossa, P.; Scarso, A.; Strukul, G., Coord. Chem. Rev. 2010, 254 (5-6), 646-660. 12. Frisone, M. D.; Pinna, F.; Strukul, G., Organometallics 1993, 12 (1), 148-156. 13. Gavagnin, R.; Cataldo, M.; Pinna, F.; Strukul, G., Organometallics 1998, 17 (4), 661-667. 14. Strukul, G.; Sinigalia, R.; Zanardo, A.; Pinna, F.; Michelin, R. A., Inorg. Chem. 1989, 28 (3), 554-559. 15. Brown, M. P.; Puddephatt, R. J.; Rashidi, M.; Seddon, K. R., Journal of the Chemical Society-Dalton Transactions 1977, (9), 951-955. 16. Li, J. J.; Li, W.; Sharp, P. R., Inorg. Chem. 1996, 35 (3), 604-613. 17. Lei, Z. Q.; Ma, G. F.; Jia, C. G., Catal. Commun. 2007, 8 (3), 305-309. 18. AIST Spectral Database for Organic Compounds, National Institute of Advanced Industrial Science and Technology http://sdbs.db.aist.go.jp/sdbs/cgi-bin/cre_index.cgi, accessed 10/02/2014. 19. Reedijk, J., Platinum Metals Review 2008, 52 (1), 2-11. 20. Brunetta, A.; Sgarbossa, P.; Strukul, G., Catal. Today 2005, 99 (1-2), 227-232. 21. Sgarbossa, P.; Scarso, A.; Michelin, R. A.; Strukul, G., Organometallics 2007, 26 (10), 2714-2719. 22. Oxley, J. C.; Smith, J. L.; Steinkamp, L.; Zhang, G., Propellants Explosives Pyrotechnics 2013, 38 (6), 841-851. 23. Matyas, R.; Pachman, J., Propellants Explosives Pyrotechnics 2010, 35 (1), 31- 37. 24. Matyas, R.; Pachman, J.; Ang, H. G., Propellants Explosives Pyrotechnics 2009, 34 (6), 484-488.

199 25. Corma, A.; Nemeth, L. T.; Renz, M.; Valencia, S., Nature 2001, 412 (6845), 423- 425. 26. Alegria, E. C. B. A.; Martins, L. M. D. R. S.; Kirillova, M. V.; Pombeiro, A. J. L., Appl Catal A - Gen 2012, 443, 27-32. 27. Rouschias, G.; Wilkinson, G., Journal of the Chemical Society A -Inorganic Physical Theoretical 1968, (3), 489-&. 28. Herrmann, W. A.; Fischer, R. W.; Correia, J. D. G., J. Mol. Catal. 1994, 94 (2), 213-223. 29. AbuOmar, M. M.; Espenson, J. H., Organometallics 1996, 15 (16), 3543-3549. 30. Zhu, Z. L.; Espenson, J. H., J. Am. Chem. Soc. 1997, 119 (15), 3507-3512. 31. Odenkirk, W.; Rheingold, A. L.; Bosnich, B., J. Am. Chem. Soc. 1992, 114 (16), 6392-6398. 32. Bull, S. D.; Davidson, M. G.; Johnson, A. L.; Mahon, M. R.; Robinson, D. E. J. E., Chemistry-an Asian Journal 2010, 5 (3), 612-620. 33. Tasca, J.; Lavat, A.; Nesprias, K.; Barreto, G.; Alvarez, E.; Eyler, N.; Canizo, A., Journal of Molecular Catalysis A-Chemical 2012, 363, 166-170. 34. Zareba, M.; Legiec, M.; Sanecka, B.; Sobczak, J.; Hojniak, M.; Wolowiec, S., Journal of Molecular Catalysis A -Chemical 2006, 248 (1-2), 144-147.

200 CHAPTER 7. Conclusion and Recommendations

This thesis has covered a diverse range of chemistry in an effort to chart a new course in achieving the safe, field relevant degradation of Organic Peroxide Explosives

(OPEs). Yet whilst focusing on an applied goal, the research has been results-led, and has explored novel reactions as they were identified. A significant addition has been made to the literature on the mechanistic pathways available to these molecules. This chapter will draw into focus the major findings, their implications on wider peroxide- degradation literature, and make recommendations on potential avenues for further study to achieve the desired endstate of safe OPE degradation.

7.1 Key Findings

7.1.1 Insights into acidic degradation

The literature surrounding degradation of OPEs has been dominated by two key themes: a focus on ketone-derived cylcic peroxides and the use of acid-catalysis. This is likely to be a response to the prominent threat posed by TATP to first responders by its precursor availability, sensitivity and volatility, combined with a good understanding of the acid- catalysed chemistry by which TATP is synthesised. As a result of this focus, the reaction of peroxides under acidic conditions has been comprehensively studied, both from a synthetic and degradation perspective. Only the thermal degradation mechanism has received similar attention, however concerns over the rate and safety of thermolysis make its relevance to laboratory or field neutralisation limited.1

Acid-catalysis is the only chemical means of degradation that has been subject to detailed, experimental, mechanistic investigations2-4 to date. Other chemical means that have been considered for the degradation of OPEs include metal salts1, 5 and oxidisable 201

metal compounds,6-7 however these studies have not provided conclusive evidence that the active reagents are reacting directly with cyclic peroxides, nor has the mechanism at work been investigated. The review in Chapter 1 of this thesis highlighted that many of the experimental conditions described in these latter reports may in fact suggest the presence of acid, although not always explicitly discussed. In this way, acid may have played a critical but often unrecognised role by generating more reactive hydroperoxides which were then reduced by the ‘active’ reagent. Furthermore, the investigations in Chapter 3 confirmed the high resistance of cyclic peroxides to oxidation compared to ROOH; both chemically and, for the first time, electrochemically. Using the molybdenum hydrogen bronze example, it was shown that degradation occurs significantly faster for hydroperoxides than cyclic peroxides. It has been demonstrated that that even when not explicitly discussed, it is probable that acid still plays a key role in most reported degradation schemes by overcoming the relative inertness of cyclic peroxides to chemical reduction.

Despite the centrality of acid-catalysis to the OPE degradation literature, a number of characteristics of this reaction limit the use of acid as a field-relevant neutralisation means. Addition of concentrated acids onto solid peroxide risks immediate detonation,5 and a large number of studies have noted that in general, the acid-catalysed degradation of TATP results in the formation of DADP which is quite resistant to further degradation,1-4, 8-11 thereby failing to remove the explosive hazard. Many reports have also observed that HCl did not produce DADP, instead generating significant quantities of chlorinated acetones,2, 4, 9, 11 which form a new hazard due to their toxic and lachrymatory effects. Whilst Armitt et al. noted that the presence of an oxidisable conjugate base (Cl-) substantially enhanced overall rate of degradation in HCl,2 this thesis has been the first to explicitly note the link between rapid degradation and the

202

absence of the acid-stable and explosive intermediate DADP. It is proposed that targeting DADP’s hydroperoxide precuror is the key to achieving facile degradation and the complete neutralisation of the organic peroxide threat.

Recent studies have not fully capitalised on Armitt’s initial insight that effective degradation requires an oxidisable substrate, however this targeted approach has formed a central theme of this thesis. In Chapter 2, studies of the TATP analogue tripentanone triperoxide (TPTP) revealed that gem-peroxyhydroperoxide acid-degradation intermediates can facilitate Baeyer-Villiger (BV) oxidations on 3-pentanone (3P) and other ketone substrates. BV oxidation provides a safer alternative to the oxidative chlorination resulting from HCl-degradation, with the ester (or lactone) products being much less hazardous than chloroacetone. The formation of these products also effectively locks away the peroxidic oxygen, preventing any subsequent formation of organic peroxides.

In chapter 6, the potential application of BV catalysts was explored with the aim of accelerating this side-reaction in acidic conditions. Whilst considerable method development is still required to allow this approach to be used in conjunction with acid catalysis, the potential for room-temperature catalysis of gem-peroxyhydroperoxide BV reactions has been demonstrated. This novel, exploratory work has opened a link between OPE degradation and the extensive literature on catalysed oxidation by ROOH, thus providing future studies with a robust strategy to achieve safe catalytic degradation.

7.1.2 Lewis acid-mediated rearrangement

In Chapter 4, a completely novel reaction of cyclic peroxides with Lewis acids was identified, and a comprehensive mechanistic explanation for the products was developed. TPTP and TiCl4 were found to undergo a particularly vigorous reaction, which was investigated in detail. The large proportion of ester products indicated BV-

203

like mechanisms; yet the presence of both oxidant and rearranged alkyl species within the parent cyclic peroxide, and the negligible impact of intermolecular BV-oxidation traps on the reaction products, provides compelling evidence for a novel, intramolecular

BV-type reaction. A lack of radical-derived products, the extremely repeatable 2:1 ester to ketone ratio, and the extremely fast initial reaction all suggest a concerted, ionic mechanism facilitated by the gem-peroxide ring structure of TPTP and the oxophilicity of TiCl4. The proposed mechanism’s suggestion of a novel Ti-OCl species, whilst not able to be definitively identified, has been suggested in other studies,12-13 and will be pursued in future studies as it has the potential to provide useful chlorination chemistry.

Further evidence for the proposed concerted BV-type mechanism was provided in

Chapter 5. Esters were again prominent in reactions of TATP and DADP with large excesses of TiCl4, suggesting the extremely uncharacteristic migration of a methyl group, an observation not reported in BV literature. As the ratio of TiCl4 is reduced, and product-binding tames the oxophilicity of the active TiIV species, the resistance of methyl to migration in TATP and DADP drives competing stepwise mechanisms, and potentially an oxacationic route, each with characteristic products. The presence of analagous stepwise products in the degradation of a mixture of MEKP isomers indicates that the BV reaction’s stereospecificity might activate the same competing mechanism where methyl falls in the migratory position of MEKP’s cyclic trimer. Rather than weaken the argument of the proposed concerted mechanism, the ambiguous product ratios in the methyl regime highlights the significance of TPTP’s strict 2:1 ratio, which is best accounted for by the triperoxide structure undergoing concerted degradation. It is interesting to note the approach of using TPTP as a model peroxide was fundamental to our elucidation of this mechanism. The strict product ratios provided by the facile

204

migration of ethyl groups was a critical insight which would not have been made had only threat compounds such as TATP and MEKP been utilised in this study.

In Chapter 4, it was identified that reactions continued to facilitate rearrangement of cyclic peroxides even when the Lewis acidity of TiCl4 was moderated by the binding of ligands (or substituting TiCl4 with SbCl3). This suggests that whilst the facile hydrolysis of TiCl4 makes it unlikely to be useful for peroxide degradation outside the laboratory, water-tolerant Lewis acid complexes may display useful chemistry in rearranging peroxides into safe ester products. That potential has been demonstrated by the experiments with soft Lewis acid catalysts in Chapter 6. Whilst the Pt-mediated rearrangements were stepwise-intermolecular instead of concerted-intramolecular, the central function of the Lewis acid in activating a Criegee-like intermediate to effect rearrangement remains the same. With this in mind, the results of the anhydrous TiCl4 reactions in Chapters 4 and 5 are fundamentally related to the water-tolerant Pt catalyst rearrangements studies in Chapter 6. Perhaps most significantly, Chapter 5 highlights the broad-based reactivity of Lewis acids with all of the threat peroxides tested. TPTP,

TATP, MEKP and HMTD all underwent effectively instantaneous degradation when a

3x excess of TiCl4 was added, with DADP only slightly less reactive. Exploiting this fundamental affinity between Lewis acids and peroxides should be a fruitful strategy with which to seek a broad-spectrum means of neutralising OPEs.

7.1.3 Other findings of note

Alongside the central explorations of BV reactions, and Lewis-acid activated rearrangement, this thesis has presented a number of smaller but useful insights into the chemistry of organic peroxides. Some of the other notable observations made in this thesis include:

205

1. Demonstrating that substituting acetone peroxides with pentanone peroxides

can reduce some risks inherent in handling organic peroxides for the purposes

of certain degradation experiments;

2. Providing a valuable illustration of the discriminatory ability of the canine

olefactory senses in being able to distinguish between TATP and TPTP,

which simultaneously highlighted:

a. the importance of training EDDs on all possible threat compounds, and

b. the inherent limitations of some patented approaches to producing safer

pseudo-scents for EDD training.

7.1.4 Summary

The difficulty of effecting the safe chemical degradation of OPEs can be summed up by observing that cyclic peroxides are a molecular paradox – chemically inert on the molecular scale, yet dangerously unstable at the macro level. This thesis has built upon the existing understanding of acidic degradation of organic peroxides provided by the literature, but taken a fresh view in interpreting the reported results. Additionally, a significant body of novel chemistry has been added to not only the literature of organic peroxide degradation, but the broader organic, inorganic and forensic chemistry literature.

7.2 Recommendations

Based on the findings of this work, it is suggested that the pursuit of an effective

OPE degradation strategy should focus on a system that allows the controlled reduction of OPEs, rather than the extant single-reagent focus. Whilst the use of acids in isolation carries unacceptable risk, their ability to generate reactive hydroperoxides from inert

206

cyclic peroxides makes them a critical component of such a system. Subsequent reduction of these hydroperoxides into safe products completes the system, avoiding the formation of other dangerous peroxides.

Such acid-redox systems have existed in the literature since at least 1963, when

14 Furuya degraded TATP in the presence of H2SO4 and acetic acid, oxidising the latter to peroxyacetic acid. Yet whilst the method has been used, the critical role of the substrate on the overall reaction outcome has remained largely undiscussed in this and many other works. This thesis presents a shift of focus from the literature’s consideration of acid catalysis or redox reaction in isolation to a cohesive system that achieves the overall intent of safe degradation.

The Lewis-acid-catalysed BV process explored in this work is just one redox reaction that might be suitable for producing safe products from the catalytic degradation of cyclic peroxides. Identifying if the BV process or some other reaction is optimal will require further study, but with the broadened scope, it is hoped that the long-sought goal of a chemical means to deal with the threat of OPEs may not be far away.

7.3 References

1. Bellamy, A. J., Journal of Forensic Sciences 1999, 44 (3), 603-608. 2. Armitt, D.; Zimmermann, P.; Ellis-Steinborner, S., Rapid Commun. Mass Spectrom. 2008, 22 (7), 950-958. 3. Tsaplev, Y. B., Kinetics and Catalysis 2012, 53 (5), 521-524. 4. Oxley, J. S., J. L.; Brady, J. E.; Steinkamp, L., Propellants Explosives Pyrotechnics 2014, 39 (2), 289-298. 5. Oxley, J. C.; Smith, J. L.; Huang, J. R.; Luo, W., Journal of Forensic Sciences 2009, 54 (5), 1029-1033. 6. Fidler, F. L., T.; Carvalho-Knighton, K.; Geiger, C.L.; Sigman, M.E.; Clausen, C. A., Degradation of TNT, RDX, and TATP using Microscale Mechanically Alloyed

207

Bimetals. In Environmental Applications of Nanoscale and Microscale Reactive Metal Particles, Geiger, C. L., et al, Ed. American Chemical Society: Washington DC, 2009; pp 117-134. 7. Apblett, A. W.; Kiran, B. P.; Malka, S.; Materer, N. F.; Piquette, A., Ceram. Trans. 2006, 172, 29-35. 8. Oxley, J. C.; Smith, J. L.; Steinkamp, L.; Zhang, G., Propellants Explosives Pyrotechnics 2013, 38 (6), 841-851. 9. Matyas, R.; Pachman, J., Propellants Explosives Pyrotechnics 2010, 35 (1), 31- 37. 10. Matyas, R.; Pachman, J.; Ang, H. G., Propellants Explosives Pyrotechnics 2009, 34 (6), 484-488. 11. Fitzgerald, M.; Bilusich, D., Journal of Forensic Sciences 2011, 56 (5), 1143- 1149. 12. El-Ahl, A. A. S.; Elbeheery, A. H.; Amer, F. A., Synth. Commun. 2011, 41 (10), 1508-1513. 13. Cavallo, L.; Jacobsen, H., Inorg. Chem. 2004, 43 (6), 2175-2182. 14. Furuya, Y.; Ogata, Y., Bull. Chem. Soc. Jpn. 1963, 36 (4), 419-422.

208