Next Generation 'Frustrated Lewis Pairs'

Home , Borole, Frustrated Lewis pair, Matthew Fuchter, Persistent carbene

NEXT GENERATION ‘FRUSTRATED LEWIS PAIRS’

Daniel J. Scott

A dissertation submitted to the

Department of Chemistry of Imperial College London

for the degree of Doctor of Philosophy

December 2016

Abstract

The past decade has seen the emergence of a new concept in main-group chemistry: ‘frustrated Lewis pairs’ (FLPs) are combinations of a Lewis acid (LA) and base (LB) that are prevented from forming the classically-expected adduct. By displaying simultaneously acidic and basic behaviour, these systems have been shown to be capable of activating a wide range of chemical bonds, in a manner highly reminiscent of transition-metal (TM) compounds.

Chief among these reactions is the activation of H2, which can then be transferred from the FLP to an appropriate substrate. FLPs have thus provided an entirely new class of homogeneous hydrogenation catalysts, which do not require the use of TMs. Nevertheless, prior to the work described herein, such catalysis had not been successfully applied to the hydrogenation of organic carbonyl functional groups, and had been found to be extremely sensitive to the presence of moisture. This thesis describes work that has successfully overcome these limitations.

Chapter 1 provides an introduction to the field of FLP chemistry, including a general summary of the topic, and a more detailed overview of catalytic hydrogenation mediated by FLPs.

Chapter 2 describes the development of novel FLPs based on the combination of strong triarylborane LAs and weak ethereal solvent LBs. These systems are shown to be highly effective for the hydrogenation of a variety of substrates, including the first examples of both FLP-catalysed hydrogenation of ketones and aldehydes, and moisture-tolerant FLP-catalysed hydrogenation.

Chapter 3 describes the design and preparation of alternative systems incorporating Sn(IV)-based LAs, whose use in FLP-mediated H2 activation has not previously been documented. These too are shown to be tolerant of moisture, and to be suitable for the hydrogenation of C=O, C=N and C=C bonds.

Chapter 4 provides full experimental details for the procedures described in Chapter 2 and Chapter 3.

Declaration

The work described herein was performed at Imperial College London during the period between October 2013 and October 2016 inclusive, under the supervision of Dr Andrew Ashley and Dr Matthew Fuchter. All results were obtained by the author (Mr Daniel Scott) unless explicitly stated otherwise, and have not previously been submitted in pursuit of any other degree at any institution.

Daniel John Scott

December 2016

The copyright of this thesis rests with the author and is made available under a Creative Commons Attribution Non-Commercial No Derivatives licence. Researchers are free to copy, distribute or transmit the thesis on the condition that they attribute it, that they do not use it for commercial purposes and that they do not alter, transform or build upon it. For any reuse or redistribution, researchers must make clear to others the licence terms of this work.

Acknowledgements

First and foremost, I would like to thank my academic supervisors for getting me to this point. Without Andy and Matt’s constant support, guidance and intellectual input this thesis would not exist.

I would also like to acknowledge all the various colleagues with whom I have been lucky enough to share an office during the course of this project: they have graciously expended a huge amount of effort first to teach me how to function in a lab setting, and then to help me stay (relatively) sane. Almost everybody who falls into this category has earned a special mention, but in order to avoid a rambling Oscars-style outpouring I’ll keep it generic: thank you all (you know who you are).

Finally, while technical support for this project has been provided by a number of people, I have to single out Peter Haycock his assistance with NMR. For the past several years he has tolerated my frankly unreasonable demands on his time, expertise, and equipment with unfailing good humour. Without his help, this project could not have been nearly as successful.

Abbreviations

2-MeTHF 2-methyltetrahydrofuran

APCI-MS atmospheric pressure chemical ionisation mass spectrometry

Ar generic aryl group

Bn benzyl; phenylmethyl

Bu n-butyl col 2,4,6-collidine; 2,4,6-trimethylpyridine

Cy cyclohexyl

DABCO 1,4-diazabicyclo[2.2.2]octane

DCB 1,2-dichlorobenzene

DFB 1,2-difluorobenzene

DFT density functional theory

Dipp 2,6,-diisopropylphenyl

DMF N,N-dimethylformamide

DMSO dimethylsulfoxide

EDG electron-donating group

Et ethyl

EWG electron-withdrawing group

FLP frustrated Lewis pair

FXyl 3,5-trifluoromethylphenyl

HMBC heteronuclear multiple bond correlation

Hex n-hexyl

HOMO highest occupied molecular orbital

I nuclear spin iBu iso-butyl iPr iso-propyl

J coupling constant

LA Lewis acid

LB Lewis base

LUMO lowest unoccupied molecular orbital lut 2,6-lutidine; 2,6-dimethylpyridine m/z mass/charge ratio

Me methyl

Mes mesityl; 2,4,6-trimethylphenyl

Ms mesyl; O2SMe

MS mass spectrometry

MS-ToF mass spectrometry – time of flight

NHC N-heterocyclic carbene

NMR nuclear magnetic resonance

Np neopentyl; CH2CMe3

Ph phenyl

Pr n-propyl

R generic organic group

F R C(CF3)3

RT room temperature

T1 spin-lattice relaxation time tBu tert-butyl

Tf trifluoromethanesulfinate; O2SCF3

THF tetrahydrofuran

TM transition metal

TMP 2,2,6,6-tetramethylpiperidyl

TMS tetramethylsilane

Trip 2,4,6-triisopropylphenyl

Ts tosyl; 4-methylphenylsulfonyl

VT variable temperature

XRD X-ray diffraction xs excess

D-CD D -cyclodextrin

G chemical shift

Hr relative permittivity

Units

°C degrees Celcius

Å Ångströms atm atmospheres bar bar cal calories d days g grams h hours

Hz Hertz

J Joules

K Kelvin

L litres

M molar (mol dm–3) mol moles ppm parts per million s seconds

Table of contents

Abstract ...... 2 Declaration ...... 3 Copyright ...... 3 Acknowledgements ...... 4 Abbreviations ...... 5 Units ...... 7 Table of contents ...... 8 List of Figures ...... 10 List of Schemes ...... 13 List of Tables ...... 19

Chapter 1 - Introduction ...... 20 Chapter 1.1 - Putting FLPs in context: main group mimics of transition metals ...... 20 Chapter 1.2 - Introduction to frustrated Lewis pair chemistry ...... 24 Chapter 1.2.1 - Stoichiometric FLP reactivity ...... 24 Chapter 1.2.2 - Catalytic FLP reactivity ...... 27 Chapter 1.3 - Some historical context for FLP chemistry ...... 30 Chapter 1.4 - Hydrogenation catalysis using FLPs ...... 32

Chapter 1.4.1 - Stoichiometric FLP H2 activation ...... 32

Chapter 1.4.2 - Catalytic FLP H2 activation ...... 35 Chapter 1.4.3 - Aspects of FLP design ...... 40 Chapter 1. 5 - Frontiers and limitations of FLP chemistry ...... 45 Chapter 1.5.1 - FLP-mediated hydrogenation of carbonyl compounds ...... 46 Chapter 1.5.2 - Moisture tolerance in FLP chemistry ...... 47 Chapter 1.5.3 - Thesis aims ...... 48 Chapter 1.6 - References for Chapter 1 ...... 49

Chapter 2 - Boron-based FLPs ...... 58 Chapter 2.1 - Introduction ...... 58 Chapter 2.2 - Catalytic hydrogenation using borane/solvent FLPs ...... 62 Chapter 2.2.1 - Catalytic hydrogenation of imines ...... 65 Chapter 2.2.2 - Catalytic hydrogenation of additional substrates ...... 72

Chapter 2.2.3 - On the stability of borane/solvent Lewis pair hydrogenation catalysts ...... 77 Chapter 2.3 - Borane-catalysed carbonyl hydrogenation ...... 81 Chapter 2.3.1 - Development of an initial catalytic system ...... 81 Chapter 2.3.2 - Substrate scope for catalytic hydrogenation of aldehydes and ketones ...... 87 Chapter 2.3.3 - Mechanistic investigations into 1a-catalysed carbonyl hydrogenation in 1,4- dioxane ...... 91 Chapter 2.3.4 - Aldehyde and ketone hydrogenation at increased pressure ...... 99 Chapter 2.3.5 - On the stability of the 1a/1,4-dioxane system for catalytic carbonyl hydrogenation ...... 101 Chapter 2.3.6 - Further development of the catalytic system ...... 105 Chapter 2.4 - Moisture-tolerant borane-catalysed hydrogenation ...... 111 Chapter 2.4.1 - Development of a moisture-tolerant system for FLP-catalysed hydrogenation 114 Chapter 2.4.2 - Application and scope of moisture-tolerant FLP-catalysed hydrogenation ...... 118 Chapter 2.5 - Conclusions ...... 123 Chapter 2.5.1 - Summary and key results ...... 123 Chapter 2.5.2 - Relation to subsequent publications ...... 123 Chapter 2.5.3 - Directions for future work ...... 132 Chapter 2.6 - References for Chapter 2 ...... 133

Chapter 3 - Tin-based FLPs ...... 138 Chapter 3.1 - Introduction ...... 138

Chapter 3.2 - Synthesis and characterisation of iPr3SnOTf ...... 144

Chapter 3.3 - H2 activation using iPr3SnOTf-based FLPs ...... 148

Chapter 3.4 - Catalytic hydrogenation using iPr3SnOTf-based FLPs ...... 153 Chapter 3.4.1 - Catalytic hydrogenation of imines ...... 153 Chapter 3.4.2 - Catalytic hydrogenation of carbonyl compounds ...... 157 Chapter 3.4.3 - Catalytic hydrogenation of other substrates ...... 163

Chapter 3.5 - Moisture tolerance in iPr3SnOTf-based hydrogenation catalysis ...... 164 Chapter 3.5.1 - Air-stability of [26b]OTf ...... 164 Chapter 3.5.2 - Moisture tolerance in [26b]OTf-based carbonyl hydrogenation ...... 168 Chapter 3.5.3 - Moisture tolerance in [26b]OTf-based imine hydrogenation ...... 169

Chapter 3. 6 - The effect of anion variation in iPr3SnX ...... 174

Chapter 3.6.1 - Hydrogenation chemistry using iPr3SnNTf2 ...... 174

F Chapter 3.6.2 - Hydrogenation chemistry using ‘[iPr3Sn][Al(OR )4]’ ...... 177 Chapter 3.7 - Conclusions ...... 182 Chapter 3.7.1 - Summary and key results ...... 182

Chapter 3.7.2 - Directions for future work ...... 183 Chapter 3.8 - References for Chapter 3 ...... 185

Chapter 4 – Experimental details ...... 190 Chapter 4.1 - General experimental details ...... 190 Chapter 4.2 - Experimental details relating to Chapter 2. 2 ...... 192 Chapter 4.3 - Experimental details relating to Chapter 2. 3 ...... 196 Chapter 4.4 - Experimental details relating to Chapter 2. 4 ...... 204 Chapter 4.5 - Experimental details relating to Chapter 3. 2 ...... 208 Chapter 4.6 - Experimental details relating to Chapter 3. 3 ...... 209 Chapter 4.7 - Experimental details relating to Chapter 3. 4 ...... 210 Chapter 4.8 - Experimental details relating to Chapter 3. 5 ...... 214 Chapter 4.9 - Experimental details relating to Chapter 3. 6 ...... 217 Chapter 4.10 – References for Chapter 4 ...... 221

Appendices ...... 222 Appendix A - X-ray crystallographic data ...... 222 Appendix B - Procedure for precise gas transfer through use of a Toepler pump ...... 228 Appendix C - Related publications ...... 230 Appendix D – Reference summary of numbered structures ...... 251

List of Figures

Chapter 1

Figure 1.1 – Examples of highly important industrial heterogeneous catalytic reactions ...... 20

Figure 1.2 – Frontier molecular orbital model for TM activation of alkenes, CO, H2 ...... 22

Figure 1.3 – TM-like H2 activation by a persistent carbene; a frontier molecular orbital model for the reaction; selected other examples of TM-like reactivity of metallylenes ...... 23

Figure 1.4 – Examples of H2 and ethylene activation across E—E multiple bonds; frontier molecular orbital models for initial H2 activation by a distannyne and a reduced diboraanthracene ...... 24

Figure 1.5 –H2 activation by FLPs, which can be explained using a molecular orbital-based model or a now-disfavoured electric field-based model ...... 26 Figure 1.6 – Some representative examples of stoichiometric P/B FLP reactivity ...... 26 Figure 1.7 – The first example of FLP-catalysed hydrogenation; a selection of compounds accessible through FLP-catalysed hydrogenation reactions ...... 28

Figure 1.8 – Examples of LAs and LBs used in FLP H2 activation ...... 33

Figure 1.9 – A thermodynamic cycle for FLP H2 activation ...... 34 Figure 1.10 – A variety of highly-electrophilic, bulky arylboranes and other B-based LAs that have been used in FLP hydrogenation catalysis ...... 36 Figure 1.11 – Qualitative effect of cumulative LA + LB strength on FLP hydrogenation catalysis ...... 40 Figure 1.12 – Qualitative effect of steric bulk on FLP hydrogenation catalysis ...... 42

Figure 1.13 – Differing reactivity of oligomethylene-bridged P/B FLPs; intermolecular H2 activation by an intramolecular FLP ...... 44 Figure 1.14 – Examples of FLP development towards practical catalytic systems ...... 45

Chapter 2

Figure 2.1 – Initial boranes studied ...... 62

11 19 Figure 2.2 – VT B and F NMR spectra for equimolar THF/1a in C7D8 ...... 63 Figure 2.3 – VT 19F and 11B NMR spectra for 1b (top) and 1c (bottom) in THF ...... 64 Figure 2.4 – Qualitative effect of steric bulk on the hydrogenation of 2a catalysed by boranes 1b-d in THF ...... 69 Figure 2.5 – Reduced catalytic activity of 1a versus 1b for the hydrogenation of 2a in THF, due to stronger binding of the solvent ...... 71

Figure 2.6 – Less favourable H2 activation accounting for the reduced rate of 1b-catalysed hydrogenation in 1,4-dioxane relative to THF ...... 71 Figure 2.7 – VT 19F and 11B NMR spectra for 1a in 1,4-dioxane , and for equimolar 1a and 1,4-dioxane in C7D8 ...... 85

Figure 2.8 – A possible account for the unsuccessful 1a-catalysed hydrogenation of 20f ...... 89 Figure 2.9 – A qualitative explanation for the effect of substrate steric bulk on the 1a-catalysed hydrogenation of aldehydes in 1,4-dioxane ...... 91 Figure 2.10 – Upfield-shifted 11B NMR resonances as an indicator of 1a·HOiPr deprotonation ...... 97 Figure 2.11 – Control experiments showing the effect of 1,4-dioxane on the 11B NMR resonance of alcohol-free 1a in DFB ...... 98 Figure 2.12 – Appearance of separate DMF and 1a·DMF resonances in the 1H NMR spectrum of DMF and 1a (2:1) in 1,4-dioxane, indicative of strong binding ...... 107 Figure 2.13 – Reduced Lewis acidity and hydride ion affinity (calculated) in a less-fluorinated analogue of 1a ...... 109

11 Figure 2.14 – B NMR spectra for addition of H2O to 1a in 1,4-dioxane ...... 112

11 Figure 2.15 – Upfield-shifted B NMR resonances as an indicator of 1a·OH2 deprotonation by 1,4- dioxane ...... 113

11 Figure 2.16 – Upfield-shifted B NMR resonances as an indicator of 1a·OH2 deprotonation by 22a ...... 114 Figure 2.17 – Low basicity as a primary requirement for successful moisture tolerance in hydrogenation reactions catalysed by 1a in 1,4-dioxane ...... 120 Figure 2.18 – Examples of possible FLP-catalysed dehydrative hydrogenation reactions, which are among the long-term goals of FLP chemistry ...... 121

Figure 2.19 – The salt [iPr2O—H—OC(Et)CH2Ph][B(C6F5)4]: a model for an activated ketone intermediate formed during catalytic hydrogenation, isolated by Stephan and Mahdi ...... 125

Chapter 3

Figure 3.1 – A conceptual approach for the development of improved ‘ROH-tolerant’ FLP catalysts ...... 140 Figure 3.2 – Some previous examples of FLPs based on heavy p-block elements ...... 141

Figure 3.3 – A comparison of calculated hydride ion affinities and aqua complex pKa values for representative triarylborane and trialkylstannylium LAs ...... 142

1 13 1 Figure 3.4 – H and C{ H} NMR spectra for [26b]OTf in CDCl3 ...... 145 Figure 3.5 – Previously-reported solid-state structures of triorganotin triflates ...... 147

1 119 1 Figure 3.6 – H and Sn{ H} NMR spectra for addition of H2 (4 bar) to an equimolar mixture of DABCO and [26b]OTf in DFB ...... 149

2 119 1 Figure 3.7 – H and Sn{ H} NMR spectra for addition of D2 (2 bar) to an equimolar mixture of DABCO and [26b]OTf in DFB ...... 150

119 1 Figure 3.8 – Sn{ H} NMR spectra for addition of H2 (10 bar) to an equimolar mixture of lut or col and [26b]OTf in DFB: an upfield shift in the [26b]OTf resonance is attributed to reversible binding of the triflate released following H2 activation ...... 152 Figure 3.9 – Autocatalysis during the [26b]OTf-catalysed hydrogenation of 2g, due to a change in the dominant reaction mechanism ...... 157

Figure 3.10 – Solid-state crystal structure of [26b·2H2O]OTf ...... 166 Figure 3.11 – Definition of the geometric parameter W for 5-coordinate complexes ...... 167 Figure 3.12 – Recent examples of TM-catalysed hydrogenative amination ...... 173

Figure 3.13 – Solid-state structure of [26b]NTf2 ...... 175 Figure 3.14 – Some possible future targets for late p-block FLP hydrogenation catalysts ...... 184

Chapter 4

1 Figure 4.1 – H NMR spectra for the 1b-catalysed hydrogenation of 2a to 3a in d8-THF ...... 193 Figure 4.2 – 1H NMR spectra for the 1a-catalysed hydrogenation of 20b to 21b in 1,4-dioxane at 4 bar ...... 197 Figure 4.3 – 1H NMR spectra for the non-anhydrous 1a-catalysed hydrogenation of 20a to 21a in 1,4- dioxane at 45 bar ...... 207 Figure 4.4 – 1H NMR spectra for the [26b]OTf-catalysed hydrogenation of 2g to 3g in DCB ...... 210 Figure 4.5 – 1H NMR spectra for the [26b]OTf-catalysed hydrogenation of 20a to 21a in DCB ...... 211

F Figure 4.6 – NMR spectra for in situ generated [26b·col][Al(OR )4] in DFB ...... 218

F Figure 4.7 – NMR spectra for in situ generated [26b·DABCO][Al(OR )4] in DFB ...... 219

Appendix B

Figure A1 – Schematic representation of the Toepler pump apparatus used for precise gas transfer ...... 228

List of Schemes

Chapter 1

Scheme 1.1 – Mechanism for alkene hydroformylation catalysed by [RhH(CO)(PPh3)3] ...... 21

Scheme 1.2 – Reversible H2 activation by a phosphine/borane, reported by Stephan et al. in 2006 .. 25 Scheme 1.3 – An ‘FLP’ cyclisation reported by Melen et al. and a similar Lewis-acid-catalysed cyclisation – also an example of FLP chemistry? ...... 27 Scheme 1.4 – A range of examples of FLP-catalysed reactions ...... 29 Scheme 1.5 – An early example of LA/LB adduct formation precluded by steric factors ...... 30 Scheme 1.6 – Alkoxide-catalysed ketone hydrogenation, which proceeds via an FLP-like mechanism ...... 30

Scheme 1.7 – Piers-type carbonyl hydrosilylation, which proceeds via an FLP Si—H bond activation mechanism ...... 31

Scheme 1.8 – FLP-like activation of H2 on a main group metal oxide surface ...... 32

Scheme 1.9 – Mechanisms for H2 activation by intramolecular and intermolecular FLPs ...... 34 Scheme 1.10 – Rare examples of boron-free FLP-catalysed hydrogenation ...... 36 Scheme 1.11 – The general mechanism for FLP-catalysed hydrogenations to date; a variation for basic substrates; and alternative mechanisms based on initial hydride transfer or LA activation of the substrate ...... 38 Scheme 1.12 – FLP-catalysed hydrogenations that proceed via atypical mechanisms ...... 39 Scheme 1.13 – Catalytic alkene hydrogenation achieved through careful choice of LB; stoichiometric hydrogenation of an aniline ...... 41 Scheme 1.14 – Improved stability of a borane/NHC Lewis pair through reversible adduct formation; a classical silylium/phosphine Lewis adduct that displays FLP-like reactivity; TM-free catalytic alkene hydrogenation based on C—B hydrogenolysis ...... 43 Scheme 1.15 – An early example of FLP-mediated carbonyl hydrogenation ...... 46 Scheme 1.16 – A mechanism for C=O hydrogenation proposed by Nyhlén and Privalov; stoichiometric

C=O hydrogenation mediated by B(C6F5)3; inhibition of FLP catalysis by oxygen-centred donors ...... 47

Scheme 1.17 – H2 activation by an N-methylacridinium-based FLP, in the presence of moisture ...... 48 Scheme 1.18 – A bench-stable FLP pre-catalyst ...... 48

Chapter 2

Scheme 2.1 – Reversibility in the binding of oxygen-centred donors allows for catalysis by B(C6F5)3 to occur in their presence ...... 58

Scheme 2.2 – Incomplete H2 transfer in an early example of FLP-mediated carbonyl hydrogenation 59 Scheme 2.3 – Irreversible inhibition of FLP reactivity by ROH ...... 59

Scheme 2.4 – Decomposition of B(C6F5)3/ROH mixtures through protodeborylation ...... 60 Scheme 2.5 – Possible mechanism for FLP-catalysed carbonyl hydrogenation, which requires use of a weak LB for substrate activation ...... 60 Scheme 2.6 – FLP-catalysed hydrogenation using a weak phosphine or ether as the LB; reversible FLP

H2 activation using the solvent as the LB ...... 61 Scheme 2.7 – Different mechanistic pathways for borane-catalysed imine hydrogenation in donor and non-donor solvents ...... 67 Scheme 2.8 – Likely causes for the reduced rates of hydrogenation of 2d, 2e and 2f relative to 2a .. 68

Scheme 2.9 – An account for differing reactivity in the 1b-catalysed hydrogenation of 2f in d8-THF and C7D8 ...... 69 Scheme 2.10 – Explaining the inability of 1d to catalyse the hydrogenation of 2a ...... 70

Scheme 2.11 – Hydrogenation of N-heteroaromatic rings catalysed by 1b in d8-THF ...... 72 Scheme 2.12 – Observation of solvent oligomerisation during the 1b-catalysed hydrogenation of 8, and attempted hydrogenation of 10...... 73

Scheme 2.13 – Hydrogenation of C=C bonds catalysed by 1b in THF ...... 74

Scheme 2.14 – Stoichiometric 1b-mediated hydrogenation of pyrroles 16 in d8-THF; and an explanation for the lack of catalytic turnover ...... 75 Scheme 2.15 – Unsuccessful hydrogenation of electron-poor pyrrole 16c, and successful hydrogenation of simple furans 18 catalysed by 1b in d8-THF ...... 76 Scheme 2.16 – One possible mechanism for 1b-mediated hydrogenation of simple pyrroles and furans ...... 77 Scheme 2.17 – Possible decomposition routes for 1/THF FLPs ...... 78 Scheme 2.18 – Factors likely to discourage THF ring-opening side-reactions ...... 79

Scheme 2.19 – Very low-level formation of short oligo(THF) compounds mediated by 1a,b under H2 80

Scheme 2.20 – Hydrogenation of 20a catalysed by 1b in d8-THF, and the presumed route for formation of side-product 22a ...... 82

Scheme 2.21 – Inhibition by products 21a and H2O of the 1b-catalysed hydrogenation of 20a in THF ...... 82 Scheme 2.22 – Hydrogenation of 20b catalysed by 1b in THF ...... 83 Scheme 2.23 – Slower hydrogenation of 20a catalysed by 1c in THF ...... 83 Scheme 2.24 – Slower hydrogenation of 20a catalysed by 1a in THF ...... 84 Scheme 2.25 – Effective hydrogenation of 20a catalysed by 1a in 1,4-dioxane ...... 84 Scheme 2.26 – Isotopic scrambling of HD due to reversible activation by 1a in 1,4-dioxane at R), as evidenced by 1H NMR spectroscopy ...... 86 Scheme 2.27 – Slower hydrogenation of 20a catalysed by 1b in 1,4-dioxane ...... 87 Scheme 2.28 – Selective hydrogenation of 20d in the presence of 20e, on the basis of differing steric bulk ...... 89 Scheme 2.29 – 1a-mediated hydrogenation of acetophenone (20j) in 1,4-dioxane, with subsequent dehydration to form styrene (23j) ...... 90 Scheme 2.30 – Proposed mechanism for 1a-catalysed hydrogenation of ketones (and possibly aldehydes) in 1,4-dioxane ...... 92 Scheme 2.31 – Stoichiometric experiments indicate that hydride transfer from [1aH]– to unactivated ketone substrates is infeasible, while transfer to 1a-activated ketones is too slow to account for the rate of the catalytic reaction, in which the concentration of [1aH]– must be very low ...... 93

Scheme 2.32 – Stoichiometric reduction of aldehydes 20k and 20n by [Bu4N][1aH] in the presence and absence of 1a ...... 95 Scheme 2.33 – Possible alternative mechanisms for 1a-catalysed hydrogenation of aldehydes in 1,4- dioxane ...... 96 Scheme 2.34 – Reversible deprotonation of intermediate 1a·ROH adducts as a likely limiting factor in the rate of 1a-catalysed carbonyl hydrogenation in 1,4-dioxane ...... 99 Scheme 2.35 – Greatly-improved rates of carbonyl hydrogenation catalysed by 1a in 1,4-dioxane at increased pressure ...... 100 Scheme 2.36 – 1a-catalysed hydrogenation of 20a in 1,4-dioxane at reduced temperature, through use of higher pressure ...... 100

Scheme 2.37 – A proposed explanation for the dramatically increased rate of 1a-catalysed carbonyl hydrogenation in 1,4-dioxane at increased pressure ...... 101 Scheme 2.38 – Catalytic or stoichiometric 1a-mediated carbonyl hydrogenation, depending on the choice of solvent ...... 102 Scheme 2.39 – The importance of relative turnover and decomposition rates in achieving 1a- catalysed carbonyl hydrogenation ...... 103 Scheme 2.40 – Mechanism for decomposition of 1a to generate borinic esters ...... 103 Scheme 2.41 – Reduced rate of 1a decomposition in 1,4-dioxane due to borane-alcohol adduct deprotonation ...... 104 Scheme 2.42 – A more general possible mechanism for H+-mediated 1a decomposition, which would be disfavoured by levelling to 1,4-dioxane ...... 104 Scheme 2.43 – A possible alternative mechanism for 1a decomposition based on aryl group migration, which would be disfavoured by more extensive Lewis acid quenching ...... 105 Scheme 2.44 – Unsuccessful hydrogenation of ester 24a using 1a in 1,4-dioxane ...... 106 Scheme 2.45 – Inferior hydrogenation of aldehyde 20n catalysed by 1a in 2-MeTHF ...... 107 Scheme 2.46 – Low-temperature hydrogenation of aldehyde 20n catalysed by 1a in 2-MeTHF, which is believed to be possible due to a lack of strong 1a-quenching interactions ...... 108 Scheme 2.47 – Less-successful hydrogenation of ketone 20a catalysed by 1b in 2-MeTHF ...... 109 Scheme 2.48 – Low-temperature hydrogenation of ketone 20a catalysed by 1e in 2-MeTHF ...... 110 Scheme 2.49 – Ester hydrogenation in 2-MeTHF mediated by boranes 1b,e ...... 110

Scheme 2.50 – FLP-catalysed hydrogenation of 20a despite the presence of H2O ...... 115 Scheme 2.51 – Moisture-tolerance in FLP-catalysed hydrogenation facilitated by addition of a desiccant ...... 115

Scheme 2.52 – Improved rate of hydrogenation when using 1a·OH2 as a precatalyst, as a result of higher reaction pressure ...... 116

Scheme 2.53 – Tolerance of multiple equivalents of H2O (with respect to 1a) in the FLP-catalysed hydrogenation of 20a at 45 bar ...... 117 Scheme 2.54 – Proposed mechanism for 1a-catalysed hydrogenation of 20a in wet 1,4-dioxane ... 118

Scheme 2.55 – A possible alternative mechanism for H2 activation by 1a in wet 1,4-dioxane ...... 118 Scheme 2.56 – Dehydrative hydrogenation of 20j to 25j catalysed by 1a in 1,4-dioxane, despite formation of H2O as a stoichiometric by-product ...... 122 Scheme 2.57 – Proposed mechanism for 1a-catalysed hydrogenation of 20j to 25j in 1,4-dioxane . 1 2 2

Scheme 2.58 – Hydrogenation of carbonyl compounds catalysed by 1a in Et2O or iPr2O, as reported by Stephan and Mahdi ...... 124 Scheme 2.59 – Hydrogenation of alkyl carbonyl compounds catalysed by 1a and (D-CD) or 4 Å molecular sieves in PhMe, as reported by Stephan and Mahdi ...... 126 Scheme 2.60 – Hydrogenation of aryl alkyl ketones catalysed by 1a and 4 Å molecular sieves in PhMe, as reported by Stephan and Mahdi ...... 126

Scheme 2.61 – Hydrogenation of diaryl ketones catalysed by 1a and 4 Å molecular sieves in PhMe, as reported by Stephan and Mahdi ...... 126 Scheme 2.62 – Mechanism for benzyl alcohol deoxygenation during ketone hydrogenation catalysed by 1a and 4 Å molecular sieves in PhMe, proposed by Stephan and Mahdi ...... 127

Scheme 2.63 – Hydrogenation of ketones catalysed by 1a and [NMe4][1aH]in PhMe, as reported by Stephan and Mahdi ...... 127

Scheme 2.64 – Mechanism for hydrogenation of ketones catalysed by 1a and [NMe4][1aH] in PhMe proposed by Stephan and Mahdi; and an alternative possibility ...... 128 Scheme 2.65 – Hydrogenation of 4-heptanone catalysed by 1a in 2,4-dimethylpentan-3-ol reported by Stephan and Mahdi, which indicates the feasibility of H2 activation by 1a/ROH Lewis pairs ...... 128 Scheme 2.66 – Moisture-tolerant hydrogenation of carbonyl compounds reported by Soós et al. using a novel borane in THF; and control experiments showing no reactivity for 1a-based systems under equivalent conditions, despite such reactivity in fact being possible ...... 130 Scheme 2.67 – Reductive amination using a hydrosilane reductant catalysed by 1a, as reported by Ingleson et al...... 131 Scheme 2.68 – Dependence on Brønsted basicity of the outcome of 1a-catalysed reductive amination, as reported by Ingleson et al...... 131 Scheme 2.69 – A simplified summary of elementary reaction steps that may be important during 1a- catalysed carbonyl hydrogenation in 1,4-dioxane ...... 133

Chapter 3

Scheme 3.1 – Organic carbonyl reduction by trialkyltin hydrides in a protic medium; or catalysed by a trialkyltin LA in an aprotic medium ...... 142

Scheme 3.2 – Successful dehydrogenation of Me2NH·BH3, but unsuccessful activation of H2, as reported by Manners et al. for a [26a]OTf/TMPH FLP ...... 143 Scheme 3.3 – Synthesis of [26b]OTf ...... 144 Scheme 3.4 – Examples of side-reactions that would be expected for a radical-mediated [26b]OTf- catalysed hydrogenation of bromoaryl imine 2i ...... 154 Scheme 3.5 – Proposed mechanisms for [26b]OTf-catalysed hydrogenation of imines ...... 155 Scheme 3.6 – Stoichiometric reactivity of [26b]H with imine 2g alone; following addition of HOTf; and following addition of [26b]OTf ...... 156 Scheme 3.7 – Hydrogenation of 20j catalysed by [26b]OTf and col, producing a mixture of products ...... 159 Scheme 3.8 – Proposed mechanism for [26b]OTf-based catalytic hydrogenation of aldehydes and ketones ...... 160 Scheme 3.9 – Stoichiometric reactivity of [26b]H with ketone 20a alone; following addition of [26b]OTf; and following addition of [colH]OTf ...... 162 Scheme 3.10 – An alternative mechanism for [26b]OTf-based catalytic hydrogenation of aldehydes and ketones ...... 162

Scheme 3.11 – Hydrogenation of 20a catalysed by [26b]OTf and [26b]OiPr, with the latter generated in situ from 20a and [26b]H ...... 163 Scheme 3.12 – Hydrogenation of enamine 27, catalysed by [26b]OTf and col; and acridine, 4, catalysed by [26b]OTf only ...... 164 Scheme 3.13 – Hydrogenation of an activated C=C bond catalysed by [26b]OTf and col ...... 164 Scheme 3.14 – Absorption of atmospheric moisture by [26b]OTf; as indicated by 1H NMR spectroscopy for a DCB solution also containing col and 10 eq. acetone, 20a ...... 165

Scheme 3.15 – Tolerance of added H2O in the [26b]OTf-based catalytic hydrogenation of model ketone 20a ...... 168 Scheme 3.16 – [26b]OTf-based catalytic carbonyl hydrogenation reactions prepared on the open bench ...... 169 Scheme 3.17 – Moisture-tolerant [26b]OTf-catalysed hydrogenation of imine 2g; whose rate is proposed to be limited due to deprotonation of the hydrated LA by the basic product 3g, and subsequent condensation of the stannol [26b]OH ...... 170 Scheme 3.18 – Moisture-tolerant [26b]OTf-based hydrogenation of imines 2d,e ...... 171 Scheme 3.19 – [26b]OTf-based hydrogenation of imine 2d, inhibited by alcohol 21a ...... 172

Scheme 3.20 – Hydrogenative amination catalysed by [26b]OTf and col, despite formation of H2O as a stoichiometric by-product ...... 173

Scheme 3.21 – Synthesis of [26b]NTf2 ...... 174

Scheme 3.22 – Hydrogenation of ketone 20a at a significantly improved rate using [26b]NTf2 ...... 176

Scheme 3.23 – Inferior hydrogenation of imine 2g catalysed by [26b]NTf2 ...... 177

F Scheme 3.24 – Salt metathesis between [26b]OTf and K[Al(OR )4] in the absence and presence of an additional LB ...... 178

F Scheme 3.25 – Hydrogenation of ketone 20a catalysed by [26b·col][Al(OR )4] ...... 179

F Scheme 3.26 – Unsuccessful H2 activation by [26b·LB][Al(OR )4] in DFB (LB = col, DABCO) ...... 179 Scheme 3.27 – A qualitative model to explain the differing ability of LB/[26b]+ Lewis pairs (LB = col,

DABCO)to activate H2, depending on the counteranion employed; and coordination of triflate to [26b←DABCO]+, indicated by 119Sn{1H} NMR spectroscopy ...... 181

F Scheme 3.28 – Successful H2 activation by [26b·LB][Al(OR )4] and an additional equivalent of LB (LB = col, DABCO) in DFB; and the proposed mechanism for these reactions ...... 182 Scheme 3.29 – To date, attempts at [26f]OTf-based catalytic ketone hydrogenation have been unsuccessful; while attempts using [26a]+-based systems have succeeded, but are believed to proceed via decomposition to form as-yet-unidentified catalytic intermediates ...... 184

List of Tables

Chapter 2

Table 2.1 – Imine hydrogenation catalysed by boranes 1 in ethereal solvents ...... 66 Table 2.2 – Carbonyl hydrogenation catalysed by 1a in 1,4-dioxane ...... 87 Table 2.3 – Hydrogenation under ‘open bench’ conditions catalysed by 1a in 1,4-dioxane ...... 119

Chapter 3

Table 3.1 – Gutmann-Beckett Lewis acidity values for some LAs relevant to Chapter 3 ...... 148

Table 3.2 – H2 activation by [26b]OTf and LBs of various strengths in DFB ...... 151 Table 3.3 – Hydrogenation of imines catalysed by [26b]OTf with or without an auxiliary base ...... 153 Table 3.4 – [26b]OTf-based catalytic hydrogenation of organic carbonyl compounds ...... 158

+ Table 3.5 – Key crystallographic bond lengths and angles for some [26·2H2O] salts ...... 167

Table 3.6 – A summary of key crystallographic bond lengths and angles for [26b]NTf2, with values for

[26b·2H2O]OTf also shown for comparison ...... 176

Chapter 1 – Introduction

Chapter 1 - Introduction

Chapter 1.1 - Putting FLPs in context: main group mimics of transition metals

The importance of catalysis in chemistry, both industrially and in academia, is hard to overstate, with crucial uses found in all fields and up to the very largest scales of production (Figure 1.1; a).1 Catalysis in solution (homogeneous catalysis) has the advantage of being able to use well- defined molecular catalysts that can be easily monitored and characterised by standard spectroscopic techniques, and which can be rationally modified and optimised to show desired reactivity and selectivity.2 This renders these catalysts particularly attractive in the preparation of pharmaceuticals, agrochemicals, and other value-added compounds (Figure 1.1; b).3

Figure 1.1 – (a) The Haber-Bosch4 (left) and Fischer-Tropsch5 (right) processes: examples of highly important industrial heterogeneous catalytic reactions. (b) Significant examples of homogeneous catalysis in industry: the Wacker process for ethylene oxidation to acetaldehyde (top left);6 the Cativa process for acetic acid production (top right);7 asymmetric catalytic hydrogenation in the synthesis of the herbicide component (S)-metolachlor (bottom).8

In principle, homogeneous catalysis can be achieved using catalysts derived from anywhere in the periodic table. Nevertheless, despite significant recent advances in the field of organocatalysis,9 for example, catalysis in solution is largely synonymous with the use of transition metal (TM)

Chapter 1 – Introduction complexes (with the notable exception of simple Brønsted or Lewis acid/base catalysts).10-11 This is particularly true of reactions that require the activation of unreactive small molecules or chemical bonds as part of the catalytic cycle, with hydroformylation catalysed by [RhH(CO)(PPh3)3] providing a good illustrative example (Scheme 1.1).12

Scheme 1.1 – Mechanism for alkene hydroformylation catalysed by [RhH(CO)(PPh3)3] (simplified; geometric fluxionality not shown).12

The ability of TM complexes to effect such transformations is a consequence of their seemingly unique capacity to engage in small molecule and bond activation reactions that form the key elementary steps of a catalytic cycle (e.g. H2, CO and alkene activation in Scheme 1.1). This in turn can be attributed to a partially occupied set of valence d-orbitals, which allows for simultaneously electrophilic/Lewis acidic and nucleophilic/Lewis basic frontier orbitals to be located on the same TM centre, separated by only a small energy difference. It is the ability of a substrate to simultaneously interact with both orbitals that allows for the activation of kinetically inert functional groups, even if these functional groups would be inert to a purely Lewis acidic or purely Lewis basic centre. Perhaps the most famous illustration of this dual donor/acceptor reactivity lies in the binding of substrates such as C2H4 and CO. In both cases the interaction can be described using a simple model (known as the Dewar-Chatt-Duncanson model for alkenes)13-14 in which binding occurs via donation of electron density to an empty TM d-orbital from a filled C=E (E = C, O) bonding orbital, with concomitant back- donation from a filled TM orbital to an empty C=E S* antibonding orbital (Figure 1.22; a, b).14 Both

Chapter 1 – Introduction interactions lead to weakening of the C—E bond, and hence to activation of the substrate. Another example, of particular relevance to this thesis, is the binding of H2. Again, complexes are formed via a combination of TM←V(H2) and TM→V*(H2) interactions, both of which lead to H—H bond weakening (Figure 1.2; c).15 In cases where these interactions are sufficiently strong complete scission of the bond is observed, leading to formation of a TM(H)2 dihydride complex, and formal +2 oxidation of the TM centre.

13-15 Figure 1.2 – Frontier molecular orbital model for TM activation of (a) alkenes, (b) CO, (c) H2.

Because the useful reactivity of TM complexes can be traced back to the occupancy of their frontier orbitals, much attention in recent decades has been directed towards the synthesis and investigation of main group compounds that possess similar electronic configurations, in the hope that they might be capable of demonstrating similar reactivity,16 and ultimately even replacing TMs in many catalytic applications (particularly where the high cost, toxicity and scarcity of many of the most active TMs can present significant problems).17 Probably the most well-known such systems are the

2 persistent singlet carbenes, R2C:, such as N-heterocyclic carbenes (NHCs), which possess both an sp lone pair and an empty p orbital located on the same carbon atom.18 This clearly mimics the electronic structure commonly found in TM complexes, and in seminal work by Bertrand et al. it was shown that stable carbenes of the type :C(NR2)R’ are consequently able to react directly with H2 (Figure 1.3; a), with computational studies supporting a similar mechanism to oxidative addition at a TM centre (Figure 1.3; b, c.f. Figure 1.2; c).19 Equivalent reactivity has subsequently been reported for a number of other ‘metallylenes’ and related coordinatively unsaturated p-block species, as have other TM-like reactions such as oxidative addition of various E—H bonds (E = N, O, Si, B, etc.).20 More reactive,

Chapter 1 – Introduction transiently-generated carbenes are also known to engage in numerous other bond insertion reactions such as cyclopropanation and C—H insertion.21-22

Figure 1.3 – (a) TM-like H2 activation by a persistent carbene and (b) a frontier molecular orbital model for the reaction; (c) selected other examples of TM-like reactivity of metallylenes.23-25

Related TM-like reactivity such as H2 cleavage and alkene activation has also been observed for a number of unsaturated p-block E—E multiply-bonded species (Figure 1.4).20, 26 Again, this is attributed to spatial and energetic closeness between the key frontier molecular orbitals located on the unsaturated moiety (Figure 1.4; b).26-28

Chapter 1 – Introduction

Figure 1.4 – (a) Examples of H2 and ethylene activation across E—E multiple bonds [Ar = 2,6-(2,6-iPr2-

C6H3)2-C6H3, Ar’ = 2,6-iPr2-C6H3, R = CCtBu]; (b) frontier molecular orbital models for initial H2 activation by a distannyne (left) and a reduced diboraanthracene (right).26-29

Chapter 1.2 - Introduction to frustrated Lewis pair chemistry

Chapter 1.2. 1 – Stoichiometric FLP reactivity

Despite impressive advances in the preparation, isolation and characterisation of unsaturated heavy p-block compounds and the realisation that their unusual electronic structures permit TM-like bond activation chemistry, the typically high reactivity and low stability of these species, and concomitant difficulties in their preparation and handling, has led to great difficulty in the development of catalytic applications. Possibly as a result, the few examples reported thus far have tended simply to exploit the Lewis acidity of these compounds, rather than more TM-like bond activation steps,20 although exceptions do exist.30

Fortunately, in 2006 Stephan et al. reported results that would lead to a valuable new approach for granting TM-like reactivity to relatively simple p-block compounds. The authors reported that addition of an H2 atmosphere to a solution of an intramolecular phosphine-borane leads reversibly to formation of the related zwitterionic phosphonium borohydride salt (Scheme 1.2).31 This

Chapter 1 – Introduction

reaction involves TM-like activation of the relatively inert, apolar H—H bond, cleaving H2 into formal ‘H+’ and ‘H–’ fragments, and was quickly generalised first to other intermolecular phosphine/borane pairs,32 and then to a much wider range of Lewis acid (LA; e.g. borenium and silylium cations)33-35 and Lewis base (LB; e.g. amines, NHCs)36-38 combinations.39-43

31 Scheme 1.2 – Reversible H2 activation by a phosphine/borane, reported by Stephan et al. in 2006.

The observed reactivity can be attributed to the inability of the LA/LB pairs employed to quench each other through formation of the adduct that would be expected classically; a result of their very high combined steric bulk. As a result these combinations, which have come to be known as ‘frustrated Lewis pairs’ (FLPs),44 possess both a high energy donor orbital (located on the LB; typically a lone pair) and a low energy acceptor orbital (located on the LA). The combined interaction

45-46 of these orbitals with the H2 V and V * orbitals can then lead to H—H cleavage, in a manner again clearly reminiscent of activation by TMs (Figure 1.5; b). Note that an alternative model for H2 activation by FLPs has also been proposed, in which an electric field established between the LA and LB polarises the H—H bond and thereby facilitates its cleavage (Figure 1.5; c).47 However, this explanation has largely fallen out of favour.48

As well as H2, FLPs have been found to be capable of activating a plethora of other small molecules and functional groups, many of which are kinetically inert and unreactive towards either acids or bases in isolation. Archetypal examples of FLP reactions include ring opening of cyclic ethers such as THF; E—H bond activation reactions; and addition across C—C and C—O multiple bonds, and small polar molecules such as nitrogen oxides (Figure 1.6).49-50 Furthermore, FLP reactivity has been observed using LAs (and, to a lesser extent, LBs) from throughout the periodic table,51 indicating very general applicability of the FLP concept.

Chapter 1 – Introduction

Figure 1.5 – H2 activation by FLPs (a), which can be explained using a molecular orbital-based model (b), or a now-disfavoured electric field-based model (c).45-47

Figure 1.6 – Some representative examples of stoichiometric P/B FLP reactivity.52-58

Chapter 1 – Introduction

It is worth noting, however, that the boundaries separating the newly-formalised field of FLP chemistry from other, more established, areas of research are not always sharp or easy to discern. For example, Melen and co-workers have recently reported the PhSeCl-induced cyclisation reaction shown in Scheme 1.3.59 This reaction is presumed to proceed via cooperative activation of the alkyne moiety by the Lewis acid, and intramolecularly by the weakly basic ester functionality, and is therefore described by the authors as an example of FLP reactivity. Nevertheless, similar cyclisations have been described without reference to the FLP concept,60 and whether it will prove useful or desirable to move these (and many other) reactions under the heading of FLP chemistry remains to be seen. This point will be touched upon again in Chapter 1.3.

Scheme 1.3 – (a) An ‘FLP’ cyclisation reported by Melen et al. and (b) a similar Lewis-acid-catalysed cyclisation – also an example of FLP chemistry?59-60

Chapter 1.2.2 – Catalytic FLP reactivity

Despite the similarities in reactivity and electronic structure, FLPs differ from other unsaturated p-block species (as discussed in Chapter 1.1) in a number of significant ways. Most notably, while both trace their interesting reactivity to a combination of energetically accessible donor and acceptor orbitals, in FLPs these orbitals are spatially separated, forming parts of discrete functional groups. This separation makes it relatively easy to tune and modify either one of these key orbitals independently of the other (e.g. Lewis basicity can readily be increased without changing Lewis acidity). Just as importantly, the functional groups most commonly used to construct FLPs (amines, phosphines, boranes) are well understood, readily incorporated into more complex structures, and chemically robust under a range of reaction conditions. Collectively, these factors imply that FLPs should be well suited to the development of catalytic reactions. In fact, just a year

31 after the first account of FLP H2 activation by Stephan and co-workers was published, the same authors were able to report the catalytic hydrogenation of simple, bulky imines using the same

Chapter 1 – Introduction

61 intramolecular P/B FLP (added as the H2 activation salt, Figure 1.7; a). While a handful of previous researchers had described very limited examples of homogeneous TM-free catalytic hydrogenation under extremely forcing conditions,62-66 this was the first example of a reaction of this type occurring under such moderate conditions. In subsequent years, rational modification of the LA and LB employed has allowed for extension of the FLP-catalysed hydrogenation methodology to a broad range of other unsaturated substrates (Figure 1.7; b). FLP chemistry has thus provided the first generally-applicable approach to TM-free catalytic hydrogenation,49-50 as will be discussed in greater detail in Chapter 1.4.

Figure 1.7 – (a) The first example of FLP-catalysed hydrogenation; (b) a selection of compounds

67-73 accessible through FLP-catalysed hydrogenation reactions (highlighted H atoms from H2).

In addition to hydrogenation, FLPs have found applications as catalysts for an ever-expanding range of other catalytic reactions.49-50 Many of these reactions are conceptually similar to hydrogenations: for example, transfer hydrogenation, dehydrogenative oxidation and dehydrocoupling, hydroborylation, and hydrosilylation reactions (Scheme 1.4). 74-77 Probably the most notable such reactions are Piers-type hydrosilylation reactions (discussed further in Chapter 1.3),

78-79 catalytic CO2 reductions, and a seminal report by Fontaine et al. of TM-free C—H/B—H cross- coupling.80 In addition, there have been reports of FLP-catalysed hydroamination,81 polymerisation,82 cyclisation,83 iodoperfluoroalkylation84 and Mannich-type reactions (Scheme 1.4).85 As with stoichiometric FLP reactivity (Chapter 1.2.1), the boundary with other areas of chemistry (particularly Lewis acid catalysis) is not always distinct. For example, Stephan et al. have recently reported LA- catalysed hydrothiolation reactions which, although not described as such, bear clear conceptual and

Chapter 1 – Introduction mechanistic similarities to related hydroelementation (e.g. hydrosilylation) reactions, for which the FLP concept is routinely invoked.86

Scheme 1.4 – A range of examples of FLP-catalysed reactions.74, 77-78, 80-85

Chapter 1 – Introduction

Chapter 1.3 – Some historical context for FLP chemistry

One of the most remarkable aspects of the FLP concept is its simplicity: fascinating reactivity can be observed using very familiar reagents simply by increasing their steric bulk to the point that strong classical adduct formation is no longer observed. Given this, along with the broad range of acids and bases to which the concept can be applied, it seems remarkable that its discovery and formalisation have occurred only so recently, and begs the question: “why?” After all, the knowledge that extreme steric bulk can inhibit classical LA/LB adduct formation is almost as old as Lewis’ concept of acids and bases itself, having been reported as early as 1942 (Scheme 1.5).87-88

Scheme 1.5 – An early example of LA/LB adduct formation precluded by steric factors.88

In fact, a re-inspection of the literature suggests that FLP chemistry has come close to being

31 discovered on a number of occasions prior to Stephan’s 2006 report of FLP H2 activation. In perhaps the most dramatic example, Berkessel and co-workers reported investigations into a reaction previously described by Walling and Bollyky, in which strong alkoxide bases are able to catalyse the hydrogenation of a small number of non-enolisable ketones at very high temperatures and pressures.62-63 After thorough analysis, the authors concluded that the reaction proceeds via the mechanism shown in Scheme 1.6, and described how the H—H bond is cleaved through ‘the joint action of a […] base and a Lewis-acid […] on the H2 molecule’. Only some years later has the generality of this mechanism come to be realised.

Scheme 1.6 – Alkoxide-catalysed ketone hydrogenation, which proceeds via an FLP-like mechanism.62-63

A more thoroughly developed example can be found in the hydrosilylation chemistry

76, 89 developed by Piers and others from the late 1990s onwards. In these reactions B(C6F5)3 (or related LAs) is used to catalyse the addition of hydrosilanes to carbonyls and other functional groups. Through a series of insightful mechanistic investigations, it was shown that these reactions proceed via LA

Chapter 1 – Introduction activation not of the substrate, but of the Si—H bond, which is then cleaved through nucleophilic attack by the substrate (Scheme 1.7).90 Were this reaction to be discovered today it would undoubtedly be described as proceeding via FLP activation of the Si—H bond; indeed, Piers has argued that ‘Si—H bond activation triggered by B(C6F5)3 in the presence of oxygen and nitrogen bases […] is a direct antecedent of the metal-free H2 activations by what have come to be known as “frustrated Lewis pairs”’.89 Again, it is only more recently that the generality of this method for bond activation has come to be appreciated.

Scheme 1.7 – Piers-type carbonyl hydrosilylation, which proceeds via an FLP Si—H bond activation mechanism.90

Other prominent areas that have foreshadowed the development FLP chemistry include the use of dual acid/base catalysis in organic chemistry,85, 91 and the development of bifunctional metal- ligand systems for catalytic bond activation reactions,92 among other more specific examples.49-50 Finally, within the last year Ozin has highlighted the connection between FLP chemistry and the chemistry of solid surfaces, where cooperative bond activation by neighbouring acidic and basic sites is well known (Scheme 1.8).93 Ozin argues that these surface reactions can therefore be considered as additional examples of FLP chemistry, although one could also argue the reverse: that FLP chemistry represents a translation of this concept into the homogeneous solution phase. In this context, it is interesting to note recent efforts by the groups of Taoufik and O’Hare, who have independently begun to investigate the chemistry of FLPs grafted onto silica surfaces (Wang, Guo et al. have also described the use of a gold surface as the LA component of an FLP).94-96

Chapter 1 – Introduction

93 Scheme 1.8 – FLP-like activation of H2 on a main group metal oxide surface.

Chapter 1.4 – Hydrogenation catalysis using FLPs

Of the numerous FLP-mediated bond activation reactions that have been reported in the past decade (Chapter 1.2.1), the one that has attracted the greatest interest is undoubtedly the cleavage

+ – of H2 into ‘H ’ and ‘H ‘. This is a result of both the ongoing novelty of being able to activate H2 without a TM, and (perhaps more importantly) the viability of incorporating this elementary step into a catalytic cycle (vide infra).

Chapter 1.4.1 – Stoichiometric FLP H2 activation

As discussed in Chapter 1.2.1., H2 activation is one of the archetypal FLP reactions, and has been described for a wide range of LA/LB pairs, with the LA ranging from neutral boranes and alanes to cationic boreniums, silyliums, phosphoniums, carbocations, zirconocenes and titanocenes; among others.50-51 The LB component, meanwhile, has included amines, phosphines, ethers, carbenes, carbanions, phosphazenes and silylenes.32, 35-38, 97-99 It is noteworthy that, in addition to their structural

100-102 diversity, the listed bases vary in strength by well over 20 pKa units, while calculated gas-phase hydride ion affinities for the acids range over more than 140 kcal/mol.103

Regardless of the precise LA/LB combination, H2 activation is believed to occur via cooperative interaction with the HOMO of the base and LUMO of the acid (Chapter 1.2.1.; Figure 1.5). For intramolecular FLPs the result is a bimolecular addition reaction (Scheme 1.9; a). For intermolecular FLPs, however, this description alone would suggest a termolecular reaction step, which would be entropically infeasible, and so suggests that there must be some form of transient interaction between two of the three reactants prior to the final H—H cleavage step. Initially, it was suggested that this might be a transient LB→H2 adduct, or a H2→LA species analogous to the SiH→LA complex found in related hydrosilylation reactions (Scheme 1.7).32 However, experimental and computational investigations into early FLP systems were unable to find evidence to support the existence of such intermediates (although LB-free H2 activation has been reported for exceedingly powerful, antiaromatic borole LAs).89, 109-110 Instead, it was proposed that, despite the absence of a strong dative

Chapter 1 – Introduction interaction, FLPs are capable of forming transient ‘encounter complexes’ held together by weak H- bonding and other non-covalent interactions.45-46 In these species the LA and LB partners are pre- organised in a way that allows for easy access by H2 into a central cavity, where it can then undergo activation. Support for the existence of such intermediates has recently been reported by Rocchigiani, Macchioni et al., using 2D-NMR correlation observations, although it should be stressed that this

111 direct experimental evidence exists only for two PR3/B(C6F5)3 FLPs.

32-33, 35-38, 97-99, 104-108 Figure 1.8 – Examples of LAs (left) and LBs (right) used in FLP H2 activation.

Alongside these kinetic aspects, the thermodynamics of the H2 activation reaction have also been studied. Particular insight was provided early on by Pápai and co-workers, who considered the

FLP H2 activation reaction as the energetic sum of five theoretical reaction steps: separation of the

+ – LA/LB pair, cleavage of H2 into H and H , attachment of these fragments to the LB and LA respectively, and ion pairing (or any other interaction) between the resulting [LB—H]+ and [LA—H]– (Figure 1.9).112

Chapter 1 – Introduction

Scheme 1.9 – Mechanisms for H2 activation by intramolecular (top) and intermolecular (bottom) FLPs.45-46, 111

112 Figure 1.9 – A thermodynamic cycle for FLP H2 activation.

Chapter 1 – Introduction

+ – Of the five steps, the energetic cost of one (H2 → H + H ) is independent of the FLP used. Meanwhile, extensive computational modelling of a large number of FLPs suggested that the ion pairing term ([LB—H]+ + [LA—H]– → [LB—H][LA—H]) does not vary significantly (though only neutral LAs and LBs were investigated). Since by design the LA/LB separation energy for an FLP must be small, this suggests that to a good approximation the thermodynamic ability of an FLP to cleave H2 is largely governed by the magnitude of the H+ + H– attachment terms (or, in other words, by the combined strength of the LB and of the LA). Experimentally-determined pKa values provide an extremely valuable proxy for the former and have been extensively tabulated (though they are highly dependent on the choice of solvent).113 By contrast, precise experimental data for the latter are far less readily available (at least for main group LAs);114 however, a large number of theoretical values have recently been reported by Heiden and Latham for FLP-relevant LAs,103 and qualitative links can often be drawn between hydride ion affinity and other experimental measures of Lewis acidity such as Gutmann- Beckett values.115 With few exceptions, Pápai’s analysis has been found to correspond very well with experimental observations.

Chapter 1.4.2 – Catalytic FLP H2 activation

Given that H2 activation by FLPs generates reagents such as borohydrides, silanes and alanes which are well known for their reducing properties, it is understandable that this reaction step has since been incorporated into a number of catalytic hydrogenation cycles. What is perhaps surprising, however, is the contrast between the wide scope of the stoichiometric FLP H2 activation reaction, and the fact that investigations into catalytic hydrogenation using FLPs have thus far focused overwhelmingly on a rather narrow range of LAs, primarily based on boron. In fact, there have only been five reports describing successful FLP-catalysed hydrogenation not using B-based LAs (excluding the results described in Chapter 3), only two of which are TM-free. These involve reduction of various substrates by zirconocene-based FLPs;116-118 of a narrow range of alkenes by a P(V) LA;105 and of a small number of imines by iBu2AlH (Scheme 1.10; although this last system was not described as an example of FLP chemistry in the initial publication, the authors subsequently included it in a review of the subject).51, 119

The focus on B-based LAs is compounded by a relative lack of variation even among these

33, 99, 120-122 catalysts; aside from a few examples using borenium ions and one using weak BR3 LAs, this chemistry has been performed exclusively using B(C6F5)3 (which has become the archetypal FLP LA), and other, closely-related, highly electrophilic arylboranes (Figure 1.10).

Chapter 1 – Introduction

Scheme 1.10 – Rare examples of boron-free FLP-catalysed hydrogenation.105, 116-119

Figure 1.10 – A variety of highly-electrophilic, bulky arylboranes (a),67, 70, 123-126 and other B-based LAs (b) that have been used in FLP hydrogenation catalysis.33, 99, 120-121

Chapter 1 – Introduction

Despite this seemingly limited approach, FLP-catalysed hydrogenation has proven to be applicable to a wide variety of different unsaturated substrates. While early reports primarily described the reduction of basic, polar substrates such as imines (and the related enamines), aziridines and N-heterocycles,43 the methodology has subsequently been expanded to cover less basic substrates such enol ethers,68 enones,126 and nitroolefins;70 and even non-polar hydrocarbons such as alkenes, alkynes and aromatics (Figure 1.7; b).127

Although detailed kinetic analysis of these reactions is still in its infancy,128-135 the vast majority of reported hydrogenations are believed to proceed via a common mechanism in which H2 activation by the FLP catalyst is followed by sequential transfer of H+ and H– to the substrate, which generates the reduced product and releases the free FLP (Scheme 1.11; a).43, 131-133, 135 For some basic substrates, hydrogenation can be performed without the addition of an auxiliary LB, with the substrate itself becoming the basic component of the FLP. In these cases a slightly modified mechanism is followed (Scheme 1.11; b).136 The tendency for H+ transfer to precede that of H– can largely be attributed to the high strength of the most common FLP LAs. This results in relatively stable hydrides with low reducing power, necessitating pre-activation of the substrate via protonation. Nevertheless, it is easy to imagine an alternative mechanism where the order of H+ and H– transfer is reversed (Scheme 1.11; c). To date this is only believed to occur for the hydrogenation of some nitroolefins and related substrates, such as acrylates.70, 124, 137 Another possibility involves Lewis, rather than Brønsted acid activation of the substrate (Scheme 1.11; d: this mechanism has to date only been proposed for reactions using stoichiometric LAs, but will be of relevance to Chapter 3 of this thesis).61 A few FLP- catalysed hydrogenations have been proposed to proceed via slightly different mechanisms still; these are summarised in Scheme 1.12.71, 117, 119, 138

Chapter 1 – Introduction

Scheme 1.11 – (a) The general mechanism for FLP-catalysed hydrogenations to date (for a generic substrate X=Y that is basic at Y); (b) a variation for basic substrates; and alternative mechanisms based on (c) initial hydride transfer or (d) LA activation of the substrate.

Chapter 1 – Introduction

Scheme 1.12 – FLP-catalysed hydrogenations that proceed via atypical mechanisms.71, 117, 119, 138

Chapter 1 – Introduction

Chapter 1.4.3 – Aspects of FLP design

Based on an understanding of the mechanisms of both stoichiometric H2 activation and catalytic hydrogenation by FLPs (vide supra), it is possible to identify a number of general factors relevant to the design and analysis of potential FLP hydrogenation catalysts:

x Importance of LA, LB strength

As described in Chapter 1.4.1, the thermodynamic ability of an FLP to cleave H2 largely correlates with the combined strength of the LA and LB (Figure 1.9). As a result, catalytic hydrogenation is likely to be infeasible if the LA and LB used are both very weak. Conversely, if both components are very strong then H2 activation is expected to be facile; however, the result will be a relatively unreactive FLP·H2 product, making further reaction unlikely (Figure 1.11).

Figure 1.11 – Qualitative effect of cumulative LA + LB strength on FLP hydrogenation catalysis.

From the general mechanism of FLP-catalysed hydrogenation (Scheme 1.11; a), it can be seen that it is also important to tailor the strength of the LB to approximate that of the substrate, in order

+ to ensure that H transfer is feasible subsequent to H2 activation. This was neatly demonstrated in the first report of FLP-catalysed hydrogenation of alkenes;69, 133 by replacing more conventional, strong

133 phosphine LBs with much weaker bases such as P(C6F5)Ph2 (pKaH = 2.56 in MeCN), the authors ensured that H2 activation generates a suitably powerful Brønsted acid to protonate the weakly basic substrate (Scheme 1.13; a). Conversely, though, this requirement can lead to problems if the product of a hydrogenation reaction is significantly more basic than the starting material. Stephan et al. have reported that B(C6F5)3 is capable of mediating the reduction of bulky anilines to cyclohexylamines via

139 an FLP mechanism. Unfortunately, further H2 activation then generates secondary alkylammonium

Chapter 1 – Introduction borohydride salts that are insufficiently acidic to protonate a second equivalent of substrate

113, 140 (pKaH = ca. 11, 3 respectively in DMSO), preventing any catalytic turnover (Scheme 1.13; b).

Scheme 1.13 – (a) Catalytic alkene hydrogenation achieved through careful choice of LB;69, 133 (b) stoichiometric hydrogenation of an aniline.139

x Importance of steric bulk (‘thermally induced frustration’)

Alongside acid/base strength, steric bulk is a key parameter for FLP systems. Central to FLP chemistry is the idea that a certain minimum level of steric bulk is necessary to prevent irreversible formation of a strong, unreactive classical adduct between the LA and LB. Nevertheless, it has become well established that adduct formation per se need not prevent FLP reactivity, provided it is reversible under the relevant reaction conditions. In particular, LA/LB pairs that form a classical adduct at room temperature may dissociate to form an active FLP at elevated temperatures; this has come to be known as ‘thermally-induced frustration’, and expands the range of systems for which FLP reactivity can be observed to those based on less bulky LAs and LBs (Figure 1.12; left).141-146 While the need for adduct dissociation can be expected to render H2 (or other bond) activation reactions slightly less favourable (the LA/LB separation term in Figure 1.9 is no longer negligible), sequestration of a highly reactive FLP as a more stable adduct can render it more robust. For example, Tamm et al. have reported that, while an NHC/B(C6F5)3 FLP undergoes rapid decomposition to an unreactive product at

Chapter 1 – Introduction

room temperature, the related classical adduct formed from the same NHC and B(FXyl)3 remains stable up to much higher temperatures, whilst retaining its FLP reactivity (FXyl = 3,5-trifluoromethylphenyl) (Scheme 1.14; a).38, 146

Figure 1.12 – Qualitative effect of steric bulk on FLP hydrogenation catalysis.

In principle, it is also possible for a classical LB→LA adduct to display FLP-like reactivity without the need for full dissociation. It has been reported that the strong adduct [iPr3Si←PtBu3][B(C6F5)4], which does not dissociate to form [iPr3Si][B(C6F5)4] + PtBu3 even at elevated temperature, is nevertheless capable of activating H2 as a result of transient lengthening and weakening of the Si←P interaction, which generates a structure similar to a conventional FLP encounter complex (Scheme 1.14; b).34 Alternatively, an analogy can be drawn with direct hydrogenolysis of polar single bonds, which has been incorporated in TM-free catalytic alkene hydrogenation (Scheme 1.14; c).147

Increasing the steric bulk of the LA and LB appreciably beyond the threshold necessary for FLP formation can also have significant consequences. Soós et al. have demonstrated that increasing the steric bulk of the LA can allow for the hydrogenation of less bulky substrates by inhibiting LA←product

123, 126, 148 adduct formation (the ‘size exclusion principle’, Figure 1.12; right). Conversely, however, excessive bulk may eventually begin to inhibit even H2 activation, while extremely bulky hydrides may not be suitable for reduction of very hindered substrates.43, 123

Chapter 1 – Introduction

Scheme 1.14 – (a) Improved stability of a borane/NHC Lewis pair through reversible adduct formation;38, 146 (b) a classical silylium/phosphine Lewis adduct that displays FLP-like reactivity;34 (c) TM-free catalytic alkene hydrogenation based on C—B hydrogenolysis.147

Chapter 1 – Introduction

x Intermolecular vs. intramolecular FLPs

By tethering the LA and LB together with a suitable linker to form an intramolecular FLP, H2 activation is transformed from a termolecular to a bimolecular process. The correspondingly decreased entropic barrier to the reaction might intuitively be expected to lead to more active optimised hydrogenation catalysts; indeed, in 2008 Erker et al. reported imine hydrogenation catalysed by Mes2PCH2CH2B(C6F5)2, which proved to be considerably more active than previously- reported P/B systems (note, though, that electronic and steric differences also need to be considered when comparing specific examples).67 Nevertheless, in practice the reactivity of intramolecular FLPs has been found to vary dramatically with the choice of linker. Among the most extensively-studied examples is the family of intramolecular P/B systems Mes2P(CH2)nB(C6F5)2 (n = 2-4), which show

67, 149-150 dramatically different reactivities towards H2 (Figure 1.13; a). In this case, the variation is attributed to differences in the strength of an internal P/B interaction. Another example is provided by the initial P/B system reported by Stephan et al..31 In this FLP, the inflexibility of the central tetrafluorophenylene linker prevents the phosphine and borane moieties of a single molecule from coming into close proximity with one another; this system must therefore activate H2 in an inter- rather than intramolecular fashion (Figure 1.13; b).46 The potential for higher activity with optimised intramolecular FLPs must thus be weighed against both the additional complexity provided by identifying the optimal linker (which will also affect the steric and electronic properties of the LA and LB), and the increased synthetic difficulty of accessing these compounds compared with analogous intermolecular systems (the components of which are often commercially available); collectively these factors can act as a barrier to rational, iterative catalyst optimisation.

Figure 1.13 – (a) Differing reactivity of oligomethylene-bridged P/B FLPs;67, 149-150 (b) intermolecular

31, 46 H2 activation by an intramolecular FLP.

Chapter 1 – Introduction

Chapter 1.5 – Frontiers and limitations of FLP chemistry

In the decade since becoming an active area of research, remarkable advances have been made in the chemistry of frustrated Lewis pairs. Nevertheless, there remain a number of serious limitations to FLP methodologies, as might be expected for such a young field. Recognition of this fact has in recent years led to a number of attempts to move on from early ‘proof-of-principle’ FLPs towards more practical systems, attractive to the wider chemical community. For example, the groups of Stephan and Crudden have independently investigated families of carbene-stabilised borenium LAs for imine hydrogenation,120-121 and through thorough screening identified highly active catalysts capable of operating at room temperature, and at either atmospheric pressure, or at TM-like loadings as low as 0.1 mol% (in contrast to more common catalytic FLP loadings of ca. 5-20 mol%). Increasing attention has also begun to be paid to the development of asymmetric versions of existing FLP- catalysed hydrogenations,151-155 to potentially provide alternatives to their very popular TM-catalysed equivalents.156 A notable early example in this area was provided by Wei and Du, who described the highly enantioselective hydrogenation of prochiral silyl enol ethers; a reaction for which there is no direct TM-catalysed alternative (Figure 1.14).157 Beyond these examples, some of the most significant limitations of the FLP-catalysed hydrogenation methodology that remain to be overcome relate to functional group tolerance and substrate scope (vide infra);43 it is these that are of greatest relevance to this thesis.

Figure 1.14 – Examples of FLP development towards practical catalytic systems.33, 68, 121, 157

Chapter 1 – Introduction

Chapter 1.5.1 – FLP-mediated hydrogenation of carbonyl compounds

FLP-mediated reduction of organic C=O bonds using H2 as the terminal reductant has been known since the earliest days of the field. Erker et al. reported the first example in 2007, describing how addition of benzaldehyde to the zwitterionic salt [Mes2P(H)CH2CH2B(H)(C6F5)2] derived from H2 activation by Mes2PCH2CH2B(C6F5)2, leads to hydride transfer and formation of the reduced alkoxide

54 (Scheme 1.15). Similarly, the intermolecular FLP TMPH/B(C6F5)3 (TMP = 2,2,6,6- tetramethylpiperidyl) has been shown to be capable of mediating the hydrogenation of CO2 to the methanol oxidation level in substoichiometic yields.158 No catalytic turnover was reported for either reaction, however. In fact, despite the ubiquity and importance of carbonyl compounds throughout the chemical sciences, prior to the start of the work described in this thesis there had not been a single example of successful catalytic hydrogenation of a C=O group using an FLP method (although the C=C bonds of enones, acrylates and malonates have been hydrogenated, leaving the C=O bonds intact).70, 126

Scheme 1.15 – An early example of FLP-mediated carbonyl hydrogenation.54

Nevertheless, Nyhlén and Privalov have used theoretical methods to propose a mechanism for LA-catalysed carbonyl hydrogenation analogous to that for imine hydrogenation (Scheme 1.11; b) or Piers-type hydrosilylation (Scheme 1.7), in which the substrate acts as the basic component of the

159 FLP during H2 activation (Scheme 1.16; a). Unfortunately, experimental investigations by Repo et al. into this system yielded only substoichiometric quantities of the reduced products with concomitant decomposition of the LA, although this outcome did at least demonstrate the ability of

160-161 very weakly basic carbonyls to engage in FLP H2 activation. Stephan et al. subsequently expanded on these results, reporting a general stoichiometric reaction of B(C6F5)3 with alkyl aldehydes and

162 ketones under H2 to give reduced borinic esters (Scheme 1.16; b). The inability to observe turnover in these systems has been attributed to the instability of highly Brønsted acidic intermediates (e.g. intermediate i in Scheme 1.16; a), which prompts decomposition prior to catalytic turnover. More generally, the failure to observe catalytic carbonyl hydrogenation using early FLPs has been attributed to the high oxophilicity of the borane LAs employed, which leads to strong binding of oxygen donors;160, 162 certainly, addition of simple alcohols or carbonyl compounds has been found to have a strong inhibitory effect on other FLP-catalysed hydrogenation reactions (Scheme 1.16; c).43 The sensitivity of FLPs to hydroxylic species such as alcohols will be described in more detail in Chapter 2.

Chapter 1 – Introduction

Scheme 1.16 – (a) A mechanism for C=O hydrogenation proposed by Nyhlén and Privalov;159 (b)

162 stoichiometric C=O hydrogenation mediated by B(C6F5)3; (c) inhibition of FLP catalysis by oxygen- centred donors.43

Chapter 1.5.2 – Moisture tolerance in FLP chemistry

Given their general sensitivity to O-donors such as alcohols, it should not be surprising that early FLPs have also proven to be highly sensitive to moisture. To date, every reported example of successful FLP-catalysed hydrogenation has had to be performed under rigorously anhydrous conditions, using thoroughly dried reagents, solvents, and H2. Again, this intolerance has generally been attributed to the oxophilicity of the borane LAs employed (which will be discussed in additional detail in Chapter 2).107, 162 As a result, Ingleson et al. have investigated the use in FLP chemistry of the N-methylacridinium cation as a LA, which it was anticipated would display reduced oxophilicity.107

Consistent with expectations, it was found that this LA is capable of activating H2 as part of an FLP even in undried commercial solvents (Scheme 1.17). Nevertheless, significant hydrolysis was also observed, and no catalytic hydrogenation was reported, presumably in part due to the reportedly poor

H2 activation kinetics.

Chapter 1 – Introduction

107 Scheme 1.17 – H2 activation by an N-methylacridinium-based FLP, in the presence of moisture.

In separate work, Stephan et al. have investigated the use of chemical drying agents as additives to allow FLP hydrogenation reactions to be prepared under non-anhydrous conditions;163 while Fontaine et al. have described the development of bench-stable FLP pre-catalysts which can be used to generate catalytically active species in situ (although anhydrous reaction conditions are still required; Scheme 1.18).164

Scheme 1.18 – A bench-stable FLP pre-catalyst.164

Chapter 1.5.3 – Thesis aims

The aim of the work described in this thesis is to overcome some of the limitations suffered by early FLP hydrogenation catalysts. In particular, there are two specific goals:

x Achieve FLP-catalysed hydrogenation of organic carbonyl compounds

Carbonyl groups and their reduction products are some of the most commonly encountered functionalities in organic chemistry. Extending the substrate scope for FLP-catalysed hydrogenation to include C=O bonds can therefore be expected to significantly increase the appeal and utility of this methodology. Although methods for C=O reduction are very well established, these typically rely either on stoichiometric reagents (which produce stoichiometric quantities of waste), or expensive and scarce precious metal catalysts.165-166 While recent years have seen extensive efforts to develop

Chapter 1 – Introduction alternative catalysts based on cheaper, earth-abundant TMs,167-169 if FLP catalysts can be applied to the same transformations then their structural and mechanistic differences are likely to lead to useful differences in scope and selectivity.

x Achieve moisture tolerance in FLP-catalysed hydrogenation reactions

The moisture-sensitivity of FLP hydrogenation catalysts reported to date leads to significant practical difficulties in their application: an inert atmosphere must be maintained throughout the reaction; solvents must be thoroughly dried before use; and catalysts and other reactants must typically be stored in gloveboxes. These requirements combine to form a significant practical barrier to the uptake of FLP hydrogenation methods beyond the specialist community, and renders these reactions particularly unappealing for any industrial application. Improving moisture tolerance is therefore a crucial requirement for the development of more attractive FLP-catalysed hydrogenation reactions that can be prepared on the open bench.170-172

Note: for the sake of clarity, this chapter has not discussed the results of several reports by Stephan et al. into FLP-catalysed carbonyl hydrogenation, and Soós et al. into moisture-tolerant FLP-catalysed hydrogenation, published in 2014-2015.173-176 These accounts relate closely to similar results obtained during the work described in this thesis, which were published near-simultaneously.177-178 Their implications will instead be discussed in the conclusion to Chapter 2.

Chapter 1.6 – References for Chapter 1

1. Leach, B. E., In Applied Industrial Catalysis, Academic Press, 1983. 2. van Leeuwen, P. W. N. M.; Chadwick, J. C., In Homogeneous Catalysts, Wiley-VCH Verlag GmbH & Co. KGaA, 2011. 3. Busacca, C. A.; Fandrick, D. R.; Song, J. J.; Senanayake, C. H., Advanced Synthesis & Catalysis 2011, 353, 1825. 4. Appl, M., In Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH & Co. KGaA, 2000. 5. Jahangiri, H.; Bennett, J.; Mahjoubi, P.; Wilson, K.; Gu, S., Catalysis Science & Technology 2014, 4, 2210. 6. Jira, R., Angewandte Chemie International Edition 2009, 48, 9034. 7. Sunley, G. J.; Watson, D. J., Catalysis Today 2000, 58, 293. 8. Blaser, H. U., Advanced Synthesis & Catalysis 2002, 344, 17.

Chapter 1 – Introduction

9. List, B., Chemical Reviews 2007, 107, 5413. 10. Bell, R. P., In The Proton in Chemistry, Springer US, 1973. 11. Yamamoto, H., In Lewis Acids in Organic Synthesis, Wiley-VCH Verlag GmbH, 2008. 12. Hartwig, J. F., In Organotransition Metal Chemistry: From Bonding to Catalysis. University Science Books, 2010. 13. Frenking, G., In Modern Coordination Chemistry: The Legacy of Joseph Chatt, The Royal Society of Chemistry, 2002. 14. Cotton, F. A.; Wilkinson, G.; Murillo, C. A.; Bochmann, M., In Advanced Inorganic Chemistry. 6th ed., Wiley-VCH Verlag GmbH & Co. KGaA, 1999. 15. Kubas, G. J., Journal of Organometallic Chemistry 2001, 635, 37. 16. Power, P. P., Nature 2010, 463, 171. 17. Anastas, P. T.; Warner, J. C., In Green Chemistry: Theory and Practice. Oxford University Press, 1998. 18. Hopkinson, M. N.; Richter, C.; Schedler, M.; Glorius, F., Nature 2014, 510, 485. 19. Frey, G. D.; Lavallo, V.; Donnadieu, B.; Schoeller, W. W.; Bertrand, G., Science 2007, 316 , 439. 20. Yadav, S.; Saha, S.; Sen, S. S., ChemCatChem 2016, 8, 486. 21. von E. Doering, W.; Hoffmann, A. K., Journal of the American Chemical Society 1954, 76, 6162. 22. Schreck, J. O., Journal of Chemical Education 1965, 42, 260. 23. Protchenko, A. V.; Schwarz, A. D.; Blake, M. P.; Jones, C.; Kaltsoyannis, N.; Mountford, P.; Aldridge, S., Angewandte Chemie International Edition 2013, 52, 568. 24. Usher, M.; Protchenko, A. V.; Rit, A.; Campos, J.; Kolychev, E. L.; Tirfoin, R.; Aldridge, S., Chemistry – A European Journal 2016, 22, 11685. 25. Protchenko, A. V.; Bates, J. I.; Saleh, L. M. A.; Blake, M. P.; Schwarz, A. D.; Kolychev, E. L.; Thompson, A. L.; Jones, C.; Mountford, P.; Aldridge, S., Journal of the American Chemical Society 2016, 138, 4555. 26. von Grotthuss, E.; Diefenbach, M.; Bolte, M.; Lerner, H.-W.; Holthausen, M. C.; Wagner, M., Angewandte Chemie International Edition 2016, 128, 14273. 27. Peng, Y.; Brynda, M.; Ellis, B. D.; Fettinger, J. C.; Rivard, E.; Power, P. P., Chemical Communications 2008, 6042. 28. Peng, Y.; Ellis, B. D.; Wang, X.; Fettinger, J. C.; Power, P. P., Science 2009, 325, 1668. 29. Rit, A.; Campos, J.; Niu, H.; Aldridge, S., Nature Chemistry 2016, 8, 1022. 30. Auth, J.; Padevet, J.; Mauleón, P.; Pfaltz, A., Angewandte Chemie International Edition 2015, 54, 9542. 31. Welch, G. C.; Juan, R. R. S.; Masuda, J. D.; Stephan, D. W., Science 2006, 314, 1124.

Chapter 1 – Introduction

32. Welch, G. C.; Stephan, D. W., Journal of the American Chemical Society 2007, 129, 1880. 33. Farrell, J. M.; Hatnean, J. A.; Stephan, D. W., Journal of the American Chemical Society 2012, 134, 15728. 34. Herrington, T. J.; Ward, B. J.; Doyle, L. R.; McDermott, J.; White, A. J. P.; Hunt, P. A.; Ashley, A. E., Chemical Communications 2014, 50, 12753. 35. Schäfer, A.; Reißmann, M.; Schäfer, A.; Schmidtmann, M.; Müller, T., Chemistry – A European Journal 2014, 20, 9381. 36. Sumerin, V.; Schulz, F.; Nieger, M.; Leskelä, M.; Repo, T.; Rieger, B., Angewandte Chemie International Edition 2008, 47, 6001. 37. Chase, P. A.; Stephan, D. W., Angewandte Chemie International Edition 2008, 47, 7433. 38. Holschumacher, D.; Bannenberg, T.; Hrib, C. G.; Jones, P. G.; Tamm, M., Angewandte Chemie International Edition 2008, 47, 7428. 39. Stephan, D. W., Organic & Biomolecular Chemistry 2008, 6, 1535. 40. Stephan, D. W., Dalton Transactions 2009, 3129. 41. Stephan, D. W., Chemical Communications 2010, 46, 8526. 42. Stephan, D. W.; Erker, G., Angewandte Chemie International Edition 2010, 49, 46. 43. Stephan, D. W.; Greenberg, S.; Graham, T. W.; Chase, P.; Hastie, J. J.; Geier, S. J.; Farrell, J. M.; Brown, C. C.; Heiden, Z. M.; Welch, G. C.; Ullrich, M., Inorganic Chemistry 2011, 50, 12338. 44. Welch, G. C.; Cabrera, L.; Chase, P. A.; Hollink, E.; Masuda, J. D.; Wei, P.; Stephan, D. W., Dalton Transactions 2007, 3407. 45. Rokob, T. A.; Hamza, A.; Stirling, A.; Soós, T.; Pápai, I., Angewandte Chemie International Edition 2008, 47, 2435. 46. Guo, Y.; Li, S., Inorganic Chemistry 2008, 47, 6212. 47. Grimme, S.; Kruse, H.; Goerigk, L.; Erker, G., Angewandte Chemie International Edition 2010, 49, 1402. 48. Rokob, T. A.; Bakó, I.; Stirling, A.; Hamza, A.; Pápai, I., Journal of the American Chemical Society 2013, 135, 4425. 49. Stephan, D. W., Journal of the American Chemical Society 2015, 137, 10018. 50. Stephan, D. W.; Erker, G., Angewandte Chemie International Edition 2015, 54, 6400. 51. A., W. S.; W., S. D., Bulletin of the Chemical Society of Japan 2015, 88, 1003. 52. Dureen, M. A.; Brown, C. C.; Stephan, D. W., Organometallics 2010, 29, 6594. 53. Welch, G. C.; Masuda, J. D.; Stephan, D. W., Inorganic Chemistry 2006, 45, 478. 54. Spies, P.; Erker, G.; Kehr, G.; Bergander, K.; Frohlich, R.; Grimme, S.; Stephan, D. W., Chemical Communications 2007, 5072.

Chapter 1 – Introduction

55. Nie, W.; Klare, H. F. T.; Oestreich, M.; Fröhlich, R.; Kehr, G.; Erker, G., Zeitschrift für Naturforschung B 2012, 67, 987. 56. Mömming, C. M.; Otten, E.; Kehr, G.; Fröhlich, R.; Grimme, S.; Stephan, D. W.; Erker, G., Angewandte Chemie International Edition 2009, 48, 6643. 57. Sajid, M.; Klose, A.; Birkmann, B.; Liang, L.; Schirmer, B.; Wiegand, T.; Eckert, H.; Lough, A. J.; Frohlich, R.; Daniliuc, C. G.; Grimme, S.; Stephan, D. W.; Kehr, G.; Erker, G., Chemical Science 2013, 4, 213. 58. Mo, Z.; Kolychev, E. L.; Rit, A.; Campos, J.; Niu, H.; Aldridge, S., Journal of the American Chemical Society 2015, 137, 12227. 59. Wilkins, L. C.; Günther, B. A. R.; Walther, M.; Lawson, J. R.; Wirth, T.; Melen, R. L., Angewandte Chemie International Edition 2016, 55, 11292. 60. Hellal, M.; Bourguignon, J. J.; Bihel, F. J. J., Tetrahedron Letters 2008, 49, 62. 61. Chase, P. A.; Welch, G. C.; Jurca, T.; Stephan, D. W., Angewandte Chemie International Edition 2007, 46, 8050. 62. Walling, C.; Bollyky, L., Journal of the American Chemical Society 1964, 86, 3750. 63. Berkessel, A.; Schubert, T. J. S.; Müller, T. N., Journal of the American Chemical Society 2002, 124, 8693. 64. Siskin, M.; Porcelli, J., Journal of the American Chemical Society 1974, 96, 3640. 65. Wristers, J., Journal of the American Chemical Society 1975, 97, 4312. 66. DeWitt, E. J.; Ramp, F. L.; Trapasso, L. E., Journal of the American Chemical Society 1961, 83, 4672. 67. Spies, P.; Schwendemann, S.; Lange, S.; Kehr, G.; Fröhlich, R.; Erker, G., Angewandte Chemie International Edition 2008, 47, 7543. 68. Wang, H.; Frohlich, R.; Kehr, G.; Erker, G., Chemical Communications 2008, 5966. 69. Greb, L.; Oña-Burgos, P.; Schirmer, B.; Grimme, S.; Stephan, D. W.; Paradies, J., Angewandte Chemie International Edition 2012, 51, 10164. 70. Greb, L.; Daniliuc, C.-G.; Bergander, K.; Paradies, J., Angewandte Chemie International Edition 2013, 52, 5876. 71. Chernichenko, K.; Madarász, Á.; Pápai, I.; Nieger, M.; Leskelä, M.; Repo, T., Nature Chemistry 2013, 5, 718. 72. Segawa, Y.; Stephan, D. W., Chemical Communications 2012, 48, 11963. 73. Geier, S. J.; Chase, P. A.; Stephan, D. W., Chemical Communications 2010, 46, 4884. 74. Maier, A. F. G.; Tussing, S.; Schneider, T.; Flörke, U.; Qu, Z.-W.; Grimme, S.; Paradies, J., Angewandte Chemie International Edition 2016, 55, 12219.

Chapter 1 – Introduction

75. Mo, Z.; Rit, A.; Campos, J.; Kolychev, E. L.; Aldridge, S., Journal of the American Chemical Society 2016, 138, 3306. 76. Oestreich, M.; Hermeke, J.; Mohr, J., Chemical Society Reviews 2015, 44, 2202. 77. Chatterjee, I.; Oestreich, M., Angewandte Chemie International Edition 2015, 54, 1965. 78. von Wolff, N.; Lefèvre, G.; Berthet, J. C.; Thuéry, P.; Cantat, T., ACS Catalysis 2016, 6, 4526. 79. Fontaine, F.-G.; Courtemanche, M.-A.; Légaré, M.-A.; Rochette, É., Coordination Chemistry Reviews, doi: /10.1016/j.ccr.2016.05.005. 80. Légaré, M.-A.; Courtemanche, M.-A.; Rochette, É.; Fontaine, F.-G., Science 2015, 349, 513. 81. Mahdi, T.; Stephan, D. W., Angewandte Chemie International Edition 2013, 52, 12418. 82. Zhang, Y.; Miyake, G. M.; Chen, E. Y. X., Angewandte Chemie International Edition 2010, 49, 10158. 83. Tamke, S.; Qu, Z.-W.; Sitte, N. A.; Flörke, U.; Grimme, S.; Paradies, J., Angewandte Chemie International Edition 2016, 55, 4336. 84. Behrends, I.; Bähr, S.; Czekelius, C., Chemistry – A European Journal 2016, 22, 17177. 85. Chan, J. Z.; Yao, W.; Hastings, B. T.; Lok, C. K.; Wasa, M., Angewandte Chemie International Edition 2016, 55, 13877. 86. Mosaferi, E.; Ripsman, D.; Stephan, D. W., Chemical Communications 2016, 52, 8291. 87. Lewi, G. N., In Valence and the Structure of Atoms and Molecules. Chemical Catalogue Company, 1923. 88. Brown, H. C.; Schlesinger, H. I.; Cardon, S. Z., Journal of the American Chemical Society 1942, 64, 325. 89. Piers, W. E.; Marwitz, A. J. V.; Mercier, L. G., Inorganic Chemistry 2011, 50, 12252. 90. Parks, D. J.; Blackwell, J. M.; Piers, W. E., The Journal of Organic Chemistry 2000, 65, 3090. 91. Shibasaki, M.; Kumagai, N., In Cooperative Catalysis, Wiley-VCH Verlag GmbH & Co. KGaA, 2015. 92. Khusnutdinova, J. R.; Milstein, D., Angewandte Chemie International Edition 2015, 54, 12236. 93. Ghuman, K. K.; Hoch, L. B.; Wood, T. E.; Mims, C.; Singh, C. V.; Ozin, G. A., ACS Catalysis 2016, 6, 5764. 94. Szeto, K. C.; Sahyoun, W.; Merle, N.; Castelbou, J. L.; Popoff, N.; Lefebvre, F.; Raynaud, J.; Godard, C.; Claver, C.; Delevoye, L.; Gauvin, R. M.; Taoufik, M., Catalysis Science & Technology 2016, 6, 882. 95. Xing, J.-Y.; Buffet, J.-C.; Rees, N. H.; Norby, P.; O'Hare, D., Chemical Communications 2016, 52, 10478.

Chapter 1 – Introduction

96. Lu, G.; Zhang, P.; Sun, D.; Wang, L.; Zhou, K.; Wang, Z.-X.; Guo, G.-C., Chemical Science 2014, 5, 1082. 97. Hounjet, L. J.; Bannwarth, C.; Garon, C. N.; Caputo, C. B.; Grimme, S.; Stephan, D. W., Angewandte Chemie International Edition 2013, 52, 7492. 98. Li, H.; Aquino, A. J. A.; Cordes, D. B.; Hung-Low, F.; Hase, W. L.; Krempner, C., Journal of the American Chemical Society 2013, 135, 16066. 99. Mummadi, S.; Unruh, D. K.; Zhao, J.; Li, S.; Krempner, C., Journal of the American Chemical Society 2016, 138, 3286. 100. Arnett, E. M.; Wu, C. Y., Journal of the American Chemical Society 1960, 82, 4999. 101. Kovacevic, B.; Baric, D.; Maksic, Z. B., New Journal of Chemistry 2004, 28, 284. 102. Kristjánsdóttir, S. S.; Norton, J. R., In Transition Metal Hydrides, Wiley-VCH Verlag GmbH & Co. KGaA, 1992. 103. Heiden, Z. M.; Lathem, A. P., Organometallics 2015, 34, 1818. 104. Ménard, G.; Stephan, D. W., Angewandte Chemie International Edition 2012, 51, 8272. 105. vom Stein, T.; Peréz, M.; Dobrovetsky, R.; Winkelhaus, D.; Caputo, C. B.; Stephan, D. W., Angewandte Chemie International Edition 2015, 54, 10178. 106. Runyon, J. W.; Steinhof, O.; Dias, H. V. R.; Calabrese, J. C.; Marshall, W. J.; Arduengo, A. J., Australian Journal of Chemistry 2011, 64, 1165. 107. Clark, E. R.; Ingleson, M. J., Angewandte Chemie International Edition 2014, 53, 11306. 108. Metters, O. J.; Forrest, S. J. K.; Sparkes, H. A.; Manners, I.; Wass, D. F., Journal of the American Chemical Society 2016, 138, 1994. 109. Houghton, A. Y.; Karttunen, V. A.; Fan, C.; Piers, W. E.; Tuononen, H. M., Journal of the American Chemical Society 2013, 135, 941. 110. Fan, C.; Mercier, L. G.; Piers, W. E.; Tuononen, H. M.; Parvez, M., Journal of the American Chemical Society 2010, 132, 9604. 111. Rocchigiani, L.; Ciancaleoni, G.; Zuccaccia, C.; Macchioni, A., Journal of the American Chemical Society 2014, 136, 112. 112. Rokob, T. A.; Hamza, A.; Pápai, I., Journal of the American Chemical Society 2009, 131, 10701. 113. Bordwell, F. G., Accounts of Chemical Research 1988, 21, 456. 114. Wiedner, E. S.; Chambers, M. B.; Pitman, C. L.; Bullock, R. M.; Miller, A. J. M.; Appel, A. M., Chemical Reviews 2016, 116, 8655. 115. Beckett, M. A.; Strickland, G. C.; Holland, J. R.; Sukumar Varma, K., Polymer 1996, 37, 4629. 116. Xu, X.; Kehr, G.; Daniliuc, C. G.; Erker, G., Journal of the American Chemical Society 2013, 135, 6465.

Chapter 1 – Introduction

117. Xu, X.; Kehr, G.; Daniliuc, C. G.; Erker, G., Journal of the American Chemical Society 2015, 137, 4550. 118. Flynn, S. R.; Metters, O. J.; Manners, I.; Wass, D. F., Organometallics 2016, 35, 847. 119. Hatnean, J. A.; Thomson, J. W.; Chase, P. A.; Stephan, D. W., Chemical Communications 2014, 50, 301. 120. Farrell, J. M.; Posaratnanathan, R. T.; Stephan, D. W., Chemical Science 2015, 6, 2010. 121. Eisenberger, P.; Bestvater, B. P.; Keske, E. C.; Crudden, C. M., Angewandte Chemie International Edition 2015, 54, 2467. 122. Lam, J.; Gunther, B. A. R.; Farrell, J. M.; Eisenberger, P.; Bestvater, B. P.; Newman, P. D.; Melen, R. L.; Crudden, C. M.; Stephan, D. W., Dalton Transactions 2016, 45, 15303. 123. Erős, G.; Nagy, K.; Mehdi, H.; Pápai, I.; Nagy, P.; Király, P.; Tárkányi, G.; Soós, T., Chemistry – A European Journal 2012, 18, 574. 124. Nicasio, J. A.; Steinberg, S.; Inés, B.; Alcarazo, M., Chemistry – A European Journal 2013, 19, 11016. 125. Sumerin, V.; Schulz, F.; Atsumi, M.; Wang, C.; Nieger, M.; Leskelä, M.; Repo, T.; Pyykkö, P.; Rieger, B., Journal of the American Chemical Society 2008, 130, 14117. 126. Erős, G.; Mehdi, H.; Pápai, I.; Rokob, T. A.; Király, P.; Tárkányi, G.; Soós, T., Angewandte Chemie International Edition 2010, 49, 6559. 127. Paradies, J., Angewandte Chemie International Edition 2014, 53, 3552. 128. Özgün, T.; Bergander, K.; Liu, L.; Daniliuc, C. G.; Grimme, S.; Kehr, G.; Erker, G., Chemistry – A European Journal 2016, 22, 11958. 129. Whittemore, S. M.; Edvenson, G.; Camaioni, D. M.; Karkamkar, A.; Neiner, D.; Parab, K.; Autrey, T., Catalysis Today 2015, 251, 28. 130. Chernichenko, K.; Kótai, B.; Pápai, I.; Zhivonitko, V.; Nieger, M.; Leskelä, M.; Repo, T., Angewandte Chemie International Edition 2015, 54, 1749. 131. Tussing, S.; Greb, L.; Tamke, S.; Schirmer, B.; Muhle-Goll, C.; Luy, B.; Paradies, J., Chemistry – A European Journal 2015, 21, 8056. 132. Tussing, S.; Kaupmees, K.; Paradies, J., Chemistry – A European Journal 2016, 22, 7422. 133. Greb, L.; Tussing, S.; Schirmer, B.; Ona-Burgos, P.; Kaupmees, K.; Lokov, M.; Leito, I.; Grimme, S.; Paradies, J., Chemical Science 2013, 4, 2788. 134. Karkamkar, A.; Parab, K.; Camaioni, D. M.; Neiner, D.; Cho, H.; Nielsen, T. K.; Autrey, T., Dalton Transactions 2013, 42, 615. 135. Whittemore, S. M.; Autrey, T., Israel Journal of Chemistry 2015, 55, 196. 136. Chase, P. A.; Jurca, T.; Stephan, D. W., Chemical Communications 2008, 1701.

Chapter 1 – Introduction

137. Morozova, V.; Mayer, P.; Berionni, G., Angewandte Chemie 2015, 127, 14716. 138. Lambic, N. S.; Sommer, R. D.; Ison, E. A., Journal of the American Chemical Society 2016, 138, 4832. 139. Mahdi, T.; Heiden, Z. M.; Grimme, S.; Stephan, D. W., Journal of the American Chemical Society 2012, 134, 4088. 140. R. Crampton, M.; A. Robotham, I., Journal of Chemical Research, Synopses 1997, 22. 141. Geier, S. J.; Stephan, D. W., Journal of the American Chemical Society 2009, 131, 3476. 142. Geier, S. J.; Gille, A. L.; Gilbert, T. M.; Stephan, D. W., Inorganic Chemistry 2009, 48, 10466. 143. Jiang, C.; Blacque, O.; Fox, T.; Berke, H., Organometallics 2011, 30, 2117. 144. Ullrich, M.; Lough, A. J.; Stephan, D. W., Organometallics 2010, 29, 3647. 145. Rokob, T. A.; Hamza, A.; Stirling, A.; Pápai, I., Journal of the American Chemical Society 2009, 131, 2029. 146. Kolychev, E. L.; Bannenberg, T.; Freytag, M.; Daniliuc, C. G.; Jones, P. G.; Tamm, M., Chemistry – A European Journal 2012, 18, 16938. 147. Wang, Y.; Chen, W.; Lu, Z.; Li, Z. H.; Wang, H., Angewandte Chemie International Edition 2013, 52, 7496. 148. Soós, T., Pure Appl. Chem. 2011, 82, 667. 149. Özgün, T.; Ye, K.-Y.; Daniliuc, C. G.; Wibbeling, B.; Liu, L.; Grimme, S.; Kehr, G.; Erker, G., Chemistry – A European Journal 2016, 22, 5988. 150. Wang, X.; Kehr, G.; Daniliuc, C. G.; Erker, G., Journal of the American Chemical Society 2014, 136, 3293. 151. Feng, X.; Du, H., Tetrahedron Letters 2014, 55, 6959. 152. Ren, X.; Li, G.; Wei, S.; Du, H., Organic Letters 2015, 17, 990. 153. Lindqvist, M.; Borre, K.; Axenov, K.; Kótai, B.; Nieger, M.; Leskelä, M.; Pápai, I.; Repo, T., Journal of the American Chemical Society 2015, 137, 4038. 154. Zhang, Z.; Du, H., Organic Letters 2015, 17, 2816. 155. Ren, X.; Du, H., Journal of the American Chemical Society 2016, 138, 810. 156. Blaser, H. U.; Spindler, F.; Studer, M., Applied Catalysis A: General 2001, 221, 119. 157. Wei, S.; Du, H., Journal of the American Chemical Society 2014, 136, 12261. 158. Ashley, A. E.; Thompson, A. L.; O'Hare, D., Angewandte Chemie International Edition 2009, 48, 9839. 159. Nyhlen, J.; Privalov, T., Dalton Transactions 2009, 5780. 160. Lindqvist, M.; Sarnela, N.; Sumerin, V.; Chernichenko, K.; Leskela, M.; Repo, T., Dalton Transactions 2012, 41, 4310.

Chapter 1 – Introduction

161. Campbell, H. J.; Edward, J. T., Canadian Journal of Chemistry 1960, 38, 2109. 162. Longobardi, L. E.; Tang, C.; Stephan, D. W., Dalton Transactions 2014, 43, 15723. 163. Thomson, J. W.; Hatnean, J. A.; Hastie, J. J.; Pasternak, A.; Stephan, D. W.; Chase, P. A., Organic Process Research & Development 2013, 17, 1287. 164. Legare, M. A.; Rochette, E.; Legare Lavergne, J.; Bouchard, N.; Fontaine, F. G., Chemical Communications 2016, 52, 5387. 165. Magano, J.; Dunetz, J. R., Organic Process Research & Development 2012, 16, 1156. 166. Clarke, M. L.; Roff, G. J., In The Handbook of Homogeneous Hydrogenation, Wiley-VCH Verlag GmbH, 2008. 167. Bullock, R. M., In Catalysis without Precious Metals, Wiley-VCH Verlag GmbH & Co. KGaA, 2010. 168. Chirik, P. J., In Catalysis without Precious Metals, Wiley-VCH Verlag GmbH & Co. KGaA, 2010. 169. Gorgas, N.; Stöger, B.; Veiros, L. F.; Kirchner, K., ACS Catalysis 2016, 6, 2664.

170. Trialkylphosphine LBs and some FLPs have also been shown to be sensitive to O2. 171. Stewart, B.; Harriman, A.; Higham, L. J., Organometallics 2011, 30, 5338. 172. Wang, T.; Kehr, G.; Liu, L.; Grimme, S.; Daniliuc, C. G.; Erker, G., Journal of the American Chemical Society 2016, 138, 4302. 173. Mahdi, T.; Stephan, D. W., Journal of the American Chemical Society 2014, 136, 15809. 174. Mahdi, T.; Stephan, D. W., Angewandte Chemie International Edition 2015, 54, 8511. 175. Gyömöre, Á.; Bakos, M.; Földes, T.; Pápai, I.; Domján, A.; Soós, T., ACS Catalysis 2015, 5, 5366. 176. Mahdi, T. Hydrogenation and Hydroamination Reactions Using Boron-Based Frustrated Lewis Pairs, University of Toronto, 2015. 177. Scott, D. J.; Fuchter, M. J.; Ashley, A. E., Journal of the American Chemical Society 2014, 136, 15813. 178. Scott, D. J.; Simmons, T. R.; Lawrence, E. J.; Wildgoose, G. G.; Fuchter, M. J.; Ashley, A. E., ACS Catalysis 2015, 5, 5540.

Chapter 2 – Boron-based FLPs

Chapter 2.1 – Introduction

As discussed in Chapter 1, the sensitivity of FLPs to oxygen-centred LBs such as alcohols and moisture has tended to be attributed in quite unspecific terms to the high oxophilicity of the

1-2 predominant LAs, of which B(C6F5)3 has become the archetype. Given its importance to this thesis, however, it is worth considering this phenomenon in rather more detail. Certainly, addition of simple organic carbonyls, alcohols, or H2O to B(C6F5)3 will lead to formation of adducts. Such adduct formation is, however, usually reversible (Scheme 2.1; top); after all, while B(C6F5)3 may be a strong

3-6 LA, these are only very weak bases (pKa < 0). These simple interactions alone are therefore not sufficient to account for the failure to observe catalytic hydrogenation in the presence of these compounds; indeed, Piers’ related hydrosilylation chemistry clearly demonstrates the feasibility of

B(C6F5)3-mediated E—H bond activation and catalysis occurring in the presence of even very unhindered oxygen-donor functionalities (Scheme 2.1; bottom).7

Scheme 2.1 – Reversibility in the binding of oxygen-centred donors (top) allows for catalysis by

7 B(C6F5)3 to occur in their presence (bottom).

A more detailed insight into the reasons why FLP-catalysed C=O hydrogenation has previously not been achieved can be found by considering again the first example of stoichiometric reduction reported by Erker et al. (Scheme 2.2).8 In this case, although H– transfer to the substrate is observed,

Chapter 2 – Boron-based FLPs there is no evidence for the subsequent H+ transfer that would be necessary to complete the catalytic cycle. As a result, both the acidic and basic components of the FLP remain thoroughly quenched, preventing any further, productive reactivity.

8 Scheme 2.2 – Incomplete H2 transfer in an early example of FLP-mediated carbonyl hydrogenation.

More generally, it can be seen that if, in an FLP, the cumulative oxophilicity of the LA and Brønsted strength of the LB are sufficiently high then any ROH species present (such as adventitious moisture, or alcohols produced via C=O hydrogenation) will react to form an inactive [LB—H][LA—OR] species. If it is irreversible, this reaction will present an unavoidable barrier to catalysis (Scheme 2.3).

This can be understood in further detail by considering relative pKa values. Studies on adducts of the type ROH·B(C6F5)3 have shown that binding of ROH to electrophilic boranes leads to dramatically increased Brønsted acidity of the O—H bond. In particular, the H2O adduct H2O·B(C6F5)3 has been reported to be an extremely powerful Brønsted acid, with strength similar to HCl (pKa = 8.4, 8.5

4 respectively in MeCN; estimated pKa < 1 for the former in H2O). This is significantly lower than the strength of commonly-employed FLP LBs based on nitrogen (e.g. col; 2,4,6-trimethylpyridine; pKaH =

9 15.0 in MeCN) or phosphorus (e.g. trialkylphosphines; pKaH > ca. 15 in MeCN). As such, an irreversible Brønsted acid/base reaction can be expected to occur between these species.

Scheme 2.3 – Irreversible inhibition of FLP reactivity by ROH.

Chapter 2 – Boron-based FLPs

One approach to overcome the additional barrier to catalysis introduced by this competing O—H activation process might be to employ significantly elevated reaction temperatures, in order to impart some reversibility to this side-reaction. Unfortunately, this strategy is limited for FLPs based on typical borane LAs by the tendency of borane/ROH mixtures to decompose via protodeborylation, which generates inactive borinic ester side-products (Scheme 2.4). This was found to be the a major limiting factor during early attempts by the groups of Repo and Stephan to observe B(C6F5)3-catalysed C=O hydrogenation,1, 3 and is readily observed by NMR spectroscopy through the formation of

11 downfield B resonances [ca. 40 ppm for typical ROB(C6F5)2 derived from B(C6F5)3 decomposition in non-donor solvents; in these reactions the C6F5H produced is also clearly visible by its characteristic multiplet around 7.2 ppm in the 1H NMR spectrum].3

Scheme 2.4 – Decomposition of B(C6F5)3/ROH mixtures through protodeborylation.

Nevertheless, if FLP-mediated O—H cleavage reaction is taken to be the root cause for the difficulty of achieving FLP-catalysed carbonyl hydrogenation, and for FLPs’ moisture-intolerance, then it becomes possible to rationally design a modified system which should be able to overcome these problems. The well-established ability of bulky, electrophilic trisarylboranes such as B(C6F5)3 to engage in FLP-catalysed hydrogenation reactions makes these particularly attractive LAs for initial study, and if H2 activation can be achieved using a sufficiently weak base then deprotonation of any highly acidic

ROH·BAr3 adducts should become reversible, making catalysis feasible. Furthermore, use of a weak LB is essential if carbonyl hydrogenation is to proceed via the most common FLP mechanism, in which the substrate is activated initially by protonation, as this will require generation of a sufficiently

+ 6 Brønsted acidic [LB—H] upon H2 activation (Scheme 2.5).

Scheme 2.5 – Possible mechanism for FLP-catalysed carbonyl hydrogenation, which requires use of a weak LB for substrate activation.

Chapter 2 – Boron-based FLPs

Reversible activation of H2 at low temperature by FLPs incorporating very weakly basic phosphines and amines (pKaH = ca. 2-5 in MeCN) has previously been reported, and these systems were shown to be competent in the hydrogenation of weakly basic alkene substrates, despite an

10-11 equilibrium that disfavours the FLP·H2 product (Scheme 2.6; a). Similar results were subsequently

12 achieved using much simpler diethyl ether (pKa = 0.2 for [Et2OH][BF4]/Et2O in MeCN) as the LB (Scheme 2.6; b).13 Previous work in the Ashley and Fuchter groups has since demonstrated that THF solutions of two trisarylboranes similar to those commonly employed in FLP chemistry are capable of

14 activating H2 without the need for an auxiliary base. In these reactions it is the solvent itself that acts as the basic component of an FLP, with H2 activation (which is rapidly reversible and appears to be only slightly thermodynamically unfavourable: 'G ≈ 0) generating a strongly acidic solvated proton

5 (pKa = –2.05 in aqueous H2SO4; c.f. –3.59 for protonated Et2O under the same conditions) alongside the conventional [LA—H]– anion (Scheme 2.6; c).

Scheme 2.6 – FLP-catalysed hydrogenation using a weak phosphine (a) or ether (b) as the LB;10, 13

14 reversible FLP H2 activation using the solvent as the LB (c).

Chapter 2 – Boron-based FLPs

These systems have several significant benefits: they are practically very simple; use an extremely cheap LB component; and display good thermal stability. Furthermore, the use of the solvent as the LB ensures that it is present at the highest possible concentration, which should positively affect the H2 activation equilibrium, and also converts formally termolecular H2 activation into a pseudo-bimolecular reaction, with likely kinetic benefits. Based on these factors, BAr3/THF FLPs were judged to be a sensible starting point for further investigations.

Chapter 2.2 – Catalytic hydrogenation using borane/solvent FLPs

Supplementary information relating primarily to Chapter 2.2 can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201405531/abstract

Initial attention was focused on the family of boranes B(C6Cl5)n(C6F5) 3-n (n = 0-3, 1a-d; Figure 2.1), whose Lewis acidic behaviour has previously been studied in some detail.15 Increasing n (i.e. increasing the number of perchlorophenyl rings, in place of perfluorophenyl) has been reported to lead to increased electrophilicity (C6Cl5 is more electron-withdrawing than C6F5), but reduced Lewis acidity due to greater steric bulk. The C6Cl5-substituted boranes 1b-d have also demonstrated appreciable stability to air and moisture, further increasing their practical appeal.

Figure 2.1 – Initial boranes studied.

To begin with, the coordination behaviour of boranes 1 in the presence of THF was studied. The commonly-employed borane 1a was found to bind strongly to THF, as shown by variable-

11 19 11 temperature (VT) B and F NMR analysis of a 1:1 mixture in C7D8. At room temperature (RT) the B NMR spectrum shows a single resonance at 3.1 ppm, which is shifted significantly upfield versus 1a

16 alone (63 ppm in C6D6), and consistent with coordination to form a 4-coordinate adduct. The observed shift does not vary significantly at lower temperature (although the resonance does become progressively broader), or at elevated temperatures up to ca. 60 °C. Above this point the resonance does begin to shift noticeably downfield, consistent with reversible dissociation (Figure 2.2; top). Similar conclusions can be drawn from the 19F spectra. The separation of meta and para 19F resonances is often considered to be an indicator for the degree of pyramidalisation of C6F5-

Chapter 2 – Boron-based FLPs substituted boranes, and in this case only begins to increase significantly above 60 °C (Figure 2.2; bottom).17-19

11 19 Figure 2.2 – VT B (top) and F (bottom) NMR spectra for equimolar THF/1a in C7D8.

The strength of this interaction suggested that a 1a/THF system might not be suitable for catalysis. Fortunately, the interaction between THF and the bulkier borane 1b was found to be significantly weaker, with similar NMR analysis showing the onset of significant dissociation at much

Chapter 2 – Boron-based FLPs lower temperature (30 °C) even in neat THF (Figure 2.3; top). This trend continues as steric bulk is increased further, with a THF solution of 1c showing evidence of an adduct only at temperatures below –40 °C (Figure 2.3; bottom). In this case adduct formation is indicated by the appearance of a second set of 19F resonances, indicating slow exchange between the free and bound forms of 1c on the NMR timescale (as well as slow rotation about the B—C6F5 bond, which accounts for the appearance of five distinct 19F resonances). For THF solutions of 1d no evidence of adduct formation is observed at any temperature between 60 °C and –100 °C.

Figure 2.3 – VT 19F (left) and 11B (right) NMR spectra for 1b (top) and 1c (bottom) in THF.

Chapter 2 – Boron-based FLPs

Before beginning detailed investigations into carbonyl hydrogenation and moisture tolerance it was decided to gain additional insight into the chemistry of 1/THF FLPs, and seek a proof-of-principle for their use in hydrogenation catalysis, by investigating the reduction of some more thoroughly- understood non-carbonyl substrates. Given the already-demonstrated ability of 1b in THF to effect H2 activation (Scheme 2.6; c), the suitably weak binding of this Lewis pair, and the anticipation that larger boranes might be kinetically ill-suited to catalysis (vide infra), this system was chosen to begin investigations.

Chapter 2.2.1 – Catalytic hydrogenation of imines

Given their prominence in the development of previous FLP hydrogenation catalysts, imines were chosen as initial substrates. Imine hydrogenation catalysed by 1a in non-donor solvents has been reported previously; however, this reaction relies on the ability of the substrate to act as an LB

20-21 for H2 activation in conjunction with borane LA. Consequently, it has been found to work poorly for electron-deficient imines such as N-tosyl imine 2a (Table 2.1, entries 1 and 2), as H2 activation becomes kinetically limiting due to the low combined strength of the 1a/2a FLP. Nevertheless, their structural and electronic similarities mean imines of this type are useful models for more difficult ketone and aldehyde substrates, and 2a was therefore chosen as an investigative starting point.

Gratifyingly, admission of H2 (ca. 4 bar; note that for simplicity and consistency, pressures at RT will be quoted throughout this thesis: pressures at elevated temperature will be slightly higher) to a gently-heated (60 °C) d8-THF solution of 2a in the presence of a catalytic quantity of 1b (5 mol%) led to rapid reduction, generating the related N-tosyl amine 3a (Table 2.1, entry 3), as indicated by clear changes in the 1H NMR spectrum of the reaction mixture (in particular, by loss of the diagnostic

22 aldimine CH resonance at 9.07 ppm, and appearance of a new PhCH2N resonance at 4.01 ppm). The product could be very easily isolated in near-quantitative yield following simple chromatographic work-up. Furthermore, the air-stability of the catalyst means that it could be conveniently stored and weighed out on the open bench prior to use (Table 2.1, entry 4), although the use of thoroughly-dried solvent and H2 unfortunately remained essential.

The success of this reaction under very mild conditions, in contrast to previous reports using 1a in solvents such as PhMe, is attributed to a change in mechanism. Whereas the reaction in a non- donor solvent must use the substrate (or, in later stages, the product, which is expected to be a similarly poor base) as the LB for H2 activation (Scheme 2.7; substrate-mediated H2 activation

23-25 pathway), in d8-THF the reaction can proceed via solvent-assisted H2 heterolysis (Scheme 2.7;

+ solvent-mediated H2 activation pathway), with subsequent H transfer from the resulting powerful

Chapter 2 – Boron-based FLPs

Brønsted acid to the substrate. Both cycles are completed by hydride transfer to generate the reduced amine product.

Table 2.1 – Imine hydrogenation catalysed by boranes 1 in ethereal solvents.

Entry Substrate Solvent [B] (mol%) T / °C t / h Conv. / %a

1b,c 2a PhMe 1a (10) 80 22 7 2b,d 2a PhMe 1a (10) 80 22 99

e 3 2a d8-THF 1b (5) 60 3 >99 (98) 4f 2a THF 1b (5) 60 3 >99 5g 2b PhMe 1a (5) 120 41 94 6g,h 2b PhMe 1a (5) 120 8 98

7 2a C7D8 1b (5) 60 3 0

8 2c d8-THF 1b (5) 60 3 >99

9 2c C7D8 1b (5) 60 3 0

10 2d d8-THF 1b (15) 60 8 91

11 2d C7D8 1b (15) 60 8 26

12 2e d8-THF 1b (5) 80 18 71

13 2e C7D8 1b (5) 80 18 79

e 14 2f d8-THF 1b (5) 60 8 >99 (99)

15 2f C7D8 1b (5) 60 8 0

16 2a d8-THF 1c (5) 60 72 90

17 2a d8-THF 1d (5) 80 72 0

18 2a d8-THF 1a (10) 80 72 84 19 2a 1,4-dioxane 1b (5) 60 41 96 aMeasured at 0.1 mmol scale by in situ 1H NMR spectroscopy, using 1,3,5-trimethoxybenzene in a capillary insert as an internal standard for integration (see Chapter 4.1). bResult reported by

21 c d e Klankermayer et al. 10 bar H2. 30 bar H2. Isolated yield at 1 mmol scale shown in parentheses. fSubstrate and catalyst stored and weighed out on open bench. gResult reported by Stephan et al.20 h With 5 mol% PMes3.

Clearly, the two mechanisms differ only in the route taken from the solvated borane (I in Scheme 2.7) to a common intermediate containing protonated substrate (II in Scheme 2.7). While the

Chapter 2 – Boron-based FLPs involvement of the solvent does not necessarily render this transformation more thermodynamically favourable (although H-bonding interactions are certainly possible), the high concentration of the THF catalyst is expected to lead to a significant kinetic improvement, thereby accounting for the overall improvement in the rate of the catalytic reaction. Note that a similar but more modest rate improvement has previously been reported when the auxiliary base PMes3 (pKaH = 2.7, 8.0 in H2O,

26 MeCN respectively for closely-related PPh3) is added to 1a-catalysed hydrogenation of weakly-basic imines (Table 2.1, entries 5 and 6; in this case the additional base is expected to make H2 activation more thermodynamically favourable).20, 27

Scheme 2.7 – Different mechanistic pathways for borane-catalysed imine hydrogenation in donor and non-donor solvents.

Consistent with the interpretation that solvent is involved in H2 cleavage, attempts to perform the equivalent reaction using 1b as a catalyst in C7D8 solution were unsuccessful (Table 2.1, entry 7); the change in rate is therefore not attributable simply to the switch from 1a to 1b as the LA. Similar results were also obtained by replacing 2a with the related imine 2c, derived from an aliphatic aldehyde. This substrate was effectively reduced at a similar rate using the same protocol in d8-THF (Table 2.1, entry 8), but again, no analogous reactivity was observed under equivalent conditions in

C7D8 (Table 2.1, entry 9). Slightly different results were obtained for the reduction of the somewhat stronger N-aryl bases 2d-f (Table 2.1, entries 10-15). In each case, hydrogenation still proceeds cleanly in d8-THF, but at lower rates than for the N-tosyl analogues. For 2d this is presumably due to reversible adduct formation between 1b and the more strongly-basic product (Scheme 2.8; top), which is supported by observation of an upfield shift in the 11B NMR spectrum during the course of the reaction

Chapter 2 – Boron-based FLPs

(from 7.0 to –1.6 ppm). Analogous adduct formation has been suggested to be the rate-limiting factor in 1a-catalysed imine hydrogenation.20 For 2e and 2f, meanwhile, increased steric bulk around the imine functionality likely hinders the final hydride transfer step to generate 3 (Scheme 2.8; bottom).

Scheme 2.8 – Likely causes for the reduced rates of hydrogenation of 2d, 2e and 2f relative to 2a.

More dramatic differences in reactivity are observed in C7D8, with appreciable hydrogenation observed for both 2d and 2e, indicating that these substrates are sufficiently basic for substrate- mediated H2 activation (Scheme 2.7; substrate-mediated H2 activation pathway) to become competitive with the solvent-mediated pathway (Scheme 2.7; solvent-mediated H2 activation pathway). Surprisingly, 2f shows no evidence for reduction in C7D8 despite being of comparable basicity to 2d/2e; clearly, substrate-mediated H2 activation is not feasible for this substrate. Since it is unlikely to be a thermodynamic result, this is most likely due to the extreme steric shielding around the basic nitrogen centre which, in combination with the very high steric bulk of 1b (even greater than

1a, for example), may preclude formation of a 1b/2f encounter complex suitable for H2 cleavage (Scheme 2.9).

Further investigations focused on establishing the generality of the borane-catalysed hydrogenation of the model substrate 2a. Replacing 1b with the bulkier borane 1c led to a significant decrease in reaction rate (72 h versus < 3 h required for 90 % conversion; Table 2.1, entry 16 versus entry 3), but otherwise had very little effect on the reaction outcome with extensive, clean reduction still eventually observed under the same reaction conditions. This result also confirms that H2 activation remains feasible for the 1c/THF Lewis pair. The lower rate is attributed to the increased

Chapter 2 – Boron-based FLPs

– bulk of the [1c—H] anion generated following H2 activation, which leads to slow hydride transfer to the substrate (Figure 2.4).

Scheme 2.9 – An account for differing reactivity in the 1b-catalysed hydrogenation of 2f in d8-THF

and C7D8.

Figure 2.4 – Qualitative effect of steric bulk on the hydrogenation of 2a catalysed by boranes 1b-d in THF.

Chapter 2 – Boron-based FLPs

The trend in catalytic activity continues as steric bulk is increased further, with no reduction at all observed using the fully chlorinated borane 1d, even under slightly more forcing conditions

(Table 2.1, entry 17). Despite this result, the ability of 1d in d8-THF to activate H2 was independently confirmed. Much as with the previously-reported para-fluorinated analogue (Scheme 2.6; c),14 addition of H2 (4 bar) followed by brief heating (60 °C, 1 h) led to the appearance of a sharp (though low intensity) doublet at –9.7 ppm [1J(11B-1H) = 91 Hz] in the 11B NMR spectrum and a singlet at 11.34

1 28 ppm in the H NMR spectrum, consistent with H2 activation. Since this initial step is clearly feasible under the reaction conditions, and protonation of the substrate should be no more difficult than when 1b or 1c are used, this confirms that the lack of catalytic reactivity using 1d must be due to unsuccessful hydride transfer (Scheme 2.10).

Scheme 2.10 – Explaining the inability of 1d to catalyse the hydrogenation of 2a.

Interestingly, a degree of catalytic activity is also retained if the steric bulk of the borane catalyst is decreased rather than increased. When the archetypal FLP LA 1a was used instead of 1b-d, slow reduction of 2a could still be observed at slightly higher temperatures (80 °C; Table 2.1, entry 18). Clearly, despite the strong binding to THF described earlier (Figure 2.2), adduct formation with the solvent is still sufficiently reversible at high temperatures to allow for transient generation of the active unsolvated LA, which can then engage in H2 activation and subsequent reduction chemistry. Even so, the increased difficulty in generating the on-cycle ‘free’ borane when using 1a instead of 1b (or, alternatively, more effective quenching of the less bulky borane by the solvent) is consistent with the lower observed rate of hydrogenation (Figure 2.5).

As well as tolerating variation of the borane 1, it was found that catalytic activity is also retained following modification of the ethereal solvent, with clean hydrogenation observed using 1b in 1,4-dioxane instead of THF (Table 2.1, entry 19). Again, however, this was associated with a significant reduction in reaction rate (41 h required for >95 % conversion, versus < 3 h in THF): in this case, attributed to both the reduced Brønsted basicity (pKaH = –2.92 vs. –2.05 in aqueous H2SO4) and

Chapter 2 – Boron-based FLPs

5, 29 lower polarity (Hr = 2.22 vs. 7.52) of 1,4-dioxane relative to THF. Both factors are expected to render

+ initial H2 activation less thermodynamically favourable (Figure 2.6); the former means the ‘H ’ formed upon H2 activation will be less stable, while the latter will disfavour the creation of the charged, ionic product (in other words, the proton attachment and ionic stabilisation steps, respectively, from Figure 1.9 will be less favourable).

Figure 2.5 – Reduced catalytic activity of 1a versus 1b for the hydrogenation of 2a in THF, due to stronger binding of the solvent.

Figure 2.6 – Less favourable H2 activation accounting for the reduced rate of 1b-catalysed hydrogenation in 1,4-dioxane relative to THF.

Chapter 2 – Boron-based FLPs

Chapter 2.2.2 – Catalytic hydrogenation of additional substrates

Having established good activity for the hydrogenation of imines it was decided to investigate the applicability of borane/solvent Lewis pair systems to the hydrogenation of a broader range of non- carbonyl substrates. As the system that had previously shown the greatest catalytic activity, 1b in THF was chosen as the focus for these studies. Previously, systems that are effective for imine hydrogenation have often also been found to catalyse the hydrogenation of related N-basic substrates

30-31 such as aziridines and conjugated aromatic N-heterocycles. Thus, exposure of a d8-THF solution of acridine, 4, to H2 (4 bar) in the presence of catalytic 1b led cleanly to reduction of the central aromatic ring to form acridane, 5, at a rate comparable to those for the imines discussed above (Scheme 2.11; top). Similar results were obtained for 8-methylquinoline, 6, with complete reduction of the heteroaromatic ring by two equivalents of H2 furnishing product 7 (Scheme 2.11; bottom).

Scheme 2.11 – Hydrogenation of N-heteroaromatic rings catalysed by 1b in d8-THF.

Rather more complicated results were obtained during the reduction of N-tosyl aziridine, 8. Initial analysis appeared promising, with 1H NMR analysis showing complete disappearance of the resonances for 8 after heating to 60 °C for 19 h under 4 bar H2 in THF, with the accompanying appearance of a pair of new alkyl resonances at 2.96 and 0.86 ppm (as well as a new set of signals for the tosyl group). Unfortunately, the final spectrum also showed large resonances at ca. 3.3 and 1.5 ppm, characteristic of THF polymerisation.32 While polymerisation of THF is well known to be initiated by strong electrophiles,32-35 previous experiments (vide supra) had shown no evidence for a similar side-reaction, indicating that this could not be due to the presence of either 1b, or solvated ‘H+’. Furthermore, close inspection of the product 1H resonances revealed different coupling constants for the quartet and triplet at 2.96 [3J(1H-1H) = 5.6 Hz, 2H] and 0.86 ppm [3J(1H-1H) = 7.3 Hz, 3H], respectively, indicating that these resonances could not correspond to the expected product 9. Instead, it is proposed that initial protonation of 8 generates an intermediate that is sufficiently

Chapter 2 – Boron-based FLPs electrophilic to initiate polymerisation of the solvent, prior to eventual termination via hydride transfer from [1b·H]–, resulting in incorporation of an oligo(THF) linker into the actual product (Scheme 2.12; a). That ring-opening kinetically prefers to use THF as the nucleophile (rather than [1b·H]–) can likely be attributed to this species’ increased concentration and decreased steric bulk relative to the borohydride anion. Note that while this reaction does not furnish the intended product, it does still involve overall hydrogenative catalytic reduction of the substrate. A more detailed discussion regarding decomposition of borane/solvent hydrogenation catalysts will be presented in Chapter 2.2.3.

Scheme 2.12 – Observation of solvent oligomerisation during the 1b-catalysed hydrogenation of 8 (a), and attempted hydrogenation of 10 (b); 1H NMR integration suggests mean m ≈ 25.

A similar model was also used to explain the outcome of unsuccessful attempts to hydrogenate the acetophenone-derived silyl enol ether 10. Instead of reduction, this reaction was observed to lead to rearrangement to form acetophenone, presumably via protonation followed by

+ 36 loss of ‘SiMe3 ’ (which is known to be labile under Brønsted acidic conditions) to the solvent. This release of the powerful silylium electrophile then initiates solvent oligomerisation, which is ultimately

Chapter 2 – Boron-based FLPs terminated by reaction with [1b·H]– (Scheme 2.12; b). Crucially, a triplet resonance is observed at 0.86 ppm [3J(1H-1H) = 7.3 Hz, 3H] in the 1H NMR spectrum of the completed reaction mixture; this is identical to the signal observed during hydrogenation of 8, and corresponds to the terminal CH3 group of the oligo(THF) unit. Note that during this reaction, no evidence was observed to indicate reduction of the carbonyl moiety in the acetophenone product, suggesting that hydrogenation of this functional group using a 1b/THF catalytic system is unfortunately not feasible (at least not for this substrate under such mild conditions; see Chapter 2.2).

Despite the failure to observe hydrogenation of the C=C bond in 10, effective hydrogenation of other activated C=C bonds could be achieved, with exposure of either n-butyl acrylate (12a) or 1,2,3,4,5-pentamethylcyclopentadiene (Cp*H; 14) to typical hydrogenation conditions leading to formation of the respective reduced species (Scheme 2.13). Note that, as in the attempted hydrogenation of 10, no evidence for reduction of the carbonyl moiety was observed during the hydrogenation of 12a.

Scheme 2.13 – Hydrogenation of C=C bonds catalysed by 1b in THF.

The ability to hydrogenate 14 (albeit with more limited turnover) provided further evidence for the suitability of borane/solvent systems for the hydrogenation of weakly basic compounds (due to the high strength of the Brønsted acid generated through H2 activation), and it was realised that this might be exploited to allow for the hydrogenation of similar substrates that have not previously been reported to be amenable to hydrogenation via FLP methods. In particular, while the reduction of strongly or moderately basic N-heterocycles is well established (vide supra), equivalent FLP- mediated hydrogenation of weakly basic pyrroles has not previously been described. Although partial hydrogenation of a few related indoles has been reported, this required the use of very high pressures

30 of H2, as well as moderately high temperatures (103 bar, 80 °C).

Chapter 2 – Boron-based FLPs

Gratifyingly, admission of H2 (4 bar) to an equimolar mixture of 1b and 2,5-dimethylpyrrole

(16a) in d8-THF led to formation of the fully-reduced pyrrolidine 17a, in the form of the H2 activation salt [17a—H]+[H—1b]–, after heating to 60 °C for 45 h (Scheme 2.14; a). Similar reactivity was observed for N-methylpyrrole (16b). Unfortunately, however, attempts to render these reactions catalytic were unsuccessful; this is attributed to the low Brønsted acidity of the [17—H]+ cation formed following hydrogenation of the first equivalent of substrate. Proton transfer from this acid to a second equivalent substrate is infeasible, preventing any further reduction (Scheme 2.14; b). An equivalent explanation is that, following stoichiometric reduction, any further ‘H+’ generated will be levelled to the basic amine product, thereby losing the strength required to protonate further substrate (for

37-38 example, the pKaH of unsubstituted pyrrole is –3.8 in aqueous H2SO4, versus 11.31 for pyrrolidine). This can be compared with a similar failure to observe catalytic turnover in the 1a-mediated hydrogenation of anilines reported by Stephan et al., as discussed in Chapter 1.4.3.39

Scheme 2.14 – Stoichiometric 1b-mediated hydrogenation of pyrroles 16 in d8-THF (a); and an explanation for the lack of catalytic turnover (b).

It was speculated that turnover for pyrrole hydrogenation might be possible following addition of an electron-withdrawing group, to render the product pyrrolidine less basic; however, attempts to reduce N-tosylpyrrole (16c) were unsuccessful, presumably because incorporation of the sulfonamide moiety reduces the already low basicity of the substrate to the extent that it cannot be protonated even by THF-solvated H+ (Scheme 2.15; a). As a result, attention was shifted to related furans as substrates. Like pyrroles, these compounds have not previously been reported as substrates

Chapter 2 – Boron-based FLPs for FLP-catalysed hydrogenation. Unlike for pyrroles, however, hydrogenation of furans should ultimately generate substituted THF derivatives, with basicities close to that of the solvent. Thus, the simple methyl-substituted furans 18a-c could all be successfully hydrogenated in a catalytic fashion, under similarly mild conditions to those discussed thus far (Scheme 2.15; b).

Scheme 2.15 – Unsuccessful hydrogenation of electron-poor pyrrole 16c (a), and successful

hydrogenation of simple furans 18 catalysed by 1b in d8-THF.

The mechanism for 1b-mediated pyrrole and furan hydrogenation has not been investigated

+ – directly; however, a plausible series of H2 activation and H /H transfer steps in shown in Scheme 2.16, with initial protonation occurring at the typically-more-nucleophilic D position. Note that while the ability of an electron-rich arene to engage in H2 activation directly as the LB component of an FLP has

39 been described previously, it seems likely on the basis of earlier results (vide supra) that H2 activation involving the solvent would be more rapid in this case.

Chapter 2 – Boron-based FLPs

Scheme 2.16 – One possible mechanism for 1b-mediated hydrogenation of simple pyrroles and furans.

Chapter 2.2.3 – On the stability of borane/solvent Lewis pair hydrogenation catalysts

As discussed briefly in the previous section (Scheme 2.12), strong Lewis and Brønsted acids have long been known to initiate the ring-opening polymerisation of THF (Scheme 2.17; a, b).32-35 Nevertheless, with the exception of the two specific examples discussed in Chapter 2.2.2, in none of the reactions discussed thus far was any evidence obtained for even low levels of solvent oligomerisation. Such decomposition tends to be very clearly noticeable, either through visible gelation of the solvent upon cooling to RT, or by the appearance of large new peaks at ca. 3.3 and 1.5 ppm in the 1H NMR spectrum (note that, while most of the reactions in Chapter 2.2.2 are described as being performed in d8-THF, in almost all cases preliminary studies were performed in cheaper proteo THF, where polymerisation would have been particularly obvious by 1H NMR analysis). Nor was any evidence obtained in any reaction to indicate stoichiometric ring opening of THF by any LA/LB combination, despite such ring-opening having been established very early on as an archetypal example of FLP reactivity (Scheme 2.17; c).40-43

Chapter 2 – Boron-based FLPs

Scheme 2.17 – Possible decomposition routes for 1/THF FLPs.

To a significant extent, the failure to observe these side-reactions can probably be attributed to relatively weak binding of the bulky boranes 1b-d to THF, particularly at elevated temperatures (as discussed at the start of this chapter). As a result, there will be only weak Lewis acid activation of any bound solvent and neither LA-initiated polymerisation, nor FLP-mediated C—O cleavage will be particularly favoured (Scheme 2.18; a, c). The latter will also have been discouraged by the relatively low nucleophilicity of most of the substrates and products investigated.

+ The failure of the solvated ‘H ’ generated following H2 activation to initiate polymerisation is less obviously explicable; after all, this is the strength to which even the strongest acids must be levelled in THF solution. One possibility is that, since the equilibrium of H2 activation strongly disfavours its formation, the ability of ‘H+’ to initiate polymerisation is limited by its low concentration. In many reactions, any ‘H+’ generated will also rapidly be levelled via attachment to a more basic substrate or product; the concomitant reduction in strength will again disfavour polymerisation. Finally, it is possible that the ‘H+’ generated in fact remains closely associated with its borohydride counteranion in solution, reducing effective Brønsted acid strength through stabilising ionic interactions (Scheme 2.18; b). For example, previous studies have suggested that significant stabilisation of FLP·H2 salts may be possible through so-called dihydrogen bonding (an unusual form

Chapter 2 – Boron-based FLPs of E—H · · · H—X hydrogen bonding in which a hydridic E—H bond acts as a hydrogen bond acceptor), both in the solid state and in solution.44-46

Scheme 2.18 – Factors likely to discourage THF ring-opening side-reactions.

While the arguments outlined thus far can be used to account for the observed stability in almost all of the reactions described in Chapter 2.2.2, there remains one exception; namely, the hydrogenation of imine 2a catalysed by 1a (Table 2.1, entry 18). While boranes 1b-d may interact only relatively weakly with THF, the equivalent interaction is much stronger for less bulky 1a (as was established at the start of this chapter). In fact, 1a has previously been reported to initiate the polymerisation of THF simply upon standing at RT.35 To confirm this result, a solution of 1a in THF was heated to 80 °C (the same as the temperature used in the catalytic reaction) for 36 h; as expected, polymerisation of the solvent was immediately obvious upon cooling to RT, both through transformation of the mixture into a viscous gel, and by 1H NMR spectroscopy (Scheme 2.19; top). No such change was detectable when the experiment was repeated using 1b in place of 1a; nor, remarkably, when an atmosphere of H2 (4 bar) was added to an otherwise identical solution of 1a in THF prior to heating.

Chapter 2 – Boron-based FLPs

Scheme 2.19 – Very low-level formation of short oligo(THF) compounds mediated by 1a,b under H2.

Insight into the origin of this unexpected variation was provided by mass spectrometry (MS- ToF). Despite the lack of evidence for solvent ring opening provided by other methods, MS analysis of the 1a/THF mixture prepared under H2 showed a sequence of peaks at m/z = 457 ± 72n (n = 0, 1, 2, 3…), consistent with formation of a distribution of short-chain THF oligomers capped at each end by H atoms, centred around the hexamer. Based on these results it can be proposed that the presence of an H2 atmosphere does not in fact prevent the initiation step of THF polymerisation. Instead, the propagating chain, once formed, is rapidly intercepted by [(THF)n—H][H—1a] before it can consume a significant proportion of the solvent, undergoing reduction to form the observed H-capped oligomers (Scheme 2.19; bottom).

Chapter 2 – Boron-based FLPs

In light of this, MS analysis was also performed on samples of 1b in THF after heating to 80 °C under both N2 and H2. While there was still no evidence of ring-opening for the sample prepared under N2 (confirming that 1b does not initiate THF polymerisation), the sample under H2 showed similar short oligomers to its 1a-based counterpart, suggesting that the [(THF)n—H][H—1b] H2 activation salt can act as an initiator under certain conditions. It should be emphasised, however, that the degree of solvent decomposition remains extremely low (below the threshold for observation by 1H NMR spectroscopy) and so does not represent a significant practical concern. This general lack of solvent decomposition will remain true for all other borane/ethereal solvent-catalysed hydrogenation reactions discussed in the remainder of Chapter 2 (the bulk of which will use 1,4-dioxane as the solvent, which is less prone than THF to ring-opening decomposition reactions due to its reduced internal strain, and lower nucleophilicity).

Chapter 2.3 – Borane-catalysed carbonyl hydrogenation

Supplementary information relating primarily to Chapter 2.3 can be found online at: http://pubs.acs.org/doi/abs/10.1021/ja5088979

Chapter 2.3.1 – Development of an initial catalytic system

Having demonstrated the ability of borane/solvent Lewis pair systems to effect the catalytic hydrogenation of a variety of unsaturated functional groups and, in particular, their suitability for the reduction of very weakly-basic substrates, attention was shifted to the primary target of achieving the FLP-catalysed hydrogenation of organic carbonyl groups. Since optimum results had previously been obtained using borane 1b in THF, this system was chosen for initial study.

Very gratifyingly, admission of H2 (4 bar) to a d8-THF solution of acetone, 20a, and 10 mol% 1b led rapidly, and under only mild heating (65 °C), to catalytic consumption of the substrate and formation of reduced products (Scheme 2.20; a). This was the first example of C=O hydrogenation to be successfully catalysed using an FLP methodology. Nevertheless, the reaction clearly suffered from a number of significant problems; the most obvious being very limited turnover, with just 3.5 equivalents of substrate consumed relative to the catalyst. Furthermore, rather than reduction proceeding cleanly to the expected alcohol product 21a, 1H NMR analysis of the final reaction mixture showed two sets of ‘iPrO—’ resonances, with the second identified as being due to formation of iPr2O (22a), presumably via acid-catalysed condensation of the initially-formed 21a, with concomitant formation of H2O (Scheme 2.20; b).

Chapter 2 – Boron-based FLPs

Scheme 2.20 – Hydrogenation of 20a catalysed by 1b in d8-THF (a), and the presumed route for formation of side-product 22a (b).

Attempts to increase catalytic turnover through the use of increased reaction times were unsuccessful. Nevertheless, 19F and 11B NMR analysis provided no indication for decomposition of the 1b catalyst, with only a single set of resonances observable. Nor were any resonances evident in the

1 H NMR spectrum that could be attributed to C6F5H or C6Cl5H, which would be formed through protodeborylation (Scheme 2.4). It therefore seemed likely that the limited turnover could be a consequence of product inhibition. In particular, it was noted that the final reaction mixture corresponded to formation of almost precisely one equivalent of 22a, and hence H2O, relative to 1b. The idea was further supported by the observation of an upfield shift in the 1b 11B NMR resonance during the course of the reaction; from 7.1 to –1.2 ppm, consistent with stronger quenching of the LA (c.f. ca. –2 ppm for typical [1a·OR]–; this will be discussed further in Chapter 2.3.3 and Chapter 2.4).47

To test this hypothesis, the hydrogenation reaction was repeated, spiking the initial mixture with one equivalent of either the alcohol product 21a, or H2O. Both reactions showed clear evidence of inhibition; however, while addition of 21b led only to a reduction in conversion (from 35 % to 20 %), addition of H2O led to a complete loss of hydrogenation activity, indicating that this is by far the more potent catalyst poison (Scheme 2.21). Intuitively this makes sense: binding of a small LB should be much more favourable for a LA with the steric bulk of 1b.

Scheme 2.21 – Inhibition by products 21a and H2O of the 1b-catalysed hydrogenation of 20a in THF.

Chapter 2 – Boron-based FLPs

Previous work in the Ashley group has established that binding of H2O to 1b in the non-donor solvent PhMe is sufficiently reversible to allow for its removal by the addition of molecular sieves.15

Nevertheless, attempts to improve turnover through addition of desiccants (3 Å molecular sieves,

MgSO4) did not lead to any improvement. Furthermore, when equivalent reaction conditions were applied to the hydrogenation of other carbonyl substrates similar or inferior results were observed. For example, switching to the bulkier substrate EtC(=O)Me (20b) was not found to have an appreciable effect on the reaction outcome, with an essentially identical final conversion (35 %) achieved within a similar time-frame (3.5 h; Scheme 2.22).

Scheme 2.22 – Hydrogenation of 20b catalysed by 1b in THF.

It was speculated that the problem of inhibition might be overcome by increasing the steric bulk of the borane catalyst, thereby rendering any adduct formation less favourable. To this end, the catalyst 1b was replaced with bulkier 1c (Scheme 2.23); unfortunately, this was found to lead to a significant reduction in reaction rate, without any increase in maximum turnover (two turnovers could be achieved using 1c, versus 3.5 using 1b). This is qualitatively consistent with previous results for imine hydrogenation (Chapter 2.2.1), and is again attributed to the increased steric bulk of the

– reductant ([1c—H] ) generated from H2 (c.f. Figure 2.4). Interestingly, this reaction showed no significant formation of the condensed side-product 22a, suggesting that the borane catalyst 1b may be directly involved in its formation (a reduced rate of condensation may also partially be due to lower 21a concentration). Turnover in this case appears to be limited by decomposition of the 1c catalyst (rather than product inhibition), as indicated by the emergence of numerous new 19F and 11B NMR resonances during the course of the reaction, consistent with B—C alcoholysis (vide supra; see also Chapter 2.3.5). That decomposition appears to be more significant in this reaction than when using 1b is presumably due to the higher reaction temperature and longer reaction time required.

Scheme 2.23 – Slower hydrogenation of 20a catalysed by 1c in THF.

Chapter 2 – Boron-based FLPs

Based on the inferior results obtained with the bulkier borane 1c it was decided to investigate the opposite approach, in the hope that a less bulky borane might show improved reactivity through formation of a more kinetically reactive borohydride reductant. When less bulky 1a was used as the LA catalyst, however, inferior results were again observed (< 3 turnovers using 1a versus 3.5 for 1b; Scheme 2.24). In this case, the reduced reactivity is attributed to strong binding of the solvent to 1a; again, this is qualitatively consistent with prior imine hydrogenation results (c.f. Figure 2.5).

Scheme 2.24 – Slower hydrogenation of 20a catalysed by 1a in THF.

It was reasoned that solvent inhibition when using 1a might be less of a problem in a weaker donor solvent, with 1,4-dioxane already having been shown to be suitable for borane/solvent hydrogenation catalysis (Table 2.1, entry 19). In line with this hypothesis, hydrogenation of 20a was found to proceed with greatly improved turnover (>15 cycles) in the presence of catalytic 1a in 1,4- dioxane, although somewhat higher temperatures were found to be necessary to ensure a reasonable reaction rate (100 °C versus 80 °C; Scheme 2.25). Crucially, good conversion could be achieved using catalyst loadings low enough to match those typically employed in FLP hydrogenation catalysis. Furthermore, clean formation of alcohol 21a was observed, with no condensation to form 22a. While the reasons for this difference have not been conclusively determined, it seems plausible that

+ condensation is acid-catalysed: either by Brønsted (H from H2 activation) or Lewis (boranes 1a,b) acids (Scheme 2.20; b). Catalysis by the former is expected to be disfavoured upon switching to 1,4-dioxane due to less favourable H2 activation (c.f. Figure 2.6), while catalysis by the latter would be limited by strong coordination to the solvent (vide infra). Condensation is also likely to proceed via an ionic SN1

29 mechanism, which should be slower in a less polar solvent (Hr = 2.22 for 1,4-dioxane vs. 7.52 for THF).

Scheme 2.25 – Effective hydrogenation of 20a catalysed by 1a in 1,4-dioxane.

To gain further insight into this effective catalytic system, VT NMR analysis was performed on a solution of 1a in 1,4-dioxane. Surprisingly, 11B and 19F spectra indicated the formation of a strong adduct even at the elevated temperatures at which hydrogenation reactivity can be observed: no downfield shift in the 11B NMR resonance was observed under these conditions, and nor was there an

Chapter 2 – Boron-based FLPs appreciable change to the separation of the meta and para 19F resonances (Figure 2.7; top). These results suggest that thermal liberation of ‘free’ uncoordinated 1a is likely to be a limiting factor in the rate of catalysis; indeed, this likely accounts for the higher reaction temperatures and longer reaction times needed compared with the earlier 1b/THF system. Nevertheless, similar VT NMR analysis of a stoichiometric mixture of 1a and 1,4-dioxane in C7D8 did confirm weaker binding to 1a for 1,4-dioxane than for THF, with evidence for the onset of adduct dissociation at significantly lower temperature (ca. RT, 60 °C respectively; Figure 2.2, Figure 2.7).

Figure 2.7 – VT 19F (left) and 11B (right) NMR spectra for 1a in 1,4-dioxane (top), and for equimolar 1a

and 1,4-dioxane in C7D8 (bottom).

The ability of 1a in 1,4-dioxane to activate H2 was also confirmed independently. For previous 1/THF systems this had typically been possible by observation of weak [1—H]– 11B NMR resonances at

Chapter 2 – Boron-based FLPs low temperature; however in this case this approach was precluded by the high melting point of 1,4- dioxane (12 °C).29 Instead, admission of an atmosphere of HD (1 bar) was observed by 1H NMR spectroscopy to lead to isotopic scrambling to form H2 (and D2) over the course of several hours at RT (Scheme 2.26). This is due to transient cleavage of the H—D linkage followed by recombination of the resulting H+, H–, D+, and D– fragments in different combinations (Scheme 2.26; a), and has been used

13, 48 to observe reversible H2 activation by other low-strength FLPs. Given sufficient time, such reactions are ultimately expected to furnish a statistical mixture of 1:1:2 H2:D2:HD, since there is no significant thermodynamic preference to form H2/D2 over HD (or vice versa).

Scheme 2.26 – Isotopic scrambling of HD due to reversible activation by 1a in 1,4-dioxane at RT (a), as evidenced by 1H NMR spectroscopy (b).

Interestingly, bulkier borane 1b also proved capable of achieving superior turnover in 1,4-dioxane relative to THF (presumably due to the lack of the condensation side-reaction), but more slowly than when using 1a (such that achieving a similar rate required twice the catalyst loading; Scheme 2.27). This stands in contrast to previous results for hydrogenation in THF, where 1b was the more active species. The reversal can again likely be attributed to weaker binding of 1,4-dioxane to 1a. In THF a general trend is observed, with reactivity improving as steric bulk decreases down the

Chapter 2 – Boron-based FLPs series 1d, 1c, 1b (Figure 2.4); however, upon reaching 1a this trend is disrupted by particularly strong binding to the solvent (Figure 2.5). In 1,4-dioxane such binding is less significant, and the trend is able to continue uninterrupted.

Scheme 2.27 – Slower hydrogenation of 20a catalysed by 1b in 1,4-dioxane.

Chapter 2.3.2 – Substrate scope for catalytic hydrogenation of aldehydes and ketones

Having established a promising system for carbonyl hydrogenation, the scope of the 1a- catalysed reaction in 1,4-dioxane was investigated. While appreciable variation of the alkyl backbone was tolerated, increasing steric bulk was found to lead to a steady decrease in reaction rate and/or turnover (Table 2.2, entries 1-3), ultimately resulting in a complete loss of hydrogenation activity (Table 2.2, entry 4). Again, this trend is entirely consistent with previous observations made for non- carbonyl substrates (c.f. Scheme 2.8, Figure 2.4, for example). It is also consistent with prior computational investigations, which anticipated that rate-limiting hydride transfer would be slower for bulkier substrates.49 Sharp size-based selectivity of this type could potentially be useful for achieving synthetic selectivity. As a proof-of-principle, the hydrogenation conditions were applied to a 1:1 mixture of 20d and 20e. Complete conversion of the slightly less bulky substrate could be achieved, without any noticeable change to its more hindered counterpart (Scheme 2.28), and without any obvious reduction in rate relative to the 20e-free reduction of 20d. Note that this result also confirms that the inability to hydrogenate 20e cannot be due to inhibition by the substrate, further supporting the idea that it is instead due to excessive steric hindrance.

Table 2.2 – Carbonyl hydrogenation catalysed by 1a in 1,4-dioxane.

[1a] / Conv. Entry Substrate Product T / °C t / h mol% / %a

1b 5 100 92 83 20a 21a

2 5 100 90 60 20c 21c

Chapter 2 – Boron-based FLPs

3 10 100 67 80 20d 21d

4 10 100 24 0 20e 21e

5 5 100 17 0 20f 21f

6 10 80 120 84 20g 21g

7 20 80 110 82 20h 21h

8 10 80 25 0 20i 21i

9 10 100 30 14 20j 23j

10 10 80 90 82 20k 21k

11 10 80 90 78 20l 21l

75 12 10 80 24 20m 21m (74)c

97 13 10 80 19 (87)d 20n 21n

14 Complex mixture 10 80 89 67 20o a 1 Measured at 0.1 mmol scale by in situ H NMR spectroscopy, using 1,3,5-trimethoxybenzene or PPh3 in a capillary insert as an internal standard for integration (see Chapter 4.1). b0.2 mmol scale, double concentration. cIsolated yield at 1 mmol scale shown in parentheses. dConversion at 1 mmol scale in parentheses.

Chapter 2 – Boron-based FLPs

Scheme 2.28 – Selective hydrogenation of 20d in the presence of 20e, on the basis of differing steric bulk.

Surprisingly, attempts to hydrogenate the cyclic ketone cyclohexanone (20f) under equivalent conditions were unsuccessful (Table 2.2, entry 5), despite its steric bulk being superficially similar to 20c (and so seemingly below the steric threshold for hydrogenation). One possible explanation it that the rigid cyclic structure of the substrate limits the ability of the alkyl side chains to favourably ‘bend away’ from the bulky borohydride nucleophile (Figure 2.8), thereby increasing its effective size.

Figure 2.8 – A possible account for the unsuccessful 1a-catalysed hydrogenation of 20f.

As well as aliphatic substrates, aromatic ketones could also be effectively hydrogenated (Table 2.2, entries 6 and 7). As with the alkyl analogues, excessive steric bulk was found to lead to a loss of reactivity, for example by installation of substituents in the ortho positions of the aromatic ring (Table 2.2, entry 8). In addition it was found that in order to ensure appreciable turnover, it was necessary to include electron-withdrawing groups on the aromatic ring (Table 2.2, entry 9). Although initial reduction was observed in their absence, this was followed by rapid (presumably acid-catalysed) dehydration of the resulting alcohol to provide an alkene in approximately stoichiometric yield (Scheme 2.29). This is presumably due to conjugative stabilisation of the intermediate carbocation that will be formed during E1 elimination (c.f. condensation of 20a observed during 1b-catalysed hydrogenation in THF, which is expected to proceed via an analogous carbocation intermediate;

Scheme 2.20), and the lack of turnover can be attributed to strong inhibition of 1a by the H2O that is

Chapter 2 – Boron-based FLPs thus released. The issue of moisture tolerance in hydrogenation reactions catalysed by 1a in 1,4- dioxane will be discussed in detail in Chapter 2.4.

Scheme 2.29 – 1a-mediated hydrogenation of acetophenone (20j) in 1,4-dioxane, with subsequent dehydration to form styrene (23j).

Slightly surprisingly, it even proved possible to apply the hydrogenation protocol to the reduction of aldehydes. Given prior observations that suggested much stronger inhibition of catalysis by H2O than by secondary alcohols, and based on the presumption that this is most likely due to steric differences between these otherwise very similar LBs, it was anticipated that primary alcohols (which would form upon aldehyde reduction) might also prove to be strong inhibitors, thereby limiting turnover. Nevertheless, attempts to hydrogenate aromatic aldehydes structurally similar to the previously-studied ketones were largely successful (Table 2.2, entries 10-13).

Note that while these results could be taken to suggest that there is little additional inhibitory effect from primary (rather than secondary) alcohols, it is also possible that any such effect is simply cancelled out by the increased electrophilicity of aldehydes relative to ketones, which could also have an effect on which elementary reaction steps are ultimately rate-limiting (vide infra). Possible support for this latter interpretation can be seen in one of the most noticeable differences between the trends in susceptibility to hydrogenation for aldehyde and ketones. As already discussed, increased steric bulk was observed to lead to slower turnover for ketone substrates, attributed to more difficult hydride transfer. For aldehydes, however, the shortest reaction times were observed to correlate with the bulkiest substrates (Table 2.2, entries 12 and 13). Most likely, this can be explained as being due to increased steric frustration of any borane-alcohol adduct, promoting dissociation and so encouraging turnover. Meanwhile, the inherently higher reactivity expected for aldehydes should help to facilitate hydride transfer for even very bulky substrates (Figure 2.9). The short reaction times for bulky aldehyde substrates were retained at larger scales, after which the alcohols can easily be purified via column chromatography, without significant loss of material (Table 2.2, entries 12 and 13).

Chapter 2 – Boron-based FLPs

Figure 2.9 – A qualitative explanation for the effect of substrate steric bulk on the 1a-catalysed hydrogenation of aldehydes in 1,4-dioxane.

Despite tolerance for steric bulk, the hydrogenation of aldehydes by 1a shows similar sensitivity to ketone hydrogenation regarding electronic factors. Again, poor reactivity was observed in the absence of electron-withdrawing groups; for example, the attempted hydrogenation of unsubstituted benzaldehyde (20o) led to appreciable turnover, but formation of a complicated mixture of mostly unidentified products (indicated by numerous 1H NMR resonances between ca. 4.5 and 5.5 ppm), presumed to arise from further reactions of the initial benzyl alcohol product (21o).

Chapter 2.3.3 – Mechanistic investigations into 1a-catalysed carbonyl hydrogenation in 1,4-dioxane

Nylhén and Privalov have previously used the outcome of computational investigations to propose that the 1a-catalysed hydrogenation of aldehydes and ketones could proceed via a mechanism analogous to 1a-catalysed imine hydrogenation, with the substrate 20 acting as the basic

20, 49 component of an FLP for H2 activation (Scheme 2.30; substrate-assisted H2 activation pathway); and the feasibility of such a pathway has certainly been supported by prior studies reported by the groups of Repo and Stephan.1, 3 Nevertheless, on the basis of the preceding studies into the hydrogenation of 2a catalysed by 1b in THF (Chapter 2.2.1), it seems perhaps more likely that carbonyl hydrogenation proceeds via initial H2 activation by 1a and the 1,4-dioxane solvent, with subsequent

Chapter 2 – Boron-based FLPs

Brønsted acid activation of the substrate (Scheme 2.30; solvent-assisted H2 activation pathway). This would be consistent with both the much higher concentration of the solvent, and its increased basicity relative to typical carbonyls (pKaH = –2.92, –7.2 for 1,4-dioxane and acetone, respectively, in aqueous

5-6 H2SO4). Given the difference in pKa values, such an activation step would probably best be characterised as a strong hydrogen-bonding interaction, in which the substrate replaces one or more of the solvent molecules in the inner coordination sphere of the acidic proton. It must be acknowledged, however, that neither pathway can be conclusively ruled out solely on the basis of current experimental evidence (vide infra). In either case, reduction of the substrate is completed by hydride transfer from the [1a—H]– anion to generate a borane-alcohol adduct 1a·ROH, with the catalytic cycle then completed by dissociation of the product to regenerate the solvated borane. It is speculated that, in addition to its role in H2 activation, the donor solvent may also facilitate turnover by acting as a nucleophilic catalyst to promote product dissociation in this final step (though again, this cannot be categorically confirmed based on current results).

Scheme 2.30 – Proposed mechanism for 1a-catalysed hydrogenation of ketones (and possibly aldehydes) in 1,4-dioxane.

In order to test the validity of the proposed mechanism for ketone hydrogenation, a number of stoichiometric reactions were performed using [Bu4N][1a·H]: a pre-formed source of the active [1a—H]– reductant, paired with a non-coordinating counter-cation.50 Firstly, equimolar mixtures of the salt with substrates 20a and 20g (chosen as a simple, representative aliphatic and aromatic substrate, respectively) were prepared in 1,4-dioxane under N2 and heated to the temperatures under which they had been observed to undergo catalytic hydrogenation (100 °C and 80 °C, respectively) for

Chapter 2 – Boron-based FLPs several hours (Scheme 2.31; a). In neither case was any evidence for hydride transfer observed by 1H, 19F or 11B NMR spectroscopy, confirming that direct reduction of the ‘naked’ substrate cannot be a viable mechanism for the hydrogenation reaction; instead, some form of prior activation must be required.

Scheme 2.31 – Stoichiometric experiments indicate that hydride transfer from [1a·H]– to unactivated ketone substrates is infeasible (a), while transfer to 1a-activated ketones (b) is too slow to account for the rate of the catalytic reaction, in which the concentration of [1a·H]– must be very low (c).

While, in the mechanism proposed in Scheme 2.30, this activation is provided by Brønsted acidic ‘H+’, another alternative might involve activation by the LA catalyst, 1a. With this in mind the reactions with [Bu4N][1a·H] were repeated in the presence of an additional equivalent of 1a (Scheme 2.31; b). This time, NMR spectroscopic analysis showed clear reduction of both substrates by the borohydride reagent, plainly demonstrating that binding of 1a can provide sufficient activation to promote hydride transfer. Nevertheless, the rate of these reactions was relatively low; for example, leading to reduction of less than half an equivalent of 20a after heating to 100 °C for 1 h. Given that the concentration of [1a·H]– in the catalytic reaction is very low (no resonances attributable to the

Chapter 2 – Boron-based FLPs anion can be observed upon independently heating a solution of 1a in 1,4-dioxane to 100 °C under 4 bar H2), this does not seem sufficient to account for the observed rate of the catalytic reaction. Again, this supports the idea that the primary substrate activation pathway involves an alternative interaction with ‘H+’ (Scheme 2.31; c). Furthermore, in the reaction containing 20a, significant

– decomposition of the 1a moiety was observed, with formation of the [B(C6F5)4] anion evident by both NMR spectroscopy and mass spectrometry (this will be discussed further in Chapter 2.3.5); no such decomposition was observed in the catalytic reaction.

While the above results strongly support the mechanism for ketone hydrogenation outlined in Scheme 2.30, very different results were obtained when equivalent reactions were performed using aldehydes as substrates (with 20k and 20n chosen as sterically-different model systems). Addition of an equivalent of [Bu4N][1a·H] to an equimolar solution of 1a and 20k or 20n in 1,4-dioxane led at RT to immediate (within minutes) disappearance of the substrate resonances from the 1H NMR spectrum, consistent with quantitative hydride transfer (Scheme 2.32; a). While the product [Bu4N][1a·OCH2Ar] salts precipitate from 1,4-dioxane solution, subsequent replacement of the solvent with CD2Cl2 and

1 re-analysis by NMR confirms the identity of the products, most obviously by the diagnostic ArCH2O H NMR resonances at ca. 4.5 ppm, and 11B resonances at ca. –2.5 ppm.

In fact, similar reactivity was observed even in the absence of 1a, with addition of one equivalent of [Bu4N][1a·H] to a 1,4-dioxane solution of 20k or 20n leading to immediate reduction (Scheme 2.32; b). While these reactions were not observed to proceed to completion (unlike their counterparts also containing 1a), this is attributed to co-precipitation of [Bu4N][1a·H] with the

[Bu4N][1a·OCH2Ar] product, which separates the reductant and substrate into different phases and thereby prevents further reaction. If the solvent is subsequently removed and replaced with CD2Cl2

(or if CD2Cl2 is used from the start of the reaction), complete reduction is immediately observed.

These results suggest that alternative mechanisms may be possible for the catalytic hydrogenation of aldehydes: either the substrate may be activated by coordination to 1a (Scheme 2.33; a), or it may undergo hydride transfer without any prior activation (Scheme 2.33; b). It is not possible to distinguish between these possibilities without further experimental investigation.

Chapter 2 – Boron-based FLPs

Scheme 2.32 – Stoichiometric reduction of aldehydes 20k and 20n by [Bu4N][1a·H] in the presence (a), and absence (b) of 1a.

Given that earlier investigations into hydrogenation using related systems had suggested an inhibitory effect by simple alcohols such as iPrOH (20b, Scheme 2.21), it seemed reasonable to assume that the formation of 1a·ROH adducts as intermediates during 1a-catalysed carbonyl hydrogenation reactions in 1,4-dioxane might be an important factor in determining overall reaction rates. Certainly, typical catalytic reaction mixtures showed appreciable upfield shifts in their 11B NMR spectra over time (for example, from 3.4 to –2.1 ppm during the hydrogenation of 20a to iPrOH, 21a), consistent with formation of a stronger adduct upon generation and binding of the alcohol product (Figure 2.10). Independent experiments confirmed that addition of iPrOH to 1a in 1,4-dioxane shifts the 11B NMR resonance appreciably upfield; for example to 0.5 ppm for a 1:1 mixture at the same concentration (the pre-formed adduct 1a·HOiPr was used in this and subsequent experiments to ensure precise stoichiometry). Surprisingly, the final 11B NMR resonances for the catalytic reactions were found to be very close to those observed for related anionic [1a—OR]– salts (c.f. –2.3 ppm for [1a—OMe]– in

Chapter 2 – Boron-based FLPs

47 CD2Cl2; similar shifts were also observed during the stoichiometric carbonyl reductions using

[Bu4N][1a·H] described earlier in this section). For comparison, solutions of 1a·HOiPr in non-donor solvents were found to give 11B resonances that lie significantly downfield (for example, 4.3 ppm in

C7D8). Remarkably, these results would appear to indicate that, despite the very low basicity of the 1a/1,4-dioxane catalytic system, the 1a·HOiPr adduct is in fact sufficiently acidic to be almost completely deprotonated by the 1,4-dioxane solvent under these conditions.

Scheme 2.33 – Possible alternative mechanisms for 1a-catalysed hydrogenation of aldehydes in 1,4- dioxane.

Chapter 2 – Boron-based FLPs

Figure 2.10 – Upfield-shifted 11B NMR resonances as an indicator of 1a·HOiPr deprotonation.47

This conclusion was further supported by the observation that addition of 1,4-dioxane to

11 1a·HOiPr in C7D8 led to clear upfield shifts in the B NMR resonance; to 2.1 and 1.1 ppm following addition of one and ten equivalents, respectively (versus 4.3 ppm for 1a·HOiPr alone). 1H NMR spectroscopy showed the resonance for 1,4-dioxane also shifting slightly (to 3.14 or 3.30 ppm respectively; c.f. 3.33 ppm for 1,4-dioxane only),51 consistent with an interaction (Figure 2.10). A

Chapter 2 – Boron-based FLPs possible alternative explanation for shifted NMR resonances might be displacement of the alcohol from the 1a·HOiPr adduct to form 1a·(1,4-dioxane); however, this possibility would seem to be inconsistent with observations in neat 1,4-dioxane (vide supra), and was further discounted on the basis of experiments performed using ‘free’ 1a in place of 1a·HOiPr. Addition of one and ten equivalents of 1,4-dioxane to 1a in non-donor 1,2-difluorobenzene (DFB) led to 11B NMR resonances at 9.8 and 5.7 ppm respectively (Figure 2.11), and 1H NMR resonances at 3.96 and 3.59 ppm, respectively. In all cases, these shifts are significantly downfield compared to when the alcohol is also present.

Figure 2.11 – Control experiments showing the effect of 1,4-dioxane on the 11B NMR resonance of alcohol-free 1a in DFB.

Based on these results it seems likely that, following the initial turnover, the bulk of 1a present in the catalytic reaction mixture will be sequestered as the off-cycle [1a—OR]– anion (Scheme 2.34). The need to re-protonate this weak base in order to re-enter the catalytic cycle will present an additional kinetic barrier to catalytic turnover, and is likely to be a factor behind the elevated reaction temperatures required before these reactions will proceed. This finding also acts as a compelling validation for the initial decision to pursue catalytic carbonyl hydrogenation through the careful exclusion of any stronger LBs; clearly the presence of even moderately strong LBs would lead to irreversible deprotonation of the very powerful 1a·ROH Brønsted acids.

Chapter 2 – Boron-based FLPs

Scheme 2.34 – Reversible deprotonation of intermediate 1a·ROH adducts as a likely limiting factor in the rate of 1a-catalysed carbonyl hydrogenation in 1,4-dioxane.

Chapter 2.3.4 – Aldehyde and ketone hydrogenation at increased pressure

While hydrogenation of suitable ketones and aldehydes 20 under 4 bar H2 was generally found to be capable of proceeding with good turnover, reactions with catalyst loadings low enough to be practically sensible typically required use of extended reaction times in order to achieve good conversions (Table 2.2). Attempts to improve reaction rates by further increasing the reaction temperature were generally hampered by decomposition of the 1a catalyst via B—C alcoholysis (vide supra), a process which begins to become prohibitive under such conditions. Such decomposition was

1 indicated by the appearance of a C6F5H multiplet at 7.25 ppm in the H NMR spectrum, and by formation of multiple sets of 19F and 11B NMR resonances, with the latter generally appearing at ca.

25-30 ppm. While these shifts are somewhat upfield of those previously reported for (C6F5)2BOR borinic esters (which typically appear around 40 ppm),3 this is attributed to weak coordination of the donor solvent [D’Alfonso, Mercandelli, et al. have previously reported similar upfield shifts upon

52 addition of related donors such as THF or MeOH to (C6F5)2BOMe].

Fortunately, even rather modest increases in H2 pressure were found to be capable of dramatically improving the rate of catalytic hydrogenation. For example, while the hydrogenation of

20a using 5 mol% 1a at 100 °C achieved only 83 % conversion after 92 h under 4 bar H2, the equivalent reaction at 10 bar was observed to reach completion within just 6 h (Scheme 2.35; top). Similarly, the aromatic substrate 20h could be hydrogenated to significantly higher conversion at a greatly decreased reaction time following the same increase in pressure (Scheme 2.35; centre). In fact, increasing the reaction pressure even seems to increase the effective scope of the 1a-catalysed hydrogenation protocol. Dramatically, despite showing no signs of reduction during previous investigations at lower pressure, cyclohexanone (20f) was found to undergo rather effective hydrogenation under 10 bar H2 (Scheme 2.35; bottom).

Chapter 2 – Boron-based FLPs

Scheme 2.35 – Greatly-improved rates of carbonyl hydrogenation catalysed by 1a in 1,4-dioxane at increased pressure.

At higher pressures still, the reaction rate can be increased sufficiently to allow for very significant reduction of the reaction temperature. For example, under 50 bar H2 the hydrogenation of 20a can be achieved at an improved rate even after a decrease in temperature to just 50 °C (Scheme 2.36; c.f. Scheme 2.25 where similar conversion at 4 bar, 100 °C required 92 h). In the longer term, the ability to perform hydrogenation at lower temperatures can be anticipated to lead to improved reaction selectivity, and to allow for the hydrogenation of less thermally robust substrates.

Scheme 2.36 – 1a-catalysed hydrogenation of 20a in 1,4-dioxane at reduced temperature, through use of higher pressure.

While an increase in reaction rate is to be expected upon increasing the concentration of one of the reagents, the magnitude of the rate accelerations observed at higher pressure is slightly surprising, and greatly exceeds what would be expected if there were a simple linear relationship between H2 concentration and reaction rate. One possible explanation is that the rate of catalytic turnover is significantly limited by reversible deprotonation of intermediate borane-alcohol adducts, as discussed in the previous section (Scheme 2.34). At higher H2 pressures the concentration of H2 in solution will be increased, and as a result the equilibrium for H2 activation by 1a/1,4-dioxane will lie further towards the ‘solvated H+’/borohydride product. The resulting increase in Brønsted acid concentration should then promote re-protonation of the off-cycle alkoxyborates, and so alter the

100

Chapter 2 – Boron-based FLPs overall 1a speciation in favour of on-cycle species, thereby increasing the rate of turnover (Scheme 2.37).

Scheme 2.37 – A proposed explanation for the dramatically increased rate of 1a-catalysed carbonyl hydrogenation in 1,4-dioxane at increased pressure.

Chapter 2.3.5 – On the stability of the 1a/1,4-dioxane system for catalytic carbonyl hydrogenation

The conditions developed for 1a-catalysed carbonyl hydrogenation are remarkably similar to those reported by the groups of Stephan and Repo for (sub-)stoichiometric hydrogenation of aldehydes and ketones, differing only in the use of an alternative solvent (1,4-dioxane versus PhMe or

1, 3 CD2Cl2) and slightly lower reaction temperatures (80-100 °C versus 110 °C). While the most important difference between the outcomes of these reactions is the observation of catalytic turnover in the donor solvent, there is also a major difference in the fate of the borane 1a. In the reactions reported by Stephan et al., aliphatic ketones undergo 1a-mediated hydrogenation to generate intermediate borane-alcohol adducts, much the same as in the catalytic reactions described above. However, rather than turnover, this initial reduction is followed by decomposition to generate a stoichiometric borinic ester product, with concomitant formation of C6F5H (Scheme 2.38). Crucially, complete decomposition was reported to occur within reaction times significantly shorter than those used for catalytic reactions, in which only relatively minor decomposition is typically observed. Repo et al. reported very similar results using benzaldehyde and benzophenone as substrates (albeit complicated by over-reduction and side-reactions resulting from reactions with the solvent), with 1a decomposition occurring on similar timescales.

101

Chapter 2 – Boron-based FLPs

Scheme 2.38 – Catalytic (top) or stoichiometric (bottom) 1a-mediated carbonyl hydrogenation, depending on the choice of solvent.3

It is therefore clear that, in addition to the factors already outlined in this chapter, the ability to observe turnover in 1a/1,4-dioxane-mediated carbonyl hydrogenation can also be attributed to the relatively high stability of the borane 1a under the reaction conditions, which helps to ensure that turnover is rapid relative to competing catalyst decomposition, thereby increasing maximum turnover. This contrasts with the stoichiometric reactions reported by Stephan and Repo where decomposition is the faster process, and catalysis is therefore precluded (Scheme 2.39; note that in addition to faster decomposition, slower turnover might hypothetically be expected in non-donor solvents; for example, due to slower and less favourable H2 activation).

To a certain extent, the improved stability of 1a in the catalytic reaction is likely to be due to the slightly lower reaction temperature. Indeed, significant decomposition was typically observed when attempts were made to run these reactions at significantly higher temperatures, as was noted in the previous section. Nevertheless, the size of the difference in behaviour suggests that additional factors are also likely to be relevant.

The mechanism for 1a decomposition proposed by Stephan et al. is outlined in Scheme 2.40, and involves elimination of C6F5H from the intermediate 1a·ROH adduct, presumably in an intramolecular fashion. Note that this mechanism is consistent with the very high Brønsted acidity of these adducts (vide supra), which will be necessary for protonation of the electron-poor arene ring. This is also suspected to be the most significant decomposition route for the catalytic reaction, and is consistent with the 1H and 11B NMR resonances seen in those reaction mixtures where 1a degradation is observed (vide supra). As discussed in Chapter 2.3.3, the concentration of any 1a·ROH adducts in the catalytic reaction mixture is actually expected to be very low, with an equilibrium instead favouring the deprotonated [1a—OR]– anion. In the absence of the acidic H atom, intramolecular elimination of

C6F5H from this anion is clearly impossible, and this is likely to be an extra factor in reducing the rate of decomposition (Scheme 2.41).

102

Chapter 2 – Boron-based FLPs

Scheme 2.39 – The importance of relative turnover and decomposition rates in achieving 1a- catalysed carbonyl hydrogenation.

Scheme 2.40 – Mechanism for decomposition of 1a to generate borinic esters.3

103

Chapter 2 – Boron-based FLPs

Scheme 2.41 – Reduced rate of 1a decomposition in 1,4-dioxane due to borane-alcohol adduct deprotonation.

More generally, it can be seen that levelling to the 1,4-dioxane solvent will place an upper limit on the strength of the Brønsted acids that can be present in the catalytic reaction mixture at any significant concentration, which should limit the rate of Brønsted acid-catalysed decomposition processes generally (Scheme 2.42). For example, in their earlier report Stephan et al. discussed an alternative decomposition pathway, based on protonolysis of the [1a·H]– anion by powerful [RR’C=OH]+ acids (although this route was ultimately judged to be less significant than the alternative described in Scheme 2.40).3

Scheme 2.42 – A more general possible mechanism for H+-mediated 1a decomposition, which would be disfavoured by levelling to 1,4-dioxane.

104

Chapter 2 – Boron-based FLPs

One final possible route for decomposition of the 1a unit can be suggested based on an observation made during the reaction of acetone (20a) with [Bu4N][1a·H] and 1a in 1,4-dioxane, which was discussed in Chapter 2.3.3 (Scheme 2.31; a). After heating this reaction mixture to 100 °C for 1 h,

– appreciable decomposition of the 1a moieties was observed, including via formation of the [B(C6F5)4] anion. This is most likely formed via aryl transfer from a 4-coordinate borate ([1a·H]– or [1a—OR]–; these anions are expected to show enhanced nucleophilicity relative to neutral 4-coordinate species) to a 1a-based LA (either 3-coordinate ‘free’ 1a, or its relatively weakly-quenched solvent adduct). In the catalytic mixture the concentration of these latter acidic species is expected to be very low (due to product binding), so decomposition by this route will also be slow (Scheme 2.43). It is possible to draw an analogy with the improved stability sometimes observed for ‘thermally induced’ rather than true FLPs (Chapter 1.4.3), where sequestration of the highly active LA as a reversible adduct can help to suppress undesirable side-reactions and so minimise the rate of decomposition.

Scheme 2.43 – A possible alternative mechanism for 1a decomposition based on aryl group migration, which would be disfavoured by more extensive Lewis acid quenching.

Chapter 2.3.6 – Further development of the catalytic system

Having established an effective system for FLP-catalysed hydrogenation of aldehydes and ketones, it was hoped that similar methods might prove to be applicable to the reduction of more challenging organic carbonyl substrates, such as esters and amides. Unfortunately, all attempts to date to achieve catalytic hydrogenation of simple esters using 1a in 1,4-dioxane have proven to be

105

Chapter 2 – Boron-based FLPs unsuccessful, and failed to achieve even stoichiometric formation of reduced species (for example: Scheme 2.44).

Scheme 2.44 – Unsuccessful hydrogenation of ester 24a using 1a in 1,4-dioxane.

Attempts to hydrogenate amides have been similarly unsuccessful; however, while the lack of ester reduction is attributed simply to the reduced electrophilicity of the carbonyl moiety in these compounds, an additional inhibitory factor is believed to be involved for many amides. 1H NMR analysis of a 2:1 mixture of 1a and a simple formamide such as DMF (N,N-dimethylformamide) in 1,4-dioxane shows two separate sets of DMF resonances (each consisting of one formamide CH and two methyl CH3 signals) in a 1:1 ratio, attributable to the free and 1a-bound species (Figure 2.12). This indicates slow exchange between these species on the NMR timescale, due to strong binding between the catalyst and potential substrate. This is consistent with the increased oxyanion character of amides relative to other organic carbonyls, and is further supported by the observation of a rather upfield-shifted 11B NMR resonance for the same mixture (0.2 ppm). Clearly, such strong binding will provide an additional barrier to productive reactivity, and so must be overcome if amide hydrogenation is to be successfully achieved. Note that in the absence of such a barrier, protonation of amides by ether-solvated protons would be expected to be very feasible (for example, pKa = 6.1 for

12, 53-54 DMF·HOTf/DMF in MeCN, versus pKa = 0.2 for [Et2OH][BF4]/Et2O in MeCN).

It was reasoned that the hydrogenation of challenging ester or amide functionalities might become feasible if the catalytic activity of borane/solvent systems could be improved further through additional, rational modification of the LA and LB components. Prior development of the 1a/1,4-dioxane system (Chapter 2.3.1) had suggested that improved reactivity could be achieved as a result of increased solvent steric bulk. In hopes that this trend might continue, it was decided to investigate the use of 2-methyltetrahydrofuran (2-MeTHF, which has recently drawn attention as an attractive ‘green’ alternative to more common ethereal solvents)55 for borane-catalysed carbonyl hydrogenation.

106

Chapter 2 – Boron-based FLPs

Figure 2.12 – Appearance of separate DMF and 1a·DMF resonances in the 1H NMR spectrum of DMF and 1a (2:1) in 1,4-dioxane, indicative of strong binding.

Unfortunately, initial investigations using 1a suggested no improvement as a result of switching from 1,4-dioxane to 2-MeTHF. For example, the hydrogenation of bulky aldehyde 20n, which had been observed to proceed very effectively in 1,4-dioxane (Table 2.2, entry 13), gave significantly inferior results under otherwise equivalent conditions, stalling after achieving just stoichiometric reduction (Scheme 2.45). The precise reason for this poorer outcome is not entirely clear (19F and 11B NMR spectroscopy did not indicate any 1a decomposition), but it should be noted that the switch of solvent will have a number of effects: it will increase the steric bulk of the LB used for H2 activation, it will alter the strength of the LB (expected to be stronger than 1,4-dioxane, by

29 analogy to unsubstituted THF), and it will increase the polarity of the reaction medium (Hr = 6.97). Each of these factors could potentially be significant; for example, increased solvent basicity and polarity are likely to render 1a·ROH deprotonation more favourable, exacerbating the problem of product inhibition (vide supra).

Scheme 2.45 – Inferior hydrogenation of aldehyde 20n catalysed by 1a in 2-MeTHF.

107

Chapter 2 – Boron-based FLPs

Far superior results were obtained using the bulkier borane 1b, however, with the same substrate observed to undergo catalytic hydrogenation at dramatically decreased temperatures; in fact, with a slight increase in pressure, excellent reactivity could be observed even at RT (Scheme 2.46).

Scheme 2.46 – Low-temperature hydrogenation of aldehyde 20n catalysed by 1a in 2-MeTHF (top), which is believed to be possible due to a lack of strong 1a-quenching interactions (bottom).

The effectiveness of this reaction is attributed to extensive steric hindrance between the 1b catalyst, the 2-MeTHF solvent, and also both the substrate and the product (although there must be at least some interaction with the last of these, based on a significant downfield shift of the 11B NMR resonance at the end of the reaction; from ca. 52.5 to < 0 ppm). Because of this unusually low degree of LA quenching, the borane catalyst is able to easily effect H2 activation (and hence promote catalytic hydrogenation) without the elevated temperatures that were required to promote dissociation when using the previous 1a/1,4-dioxane system. Unfortunately, the activity of the 1b/2-MeTHF system was therefore found to be highly dependent on the substrate being reduced. When the aliphatic ketone substrate 20a was used instead of 20n, significantly elevated reaction temperatures (80 °C) were once again required in order to observe good turnover (Scheme 2.47). This is attributed to stronger inhibition by the less bulky substrate and product (again, supported by 11B NMR spectroscopy, which shows resonances further upfield: at 26.4 and –1.9 ppm at the start and end of the reaction, respectively). At 80 °C this reaction also shows formation of the condensation product 22a in a manner analogous to the similar 1b/THF system previously studied (Scheme 2.20), which will lead to

108

Chapter 2 – Boron-based FLPs

additional H2O-based inhibition (at 40 °C, approximately stoichiometric reduction was observed without condensation).

Scheme 2.47 – Less-successful hydrogenation of ketone 20a catalysed by 1b in 2-MeTHF.

As an alternative strategy, it was reasoned that quenching of the borane catalyst could be minimised if its Lewis acidity was reduced. Note that while use of a weaker LA should also render H2 activation less favourable, it was anticipated that this would largely be balanced out by the generation of a more powerful borohydride reductant (which might be expected to be particularly beneficial for the hydrogenation of weaker electrophiles, such as esters).

The Lewis acid tris(2,4,6-trifluorophenyl)borane (1e, Figure 2.13) has previously been used in

FLP chemistry and found to activate H2 and effect catalytic hydrogenation in conjunction with typical

24, 56, 60 nitrogen-centred LBs such as DABCO (1,4-diazabicyclo[2.2.2]octane, pKaH = 18.29 in MeCN). Due to the reduced degree of fluorination, 1e is a weaker LA than its close analogue 1a, with measurements performed using the common spectroscopic Childs method suggesting its strength is reduced by approximately 30 %.57-58 Correspondingly, theoretical investigations have suggested a ca. 13 kcal/mol

59 reduction in hydride ion affinity, which in turn leads to less favourable H2 activation in conjunction with any specific LB. For example, Paradies et al. have calculated that while H2 activation by a

25 combination of 1a and the imine PhCH=NtBu (2g, calculated pKaH = 15.3 in MeCN) is close to being energetically neutral, the equivalent reaction using 1e is almost 10 kcal/mol less favourable.24

Figure 2.13 – Reduced Lewis acidity and hydride ion affinity (calculated) in a less-fluorinated analogue of 1a. 24, 56-59

109

Chapter 2 – Boron-based FLPs

Gratifyingly, when 1e was used in 2-MeTHF, the hydrogenation of 20a could be achieved at a reasonable rate at temperatures as low as 40 °C. No alcohol condensation was observed, possibly due to the low reaction temperature, or alternatively because the reduced LA strength leads to a less acidic and less reactive 1·ROH adduct.

Scheme 2.48 – Low-temperature hydrogenation of ketone 20a catalysed by 1e in 2-MeTHF.

Since the 1b/2-MeTHF and 1e/2-MeTHF systems had been found to be capable of catalysing the hydrogenation of aldehydes and ketones under only mild heating, it was hoped that at elevated temperatures they might be able to mediate similar reduction of less reactive ester substrates. Unfortunately, all attempts to realise such reactivity yielded results too poor to be of practical value.

Nevertheless, formal catalytic turnover was observed (ca. 3 cycles) for the hydrogenation of HCO2Me (24b) in 2-MeTHF mediated by 1b at 100°C and 10 bar pressure (Scheme 2.49; top).

Scheme 2.49 – Ester hydrogenation in 2-MeTHF mediated by boranes 1b,e.

So far, such turnover has been limited to this specific example (presumably due to the substrate’s minimal steric bulk, which aids hydride transfer from bulky [1b·H]–), whose reduction produces a number of ‘MeO’ species visible by 1H NMR (several singlets are observed to appear in the range 3.1-3.3 ppm, with a concomitant decrease in intensity for the starting material resonances);

110

Chapter 2 – Boron-based FLPs believed to be the expected alcohol (MeOH), as well as MeO-substituted borinic/boronic esters resulting from borane decomposition (supported by the appearance of 11B NMR resonances at 26 and 39 ppm). Despite these limitations, this represents a clear improvement over the 1a/1,4-dioxane system (for which catalytic turnover in ester hydrogenation has yet to be observed). While the goal of FLP-catalysed hydrogenation of esters, amides and many other carbonyl compounds has yet to be achieved, there remains enormous scope for further rational modification of the borane/solvent systems that have been developed thus far.

Chapter 2.4 – Moisture-tolerant borane-catalysed hydrogenation

Supplementary information relating primarily to Chapter 2.4 can be found online at: http://pubs.acs.org/doi/abs/10.1021/acscatal.5b01417

At several points during the development of systems for ketone and aldehyde hydrogenation, specific observations had been made suggesting that the presence or formation of H2O can have a strong inhibitory effect on hydrogenations reactions catalysed by borane/solvent FLPs (see, for example, Scheme 2.21). Nevertheless, these reactions had also clearly demonstrated appreciable tolerance of closely-related hydroxylic species (primary and secondary alcohols), and it seemed likely that with further refinement it might also be possible to use these protocols to achieve moisture tolerance in FLP-catalysed hydrogenation. To facilitate this goal, it was decided to investigate the behaviour of the most thoroughly-studied of these systems, 1a in 1,4-dioxane, upon addition of H2O.

11 Addition of one equivalent of H2O to 1a in 1,4-dioxane leads to an upfield shift in the B NMR resonance from 5.5 ppm to –2.1 ppm; this shift remains unchanged upon addition of a second equivalent of H2O (Figure 2.14). This is clearly similar to the behaviour that was observed previously with iPrOH (21a; Figure 2.10); however, the increased magnitude of the shift and the lack of change upon addition of further H2O are both indicative of a stronger (less reversible) donor-acceptor interaction. This is presumably a consequence of the reduced steric bulk of H2O, which facilitates its

11 binding to the bulky borane. Interestingly, if only half an equivalent of H2O is added, B NMR analysis shows separate resonances for 1a·(1,4-dioxane) and 1a·OH2, both of which are observed at the same shifts as previously (19F NMR spectroscopy also shows two sets of resonances). This indicates that exchange of the H2O ligand between 1a centres is slow on the NMR timescale (which will prove useful for later quantification of H2O content in ‘non-anhydrous’ reaction mixtures).

111

Chapter 2 – Boron-based FLPs

11 Figure 2.14 – B NMR spectra for addition of H2O to 1a in 1,4-dioxane.

11 Just as with iPrOH, the very upfield B NMR resonance observed for 1a·OH2 would seem to indicate extensive deprotonation in 1,4-dioxane solution. The shift is in excellent agreement with

– literatures values reported for [1a·OH] (e.g. –2.1 ppm for [NMe4][1a·OH] in CD2Cl2, with similar values for other salts containing the same anion),61-62 and contrasts with a much more downfield shift of 4.6 ppm for 1a·OH2 in non-donor DFB. Again, addition of 1,4-dioxane to 1a·OH2 (the pre-formed adduct was used to ensure precise equimolar stoichiometry) in a non-donor medium led to progressive shifts in the 11B and 1H NMR spectra consistent with an acid-base interaction (Figure 2.15; the 11B NMR resonance shifts upfield to –0.6 and –2.1 ppm for 1 and 10 equivalents of 1,4-dioxane, respectively; while the 1,4-dioxane 1H NMR resonance shifts downfield to 3.59 ppm for an equimolar mixture).

112

Chapter 2 – Boron-based FLPs

11 Figure 2.15 – Upfield-shifted B NMR resonances as an indicator of 1a·OH2 deprotonation by 1,4- dioxane.61-62

Comparably-shifted NMR resonances were also observed when the model substrate acetone,

11 20a, was added to 1a·OH2 in a similar manner. Addition of 1 or 10 equivalents led to upfield B NMR resonances at –0.4 and –1.9 ppm (Figure 2.16; note that even in neat 20a, pure, dry 1a gives an 11B NMR resonance that is significantly less upfield, at 2.3 ppm), while the 1H NMR resonance for 20a was shifted downfield by ca. 0.1 ppm for a 1:1 mixture. These observations indicate that the carbonyl substrate is capable of deprotonating the 1a·OH2 adduct in a similar manner to 1,4-dioxane, despite

5-6 nominally being a significantly weaker base (pKa = –7.2, –2.92 respectively, in aqueous H2SO4). This is consistent with suggestions from the literature that the effective basicity of carbonyl compounds can be significantly increased in hydrogen-bonding environments,63 and provides indirect support for the idea that these substrates can be activated by solvated ‘H+’ during 1a-catalysed hydrogenation (Scheme 2.30).

113

Chapter 2 – Boron-based FLPs

11 61- Figure 2.16 – Upfield-shifted B NMR resonances as an indicator of 1a·OH2 deprotonation by 20a. 62

Chapter 2.4.1 – Development of a moisture-tolerant system for FLP-catalysed hydrogenation

Despite preliminary NMR studies suggesting a strong interaction of H2O with 1a in 1,4- dioxane, it was nevertheless decided to continue investigations by examining the effect of moisture on a catalytic reaction. Since 20a had been extensively studied as a model substrate its hydrogenation was repeated in a reaction that used pre-formed 1a·OH2 in place of the ‘free’ 1a catalyst, but which was otherwise still prepared under anhydrous conditions. Despite the presence of a full equivalent of

H2O with respect to the catalyst, the substrate was still observed to undergo hydrogenation, in a reaction that was able to achieve relatively good catalytic turnover (Scheme 2.50). This represented the first example of an FLP-catalysed hydrogenation reaction that is capable of tolerating the presence of moisture without complete loss of activity.

114

Chapter 2 – Boron-based FLPs

Scheme 2.50 – FLP-catalysed hydrogenation of 20a despite the presence of H2O.

Though good turnover was observed, the hydrogenation of 20a was noted to proceed at a significantly reduced rate when compared with anhydrous conditions (c.f. Scheme 2.25, where similar conversion was achieved within 92 h for the equivalent anhydrous reaction). Clearly H2O does have a strong inhibitory effect on catalysis, consistent with the established strong binding to 1a. Equally clear, however, is that this inhibition is reversible (at least at elevated temperature), allowing for a noticeable degree of catalytic activity to be retained.

Since re-entry of 1a into the catalytic cycle must presumably require at least transient dissociation of H2O, it was reasoned that improved catalytic activity might be realised through addition of molecular sieves as a desiccant, to remove this ‘free’ moisture from the reaction solution and so limit its inhibitory effect.64 Indeed, moderate turnover could be observed after appreciably reduced reaction times (61 % after 3 days) following addition of 4 Å molecular sieves to the 1a·OH2-based hydrogenation of ketone 20a (Scheme 2.51; top).

Scheme 2.51 – Moisture-tolerance in FLP-catalysed hydrogenation facilitated by addition of a desiccant.

Similar conditions also allowed for successful turnover in the hydrogenation of bulky aldehyde 20n (Scheme 2.51; bottom). Nevertheless, these reactions were still significantly slower than their anhydrous counterparts, suggesting that moisture is not efficiently removed from the reaction mixtures (this may be due to the reduced ability of molecular sieves to absorb moisture at elevated temperatures).65 Furthermore, increased reaction times failed to improve overall turnover, with 1H, 19F and 11B NMR spectroscopy suggesting that this is likely the result of decomposition of the

115

Chapter 2 – Boron-based FLPs

1a catalyst (indicated by formation of C6F5H, for example; vide supra). Given these limitations, as well as the fact that use of additional desiccants is far from ideal as a practical solution, it was decided to focus further efforts on instead improving catalytic performance through variation of the reaction conditions. From a more academic perspective, it can also be argued that the use of desiccants provides a means for side-stepping the problem of moisture-tolerance, rather than truly overcoming it.

Much as had previously been observed for ‘anhydrous’ hydrogenation, attempts to improve the rate of hydrogenation through the use of increased temperatures were largely hampered by catalyst decomposition; however, increased pressures were found to have a highly beneficial effect. Increasing the reaction pressure from 4 to 10 bar led to a much greater rate of turnover for the hydrogenation of 20a, allowing for a significantly reduced reaction time even at lower catalyst loadings (Scheme 2.52; top). Substrate 20n was also effectively hydrogenated under the same conditions (Scheme 2.52; bottom); interestingly, however, it was found to require a longer reaction time. This contrasts with behaviour previously observed under anhydrous conditions, where hydrogenation of 20n was found to proceed particularly rapidly (Table 2.2, entry 13). While at first these results might appear to be contradictory, it should be noted that the effectiveness of 20n hydrogenation was previously attributed to the steric bulk of the product alcohol, which limits the strength of binding to the 1a catalyst, and so minimises product inhibition (Figure 2.9). In ‘wet’ conditions, inhibition is expected to result primarily from binding to H2O, and so this advantage is eliminated. The reduced rate may then simply reflect more difficult underlying hydride transfer to the bulky substrate.

Scheme 2.52 – Improved rate of hydrogenation when using 1a·OH2 as a precatalyst, as a result of higher reaction pressure.

The improved rate of the ‘wet’ reaction at higher pressure is presumed to arise for the same reasons as in the anhydrous system. Increased H2 concentration in solution will increase the reaction rate directly, but will also do so indirectly by generating a more acidic reaction medium, and so

116

Chapter 2 – Boron-based FLPs favouring protonation of strongly quenched [1a·OR]– (R = H, alkyl) adducts in favour of more labile and hence more reactive 1a·ROH adducts (c.f. Scheme 2.37).

Further increasing the pressure to 45 bar led to additional improvement in the reaction time required for 20a hydrogenation (again, even at decreased catalyst loading). As a result, greater quantities of H2O (relative to 1a) could be added to the reaction mixture; while this was observed to lead to greater inhibition (as expected), excellent turnover numbers could still ultimately be achieved (up to at least 39; Scheme 2.53).

Scheme 2.53 – Tolerance of multiple equivalents of H2O (with respect to 1a) in the FLP-catalysed hydrogenation of 20a at 45 bar.

The mechanism of 20a hydrogenation in the above reactions is believed to be essentially identical to that under anhydrous conditions (Scheme 2.54; c.f. Scheme 2.30). It is proposed that the role of H2O is primarily that of an additional, strong inhibitor of the 1a LA, which suppresses the rate

– of catalysis through formation of off-cycle 1a·OH2 and [1a·OH] . This additional inhibition accounts for the reduced rate of the ‘wet’ versus the anhydrous reactions. 11B NMR spectroscopic analysis again supports the idea that in the catalytic reaction 1a is largely sequestered in the form of 4-coordinate anionic borates: resonances for the reaction mixtures are typically observed to be very upfield, around –3 ppm.

Nevertheless, an alternative possibility that cannot be ignored is that the H2O present could actually be involved in the key FLP H2 activation step (Scheme 2.55). Certainly, H2O is expected to be

5, 29 a stronger base than 1,4-dioxane based on aqueous pKa values (–1.7 and –2.92, respectively), and as such ought to be a more effective FLP component for H2 cleavage. Unfortunately, attempts thus far have failed to produce conclusive proof for the feasibility of such reactivity (for example, by attempting to observe HD formation upon admission of H2 to mixtures of 1a and D2O; though it should be noted that 1a is not appreciably soluble in this solvent), and so this is currently judged to be less

117

Chapter 2 – Boron-based FLPs

+ likely than solvent-assisted H2 activation (although even with this mechanism, subsequent H transfer to H2O may still be relevant). The much higher concentration of 1,4-dioxane also supports its proposed involvement in the H2 activation step.

Scheme 2.54 – Proposed mechanism for 1a-catalysed hydrogenation of 20a in wet 1,4-dioxane.

Scheme 2.55 – A possible alternative mechanism for H2 activation by 1a in wet 1,4-dioxane (likely hydrogen bonding interactions not shown).

Chapter 2.4.2 – Application and scope of moisture-tolerant FLP-catalysed hydrogenation

Given the established ability of the 1a/1,4-dioxane catalytic hydrogenation system to tolerate the presence of multiple equivalents of H2O at higher pressure, it was reasoned that it should be possible to prepare these reactions under ‘open bench’ conditions, thereby considerably improving their practicality. Gratifyingly, it was found that the hydrogenation of 20a could be successfully

118

Chapter 2 – Boron-based FLPs achieved using non-anhydrous commercial grade solvent (which was not subjected to any further drying or other purification), following preparation of the initial reaction mixture under open air (Table 2.3, entry 1).

Table 2.3 – Hydrogenation under ‘open bench’ conditions catalysed by 1a in 1,4-dioxane.

[1a] / Conv. / Entry Substrate Product t / h mol% %a

1 2.5 15 94 20a 21a

2b 5 20 50 3c 20g 21g 10 17 >99

4 5 20 77 5b 20k 21k 10 17 >99

6b 5 17 >99

20n 21n

7c 5 18 0 20f 21f

8 2.5 21 48 12b 13b

9c 10 44 >99d 14 15 aReactions performed in Parr reactors using 0.8 mmol substrate at 0.53 M; conversion measured by 1H NMR integration of aliquots. b0.4 mmol scale. c0.2 mmol scale. dUnquantified mixture of isomers.

Similar results were observed for other carbonyl substrates such as the aromatic ketone 20g and the aldehydes 20k and 20n (Table 2.3, entries 2-6), although slightly higher catalyst loadings were typically necessary to ensure good conversion within a similar timeframe. Qualitatively, the scope of this ‘wet’ reaction appears to mirror that of the initial low-pressure anhydrous system that was discussed in Chapter 2.3.2. For example, attempts to hydrogenate cyclohexanone (20f) using the same

119

Chapter 2 – Boron-based FLPs protocol were unsuccessful, which is consistent with prior observations that anhydrous hydrogenation of 20f only proceeded at higher pressures than those required for 20a,g,k,n (c.f. Table 2.2, entry 5).

Much as for ROH tolerance in the lower-pressure anhydrous system, the ability to tolerate

H2O at elevated pressure is attributed to the absence of any strongly basic species in the reaction mixture. As a result, any deprotonation of the highly acidic 1a·OH2 adduct remains reversible, and so irreversible quenching of the LA catalyst is avoided. Based on this analysis, moisture tolerance ought to be achievable for substrates other than carbonyls, provided that neither they nor their hydrogenation products show appreciable Brønsted basicity (Figure 2.17), or are overly susceptible to hydrolysis (e.g. imines). Thus, without any further optimisation, the same catalytic protocol was found to be capable of mediating the catalytic hydrogenation of C=C bonds in the weakly basic substrates 12b and 14 (Table 2.3, entries 8 and 9).

Figure 2.17 – Low basicity as a primary requirement for successful moisture tolerance in hydrogenation reactions catalysed by 1a in 1,4-dioxane.

In order to confirm that the transformations collected in Table 2.3 are indeed examples of moisture-tolerant FLP catalysis, it was important to confirm that the quantity of H2O in these reactions exceeded the quantity of the 1a catalyst (if not, even irreversible adduct formation could still leave an excess of unquenched 1a capable of preforming catalysis). Karl-Fischer titration of the reaction

66 solvent indicated an H2O content of 220 ppm w/w, corresponding to ca. 2.5 mol% relative to substrate at the concentration used (significantly less than the catalyst loading for many of the reactions); however, this measurement alone does not account for many other possible sources of

H2O in the reaction mixture. Such sources include moisture absorbed from the inner lining of the reactor body, from the atmosphere during preparation of the initial reaction mixture, or from the H2

120

Chapter 2 – Boron-based FLPs subsequently admitted (which was not dried beyond the commercial grade which was supplied). With this in mind, control experiments were performed in which solutions of 1a in 1,4-dioxane were prepared on the open bench in a manner identical to the catalytic reaction mixtures. No substrate was added, but the concentration of 1a was chosen to correspond to that used in the reactions with the highest catalyst loading (0.053 M, corresponding to a 10 mol% loading). The reaction vessel was

19 11 then pressurised with H2 (45 bar), and the solution stirred for 45 minutes. In each case, F and B analysis performed on an aliquot taken from the bulk solution showed complete sequestration of 1a as the H2O adduct, confirming that [H2O]>[1a]. By contrast, aliquots taken prior to admission of H2

19 showed a mixture of 1a·OH2 and solvated 1a, with F NMR integration suggesting that at this point

[1a]/[H2O] ≈ 2.2.

As an alternative to permitting reactions to be prepared without the need to rigorously maintain an inert atmosphere, it should also be possible to exploit moisture tolerance to achieve turnover in any hydrogenation reactions that might produce H2O as a stoichiometric side-product. In the long term, reactions of this type could potentially be very important applications for FLP chemistry.

For example, the (sub-)stoichiometric hydrogenation of CO2 to form MeOH was established fairly early on in the development of FLP chemistry,47 but has not yet led to the development of an effective catalytic protocol. This is largely attributed to strong inhibition by the small, hydroxylic products of the reaction: H2O and MeOH. Other useful substrates that might be reduced with release of H2O include chemically-ubiquitous amides and carboxylic acids (Figure 2.18).

Figure 2.18 – Examples of possible FLP-catalysed dehydrative hydrogenation reactions, which are among the long-term goals of FLP chemistry.

To provide a proof-of-principle demonstration of such FLP-catalysed dehydrative hydrogenation, the reduction of acetophenone (20j) was re-examined under higher pressure

121

Chapter 2 – Boron-based FLPs

conditions. After heating to 100 °C under 45 bar H2 for 61 h, complete deoxygenative reduction was observed to ultimately form ethylbenzene 25j (Scheme 2.56). Crucially, clear catalytic turnover was observed, despite the reaction necessarily producing H2O as a stoichiometric by-product. Interestingly, 1H NMR spectroscopy indicated very clean reduction, with 25j the only species observable in the final reaction mixture (aside from the solvent). This contrasts with the results previously obtained at 4 bar pressure, where roughly-stoichiometric reduction yielded styrene (23j) as the final product (Scheme 2.29). In fact, NMR analysis of an aliquot taken from the higher pressure reaction at partial conversion did show 1H resonances consistent with styrene, as well as additional signals very similar to those for the alcohol 21j, which are attributed to formation of the condensed ether product 22j. A simple mechanistic explanation for these observations is outlined in Scheme 2.57, with formation of 22j and 23j proposed to be sufficiently reversible to allow for further reduction at higher pressure.

Scheme 2.56 – Dehydrative hydrogenation of 20j to 25j catalysed by 1a in 1,4-dioxane, despite

formation of H2O as a stoichiometric by-product.

Scheme 2.57 – Proposed mechanism for 1a-catalysed hydrogenation of 20j to 25j in 1,4-dioxane.

122

Chapter 2 – Boron-based FLPs

Chapter 2.5 – Conclusions

Chapter 2.5.1 – Summary and key results

The results described in this chapter demonstrate that it is possible to achieve both of the primary goals of this thesis – FLP-catalysed carbonyl hydrogenation and moisture-tolerant FLP catalysis – through the use of very simple FLP systems based on common FLP borane LAs, including the commercially-available archetype of this family of compounds: B(C6F5)3 (1a). Solutions of these boranes in typical ethereal solvents have proven to be very effective for both the activation of H2 and subsequent catalytic hydrogenation of a wide variety of unsaturated substrates; from simple imines and N-heterocycles to the first examples of FLP-catalysed hydrogenation of furans, aldehydes and ketones. Particularly effective is the hydrogenation of weakly basic substrates, which is attributed to the very high strength of the Brønsted acid generated following H2 heterolysis.

Perhaps the most important aspect of the borane/solvent systems investigated is their ability to perform effective catalytic hydrogenation in the complete absence of even moderately strong Brønsted bases. As a result they are able to avoid the irreversible formation of inactive oxyborate salts that is typically observed when simple FLPs are exposed to hydroxylic species such as alcohols and H2O. The resulting tolerance for these compounds can then be exploited to allow for catalytic turnover in the hydrogenation of aldehydes and ketones to alcohols; or to remove the need for rigorously anhydrous reaction conditions in a variety of catalytic hydrogenation reactions.

Early development has identified 1a in 1,4-dioxane as a successful and broadly-applicable catalytic system, but further investigations have shown that there remains considerable scope for improvement through additional variation of both the borane and solvent. Variations between different catalytic systems can typically be rationalised by considering simple steric and electronic factors; however, the understanding of these reactions is unfortunately not yet at the level where the effect of such factors can reliably be predicted in advance. The need to avoid the presence of strong bases or Brønsted acid-sensitive functional groups also places inevitable limitations on substrate scope for both anhydrous hydrogenation of carbonyl substrates, and moisture-tolerant catalytic hydrogenation in general.

Chapter 2.5.2 – Relation to subsequent publications

At the time it was performed, much of the work described in this chapter was without precedent in the literature. This includes the development of protocols for successful FLP-catalysed hydrogenation of aldehydes and ketones, and for moisture-tolerant FLP-catalysed hydrogenation of

123

Chapter 2 – Boron-based FLPs various weakly-basic substrates. Nevertheless, since this work was completed several reports have been published describing the development of closely related chemistry, and it is useful to draw comparisons between the two bodies of work.

x ‘Enabling Catalytic Ketone Hydrogenation by Frustrated Lewis Pairs’: Stephan et al., 201467

In the first of these relevant reports, which was published back-to-back with most of the results described in Chapter 2.3,68 Stephan and Mahdi described the development of a very similar system for FLP-catalysed ketone hydrogenation (Scheme 2.58). While the most-thoroughly studied system in Chapter 2.3 used 1a as a catalyst for aldehyde and ketone hydrogenation in 1,4-dioxane, these authors reported that solutions of the same borane in Et2O or iPr2O are capable of catalysing the hydrogenation of 16 different aliphatic ketones, as well as one aromatic ketone and one aliphatic aldehyde [inferior results were also reported using Ph2O or (Me3Si)2O as solvent, or using roughly stoichiometric quantities of Et2O in C7D8]. Apart from the choice of ethereal solvent the only significant difference between the two systems was the choice of reaction conditions, with the reactions in

Et2O/iPr2O typically reported to proceed more rapidly and at slightly lower temperatures, but at the cost of significantly higher reaction pressures (60 bar). Unfortunately, these differences preclude a precise comparison of the activities of these different systems.

Scheme 2.58 – Hydrogenation of carbonyl compounds catalysed by 1a in Et2O or iPr2O, as reported by Stephan and Mahdi.67

Given the clear similarity between the two protocols, it is perhaps unsurprising that they were reported to show broadly similar reactivity. While minor differences can be noted (for example, Stephan and Mahdi reported successful hydrogenation of acetophenone 20j to the alcohol 21j, without significant formation of styrene, 23j), for the most part comparable steric and functional group tolerances were reported. In particular, neither system was successfully applied to the hydrogenation of model esters, amides, or amine-containing substrates. The results of this paper can therefore be taken as evidence that further supports the conclusions drawn in this chapter (including

124

Chapter 2 – Boron-based FLPs the importance of excluding strong bases when trying to achieve C=O hydrogenation, although this aspect was not discussed by the authors in their report).

The mechanism the authors proposed for carbonyl hydrogenation in Et2O/iPr2O is identical to that suggested in Scheme 2.30; including the suggestion that the solvent is the primary base involved in H2 activation, and that the “resulting protonated ether then associates with [the] ketone via a hydrogen-bonding interaction [which] activates the carbonyl fragment”. Interestingly, while direct mechanistic studies were not reported, support for this activation mechanism was provided by independent isolation and X-ray analysis of the related salt [iPr2O—H—OC(Et)CH2Ph][B(C6F5)4] (Figure 2.19). Based on the observed O—H distances it was judged that the cation can best be described as

+ ‘[iPr2OH] ’ hydrogen bonded to a ketone. The authors also highlighted the likely importance of the ethereal solvent in stabilising highly acidic protonated intermediates (c.f. Scheme 2.42). A more recent DFT study into the 1a/Et2O system has provided further strong support for the proposed mechanism.69

Figure 2.19 – The salt [iPr2O—H—OC(Et)CH2Ph][B(C6F5)4]: a model for an activated ketone intermediate formed during catalytic hydrogenation, isolated by Stephan and Mahdi.67

x ‘Facile Protocol for Catalytic Frustrated Lewis Pair Hydrogenation and Reductive Deoxygenation of Ketones and Aldehydes’: Stephan et al., 201570

In a follow-up to their first report, the same authors were able to expand the range of LBs suitable for use in the FLP-catalysed hydrogenation of carbonyls to alcohols. Instead of ethereal solvents, D-cyclodextrin (D-CD) or 4 Å molecular sieves were used in PhMe to effect the hydrogenation of a very similar range of alkyl substrates under essentially identical conditions, via a mechanism proposed to be essentially identical (Scheme 2.59). Again, the reactions can be seen to include only very weak oxygen-centred LBs (specifically, the LBs in these reactions are the hydroxyl groups found on the D-CD, or on the surface of the molecular sieves); based on the conclusions drawn in this chapter, this is expected to be a crucial factor in their success.

125

Chapter 2 – Boron-based FLPs

Scheme 2.59 – Hydrogenation of alkyl carbonyl compounds catalysed by 1a and (D-CD) or 4 Å molecular sieves in PhMe, as reported by Stephan and Mahdi.70

Interestingly, when the molecular sieve-based protocol was applied to the hydrogenation of aryl alkyl ketones, catalytic formation of dehydrated alkene products was observed (Scheme 2.60). This can be contrasted with the hydrogenation of acetophenone 20j in 1,4-dioxane, which gave either stoichiometric styrene (23j; Scheme 2.29) or catalytic ethylbenzene (25j; Scheme 2.56) depending on the pressure of H2. Whether turnover in this molecular sieve-based system is due to turnover in the catalytic hydrogenation step occurring prior to subsequent dehydration, to removal of H2O by the desiccant LB, or to inherent moisture tolerance in this catalytic system, was not discussed in this report.

Scheme 2.60 – Hydrogenation of aryl alkyl ketones catalysed by 1a and 4 Å molecular sieves in PhMe, as reported by Stephan and Mahdi.70

Different results still were reported for the same system using diaryl ketone substrates, which underwent complete deoxygenative hydrogenation (Scheme 2.61; c.f. Scheme 2.56). Again, it was not discussed whether tolerance of the moisture formed is due to its removal from solution by the molecular sieves (perhaps the more likely possibility), or more fundamental moisture tolerance in this system.

Scheme 2.61 – Hydrogenation of diaryl ketones catalysed by 1a and 4 Å molecular sieves in PhMe, as reported by Stephan and Mahdi.70

Surprisingly, mechanistic investigations suggested that the ketone starting materials play a crucial role in the deoxygenation step of these reactions, with control experiments showing no further reaction of benzyl alcohols in their absence (Scheme 2.62). Given the similarity of the reactions, it

126

Chapter 2 – Boron-based FLPs seems feasible that the substrate may behave similarly during deoxygenative hydrogenation of 20j in 1,4-dioxane.

Scheme 2.62 – Mechanism for benzyl alcohol deoxygenation during ketone hydrogenation catalysed by 1a and 4 Å molecular sieves in PhMe, proposed by Stephan and Mahdi.70

x Tayseer Mahdi PhD thesis, 201571

A final development in this field by the same group was reported in the PhD thesis of Tayseer Mahdi. In another variation of the same 1a-based hydrogenation protocol, the oxygen-centred Lewis bases were replaced with a co-catalytic quantity of the hydride [NMe4][1a·H]. Following in situ hydride transfer a four-coordinate 1a-based alkoxide is formed, which can act as another viable LB component

+ – for H2 activation in conjunction with ‘free’ 1a. Subsequent H and H transfer steps then effect hydrogenation of carbonyl substrates (Scheme 2.63). The precise mechanism that was proposed for this catalytic reaction is shown in Scheme 2.64, and involves LA activation of the substrate by 1a. Based on the results described in Chapter 2.3.3, which suggest that such activation may be unlikely for ketone substrates, an alternative mechanism based on Brønsted acid activation can also be proposed, which more closely mirrors those for R2O or molecular sieve-based reactions.

Scheme 2.63 – Hydrogenation of ketones catalysed by 1a and [NMe4][1a·H]in PhMe, as reported by Stephan and Mahdi.71

127

Chapter 2 – Boron-based FLPs

Scheme 2.64 – Mechanism for hydrogenation of ketones catalysed by 1a and [NMe4][1a·H] in PhMe proposed by Stephan and Mahdi (a);71 and an alternative pathway (b).

The thesis also reported that hydrogenation of 4-heptanone can be catalysed by 1a in the solvent 2,4-dimethylpentan-3-ol, instead of the ethers already discussed. This result clearly indicates the ability of 1a/ROH combinations to effect H2 activation, and suggests that such a reaction step might be feasible in other hydrogenation reactions that produce alcohol products, including the 1a- catalysed hydrogenation of carbonyls in 1,4-dioxane described in Chapter 2.3.2. Nevertheless, it is still judged that this is likely to be less significant than solvent-mediated H2 activation in this system, based on the relative concentration of the two bases.

Scheme 2.65 – Hydrogenation of 4-heptanone catalysed by 1a in 2,4-dimethylpentan-3-ol reported

71 by Stephan and Mahdi (a), which indicates the feasibility of H2 activation by 1a/ROH Lewis pairs (b).

128

Chapter 2 – Boron-based FLPs

x ‘Moisture-Tolerant Frustrated Lewis Pair Catalyst for Hydrogenation of Aldehydes and Ketones’: Soós et al., 201572

In a subsequent report that was published near-simultaneously to most of the results described in Chapter 2.4,73 the Soós group described the development of a new borane/solvent FLP for the catalytic hydrogenation of simple aldehydes and ketones. Unlike previous systems, which had been based on commercially-available 1a in various ethereal solvents, this new system employed rather more elaborate, purpose-designed boranes, with optimal results achieved in THF solvent. Computational investigations were carried out into the mechanism of these reactions, with the results strongly supporting the mechanisms proposed for earlier systems (for example: Scheme 2.30), as well as highlighting the potential for stabilisation of strongly Brønsted acidic intermediates and transition states through hydrogen-bonding interactions (c.f. Scheme 2.42).

The most notable aspect reported for this new system was its ability to tolerate the presence of moisture (more than an equivalent relative to the borane) while still demonstrating catalytic activity (Scheme 2.66; a), thus allowing for some successful non-anhydrous reactions to be prepared on the open bench. These could be run at considerably lower temperature than the open bench 1a/1,4- dioxane reactions whose development has been described in this chapter (50 °C versus 100 °C), although they required significantly higher reaction pressures (100 bar versus 45 bar), and the reported scope was limited to a handful of aromatic aldehydes (no ketones or non-carbonyl substrates).

While the possibility of irreversible deprotonation of borane-alcohol adducts was briefly acknowledged in their introduction (and computational studies suggested that these adducts can be appreciably stabilised by hydrogen-bonding), the authors ultimately attributed the ability of their system to tolerate moisture to careful tailoring of the precise steric and electronic parameters of their borane catalysts. In fact, a direct comparison was even made to unsuccessful results obtained using 1a under equivalent conditions, in order to emphasise the perceived importance of this LA design (unfortunately, these 1a-based control reactions were all performed at 50 °C: significantly below the temperatures generally required if effective hydrogenation is to be observed with 1a in 1,4-dioxane). Given the fact that moisture tolerance using 1a-based systems is, actually, possible (as established in the previous section), it seems plausible that the significant moisture tolerance observed by Soós et al. may to some extent be due to an identical explanation (i.e. to the absence of strong bases, which avoids irreversible O—H activation), in addition to any specific benefits that may arise from the careful choice of LA catalyst.

129

Chapter 2 – Boron-based FLPs

Scheme 2.66 – Moisture-tolerant hydrogenation of carbonyl compounds reported by Soós et al. using a novel borane in THF (a);72 and control experiments showing no reactivity for 1a-based systems under equivalent conditions (b), although such reactivity is in fact possible (c).

74 x ‘B(C6F5)3-Catalyzed Reductive Amination using Hydrosilanes’: Ingleson et al., 2016

The final and most recently-published report that will be discussed in this section relates to the development by the Ingleson group of a 1a-catalysed protocol for reductive amination using a hydrosilane reductant (Scheme 2.67). The reaction involves in situ condensation of amines and aldehydes/ketones to form imines, which then undergo 1a-catalysed hydrosilylation to generate the final product amines. The reactions could be prepared under non-anhydrous conditions, and also produce H2O as an intermediate by-product prior to the 1a-catalysed reduction step; they thus represent additional examples of moisture-tolerant FLP-type reactivity (the possibility that this tolerance is simply due to removal of H2O from the reaction mixture through 1a-catalysed dehydrocoupling with the silane was considered by the authors, but discounted based on the much lower rate of this side-reaction).

Much as in the development of moisture-tolerant FLP-catalysed hydrogenation described in Chapter 2.4, more forcing reaction conditions were required when compared with equivalent

75 anhydrous reductions, due to reversible inhibition by H2O. Even more significantly, the authors concluded that formation of inactive [1a·OH]– anion plays a significant role in the reaction, with re-

130

Chapter 2 – Boron-based FLPs protonation crucial to ensure at least transient re-release of catalytically-active 1a. As a result, the success of this reaction is highly dependent on the basicity of the amines (and imines) employed, with the authors noting that “the amine and corresponding imine need to be of sufficiently low Brønsted basicity to reversibly deprotonate [1a·OH2]” (Scheme 2.68). Again, this closely echoes one of the key conclusions drawn in this chapter regarding moisture-tolerant FLP-catalysed hydrogenation (c.f. Figure 2.17).

Scheme 2.67 – Reductive amination using a hydrosilane reductant catalysed by 1a, as reported by Ingleson et al.74

Scheme 2.68 – Dependence on Brønsted basicity of the outcome of 1a-catalysed reductive amination, as reported by Ingleson et al.74

The absolute strength of the Brønsted bases tolerated is clearly higher for this 1a-catalysed

+ 76 reductive amination than for 1a-catalysed hydrogenation (e.g. pKa = 4.6 for PhNH3 in H2O). Indeed,

131

Chapter 2 – Boron-based FLPs

the authors did attempt to replace the hydrosilane reductant in their reaction with H2, but without success. This was attributed to inherently higher activity of 1a as a hydrosilylation rather than hydrogenation catalyst, which is expected to compensate for less favourable release of ‘free’ 1a from the [1a·OH]– anion.

Chapter 2.5.3 – Directions for future work

Between the work described throughout this chapter and related research efforts by other groups, the past several years have seen dramatic progress in the development of alcohol and moisture tolerant FLP hydrogenation catalysis using borane LAs. Alongside the development of several successful systems for carbonyl hydrogenation and open bench catalysis, preliminary mechanistic investigations (both experimental and computational) have provided numerous important insights into the elementary reaction steps (both on- and off-cycle) that control the outcome of these previously-unprecedented reactions.

Nevertheless, there remains enormous scope for further study and optimisation of this chemistry. Results so far support the idea that success is largely dependent on the use FLPs that avoid the use of strong Brønsted bases, yet so far only a handful of such systems have been investigated in any detail. As a result, there remains enormous scope for further variation and improvement of these reactions. For example, the results outlined in Chapter 2.3.6 suggest that dramatic improvements in reactivity may still be achievable through further optimisation of borane/solvent FLPs. Meanwhile, results reported by Stephan et al. have demonstrated the feasibility of using a wider range of weak LBs than just simple ethers. To date these have exclusively been oxygen-centred; however, similar results may also be achievable using weak phosphine or amine bases, for example.10-11, 77 While such compounds may be less synthetically convenient than simple ethers, they should also be better suited to rational modification of their steric and electronic properties: potentially a great advantage for the iterative design of optimised catalytic systems. Ultimately, the development of improved hydrogenation protocols may also allow for their extension beyond simple ketone and aldehyde substrates to less reactive substrates such as esters, carboxylic acids, or even CO2.

Alongside the development of improved and more diverse FLP catalysts, improved understanding could also be gained through additional mechanistic investigations into established catalysts. While preliminary investigations have provided strong support for a basic common mechanism, it would be incredibly useful to supplement these results with a more detailed kinetic analysis. As well as confirming the reaction mechanism (including potentially for aldehydes, where multiple possibilities exist: Scheme 2.33), such analysis could also help to establish both the rate- limiting step, and any possible mass-transfer limitations in these multi-phasic (gas/liquid interface)

132

Chapter 2 – Boron-based FLPs reactions. To emphasise the value of such studies, Scheme 2.69 shows a summary of many of the numerous elementary reaction steps that may be involved in the 1a-catalysed hydrogenation of carbonyls 20 in 1,4-dioxane. Further mechanistic analysis could aid in identifying the most important of these steps.

Scheme 2.69 – A simplified summary of elementary reaction steps that may be important during 1a- catalysed carbonyl hydrogenation in 1,4-dioxane (additional possible H-bonding arrangements not shown).

Chapter 2.6 – References for Chapter 2

1. Lindqvist, M.; Sarnela, N.; Sumerin, V.; Chernichenko, K.; Leskela, M.; Repo, T., Dalton Transactions 2012, 41, 4310. 2. Clark, E. R.; Ingleson, M. J., Angewandte Chemie International Edition 2014, 53, 11306. 3. Longobardi, L. E.; Tang, C.; Stephan, D. W., Dalton Transactions 2014, 43, 15723. 4. Bergquist, C.; Bridgewater, B. M.; Harlan, C. J.; Norton, J. R.; Friesner, R. A.; Parkin, G., Journal of the American Chemical Society 2000, 122, 10581. 5. Arnett, E. M.; Wu, C. Y., Journal of the American Chemical Society 1960, 82, 4999. 6. Campbell, H. J.; Edward, J. T., Canadian Journal of Chemistry 1960, 38, 2109. 7. Parks, D. J.; Blackwell, J. M.; Piers, W. E., The Journal of Organic Chemistry 2000, 65, 3090.

133

Chapter 2 – Boron-based FLPs

8. Spies, P.; Erker, G.; Kehr, G.; Bergander, K.; Frohlich, R.; Grimme, S.; Stephan, D. W., Chemical Communications 2007, 5072. 9. Haav, K.; Saame, J.; Kütt, A.; Leito, I., European Journal of Organic Chemistry 2012, 2167. 10. Greb, L.; Oña-Burgos, P.; Schirmer, B.; Grimme, S.; Stephan, D. W.; Paradies, J., Angewandte Chemie International Edition 2012, 51, 10164. 11. Greb, L.; Tussing, S.; Schirmer, B.; Ona-Burgos, P.; Kaupmees, K.; Lokov, M.; Leito, I.; Grimme, S.; Paradies, J., Chemical Science 2013, 4, 2788. 12. Morris, R. H., Chemical Reviews 2016, 116, 8588. 13. Hounjet, L. J.; Bannwarth, C.; Garon, C. N.; Caputo, C. B.; Grimme, S.; Stephan, D. W., Angewandte Chemie International Edition 2013, 52, 7492. 14. Scott, D. J. Next Generation ‘Frustrated’ Lewis Pair Catalysts: Towards Reduction of Amides from Regenerable Hydrides and Catalytic C-H Bond Reactions, Imperial College London, 2013. 15. Ashley, A. E.; Herrington, T. J.; Wildgoose, G. G.; Zaher, H.; Thompson, A. L.; Rees, N. H.; Krämer, T.; O’Hare, D., Journal of the American Chemical Society 2011, 133, 14727. 16. Lancaster, S., ChemSpider Synthetic Pages 2003, 215, http://cssp.chemspider.com/215. 17. Massey, A. G.; Park, A. J., Journal of Organometallic Chemistry 1966, 5, 218. 18. Horton, A. D.; de With, J.; van der Linden, A. J.; van de Weg, H., Organometallics 1996, 15, 2672. 19. Horton, A. D.; de With, J., Chemical Communications 1996, 1375. 20. Chase, P. A.; Jurca, T.; Stephan, D. W., Chemical Communications 2008, 1701. 21. Chen, D.; Klankermayer, J., Chemical Communications 2008, 2130. 22. Johnson Ii, D. C.; Widlanski, T. S., Tetrahedron Letters 2004, 45, 8483. 23. Rokob, T. A.; Hamza, A.; Stirling, A.; Pápai, I., Journal of the American Chemical Society 2009, 131, 2029. 24. Tussing, S.; Greb, L.; Tamke, S.; Schirmer, B.; Muhle-Goll, C.; Luy, B.; Paradies, J., Chemistry – A European Journal 2015, 21, 8056. 25. Tussing, S.; Kaupmees, K.; Paradies, J., Chemistry – A European Journal 2016, 22, 7422. 26. Abdur-Rashid, K.; Fong, T. P.; Greaves, B.; Gusev, D. G.; Hinman, J. G.; Landau, S. E.; Lough, A. J.; Morris, R. H., Journal of the American Chemical Society 2000, 122, 9155. 27. Rokob, T. A.; Hamza, A.; Pápai, I., Journal of the American Chemical Society 2009, 131, 10701. 28. Travis, A. L.; Binding, S. C.; Zaher, H.; Arnold, T. A. Q.; Buffet, J.-C.; O'Hare, D., Dalton Transactions 2013, 42, 2431. 29. In Handbook of Chemistry and Physics. 94th ed.; CRC, 2013.

134

Chapter 2 – Boron-based FLPs

30. Stephan, D. W.; Greenberg, S.; Graham, T. W.; Chase, P.; Hastie, J. J.; Geier, S. J.; Farrell, J. M.; Brown, C. C.; Heiden, Z. M.; Welch, G. C.; Ullrich, M., Inorganic Chemistry 2011, 50, 12338. 31. Geier, S. J.; Chase, P. A.; Stephan, D. W., Chemical Communications 2010, 46, 4884. 32. Pruckmayr, G.; Wu, T. K., Macromolecules 1973, 6, 33. 33. Pruckmayr, G.; Wu, T. K., Macromolecules 1978, 11, 662. 34. Johnson, F., In Friedel-Crafts and Related Reactions., Vol IV, Interscience, 1965. 35. Chivers, T.; Schatte, G., European Journal of Inorganic Chemistry 2003, 3314. 36. Wuts, P. G. M.; Greene, T. W., In Protective Groups in Organic Synthesis. 4th ed.; John Wiley & Sons, 2007. 37. Chiang, Y.; Whipple, E. B., Journal of the American Chemical Society 1963, 85, 2763. 38. R. Crampton, M.; A. Robotham, I., Journal of Chemical Research, Synopses 1997, 22. 39. Mahdi, T.; Heiden, Z. M.; Grimme, S.; Stephan, D. W., Journal of the American Chemical Society 2012, 134, 4088. 40. Welch, G. C.; Masuda, J. D.; Stephan, D. W., Inorganic Chemistry 2006, 45, 478. 41. Birkmann, B.; Voss, T.; Geier, S. J.; Ullrich, M.; Kehr, G.; Erker, G.; Stephan, D. W., Organometallics 2010, 29, 5310. 42. Holschumacher, D.; Bannenberg, T.; Hrib, C. G.; Jones, P. G.; Tamm, M., Angewandte Chemie International Edition 2008, 47, 7428. 43. Chapman, A. M.; Haddow, M. F.; Wass, D. F., Journal of the American Chemical Society 2011, 133, 18463. 44. Zaher, H.; Ashley, A. E.; Irwin, M.; Thompson, A. L.; Gutmann, M. J.; Kramer, T.; O'Hare, D., Chemical Communications 2013, 49, 9755. 45. Custelcean, R.; Jackson, J. E., Chemical Reviews 2001, 101, 1963. 46. Greb, L.; Daniliuc, C.-G.; Bergander, K.; Paradies, J., Angewandte Chemie International Edition 2013, 52, 5876. 47. Ashley, A. E.; Thompson, A. L.; O'Hare, D., Angewandte Chemie International Edition 2009, 48, 9839. 48. Miller, A. J. M.; Labinger, J. A.; Bercaw, J. E., Journal of the American Chemical Society 2010, 132, 3301. 49. Nyhlen, J.; Privalov, T., Dalton Transactions 2009, 5780. 50. Blackwell, J. M.; Morrison, D. J.; Piers, W. E., Tetrahedron 2002, 58, 8247. 51. Fulmer, G. R.; Miller, A. J. M.; Sherden, N. H.; Gottlieb, H. E.; Nudelman, A.; Stoltz, B. M.; Bercaw, J. E.; Goldberg, K. I., Organometallics 2010, 29, 2176.

135

Chapter 2 – Boron-based FLPs

52. Donghi, D.; Maggioni, D.; Beringhelli, T.; D'Alfonso, G.; Mercandelli, P.; Sironi, A., European Journal of Inorganic Chemistry 2008, 1645. 53. Kolthoff, I. M.; Chantooni, M. K.; Bhowmik, S., Analytical Chemistry 1967, 39, 1627. 54. Kilgore, U. J.; Roberts, J. A. S.; Pool, D. H.; Appel, A. M.; Stewart, M. P.; DuBois, M. R.; Dougherty, W. G.; Kassel, W. S.; Bullock, R. M.; DuBois, D. L., Journal of the American Chemical Society 2011, 133, 5861. 55. Pace, V.; Hoyos, P.; Castoldi, L.; Domínguez de María, P.; Alcántara, A. R., ChemSusChem 2012, 5, 1369. 56. Nicasio, J. A.; Steinberg, S.; Inés, B.; Alcarazo, M., Chemistry – A European Journal 2013, 19, 11016. 57. Sivaev, I. B.; Bregadze, V. I., Coordination Chemistry Reviews 2014, 270–271, 75. 58. Childs, R. F.; Mulholland, D. L.; Nixon, A., Canadian Journal of Chemistry 1982, 60, 801. 59. Heiden, Z. M.; Lathem, A. P., Organometallics 2015, 34, 1818. 60. Coetzee, J. F.; Padmanabhan, G. R., Journal of the American Chemical Society 1965, 87, 5005. 61. Hewavitharanage, P.; Danilov, E. O.; Neckers, D. C., The Journal of Organic Chemistry 2005, 70, 10653. 62. Bibal, C.; Santini, C. C.; Chauvin, Y.; Vallee, C.; Olivier-Bourbigou, H., Dalton Transactions 2008, 2866. 63. Palm, V. A.; Haldna, Ü. L.; Talvik, A. J., In The Carbonyl Group, John Wiley & Sons, Ltd., 1966. 64. Thomson, J. W.; Hatnean, J. A.; Hastie, J. J.; Pasternak, A.; Stephan, D. W.; Chase, P. A., Organic Process Research & Development 2013, 17, 1287. 65. Breck, D. W.; Eversole, W. G.; Milton, R. M.; Reed, T. B.; Thomas, T. L., Journal of the American Chemical Society 1956, 78, 5963. 66. Fischer, K., Angewandte Chemie 1935, 48, 394. 67. Mahdi, T.; Stephan, D. W., Journal of the American Chemical Society 2014, 136, 15809. 68. Scott, D. J.; Fuchter, M. J.; Ashley, A. E., Journal of the American Chemical Society 2014, 136, 15813. 69. Pati, S. K.; Das, S., Chemistry – A European Journal 2017, 23, 1078. 70. Mahdi, T.; Stephan, D. W., Angewandte Chemie International Edition 2015, 54, 8511. 71. Mahdi, T. Hydrogenation and Hydroamination Reactions Using Boron-Based Frustrated Lewis Pairs, University of Toronto, 2015. 72. Gyömöre, Á.; Bakos, M.; Földes, T.; Pápai, I.; Domján, A.; Soós, T., ACS Catalysis 2015, 5, 5366. 73. Scott, D. J.; Simmons, T. R.; Lawrence, E. J.; Wildgoose, G. G.; Fuchter, M. J.; Ashley, A. E., ACS Catalysis 2015, 5, 5540.

136

Chapter 2 – Boron-based FLPs

74. Fasano, V.; Radcliffe, J. E.; Ingleson, M. J., ACS Catalysis 2016, 6, 1793. 75. Blackwell, J. M.; Sonmor, E. R.; Scoccitti, T.; Piers, W. E., Organic Letters 2000, 2, 3921. 76. Jensen, J. L.; Gardner, M. P., The Journal of Physical Chemistry 1973, 77, 1557. 77. vom Stein, T.; Peréz, M.; Dobrovetsky, R.; Winkelhaus, D.; Caputo, C. B.; Stephan, D. W., Angewandte Chemie International Edition 2015, 54, 10178.

137

Chapter 3 – Tin-based FLPs

Supplementary information relating to Chapter 3 can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201606639/abstract

Chapter 3.1 – Introduction

Among the most important conclusions drawn from the work described in the previous chapter is that the ability of an FLP catalyst to tolerate hydroxylic species such as alcohols or H2O is heavily dependent upon ensuring that deprotonation of any LA·ROH adduct by the LB is reversible. If not, then this side-reaction will lead to rapid and complete quenching of both the LA and LB components of the FLP, thereby eliminating any possibility of productive catalysis. In Chapter 2 it was demonstrated that this can be achieved using archetypal BAr3 LAs (Ar = electron-poor aryl) by combining them with weak ethereal solvent LBs, which led to the development of the first protocols for FLP-catalysed hydrogenation of aldehydes and ketones, and moisture-tolerant FLP-catalysed hydrogenation.

Despite the undeniable success of this approach, it must be acknowledged that it still suffers from some fundamental limitations. Chief among these is the need to avoid the presence of even moderately strong LBs, which has inevitable implications for the long-term functional group tolerance and substrate scope of these catalytic systems. This problem is attributable to the strength of the binding between ROH species and electrophilic BAr3 LAs, which leads to a dramatic increase in the Brønsted acid strength of the O—H bond.1 Conceptually, it can be seen that this situation might be ameliorated if the strength of the ROH→LA interaction could be weakened; however, this must be achieved whilst maintaining the ability of the LA to engage in H2 activation as part of an FLP. In other words, the acidity of the LA towards oxygen-centred donors must be reduced without this resulting in an equivalent loss of hydride ion affinity.

According the qualitative Pearson model,2 acids and bases can be classified as either ‘hard’ or ‘soft’. For the former, high charge density and low polarizability cause bonding interactions to be dominated by ionic contributions; for the latter, more accessible frontier molecular orbitals and higher polarizability result in bonds with far more covalent character. In order to maximise the strength of the more important of these bonding modes (ionic or covalent) bases will preferentially bind to acids of the same type. For example, addition of a soft cyanide base (NC–) to an aqueous mixture of hard ‘H+’ and soft ‘MeHg+’ will lead to selective formation of the soft/soft adduct MeHgCN, as this results

138

Chapter 3 – Tin-based FLPs in the strongest possible covalent interaction (due to the more energetically and spatially accessible acceptor orbitals on the heavy Hg atom).

Thus, it should be possible to minimise the strength of binding to hard oxygen-centred donors, while retaining a strong interaction with soft hydride ligands, by pursuing the use of softer LAs for FLP chemistry (Figure 3.1). Unfortunately, as was discussed in Chapter 1.4.2, FLP-catalysed hydrogenation reactions have only previously been achieved using LAs based on a very narrow range of chemical elements. Aside from B-based LAs, the only such have examples have exploited Al-, P- or Zr-based acids;3-7 in each case these are based on hard, oxophilic early elements. Instead, it would be preferable to use FLP hydrogenation catalysts incorporating heavier p-block element LAs, whose lower electronegativity and higher polarizability should result in much ‘softer’ character. Unfortunately, such systems have never previously been shown to successfully activate H2, and even their more general FLP chemistry has attracted only minimal interest, with attention to date instead focusing on their lighter congeners.8

Several of the reports that do exist have involved the use of Sn(II) LAs, which were incorporated by Wesemann et al. into intramolecular Sn(II)/P systems. These compounds, which were constructed with a variety of different linkers, were shown to be capable of several typical FLP-type reactions, such as addition to carbon-carbon multiple bonds (Figure 3.2; a).9-11 Zhu et al. subsequently described similar reactivity based on related E(IV)/P systems (E = Ge, Sn; Figure 3.2; b),12 while the

Manners group have described the use of FLPs based on Bu3SnOTf ([26a]OTf; Bu = n-butyl) to mediate

13 the dehydrogenation of Me2NH·BH3 (vide infra). More recently, Gabbai and Tofan have demonstrated FLP reactivity using Sb(V) LAs (which were found to have potential for use as colourimetric sensors for aqueous formaldehyde: Figure 3.2; c);14 and Melen et al. have investigated FLP-type reactivity using PhSeCl as the LA (see Chapter 1.2.1), and highlighted the likelihood that different reactivity will be possible using FLPs based on heavy rather than light element LAs.15 Finally, Stephan et al. have published a pair of reports showing how a heavy p-block element can be used as the Lewis basic centre of an FLP, describing alkyne addition to intramolecular B/Te FLPs (Figure 3.2; d).16-17

139

Chapter 3 – Tin-based FLPs

Figure 3.1 – A conceptual approach for the development of improved ‘ROH-tolerant’ FLP catalysts.

Despite the lack of prior reports, it was reasoned that H2 activation (and thence hydrogenation catalysis) ought to be possible for FLPs incorporating late p-block LAs, provided that they satisfy the same conditions that have been established to be necessary for success using lighter elements; for example, sufficient cumulative LA and LB strength, a suitable steric profile, and the ability to exist without decomposition in the presence of a suitable LB partner (see Chapter 1.4.3 for a more detailed discussion).

140

Chapter 3 – Tin-based FLPs

Figure 3.2 – Some previous examples of FLPs based on heavy p-block elements.9-10, 12

In principle, it seems likely that these requirements could be satisfied using examples from many different families of late p-block LAs. Nevertheless, it was decided to focus initial proof-of-

+ + concept studies on the use of LAs based on stannylium ions: ‘R3Sn ’ ([26] ; note that in standard

+ nomenclature, “stannylium” is used to refer to three-coordinate R3Sn cations, while “stannyl” refers

+ + 18 to ligated [R3Sn·L] or [R3Sn·L2] cations). These compounds were attractive for a number of reasons, the simplest of which relate to their ease of synthesis using very cheap precursors, and a chemically robust core structure (vide infra). Furthermore, although FLP H2 activation using these compounds has not been extensively studied, the subsequent reduction chemistry of the related hydrides (R3SnH, [26]H) is well established, with these reagents well known to be capable of effecting the reduction of numerous organic substrates,19 including in protic media, and via ionic mechanisms that closely resemble the reduction steps of FLP-catalysed hydrogenation (Scheme 3.1; a).20 One particularly relevant example is the hydrostannylation of organic carbonyls, which it is known can be catalysed by LAs including the stannylium salt [26a]OTf (Scheme 3.1; b).21 Note that while the use of organotin reagents can sometimes present concerns relating to toxicity, in practice the risk related to these

141

Chapter 3 – Tin-based FLPs compounds is highly dependent on the nature of the attached organyl framework (with increased molecular weight typically leading to reduced toxicity), meaning that in the long term it could potentially be mitigated by careful LA design.22-23 The mammalian toxicity of these compounds is also typically lower than that towards other organisms.24

Scheme 3.1 – Organic carbonyl reduction by trialkyltin hydrides in a protic medium (a);20 or catalysed by a trialkyltin LA in an aprotic medium (b).21

More specific motivation for the study of ‘[26]+’ LAs was provided by theoretical studies reported by Heidem and Latham, in which DFT calculations suggested that the stannylium cation [26a]+ should possess a hydride ion affinity remarkably similar to that for the isolobal borane 1a, which in turn indicated that they ought to capable of mediating similar FLP H2 activation and hydrogenation

25 chemistry (Figure 3.3; top). By contrast, comparison of the H2O adducts of these two LAs shows a

1 clear difference: while 1a·OH2 is a very powerful Brønsted acid (estimated aqueous pKa < 1), the cation

+ 26 [26a·x(H2O)] is much weaker (aqueous pKa = 6.25), suggesting that this LA (and structurally-related analogues) should be much less prone to deactivation by H2O (or other ROH) in the presence of moderately strong LBs (Figure 3.3; bottom).

Figure 3.3 – A comparison of calculated hydride ion affinities and aqua complex pKa values for representative triarylborane and trialkylstannylium LAs.1, 25-26

142

Chapter 3 – Tin-based FLPs

Nevertheless, it was anticipated that in practice the use of truly ‘naked’ stannylium cations for useful FLP chemistry is likely to be infeasible. Previous experimental investigations have shown that these species are extremely reactive electrophiles; unlike for analogous carbocations, stabilisation through (hyper)conjugation is precluded by the increased length of the Sn—C bond, and by less efficient overlap with the more diffuse vacant p orbital on Sn. Thus, crystallographic analysis has shown that the [26a]+ cation will interact even with incredibly weakly basic anions such as the

– 27 carborane [CB11Me12] , and the only previously-isolated stannylium salts have required exceedingly bulky ligands such as Mes (mesityl; 2,4,6-triphenylmethyl) or Trip (2,4,6-triisopropylphenyl) to prevent

28-30 close approach of possible donors. Instead, it was decided to target R3SnX salts containing slightly coordinating counteranions such as triflate (X = OTf), with the expectation that this would result in

+ improved stability and ease of synthesis without excessively quenching the Lewis acidity of the ‘R3Sn ’ core. It is worth noting that in their hydride ion affinity calculations Heiden and Latham used a polarized continuum model to account for stabilisation of [26a]+ by polar MeCN solvent, and some weak stabilisation of the stannylium is thus implicitly built into their results (in fact, equivalent calculations in the gas phase suggested a hydride ion affinity for ‘naked’ [26a]+ that is ca. 85 kcal/mol higher than for 1a).25

The use in FLP chemistry of butyl-substituted [26a]OTf (which can be considered as a surrogate for the [26a]+ cation) has briefly been studied by Manners et al., who reported that this LA forms an FLP in combination with TMPH, and that this FLP is capable of effectively mediating the

13 dehydrogenation of Me2NH·BH3 (Scheme 3.2; a). Unfortunately, the authors also reported that attempts to use this FLP to activate H2 were unsuccessful (Scheme 3.2; b). This contrasts with facile

31 H2 activation by 1a/TMPH, and can likely be attributed to an overly-strong interaction between the ‘[26a]+’ and ‘TfO–’ moieties, which will reduce the effective strength of the LA considerably below that of the ‘naked’ stannylium.

Scheme 3.2 – Successful dehydrogenation of Me2NH·BH3 (a), but unsuccessful activation of H2 (b), as reported by Manners et al. for a [26a]OTf/TMPH FLP.13

143

Chapter 3 – Tin-based FLPs

It was anticipated that the strength of the Sn—OTf interaction might be weakened by increasing the steric bulk of the attached alkyl groups, with the resulting increase in ‘internal

32 frustration’ raising LA strength to the level necessary to engage in useful H2 activation. Precedent

+ for this strategy can be found in the work of Ghosez et al. who showed that the ‘Me3Si ’ surrogate

33 – Me3SiNTf2 is a stronger LA than Me3SiOTf. This was attributed to the greater steric bulk of the Tf2N

+ anion, which weakens its interaction with the ‘Me3Si ’ moiety and so increases the cationic character of the Si atom. To this end, it was decided to target the more sterically-encumbered isopropyl- substituted analogue iPr3SnOTf ([26b]OTf) for initial study.

Chapter 3.2 – Synthesis and characterisation of iPr3SnOTf

The LA [26b]OTf could readily be prepared via a simple, two-step procedure (Scheme 3.3;

34 developed in cooperation with Mr. Joshua Sapsford). Treatment of SnCl4 with an excess of an isopropyl Grignard reagent (commercially available, or generated from iPrCl and Mg) gave the tetrasubstituted stannane iPr4Sn, which could be isolated and purified via distillation under reduced pressure (1 mbar, 110 °C). Subsequent treatment with a single equivalent of the acid HOTf then led to selective cleavage of a single C—Sn bond, and furnished the target compound as a pure, white solid after removal of solvent in vacuo and washing with pentane. Both steps proceeded in reasonable yield, and could be performed on multi-gram scale.

Scheme 3.3 – Synthesis of [26b]OTf.

[26b]OTf shows moderate-to-high solubility in polar halogenated or donor solvents (including

H2O), but is only sparingly soluble in most other common solvents. NMR spectroscopic analysis shows features consistent with the expected structure (Figure 3.4). For example, the 1H NMR spectrum shows the septet and doublet expected for the isopropyl moieties [2.07 and 1.44 ppm respectively in

3 1 1 CDCl3, integrating in a 6:1 ratio, with J( H- H) = 7.6 Hz]. Both resonances show clear satellites due to coupling to NMR-active (I = ½)35 117Sn and 119Sn nuclei [the 2J(119Sn-1H) coupling was measured as 39 Hz via a 2-dimensional 119Sn/1H HMBC experiment], and similar satellites are apparent in the 13C{1H} NMR spectrum. The presence of the –OTf anion is confirmed by the presence of a 19F resonance at – 76.7 ppm, as well as a 13C{1H} quartet at 119.0 ppm [1J(19F-13C) = 319 Hz]. Finally, the 119Sn{1H}

144

Chapter 3 – Tin-based FLPs

spectrum shows a single, broad signal at 156 ppm ('v½ = 130 Hz; 0.06 M in CDCl3). The downfield resonance is consistent with a highly deshielded (and hence Lewis acidic) stannylium core. Nevertheless, it comes at a significantly lower chemical shift than that previously reported for

[26a][CB11Me12], which contains a weakly-coordinating carborane anion (454 ppm in C6D12; note that even this shift is very significantly upfield of that expected for a true ‘naked’ stannylium),27, 29, 36 and is

37 similar to that for butyl-substituted [26a]OTf (168 ppm in CDCl3), suggesting that [26b]OTf must retain an appreciable Sn—OTf interaction, despite the increased steric bulk of the alkyl side-chains.

1 13 1 Figure 3.4 – H (top) and C{ H} (bottom) NMR spectra for [26b]OTf in CDCl3.

145

Chapter 3 – Tin-based FLPs

The coupling constant 1J(119Sn-13C) has previously been used as a guide to the degree of trigonalisation/pyramidalisation at Sn,37 with larger values taken to indicate increased s-orbital character for the Sn—C bonding interaction and so a more planar ‘R3Sn’ moiety, which would be consistent with increased stannylium character. Surprisingly, in this case, the value for [26b]OTf is lower than that for n-butyl-substituted [26a]OTf (316 Hz vs. 383 Hz) which might appear to suggest reduced stannylium character, contrary to expectations. It should be noted, however, that additional trigonalisation of compounds similar to [26a]OTf has been proposed to arise due to formation of 5- coordinate trigonal bipyramidal structures (vide infra), which complicates this analysis. Increased steric bulk for [26b]OTf should reduce the degree of such hypercoordination in solution, which could account for the smaller coupling constant.

The structure of [26b]OTf was also supported by mass spectrometry (APCI-MS showed peaks at m/z = 327 and 249 consistent with loss of CF3 and OTf, respectively, from the molecular ion) and elemental analysis. Unfortunately, attempts at crystallographic characterisation of the solid-state structure have thus far been unsuccessful. While single crystals seemingly suitable for analysis by X- ray crystallography (XRD) have been obtained on a number of occasions (e.g. via slow diffusion of pentane into a CHCl3 solution), these have proven to be extremely sensitive to fracturing in the diffractometer cryostream (as well as more generally during mechanical manipulation), even after raising the temperature of the N2 flow to as high as RT. Fortunately, the crystal structures of two other triorganotin triflates have been reported previously. Bulky [(Me3Si)2CH]3SnOTf ([26c]OTf) was reported by Westerhausen and Schwarz to adopt a fairly unremarkable 4-coordinate molecular structure,38 as shown in Figure 3.5. In contrast, Beckmann and co-workers have subsequently described a polymeric structure for Ph3SnOTf ([26d]OTf), in which the triflate moieties each bridge between two separate Ph3Sn centres, which feature 5-coordinate Sn atoms in a trigonal bipyramidal geometry (with the triflate oxygens occupying the axial positions).39

These structural differences are attributed to the differing steric bulk of the organic groups, with very large [(Me3Si)2CH] substituents disfavouring higher coordination numbers. Electronic factors may also play a role, with aryl-substituted stannyliums expected to be stronger electrophiles than their alkyl-substituted equivalents (due to weaker donation of electron density to the Sn centre from sp2 carbons than from sp3).25 Since the steric profile of [26b]OTf is expected to more closely mirror that of [26d]OTf than [26c]OTf it seems likely that it would also adopt a polymeric structure in the solid state, which might account for its relatively low solubility in many common solvents. This is also supported by the observation of a similar structure for related [26b]NTf2 (discussed later, in Chapter 3.5).

146

Chapter 3 – Tin-based FLPs

Figure 3.5 – Previously-reported solid-state structures of triorganotin triflates.38-39 Thermal ellipsoids shown at 50% probability. C atoms shown in blue; F in green; O in red; S in yellow; Si in orange; Sn in magenta. H atoms omitted for clarity.

The Lewis acidity of [26b]OTf was probed during separate work within the Ashley group,34 using the established Gutmann-Beckett method (Table 3.1).40-41 The results indicate that it is an appreciably stronger acid than butyl-substituted [26a]OTf (consistent with the expected effect of the bulkier iPr ligand), but remains significantly weaker than the borane 1a.42 Note, however, that this measure of abstract ‘Lewis acidity’ is based on coordination of an electron-rich oxygen-centred LB

147

Chapter 3 – Tin-based FLPs

(Et3PO); given the expectation that Sn-based LAs will be less oxophilic than their B-based counterparts (vide supra), this may mean that there is a poor correlation between these values and other parameters that are more pertinent in the context of FLP H2 activation, such as hydride ion affinities.

Table 3.1 – Gutmann-Beckett Lewis acidity values for some LAs relevant to Chapter 3.34, 42

Relative LA strength LA Acceptor number (normalised to 1a)

64.2 0.82 [26a]OTf

68.0 0.87 [26b]OTf

78.1 1 1a

Chapter 3.3 – H2 activation using iPr3SnOTf-based FLPs

Having prepared the target LA, the ability of [26b]OTf to activate H2 in combination with appropriate bases was examined. Addition of 1,4-diazabicyclo[2.2.2]octane (DABCO) to a DFB solution in a 1:1 ratio led to large upfield shift in the 119Sn{1H} NMR spectrum: to 39 ppm (the resonance remained similarly broad), indicating formation of an adduct (the precise structure of the adduct will be discussed further in Chapter 3.6.2). Nevertheless, 1H NMR spectroscopy showed only a single resonance for the DABCO protons, suggesting that this adduct must undergo rapid exchange, plausibly via transient re-dissociation to the free LA and LB. Gratifyingly, admission of H2 (4 bar, RT) to this solution led to clear changes in its NMR spectra, with a new resonance appearing at —46 ppm in the 119Sn{1H} spectrum alongside a second, broad set of isopropyl signals in the 1H spectrum, accompanied by additional downfield resonances at 10.93 and 5.12 ppm. These changes are consistent with

34 formation of DABCO·HOTf (NH at 10.93 ppm) and iPr3SnH ([26b]H; SnH at 5.12 ppm), through activation of H2 by the DABCO/[26b]OTf Lewis pair (Figure 3.6). It was also possible to observe 117Sn/119Sn satellites [1J(117Sn-1H) = 1405 Hz, 1J(119Sn-1H) = 1471 Hz] that are highly diagnostic for the hydridic SnH hydrogen. Further, conclusive proof for this reactivity was provided when the reaction was repeated using D2 (ca. 2 bar) in place of H2 (Figure 3.7). While similar changes were observed in the aliphatic region of the 1H NMR spectrum, no new downfield resonances were observed; instead, equivalent signals were observed in the 2H NMR spectrum, with satellites again observable for the SnD

148

Chapter 3 – Tin-based FLPs peak [the 1J(117/119Sn-2H)/ 1J(117/119Sn-1H) ratio closely matches the J(2H)/J(1H) gyromagnetic ratio of ca. 0.15].35 Meanwhile, the singlet previously observed in the 119Sn{1H} NMR spectrum was replaced by a 1:1:1 triplet at the same shift, due to coupling between the 119Sn and 2H nuclei (I = 1).35 These results confirm that this reaction is the first example of successful H2 activation using an FLP incorporating an LA based on a heavy p-block element.

1 119 1 Figure 3.6 – H (top) and Sn{ H} (bottom) NMR spectra for addition of H2 (4 bar) to an equimolar mixture of DABCO and [26b]OTf in DFB.

149

Chapter 3 – Tin-based FLPs

2 119 1 Figure 3.7 – H (top) and Sn{ H} (bottom) NMR spectra for addition of D2 (2 bar) to an equimolar mixture of DABCO and [26b]OTf in DFB.

1 Integration of the H NMR spectrum taken under H2 suggests that the H2 activation reaction does not proceed to completion; in fact, the observation of ca. 50 % conversion would appear to suggest that it is close to thermoneutral under these conditions ('G ≈ 0; note that accurate integration of the SnH resonance requires an extended delay: the T1 relaxation constant was measured in situ as

150

Chapter 3 – Tin-based FLPs approximately 15 s). This is also supported by the persistence of two sets of isopropyl resonances, as well as of the signal attributed to [DABCO→26b]OTf in the 119Sn{1H} spectrum.

43 Similar results were obtained using the weaker LBs 2,4,6-collidine (col; pKa = 14.98 in MeCN)

43 44 and 2,6-lutidine (lut; pKa = 14.13 in MeCN) in place of DABCO (pKa = 18.29 in MeCN). Note that the initial 119Sn{1H} resonances for [26b]OTf were nearly unchanged in the presence of these bases (no significant change in linewidth, and only minor upfield shifts: to 142 ppm and 151 ppm, respectively), suggesting that there is no more than a minimal interaction. As expected, reduced basicity leads to less favourable H2 activation, with appreciable formation of [26b]H only observed in these systems at slightly higher pressure (Table 3.2).

Table 3.2 – H2 activation by [26b]OTf and LBs of various strengths in DFB.

LB x pKa (MeCN) Conversion to [26b]H

DABCO 4 18.2944 54 % 2,4,6-collidine (col) 10 14.9843 20 % 2,6-lutidine (lut) 10 14.1343 15 %

These results contrast with the outcomes of similar reactions using 1a, whose H2 activation reactions have been reported to proceed to completion when using similarly-strong LBs.45-46 This would seem to suggest a lower hydride ion affinity for [26b]OTf, which is consistent with prior Lewis acidity measurements (Table 3.1), but contrasts with the prediction of similar hydride ion affinity for 1a and [26a]+ (Figure 3.3).25 Assuming this discrepancy in not the result of limitations in the predictive methodology (which might arise when trying to compare such structurally-different LAs), it could be

+ attributed to partial quenching of the ‘R3Sn ’ moiety by the triflate counterion, or to the effect of the bulkier alkyl groups (which should disfavour 4-coordinate structures relative to 3-coordinate ones). Nevertheless, it is important to acknowledge an additional factor which may complicate this

119 1 explanation for incomplete reactivity. This is indicated by the Sn{ H} spectra for H2 activation involving col and lut: in both of these reactions, the signal for the LA is observed to move steadily upfield during the course of the reaction, as shown in Figure 3.8 (that a comparable shift is not observed when using DABCO is presumably due to stronger Lewis acid/base adduct formation).

151

Chapter 3 – Tin-based FLPs

Previous work in the Ashley group has suggested that this may be due to coordination of the remaining

34 [26b]OTf to triflate anions released upon H2 activation. This quenching of the remaining LA will inhibit further H2 activation, and so disfavour completion of the reaction. In particular, this is consistent with the inability of H2 activation by [26b]OTf/DABCO to proceed beyond ca. 50% conversion. This interpretation was further supported by the failure of attempts to observe HD scrambling using col or lut under equivalent conditions, which suggests that despite not proceeding to full conversion, H2 activation is not reversible.

119 1 Figure 3.8 – Sn{ H} NMR spectra for addition of H2 (10 bar) to an equimolar mixture of lut (a) or col (b) and [26b]OTf in DFB: an upfield shift in the [26b]OTf resonance is attributed to reversible binding

of the triflate released following H2 activation (c).

152

Chapter 3 – Tin-based FLPs

Chapter 3.4 – Catalytic hydrogenation using iPr3SnOTf-based FLPs

Chapter 3.4.1 – Catalytic hydrogenation of imines

Following the successful use of [26b]OTf-based FLPs for H2 activation, investigations proceeded to examine the use of this LA in TM-free hydrogenation catalysis, with initial attention focused on archetypal imine substrates. Gratifyingly, when H2 (10 bar) was added to a solution of imine 2g and [26b]OTf (10 mol%) in DCB and the mixture subsequently heated to 120 °C, effective hydrogenation was observed, with clean formation of the related amine 3g (97 % conversion after 12 h; Table 3.3, entry 1). This is only the third example of successful FLP-type hydrogenation to use a p-block LA catalyst based on an element other than boron,3-4 and the first example based on such a heavy element. The result can be compared with the equivalent reaction catalysed by 1a, which proceeds under similar (though less forcing: complete conversion after 2 h at 80 °C, 1 bar H2, 5 mol%) conditions.47 Interestingly, while other donors typically have a dramatic solubilising effect on [26b]OTf, in this reaction mixture the LA catalyst is only sparingly soluble prior to heating, suggesting that there is no significant initial interaction with the substrate (vide infra).

Table 3.3 – Hydrogenation of imines catalysed by [26b]OTf with or without an auxiliary base.

Conv. / Entry Substrate R R’ R” LB t / h %a

1 2g H H tBu - 12 97 2 2h H Me tBu - 16 85 3 2d H H Ph - 16 4 4 2d H H Ph col 24 >99 5 2e H Me Ph col 32 96 6 2a H H Ts col 80 65 7 2i 4-Br H tBu col 16 96 aMeasured by 1H NMR integration.

Alongside the aldimine 2g, the related ketimine 2h was also hydrogenated under the same conditions, and at a comparable rate (85 % conversion after 16 h; Table 3.3, entry 2). In contrast, the

153

Chapter 3 – Tin-based FLPs same reaction protocol was found to lead to only minimal reduction (4 %) of the N-phenyl analogue 2d, which is attributed to the reduced basicity of this substrate. In the absence of any other base, initial H2 activation must proceed using the substrate as the LB component of an FLP (vide infra), and this step will become prohibitively unfavourable for weakly basic substrates. Consistent with this analysis, addition of a co-catalytic quantity of col to the reaction mixture led to dramatically improved reactivity (full conversion after 24 h; Table 3.3, entry 3), which is attributed to more favourable H2 activation using this auxiliary LB. This effect can be compared to the role of the solvent in mediating

H2 activation in the 1b/THF system described in Chapter 2.2, or to the observation by Stephan et al.

47 that addition of P(Mes)3 can speed the 1a-catalysed hydrogenation of the weakly-basic imine 2b.

In the presence of col, the N-phenyl ketimine 2e could also be hydrogenated, as could the N-tosyl aldimine 2a, although reduction of the latter was appreciably slower (only 65 % conversion after 80 h; Table 3.3, entry 6). This is attributed to the even lower basicity of this substrate, which will

+ + hinder H transfer from [col—H] (generated from H2 activation). Finally, the brominated substrate 2i was observed to undergo very clean hydrogenation, with the final NMR spectra showing no indication of side-reactions such as hydrodebromination (to form 2g/3g), nor formation of Sn-based compounds

48 such as [26b]Br or [26b]2. The absence of such processes suggests that hydrogenation does not occur via a radical mechanism (Scheme 3.4), despite the known ability of R3SnH reagents to act as reductants through radical chain reaction mechanisms,49 which are made possible by low homolytic Sn—H bond strengths (e.g. 78 kcal/mol for [26a]H).50 This conclusion is also supported by the lack of a specific chemical radical initiator (meaning initiation would have to be spontaneous or UV- mediated).

Scheme 3.4 – Example of side-reactions that would be expected for a radical-mediated [26b]OTf- catalysed hydrogenation of bromoaryl imine 2i.

154

Chapter 3 – Tin-based FLPs

Instead, [26b]OTf-catalysed imine hydrogenation is proposed to proceed via a typical FLP

47 mechanism analogous to that for 1a-catalysed reactions. In the absence of other LBs, H2 activation by a substrate/[26b]OTf Lewis pair will proceed to directly generate protonated substrate and the hydride source [26b]H (Scheme 3.5; substrate-mediated H2 activation pathway). Hydride transfer will then furnish the reduced amine product, which can dissociate from the LA catalyst to close the catalytic cycle.

Scheme 3.5 – Proposed mechanisms for [26b]OTf-catalysed hydrogenation of imines.

The feasibility of this mechanism was confirmed by independent stoichiometric reduction experiments using the pre-formed reductant [26b]H.34 In the first of these reactions, the substrate 2g was combined with [26b]H in DCB (Scheme 3.6; a). No reduction of the imine was observed even after prolonged heating to 120 °C (the temperature of the catalytic reaction), confirming that hydride transfer from [26b]H to the unactivated substrate is not a feasible reaction step (the [26b]H is robust under these conditions and does not decompose). By contrast, when the imine was treated with an

155

Chapter 3 – Tin-based FLPs equivalent of HOTf prior to addition of [26b]H, significant immediate reduction was observed even at RT (Scheme 3.6; b), consistent with hydride transfer to the protonated substrate being sufficiently rapid to facilitate the catalytic reaction. Interestingly, while this reaction proceeds very quickly at first, its rate drops off sharply following initial reduction (for example, almost 50% conversion was observed within minutes, but only 65% conversion after 17 h). This is attributed to the increased basicity of the product amine relative to the imine,51-52 which leads to near-complete deprotonation of the remaining substrate and so to slower further reduction. Finally, the reaction of 2g and [26b]H was repeated in the presence of an additional equivalent of the [26b]OTf LA (Scheme 3.6; c). Again, no reduction was observed, even at 120 °C, thereby ruling out the possibility that the substrate might be activated by the LA, rather than ‘H+’.

Scheme 3.6 – Stoichiometric reactivity of [26b]H with imine 2g alone (a); following addition of HOTf (b); and following addition of [26b]OTf (c).

The direct substrate-mediated H2 activation mechanism will become slow for less basic substrates such as 2d, as the reduced cumulative LA/LB strength of the Lewis pair causes H2 activation to become overly disfavoured. However, this can be compensated for through addition of a slightly stronger LB such as col. Due to its increased strength, H2 activation using the auxiliary base is much more rapid, and subsequent H+ transfer to the substrate completes an alternative (and more

156

Chapter 3 – Tin-based FLPs favourable) route to the shared [2—H]OTf + [26b]H intermediate (Scheme 3.5; auxiliary base- mediated H2 activation pathway). For even less basic substrates such as 2a, even this pathway will begin to become infeasible, as the H+ transfer step becomes highly unfavourable.

Interestingly, closer inspection of the hydrogenation of 2g (without any other LB) suggested a degree of autocatalysis during the course of the reaction. For example, while only 16% conversion was observed after heating for 3 h, the same sample showed 60 % conversion after 6 h. This is attributed to a third mechanism by which H2 can be activated, via addition to a Lewis pair made up of the [26b]OTf LA and the amine product, 3g (Scheme 3.5; product-mediated H2 activation). Since the amine is expected to be a stronger base than the imine (calculated pKaH = 17.3, 15.3 respectively in

52 MeCN), H2 activation (and hence catalytic hydrogenation) will be more favourable using this mechanism. As a result, the increase in 3g concentration in the early stages of the reaction leads to an increasing rate of reaction (Figure 3.9; this is essentially equivalent to the effect of adding col to the hydrogenation of 2d, for example). Paradies et al. have observed similar autocatalysis in related borane-catalysed hydrogenation of imines (including 2g), and found that it is particularly significant for catalysts that (like [26b]OTf) are significantly weaker LAs than borane 1a.51-52 This is because the inherently more difficult H2 activation using these LAs makes them significantly more sensitive to any change in LB strength.

Figure 3.9 – Autocatalysis during the [26b]OTf-catalysed hydrogenation of 2g, due to a change in the dominant reaction mechanism.

Chapter 3.4.2 – Catalytic hydrogenation of carbonyl compounds

The successful hydrogenation of imines provided a positive proof-of-principle for the use of [26b]OTf as a LA in FLP-catalysed hydrogenation chemistry. As a result, attention was shifted from these simple substrates and towards the more challenging organic carbonyls that are the primary

157

Chapter 3 – Tin-based FLPs focus of this thesis. Given the successful use of [26b]OTf and col for the hydrogenation of N-tosyl imine 2a, this system was used as the starting point for investigation (following the same logic as in Chapter 2, where the high catalytic activity of borane/solvent systems for this hydrogenation reaction was used to guide the development of boron-based carbonyl hydrogenation catalysis protocols).

Gratifyingly, when 2a was replaced with acetone (20a) under essentially identical conditions, clean hydrogenation to the alcohol 21a was observed (78 % conversion after 96 h; Table 3.4, entry 1). Notably, this was the first example of successful hydrogenation of a carbonyl group to be catalysed by the ‘conventional’ FLP combination of a reasonably strong LB and LA, rather than the more unusual combinations of a strong LA and very weak LB discussed in Chapter 2. This is consistent with the design principles earlier in this chapter (e.g. in Figure 3.1); and in particular with the idea that the [26b]OTf LA should form a less acidic adduct with ROH products, which will be less prone to irreversible deactivation in the presence of strong LBs. The speed with which this [26b]OTf-based system could be successfully applied to the hydrogenation of a carbonyl group can be compared with the rather greater difficulties encountered while developing boron-based FLPs for the same transformation.53-55

Table 3.4 – [26b]OTf-based catalytic hydrogenation of organic carbonyl compounds.

Entry Substrate R R’ LB t / h Conv. / %a

1b 20a Me Me col 96 78 2c 20a Me Me col 32 97 3 c 20j Ph Me col 48 91d

c 4 20n 2,6-Cl2C6H3 H col 32 91 5 c 20p tBu H col 48 79 6 20a Me Me col 16 57 7 20a Me Me lut 16 48 8 20a Me Me DABCO 16 14 9 20a Me Me [26b]OiPre 16 32 aMeasured by 1H NMR integration. bReaction performed at 120 °C and repressurised after 48 h. cRepressurised at 16 h intervals. dBased on consumption of substrate (products are 21j, 22j and 23j in a ca. 74:8:18 molar ratio). eGenerated in situ from 20a and [26b]H.

158

Chapter 3 – Tin-based FLPs

While the hydrogenation of 20a by [26b]OTf and col at 120 °C was observed to proceed with good turnover, the rate of the reaction was relatively low. Fortunately, it was found that the reaction time could be reduced significantly simply by increasing the reaction temperature (for example, near- quantitative conversion could be achieved after 32 h at 180 °C; Table 3.4, entry 2). The [26b]OTf catalyst was found to be remarkably robust, with NMR analysis of the final reaction mixture showing no evidence for decomposition, despite having undergone extended heating to 180 °C. In particular, 1H NMR spectroscopy showed only a single set of resonances for the isopropyl groups of the [26b]+ moiety, while the 119Sn{1H} spectrum showed only a single broad signal, at around 30 ppm (an essentially identical signal was observed following independent preparation of a solution of [26b]OTf, col, and 10 equivalents of the product alcohol, 21a). The very high stability of this catalytic system can be compared with the observation of significant decomposition for borane/solvent systems even at much lower temperature (discussed in detail in Chapter 2.3.5). In fact, there have not been any previous reports of FLP hydrogenation catalysts that are capable of operating at such high reaction temperatures.

As well as 20a, the combination of [26b]OTf and col has been found to be capable of catalysing the hydrogenation of other carbonyl compounds. While the scope of this reaction has not yet been studied in as much detail as for 1a-catalysed hydrogenation in 1,4-dioxane, it has been possible to successfully hydrogenate one example each of an aliphatic and aromatic ketone and aldehyde (Table 3.4, entries 2-5). In the case of the aromatic ketone 20j, hydrogenation was observed to give a mixture of products including styrene (much as when previously using 1a in 1,4-dioxane), but this was not found to prevent good conversion (91 % after 48 h; Scheme 3.7). No evidence for hydrogenation or hydrostannylation of the styrene side-product was observed, presumably because the reduced polarity and basicity of the C=C bond renders it less susceptible than the C=O bond in 20j to initial activation by the LA catalyst (vide infra).

Scheme 3.7 – Hydrogenation of 20j catalysed by [26b]OTf and col, producing a mixture of products.

Further investigation of the model hydrogenation of 20a showed that this reaction is highly sensitive to the choice of LB co-catalyst. When col was replaced with either slightly weaker lut, or slightly stronger DABCO, significantly inferior results were observed under otherwise identical

159

Chapter 3 – Tin-based FLPs

conditions (Table 3.4, entries 6-8). These results are attributed to less favourable H2 activation and less favourable H+ transfer, respectively (vide infra).

While the 1a-catalysed hydrogenation of carbonyl compounds in 1,4-dioxane (discussed in Chapter 2.3) was proposed to proceed via Brønsted acid activation of the substrate, this mechanism is unlikely to be feasible for the same reaction catalysed by [26b]OTf, given the much stronger LBs used (and hence much weaker Brønsted acids generated). Instead, it seems likely that in these reactions the substrate is activated by coordination of the LA. Subsequent H– and H+ transfer steps and product dissociation will then complete the catalytic cycle (Scheme 3.8). This contrasts with the mechanism proposed for imine hydrogenation in the previous section (c.f. Scheme 3.5), but is consistent with previously-reported [26a]OTf-catalysed reduction of carbonyls using [26a]H.21

Scheme 3.8 – Proposed mechanism for [26b]OTf-based catalytic hydrogenation of aldehydes and ketones.

Clear evidence for the ability of [26b]OTf to bind a carbonyl substrate was provided by independent addition of substrate 20a (10 equivalents) to a solution of the LA in DCB. 119Sn{1H} NMR spectroscopy showed a large upfield shift for the [26b]OTf resonance (to 92 ppm; c.f. 156 ppm for

[26b]OTf only in CDCl3), consistent with a significant Sn←O binding interaction. The proposed mechanism also clearly accounts for the observed effect of LB strength on reaction performance. 119Sn{1H} NMR spectra showed no evidence for accumulation of the hydride [26b]H during the hydrogenation of any of the carbonyl substrates; instead, resonances were typically observed around 30 ppm: consistent with formation of the intermediate [26b]-alkoxide as the main resting state (vide infra). This suggests that the rate of turnover is primarily limited not by the rate of hydride transfer, but by the ability of subsequent proton transfer to release the alcohol product and transiently generate ‘free’ LA that can engage in H2 activation. Clearly, if the strength of the auxiliary LB used is

160

Chapter 3 – Tin-based FLPs

too low then such H2 activation will be highly disfavoured, and so will limit the rate of turnover. Conversely, if the LB is too strong, then initial H+ transfer from [LB—H]OTf to the Sn alkoxide will become infeasible, and the reaction rate will again be limited. This latter case can be considered as another example of an FLP being deactivated through O—H bond activation, which acts to inhibit catalytic turnover.

Again, further support for the proposed hydrogenation mechanism was obtained by investigating stoichiometric reductions using pre-formed [26b]H, in this case of the model ketone 20a. Simple addition of [26b]H of 20a in DCB led to no reaction (Scheme 3.9; a), even after heating to 120 °C (at which temperature catalytic hydrogenation of 20a had been observed to proceed previously), confirming that activation of the carbonyl is required prior to hydride transfer. If an equivalent of [26b]OTf was also added immediate, significant reduction was observed at RT, clearly indicating the ability of the LA catalyst to provide this necessary degree of activation (Scheme 3.9; b). The rate of further reaction appeared to be limited by the low solubility of [26b]OTf under these conditions, but complete conversion was still achieved after standing overnight. When [26b]OTf was replaced with the Brønsted acid [col—H]OTf only slow release of H2 was observed (Scheme 3.9; c). Slow reduction was then observed when this mixture was subsequently heated to 120 °C; however, this seems likely to be due to substrate activation by the [26b]OTf released following loss of H2, rather than activation by remaining [col—H]OTf.

An important factor to note from the proposed mechanism is that, in order for the proton transfer step to be accessible, the intermediate [26b]-bound alkoxide must possess a Brønsted basicity similar to the LB employed (this is consistent with prior discussion: for example, the aqueous pKa of

+ 26, 43 [26a] ·xH2O is 6.25, versus 7.45 and 6.70 for protonated col and lut, respectively). This suggests that an alternative reaction pathway might be possible, in which this intermediate acts directly as the

LB involved in H2 activation (Scheme 3.10). Indeed, when the auxiliary LB was replaced in the initial reaction mixture with an additional equivalent of 20a and [26b]H, catalytic hydrogenation of 20a was still observed (Table 3.4, entry 9). Under these conditions, the additional 20a and [26b]H react to form

[26b]OiPr in situ prior even to admission of H2 (Scheme 3.11); the alkoxide must then act as the LB for

H2 activation, confirming the feasibility of this alternative mechanism. This is highly reminiscent of the

56 1a/[NMe4][1aH] catalytic system developed by Stephan and Mahdi, which was discussed in Chapter 2.5.2. In support of the idea that it a stronger base than [26b]OiPr, addition of col to a DCB solution of [26b·HOiPr]OTf ([26b←21a]OTf, generated in situ by addition of [26b]OTf to iPrOH, 21a, in a 1:10 ratio) was observed to lead to an upfield-shifted 119Sn{1H} NMR resonance (from 65 ppm to 28 ppm), and downfield-shifted col 1H NMR resonances (from 6.56, 2.41 and 2.09 ppm; to 6.66, 2.42 and 2.11 ppm), as would be expected for deprotonation of the [26b←21a]OTf adduct.

161

Chapter 3 – Tin-based FLPs

Scheme 3.9 – Stoichiometric reactivity of [26b]H with ketone 20a alone (a); following addition of [26b]OTf (b); and following addition of [colH]OTf (c).

Scheme 3.10 – An alternative mechanism for [26b]OTf-based catalytic hydrogenation of aldehydes and ketones.

162

Chapter 3 – Tin-based FLPs

Scheme 3.11 – Hydrogenation of 20a catalysed by [26b]OTf and [26b]OiPr, with the latter generated in situ from 20a and [26b]H.

Nevertheless, this [26b]OiPr-based reaction proceeds more slowly than when the optimal LB col is used. This is attributed to increased basicity of col, which aids in H2 activation, and clearly mirrors the similar beneficial effect that was observed upon addition of the same LB to the hydrogenation of weakly-basic imines (described in Chapter 3.4.1).

Chapter 3.4.3 – Catalytic hydrogenation of other substrates

Having established that [26b]OTf-based FLPs can be successfully applied to the catalytic hydrogenation of more than one type of functional group, preliminary investigations were carried out into their broader substrate scope. Early investigations in the field of FLP hydrogenation showed that successful catalysts for imine hydrogenation can often be applied to the reduction of related enamines and N-heterocycles.57-59 Thus, it was found that the substrate 27 could very rapidly (within 4 h) be completely hydrogenated to the amine 28, using [26b]OTf and col under identical conditions to those previously used for imines 2 (Scheme 3.12; a). Meanwhile, hydrogenation of acridine, 4, was observed to proceed effectively using [26b]OTf as the only catalyst (Scheme 3.12; b), without the need for an

35, 43 additional LB (aqueous pKaH = 5.58; c.f. 6.70 for lut).

The hydrogenation of butyl acrylate 12a was also investigated, in hopes that hydrogenation using [26b]OTf-based FLPs might also applicable to C=C reduction (in addition to C=N and C=O). Gratifyingly, clean reduction to the aliphatic ester 13a was observed using catalytic [26b]OTf and col without the need for any optimisation of the catalytic protocol (83 % conversion after 80 h; Scheme 3.13). Note that no evidence for reduction of the ester functional group was observed (indeed, all attempts at ester hydrogenation using [26b]OTf to date have unfortunately been unsuccessful).

163

Chapter 3 – Tin-based FLPs

Scheme 3.12 – Hydrogenation of enamine 27, catalysed by [26b]OTf and col (a); and acridine, 4, catalysed by [26b]OTf only (b).

Scheme 3.13 – Hydrogenation of an activated C=C bond catalysed by [26b]OTf and col.

Chapter 3.5 – Moisture tolerance in iPr3SnOTf-based hydrogenation catalysis

Chapter 3.5.1 – Air-stability of [26b]OTf

Having established their utility for the catalytic hydrogenation of aldehydes and ketones to alcohols, it was decided to investigate whether the tolerance of [26b]OTf-based FLP hydrogenation catalysts to simple alcohols also translates into appreciable moisture tolerance. To begin, the effect of exposure to air on [26b]OTf was examined. After standing on the open bench for several weeks, samples of solid [26b]OTf did not appear to undergo any visible physical changes, still appearing as pristine, pure-white powders. Nevertheless, attempts to re-dissolve these samples for NMR spectroscopic analysis showed that changes must have taken place, with the solids found to no longer be appreciably soluble in polar halogenated solvents such as DFB and CHCl3. Fortunately, when later attempts were made to use these samples to make up catalytic reaction mixtures (vide infra), it was found that in some cases the other reaction components had a solubilising effect (particularly donors

164

Chapter 3 – Tin-based FLPs such as carbonyl substrates, presumably as a result of Brønsted and/or Lewis acid/base interactions). NMR comparison of these initial mixtures with their equivalents prepared using ‘anhydrous’ [26b]OTf showed clear differences, with new downfield 1H resonances observed (e.g. at ca. 5.3 ppm for a mixture of col, ‘open bench’ [26b]OTf, and 10 equivalents of 20a), as well as more upfield 119Sn{1H} shifts (to ca. 20 ppm for the same mixture). These changes are attributed to absorption of moisture from the atmosphere, which coordinates to the acidic Sn centres (Scheme 3.14; a). Note that aside from this, no evidence for decomposition of the LA was observed, with only a single set of 1H and 119Sn{1H} NMR resonances observed for the ‘[26b]+’ moiety in each case.

Scheme 3.14 – Absorption of atmospheric moisture by [26b]OTf (a); as indicated by 1H NMR spectroscopy for a DCB solution also containing col and 10 eq. acetone, 20a (b).

Integration of the new 1H signals relative to the ‘[26b]’ isopropyl resonances indicated that two equivalents of H2O are absorbed per Sn centre (Scheme 3.14; b), suggesting formation of a 5-

+ coordinate trigonal bipyramidal [26b·2H2O] cation, similar to those in previously reported

60-64 [R3Sn·2H2O][X] salts. In fact, slow evaporation under air of an aqueous solution of [26b]OTf yielded single crystals suitable for XRD analysis, which revealed just such a structure (in the solid state, at least). The crystallographic analysis showed that the triflate counterion is well separated from the

Lewis acidic Sn centre, but forms an extensive hydrogen-bonding network with the H2O ligands, with every O—H bond and every triflate oxygen atom engaged in at least one such interaction. This can be compared with H2O·B(C6F5)3 (H2O·1a), which also tends to form O—H · · · X interactions in its solid-

65-67 + state structures. Meanwhile, the [26b·2H2O] cation adopts a near-perfect trigonal bipyramidal geometry with the H2O ligands in the axial positions (Figure 3.10). The three equatorial Sn—C bond lengths are identical to within error (average 2.151 Å), although there is an appreciable difference in

165

Chapter 3 – Tin-based FLPs the two Sn—O distances [2.298(2) Å versus 2.322(3) Å; presumably due to crystal packing effects, or asymmetry in the hydrogen-bonding network]. The degree of distortion from an idealised trigonal bipyramidal structure can be quantified using the parameter W, which is defined as shown in Figure

68 3.11. The value of 0.94 obtained for [26b·2H2O]OTf indicates a close-to-perfect trigonal bipyramidal geometry (idealised W = 1, vs. W = 0 for square-based-pyramidal).

Figure 3.10 – Solid-state crystal structure of [26b·2H2O]OTf (top), with an expanded view highlighting inter-ionic hydrogen bonding interactions (bottom). For clarity, H atoms not involved in hydrogen bonding are omitted from the latter. Thermal ellipsoids are shown at 50% probability. C atoms are shown in blue; H in white; F in green; O in red; S in yellow; Sn in magenta.

166

Chapter 3 – Tin-based FLPs

Figure 3.11 – Definition of the geometric parameter W for 5-coordinate complexes.

The structure of [26b·2H2O]OTf can be compared with those of other [R3Sn·2H2O][X] salts whose solid-state structures have been reported previously; namely, butyl-substituted

61-62 [26a·2H2O][C5(CO2Me)5], and methyl-substituted [Me3Sn·2H2O]X ([26e·2H2O]X; X = NTf2, NMs2; Ms

63-64 = MeSO2) congeners. These compounds all adopt very similar structures, with trigonal-bipyramidal

+ geometries at Sn (W > 0.9), axial H2O ligands, and well-separated counteranions. The [26e·2H2O] salts both also show numerous hydrogen-bonding interactions, comparable to those in [26b·2H2O]OTf. Surprisingly, despite the increased steric bulk of the isopropyl groups, the Sn—C distances in

[26b·2H2O]OTf are only slightly longer than those in [26e·2H2O]X, and not measurably different from those in [26a·2H2O][C5(CO2Me)5]; while Sn—O distances vary across the series by less than their variation within each individual structure (Table 3.5). This stands in contrast with the expectation that

– bulkier alkyl groups should lead to weaker binding of donors such as H2O or OTf (vide supra), which may be attributable to the low steric profile of the aqua ligands, as well as to relatively long Sn—O and Sn—C bonds, which will help to minimise the steric clash between neighbouring groups.

+ 61-64 Table 3.5 – Key crystallographic bond lengths (in Å) and angles (in °) for some [26·2H2O] salts.

[26a·2H2O]

[26b·2H2O]OTf [26e·2H2O]NTf2 [26e·2H2O]NMs2 [C5(CO2Me)5]

Sn—O 2.298(2) 2.295(4) 2.306(3) 2.254(2) bond length / Å 2.322(3) 2.326(5) 2.335(3) 2.327(3) 2.155(4) 2.190(14) 2.104(4) 2.110(4) Sn—C 2.149(4) 2.155(12) 2.115(4) 2.111(5) bond length / Å 2.150(4) 2.189(11) 2.120(4) 2.108(5) O—Sn—O 174.49(10) 178.4(2) 175.94(11) 177.0(1) bond angle / ° 117.30(15) 119.2(6) 117.8(2) 121.4(2) C—Sn—C 118.23(16) 120.8(4) 120.1(2) 121.1(2) bond angle / ° 124.33(16) 120.0(5) 122.1(2) 117.3(2)

167

Chapter 3 – Tin-based FLPs

Chapter 3.5.2 – Moisture tolerance in [26b]OTf-based carbonyl hydrogenation

Studies into the effect of moisture on [26b]OTf-based hydrogenation catalysis began with investigations into the reduction of carbonyl compounds, since these had already shown tolerance of the O—H bonds in their alcohol products. To this end, the hydrogenation of model ketone 20a was repeated, with H2O (two equivalents) added to the otherwise-anhydrous reaction mixture (Scheme 3.15). Remarkably, not only was effective catalytic hydrogenation still observed, but the reaction was found to proceed with only a marginal decrease in reaction rate (90% conversion versus 97%, after 32 h). This represents a dramatic improvement over even the previously-developed moisture-tolerant systems based on boron that were discussed in Chapter 2,69-71 where the presence of moisture led to a requirement for significantly more forcing reaction conditions if reduction was to be observed within a similar timeframe. When further comparing the two systems, it is also worth emphasising once again the clear tolerance of a moderately strong base in the Sn-based protocol; as well as the relative ease with which its H2O-tolerance could be established.

Scheme 3.15 – Tolerance of added H2O in the [26b]OTf-based catalytic hydrogenation of model ketone 20a.

Increasing the amount of H2O present to 5 equivalents did lead to a clear reduction in rate for the same reaction, but comparable turnover could still be achieved with increased reaction time

(80 h). Interestingly, a further increase to 10 equivalents of H2O seemed to have only a more minor further impact on reaction, which is attributed to limited solubility of H2O in the DCB reaction solvent.

In fact, for both of the reactions containing 5 or 10 equivalents of H2O, clear phase separation of an aqueous phase was visible for the initial mixture at RT. This contrasts with the ethereal solvents used in Chapter 2 which are typically fully miscible with H2O, and so demonstrates a clear advantage to being able to achieve O—H-tolerant hydrogenation catalysis in non-donor solvents.

168

Chapter 3 – Tin-based FLPs

Based on this established moisture-tolerance, the hydrogenation of C=O bonds using reaction mixtures prepared under open bench conditions was investigated. Satisfyingly, when a sample of [26b]OTf that had been allowed to stand under air for over a week was used in combination with undried, ‘non-anhydrous’ commercial-grade DCB, H2, and substrate, essentially no drop in performance relative to the anhydrous reaction was observed (Scheme 3.16; a). Catalytic hydrogenation of aldehyde 20n could also be observed following equivalent preparation on the open bench (Scheme 3.16; b).

Scheme 3.16 – [26b]OTf-based catalytic carbonyl hydrogenation reactions prepared on the open bench.

Chapter 3.5.3 – Moisture tolerance in [26b]OTf-based imine hydrogenation

Given the extremely positive results obtained using carbonyl substrates, research was advanced to establish the moisture tolerance of the related [26b]OTf-based hydrogenation of imines. To begin with, the hydrogenation of model imine 2g was investigated, with its hydrogenation repeated using ‘wet’ [26b]OTf that had previously been left to stand under air. Unlike the previous experiments using ketone 20a, hydrogenation of 2g was found to be considerably slower than under anhydrous conditions. Nevertheless, good turnover could still be observed following an increase in reaction temperature to 180 °C (Scheme 3.17; a). This is the first example of moisture tolerance in any FLP- type hydrogenation of an imine (or, indeed, any such basic substrate), and proceeds despite the formation of a rather powerfully basic 2° amine. Formation of this LB, which is expected to be stronger

43, 52 than the col that was used during 20a hydrogenation (calculated pKaH = 17.3 vs. 14.98 in MeCN), is nevertheless likely to lead to extensive deprotonation of hydrated [26b]OTf, and so to account for the reduced rate of the ‘wet’ reaction (Scheme 3.17; b). Note that unlike the [26b]OR intermediates

169

Chapter 3 – Tin-based FLPs formed during anhydrous carbonyl hydrogenation (Scheme 3.8), the stannol [26b]OH formed in this non-anhydrous reaction is expected to exist in equilibrium with the condensed stannoxane [26b]2O (Scheme 3.17; c), which is likely to cause a further reduction in the rate of catalysis. Indeed, 119Sn{1H} NMR spectroscopy of intermediate reaction mixtures shows two separate resonances (at 41 ppm and 31 ppm) in the chemical shift range expected for these compounds.34 Interestingly, while the initial 1H NMR spectrum for the reaction indicated appreciable hydrolysis of the imine starting material (>10%, most obviously indicated by appearance of a resonance at 9.84 ppm attributed to benzaldehyde), the final spectrum showed rather clean formation of the product 3g (< 5% BnOH from PhCHO reduction), suggesting that the imine is reduced preferentially under these reaction conditions.

Scheme 3.17 – Moisture-tolerant [26b]OTf-catalysed hydrogenation of imine 2g (a); whose rate is proposed to be limited due to deprotonation of the hydrated LA by the basic product 3g (b), and subsequent condensation of the stannol [26b]OH (c).

Similar results were observed for the hydrogenation of the slightly less basic N-phenyl imines

2d and 2e catalysed by [26b]OTf and col, with more forcing reaction conditions required if H2O was added to the initial reaction mixture, or if it was prepared on the open bench (Scheme 3.18). Nevertheless, the rate of these reactions at 180 °C exceeded those for 2g at the same temperature. This contrasts with the relative rates of the equivalent anhydrous reactions at 120 °C (where 2g was reduced more rapidly; Table 3.3), and is attributed to the reduced basicity of substrates 2d,e, products

3d,e, and auxiliary LB col, which results in slightly less dramatic inhibition by H2O.

170

Chapter 3 – Tin-based FLPs

Scheme 3.18 – Moisture-tolerant [26b]OTf-based hydrogenation of imines 2d,e.

While for 2d complications due to hydrolysis were not observed to be any more serious than for 2g, hydrogenation of 2e was found to be significantly less clean in the presence of H2O (Scheme 3.18; c). Although complete conversion was observed at a comparable rate to 2d, the final reaction mixture was observed to contain a mixture of styrene (23j; ca. 6 %) and alcohol 21j (ca. 7 %) in addition to the expected 3e, presumably due to reduction of PhCOMe (20j). The reduction could be made to

171

Chapter 3 – Tin-based FLPs proceed somewhat more selectively through addition of dry 5 Å molecular sieves at the start of the reaction, in order to remove H2 and so disfavour hydrolysis (Scheme 3.18; d). This also allowed the reaction to be run at a comparable rate under a reduced pressure of H2 (4 bar). Nevertheless, formation of some 21j and 23j was still observed, possibly because the high temperature limits the effectiveness of the desiccant.72 Addition of molecular sieves was also found to allow for lower pressures in the hydrogenation of 2d (Scheme 3.18; e).

It is interesting to consider why the hydrogenation of imines 2d and 2e should be so significantly slowed by the presence of moisture. After all, the strongest base present in these reactions is expected to be the col co-catalyst: exactly as in the hydrogenation of ketone 20a. It might therefore have been expected that these reactions would be similarly susceptible to O—H bond inhibition, and so similarly sensitive to the presence of moisture. A possible partial explanation is that, unlike for imine hydrogenation, the anhydrous hydrogenation of 20a already involves the presence of a hydroxylic species; namely, the alcohol product 21a. This is expected to have a similar inhibitory effect to H2O, and so inhibition by O—H groups is already ‘built in’ to the anhydrous reaction, and accounted for in its observed rate. Imine hydrogenation has no equivalent inherent inhibition (which likely accounts for the lower reaction temperatures that can be used), and so the presence of moisture has a much more dramatic effect. In support of this explanation, it was observed that the presence of the alcohol 21a leads to an appreciable decrease in the rate of hydrogenation of the imine 2d (Scheme 3.19).

Scheme 3.19 – [26b]OTf-based hydrogenation of imine 2d, inhibited by alcohol 21a.

Successful moisture tolerance in the hydrogenation of imines suggested that it might be possible to extend the application of this [26b]OTf-based system beyond simple reduction of pre- formed substrates, to reductive amination. This is one of the most important families of C—N bond- forming reaction, in which an aldehyde or ketone and amine react to form an imine, followed by in situ reduction.73 Despite the significance of these reactions, precious-metal-free protocols for reductive amination using H2 as the reductant are rather under-developed; several recently-reported TM-based protocols are summarised in Figure 3.12, all of which require rather forcing reaction temperatures and pressures.74-76

172

Chapter 3 – Tin-based FLPs

Figure 3.12 – Recent examples of TM-catalysed hydrogenative amination.74-76

Addition of benzaldehyde and aniline (instead of 2d) to the catalytic reaction mixture led to immediate imine formation as indicated by both 1H NMR spectroscopy, and visible formation of a separate aqueous phase on top of the reaction solution. Remarkably, complete reduction was observed after heating to 180 °C under 10 bar H2 (Scheme 3.20). Only a slight increase in reaction time was required relative to the open bench hydrogenation of pre-formed 2d (48 h versus 32 h), with no significant difference in the composition of the final product mixture. This is despite a significant increase in the amount of H2O present, and is presumably due to the observed phase separation. While the scope of this reaction has not yet been investigated, it provides an important proof-of- principle not just for FLP-catalysed hydrogenative amination, but for the more general possibility of incorporating FLP-catalysed reduction processes into more elaborate multi-step organic transformations. It is also worth contrasting the success of this reaction with the failure by Ingleson et al. to observe analogous chemistry using 1a as a catalyst (albeit under less forcing conditions), as was discussed in Chapter 2.5.2.71

Scheme 3.20 – Hydrogenative amination catalysed by [26b]OTf and col, despite formation of H2O as a stoichiometric by-product.

173

Chapter 3 – Tin-based FLPs

Chapter 3.6 – The effect of anion variation in iPr3SnX

When designing [26b]OTf as the target R3SnX LA for initial study it was argued that the strength of the LA, and hence its suitability for FLP hydrogenation catalysis, could be improved by reducing the strength of the interaction between the counteranion and the Lewis acidic Sn centre (Chapter 3.1). Despite this, 119Sn{1H} NMR spectroscopy and Lewis acidity measurements on [26b]OTf both suggest that an appreciable degree of coordination remains, despite the relatively bulky isopropyl ligands (Chapter 3.2). As a result, it was decided to perform preliminary investigations into the FLP-hydrogenation chemistry of some related [26b]X species for which any Sn—X bonding was expected to be weaker, in the hope that further improvements in catalytic activity might be observed.

Chapter 3.6.1 – Hydrogenation chemistry using iPr3SnNTf2

– The first analogue targeted was iPr3SnNTf2 ([26b]NTf2). While the triflimide anion (NTf2 ) is a

77 stronger Brønsted base than triflate (pKaH = 7.8 vs. 4.2 in acetic acid), it was anticipated that it would bind the [26b]+ cation more weakly due to its increased steric bulk. This has precedent in the chemistry of trialkylsilylium LAs as was discussed earlier in this chapter, 33 and is further supported by previous studies in the Ashley group, which showed higher Lewis acidity for [26a]NTf2 than [26a]OTf, as assessed by the Gutmann-Beckett method.34

[26b]NTf2 could be prepared from iPr4Sn in an analogous manner to [26b]OTf, via reaction with HNTf2 in CHCl3 (Scheme 3.21; this reaction was developed in collaboration with Mr Joshua Sapsford).34 A slightly higher reaction temperature was needed in this case, which is consistent with the lower strength of the Brønsted acid.

Scheme 3.21 – Synthesis of [26b]NTf2.

1 13 1 While H and C{ H} NMR spectra for [26b]NTf2 are broadly similar to those for [26b]OTf, the very low solubility of this compound in even highly polar non-donor solvents means it has not been possible to resolve either a clear 119Sn{1H} signal or a 1J(119Sn-13C) coupling constant in the absence of external donors (although a single, sharp peak was observed at –253 ppm for a solution in CD3CN). The same problem has also hampered efforts within the Ashley group to quantify Lewis acidity through

34 the standard Gutmann-Beckett method. Thus, while [26b]NTf2 is presumed to be a stronger LA than [26b]OTf, this has yet to be confirmed by direct experimental evidence.

174

Chapter 3 – Tin-based FLPs

Unlike for [26b]OTf, it has been possible to perform XRD analysis on single crystals of [26b]NTf2

(grown via slow diffusion of pentane into a dilute CHCl3 solution), and so to establish its solid-state structure (Figure 3.13). The compound was found to adopt a polymeric arrangement, with each triflimide anion bridging between two Sn centres, similar to the structure of [26d]OTf (Figure 3.13; bottom: c.f. Figure 3.5); this likely accounts for the compound’s low solubility in most solvents, in the absence of additional donors. The anion binds to Sn through its oxygen atoms (rather than N, possibly for steric reasons), which occupy axial positions around the trigonal-bipyramidal metal

68 centre (W = 0.86). The Sn—C bond lengths are identical to those in [26b·2H2O]OTf (to within error;

Figure 3.13 – Solid-state structure of [26b]NTf2. Thermal ellipsoids shown at 30% probability. H atoms omitted for clarity. C atoms shown in blue; F in green; N in purple; O in red; S in yellow; Sn in grey.34

Table 3.6), but the Sn—O bond length is significantly longer, consistent with the triflimide anion being a poorer ligand than H2O (note that the Sn—O bonds are all crystallographically equivalent).

175

Chapter 3 – Tin-based FLPs

Table 3.6 – A summary of key crystallographic bond lengths (in Å) and angles (in °) for [26b]NTf2,

with values for [26b·2H2O]OTf also shown for comparison.

[26b]NTf2 [26b·2H2O]OTf

Sn—O 2.298(2) 2.346(3) bond length / Å 2.322(3) 2.155(4) Sn—C 2.137(5) 2.149(4) bond length / Å 2.115(15) 2.150(4) O—Sn—O 178.8(2) 174.49(10) bond angle / °

176

Chapter 3 – Tin-based FLPs

127.2(4) 117.30(15) C—Sn—C 117.7(11) 118.23(16) bond angle / ° 113.9(10) 124.33(16)

Having isolated [26b]NTf2, its use in model FLP-catalysed hydrogenation reactions (in place of [26b]OTf) was investigated, beginning with the hydrogenation of ketone 20a. This switch was observed to lead to a remarkable improvement in catalytic performance, with near-complete conversion achieved after a significantly decreased reaction time, even when the reaction was performed at 60 °C below the standard temperature (18 h for [26b]NTf2 at 120 °C, versus 32 h for [26b]OTf at 180 °C; Scheme 3.22).

Scheme 3.22 – Hydrogenation of ketone 20a at a significantly improved rate using [26b]NTf2.

The greatly improved catalytic activity observed for this reaction clearly demonstrates the

+ potential for further improvement of ‘R3Sn ’-based FLP hydrogenation catalysts through iterative rational modification. Nevertheless, when the same catalyst was applied to the hydrogenation of model imine 2g, a significantly decreased rate of turnover was observed relative to [26b]OTf (39 % conversion after 32 h, versus 97 % conversion after 12 h; Scheme 3.23). This may simply be due to the low solubility of the LA under these reaction conditions (a significant quantity of undissolved solid was observed to persist throughout the reaction; as with [26b]OTf this suggests that there is no significant interaction of the imine substrate with the LA), but it could also reflect mechanistic differences between the two model reactions (discussed in Chapter 3.4).

Scheme 3.23 – Inferior hydrogenation of imine 2g catalysed by [26b]NTf2.

177

Chapter 3 – Tin-based FLPs

F Chapter 3.6.2 – Hydrogenation chemistry using ‘[iPr3Sn][Al(OR )4]’

Given the greatly improved rate of 20a hydrogenation using [26b]NTf2 rather than [26b]OTf, which is presumed to be due to a weaker Sn—O interaction, it was decided to subsequently target a salt containing [26b]+ paired with a much more inert ‘non-coordinating’ anion. While many such anions have been described in the literature (e.g. carboranes, borates, oxymetallates) it was decided

– F 78 to specifically investigate the alkoxyaluminate [Al(OC{CF3}3)4] (Al{OR }4). This anion benefits from relatively simple synthetic access to its base-free alkali metal salts (which are key starting reagents),79 and has previously been shown to be stable in the presence of many other powerful cationic

+ 80 + 81 + 82 electrophiles, including [Ag(P4)2] , [PX4] (X = I, Br), and [CI3] . It was anticipated that the target

F 79 stannylium salt might be accessible through metathesis of [26b]OTf with K[Al(OR )4], in a reaction that would be driven by formation and precipitation of KOTf. Indeed, addition of [26b]OTf to a

F solution of K[Al(OR )4] in DFB (chosen to ensure full solubility of both reagents) led to immediate formation of a fine precipitate, consistent with this expected reactivity. Unfortunately, 19F NMR spectroscopy indicated that this reaction does not proceed to completion, with a significant resonance for the triflate anion observed to remain at – 77.7 ppm (integration suggested ca. 50 % of the triflate remained in solution). This is attributed to the high Lewis acidity expected from a truly ‘naked’ trialkylstannylium moiety (vide supra); it is likely that any ‘[26b]+’ transiently formed is stabilised via coordination to an additional equivalent of [26b]OTf, which renders loss of the resulting bringing triflate anion highly unfavourable (Scheme 3.24; a). 119Sn{1H} NMR spectroscopic analysis of the reaction mixture was able to detect a single very downfield resonance at 325 ppm, consistent with

+ the very high level of stannylium character that would be expected from this bridged [(26b)2·OTf] structure.

178

Chapter 3 – Tin-based FLPs

F Scheme 3.24 – Salt metathesis between [26b]OTf and K[Al(OR )4] in the absence (a) and presence (b) of an additional LB.

Fortunately, the above analysis suggested that it ought to be possible to drive the metathesis reaction to completion through addition of a suitable alternative donor, to stabilise the [26b]+ moiety. Indeed, when the reaction was repeated in the presence of an equivalent of col or DABCO (these bases were chosen because of their established use in H2 activation chemistry when combined with [26b]OTf), complete loss of triflate anion from solution was observed by 19F NMR spectroscopy. In situ 1H NMR spectroscopy suggested that these reactions proceed cleanly, with only a single significant set of resonances for both the LB and ‘[26b]+’; meanwhile, the 119Sn{1H} spectrum showed a single downfield signal at 115 ppm when using DABCO (no clear resonance was observed when using col).

F Collectively, these data are consistent with proposed formation of LB-stabilised [26b←LB][Al(OR )4] (Scheme 3.24; b). Note that 19F and 27Al NMR spectra showed no evidence for decomposition of the anion [G(19F) = –75.5 ppm, G(27Al) = 35.0 ppm], including during subsequent reactivity studies.

Since the equivalent [26b]OTf-based reaction also contained col as the LB, the use of

F [26b·col][Al(OR )4] (generated in situ) in the hydrogenation of model ketone 20a was investigated. Surprisingly, despite the use of an anion expected to be much more weakly coordinating, the rate of this reaction was not observed to be significantly greater than for that using [26b]OTf, and was much slower than when using [26b]NTf2 (44 % conversion after 48 h at 120 °C, versus 95 % after 18 h for

[26b]NTf2; Scheme 3.25, c.f. Scheme 3.22). Clearly, the optimal catalytic system is not simply the one

179

Chapter 3 – Tin-based FLPs containing the most inert anion, and this result suggests that there may actually be a beneficial effect to the presence of an anion that retains some weak donor reactivity.

F Scheme 3.25 – Hydrogenation of ketone 20a catalysed by [26b·col][Al(OR )4].

F In an attempt to understand the poor outcome of the [26b·col][Al(OR )4]-based reaction, the

F reactivity of the salts [26b·LB][Al(OR )4] (LB = col, DABCO) towards H2 was investigated in the absence of any substrate. Surprisingly, when DFB solutions of these compounds were monitored by 119Sn{1H}

NMR spectroscopy, no resonances attributable to [26b]H were observed following addition of H2 (4 bar for DABCO, 10 bar for col), even after several days at RT (Scheme 3.26). This clearly contrasts with the equivalent reactions using [26b]OTf and LB Lewis pairs, where rapid H2 activation was observed, and formation of [26b]H was clearly indicated by the 119Sn{1H} spectra (see Figure 3.6 and Table 3.2, for example).

F Scheme 3.26 – Unsuccessful H2 activation by [26b·LB][Al(OR )4] in DFB (LB = col, DABCO).

+ Clearly the anion must play a crucial ability in facilitating H2 activation by LB/[26b] ; however, it seems unlikely that this is due to thermodynamic differences. While interaction with a more

+ coordinating anion could potentially stabilise the [LB—H] cation formed following H2 activation (for

180

Chapter 3 – Tin-based FLPs example via hydrogen bonding), it seems likely that this would be less significant than stabilisation of

+ [26b] prior to H2 activation; consequently, a more weakly coordinating anion is actually expected to make H2 activation more favourable. As a result, it is anticipated that the failure to observe H2

F activation using [26b·LB][Al(OR )4] is probably due to kinetic factors. A plausible explanation is that, in the absence of a coordinating anion, separation of the [26b]+←LB adduct is highly unfavourable, as this necessitates formation of a highly reactive ‘naked’ [26b]+ cation. Failure to dissociate the adduct

+ into a reactive [26b] /LB FLP therefore prohibits FLP-type H2 activation (Scheme 3.27; a). By contrast, in the presence of a slightly more coordinating anion (such as triflate or triflimide), interaction with [26b]+ provides sufficient stabilisation for such dissociation to become energetically feasible, but not so much that the LA is rendered too weak for subsequent H2 activation (Scheme 3.27; b).

In this context, it is interesting to compare the 119Sn{1H} NMR spectra for solutions of

F [26b·DABCO][Al(OR )4], or [26b]OTf and DABCO, in DFB. While the former shows a relatively downfield signal at 115 ppm, the resonance for the latter is shifted dramatically upfield, to 39 ppm. Since the only relevant difference between these two systems is the presence or absence of the triflate anion, this discrepancy suggests that a significant Sn—O interaction is retained in the [26b·DABCO]OTf adduct (Scheme 3.27; d). Alternatively, it is possible that [26b·DABCO]OTf could be in rapid equilibrium with the ligand-scrambled salt [26b·DABCO][26b(OTf)2]and free DABCO, with the observed chemical shift representing a weighted average of the various 119Sn environments. By

F contrast, addition of a second equivalent of DABCO to [26b·DABCO][Al(OR )4] has no observable effect

119 1 + on the Sn{ H} resonance, suggesting that formation of a 5-coordinate [26b·(DABCO)2] cation is not feasible, possibly for steric reasons.

+ It seems likely that the proposed stabilisation of [26b] by coordinating anions en route to H2 activation could also be performed by other donors, including neutral species such as carbonyls and alcohols; this likely accounts for the observation of catalytic hydrogenation of 20a by

F [26b·DABCO][Al(OR )4], which obviously requires H2 activation to occur as one of the elementary

F reaction steps. Indeed, when an additional equivalent was LB was added, the salts [26b·LB][Al(OR )4]

119 1 (LB = col, DABCO) were both observed by Sn{ H} NMR spectroscopy to activate H2 in DFB (with clear formation of a [26b]H resonance at –46 ppm) after standing at RT overnight (Scheme 3.28; a).

+ Presumably, in these reactions the [26b·LB] cation acts as the LA to activate H2, in combination with the additional LB (Scheme 3.28; b).

181

Chapter 3 – Tin-based FLPs

Scheme 3.27 – A qualitative model to explain the differing ability of LB/[26b]+ Lewis pairs (LB = col,

DABCO)to activate H2, depending on the counteranion employed (a-c); and coordination of triflate to [26b←DABCO]+, indicated by 119Sn{1H} NMR spectroscopy (d).

The increased difficulty of activating H2 likely accounts for the reduced rate of hydrogenation

F of 20a using [26b·col][Al(OR )4] relative to [26b]NTf2 and col. Nevertheless, the possibility cannot be ignored that anion binding to other intermediate structures might also lead to significant changes in the rates of other elementary steps in the catalytic cycle (for example, binding of triflate to [26b]OTf was invoked earlier in this chapter to account for incomplete H2 activation by [26b]OTf/LB Lewis pairs; Figure 3.8). Indeed, the mechanistic proposals advanced so far in this chapter have largely ignored

182

Chapter 3 – Tin-based FLPs the possible role of hypervalent 5-coordinate Sn compounds, which could plausibly act as significantly- quenched off-cycle intermediates, for example.

F Scheme 3.28 – Successful H2 activation by [26b·LB][Al(OR )4] and an additional equivalent of LB (LB = col, DABCO) in DFB (a); and the proposed mechanism for these reactions (b).

Chapter 3.7 – Conclusions

Chapter 3.7.1 – Summary and key results

This chapter has described the development of FLP-catalysed hydrogenation reactions incorporating trialkylstannylium-based LAs, and their application to H2 activation (the first examples of such FLP reactivity using a heavy element LA); catalytic hydrogenation of C=N, C=O and C=C bonds (the first example of an FLP hydrogenation catalyst incorporating a boron-free p-block LA that can be successfully applied to the reduction of multiple different functional groups); and moisture-tolerant

FLP catalysis (including the first examples of imine hydrogenation in the presence of H2O, and of FLP- catalysed hydrogenative amination). These systems were found to be particularly effective for the last of these, with ‘open-bench’ hydrogenation of acetone found to proceed without a reduction in rate relative to the same reaction performed under rigorously anhydrous conditions.

Much as for the borane/solvent systems described in Chapter 2, the ability of [26]X/LB FLPs both to catalyse the hydrogenation of organic carbonyls, and to tolerate the presence of moisture, is attributed to reversibility in the deprotonation of any adducts formed between the LA and ROH (R = alkyl, H). Unlike those systems, however, this reversibility is not dependent on the absence of

183

Chapter 3 – Tin-based FLPs moderately strong LBs. Instead, relatively weak Sn—O interactions lead to only mildly acidic ROH→Sn adducts, which are able to tolerate many of the relatively basic N-centred substrates, products, and LBs commonly encountered in FLP chemistry.

Chapter 3.7.2 – Directions for future work

While the [26b]OTf-based catalytic hydrogenation reactions described in this chapter typically require slightly more forcing conditions than their boron-based counterparts, it should be emphasised that studies to date have focused almost exclusively on this specific LA. There remains enormous scope for modification and improvement of trialkylstannylium LAs for FLPs hydrogenation catalysis, and this should be a key focus for future investigations. Work to date has already demonstrated that major differences in reactivity can be observed upon variation of the anion in [26b]X, potentially leading to significantly increased catalytic activity. Importantly, this anion dependence has been found to more complex than the simple correlation with LA strength that had initially been expected.

Preliminary investigations have also indicated that variation of the organyl groups attached to Sn can have a similarly dramatic effect on catalytic performance (note that these studies are currently being continued by other researchers within the Ashley and Fuchter groups, and will not be discussed in detail in this thesis). For example, attempts have been made to use Np3SnOTf ([26f]OTf; Np = neopentyl; this compound can be prepared in an analogous manner to [26b]OTf) in place of [26b]OTf for the hydrogenation of ketone 20a, but have thus far failed to yield even stoichiometric reduction (Scheme 3.29; a). Alternatively, attempts have been made to use commercially-available ‘[26a]’ moieties in FLP hydrogenation catalysis. Remarkably, many of these attempts have led to successful catalytic hydrogenation of substrates including 20a; however, close inspection has suggested that these reactions do not proceed via a simple FLP mechanism (analogous to that for [26b]OTf), with decomposition of the [26a]+ moiety instead leading to an unknown catalytically-active mixture (Scheme 3.29; b).

Alongside investigation of a more diverse range of intermolecular R3SnX/LB FLPs, improved reactivity may be achievable using intramolecular systems (Figure 3.14; left: see Chapter 1.4.3 for a more extensive discussion of intermolecular versus intramolecular FLPs). In the longer term, it is also hoped that the demonstration of successful FLP-catalysed hydrogenation using Sn(IV) LAs will prompt interest in the study of other heavy p-block LAs for similar chemistry. For example, Gabbai et al. have described a range of Sb(V)-based LAs, including examples with similar strength to 1a, that are moisture tolerant, or that can engage in FLP-type reactivity (Figure 3.14; right);14, 83-84 however, investigations into their use for H2 activation or catalytic hydrogenation have yet to be reported.

184

Chapter 3 – Tin-based FLPs

Scheme 3.29 – To date, attempts at [26f]OTf-based catalytic ketone hydrogenation have been unsuccessful (a); while attempts using [26a]+-based systems have succeeded, but are believed to proceed via decomposition to form as-yet-unidentified catalytic intermediates (b).

Figure 3.14 – Some possible future targets for late p-block FLP hydrogenation catalysts.14, 83

Alongside the investigation of alternative LAs, it would be valuable to pursue additional investigations into the scope and mechanism of [26b]X-based hydrogenation reactions. While preliminary studies have provided some insight into the latter, many aspects still remain ambiguous, particularly relating to the precise coordination environment at Sn for key intermediates and transition states, and the precise role played by the counteranion (as discussed in Chapter 3.6); a greater understanding of these aspects could aid in the development of improved catalytic protocols.

185

Chapter 3 – Tin-based FLPs

Chapter 3.8 – References for Chapter 3

1. Bergquist, C.; Bridgewater, B. M.; Harlan, C. J.; Norton, J. R.; Friesner, R. A.; Parkin, G., Journal of the American Chemical Society 2000, 122, 10581. 2. Pearson, R. G., Journal of Chemical Education 1968, 45, 581. 3. Hatnean, J. A.; Thomson, J. W.; Chase, P. A.; Stephan, D. W., Chemical Communications 2014, 50, 301. 4. vom Stein, T.; Peréz, M.; Dobrovetsky, R.; Winkelhaus, D.; Caputo, C. B.; Stephan, D. W., Angewandte Chemie International Edition 2015, 54, 10178. 5. Xu, X.; Kehr, G.; Daniliuc, C. G.; Erker, G., Journal of the American Chemical Society 2013, 135, 6465. 6. Xu, X.; Kehr, G.; Daniliuc, C. G.; Erker, G., Journal of the American Chemical Society 2015, 137, 4550. 7. Flynn, S. R.; Metters, O. J.; Manners, I.; Wass, D. F., Organometallics 2016, 35, 847. 8. A., W. S.; W., S. D., Bulletin of the Chemical Society of Japan 2015, 88, 1003. 9. Freitag, S.; Henning, J.; Schubert, H.; Wesemann, L., Angewandte Chemie International Edition 2013, 52, 5640. 10. Freitag, S.; Krebs, K. M.; Henning, J.; Hirdler, J.; Schubert, H.; Wesemann, L., Organometallics 2013, 32, 6785. 11. Krebs, K. M.; Maudrich, J. J.; Wesemann, L., Dalton Transactions 2016, 45, 8081. 12. Yu, Y.; Li, J.; Liu, W.; Ye, Q.; Zhu, H., Dalton Transactions 2016, 45, 6259. 13. Whittell, G. R.; Balmond, E. I.; Robertson, A. P. M.; Patra, S. K.; Haddow, M. F.; Manners, I., European Journal of Inorganic Chemistry 2010, 3967. 14. Tofan, D.; Gabbai, F. P., Chemical Science 2016, 7, 6768. 15. Wilkins, L. C.; Günther, B. A. R.; Walther, M.; Lawson, J. R.; Wirth, T.; Melen, R. L., Angewandte Chemie International Edition 2016, 55, 11292. 16. Tsao, F. A.; Stephan, D. W., Dalton Transactions 2015, 44, 71. 17. Tsao, F. A.; Cao, L.; Grimme, S.; Stephan, D. W., Journal of the American Chemical Society 2015, 137, 13264. 18. In A Guide to IUPAC Nomerclature of Organic Compounds (Recommendations 1993). Blackwell Scientific Publications, 1993. 19. Neumann, W. P., Synthesis 1987, 665. 20. Kamiura, K.; Wada, M.,Tetrahedron Letters 1999, 40, 9059. 21. Yang, T. X.; Four, P.; Guibé, F.; Balavoince, G., New. J. Chem. 1984, 8, 611. 22. Gadja, T.; Jancsó, A., Met. Ions Life Sci. 2010, 7, 111.

186

Chapter 3 – Tin-based FLPs

23. In ITRI Publication #538. International Research Institute, 1977. 24. Davies, A., In Tin Chemistry: Fundamentals, Frontiers, and Applications, John Wiley and Sons, 2008. 25. Heiden, Z. M.; Lathem, A. P., Organometallics 2015, 34, 1818. 26. Arnold, C. G.; Weidenhaupt, A.; David, M. M.; Müller, S. R.; Haderlein, S. B.; Schwarzenbach, R. P., Environmental Science & Technology 1997, 31, 2596. 27. Zharov, I.; King, B. T.; Havlas, Z.; Pardi, A.; Michl, J., Journal of the American Chemical Society 2000, 122, 10253. 28. Schäfer, A.; Saak, W.; Haase, D.; Müller, T., Journal of the American Chemical Society 2011, 133, 14562. 29. Lambert, J. B.; Zhao, Y.; Wu, H.; Tse, W. C.; Kuhlmann, B., Journal of the American Chemical Society 1999, 121, 5001. 30. Lambert, J. B.; Lin, L.; Keinan, S.; Müller, T., Journal of the American Chemical Society 2003, 125, 6022. 31. Sumerin, V.; Schulz, F.; Nieger, M.; Leskelä, M.; Repo, T.; Rieger, B., Angewandte Chemie International Edition 2008, 47, 6001. 32. Herrington, T. J.; Ward, B. J.; Doyle, L. R.; McDermott, J.; White, A. J. P.; Hunt, P. A.; Ashley, A. E., Chemical Communications 2014, 50, 12753. 33. Mathieu, B.; Ghosez, L., Tetrahedron 2002, 58, 8219.

34. Sapsford, J., Investigating The Direct Hydrogenation of CO2 Using Bulky Stannyl Cations, Imperial College London, 2016. 35. In Handbook of Chemistry and Physics. 94 ed.; CRC, 2013. 36. Arshadi, M.; Johnels, D.; Edlund, U., Journal of the Chemical Society, Chemical Communications 1996, 1279. 37. Farah, D.; Swami, K.; Kuivila, H. G., Journal of Organometallic Chemistry 1992, 429, 311. 38. Westerhausen, M.; Schwarz, W., Main Group Metal Chemistry, 1997, 20, 351. 39. Beckmann, J.; Lork, E.; Mallow, O., Main Group Metal Chemistry, 2012, 35, 183. 40. Beckett, M. A.; Strickland, G. C.; Holland, J. R.; Sukumar Varma, K., Polymer 1996, 37, 4629. 41. Welch, G. C.; Cabrera, L.; Chase, P. A.; Hollink, E.; Masuda, J. D.; Wei, P.; Stephan, D. W., Dalton Transactions 2007, 3407. 42. Ashley, A. E.; Herrington, T. J.; Wildgoose, G. G.; Zaher, H.; Thompson, A. L.; Rees, N. H.; Krämer, T.; O’Hare, D., Journal of the American Chemical Society 2011, 133, 14727. 43. Kaljurand, I.; Kütt, A.; Sooväli, L.; Rodima, T.; Mäemets, V.; Leito, I.; Koppel, I. A., The Journal of Organic Chemistry 2005, 70, 1019.

187

Chapter 3 – Tin-based FLPs

44. Coetzee, J. F.; Padmanabhan, G. R., Journal of the American Chemical Society 1965, 87, 5005. 45. Geier, S. J.; Stephan, D. W., Journal of the American Chemical Society 2009, 131, 3476. 46. Rokob, T. A.; Hamza, A.; Pápai, I., Journal of the American Chemical Society 2009, 131, 10701. 47. Chase, P. A.; Jurca, T.; Stephan, D. W. Chemical Communications 2008, 1701. 48. Kitching, W.; Olszowy, H. A.; Drew, G. M., Organometallics 1982, 1, 1244. 49. RajanBabu, T. V.; Bulman Page, P. C.; Buckley, B. R., In Encyclopedia of Reagents for Organic Synthesis, John Wiley & Sons, Ltd, 2001. 50. Laarhoven, L. J. J.; Mulder, P.; Wayner, D. D. M., Accounts of Chemical Research 1999, 32, 342. 51. Tussing, S.; Greb, L.; Tamke, S.; Schirmer, B.; Muhle-Goll, C.; Luy, B.; Paradies, J., Chemistry – A European Journal 2015, 21, 8056. 52. Tussing, S.; Kaupmees, K.; Paradies, J., Chemistry – A European Journal 2016, 22, 7422. 53. Chase, P. A.; Welch, G. C.; Jurca, T.; Stephan, D. W., Angewandte Chemie International Edition 2007, 46, 8050. 54. Mahdi, T.; Stephan, D. W., Journal of the American Chemical Society 2014, 136, 15809. 55. Scott, D. J.; Fuchter, M. J.; Ashley, A. E., Journal of the American Chemical Society 2014, 136, 15813. 56. Mahdi, T, Hydrogenation and Hydroamination Reactions Using Boron-Based Frustrated Lewis Pairs. University of Toronto, 2015. 57. Spies, P.; Schwendemann, S.; Lange, S.; Kehr, G.; Fröhlich, R.; Erker, G., Angewandte Chemie International Edition 2008, 47, 7543. 58. Geier, S. J.; Chase, P. A.; Stephan, D. W., Chemical Communications 2010, 46, 4884. 59. Stephan, D. W.; Greenberg, S.; Graham, T. W.; Chase, P.; Hastie, J. J.; Geier, S. J.; Farrell, J. M.; Brown, C. C.; Heiden, Z. M.; Welch, G. C.; Ullrich, M., Inorganic Chemistry 2011, 50, 12338. 60. Wada, M.; Okawara, R., Journal of Organometallic Chemistry 1965, 4, 487. 61. Davies, A. G.; Goddard, J. P.; Hursthouse, M. B.; Walker, N. P. C., Journal of the Chemical Society, Chemical Communications 1983, 597. 62. Davies, A. G.; Goddard, J. P.; Hursthouse, M. B.; Walker, N. P. C., Journal of the Chemical Society, Dalton Transactions 1986, 1873. 63. Blaschette, A.; Schomburg, D.; Wieland, E., Zeitschrift für anorganische und allgemeine Chemie 1989, 571, 75. 64. Vij, A.; Wilson, W. W.; Vij, V.; Corley, R. C.; Tham, F. S.; Gerken, M.; Haiges, R.; Schneider, S.; Schroer, T.; Wagner, R. I., Inorganic Chemistry 2004, 43, 3189. 65. H. Doerrer, L.; L. H. Green, M., Journal of the Chemical Society, Dalton Transactions 1999, 4325.

188

Chapter 3 – Tin-based FLPs

66. Janiak, C.; Braun, L.; Scharmann, T. G.; Girgsdies, F., Acta Crystallographica Section C 1998, 54, 1722. 67. Danapoulos, A. A.; Galsworthy, J. R.; Green, M. L. H.; Cafferkey, S.; Doerrer, L. H.; Hursthouse, M. B., Chemical Communications 1998, 2529.

68. Addison, A. W.; Rao, T. N.; Reedijk, J.; van Rijn, J.; Verschoor, G. C., Journal of the Chemical Society, Dalton Transactions 1984, 1349. 69. Scott, D. J.; Simmons, T. R.; Lawrence, E. J.; Wildgoose, G. G.; Fuchter, M. J.; Ashley, A. E., ACS Catalysis 2015, 5, 5540 70. Gyömöre, Á.; Bakos, M.; Földes, T.; Pápai, I.; Domján, A.; Soós, T., ACS Catalysis 2015, 5, 5366. 71. Fasano, V.; Radcliffe, J. E.; Ingleson, M. J., ACS Catalysis 2016, 6, 1793. 72. Breck, D. W.; Eversole, W. G.; Milton, R. M.; Reed, T. B.; Thomas, T. L., Journal of the American Chemical Society 1956, 78, 5963. 73. Ohkuma, T.; Noyori, R., In Comprehensive Asymmetric Catalysis, Suppl. 1, Springer, 2004. 74. Werkmeister, S.; Junge, K.; Beller, M., Green Chemistry 2012, 14, 2371. 75. Fleischer, S.; Zhou, S.; Junge, K.; Beller, M., Chemistry – An Asian Journal 2011, 6, 2240. 76. Bhor, M. D.; Bhanushali, M. J.; Nandurkar, N. S.; Bhanage, B. M., Tetrahedron Letters 2008, 49, 965. 77. Foropoulos, J.; DesMarteau, D. D., Inorganic Chemistry 1984, 23, 3720. 78. Krossing, I.; Raabe, I., Angewandte Chemie International Edition 2004, 43, 2066. 79. Jaroń, T.; Orłowski, P. A.; Wegner, W.; Fijałkowski, K. J.; Leszczyński, P. J.; Grochala, W., Angewandte Chemie International Edition 2015, 54, 1236. 80. Krossing, I., Journal of the American Chemical Society 2001, 123, 4603. 81. Gonsior, M.; Krossing, I.; Müller, L.; Raabe, I.; Jansen, M.; van Wüllen, L., Chemistry – A European Journal 2002, 8, 4475. 82. Krossing, I.; Bihlmeier, A.; Raabe, I.; Trapp, N., Angewandte Chemie International Edition 2003, 42, 1531. 83. Pan, B.; Gabbaï, F. P., Journal of the American Chemical Society 2014, 136, 9564. 84. Hirai, M.; Cho, J.; Gabbaï, F. P., Chemistry – A European Journal 2016, 22, 6537.

189

Chapter 4 – Experimental details

Supplementary information relating to this thesis, including NMR spectra for catalytic and other reactions, can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201405531/abstract http://pubs.acs.org/doi/abs/10.1021/ja5088979 http://pubs.acs.org/doi/abs/10.1021/acscatal.5b01417 http://onlinelibrary.wiley.com/doi/10.1002/anie.201606639/abstract

Chapter 4.1 – General experimental details

Unless stated otherwise, all reactions and other procedures were performed as outlined in this section.

All reactions were performed under an N2 atmosphere, with all manipulations carried out either in an MBraun Labmaster DP glovebox or by using standard Schlenk line techniques. All glassware was dried by heating to 170 °C overnight before use. All solvents were degassed and dried before use: THF was distilled under N2 from Na / fluorenone and stored over 4 Å molecular sieves; 1,4- dioxane was distilled under N2 from Na / benzophenone and stored over 4 Å molecular sieves; pentane, hexane and toluene were dried using an Innovative Technology Pure Solv™ SPS-400 and stored over K; CHCl3, CH2Cl2 and DMF were dried using an Innovative Technology Pure Solv™ SPS-400 and stored over 3 Å molecular sieves; DFB was dried by refluxing over CaH2, distilled, and stored over 4 Å molecular sieves; DCB was purchased anhydrous from Sigma-Aldrich and further dried and stored over 5 Å molecular sieves; 2-MeTHF was purchased anhydrous from Sigma-Aldrich and further dried and stored over 4 Å molecular sieves; d8-THF, CDCl3 and CD2Cl2 were freeze-pump-thaw degassed and dried and stored over 4 Å molecular sieves; C7D8 was freeze-pump-thaw degassed and dried and stored over K. Molecular sieves were dried by heating under vacuum overnight (ca. 200 °C, 10–1 mbar).

1-3 4 5 6 7 8 Boranes 1a-e; borane-water adduct 1a·OH2; borohydride salt [Bu4N][1a·H]; imines 2c, 2e, 2h,

9 10 11 F 12 and 2i; hydrocarbon substrate 14; stannane [26b]H; and metathesis reagent K[Al(OR )4] were all prepared following previously-reported procedures. Isopropanol and diisopropyl ether were degassed and dried over 4 Å molecular sieves. Acetone was degassed, dried over B2O3 and distilled.

Mg turnings were heated to 170 °C overnight before use. H2 was purchased from BOC (research grade)

190

Chapter 4 – Experimental details and dried by passage through a Matheson Tri-Gas WeldassureTM Purifier drying column, except when used at pressures above 10 bar. D2 (99.8% D) was purchased from Cambridge Isotope Laboratories and dried by standing over 3 Å molecular sieves. HD (96 mol% HD, 98% D) was purchased from Sigma

Aldrich and dried by storing over 3 Å molecular sieves. iPrCl, SnCl4, HOTf and HNTf2 were purchased from major suppliers and used as provided. All other compounds (for example, any substrates not listed above) were purchased from major suppliers: solids were dried under vacuum, while liquids were degassed and dried over 4Å molecular sieves.

Unless stated otherwise, all pressures quoted in this thesis are those of the closed system at RT. In practice, pressures will typically be slightly higher upon heating to the relevant reaction temperature.

Elemental analysis was performed by Stephen Boyer of London Metropolitan University. Crystallographic data were acquired and processed by Mr Andrew Crawford (see Appendix A). NMR spectra were recorded on Bruker AV-400, AV-500 and DRX-400 spectrometers. 1H and 2H spectra were referenced internally to residual solvent signals, while 27Al, 11B, 19F and 119Sn spectra were referenced externally to AlCl3, BF3·OEt2, CFCl3 and SnMe4 respectively. Chemical shifts are stated in ppm (s = singlet, d = doublet, q = quartet, sp = septet, m = multiplet, br = broad).

Conversions were calculated by 1H NMR integration; either by relative integration of product and starting material resonances (in cases where no other species were observed); or by integration relative to a capillary insert containing PPh3 or 1,3,5-trimethoxybenzene in C6D6, or to SiMe4 added as an internal standard. In cases where the final reaction mixture was not fully homogeneous at RT, spectra were also acquired of homogeneous solutions at elevated temperature. In order to minimise any errors, integrations were performed on the most intense product/substrate resonances wherever possible, and only on signals well separated from other peaks. Typically, the intensity of a particular product resonance was compared to the intensity of the resonance for the same protons in the starting material. T1 relaxation measurements were used to ensure accurate linear correlation between integrals and relative concentrations. For illustrative purposes, representative examples of initial and final NMR spectra for catalytic reactions are included later in this chapter. Additional spectra have been omitted in the interests of space, but can easily be found online by following the links listed at the start of this chapter (which broadly correspond to the material covered in Chapter 2.2, Chapter 2.3, Chapter 2.4, and Chapter 3, respectively).

191

Chapter 4 – Experimental details

Chapter 4.2 – Experimental details relating to Chapter 2.2

Supplementary information relating to this section, including NMR spectra for catalytic and other reactions, can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201405531/abstract

Chapter 4.2.1 – Synthesis of imine 2f (PhCH=NDipp)

Benzaldehyde (1.0 mL, 9.6 mmol) was added to a mixture of 2,6-diisopropylaniline (DippNH2,

2.0 mL, 10.6 mmol) and 4 Å molecular sieve beads (5.0 g) in CH2Cl2 (20 mL), and the mixture stirred at RT for 24 h. Purification was performed on the open bench. The reaction mixture was filtered, the filtrate washed with additional CH2Cl2 (20 mL), and the combined organic phase evaporated to dryness under reduced pressure, to yield an oily amber liquid. The crude product was purified by flash column chromatography (SiO2, 1:1; CH2Cl2/pentane as eluent), and recystallised from hexane, to yield the target as a yellow crystalline solid (0.95 g, 37 %). NMR data were consistent with previous reports.13

Chapter 4.2.2 – Typical procedure for hydrogenation of imines, 2 (0.1 mmol scale)

Inside a glovebox borane 1 and imine 2 (0.1 mmol) were dissolved in d8-THF or C7D8 (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing 1,3,5-trimethoxybenzene in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Table 2.1 and re-analysed. Reaction yield was determined by integration of 1H resonances for the product amine, 3, relative to the capillary insert, as described in Chapter 4.1.

192

Chapter 4 – Experimental details

1 Figure 4.1 – H NMR spectra for the 1b-catalysed hydrogenation of 2a to 3a in d8-THF (* = 1,3,5- trimethoxybenzene in capillary insert; ** = residual proteo solvent)

Chapter 4.2.3 – Glovebox-free procedure for hydrogenation of 2a (0.1 mmol scale)

B(C6Cl5)(C6F5)2 (1b) and N-benzylidene-4-toluenesulfonamide (2a) were stored under air in sealed screw-cap vials prior to use.

Under air, 1b (29.6 mg, 0.05 mmol) and 2a (260 mg, 1 mmol) were placed in an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing 1,3,5- trimethoxybenzene in C6D6. Dry THF (0.4 mL) was added via syringe and the NMR tube rapidly sealed [care must be taken to avoid admission of excess moisture, which slows the reaction via formation of

2 the H2O·B(C6Cl5)(C6F5)2 adduct]. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath to 60 °C for 3 hours and re-analysed. Complete conversion was determined based upon integration of 1H resonances for the product amine, 3a, relative to the capillary insert, as described in Chapter 4.1.

193

Chapter 4 – Experimental details

Chapter 4.2.4 – Typical procedure for hydrogenation of imines, 2 (1 mmol scale)

Inside a glovebox B(C6Cl5)(C6F5)2 (1b, 29.6 mg, 0.05 mmol) and imine 2 (1 mmol) were dissolved in THF (4 mL) and transferred into a Rotaflo ampoule also containing a magnetic stirrer bar. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT). The reaction vessel was heated as indicated in Table 2.1.

Subsequent work-up was performed in air. The reaction mixture was transferred directly onto a SiO2 column and eluted with a 2:1 mixture of pentane/ethyl acetate. Spectroscopically pure amine 3 was isolated following solvent removal under reduced pressure.

N-benzyl-4-toluenesulfonamide, 3a14

1 Isolated as a white powder (259 mg, 99%). H NMR (400 MHz, CDCl3) G : 7.77 (d, J = 8.3 Hz, 2H), 7.37- 7.12 (m, 7H), 4.60 (t, J = 6.1 Hz, 1H), 4.13 (d, J = 6.1 Hz, 2H), 2.45 (s, 3H).

N-benzyl-2,6-diisopropylaniline, 3f15

1 Isolated as a colourless oil (263 mg, 98%). H NMR (400 MHz, CDCl3) G : 7.53-7.27 (m, 5H), 7.17-7.08 (m, 3H), 4.05 (s, 2H), 3.32 (septet, J = 6.9 Hz, 2H), 3.16 (s, 1H), 1.25 (d, J = 6.9 Hz, 12H).

Chapter 4.2.5 – Procedure for hydrogenation of basic N-heterocycles 4 and 6

Inside a glovebox B(C6Cl5)(C6F5)2 (1b) and the substrate (4 or 6, 0.1 mmol) were dissolved in d8-THF (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing 1,3,5-trimethoxybenzene in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath to 60 °C for the period indicated in Scheme 2.11 (22 h or 17 h, respectively) and re-analysed. Reaction yields (95 % and 55 % for reduced products 5 and 7, respectively) were determined by integration of 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.2.6 – Procedure for hydrogenation of C=C bonds in substrates 12a and 14

Inside a glovebox B(C6Cl5)(C6F5)2 (1b) and the substrate (12a or 14, 0.1 mmol) were dissolved in d8-THF (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing 1,3,5-trimethoxybenzene in C6D6. H2 was admitted via a

194

Chapter 4 – Experimental details freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Scheme 2.13 and re-analysed. Reaction yields (94 % and 80 % for reduced products 13a and 15, respectively) were determined by integration of 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.2.7 – Procedure for hydrogenation of pyrroles 16

Inside a glovebox B(C6Cl5)(C6F5)2 (1b, 29.6 mg, 0.05 mmol) and pyrrole 16 (0.05 mmol) were dissolved in d8-THF (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing 1,3,5-trimethoxybenzene in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Scheme 2.14 and re-analysed. Reaction yields (95 % and 67 % for pyrrolidinium salts [17aH][H1b] and [17bH][H1b], respectively) were determined by integration of 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.2.8 – Procedure for hydrogenation of furans 18

Inside a glovebox B(C6Cl5)(C6F5)2 (1b, 14.8 mg, 0.025 mmol) and furan 18 (0.1 mmol) were dissolved in d8-THF (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing 1,3,5-trimethoxybenzene in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Scheme 2.15 and re-analysed. Reaction yields (81 %, 75 % and 64 % for substituted tetrahydrofurans 19a, 19b and 19c, respectively) were determined by integration of 1H resonances relative to the capillary insert, as described in Chapter 4.1.

195

Chapter 4 – Experimental details

Chapter 4.3 – Experimental details relating to Chapter 2.3

Supplementary information relating to this section, including NMR spectra for catalytic and other reactions, can be found online at: http://pubs.acs.org/doi/abs/10.1021/ja5088979

Chapter 4.3.1 – Typical procedure for hydrogenation of carbonyl compounds in THF

Inside a glovebox borane 1 and substrate 20 (0.1 mmol) were dissolved in d8-THF (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Chapter 2.3.1, without active mixing, and re-analysed. Reaction outcome was determined by integration of 1H resonances for substrate 20 and reduced products 21 and 22 relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.2 – Typical procedure for hydrogenation of acetone, 20a, in 1,4-dioxane (4 bar)

Inside a glovebox borane 1 and 20a (14.7 PL, 0.2 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Chapter 2.3.1, without active mixing, and re-analysed. Reaction yields were determined by integration of 1H resonances for the product alcohol 21a relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.3 – HD scrambling by B(C6F5)3 (1a) in 1,4-dioxane

Inside a glovebox B(C6F5)3 (1a, 15.4 mg, 0.03 mmol) was dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. HD (1 bar) was admitted through use of a Toepler pump (see Appendix B) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR

196

Chapter 4 – Experimental details spectroscopy. The reaction was left at room temperature without active mixing and periodically re- analysed.

Chapter 4.3.4 – Typical procedure for hydrogenation of aldehydes and ketones 20 in 1,4-dioxane (4 bar)

Inside a glovebox borane 1a and 20a (0.1 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Table 2.2, without active mixing, and re-analysed. Reaction yields were determined by integration of 1H resonances for the product alcohols 21 relative to the capillary insert, as described in Chapter 4.1.

Figure 4.2 – 1H NMR spectra for the 1a-catalysed hydrogenation of 20b to 21b in 1,4-dioxane at 4 bar (* = 1,3,5-trimethoxybenzene in capillary insert)

197

Chapter 4 – Experimental details

Chapter 4.3.5 – Typical procedure for hydrogenation of aldehydes 20 in 1,4-dioxane at 1 mmol scale

Inside a glovebox B(C6F5)3 (51 mg, 0.1 mmol) and aldehyde 20 (1 mmol) were dissolved in 1,4- dioxane (4 mL) and transferred into a Rotaflo ampoule also containing a magnetic stirrer bar. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction was heated in an oil bath to 80 °C as indicated in Table 2.2, with active stirring. Inside a glovebox an aliquot (0.4 mL) of the resulting homogeneous mixture was transferred to an NMR tube fitted with a J. Young’s valve, to which was also added a

1 19 11 sealed capillary insert containing PPh3 in C6D6, and the sample was analysed by H, F and B NMR spectroscopy. For 2,6-dichlorobenzaldehyde, 21n, reaction yield was determined by relative integration of 1H resonances for the product relative to the starting material, as described in Chapter 4.1

For pentafluorobenzaldehyde, 20m, the aliquot and main reaction mixture were recombined, and subsequent workup was performed under air. Following removal of all volatiles under reduced pressure, the crude reaction product was purified by flash column chromatography (SiO2; eluted using increasingly polar mixtures of pentane and ethyl acetate, increasing from an initial ratio of 14:1 respectively until reaching neat ethyl acetate). The product, (pentafluorophenyl)methanol, was isolated as a colourless oil (147 mg, 74%).

(Pentafluorophenyl)methanol, 21m16

1 19 H (CDCl3, 400 MHz) G: 4.77 (s, 2H), 2.97 (br s, 1H). F (CDCl3, 376 MHz) G: ˗144.6, -154.4, -162.2.

Chapter 4.3.6 – Procedure for selective hydrogenation of 3-methyl-2-butanone, 20d, in the presence of 3,3-dimethyl-2-butanone, 20e

Inside a glovebox B(C6F5)3 (5.1 mg, 0.01 mmol), 3-methyl-2-butanone (5.3PL, 0.05 mmol) and 3,3-dimethyl-2-butanone (6.3 PL, 0.05 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath to 100 °C for 25 hours without active mixing, and re-analysed. Reaction outcome

198

Chapter 4 – Experimental details

(full reduction of 20d to alcohol 21d, no reduction of 20e) was determined by integration of 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.7 – Reaction of [Bu4N][HB(C6F5)3]/B(C6F5)3 ([Bu4N][1a·H]/1a) with acetone, 20a

Inside a glovebox B(C6F5)3 (1a, 10.2 mg, 0.02 mmol) and acetone (20a, 1.5 PL, 0.02 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. The reaction mixture was

1 19 11 analysed by H, F and B NMR spectroscopy. [Bu4N][HB(C6F5)3] ([Bu4N][1a·H], 15.1 mg, 0.02 mmol) was added, and the reaction mixture was heated to 100 °C for one hour before being reanalysed. All volatiles were removed in vacuo, CD2Cl2 (0.4 mL) was added to give a homogeneous solution, and the reaction mixture was again reanalysed. Reaction outcome (44 % reduction of the carbonyl moiety) was determined by integration of product 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.8 – Reaction of [Bu4N][HB(C6F5)3]/B(C6F5)3 ([Bu4N][1a·H]/1a) with 4ʹ-nitroacetophenone, 20g

Inside a glovebox B(C6F5)3 (1a, 10.2 mg, 0.02 mmol) and 4ʹ-nitroacetophenone (20g, 3.3 mg, 0.02 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J.

Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. The reaction

1 19 11 mixture was analysed by H, F and B NMR spectroscopy. [Bu4N][HB(C6F5)3] ([Bu4N][1a·H], 15.1 mg, 0.02 mmol) was added, and the reaction mixture was heated to 80 °C for 21 hours before being reanalysed. All volatiles were removed in vacuo, CD2Cl2 (0.4 mL) was added to give a homogeneous solution, and the reaction mixture was again reanalysed. Reaction outcome (53 % reduction of the carbonyl moiety) was determined by integration of product 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.9 – Reaction of [Bu4N][HB(C6F5)3] ([Bu4N][1a·H]) with 4-nitrobenzaldehyde, 20k

Inside a glovebox 4-nitrobenzaldehyde (20k, 3.0 mg, 0.02 mmol) was dissolved in CD2Cl2 (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed

1 19 11 capillary insert containing PPh3 in C6D6. The reaction mixture was analysed by H, F and B NMR spectroscopy. [Bu4N][HB(C6F5)3] ([Bu4N][1a·H], 15.1 mg, 0.02 mmol) was added, and the mixture was

199

Chapter 4 – Experimental details immediately reanalysed. Reaction outcome (full reduction of the carbonyl moiety) was determined based upon complete loss of the starting material 1H resonances.

Chapter 4.3.10 – Reaction of [Bu4N][HB(C6F5)3] ([Bu4N][1a·H]) with 2,6-dichlorobenzaldehyde, 20n

Inside a glovebox 2,6-dichlorobenzaldehyde (20n, 3.5 mg, 0.02 mmol) was dissolved in CD2Cl2 (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a

1 19 11 sealed capillary insert containing PPh3 in C6D6, and the solution was analysed by H, F and B NMR spectroscopy. [Bu4N][HB(C6F5)3] ([Bu4N][1a·H], 15.1 mg, 0.02 mmol) was added, and the mixture was immediately reanalysed. Reaction outcome (full reduction of the carbonyl moiety) was determined based upon complete loss of the starting material 1H resonances.

Chapter 4.3.11 – Reaction of [Bu4N][HB(C6F5)3]/B(C6F5)3 ([Bu4N][1a·H]/1a) with 4-nitrobenzaldehyde, 20k

Inside a glovebox B(C6F5)3 (1a, 10.2 mg, 0.02 mmol) and 4-nitrobenzaldehyde (20k, 3.0 mg, 0.02 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J.

Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. The reaction

1 19 11 mixture was analysed by H, F and B NMR spectroscopy. [Bu4N][HB(C6F5)3] ([Bu4N][1a·H], 15.1 mg, 0.02 mmol) was added, and the reaction mixture was immediately reanalysed. All volatiles were removed in vacuo, CD2Cl2 (0.4 mL) was added to give a homogeneous solution, and the reaction mixture was again reanalysed. Reaction outcome (complete reduction of the carbonyl moiety) was determined by integration of product 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.12 – Reaction of [Bu4N][HB(C6F5)3]/B(C6F5)3 ([Bu4N][1a·H]/1a) with 2,6- dichlorobenzaldehyde, 20n

Inside a glovebox B(C6F5)3 (1a, 10.2 mg, 0.02 mmol) and 2,6-dichlorobenzaldehyde (20n, 3.5 mg, 0.02 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing 1,3,5-

1 19 11 trimethoxybenzene in C6D6. The reaction mixture was analysed by H, F and B NMR spectroscopy.

[Bu4N][HB(C6F5)3] ([Bu4N][1a·H], 15.1 mg, 0.02 mmol) was added, and the reaction mixture was immediately reanalysed. All volatiles were removed in vacuo, CD2Cl2 (0.4 mL) was added to give a

200

Chapter 4 – Experimental details homogeneous solution, and the reaction mixture was again reanalysed. Reaction outcome (complete reduction of the carbonyl moiety) was determined by integration of product 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.13 – Synthesis of iPrOH·B(C6F5)3 (iPrOH·1a)

This procedure is based on a literature procedure for preparation of similar species.4 To a solution of B(C6F5)3 (1a, 500 mg, 0.98 mmol) in pentane (30 mL) was added iPrOH (21a, 74 PL, 0.98 mmol)). The solution was stirred at RT overnight, during which time a white precipitate was observed to form. This solid was isolated by filtration, washed with additional pentane (2 mL), and dried under vacuum, to yield the target compound as a white powder (300 mg, 54 %).

1 3 1 1 3 1 1 19 H (CH2Cl2, 400 MHz) G: 4.50 [sp, J( H- H) = 6.3 Hz, 1H], 1.40 [d, J( H- H) = 6.3 Hz, 6H]. F (CH2Cl2,

11 376 MHz) G: ˗133.4, -155.6, -163.1. B (CH2Cl2, 128 MHz) G: 4.2.

Chapter 4.3.14 – Interaction of iPrOH·B(C6F5)3 (iPrOH·1a) with 1,4-dioxane

Inside a glovebox iPrOH·B(C6F5)3 (iPrOH·1a, 10.6 mg, 0.020 mmol) was dissolved in C7D8 (0.5 mL), and the solution transferred into an NMR tube fitted with a J. Young’s valve, to which was also

1 19 added a sealed capillary insert containing PPh3 in C6D6. The reaction mixture was analysed by H, F and 11B NMR spectroscopy. 1,4-dioxane (1.7 PL, 0.020 mmol) was added and the reaction mixture re- analysed. Further 1,4-dioxane (15.3 PL, 0.18 mmol) was added, and the reaction mixture again re- analysed.

Chapter 4.3.15 – Interaction of B(C6F5)3 (1a) with 1,4-dioxane

Inside a glovebox B(C6F5)3 (1a, 10.2 mg, 0.020 mmol) was dissolved in DFB (0.4 mL), and the solution transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed

1 19 11 capillary insert containing PPh3 in C6D6. The reaction mixture was analysed by H, F and B NMR spectroscopy. 1,4-dioxane (1.7 PL, 0.020 mmol) was added and the reaction mixture re-analysed. Further 1,4-dioxane (15.3 PL, 0.18 mmol) was added, and the reaction mixture again re-analysed.

201

Chapter 4 – Experimental details

Chapter 4.3.16 – Procedure for hydrogenation of acetone, 20a, in 1,4-dioxane (10 bar)

Inside a glovebox B(C6F5)3 (1a, 5.1 mg, 0.01 mmol) and acetone (20a, 14.7 PL 0.2 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into a Wilmad high pressure NMR tube fitted with a

PV-ANV PTFE valve. H2 was admitted at RT to a pressure of 10 bar and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath to 100 °C for 6 h without active mixing, and re-analysed. Reaction yield (complete conversion to alcohol 21a) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.3.17 – Typical procedure for hydrogenation of other ketones 20 in 1,4-dioxane (10 bar)

Inside a glovebox B(C6F5)3 (1a, 10.2 mg, 0.02 mmol) and ketone 20 (0.1 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into a Wilmad high pressure NMR tube fitted with a PV-ANV

PTFE valve. H2 was admitted at RT to a pressure of 10 bar and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath to 80 °C as indicated in Scheme 2.35 (24 h and 39 h for 20h and 20f, respectively), without active mixing, and re-analysed. Reaction yield (>99 % and 92 % conversion to alcohols 21h and 21f, respectively) was determined by relative integration of product and starting material 1H resonances as described in Chapter 4.1.

Chapter 4.3.18 – Procedure for hydrogenation of acetone, 20a, in 1,4-dioxane (50 bar)

This experiment was performed at the University of East Anglia with the assistance of

Professor Gregory G. Wildgoose and Dr Colin McDonald. Inside a glovebox, B(C6F5)3 (1a, 14.6 mg, 0.029 mmol) and acetone (20a, 42 PL, 0.57 mmol) were dissolved in 1,4-dioxane (2 mL), and transferred into a 10mm diameter Sapphire NMR tube. H2 was admitted at RT to a pressure of 50 bar and the reaction mixture was briefly shaken by hand before being analysed by 1H NMR spectroscopy. The sample was heated to 50 °C inside the NMR spectrometer and re-analysed periodically. Reaction yield (85 % conversion to alcohol 21a after 19 h) was determined by relative integration of product and starting material 1H resonances as described in Chapter 4.1.

202

Chapter 4 – Experimental details

Chapter 4.3.19 – Typical procedure for hydrogenation of aldehyde 20n catalysed by boranes 1 in 2- MeTHF (4 bar)

Inside a glovebox borane 1 (0.02 mmol) and aldehyde 20n (35.0 mg, 0.2 mmol) were dissolved in 2-MeTHF (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Chapter 2.3.6, without active mixing, and re-analysed. Reaction yields were determined by integration of 1H resonances for the product alcohol 21n relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.3.20 – Typical procedure for hydrogenation of carbonyls 20 catalysed by boranes 1 in 2- MeTHF (10 bar)

Inside a glovebox borane 1 (0.02 mmol) and carbonyl 20 (0.2 mmol) were dissolved in 2- MeTHF (0.7 mL) and transferred into a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted at RT to a pressure of 10 bar and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Chapter 2.3.6, without active mixing, and re-analysed. Reaction yield was determined by relative integration of 1H resonances for the reduced products (21 and 22) and starting materials, as described in Chapter 4.1.

Chapter 4.3.21 – Procedure for hydrogenation of ester 24b catalysed by borane 1b in 2-MeTHF

Inside a glovebox borane 1b (11.9 mg, 0.02 mmol) and methyl formate (24b, 12 PL, 0.2 mmol) were dissolved in 2-MeTHF (0.7 mL) and transferred into a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted at RT to a pressure of 10 bar and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath to 100 °C for 3 days without active mixing, and re-analysed. Reaction yield (31 % conversion to various reduced, methoxide-containing species) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

203

Chapter 4 – Experimental details

Chapter 4.4 – Experimental details relating to Chapter 2.4

Supplementary information relating to this section, including NMR spectra for catalytic and other reactions, can be found online at: http://pubs.acs.org/doi/abs/10.1021/acscatal.5b01417

Reactions described in this section as being ‘non-anhydrous’ were prepared using ACS reagent grade 1,4-dioxane purchased from Sigma-Aldrich and used without any further drying, degassing or other purification. Karl-Fischer titration (performed by Mr Damion Box) indicated that this solvent contained 220 ppm (0.022 % w/w) H2O (average of three measurements).

Reactions in the section performed at 45 bar pressure were performed in Parr reactors (25 mL internal volume for 2 mmol scale reactions; 100 mL for 8 mmol scale; either size for 4 mmol scale).

Chapter 4.4.1 – Interaction of H2O·B(C6F5)3 (1a·OH2) with 1,4-dioxane

Inside a glovebox H2O·B(C6F5)3 (1a·OH2, 10.6 mg, 0.02 mmol) was dissolved in DFB (0.4 mL), and the solution transferred into an NMR tube fitted with a J. Young’s valve, to which was also added

1 19 11 a sealed capillary insert containing PPh3 in C6D6. The reaction mixture was analysed by H, F and B NMR spectroscopy. 1,4-dioxane (1.7 PL, 0.02 mmol) was added and the reaction mixture re-analysed. Further 1,4-dioxane (15.3 PL, 0.18 mmol) was added, and the reaction mixture again re-analysed.

Chapter 4.4.2 – Interaction of H2O·B(C6F5)3 (1a·OH2) with acetone (20a)

Inside a glovebox H2O·B(C6F5)3 (1a·OH2, 10.6 mg, 0.02 mmol) was dissolved in DFB (0.4 mL), and the solution transferred into an NMR tube fitted with a J. Young’s valve, to which was also added

1 19 11 a sealed capillary insert containing PPh3 in C6D6. The reaction mixture was analysed by H, F and B NMR spectroscopy. Acetone (20a, 1.7 PL, 0.02 mmol) was added and the reaction mixture re- analysed. Further 20a (15.3 PL, 0.18 mmol) was added, and the reaction mixture again re-analysed.

Chapter 4.4.3 – Procedure for hydrogenation of acetone (20a) using H2O·B(C6F5)3 (1a·OH2) at 4 bar

Inside a glovebox H2O·B(C6F5)3 (1a·OH2, 5.3 mg, 0.01 mmol) and acetone (20a, 7.3 PL, 0.1 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which was also added a sealed capillary insert containing PPh3 in C6D6. H2 was admitted via

204

Chapter 4 – Experimental details a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath to 100 °C for 13 days without active mixing, and re-analysed. Reaction yield (75 % conversion to alcohol 21a) was determined by integration of product 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.4.4 – Typical procedure for hydrogenation of carbonyls 20a and 20n using H2O·B(C6F5)3

(1a·OH2) with added molecular sieves

Inside a glovebox H2O·B(C6F5)3 (1a·OH2, 5.3 mg, 0.01 mmol) and carbonyl 20 (0.1 mmol) were dissolved in 1,4-dioxane (0.4 mL) and transferred into an NMR tube fitted with a J. Young’s valve, to which were also added three 4 Å molecular sieve beads and a sealed capillary insert containing PPh3 in C6D6. H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT) and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction was heated in an oil bath as indicated in Scheme 2.51, without active mixing, and re-analysed. Reaction yields (61 % and 60 % conversion to alcohols 21a and 21n respectively) were determined by integration of product 1H resonances relative to the capillary insert, as described in Chapter 4.1.

Chapter 4.4.5 – Typical procedure for hydrogenation of carbonyls 20a and 20n using H2O·B(C6F5)3

(1a·OH2) at 10 bar

Inside a glovebox, carbonyl 20 (0.2 mmol) was added to a solution of H2O·B(C6F5)3 (1a·OH2) in anhydrous 1,4-dioxane (0.7 mL), and the reaction mixture was transferred into a Wilmad high pressure

NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted at RT to a pressure of 10 bar and the reaction mixture was briefly shaken by hand before being analysed by 1H, 19F and 11B NMR spectroscopy. The reaction vessel was heated to 100 °C as indicated in Scheme 2.52 (70 h and 5 days for 20aand 20n, respectively), without active mixing, and re-analysed. Reaction yields (94 % and complete conversion to alcohols 21a and 21n respectively) were determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

205

Chapter 4 – Experimental details

Chapter 4.4.6 – Typical procedure for hydrogenation of acetone (20a) in ‘anhydrous’ solvent at 45 bar, following addition of H2O

Inside a glovebox, acetone (20a, 0.29 mL, 4 mmol) and B(C6F5)3 (1a, 51 mg, 0.1 mmol) were dissolved in anhydrous 1,4-dioxane (15 mL). H2O was subsequently added, and the reaction mixture transferred into a stainless steel Parr reactor (which had previously been evacuated and back-filled

1 19 11 with dry N2 three times). An aliquot was taken via cannula, and analysed by H, F and B NMR spectroscopy. The reaction vessel was filled with H2 to a pressure of 45 bar, before being heated with stirring to 100 °C as indicated in Scheme 2.53 (39 h, 84 h and 108 h for 0.1, 0.2 and 0.5 mmol H2O, respectively). After returning to room temperature the reaction vessel was cooled in an ice bath, and the excess pressure slowly vented over the course of 30 minutes. Upon re-warming a new aliquot was removed for analysis. Reaction yields (92 %, 98 % and 92 % conversion to alcohol 21a, respectively) were determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.4.7 – Typical procedure for non-anhydrous hydrogenation

Inside a glovebox B(C6F5)3 (1a) and the substrate were weighed out into a vial. Undried 1,4- dioxane was added on the open bench (to give a final substrate concentration of 0.53 M), and the resulting solution quickly transferred into a stainless steel Parr reactor. An aliquot was taken via

1 19 11 cannula, and analysed by H, F and B NMR spectroscopy. The reaction vessel was filled with H2 to a pressure of 45 bar, before being heated with stirring as indicated in Table 2.3. After returning to room temperature the reaction vessel was cooled in an ice bath, and the excess pressure slowly vented over the course of 30 minutes. Upon re-warming a new aliquot was removed for analysis. Reaction yields were determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

The reactions corresponding to entries 2, 4 and 8 in Table 2.3 were kindly performed (under the direction of the author) by Trevor R. Simmons and Elliot J. Lawrence, working in the research group of Professor Gregory G. Wildgoose at the University of East Anglia.

206

Chapter 4 – Experimental details

Figure 4.3 – 1H NMR spectra for the non-anhydrous 1a-catalysed hydrogenation of 20a to 21a in 1,4- dioxane at 45 bar

Chapter 4.4.8 – Procedure to establish moisture content during non-anhydrous hydrogenation

A 0.053 M solution of B(C6F5)3 (1a) in undried 1,4-dioxane was prepared in the same manner as for an initial non-anhydrous reaction mixture (see Chapter 4.4.7), and transferred into a stainless steel Parr reactor. After an initial aliquot was taken for analysis by 1H, 19F and 11B NMR spectroscopy the reactor was pressurised with H2 to 45 bar, and the reaction mixture stirred at room temperature for 90 minutes. The reactor was then cooled, vented and re-warmed in the manner described in section 4, above, and a final aliquot was removed under a flush of dry N2. For both aliquots, NMR tubes fitted with J. Young’s valves were used. The tubes were previously flame-dried under vacuum, back-filled with dry N2 and kept sealed until use.

Chapter 4.4.9 – Procedure for deoxygenative hydrogenation of acetophenone, 20j

Inside a glovebox, acetophenone (20j, 118 PL, 2 mmol) and B(C6F5)3 (1a, 205 mg, 0.4 mmol) were dissolved in anhydrous 1,4-dioxane (3.75 mL). The mixture was then transferred into a stainless steel Parr reactor (which had previously been evacuated and back-filled with dry N2 three times). An aliquot was taken via cannula, and analysed by 1H, 19F and 11B NMR spectroscopy. The reaction vessel

207

Chapter 4 – Experimental details

was filled with H2 to a pressure of 45 bar, before being heated with stirring to 100 °C for 61 h. After returning to room temperature the reaction vessel was cooled in an ice bath, and the excess pressure slowly vented over the course of 30 minutes. Upon re-warming a new aliquot was removed for analysis. Reaction yield (complete conversion to ethylbenzene, 25j) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.5 – Experimental details relating to Chapter 3.2

Supplementary information relating to this section, including NMR spectra, can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201606639/abstract

Chapter 4.5.1 – Synthesis of iPr4Sn

This reaction was developed in cooperation with Mr Joshua Sapsford.11

To a suspension of Mg turnings (5.64 g, 232 mmol) in THF (40 mL) was added dropwise a solution of iPrCl (21.2 mL, 232 mmol) in THF (80 mmol) at RT (maintained through use of a water bath).

After stirring for 20 h the solution was filtered dropwise over 4 h onto a stirred suspension of SnCl4 (13.4 g, 51.5 mmol) in THF (120 mL), which was maintained at 0 °C through use of an ice bath. The solid residue was washed with further THF (30 mL), which was filtered across in an identical manner. The resulting suspension was heated to 60 °C for 25 h, cooled to RT, and extracted into pentane (3 x

150 mL). The remaining work-up was performed under air: the solution was dried over MgSO4 and filtered, and the solvent removed under reduced pressure. The resulting oil was distilled (110 °C, 1

17 mbar) to afford iPr4Sn as a colourless oil (10.4 g, 69 %).

1 3 1 1 3 117 1 3 119 1 H NMR (400 MHz, CDCl3) G : 1.32 [6H, d, J( H- H) = 7.2 Hz, J( Sn- H) = 29 Hz, J( Sn- H) = 30 Hz,

119 1 CH3], 1.42-1.55 [1H, m, CH]. Sn{ H} NMR (149 Hz, CDCl3) G: -42.9 (s).

Chapter 4.5.2 – Synthesis of iPr3SnOTf ([26b]OTf)

This reaction was developed in cooperation with Mr Joshua Sapsford.11

To a solution of iPr4Sn (8.0 g, 27.5 mmol) in CHCl3 (80 mL) was added HOTf (3.9 g, 26.2 mmol) in CHCl3 (40 mL). The mixture was stirred at RT for 5 days before the solvent was removed in vacuo and the resulting solid washed with pentane (2 x 15 mL), affording iPr3SnOTf as a white solid (6.4 g, 61 %).

208

Chapter 4 – Experimental details

1 3 1 1 3 117 1 3 119 1 H NMR (400 MHz, CDCl3) G: 1.44 [6H, d, J( H- H) = 7.6 Hz, J( Sn- H) = 86 Hz, J( Sn- H) = 90 Hz,

3 1 1 2 119 1 13 1 CH3], 2.07 [1H, sp, J( H- H) = 7.6 Hz, J( Sn- H) = 39 Hz, CH]. C{ H} NMR (101 MHz, CDCl3) G: 20.7

2 117/119 13 1 117 13 1 119 13 [s, J( Sn- C) = 16 Hz, CH3], 27.2 [s, J( Sn- C) = 302 Hz, J( Sn- C) = 316 Hz, CH], 119.0 [q,

1 19 13 19 119 1 J( F- C) = 319 Hz, CF3]. F NMR (376 MHz, CDCl3) G: -76.7 (s). Sn{ H} NMR (149 MHz, CDCl3, 0.06

+ + M) G: 156 [br s, 'v1/2 = 130 Hz]. MS (APCI) m/z: 327 (iPr3SnOSO2 ), 249 (iPr3Sn ). Anal. calcd. for

C10H21F3O3SSn: C, 30.25; H, 5.33. Found: C, 30.08; H, 5.45.

Chapter 4.6 – Experimental details relating to Chapter 3.3

Supplementary information relating to this section, including NMR spectra, can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201606639/abstract

Chapter 4.6.1 – H2 activation by DABCO and iPr3SnOTf ([26b]OTf)

[26b]OTf (15.9 mg, 0.04 mmol) and DABCO (4.5 mg, 0.04 mmol) were dissolved in DFB (0.7 mL) in an NMR tube fitted with a J. Young’s valve, to which was also added a capillary insert containing

1 119 1 PPh3 in C6D6 (to provide a lock and reference). Initial H and Sn{ H} NMR spectra were recorded, H2 was added via a freeze-pump-thaw method (1 bar at -196 °C, ca. 4 bar at RT), and the sample allowed to stand for 2 days (with occasional agitation) before being re-analysed. Note that due to slow relaxation of the SnH resonance (ca. 15 s, measured in situ), the final 1H NMR spectrum was recorded using an extended delay of 100s. Relative integration of the SnH and DABCO (CH2) resonances indicated 54 % conversion (to [26b]H and DABCO·HOTf).

Chapter 4.6.2 – D2 activation by DABCO and iPr3SnOTf ([26b]OTf)

[26b]OTf (15.9 mg, 0.04 mmol) and DABCO (4.5 mg, 0.04 mmol) were dissolved in DFB (0.7 mL) in an NMR tube fitted with a J. Young’s valve. Initial 1H, 2H and 119Sn{1H} NMR spectra were recorded, D2 was added through use of a Toepler pump (ca. 2 bar), and the sample allowed to stand for 2 days (with occasional agitation) to allow it to reach equilibrium before being re-analysed.

Chapter 4.6.3 – H2 activation by col or lut and iPr3SnOTf ([26b]OTf)

[26b]OTf (15.9 mg, 0.04 mmol) and collidine (col, 5.3 PL, 0.04 mmol) or lutidine (lut, 4.7 PL, 0.04 mmol) were dissolved in DFB (0.7 mL) in a Wilmad high pressure NMR tube fitted with a PV-ANV

209

Chapter 4 – Experimental details

PTFE valve. H2 was admitted up to a pressure of 10 bar (at RT), and the sample was allowed to stand for 20 hours (with occasional agitation) before being re-analysed. Note that due to slow relaxation of the SnH resonance (ca. 14 s, measured in situ), the final 1H NMR spectrum was recorded using an extended delay of 100s. Relative integration of the SnH and col or lut (CH3) resonances indicated 20 % and 15 % conversion, respectively (to [26b]H and col·HOTf or lut·HOTf).

Chapter 4.7 – Experimental details relating to Chapter 3.4

Supplementary information relating to this section, including NMR spectra for catalytic and other reactions, can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201606639/abstract

Chapter 4.7.1 – Typical procedure for hydrogenation of imines 2 catalysed by iPr3SnOTf ([26b]OTf)

A solution of imine 2 (0.2 mmol) and, if necessary, collidine (2.6 PL, 0.02 mmol) in DCB (0.7 mL) was added to iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted to a pressure of 10 bar (at RT). The reaction mixture was heated in an oil bath to 120 °C as indicated in Table 3.3. Reaction yields were determined by relative integration of product (amine 3) and starting material 1H resonances, as described in Chapter 4.1.

Figure 4.4 – 1H NMR spectra for the [26b]OTf-catalysed hydrogenation of 2g to 3g in DCB (final spectrum taken at 70 °C to ensure homogeneity).

210

Chapter 4 – Experimental details

Chapter 4.7.2 – Stoichiometric reaction of iminium 2g·HOTf with iPr3SnH ([26b]H)

To a solution of HOTf (3.0 mg, 0.02 mmol) in DCB (0.35 mL) in an NMR tube fitted with a J. Young’s valve was added imine 2g (3.6 PL, 0.02 mmol). The resulting mixture was agitated by hand, resulting in immediate formation of a colourless precipitate, and left to stand at RT for 24 h. To this was added [26b]H (5.0 mg, 0.02 mmol) in DCB (0.35mL). The mixture was again agitated by hand, becoming fully homogeneous within ca. 1 min. Reaction yield (65 % reduction of the substrate after 17 h) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.7.3 – Typical procedure for hydrogenation of aldehydes and ketones 20 catalysed by iPr3SnOTf ([26b]OTf)

A solution of carbonyl 20 (0.4 mmol) and Lewis base (0.04 mmol) in DCB (0.7 mL) was added to iPr3SnOTf ([26b]OTf, 15.9 mg, 0.04 mmol) in a Wilmad high pressure NMR tube fitted with a PV-

ANV PTFE valve. For 20j and 20p a drop of SiMe4 was also added to act as an internal integration standard. H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an Al bead bath as indicated in Table 3.4. In certain cases the reaction was periodically removed from the heating bath and re-pressurised to 10 bar with H2. Reaction yields were determined by relative integration of product (alcohol 21) and starting material 1H resonances (or integration relative to the internal standard for 20j and 20p), as described in Chapter 4.1

Figure 4.5 – 1H NMR spectra for the [26b]OTf-catalysed hydrogenation of 20a to 21a in DCB

211

Chapter 4 – Experimental details

Chapter 4.7.4 – Stoichiometric reaction of acetone (20a) with iPr3SnH ([26b]H) and iPr3SnOTf ([26b]H)

To a suspension of [26b]OTf (7.9 mg, 0.02 mmol) and acetone (20a, 1.5 PL, 0.02 mmol), in DCB (0.7 mL) in an NMR tube fitted with a J. Young’s valve was added [26b]H (5.5 mg, 0.022 mmol). NMR spectroscopic analysis showed immediate reduction. Further reduction was significantly slower, presumably due to the poor solubility of [26b]OTf under these conditions, but still proceeded to completion overnight at RT (based on relative integration of product and starting material resonances, as described in Chapter 4.1).

Chapter 4.7.5 – Stoichiometric reaction of acetone (20a) with iPr3SnH ([26b]H) and col·HOTf

To a solution of HOTf (3.0 mg, 0.02 mmol) and collidine (2.6 PL, 0.02 mmol) in DCB (0.7 mL) in an NMR tube fitted with a J. Young’s valve was added acetone, (20a, 1.5 PL, 0.02 mmol), followed by [26b]H (5.0 mg, 0.02 mmol). The resulting mixture was left to stand at RT for 24 h. No evidence for reduction of 20a was observed by NMR spectroscopy, although the 1H NMR spectrum did indicate some release of H2. Slow reduction was observed upon subsequent heating to 120 °C (43 % conversion after 17 h, based upon relative integration of product and starting material 1H resonances, as described in Chapter 4.1).

Chapter 4.7.6 – Procedure for hydrogenation of acetone (20a) catalysed by iPr3SnOTf ([26b]OTf) and iPr3SnOiPr ([26b]OiPr)

To a solution of acetone (32 PL, 0.44 mmol) and [26b]OTf (15.9 mg, 0.04 mmol) in DCB (0.7 mL) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve, was added [26b]H (10.0 mg, 0.04 mmol). H2 was admitted up to a pressure of 10 bar (at RT). Initial NMR spectroscopic analysis indicated complete consumption of [26b]H to form [26b]OiPr. The reaction mixture was heated in an Al bead bath to 180 °C for 16 h. Reaction yield (32 % reduction to alcohol 21a after 16 h) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

212

Chapter 4 – Experimental details

Chapter 4.7.7 – Procedure for hydrogenation of 1-piperidino-1-cyclohexene (27) catalysed by iPr3SnOTf ([26b]OTf)

A solution of 1-piperidino-1-cyclohexene (27, 33.1 mg, 0.2 mmol) and collidine (2.6 PL, 0.02 mmol) in DCB (0.7 mL) was added to iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an oil bath to 120 °C for 4 h. 1H NMR spectroscopic analysis indicated >99 % conversion to 1-cyclohexyl piperidine based on consumption of starting material.

To confirm the identity of the product, an authentic sample of 1-cyclohexyl piperidine (28, 37 PL, 0.2 mmol) was combined with collidine (2.6 PL, 0.02 mmol) and [26b]OTf (7.9 mg, 0.02 mmol) in

1 DCB (0.7 mL), and H2 was admitted up to a pressure of 10 bar (at RT). The H NMR spectrum for this mixture closely matched that for the hydrogenation reaction.

Chapter 4.7.8 – Procedure for hydrogenation of acridine (4) catalysed by iPr3SnOTf ([26b]OTf)

A solution of acridine (4, 35.8 mg, 0.2 mmol) in DCB (0.7 mL) was added to iPr3SnOTf ([26b]OTf,

7.9 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an oil bath to 120 °C for 32 h. 1H NMR spectroscopic analysis indicated 84 % conversion to acridane (5, final NMR spectrum recorded at 70 °C to ensure homogeneity), based upon relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.7.9 – Procedure for hydrogenation of butyl acrylate (12a) catalysed by iPr3SnOTf ([26b]OTf)

A solution of butyl acrylate (12a, 29 PL, 0.2 mmol) and collidine (2.6 PL, 0.02 mmol) in DCB

(0.7 mL) was added to iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an oil bath to 120 °C for 80 h. 1H NMR spectroscopic analysis indicated 83 % conversion to n-butyl propanoate (13a), based upon relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

213

Chapter 4 – Experimental details

Chapter 4.8 – Experimental details relating to Chapter 3.5

Supplementary information relating to this section, including NMR spectra, can be found online at: http://onlinelibrary.wiley.com/doi/10.1002/anie.201606639/abstract

For reactions described in this section as ‘open bench’ DCB, col, SiMe4 and substrates were purchased from major suppliers (non-anhydrous grades) and used as supplied without further drying or other purification. H2 was purchased from BOC (research grade) and used without further drying. iPr3SnOTf ([26b]OTf) was stored in an open container under air for a period of at least a week prior to use.

Chapter 4.8.1 – Typical procedure for hydrogenation of acetone (20a) in ‘anhydrous’ solvent, following addition of H2O

Inside a glovebox, a solution of acetone (20a, 29 PL , 0.4 mmol) and col (5.3 PL, 0.04 mmol) in

DCB (0.7 mL) was added to iPr3SnOTf ([26b]OTf, 15.9 mg, 0.04 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2O was subsequently added, and H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated to 180 °C in an Al bead bath as indicated in Scheme 3.15. The reactions were periodically removed from the heating bath in order to monitor their progress, and were re-pressurised upon reaching ca. 50 % conversion. Reaction yield was determined by relative integration of product (alcohol 21a) and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.8.2 – Typical procedure for ‘open bench’ hydrogenation of carbonyls 20a and 20n

To a sample of iPr3SnOTf ([26b]OTf, 15.9 mg, 0.04 mmol) were added DCB (0.7 mL), col (5.3 PL, 0.04 mmol) and carbonyl 20 (0.4 mmol) on the open bench. After shaking thoroughly the resulting homogeneous solution was transferred to a Wilmad high pressure NMR tube fitted with a PV-ANV

PTFE valve (still under air), to which H2 was then added up to a pressure of 10 bar (at RT). The reaction mixture was heated in an Al bead bath to 180 °C for 16 h, before being re-pressurised with H2 and heated for a further 16 h (20a) or 32 h (20n). Reaction yield (95 % and 65 % conversion to alcohols 20a and 20n, respectively) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

214

Chapter 4 – Experimental details

Chapter 4.8.3 – Procedure for moisture-tolerant hydrogenation of imine 2g

A sample of iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) was removed from a glovebox and left to sand under air in an open contained for 1 week. The sample was flushed back into the glovebox (without being exposed to vacuum), where a solution of imine 2g (36 PL, 0.2 mmol) and a drop of

SiMe4 (added as an internal standard) in DCB (0.7 mL) was added. The resulting mixture was transferred into a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve, and H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated to 180 °C for 5 days. Reaction yield (complete reduction to amine 3g) was determined by integration of product 1H resonances relative to the internal standard, as described in Chapter 4.1.

Chapter 4.8.4 – Procedure for moisture-tolerant hydrogenation of imine 2d

A solution of imine 2d (36.2 mg, 0.2 mmol) and collidine (2.6 PL, 0.02 mmol) in DCB (0.7 mL) was added to iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. A drop of SiMe4 was also added to act as an internal integration standard. H2O was subsequently added (0.8 PL, 0.04 mmol), and H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an Al bead bath to 180 °C for 32 h without stirring. Reaction yield (complete reduction to amine 3g) was determined by integration of product 1H resonances relative to the internal standard, as described in Chapter 4.1.

Chapter 4.8.5 – Typical procedure for ‘open bench’ hydrogenation of imines 2d and 2e

To a sample of iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) were added DCB (0.7 mL), col (2.6 PL,

0.02 mmol) and imine 2 (0.2 mmol) on the open bench. A drop of SiMe4 was also added to act as an internal integration standard. After shaking thoroughly the resulting mixture was transferred to a

Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve (still under air), to which H2 was then added up to a pressure of 10 bar (at RT). The reaction mixture was heated in an Al bead bath to 180 °C for 32 h without stirring. Reaction yield (complete reduction to form amine 3d or a 72:5:23 ratio of 3e, 23j and 21j, respectively) was determined by integration of product 1H resonances relative to the internal standard, as described in Chapter 4.1.

215

Chapter 4 – Experimental details

Chapter 4.8.6 – Typical procedure for ‘open bench’ hydrogenation of imines 2d and 2e in the presence of molecular sieves

To a sample of iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) were added DCB (0.7 mL), col (2.6 PL,

0.02 mmol) and imine 2 (0.2 mmol) on the open bench. A drop of SiMe4 was also added to act as an internal integration standard. After shaking thoroughly the resulting mixture was transferred to an NMR tube fitted with a J. Young’s valve; containing three 5 Å molecular sieve beads; and that had previously been flame-dried under vacuum, back-filled with dry N2 and kept sealed until use. A sealed capillary insert containing PPh3 in C6D6 was also added, and H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT). The reaction mixture was heated in an Al bead bath to 180 °C as indicated in Scheme 3.18 (48 h and 49 h for 2d and 2e,respectively), without stirring. Reaction yield (complete reduction to form amine 3d or an 88:6:7 ratio of 3e, 23j and 21j, respectively) was determined by integration of product 1H resonances relative to the internal standard, as described in Chapter 4.1.

Chapter 4.8.7 – Procedure for hydrogenation of imine 2d in the presence iPrOH (21a)

A solution of imine 2d (36.2 mg, 0.2 mmol), collidine (2.6 PL, 0.02 mmol), and iPrOH (21a, 3.1

PL, 0.04 mmol) in DCB (0.7 mL) was added to iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an Al bead bath to 180 °C for 32 h without stirring. Reaction yield (58 % reduction to amine 3d) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.8.8 – Procedure for hydrogenative amination of benzaldehyde and aniline

To a sample of iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) were added DCB (0.7 mL), col (2.6 PL, 0.02 mmol), benzaldehyde (20o, 20 PL, 0.2 mmol) and aniline (18 PL, 0.2 mmol) on the open bench.

A drop of SiMe4 was also added to act as an internal integration standard. After shaking thoroughly the resulting mixture was transferred to a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve (still under air), to which H2 was then added up to a pressure of 10 bar (at RT). The reaction mixture was heated in an Al bead bath to 180 °C for 48 h, without stirring. Reaction yield (complete conversion to amine 3d) was determined by 1H integration of product resonances relative to the internal standard, as described in Chapter 4.1.

216

Chapter 4 – Experimental details

Chapter 4.9 – Experimental details relating to Chapter 3.6

Chapter 4.9.1 – Synthesis of iPr3SnNTf2 ([26b]NTf2)

This reaction was developed in cooperation with Mr Joshua Sapsford.11

To a solution of iPr4Sn (84.6 mg, 0.291 mmol) in CHCl3 (2 mL) was added HNTf2 (81.8 mg, 0.291 mmol). The mixture was stirred at 60 °C for 48 h before the solvent was removed in vacuo and the resulting oily solid washed with pentane (2 x 5 mL), affording [26b]NTf2 as a white microcrystalline solid (98.2 mg, 64 %).

1 3 1 1 3 117 1 3 119 1 H NMR (400 MHz, CDCl3) G: 1.47 [6H, d, J( H- H) = 7.2 Hz, J( Sn- H) = 81 Hz, J( Sn- H) = 96 Hz,

13 1 1 117 13 CH3], 2.04-2.21 [1H, m, CH]. C{ H} NMR (101 MHz, CD3CN) G: 19.7 [s, CH3], 37.3 [s, J( Sn- C) = 587

1 119 13 1 19 13 19 Hz, J( Sn- C) = 613 Hz, CH], 120.9 [q, J( F- C) = 322 Hz, CF3]. F NMR (376 MHz, CDCl3) G: -77.4

119 1 + + (s). Sn{ H} NMR (149 MHz, CD3CN) G: –253 (s). MS (APCI) m/z: 530 (iPr3SnNTf2·H ), 249 (iPr3Sn ).

Single crystals of [26b]NTf2 suitable for analysis by X-ray crystallography were obtained by slow diffusion of pentane into a dilute CHCl3 solution (see Appendix A).

Chapter 4.9.2 – Procedure for hydrogenation of acetone (20a) catalysed by iPr3SnNTf2 ([26b]NTf2)

A solution of acetone (20a, 29 PL, 0.4 mmol) and col (5.3 PL, 0.04 mmol) in DCB (0.7 mL) was added to iPr3SnNTf2 ([26b]NTf2, 10.6 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an Al bead bath to 120 °C for 6 h without stirring, before being re-pressurised with H2 and heated for a further 12 h. Reaction yield (95 % reduction to alcohol 21a) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

Chapter 4.9.3 – Procedure for hydrogenation of imine 2g catalysed by iPr3SnNTf2 ([26b]NTf2)

A solution of imine 2g (36 PL, 0.2 mmol) in DCB (0.7 mL) was added to iPr3SnNTf2 ([26b]NTf2,

10.6 mg, 0.02 mmol) in a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve. H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was heated in an oil bath to 120 °C for 32 h without stirring. Reaction yield (39 % reduction to amine 3g) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

217

Chapter 4 – Experimental details

F F Chapter 4.9.4 – Procedure for in situ formation of [iPr3Sn·LB][Al(OR )4] ([26b·LB][Al(OR )4])

To a solution of LB (0.02 mmol) and iPr3SnOTf ([26b]OTf, 7.9 mg, 0.02 mmol) in DFB (0.7 mL)

F was added K[Al(OR )4] (20.1 mg, 0.04 mmol). The resulting mixture was shaken thoroughly for several minutes during which time a colourless precipitate was observed to form, which was removed via filtration.

F Figure 4.6 – NMR spectra for in situ generated [26b·col][Al(OR )4] in DFB (downfield aromatic col resonance is obscured by solvent signals)

218

Chapter 4 – Experimental details

F Figure 4.7 – NMR spectra for in situ generated [26b·DABCO][Al(OR )4] in DFB

219

Chapter 4 – Experimental details

F Chapter 4.9.5 – Procedure for hydrogenation of acetone (20a) catalysed by [iPr3Sn·col][Al(OR )4]

F ([26b·col][Al(OR )4])

F Following in situ formation of [iPr3Sn·col][Al(OR )4] in DFB (at slightly larger scale: 0.04 mmol of reagents in 1 mL DFB), and transfer to a Wilmad high pressure NMR tube fitted with a PV-ANV PTFE valve, all volatiles were removed in vacuo (to yield an oil), followed by addition of acetone (20a, 29

PL, 0.4 mmol) in DCB (0.7 mL). H2 was admitted up to a pressure of 10 bar (at RT). The reaction mixture was then heated in an Al bead bath to 120 °C for 48 h without stirring. Reaction yield (44 % reduction to alcohol 21a) was determined by relative integration of product and starting material 1H resonances, as described in Chapter 4.1.

F Chapter 4.9.6 – Typical procedure for activation of H2 using in situ-formed [iPr3Sn·DABCO][Al(OR )4]

F ([26b·DABCO][Al(OR )4])

F Following in situ formation of [iPr3Sn·DABCO][Al(OR )4] in DFB (as well as addition of an additional equivalent of DABCO, if necessary) and transfer to an NMR tube fitted with a J. Young’s valve, H2 was admitted via a freeze-pump-thaw method to a pressure of 1 bar at −196 °C (which equates to a pressure of approximately 4 bar at RT). In the absence of additional DABCO 1H, 19F, 27Al and 119Sn{1H} NMR spectra were unchanged after standing for several days at RT. By contrast in its

1 119 1 presence H2 activation was clearly indicated in the H and Sn{ H} NMR spectra, particularly through the appearance of resonances due to [26b]H.

F Chapter 4.9.7 – Typical procedure for activation of H2 using in situ-formed [iPr3Sn·col][Al(OR )4]

F ([26b·col][Al(OR )4])

F Following in situ formation of [iPr3Sn·col][Al(OR )4] in DFB (as well as addition of an additional equivalent of col, if necessary) and transfer to a Wilmad high pressure NMR tube fitted with a PV-ANV

1 19 PTFE valve, H2 was admitted up to a pressure of 10 bar (at RT). In the absence of additional col H, F, 27Al and 119Sn{1H} NMR spectra were unchanged after standing for several days at RT. By contrast in

1 119 1 its presence H2 activation was clearly indicated in the H and Sn{ H} NMR spectra, particularly through the appearance of resonances due to [26b]H.

220

Chapter 4 – Experimental details

Chapter 4.10 – References for Chapter 4

1. Lancaster, S., ChemSpider Synthetic Pages 2003, 215. 2. Ashley, A. E.; Herrington, T. J.; Wildgoose, G. G.; Zaher, H.; Thompson, A. L.; Rees, N. H.; Krämer, T.; O’Hare, D., Journal of the American Chemical Society 2011, 133, 14727. 3. Nicasio, J. A.; Steinberg, S.; Inés, B.; Alcarazo, M., Chemistry – A European Journal 2013, 19, 11016. 4. Bergquist, C.; Bridgewater, B. M.; Harlan, C. J.; Norton, J. R.; Friesner, R. A.; Parkin, G., Journal of the American Chemical Society 2000, 122, 10581. 5. Blackwell, J. M.; Morrison, D. J.; Piers, W. E., Tetrahedron 2002, 58, 8247. 6. Cui, Z.; Yu, H.-J.; Yang, R.-F.; Gao, W.-Y.; Feng, C.-G.; Lin, G.-Q., Journal of the American Chemical Society 2011, 133, 12394. 7. Weemers, J. J. M.; Sypaseuth, F. D.; Bäuerlein, P. S.; van der Graaff, W. N. P.; Filot, I. A. W.; Lutz, M.; Müller, C., European Journal of Organic Chemistry 2014, 350. 8. Barluenga, J.; Jiménez-Aquino, A.; Aznar, F.; Valdés, C., Journal of the American Chemical Society 2009, 131, 4031. 9. Kloc, K.; Kubicz, E.; Młochowski, J.; Syper, L., Synthesis 1987, 1084. 10. Fendrick, C. M.; Schertz, L. D.; Mintz, E. A.; Marks, T. J.; Bitterwolf, T. E.; Horine, P. A.; Hubler, T. L.; Sheldon, J. A.; Belin, D. D., In Inorganic Syntheses, John Wiley & Sons, Inc., 2007, 193.

11. Sapsford, J., Investigating the Direct Hydrogenation of CO2 Using Bulky Stannyl Cations, Imperial College London, 2016. 12. Jaroń, T.; Orłowski, P. A.; Wegner, W.; Fijałkowski, K. J.; Leszczyński, P. J.; Grochala, W., Angewandte Chemie International Edition 2015, 54, 1236. 13. Mungwe, N.; Swarts, A. J.; Mapolie, S. F.; Westman, G., Journal of Organometallic Chemistry 2011, 696, 3527. 14. Johnson Ii, D. C.; Widlanski, T. S., Tetrahedron Letters 2004, 45, 8483. 15. Mao, F.; Sui, D.; Qi, Z.; Fan, H.; Chen, R.; Huang, J., RSC Advances 2016, 6, 94068. 16. Zhang, D.; Chen, Z.; Cai, H.; Zou, X., Journal of Fluorine Chemistry 2009, 130, 938. 17. Kitching, W.; Olszowy, H. A.; Drew, G. M., Organometallics 1982, 1, 1244.

221

Appendices

Appendix A – X-ray crystallographic data

The crystallographic data in this section were acquired and processed by Mr Andrew Crawford. X-ray quality single crystals of iPr3SnNTf2 ([26b]NTf2) were obtained during collaborative work with Mr Joshua Sapsford.

Crystallographic data for [iPr3Sn·2H2O]OTf ([26b·2H2O]OTf)

The crystal used for this analysis was grown through slow evaporation of an aqueous solution. Single crystal X-ray diffraction data was collected with an Oxford Diffraction Xcalibur unit; crystals were mounted on a nylon MicroLoop™ using perfluoropolyether oil and measured in a stream of N2 at 173 K. The structure was solved in Olex21 by charge flipping using Superflip,2 and subsequently refined with the ShelXL3 refinement package.

The O-H hydrogen atoms H19A, H19B, H20A, and H20B were located from a ΔF map and refined subject to constraints on the O-H and H-H distances. Idealised distances were taken from I. Guzei’s fragment library.4 These hydrogen atoms all participate in H-bonding to the O-atoms of the triflate anion, which form an extensive network in the solid state structure of [iPr3Sn·2H2O]OTf.

The absolute structure of [iPr3Sn·2H2O]OTf was determined using the Flack Parameter (0.002(13)).

Crystal data and structure refinement for [iPr3Sn·2H2O]OTf

Formula C10H25F3O5SSn

Formula Weight 433.05

Temperature 173.00(10) K

Diffractometer, Wavelength Agilent Xcalibur 3E, 0.71073 Å

Crystal system, space group Orthorhombic, P212121

Unit cell dimensions

222

Appendices a 10.3086(3) Å b 12.8043(4) Å c 13.4864(4) Å

α 90°

β 90°

γ 90°

Volume, Z 1780.15(9) Å, 4

Density (calculated) 1.616 g/cm3

Absorption coefficient 1.593 mm-1

F(000) 872.0

Crystal colour / morphology Colourless clear needles

Crystal size 0.4607 × 0.1607 × 0.1444 mm3

2θ range for data collection 5.904 to 56.666°

Index Ranges -12 ≤ h ≤ 13, -16 ≤ k ≤ 16, -15 ≤ l ≤ 17

Reflections collected 10254

Independent Reflections 3726 [Rint = 0.0241, Rsigma = 0.0327]

Absorption correction Analytical

Refinement method Full-matrix least-squares on F2

Data / restraints / parameters 3726/6/203

Goodness-of-fit on F2 1.052

Final R indices [I>=2σ (I)] R1 = 0.0229, wR2 = 0.0428

R indices (all data) R1 = 0.0257, wR2 = 0.0439

Largest diff. peak, hole 0.44, -0.32 eÅ-3

Flack Parameter -0.002(13)

223

Appendices

Bond lengths for [iPr3Sn·2H2O]OTf

Atom Atom Length/Å Atom Atom Length/Å Sn(1) O(19) 2.298(2) F(16) C(15) 1.322(4) Sn(1) C(5) 2.155(4) F(17) C(15) 1.310(5) Sn(1) C(8) 2.149(4) C(4) C(2) 1.518(6) Sn(1) C(2) 2.150(4) C(5) C(7) 1.524(6) Sn(1) O(20) 2.322(3) C(5) C(6) 1.525(6) S(11) O(13) 1.446(3) C(15) F(18) 1.321(5) S(11) O(12) 1.435(3) C(8) C(9) 1.524(6) S(11) O(14) 1.431(3) C(8) C(10) 1.508(6) S(11) C(15) 1.814(4) C(2) C(3) 1.520(6)

Bond angles for [iPr3Sn·2H2O]OTf Atom Atom Atom Angle/˚ Atom Atom Atom Angle/˚ O(19) Sn(1) O(20) 174.49(10) C(7) C(5) Sn(1) 110.4(3) C(5) Sn(1) O(19) 87.33(12) C(7) C(5) C(6) 110.9(3) C(5) Sn(1) O(20) 90.60(12) C(6) C(5) Sn(1) 114.1(3) C(8) Sn(1) O(19) 95.37(13) F(16) C(15) S(11) 111.3(3) C(8) Sn(1) C(5) 117.30(15) F(17) C(15) S(11) 111.4(3) C(8) Sn(1) C(2) 118.23(16) F(17) C(15) F(16) 108.5(3) C(8) Sn(1) O(20) 90.13(14) F(17) C(15) F(18) 107.6(4) C(2) Sn(1) O(19) 91.25(13) F(18) C(15) S(11) 109.8(3) C(2) Sn(1) C(5) 124.33(16) F(18) C(15) F(16) 108.1(4) C(2) Sn(1) O(20) 85.69(13) C(9) C(8) Sn(1) 114.4(3) O(13) S(11) C(15) 104.07(17) C(10) C(8) Sn(1) 111.7(3) O(12) S(11) O(13) 113.20(17) C(10) C(8) C(9) 111.9(4) O(12) S(11) C(15) 103.9(2) C(4) C(2) Sn(1) 111.8(3) O(14) S(11) O(13) 114.50(16) C(4) C(2) C(3) 111.4(4) O(14) S(11) O(12) 115.62(19) C(3) C(2) Sn(1) 113.2(3) O(14) S(11) C(15) 103.64(18)

224

Appendices

Crystallographic data for iPr3SnNTf2 ([26b]NTf2)

Single crystals of C11H21F6NO4S2Sn were obtained from a CHCl3 solution layered with pentane. Single crystal X-ray diffraction data was collected with an Oxford Diffraction Xcalibur unit; crystals were mounted on a nylon MicroLoop™ using perfluoropolyether oil and measured in a stream of N2 at 173 K. The structure was solved in Olex21 by charge flipping using Superflip,2 and subsequently refined with the ShelXL3 refinement package.

Crystal data and structure refinement for iPr3SnNTf2

Empirical formula C11H21F6NO4S2Sn

Formula weight 528.10

Temperature 172.95(10) K

Radiation MoKα (λ = 0.71073)

Crystal system, space group Tetragonal, I41/acd

Unit cell dimensions

a 16.7483(4) Å

b 16.7483(4) Å

c 28.9201(7) Å

α 90°

β 90°

γ 90°

Volume, Z 8112.2(4) Å3, 16

Density (calculated) 1.730 g/cm3

Absorption coefficient 1.533 mm-1

225

Appendices

F(000) 4192.0

Crystal size 0.5171 × 0.4833 × 0.2545 mm3

2Θ range for data collection 5.618 to 56.542°

Index ranges -21 ≤ h ≤ 20, -21 ≤ k ≤ 21, -35 ≤ l ≤ 31

Reflections collected 21697

Independent reflections 2359 [Rint = 0.0305, Rsigma = 0.0272]

Data/restraints/parameters 2359/0/132

Goodness-of-fit on F2 1.036

Final R indexes [I>=2σ (I)] R1 = 0.0390, wR2 = 0.0817

Final R indexes [all data] R1 = 0.0686, wR2 = 0.0973

Largest diff. peak, hole 0.41, -0.41 e Å-3

Bond lengths for iPr3SnNTf2 Atom Atom Length/Å Atom Atom Length/Å Sn(1) O(1)1 2.346(3) F(2) C(1) 1.301(6) Sn(1) O(1) 2.346(3) N(3) S(1)2 1.564(2) Sn(1) C(2) 2.137(5) F(4) C(1) 1.301(6) Sn(1) C(2)1 2.137(5) F(1) C(1) 1.325(6) Sn(1) C(7) 2.115(15) C(2) C(3) 1.516(9) S(1) O(0AA) 1.401(4) C(2) C(4) 1.506(10) S(1) N(3) 1.564(2) C(5) C(7) 1.74(4) S(1) C(1) 1.835(6) C(6) C(7) 1.38(2) S(1) O(1) 1.431(4)

11/4-Y,1/4-X,1/4-Z; 21/2-X,+Y,-Z

226

Appendices

Bond angles for iPr3SnNTf2 Atom Atom Atom Angle/˚ Atom Atom Atom Angle/˚ O(1) Sn(1) O(1)1 178.6(2) O(1) S(1) C(1) 100.6(2) C(2)1 Sn(1) O(1) 88.39(17) S(1) N(3) S(1)2 124.1(3) C(2)1 Sn(1) O(1)1 90.99(16) F(2) C(1) S(1) 110.8(4) C(2) Sn(1) O(1)1 88.39(17) F(2) C(1) F(1) 108.8(5) C(2) Sn(1) O(1) 90.99(16) F(4) C(1) S(1) 111.3(4) C(2)1 Sn(1) C(2) 127.2(4) F(4) C(1) F(2) 108.8(5) C(7) Sn(1) O(1) 78.5(6) F(4) C(1) F(1) 109.0(5) C(7) Sn(1) O(1)1 102.9(6) F(1) C(1) S(1) 108.2(4) C(7) Sn(1) C(2) 117.7(11) S(1) O(1) Sn(1) 148.9(2) C(7) Sn(1) C(2)1 113.9(10) C(3) C(2) Sn(1) 110.1(4) O(0AA) S(1) N(3) 117.2(3) C(4) C(2) Sn(1) 111.8(5) O(0AA) S(1) C(1) 107.0(3) C(4) C(2) C(3) 114.6(7) O(0AA) S(1) O(1) 118.1(3) C(5) C(7) Sn(1) 106.4(15) N(3) S(1) C(1) 104.97(18) C(6) C(7) Sn(1) 116.4(13) O(1) S(1) N(3) 107.0(2) C(6) C(7) C(5) 105.7(17)

11/4-Y,1/4-X,1/4-Z; 21/2-X,+Y,-Z

References for Appendix A

1. Dolomanov, O.V., Bourhis, L.J., Gildea, R.J, Howard, J.A.K., and Puschmann, H., Journal of Applied Crystallography, 2009, 42, 339.

2. Palatinus, L., Prathapa, S. J. and van Smaalen, S., Journal of Applied Crystallography, 2012, 45, 575.

3. Sheldrick, G.M. Acta Crystallographica Section C, 2015, 71, 3.

4. Guzei, I. A. Journal of Applied Crystallography 2014, 47, 806.

227

Appendices

Appendix B – Procedure for precise gas transfer through use of a Toepler pump

The Toepler pump is an apparatus that allows for the precise transfer of a known quantity of gas from a supply bulb into a receiver vessel. Because very little gas is lost or wasted during the transfer process, it is particularly convenient for the manipulation of expensive reagent gases, including isotopically-enriched samples such as HD or D2.

A simplified schematic representation of the Toepler set-up is shown in Figure A1. Key components include a high-performance vacuum pump, mercury piston pump, manometer and a pair of valves that permit gas transfer in one direction only.

Figure A1 – Schematic representation of the Toepler pump apparatus used for precise gas transfer.

The procedure for gas transfer begins with attachment of a glass bulb containing the gas of interest, as well as the desired receiving vessel (connected using standard, greased ground-glass joints). The high-performance vacuum pump is then used to thoroughly evacuate the entire volume of the apparatus (with the exception of the supply bulb), including the receiving vessel (any solution in this vessel in frozen in liquid N2 to prevent its evaporation). Having removed any previous atmosphere, the desired quantity of gas is bled into section A of the apparatus, which is connected to the manometer. Because the volume of this section (including the side-arm connecting it to the supply

228

Appendices bulb) has been measured previously, the observed pressure can be used to easily calculate the molar quantity of gas released.

At this point, tap 1 is opened, connecting section A to section B and the mercury piston pump. This pump works by varying the pressure beneath a reservoir of liquid Hg, which raises or lowers its level. As the level rises, the connection between sections A and B is cut off, and the portion of gas trapped in section B is forced through a one-way valve and into the receiver section of the apparatus. When the level is lowered again the connection is re-established, and by repeating this cycle several times the vast majority of the gas that was released from the bulb will be transferred into the receiver section (note that this section consists of both the receiving vessel and the side-arm to which it is attached: if the gas being transferred is not condensable, then this means a portion will be left behind in the side-arm, which must be accounted for when calculating the initial amount of gas to release).

229

Appendices

Appendix C – Related publications

Significant portions of the work described in this thesis have previously been published as academic journal articles, which are reproduced here.

“Metal-Free Hydrogenation Catalyzed by an Air-Stable Borane: Use of Solvent as a Frustrated Lewis Base” by Daniel J Scott, Matthew J. Fuchter and Andrew A. Ashley was published in Angewandte Chemie International Edition, 2014, 53, 10218-10222 (doi: 10.1002/anie.201405531; http://onlinelibrary.wiley.com/doi/10.1002/anie.201405531/abstract), and is reproduced here without modification under the terms of the Creative Commons Attribution License.

Abstract: “In recent years ‘frustrated Lewis pairs’ (FLPs) have been shown to be effective metal-free catalysts for the hydrogenation of many unsaturated substrates. Even so, limited functional-group tolerance restricts the range of solvents in which FLP-mediated reactions can be performed, with all FLP-mediated hydrogenations reported to date carried out in non-donor hydrocarbon or chlorinated solvents. Herein we report that the bulky Lewis acids B(C6Cl5)x(C6F5)3−x (x = 0–3) are capable of heterolytic H2 activation in the strong-donor solvent THF, in the absence of any additional Lewis base. This allows metal-free catalytic hydrogenations to be performed in donor solvent media under mild conditions; these systems are particularly effective for the hydrogenation of weakly basic substrates, including the first examples of metal-free catalytic hydrogenation of furan heterocycles. The air- stability of the most effective borane, B(C6Cl5)(C6F5)2, makes this a practically simple reaction method.”

“Nonmetal Catalyzed Hydrogenation of Carbonyl Compounds” by Daniel J Scott, Matthew J. Fuchter and Andrew A. Ashley was published in Journal of the American Chemical Society, 2014, 136, 15813- 15816 (doi: 10.1021/ja5088979; http://pubs.acs.org/doi/abs/10.1021/ja5088979), and is reproduced here without modification under the terms of the Creative Commons Attribution (CC-BY) License, the relevant details of which can be found online at http://pubs.acs.org/page/policy/authorchoice_ccby_termsofuse.html.

Abstract: “Solutions of the Lewis acid B(C6F5)3 in 1,4-dioxane are found to effectively catalyze the hydrogenation of a variety of ketones and aldehydes. These reactions, the first to allow entirely metal- free catalytic hydrogenation of carbonyl groups under relatively mild reaction conditions, are found to proceed via a “frustrated Lewis pair” mechanism in which the solvent, a weak Brønsted base yet moderately strong donor, plays a pivotal role.”

230

Appendices

“Facile Protocol for Water-Tolerant “Frustrated Lewis Pair”-Catalyzed Hydrogenation” by Daniel J Scott, Trevor R. Simmons, Elliot J. Lawrence, Gregory G. Wildgoose Matthew J. Fuchter and Andrew A. Ashley was published in ACS Catalysis, 2015, 5, 5540-5544 (doi: 10.1021/acscatal.5b01417; http://pubs.acs.org/doi/abs/10.1021/acscatal.5b01417), and is reproduced here without modification under the terms of the Creative Commons Attribution (CC-BY) License, the relevant details of which can be found online at http://pubs.acs.org/page/policy/authorchoice_ccby_termsofuse.html.

Abstract: “Despite rapid advances in the field of metal-free, “frustrated Lewis pair” (FLP)-catalyzed hydrogenation, the need for strictly anhydrous reaction conditions has hampered wide-scale uptake of this methodology. Herein, we report that, despite the generally perceived moisture sensitivity of

FLPs, 1,4-dioxane solutions of B(C6F5)3 actually show appreciable moisture tolerance and can catalyze hydrogenation of a range of weakly basic substrates without the need for rigorously inert conditions. In particular, reactions can be performed directly in commercially available nonanhydrous solvents without subsequent drying or use of internal desiccants.”

“Versatile Catalytic Hydrogenation Using A Simple Tin(IV) Lewis Acid” by Daniel J Scott, Nicholas A. Phillips, Joshua S. Sapsford, Arron C. Deacy, Matthew J. Fuchter and Andrew A. Ashley was published in Angewandte Chemie International Edition, 2016, 55, 14738-14742 (doi: 10.1002/anie.201606639; http://onlinelibrary.wiley.com/doi/10.1002/anie.201606639/abstract), and is reproduced here without modification under the terms of the Creative Commons Attribution License.

Abstract: “Despite the rapid development of frustrated Lewis pair (FLP) chemistry over the last ten years, its application in catalytic hydrogenations remains dependent on a narrow family of structurally similar early main-group Lewis acids (LAs), inevitably placing limitations on reactivity, sensitivity and substrate scope. Herein we describe the FLP-mediated H2 activation and catalytic hydrogenation

+ activity of the alternative LA iPr3SnOTf, which acts as a surrogate for the trialkylstannylium ion iPr3Sn , and is rapidly and easily prepared from simple, inexpensive starting materials. This highly thermally robust LA is found to be competent in the hydrogenation of a number of different unsaturated functional groups (which is unique to date for main-group FLP LAs not based on boron), and also displays a remarkable tolerance to moisture.”

231

Angewandte. Communications

DOI: 10.1002/anie.201405531 H2 Activation Hot Paper Metal-Free Hydrogenation Catalyzed by an Air-Stable Borane: Use of Solvent as a Frustrated Lewis Base** Daniel J. Scott, Matthew J. Fuchter, and Andrew E. Ashley*

Abstract: In recent years frustrated Lewis pairs (FLPs) have been shown to be effective metal-free catalysts for the hydrogenation of many unsaturated substrates. Even so, limited functional-group tolerance restricts the range of solvents in which FLP-mediated reactions can be performed, with all FLP-mediated hydrogenations reported to date carried out in non-donor hydrocarbon or chlorinated solvents. Herein we = report that the bulky Lewis acids B(C6Cl5)x(C6F5)3Àx (x 0–3)

are capable of heterolytic H2 activation in the strong-donor solvent THF, in the absence of any additional Lewis base. This Scheme 1. Some examples of ether CÀO cleavage by FLPs.[4b, c] allows metal-free catalytic hydrogenations to be performed in donor solvent media under mild conditions; these systems are particularly effective for the hydrogenation of weakly basic and derivatives thereof, has significantly limited the use of substrates, including the first examples of metal-free catalytic donor solvents, such as ethers, which tend to form strong hydrogenation of furan heterocycles. The air-stability of the classical donor–acceptor adducts. For many FLPs this coor-

most effective borane, B(C6Cl5)(C6F5)2, makes this a practically dination is followed by nucleophilic cleavage of the activated simple reaction method. CÀO bond (Scheme 1). In particular, ring-opening of THF was one of the first reported FLP-mediated transformations, Since the initial reports into their reactivity by Stephan and as such is often viewed as an archetypal FLP reaction.[4c] et al., frustrated Lewis pairs (FLPs) have attracted great Consequently, only a few explicit reports exist of H2 activation interest for their ability to act as metal-free polar hydro- by FLPs in donor-solvent media, all of which were based on genation catalysts.[1] By rational modification of both the stoichiometric phosphine or amine bases, and none of which Lewis acidic and Lewis basic components, FLPs have been described any subsequent catalytic hydrogenation reactivity.[8] developed that are effective for the reduction of a wide range Recent work has shown that near-stoichiometric mixtures

of unsaturated substrates, both polar (e.g. imines, enol of 1a (Figure 1) and specific ethers (Et2O, crown ethers) are ethers)[2] and non-polar (e.g. 1,1-diphenylethylene).[3] capable of acting as hydrogenation catalysts in non-donor

In addition to H2, FLPs have been shown to readily react solvents, such as CD2Cl2, neatly demonstrating that such

with a wide variety of other functional groups including ethers are not fundamentally incompatible with FLP H2 ethers,[4] carbonyls,[5] and weakly acidic CÀH[6] and NÀH activation chemistry.[9] Meanwhile, Paradies and co-workers [7] bonds. Though impressive, this diverse reactivity has gen- have reported use of the THF adduct of B(2,6-F2C6H3)3 as erally rendered FLPs incompatible with many common a convenient source of the borane for certain P/B and N/B organic solvents. In particular, the ubiquity in FLP chemistry FLP-catalyzed hydrogenations.[10] These results led us to

of very strong, air-sensitive, Lewis acids, such as B(C6F5)3 (1a) speculate that, with an appropriate Lewis acid, not only should FLP-mediated hydrogenation be possible in stronger donor ethereal solvents, but such solvents might remove the [*] D. J. Scott, Dr. M. J. Fuchter, Dr. A. E. Ashley need for an additional “frustrated” Lewis base, by performing Department of Chemistry Imperial College London that role themselves. London, SW7 2AZ (UK) The use of reaction media other than hydrocarbons and E-mail: [email protected] chlorinated solvents is inherently appealing; the low polarity Homepage: http://www3.imperial.ac.uk/people/a.ashley of the hydrocarbons limits their effectiveness at solubilizing e = e = [**] We would like to thank GreenCatEng, Eli Lilly (Pharmacat con- many potential polar substrates ( PhMe 2.38, c.f. THF 7.52, sortium), and the EPSRC for providing funding for a PhD student- e = [11] DCM 8.93), while chlorinated solvents have become ship (D.J.S.), and the Royal Society for a University Research increasingly unattractive as chemists become more concerned Fellowship (A.E.A.). about the greenness of their reactions.[12] Supporting information for this article is available on the WWW Previously, we have investigated the extremely hindered under http://dx.doi.org/10.1002/anie.201405531. boranes B(C Cl ) (C F ) À (x = 1–3, Figure 1) and found that 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. 6 5 x 6 5 3 x although electrophilicity increases with the number of KGaA. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and perchlorophenyl groups, Lewis acidity decreases as a result [13] reproduction in any medium, provided the original work is properly of increasing steric hindrance. Significantly, and unlike 1a, cited. these boranes were also found to demonstrate appreciable

10218 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 10218 –10222 Angewandte Chemie

Figure 1. Boranes 1a–1d, studied for hydrogenation efficacy in THF solvent.

stability to air and moisture. Herein we describe investigations into the behavior of this family of boranes in the donor- solvent THF, and report the ability of such solutions to Scheme 2. a) Reversible H2 activation by B(C6Cl5)(C6F5)2 in THF and effectively catalyze the hydrogenation of even weakly basic b) potential hydride abstraction from THF, which is not observed. substrates, using an operationally simple method that does not require the addition of an auxiliary Lewis base. Although 1a binds strongly to THF, we envisioned that to [1b·H]À are apparent in the 1H NMR spectrum, this can be the strength of this interaction might be reduced by increasing attributed to line broadening as a result of the quadrupolar steric bulk. Rational modification of the Lewis acid has been 10B/11B nuclei, in addition to broadening arising from dynamic shown to lead to improved functional-group tolerance in dihydrogen bonding, which may be expected in the Brønsted [10,14] [18,21] À FLP-catalyzed hydrogenation reactions. Thus B(C6Cl5)- acidic medium. The possibility that [1b·H] is formed [13] (C6F5)2 (1b), though more electrophilic than 1a, is found to instead as a result of hydride abstraction from the solvent can bind the solvent only weakly when dissolved in neat THF. The be discounted based on the observation of the 11B borohy- reversibility of the binding is clear from variable-temperature dride resonance signal as a doublet in both proteo and (VT) NMR analysis of THF solutions of 1b; below 0 8C the deutero THF, as well as the lack of any reaction in the absence 11 d = B NMR shift remains constant at 3.8 ppm, consistent of H2 (Scheme 2b). Conclusive evidence is provided by using d = 11 d = with the four-coordinate 1b·THF adduct (c.f. 3.3 ppm for D2 in place of H2, which replaces the B doublet at [15] À 1a·THF in CD2Cl2). Upon warming, however, the reso- 19.6 ppm with a singlet at the same shift, and a comparable nance signal moves progressively downfield, reaching d = signal in the 2H spectrum diagnostic of [THF-D]+, or a solvate 23.9 ppm at 608C, indicative of a shift in the equilibrium thereof (Figure 2). towards free, uncoordinated 1b (c.f. d = 63.6 ppm for free 1b in PhMe, see Supporting Information). A similar trend is observed in the 19F NMR spectrum over the same temperature range, with the para fluorine resonance signal shifting d = À 8 Dd = d = from 158.0 ppm at 0 C( m,p 7.1 ppm) to À Dd = 8 153.3 ppm ( m,p 10.9 ppm) at 60 C. The increased separation of the meta and para resonances is consistent with a move away from four-coordinate and towards three- Dd = [16] coordinate boron (c.f. m,p 18.3 ppm for 1b in PhMe). Based on these results the 1b/THF system can be considered to be on the borderline between a classical and a frustrated Lewis pair.[17]

THF solutions of B(C6Cl5)2(C6F5)(1c), which is bulkier still, show no sign of coordination at all at room temperature (11B d = 63.5 ppm, c.f. d = 64.1 ppm in PhMe). Only upon cooling to À408C do signals consistent with a THF adduct become apparent in the 19F NMR (see Supporting Informa- 1 2 1b Figure 2. H and H NMR spectra of in [D8]THF under H2, and in tion). We observed no evidence for adduct formation with 11 11 1 proteo THF under D , respectively (inset: B and B{H} spectra at À 8 8 2 B(C6Cl5)3 (1d) in THF between 100 C and 60 C. À258C). Admission of H2 (4 bar) to a THF solution of 1b at room temperature leads to immediate appearance of a resonance d = 1 signal at 11.19 ppm in the H NMR spectrum. Upon Further evidence for H2 activation is provided by THF À 8 1 8 cooling to 25 C a new doublet (singlet in the H-decoupled solutions of B(C6Cl5)3 (1d). After heating to 60 C for 1h d = À spectrum) can also be resolved at 19.6 ppm in the under H2 (4 bar), new resonance signals can clearly be 11B NMR spectrum (J = 90 Hz). The 11B NMR data is con- observed at d = 11.34 ppm and d = À8.7 ppm (d, J = sistent with previous reports of the borohydride anion 91 Hz)[8c] in the room temperature 1H and 11B NMR spectra, [1b·H]À ,[18] while the new 1H NMR resonance lies within respectively. [19] the range reported for protonated THF. These results are Clearly H2 activation in this manner generates a substan-

therefore consistent with reversible H2 activation by an FLP- tially acidic proton (the pKa of protonated THF has been À [22] type mechanism, with THF acting as the Lewis base measured as 2.05 in aqueous H2SO4). Strong Brønsted (Scheme 2a).[20] Although no resonance signals attributable acids can initiate polymerization of THF,[19b,c] as can strong

Angew. Chem. Int. Ed. 2014, 53, 10218 –10222 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org 10219 Angewandte. Communications

Lewis acids, including 1a.[23] Nevertheless, during the course reduced (Table 1, entries 6 and 7), as was the less bulky N-aryl of our studies no evidence for borane or proton-catalyzed imine 2e, although in this final case slightly higher catalyst polymerization of THF was detected for solutions of 1a–d loadings were necessary to achieve complete conversion, [24] under H2, even after prolonged heating. Nor, during our owing to reversible binding of 1b to the product 3e (Table 1, subsequent investigations into catalytic hydrogenation, was entry 8). any FLP-mediated ring-opening of the solvent observed, even Notably, when the hydrogenation experiments were

in the presence of relatively basic imines. repeated in a non-basic solvent (C7D8) rather than in

1ahas been shown to catalyze the hydrogenation of bulky [D8]THF, under otherwise identical conditions, the weakly imines in PhMe through a FLP mechanism.[25] However, since basic substrates 2a and 2b showed no evidence of hydro- the reaction relies on the substrate to act as the frustrated genation (Table 1, entries 9 and 10). Conversely, the relatively

Lewis base for initial H2 activation, it works relatively poorly basic imines 2d and 2e both show appreciable conversions in

for less electron-rich, and hence less basic, imines. The bulky C7D8 (Table 1, entries 12 and 13). This divergent reactivity is electron-deficient N-tosyl imine 2a, for example, was consistent with hydrogenation occurring by two distinct

reported to require forcing conditions, in particular high H2 mechanisms. In the first, H2 activation by 1b/THF is followed pressures, to achieve appreciable conversion (Table 1, by sequential proton and hydride transfer to generate the entries 1 and 2). product amine (Scheme 3, route a). In the second mechanism, In contrast, the same imine was rapidly reduced in the

presence of 1b in [D8]THF under much milder conditions 8 (5 mol% 1b,60 C, 4 bar H2, 3 h), as was the related substrate 2b (Table 1, entries 3 and 4). Furthermore, the air-stability of 1b meant the initial reaction mixture could be conveniently prepared under air using pre-dried solvent, without the need for use of a glovebox (Table 1, entry 5). In addition to 2a and 2b the bulky N-aryl imines 2c and 2d were also successfully

Table 1: FLP-mediated hydrogenation of imines.

Scheme 3. Proposed mechanisms for hydrogenation of imines by activation of H using either a) THF solvent or b) substrate as T 8 t [a] 2 Entry Substrate Solvent [ C] [B] (mol%) Yield [%] a frustrated Lewis base. [h] [b,c] 2a 1a 1 C7H8 80 (10) 22 7 [b,d] 2a 1a H2 is activated instead by a 1b/substrate FLP in the manner 2 C7H8 80 (10) 22 99 2a 1b > [e] described by Stephan et al., with subsequent transfer of 3 [D8]THF 60 (5) 3 99 (98) 2b 1b > [25b] 4 [D8]THF 60 (5) 3 99 hydride to the protonated imine (Scheme 3, route b). The 2a 1b > [f] 5 THF 60 (5) 3 99 reduction of 2d and 2e in non-donor solvent (C7D8) clearly 2c 1b > [e] 6 [D8]THF 60 (5) 8 99 (99) demonstrates the feasibility of the route b mechanism. By 2d 1b 7 [D8]THF 80 (5) 18 71 contrast the lack of reactivity for the more weakly basic 2e 1b 8 [D8]THF 60 (15) 8 91 2a 1b substrates 2a and 2b in C7D8, suggests that their reduction in 9 C7D8 60 (5) 3 0 2b 1b THF occurs solely by solvent-mediated hydrogen activation. 10 C7D8 60 (5) 3 0 2c 1b The different reactivity is consistent with other observations 11 C7D8 60 (5) 8 0 2d 1b 12 C7D8 80 (5) 18 79 and can be understood intuitively: H2 activation using the 2e 1b 13 C7D8 60 (15) 8 26 substrate as the frustrated Lewis base will become less 14 2a Dioxane 60 1b (5) 41 96 favorable as the substrate becomes less basic. However, the 2a 1c 15 [D8]THF 60 (5) 72 90 high Brønsted acidity of protonated THF allows for levelling 16 2a [D ]THF 80 1a (10) 72 84 8 even to relatively electron-poor substrates. Interestingly, 2c 17 2a [D ]THF 80 1d (5) 72 0 8 also fails to undergo hydrogenation in C D , despite being of 1 7 8 [a] Yields measured by in situ H NMR spectroscopy, using 1,3,5- similar basicity to 2e (Table 1, entry 11). In this case steric trimethoxybenzene in C6D6 in a capillary insert as an internal integration [25a] shielding of the basic nitrogen atom presumably inhibits standard. [b] Result reported by Klankermayer and Chen. [c] 10 bar H2. direct H2 activation. [d] 30 bar H2. [e] Number in parentheses is yield isolated after increasing to 1 mmol scale (see Supporting Information). [f] Initial reaction mixture The hydrogenation mechanism (route a), where H2 acti- prepared using pre-dried solvent under air (see Supporting Information). vation is mediated by the Lewis acid and the solvent, is also

10220 www.angewandte.org 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 10218 –10222 Angewandte Chemie feasible for other ethereal solvents. Solutions of 1b in 1,4- it should be noted that the reduction of the pyrroles 4 to the dioxane catalyze the hydrogenation of 2d under identical corresponding pyrrolidines, 5, does require the use of two conditions to solutions in [D8]THF, albeit more slowly equivalents of H2). Similar limitations have been reported for (Table 1, entry 14). The lower rate is consistent with the the FLP-mediated hydrogenation of anilines to much more = À [27] lower basicity of 1,4-dioxane (pKaH 2.92 in aqueous basic cyclohexylamines. [22,26] H2SO4), but may also partially be attributed to its It was anticipated that the use of furans instead of pyrroles e = e = reduced polarity relative to THF ( dioxane 2.22, THF might lead to superior results; the substituted tetrahydrofuran [11] + À 7.52), which will make cleavage of H2 into ionic H /H products ought to be no more basic than the solvent, and so adducts less favorable (Scheme 3, route a). Some variation of should not prevent catalytic turnover. Indeed, although the borane is also tolerated: use of 1c leads to a reduction in attempts to hydrogenate furan itself were unsuccessful, reaction rate, but otherwise only a minor change in outcome several more electron-rich methyl-substituted furans, 6, did (Table 1, entry 15). In fact, even 1a is observed to effectively undergo catalytic hydrogenation (Scheme 4), despite the fact catalyze hydrogenation at slightly higher temperatures that such compounds are extremely weak bases.[28] This (Table 1, entry 16); clearly under these conditions, coordina- represents the first reported example of FLP-catalyzed tion of THF is sufficiently reversible to allow some H2 hydrogenation of aromatic O-heterocyclic rings, and nicely activation to occur. No reaction is observed with 1d, demonstrates the value of the borane/solvent systems de- suggesting [1d·H]À to be a much poorer hydride donor. scribed. In addition to these novel results, attempts to reduce Given that 11B NMR spectroscopic analysis suggests the compounds from a variety of previously-studied substrate À equilibrium between 1d and [1d·H] under H2 favors 1d, classes were also successful, under similar conditions (Sche- this lack of reactivity is most likely due to kinetic (steric) me 4).[1b,c] rather than thermodynamic factors (Table 1, entry 17). In conclusion, we have shown that THF solutions of

Given the success of 1b as a hydrogenation catalyst for boranes 1 are capable of effecting H2 activation in the absence electron-poor imines we were interested in its ability to effect of any additional Lewis base. Solutions of 1bin particular are hydrogenation of other weakly basic substrates. To date the effective catalysts for the metal-free hydrogenation of a vari- only reported example of FLP-mediated hydrogenation of ety of substrates by a solvent-assisted mechanism. Compound a weakly basic aromatic heterocycle describes the reduction 1b shows appreciable stability in air, which further increases [2] of indoles under very high pressures of H2. Nevertheless, the practicality of this system relative to the 1a-derived admission of just 5 bar H2 to a mixture of 1b and N-methyl alternatives. pyrrole (4a) or 2,5-dimethylpyrrole (4b) in THF led to formation of the reduced species [5·H]+[1b·H]À (Scheme 4). Received: May 22, 2014 No catalytic turnover was observed due to the relatively low Revised: July 7, 2014 acidity of the pyrrolidinium borohydride products (although Published online: August 11, 2014 .Keywords: boranes · frustrated Lewis pairs · heterocycles · hydrogenation · solvent effects

[1] a) G. C. Welch, R. R. S. Juan, J. D. Masuda, D. W. Stephan, Science 2006, 314, 1124 – 1126; b) J. Paradies, Synlett 2013, 777 – 780; c) L. J. Hounjet, D. W. Stephan, Org. Process Res. Dev. 2014, 18, 385 – 391; Also relevant to this field is earlier work on

B(C6F5)3-catalyzed hydrosilylation. See: d) D. J. Parks, W. E. Piers, J. Am. Chem. Soc. 1996, 118, 9440 – 9441; e) D. J. Parks, J. M. Blackwell, W. E. Piers, J. Org. Chem. 2000, 65, 3090 – 3098; f) J. M. Blackwell, E. R. Sonmor, T. Scoccitti, W. E. Piers, Org. Lett. 2000, 2, 3921 – 3923; g) W. E. Piers, A. J. V. Marwitz, L. G. Mercier, Inorg. Chem. 2011, 50, 12252 – 12262. [2] D. W. Stephan, S. Greenberg, T. W. Graham, P. Chase, J. J. Hastie, S. J. Geier, J. M. Farrell, C. C. Brown, Z. M. Heiden, G. C. Welch, M. Ullrich, Inorg. Chem. 2011, 50, 12338 – 12348. [3] a) L. Greb, P. Ona-Burgos, B. Schirmer, S. Grimme, D. W. Stephan, J. Paradies, Angew. Chem. 2012, 124, 10311 – 10315; Angew. Chem. Int. Ed. 2012, 51, 10164 – 10168; b) Y. Segawa, D. W. Stephan, Chem. Commun. 2012, 48, 11963 – 11965. [4] a) B. Birkmann, T. Voss, S. J. Geier, M. Ullrich, G. Kehr, G. Erker, D. W. Stephan, Organometallics 2010, 29, 5310 – 5319; b) A. M. Chapman, M. F. Haddow, D. F. Wass, J. Am. Chem. Soc. 2011, 133, 18463 – 18478; c) G. C. Welch, J. D. Masuda, D. W. Stephan, Inorg. Chem. 2006, 45, 478 – 480; d) D. Holschumacher, T. Bannenberg, C. G. Hrib, P. G. Jones, M. Tamm, Angew. Chem.

Scheme 4. B(C6Cl5)(C6F5)2-mediated hydrogenations performed in Int. Ed. 2008, 47, 7428 – 7432; Angew. Chem. 2008, 120, 7538 –

[D8]THF. 7542.

Angew. Chem. Int. Ed. 2014, 53, 10218 –10222 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org 10221 Angewandte. Communications

[5] a) C. M. Mçmming, S. Froemel, G. Kehr, R. Froehlich, S. [16] a) A. G. Massey, A. J. Park, J. Organomet. Chem. 1966, 5, 218 – Grimme, G. Erker, J. Am. Chem. Soc. 2009, 131, 12280 – 12289; 225; b) A. D. Horton, J. de With, A. J. van der Linden, H. b) C. M. Mçmming, G. Kehr, B. Wibbeling, R. Froehlich, G. van de Weg, Organometallics 1996, 15, 2672 – 2674; c) A. D. Erker, Dalton Trans. 2010, 39, 7556 – 7564; c) S. Moebs-Sanchez, Horton, J. de With, Chem. Commun. 1996, 1375 – 1376. G. Bouhadir, N. Saffon, L. Maron, D. Bourissou, Chem. [17] Because the limiting 19For11B resonance signals of free 1b in Commun. 2008, 29, 3435 – 3437; d) W. Uhl, C. Appelt, Organo- THF are not known, it is unfortunately not possible to extract metallics 2013, 32, 5008 – 5014. thermodynamic activation parameters for the reversible binding [6] a) S. D. Tran, T. A. Tronic, W. Kaminsky, D. M. Heinekey, J. M. of THF to 1b from these spectra. Mayer, Inorg. Chim. Acta 2011, 369, 126 – 132; b) D. Chakra- [18] H. Zaher, A. E. Ashley, M. Irwin, A. L. Thompson, M. J. borty, E. Y. X. Chen, Macromolecules 2002, 35, 13 – 15. Gutmann, T. Kramer, D. OHare, Chem. Commun. 2013, 49, [7] a) P. A. Chase, D. W. Stephan, Angew. Chem. Int. Ed. 2008, 47, 9755 – 9757. 7433 – 7437; Angew. Chem. 2008, 120, 7543 – 7547. [19] a) G. A. Olah, P. J. Szilagyi, J. Org. Chem. 1971, 36, 1121 – 1126; [8] a) T. J. Herrington, A. J. W. Thom, A. J. P. White, A. E. Ashley, b) G. Pruckmayr, T. K. Wu, Macromolecules 1978, 11, 662 – 668; Dalton Trans. 2012, 41, 9019 – 9022; b) Z. Lu, Z. Cheng, Z. Chen, c) G. Pruckmayr, T. K. Wu, Macromolecules 1973, 6, 33 – 38. L. Weng, Z. H. Li, H. Wang, Angew. Chem. Int. Ed. 2011, 50, [20] Although the number of THF molcules coordinated to the 12227 – 12231; Angew. Chem. 2011, 123, 12435 – 12439; c) A. L. proton has not been determined, a coordination number of two Travis, S. C. Binding, H. Zaher, T. A. Q. Arnold, J. C. Buffet, D. would be consistent with previous observations.[9] See also: I. OHare, Dalton Trans. 2013, 42, 2431 – 2437. Krossing, A. Reisinger, Eur. J. Inorg. Chem. 2005, 1979 – 1989, [9] L. J. Hounjet, C. Bannwarth, C. N. Garon, C. B. Caputo, S. and references therein. Grimme, D. W. Stephan, Angew. Chem. Int. Ed. 2013, 52, 7492 – [21] F. Schulz, V. Sumerin, S. Heikkinen, B. Pedersen, C. Wang, M. 7495; Angew. Chem. 2013, 125, 7640 – 7643. Atsumi, M. Leskel, T. Repo, P. Pyykkç, W. Petry, B. Rieger, J. [10] L. Greb, C. G. Daniliuc, K. Bergander, J. Paradies, Angew. Am. Chem. Soc. 2011, 133, 20245 – 20257. Chem. Int. Ed. 2013, 52, 5876 – 5879; Angew. Chem. 2013, 125, [22] E. Arnett, C. Y. Wu, J. Am. Chem. Soc. 1960, 82, 4999 – 5000. 5989 – 5992. [23] T. Chivers, G. Schatte, Eur. J. Inorg. Chem. 2003, 3314 – 3317.

[11] Handbook of Chemistry and Physics (Ed.: W. M. Haynes), [24] In fact, it appears that the presence of an atmosphere of H2 94edth edCRC, Boca Raton, 2013. inhibits polymerization of THF by 1a (see Supporting Informa- [12] K. Alfonsi, J. Colberg, P. J. Dunn, T. Fevig, S. Jennings, T. A. tion). Johnson, H. P. Kleine, C. Knight, M. A. Nagy, D. A. Perry, M. [25] a) D. Chen, J. Klankermayer, Chem. Commun. 2008, 2130 – Stefaniak, Green Chem. 2008, 10, 31 – 36. 2131; b) P. A. Chase, T. Jurca, D. W. Stephan, Chem. Commun. [13] A. E. Ashley, T. J. Herrington, G. G. Wildgoose, H. Zaher, A. L. 2008, 1701 – 1703.

Thompson, N. H. Rees, T. Kraemer, D. OHare, J. Am. Chem. [26] pKa differences of this magnitude have been shown to signifi- Soc. 2011, 133, 14727 – 14740. cantly affect the rate of alkene hydrogenation by FLP catalysts [14] a) G. Ero˝s, H. Mehdi, I. Ppai, T. A. Rokob, P. Kirly, G. based on weakly basic phosphines. See: L. Greb, S. Tussing, B. Trknyi, T. Sos, Angew. Chem. Int. Ed. 2010, 49, 6559 – 6563; Schirmer, P. OÇa-Burgos, K. Kaupmees, M. Lkov, I. Leito, S. Angew. Chem. 2010, 122, 6709 – 6713; b) G. Ero˝ s, K. Nagy, H. Grimme, J. Paradies, Chem. Sci. 2013, 4, 2788 – 2796. Mehdi, I. Ppai, P. Nagy, P. Kirly, G. Trknyi, T. Sos, Chem. [27] T. Mahdi, Z. M. Heiden, S. Grimme, D. W. Stephan, J. Am. Eur. J. 2012, 18, 574 – 585. Chem. Soc. 2012, 134, 4088 – 4091. [15] C. Lorber, R. Choukroun, L. Vendier, Organometallics 2008, 27, [28] M. P. Carmody, M. J. Cook, R. D. Tack, Tetrahedron 1976, 32, 5017 – 5024. 1767 – 1771.

10222 www.angewandte.org 2014 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2014, 53, 10218 –10222 This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited.

Communication

pubs.acs.org/JACS

Nonmetal Catalyzed Hydrogenation of Carbonyl Compounds Daniel J. Scott, Matthew J. Fuchter, and Andrew E. Ashley* Department of Chemistry, Imperial College London, London SW7 2AZ, U.K.

*S Supporting Information

Scheme 1. Proposed Mechanism of B(C6F5)3-Catalyzed ABSTRACT: Solutions of the Lewis acid B(C6F5)3 in 1,4- Hydrogenation from Theoretical Studies by Nyhleń and dioxane are found to effectively catalyze the hydrogenation 8 Privalov, Involving Direct H2 Activation by the Substrate of a variety of ketones and aldehydes. These reactions, the first to allow entirely metal-free catalytic hydrogenation of carbonyl groups under relatively mild reaction conditions, are found to proceed via a “frustrated Lewis pair” mechanism in which the solvent, a weak Brønsted base yet moderately strong donor, plays a pivotal role. Scheme 2. Example of Stoichiometric FLP-Mediated Carbonyl Hydrogenation16 atalytic hydrogenations represent one of the most C important families of all chemical transformations and are routinely employed at all scales of chemical production.1 The catalysts that facilitate these reactions are predominantly based on rare, expensive, and often toxic transition metals (TMs); consequently there exists a strong incentive for chemists to develop new catalysts based on more abundant and benign elements, which mimic the reactivity of existing systems. In recent years this has led researchers to investigate potential have been observed upon reaction with prehydrogenated FLP catalysts based on inexpensive and readily available TMs such as systems, with a representative example shown in Scheme 2 − iron and systems that consist solely of main group elements. 2.4,16 18 No turnover is observed, which has been attributed to In the latter category the most notable successes have been the strength of the B−O bonding interaction.19 This could also “ ” 3,4 achieved using frustrated Lewis pairs (FLPs). Rational be explained by the weak conjugate acids of the phosphine and design of these systems, in which H2 is activated in a cooperative amine Brønsted bases employed, which prevents protonation of manner by Lewis acidic and Lewis basic moieties, has led to the the reduced alkoxyborate product; this ensures the organic group ff development of metal-free compounds capable of e ecting exists as a potent Lewis basic alkoxide moiety (strongly bound to catalytic hydrogenation of many unsaturated organic substrates the borane) rather than the less basic alcohol,20,21 and hence including imines, enamines, aziridines, enol ethers, alkenes, and 5,6 leads to strong product inhibition. Indeed, hydroxylic substrates aromatics. Nevertheless, one important class of functional have been shown to be activated via coordination to 1a group remains conspicuous by its absence from this list: the C generating strong acids of comparable pKa to HCl (8.4 in O bond in organic carbonyl compounds. In fact, Wei and Du MeCN).22 Clearly protonation of an alkoxyborate in an FLP- ‘ have very recently stated that the direct hydrogenation of ketones to mediated catalytic hydrogenation cycle would require a very secondary alcohols under FLP catalysis still remains as an unsolved weak Lewis base as one component. 7 problem’. We have recently reported that THF and 1,4-dioxane solutions In 2009, Nyhleń and Privalov reported the results of a − of the boranes B(C6Cl5)x(C6F5)3−x (x =0 3) are capable of theoretical study into possible B(C6F5)3 (1a) catalyzed hydro- reversibly cleaving H to generate the related borohydride 8 2 genation of simple aldehydes and ketones, suggesting that an anions, in addition to strongly Brønsted acidic solvated protons FLP mechanism analogous to that for the related hydrogenation (pK < −2.5 in H O;23 Scheme 3).24 Although the equilibrium 9−13 a 2 of imines (and hydrosilylation of carbonyl compounds) greatly disfavors the hydrogen activation products, these stable ought to be kinetically accessible (Scheme 1). Nevertheless, attempts to realize this prediction experimentally have thus far Scheme 3. Metal-Free Hydrogenation of an Enolizable been unsuccessful. Repo et al. have reported that the 1a- Ketone Catalyzed by B(C6Cl5)(C6F5)2 mediated hydrogenations of benzaldehyde and benzophenone proceed only substoichiometrically in the noncoordinating 14 solvents d8-toluene and CD2Cl2, due to rapid decomposition of the borane. More recently, Stephan et al. have reported similar results using aliphatic ketones.15 Other attempts at FLP-catalyzed carbonyl hydrogenation have Received: August 28, 2014 also been unsuccessful. A number of stoichiometic reductions Published: October 21, 2014

© 2014 American Chemical Society 15813 dx.doi.org/10.1021/ja5088979 | J. Am. Chem. Soc. 2014, 136, 15813−15816 Journal of the American Chemical Society Communication systems were found to be effective at catalyzing the hydro- hence increased Lewis acidity of this borane, leads to stronger genation of weakly basic substrates, including electron-poor coordination to the solvent and hence a need for relatively higher imines, which are electronically very similar to organic carbonyls. reaction times and temperatures. In order to circumvent this Given the electronic similarities, we reasoned that these might problem, THF was replaced with 1,4-dioxane,29 which is a also be good systems to investigate for CO hydrogenation. In weaker donor (and which has also previously been shown to be a 24 particular, these systems have already proven capable of viable component for borane/solvent H2 activation). generating powerful Brønsted acids without suffering from Gratifyingly, the 1a/1,4-dioxane system demonstrated sig- borane decomposition. Furthermore, we anticipated that a donor nificantly improved turnover (albeit at the cost of increased solvent might competitively bind the Lewis acid, thereby aiding reaction times; Table 1, entry 1). Also, iPrOH was produced as product dissociation and facilitating catalytic turnover. the only product, with no evidence for formation of iPr O.30 fi 2 Our previous work identi ed B(C6Cl5)(C6F5)2 (1b) in THF It must be noted that the significantly improved reactivity of 1a as the most catalytically competent system, and consequently relative to 1b in this 1,4-dioxane-based system stands in contrast this was selected for initial investigation. Gratifyingly, admission to the results of our previous investigations using THF, where 1a of H2 to a THF solution of acetone with 1b (10 mol %) led to catalytic consumption of the starting material under mild Table 1. B(C F ) -Catalyzed Hydrogenation of Aldehydes ° 6 5 3 conditions (4 bar of H2,65 C; Scheme 3). To the best of our and Ketones knowledge this is the first example of TM-free catalytic hydrogenation of an enolizable ketone or of any organic carbonyl under such mild conditions.25,26 Yet, although technically catalytic, the reaction proceeds with limited turnover, and the conversion is not significantly improved with increased reaction times. This limited turnover may partially be attributed to inhibition by the product, whereby the alcohol reversibly binds to the highly electrophilic Lewis acid; this is supported by the observation that addition of stoichiometric iPrOH (relative to 1b) at the start of the reaction leads to a significant decrease in conversion (ca. 20%). Analysis of the initial 11B NMR spectrum for this reaction shows a slight upfield shift for the borane resonance from 8.2 ppm in the absence of iPrOH to 7.1 ppm, which also indicates some interaction. The 1H NMR spectrum of the reaction mixture showed the formation of an additional set of isopropyl methine 27 resonances, consistent with formation of iPr2O, which presumably results from acid-catalyzed condensation of iPrOH ff and must result in formation of H2O. To examine the e ect that H2O may have on the catalysis, 1 equiv of H2O relative to 1b was added at the start of the reaction, which led to a complete loss of hydrogenation reactivity, demonstrating that its formation has a ff potent inhibitory e ect. Although coordination of H2Oto1b is 28 · known to be reversible in toluene, the related adduct 1a OH2 can form a variety of H-bonding interactions in the presence of hydroxylic species such as H2O and simple alcohols, which stabilize it significantly;22 it is likely that in the reaction mixture · the 1b OH2 adduct is stabilized to a similar extent. Initially we speculated that the specific problem of product inhibition might be resolved by slightly increasing the steric bulk of the borane catalyst; however, when 1b was replaced with the larger borane B(C6Cl5)2(C6F5)(1c) minimal reduction was observed (<5% after 60 h at 80 °C). This reduced reactivity relative to 1b is consistent with previous observations regarding the reduction of imines (though in this case the difference is far more pronounced)24 and is attributed to the increased steric bulk of the [1c·H]− reducing agent, which prevents close approach of the substrate. Based on this analysis we reasoned the reverse strategy might be more effective and that reducing the bulk of the borane catalyst might lead to generation of a less hindered, and hence kinetically more reactive borohydride intermediate. Our further investigations therefore focused on commercially aReactions typically performed on 0.1 mmol scale in 0.4 mL of solvent available 1a, which has been studied extensively for its use in under 5 bar of H . bAll conversions measured by 1H NMR integration metal-free hydrogenation chemistry.19 2 (capillary insert containing either 1,3,5-trimethoxybenzene or PPh3 in c Our previous investigations had demonstrated that 1a in THF C6D6 typically used as internal standard; see SI). 0.2 mmol of ff d − e f is capable of e ecting catalytic hydrogenation in the same substrate. 12 13 bar of H2. With respect to iPrCOMe. 1 mmol manner as 1b and 1c. However, the reduced steric bulk, and scale. gIsolated yield.

15814 dx.doi.org/10.1021/ja5088979 | J. Am. Chem. Soc. 2014, 136, 15813−15816 Journal of the American Chemical Society Communication

24 was found to give inferior results. This apparent discrepancy Scheme 4. Proposed Mechanism of B(C6F5)3-Catalyzed can be explained by considering the different basicities of the two Hydrogenation of Ketones (a), and Possible Alternative solvents: in THF very little uncoordinated 1a is ever present, Mechanisms for Hydrogenation of Aldehydes (b, c) even at elevated temperatures, and so the extent of H2 activation is low and hydrogenation occurs only very slowly. In the weaker donor 1,4-dioxane, more free 1a can be formed, and hence hydrogenation occurs more readily (by contrast, 1b, which is sterically more demanding, dissociates appreciably in either solvent as shown by VT 11B NMR; see Supporting Information (SI)). The weaker binding of 1,4-dioxane relative to THF is demonstrated by variable temperature NMR studies, which show a large downfield shift for the 11B resonance of a stoichiometric 1a/1,4-dioxane mixture in C7D8 at higher temperatures, and only a much smaller shift for 1a/THF (see SI). Even so, the absolute degree of dissociation for 1a in neat 1,4-dioxane must still be low; for this system no significant perturbation in the 11B and 19F NMR resonances is observed at elevated temperatures (up to 100 °C).24 Indeed, this low degree of dissociation likely explains the reduced initial rate of this hydrogenation reaction relative to the 1b/THF system. Other simple aliphatic ketones of moderate steric bulk were hydrogenated effectively under identical conditions (Table 1, entries 3 and 4). More hindered substrates were not reduced, in line with the steric arguments outlined earlier (Table 1, entries 5 and 7). These observations are qualitatively consistent with theoretical calculations (vide supra), which predicted a much larger Gibbs free energy barrier for hydride transfer from [1a· H]− to more sterically hindered ketones,8 and can be exploited to allow selective reduction of a smaller ketone in the presence of a more hindered substrate (Table 1, entry 6). The system can also be applied to aromatic ketones, subject to similar steric limitations (Table 1, entries 9, 10, and 12), and aromatic aldehydes (Table 1, entries 13−16), with a range of electron- poor carbonyl compounds reduced in good to excellent yields. Reduction of ortho-substituted aldehydes was particularly mixture contains only very small amounts of [1a·H]−,32 effective; presumably the increased steric bulk facilitates reduction by this mechanism does not seem rapid enough to dissociation of the primary alcohol. Reactions were less clean account for the rate of the catalytic hydrogenation. Furthermore, for more electron-rich aromatic substrates. For example, significant decomposition is observed during the reaction of · reduction of acetophenone (Table 1, entry 17) is followed by acetone with 1a/[nBu4N][1a H], most likely via C6F5 group dehydration, with the resultant H2O limiting the observed transfer to 1a, as indicated by the observation of species such as − − 11 turnover, as previously seen for 1b/THF. B(C6F5)4 by mass spectrometry (ES ) and B NMR; it should Although many of the above hydrogenations require be noted that these species are not observed in the catalytic significant reactions times to achieve good conversion under 5 reaction. Similar results are obtained using the aromatic substrate bar of H2, it should be noted that these can be shortened 4′-nitroacetophenone in place of acetone; no reaction is · ° substantially by increasing the partial pressure of H2. Even a observed with [nBu4N][1a H] after heating to 80 C in 1,4- relatively modest increase can lead to dramatically improved dioxane for 16 h, and only slow reduction is observed when 1a is reaction rates (Table 1, entries 2 and 11). Higher pressures also also added (although no borane decomposition is observed in allow reduction of some substrates that are not transformed this case; see SI). Collectively, these results suggest that ketone under milder conditions (Table 1, entry 8). activation occurs not by Lewis acid catalysis and that instead the By analogy with our previous work, we propose that ketone reaction proceeds via Brønsted acid activation of the substrate.33 hydrogenation occurs via a mechanism in which the carbonyl Note, however, that these observations could also be consistent substrate is activated by coordination to a solvated proton with the mechanism proposed by Nyhleń and Privalov, which 31 (generated by activation of H2 by 1a/1,4-dioxane) prior to differs only in the means of generation of the activated carbonyl hydride transfer (Scheme 4a, solvent-assisted pathway); intermediate I (see Schemes 1 and 4a, direct activation pathway). subsequent displacement by the solvent facilitates dissociation Slightly different results are observed when ketones are of the alcohol product. This proposal is supported by some replaced with aldehydes in the above experiments. Addition of · · preliminary mechanistic studies. [nBu4N][1a H] is not observed [nBu4N][1a H] to a mixture of 1a and 4-nitrobenzaldehyde or to reduce acetone even after several hours at 100 °C in 1,4- 2,6-dichlorobenzaldehyde in 1,4-dioxane at rt leads to immediate dioxane, indicating a need for O-activation of the carbonyl. reduction of the carbonyl, as evidenced by the disappearance of Addition of 1a to this reaction mixture does lead to some the resonances attributed to the aldehyde in the 1H NMR reduction, suggesting that sufficient activation can occur via spectrum, concomitant with the appearance of sharp resonances coordination of 1a. However, in this reaction <0.5 equiv of at ca. −2.5 ppm in the 11B NMR spectra, consistent with acetone is consumed after 1 h; given that the catalytic reaction formation of the 1a·alkoxide adducts;34 this suggests an

15815 dx.doi.org/10.1021/ja5088979 | J. Am. Chem. Soc. 2014, 136, 15813−15816 Journal of the American Chemical Society Communication alternative, Lewis acid catalyzed reaction pathway may also be (17) Ashley, A. E.; Thompson, A. L.; O’Hare, D. Angew. Chem., Int. Ed. feasible for these substrates (Scheme 4b). In fact, similar 2009, 48, 9839. reactivity is observed for these aldehydes even without addition (18) Sumerin, V.; Schulz, F.; Nieger, M.; Leskela,̈ M.; Repo, T.; Rieger, of 1a,35 suggesting that their reductions may even proceed B. Angew. Chem., Int. Ed. 2008, 47, 6001. without any prior activation of the carbonyl (Scheme 4c). (19) Stephan, D. W.; Greenberg, S.; Graham, T. W.; Chase, P.; Hastie, J. J.; Geier, S. J.; Farrell, J. M.; Brown, C. C.; Heiden, Z. M.; Welch, G. C.; Nevertheless, Brønsted or Lewis acid catalysis cannot be ruled fi Ullrich, M. Inorg. Chem. 2011, 50, 12338. out, and further studies are needed to con rm the validity and (20) Li, H.; Zhao, L.; Lu, G.; Huang, F.; Wang, Z. X. Dalton Trans. generality of our proposed mechanisms. 2010, 39, 5519. In conclusion, we have developed a protocol for TM-free, (21) Zhao, L.; Lu, G.; Huang, F.; Wang, Z. X. Dalton Trans. 2012, 41, FLP-mediated catalytic hydrogenation of aliphatic and aromatic 4674. ketones and aldehydes to their respective alcohols. Preliminary (22) Bergquist, C.; Bridgewater, B. M.; Harlan, C. J.; Norton, J. R.; mechanistic studies suggest that ketone reduction likely occurs Friesner, R. A.; Parkin, G. J. Am. Chem. Soc. 2000, 122, 10581. via Brønsted acid activation of the substrate followed by hydride (23) Arnett, E.; Wu, C. Y. J. Am. Chem. Soc. 1960, 82, 4999. (24) Scott, D. J.; Fuchter, M. J.; Ashley, A. E. Angew. Chem., Int. Ed. transfer, but that alternative mechanisms may be feasible for − more electrophilic aldehyde substrates. We anticipate that, with 2014, 53, 10218 10222. (25) Walling, C.; Bollyky, L. J. Am. Chem. Soc. 1964, 86, 3750. further rational variation of both the solvent and borane catalyst, (26) Berkessel, A.; Schubert, T. J. S.; Müller, T. N. J. Am. Chem. Soc. hydrogenation of more challenging carbonyl substrates should be 2002, 124, 8693. fi possible using systems of this type. Investigations in this area are (27) Con rmed by independent addition of iPrOH and iPr2Oto1b in ongoing, and will be reported in due course. d8-THF in the appropriate molar ratios. (28) Ashley, A. E.; Herrington, T. J.; Wildgoose, G. G.; Zaher, H.; ■ ASSOCIATED CONTENT Thompson, A. L.; Rees, N. H.; Kraemer, T.; O’Hare, D. J. Am. Chem. Soc. 2011, 133, 14727. *S Supporting Information (29) At no point during the course of our studies was any evidence for Supplementary information includes full experimental details ring-opening or polymerization of the solvent observed. and spectroscopic characterization of products. This material is (30) The different outcome in 1,4-dioxane may simply be attributable ε ε available free of charge via the Internet at http://pubs.acs.org. to its reduced polarity ( 1,4‑dioxane = 2.22, THF = 7.52) and Brønsted basicity relative to THF. This should reduce the concentration of ionic ■ AUTHOR INFORMATION species, including solvated ‘H+’, present in the reaction mixture. Since Corresponding Author the condensation mechanisms are likely to proceed via carbocationic intermediates in the acidic media, condensation/dehydration pathways [email protected] are more likely to be supressed in 1,4-dioxane. Notes (31) For our previous 1b/THF system the hydrogen activation · The authors declare no competing financial interest. product [(THF)nH][H 1b] could be observed directly by low temperature 11B NMR.24 Such direct observation is not possible in ■ ACKNOWLEDGMENTS this case due to the high melting point of 1,4-dioxane. Yet, admission of HD (1 bar) to a solution of 1a in 1,4-dioxane leads to formation of H2 We would like to thank GreenCatEng, Eli Lilly (Pharmacat (clearly visible by 1H NMR) over several hours at rt, clearly consortium), and the EPSRC for providing funding for a PhD demonstrating that reversible activation must occur (see SI). studentship (D.J.S.) and the Royal Society for a University (32) No resonances attributable to [1a·H]− are observed by 1H, 19F, or 11 ° Research Fellowship (A.E.A.). B NMR for solutions of 1a in 1,4-dioxane under H2 (5 bar) at 100 C. (33) This activation is perhaps best characterized as a H-bonding ■ REFERENCES interaction in which the substrate enters the inner coordination sphere of the solvated proton. The pK of protonated acetone, for example, is (1) de Vries, J. G.; Elsevier, C. J. The Handbook of Homogeneous a much lower than that of protonated 1,4-dioxane (−7.2 vs −2.9223 in Hydrogenation; Wiley-VCH: Weinheim, Germany, 2008. H O), but its acidity is known to drop appreciably upon interaction with (2) Darwish, M.; Wills, M. Catal. Sci. Technol. 2012, 2, 243. 2 H-bond acceptors such as H O. See: Campbell, H. J.; Edward, J. T. Can. (3) Welch, G. C.; Juan, R. R. S.; Masuda, J. D.; Stephan, D. W. Science 2 J. Chem. 1960, 38, 2109. Palm, V. A.; Haldna, U. L.; Talvik, A. J.; Patai, S. 2006, 314, 1124. Basicity of carbonyl compounds; The Carbonyl Group: Vol. 1; John (4) Chase, P. A.; Welch, G. C.; Jurca, T.; Stephan, D. W. Angew. Chem., Wiley & Sons, Ltd.: Chichester, U.K., 1966 and references therein. Int. Ed. 2007, 46, 8050. (34) The product resonances are not observed in the 1H NMR (5) Hounjet, L. J.; Stephan, D. W. Org. Process Res. Dev. 2014, 18, 385. spectrum, due to precipitation of [nBu N][1a·OCH Ar] from the (6) Paradies, J. Synlett 2013, 24, 777. 4 2 reaction mixture; yet, removal of the solvent in vacuo and subsequent (7) Wei, S.; Du, H. J. Am. Chem. Soc. 2014, 136, 12261−12264. addition of CD Cl allow the products to be clearly observed, most (8) Nyhlen, J.; Privalov, T. Dalton Trans. 2009, 29, 5780. 2 2 notably by their diagnostic CH singlet resonances at ∼4.5 ppm. (9) Chase, P. A.; Jurca, T.; Stephan, D. W. Chem. Commun. 2008, 14, 2 (35) For 2,6-dichlorobenzaldehyde the reduction is significantly slower 1701. in the absence of additional 1a, and for neither substrate does the (10) Chen, D.; Klankermayer, J. Chem. Commun. 2008, 18, 2130. reaction proceed to completion. Both observations may simply be (11) Parks, D. J.; Piers, W. E. J. Am. Chem. Soc. 1996, 118, 9440. attributable to coprecipitation of [nBu N][1a·OCH Ar] and [nBu N] (12) Parks, D. J.; Blackwell, J. M.; Piers, W. E. J. Org. Chem. 2000, 65, 4 2 4 [1a·H] from the 1,4-dioxane solvent, which results in separation of the 3090. substrate and reductant into different phases. Repeating the experiments (13) Piers, W. E.; Marwitz, A. J. V.; Mercier, L. G. Inorg. Chem. 2011, using CD Cl as the solvent (or removing 1,4-dioxane in vacuo and 50, 12252. 2 2 replacing it with CD Cl ) prevents phase separation, and complete (14) Lindqvist, M.; Sarnela, N.; Sumerin, V.; Chernichenko, K.; 2 2 reduction is observed to occur immediately for both substrates. Leskela, M.; Repo, T. Dalton Trans. 2012, 41, 4310. (15) Longobardi, L. E.; Tang, C.; Stephan, D. W. Dalton Trans. 2014, DOI: 10.1039/C4DT02648A. (16) Spies, P.; Erker, G.; Kehr, G.; Bergander, K.; Frohlich, R.; Grimme, S.; Stephan, D. W. Chem. Commun. 2007, 47, 5072.

15816 dx.doi.org/10.1021/ja5088979 | J. Am. Chem. Soc. 2014, 136, 15813−15816 This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited.

Letter

pubs.acs.org/acscatalysis

Facile Protocol for Water-Tolerant “Frustrated Lewis Pair”-Catalyzed Hydrogenation † ‡ ‡ ‡ † Daniel J. Scott, Trevor R. Simmons, Elliot J. Lawrence, Gregory G. Wildgoose, Matthew J. Fuchter, † and Andrew E. Ashley*, † Department of Chemistry, Imperial College London, London, SW7 2AZ, United Kingdom ‡ Energy Materials Laboratory, School of Chemistry, University of East Anglia, Norwich Research Park, Norwich, NR4 7TJ, United Kingdom

*S Supporting Information

ABSTRACT: Despite rapid advances in the ﬁeld of metal-free, “frustrated Lewis pair” (FLP)-catalyzed hydrogenation, the need for strictly anhydrous reaction conditions has hampered wide-scale uptake of this methodology. Herein, we report that, despite the generally perceived moisture sensitivity of FLPs, 1,4-dioxane solutions of B(C6F5)3 actually show appreciable moisture tolerance and can catalyze hydrogenation of a range of weakly basic substrates without the need for rigorously inert conditions. In particular, reactions can be performed directly in commercially available nonanhydrous solvents without subsequent drying or use of internal desiccants. KEYWORDS: “frustrated Lewis pairs”, catalytic hydrogenation, water tolerance, solvent eﬀects, metal-free

“ ” 11 ith the advent of frustrated Lewis pair (FLP) B(C6Cl5)(C6F5)2 as a bench-stable Lewis acid for FLP W chemistry, metal-free catalytic hydrogenation has chemistry;12 however, this system still required the use of managed to progress, in less than a decade, from the first rigorously dry reaction conditions (including freshly distilled, 1 example of reversible metal-free H2 activation, into an 2 anhydrous solvent) during catalytic hydrogenations. Separately, established area of active research. In this remarkably short Ingleson et al. have described the use of the N-methylacridi- span of time, the substrate scope of FLP hydrogenation nium cation as a Lewis acid for FLP H2 activation in wet 1,2- methodologies (in which H2 is activated through the dichlorobenzene,13 yet these reaction conditions were not cooperative action of a Lewis acid/base pair precluded from extended to subsequent hydrogenation catalysis. In addition, forming a strong classical adduct) has expanded from the initial fi basic imines and aziridines3 to include a wide variety of other signi cant hydrolysis (40%) was observed. More recently, functional groups,2 including heterocycles,4 simple alkenes, and Stephan et al. have reported an FLP-catalyzed deoxygenation of 5 6 alkynes, and most recently, aldehydes and ketones. aryl ketones and, although this reaction produces H2Oasa As FLP chemistry becomes more established, focus has byproduct, the use of molecular sieves as a reagent ensures that inevitably begun to shift from early, “proof-of-concept” studies the reaction mixture remains strictly anhydrous.6c As such, the to the aim of developing more practical and widely applicable challenge of developing simple, H2O-tolerant FLP catalytic FLP catalysts. To this end, recent reports have been seen systems remains an open one.14 emphasizing the development of FLPs that can operate with The origins of FLPs’ moisture sensitivity can be understood 5a,7 8 low catalyst loadings, or with high enantioselectivity. by considering as an example the archetypal Lewis acid of FLP Despite these advances, one factor that seriously limits the chemistry: commercially available B(C6F5)3 (1). Because of its attractiveness of the FLP catalysts reported to date is their 9 high Lewis acidity, complexation of H2O not only is strong, but extreme sensitivity (and perceived sensitivity) to moisture (as fi fi · 4,9,10 also leads to signi cant Brønsted acidi cation (the pKa of [1 well as many other functional groups), requiring these 15 OH ] has been found to be comparable to that of HCl) and reagents to be handled and employed under rigorously inert 2 conditions through use of gloveboxes and Schlenk lines. This accordingly, deprotonation by even moderately strong bases fi (including the amines and phosphines typically employed in lack of H2O tolerance represents a signi cant practical barrier to the uptake of FLP catalysis by the broader chemical FLP chemistry) is irreversible. Under more forcing conditions, community that must be overcome if FLPs are truly to become viable alternatives to transition metal hydrogenation catalysts. Received: July 7, 2015 Several attempts have been made to mitigate this problem in Revised: August 11, 2015 recent years. For example, we have reported the use of Published: August 17, 2015

© 2015 American Chemical Society 5540 DOI: 10.1021/acscatal.5b01417 ACS Catal. 2015, 5, 5540−5544 ACS Catalysis Letter · − the adduct [1 OH2] is also prone to decomposition via B C Table 1. Metal-free catalytic hydrogenation of acetone in the 16 bond protonolysis (Scheme 1). presence of various amounts of H2O

Scheme 1. Pathways for Deactivation of 1 by H2O

[1], p, conv, a entry [Me2CO], M mol % [H2O]/[1] bar t,h % TON 1b,c 0.50 5 0 13 6 99 20 2c 0.29 5 1d 13 70 94 19 3e 0.27 2.5 1 50 39 92 37 4e 0.27 2.5 2 50 84 98 39 5e 0.27 2.5 5 50 108 92 37 Recently, we6a and Stephan et al.6b have independently 6f 0.53 2.5 >1g 50 15 94 38 reported the hydrogenation of ketones and aldehydes to aAll conversions measured by 1H NMR integration. bResult taken 6a c d · e alcohols, catalyzed by 1 in ethereal solvents (Scheme 2). These from Ashley et al. 0.2 mmol acetone. Added as [1 OH2]. 4.0 reactions are clearly tolerant of hydroxylic functionalities, which mmol acetone. f8 mmol acetone, ACS reagent grade commercial g fi is attributed to the weak Brønsted basicity of the ethers solvent (Sigma-Aldrich). Con rmed by control experiments (see the employed, which inhibits irreversible deprotonation of the SI). highly acidic [1·ROH] adducts. · Following on from these results, we reasoned that these deprotonation of [1 OH2] does not occur (c.f. Scheme 1). Even systems might also be capable of tolerating the presence of so, it seems possible that reversible deprotonation does occur, H2O, for similar reasons, which would represent an important which could be consistent with the acidity of this adduct (vide advance for the field of FLP chemistry. Initial investigations supra).19 Evidence for deprotonation comes from 11B NMR · 11 focused on the hydrogenation of acetone (the simplest spectroscopy: although [1 OH2] alone shows an B NMR substrate previously examined under anhydrous conditions). resonance at 4.6 ppm in the noncoordinating yet highly polar Gratifyingly, when 1 was replaced with the preformed adduct 1· solvent 1,2-difluorobenzene (DFB), addition of 1,4-dioxane 15 fi − OH2 under identical conditions, selective hydrogenation to 2- leads to a clear up eld shift, to 0.6 ppm with 1 equiv of propanol was observed to proceed cleanly and with good ethereal base, and −2.1 ppm with 10 equiv (Figure 1).20 turnover number (Table 1, entry 2). Although the rate of Typical nonanhydrous reaction mixtures, which contain [1· fi fi reaction is reduced (c.f. Table 1, entry 1), con rming that H2O OH2]inneat 1,4-dioxane, produce resonances farther up eld fi − 11 acts as a signi cant catalyst poison, inhibition by H2Ois still, at 3.0 ppm. For comparison, the B NMR shift of · “ ” · nonetheless clearly reversible. At the time, this represented the [NMe4][1 OH], which contains the free conjugate base of [1 fi − rst example of an FLP-catalyzed hydrogenation that is tolerant OH2], has been reported as 2.1 ppm in CD2Cl2, with similar 21 1 of stoichiometric (relative to catalyst) H2O. shifts for related salts. H NMR analysis also suggests an fi · Increasing the H2 pressure led to a signi cant rate interaction between [1 OH2] and 1,4-dioxane, with the 1,4- enhancement, even at lower catalyst loadings (Table 1, entry dioxane resonance shifted slightly downfield, from 3.56 ppm in · 3) and without any detectable catalyst decomposition the absence of [1 OH2], to 3.59 ppm in the presence of 1 equiv, 19 11 (ascertained by F and B NMR spectroscopy). Under these indicating overall deshielding. Although displacement of H2O conditions, several more equivalents of H2O could be tolerated, for 1,4-dioxane could also potentially explain these shifts in the with an attendant decrease in rate yet otherwise no major NMR spectra, this possibility can be discounted: addition of 1 difference in reaction outcome (Table 1, entries 4 and 5). and 10 equiv of 1,4-dioxane to dry 1 in DFB leads to 11B NMR On the basis of the ability of this system to tolerate multiple resonances at 9.8 and 5.7 ppm, respectively, and 1,4-dioxane 1H equivalents of water, we reasoned that the use of “undried” resonances at 3.96 and 3.59 ppm, respectively; in all cases fi fi solvents ought to be achievable; impressively, the reaction signi cantly farther down eld than when H2O is also present. ff · · could be performed very e ectively in nonanhydrous Reversible deprotonation of [1 OH2] and the related [1 commercial solvent17 even without any need for subsequent ROH] adducts may partially explain the large effect that drying, degassing, or other purification. Furthermore, doubling pressure has been observed to have on both the anhydrous and the substrate and catalyst concentrations allowed for a nonanhydrous reactions. Increasing H2 pressure will increase fi signi cant decrease in reaction time (Table 1, entry 6). the solution concentration of H2 and, in turn, the degree of H2 The mechanism by which the hydrogenation is believed to activation. The resulting increase in Brønsted acid concen- · · occur is identical to that proposed for the anhydrous reaction, tration should perturb the equilibrium between [1 OH2]/[1 · ff 18 · − · − with [1 OH2] acting as an o -cycle resting state (Scheme 3). iPrOH] and [1 OH] /[1 OiPr] in favor of the more weakly As with previously observed alcohol tolerance, H2O tolerance is bound neutral adducts, hence facilitating catalytic turnover via attributed to the lack of any strong base, meaning irreversible ROH/H2O dissociation from 1.

a Scheme 2. Previously-Reported Hydrogenation of Aldehydes and Ketones Catalyzed by 1.6a,b

aR = aryl, alkyl; R′ = alkyl, H.

5541 DOI: 10.1021/acscatal.5b01417 ACS Catal. 2015, 5, 5540−5544 ACS Catalysis Letter

a Scheme 3. Proposed Mechanism for Moisture-Tolerant Hydrogenation of Acetone

a · · Possible hydrogen bonding of solvent with [1 ROH] and [1 OH2] omitted for clarity.

11 · Figure 1. Variation in B NMR chemical shift of [1 OH2] upon addition of 1,4-dioxane in DFB.

Given that the rationale for H2O tolerance in this system other carbonyl substrates. Collectively, these results suggest a depends primarily on the absence of a strong base, it seemed similar substrate scope for the anhydrous and nonanhydrous reasonable that catalytic hydrogenation without the need for reaction protocols. rigorously dry conditions might also be achievable for other In addition to the advantages already discussed, moisture- weakly basic substrates. To this end, a variety of additional tolerance should also allow FLPs to catalyze reactions that substrates (both carbonyls and noncarbonyls) were treated produce H2O. To this end, acetophenone was exposed to our under similar conditions. For most, clean, eﬃcient catalytic anhydrous reaction conditions. Clean, catalytic reductive reduction was observed without the need for extensive further deoxygenation to ethylbenzene23 was observed, despite the − optimization (Table 2, entries 1 7), clearly indicating the concomitant formation of H2O (5 equiv relative to catalyst) generality of this water-tolerant methodology. As well as using that necessarily occurs (Scheme 4). undried solvent, reactions could be performed under open In conclusion, we have demonstrated a number of examples bench conditions without the need for inert atmosphere of FLP-catalyzed hydrogenation reactions demonstrating containment systems22 (although long-term storage of 1 is best appreciable water tolerance. As a result, a variety of weakly carried out in a glovebox). By contrast, the attempted reduction basic substrates can be hydrogenated cleanly and in high yield, of cyclohexanone under the same conditions was unsuccessful in the absence of a transition metal catalyst, in commercial- (Table 2, entry 8), which is qualitatively consistent with grade “bench” (undried) solvents. By removing the need for previous observations made under dry conditions,6a for which both extensive, laborious drying of reaction solvents and inert higher pressures were required for hydrogenation than for the atmosphere reaction techniques, this development signiﬁcantly

5542 DOI: 10.1021/acscatal.5b01417 ACS Catal. 2015, 5, 5540−5544 ACS Catalysis Letter

Table 2. Metal-Free Hydrogenation of Weakly-Basic Substrates in Non-Anhydrous Solvent

aStandard conditions: 4.0 mmol substrate, 0.53 M. bAll conversions measured by 1H NMR integration. c2.0 mmol substrate. d8.0 mmol substrate. eUnquantiﬁed mixture of isomers.

Scheme 4. Hydrogenation of Acetophenone to Ethylbenzene Notes fi and H2O Catalyzed by 1 The authors declare no competing nancial interest. ■ ACKNOWLEDGMENTS We would like to thank GreenCatEng, Eli Lilly (Pharmacat consortium), and the EPSRC for providing funding for a Ph.D. studentship (D.J.S.), and Dr. George Britovsek and Prof. Ramon Vilar for access to high-pressure equipment. G.G.W. and A.E.A. thank the Royal Society for financial support increases the practicality of the FLP hydrogenation method- through University Research Fellowships. The research leading fi ology. These ndings extend the current reach of FLP to these results has received funding from the European hydrogenation catalysis from rigorously anhydrous research Research Council under the ERC Starting Grant Agreement laboratory conditions into industrially relevant, commercially No. 307061 (PiHOMER) and ERC PoC Grant Agreement No. available solvent grades and reaction conditions. 640988 (FLPower). ■ ASSOCIATED CONTENT ■ REFERENCES *S Supporting Information (1) Welch, G. C.; Juan, R. R. S.; Masuda, J. D.; Stephan, D. W. Science The Supporting Information is available free of charge on the 2006, 314, 1124−1126. ACS Publications website at DOI: 10.1021/acscatal.5b01417. (2) For recent reviews, see: (a) Stephan, D. W.; Erker, G. Angew. Full experimental details (PDF) Chem., Int. Ed. 2015, 54, 6400−6441. (b) Stephan, D. W. Acc. Chem. Res. 2015, 48, 306−316. ■ AUTHOR INFORMATION (3) Chase, P. A.; Welch, G. C.; Jurca, T.; Stephan, D. W. Angew. Chem., Int. Ed. 2007, 46, 8050−8053. Corresponding Author (4) Stephan, D. W.; Greenberg, S.; Graham, T. W.; Chase, P.; Hastie, *Phone: +44 (0)20 759 45810. E-mail: [email protected]. J. J.; Geier, S. J.; Farrell, J. M.; Brown, C. C.; Heiden, Z. M.; Welch, G. uk. C.; Ullrich, M. Inorg. Chem. 2011, 50, 12338−12348.

5543 DOI: 10.1021/acscatal.5b01417 ACS Catal. 2015, 5, 5540−5544 ACS Catalysis Letter

(5) (a) Xu, X.; Kehr, G.; Daniliuc, C. G.; Erker, G. J. Am. Chem. Soc. observations made at lower pressure.6a At completion, only ethyl- 2015, 137, 4550−4557. (b) Chernichenko, K.; Madarasz,́ Á.; Papai,́ I.; benzene was observed, however, indicating that the alkene undergoes Nieger, M.; Leskela,̈ M.; Repo, T. Nat. Chem. 2013, 5, 718−723. further reduction (either directly, or via rehydration to the alcohol). (6) (a) Scott, D. J.; Fuchter, M. J.; Ashley, A. E. J. Am. Chem. Soc. 2014, 136, 15813−15816. (b) Mahdi, T.; Stephan, D. W. J. Am. Chem. Soc. 2014, 136, 15809−15812. (c) Mahdi, T.; Stephan, D. W. Angew. Chem., Int. Ed. 2015, 54, 8511−8514. (7) Farrell, J. M.; Posaratnanathan, R. T.; Stephan, D. W. Chem. Sci. 2015, 6, 2010−2015. (8) (a) Lindqvist, M.; Borre, K.; Axenov, K.; Kotai,́ B.; Nieger, M.; Leskela,̈ M.; Papai,́ I.; Repo, T. J. Am. Chem. Soc. 2015, 137, 4038− 4041. (b) Zhang, Z.; Du, H. Org. Lett. 2015, 17, 2816−2819. (c) Ren, X.; Li, G.; Wei, S.; Du, H. Org. Lett. 2015, 17, 990−993. (d) Wei, S.; Du, H. J. Am. Chem. Soc. 2014, 136, 12261−12264. (9) Thomson, J. W.; Hatnean, J. A.; Hastie, J. J.; Pasternak, A.; Stephan, D. W.; Chase, P. A. Org. Process Res. Dev. 2013, 17, 1287− 1292. (10) (a) Erős, G.; Mehdi, H.; Papai,́ I.; Rokob, T. A.; Kiraly,́ P.; Tarká nyi,́ G.; Soos,́ T. Angew. Chem., Int. Ed. 2010, 49, 6559−6563. (b) Erős, G.; Nagy, K.; Mehdi, H.; Papai,́ I.; Nagy, P.; Kiraly,́ P.; Tarká nyi,́ G.; Soos,́ T. Chem. - Eur. J. 2012, 18, 574−585. (c) Greb, L.; Daniliuc, C. G.; Bergander, K.; Paradies, J. Angew. Chem., Int. Ed. 2013, 52, 5876−5879. (11) Ashley, A. E.; Herrington, T. J.; Wildgoose, G. G.; Zaher, H.; Thompson, A. L.; Rees, N. H.; Kramer,̈ T.; O’Hare, D. J. Am. Chem. Soc. 2011, 133, 14727−14740. (12) Scott, D. J.; Fuchter, M. J.; Ashley, A. E. Angew. Chem., Int. Ed. 2014, 53, 10218−10222. (13) Clark, E. R.; Ingleson, M. J. Angew. Chem., Int. Ed. 2014, 53, 11306−11309. (14) While this manuscript was under review, Sooś et al. reported a moisture-tolerant protocol for FLP-catalysed hydrogenation of aldehydes and ketones using novel boranes speciﬁcally designed for that purpose. Their approach (targeted design of new catalysts, rather than adaptation of existing FLP hydrogenation protocols) comple- ments the work reported herein. Gyömöre, Á.; Bakos, M.; Földes, T.; Papai,́ I.; Domjan,́ A.; Soos,́ T. ACS Catal. 2015, DOI: 10.1021/ acscatal.5b01299. (15) pKa = 8.4 in MeCN, c.f. 8.5 for HCl: Bergquist, C.; Bridgewater, B. M.; Harlan, C. J.; Norton, J. R.; Friesner, R. A.; Parkin, G. J. Am. Chem. Soc. 2000, 122, 10581−10590. (16) (a) Bradley, D. C.; Harding, I. S.; Keefe, A. D.; Motevalli, M.; Zheng, D. H. J. Chem. Soc., Dalton Trans. 1996,3931−3936. (b) Ashley, A. E.; Thompson, A. L.; O’Hare, D. Angew. Chem., Int. Ed. 2009, 48, 9839−9843. (17) ACS reagent grade 1,4-dioxane purchased from Sigma Aldrich was used. Karl Fischer titration indicated 220 ppm/0.022% H2O (w/ w) (advertised as up to 0.05%). (18) We propose that 1,4-dioxane acts as the main base for hydrogen activation due to the vast excess (12 M) in which it is present; however, in principle, this role could also be played by water or 2- propanol. − (19) The pKa of protonated 1,4-dioxane has been measured as 2.92 · in aqueous solution. Although the aqueous pKa of [1 OH2] has not been measured,15 our results indicate that it may be comparable. See: Arnett, E.; Wu, C. Y. J. Am. Chem. Soc. 1960, 82, 4999−5000. . (20) It should be noted that these observations could also be consistent with strong hydrogen bonding to the solvent rather than full formal deprotonation; in practice, the true situation might be expected to lie somewhere between these two extremes. (21) (a) Hewavitharanage, P.; Danilov, E. O.; Neckers, D. C. J. Org. Chem. 2005, 70, 10653−10659. (b) Bibal, C.; Santini, C. C.; Chauvin, Y.; Vallee, C.; Olivier-Bourbigou, H. Dalton Trans. 2008, 37, 2866− 2870. (22) For all experiments [H2O] > [1], as indicated by control experiments: see SI. (23) 1H NMR analysis of an aliquot taken at partial conversion showed formation of styrene (in addition to ethylbenzene and the expected alcohol intermediate), which is consistent with previous

5544 DOI: 10.1021/acscatal.5b01417 ACS Catal. 2015, 5, 5540−5544 Angewandte Communications Chemie

International Edition: DOI: 10.1002/anie.201606639 FLP Hydrogenation German Edition: DOI: 10.1002/ange.201606639 Versatile Catalytic Hydrogenation Using A Simple Tin(IV) Lewis Acid Daniel J. Scott, Nicholas A. Phillips, Joshua S. Sapsford, Arron C. Deacy, Matthew J. Fuchter, and Andrew E. Ashley*

Abstract: Despite the rapid development of frustrated Lewis LAs; thus far this has exclusively been achieved using B- pair (FLP) chemistry over the last ten years, its application in based acceptors[6] [predominantly (fluoroaryl)borane deriva- [7] catalytic hydrogenations remains dependent on a narrow tives, of which B(C6F5)3 is prevalent], with the exception of family of structurally similar early main-group Lewis acids a single report using P-based LAs (for a limited range of (LAs), inevitably placing limitations on reactivity, sensitivity activated olefins).[8] This constrained focus is far from ideal, as

and substrate scope. Herein we describe the FLP-mediated H2 examining and developing a wider variety of LAs can be activation and catalytic hydrogenation activity of the alterna- expected to produce novel FLP-catalyzed protocols that

tive LA iPr3SnOTf, which acts as a surrogate for the display different substrate scope and/or more favorable + [9] trialkylstannylium ion iPr3Sn , and is rapidly and easily functional group tolerance. For example, the application prepared from simple, inexpensive starting materials. This of highly Lewis acidic boranes to the FLP-catalyzed hydro- highly thermally robust LA is found to be competent in the genation of organic carbonyls has been notably challenging: hydrogenation of a number of different unsaturated functional whilst stoichiometric reductions were reported as early as groups (which is unique to date for main-group FLP LAs not 2007,[9] it took until 2014 until catalytic protocols were based on boron), and also displays a remarkable tolerance to developed.[10] This difficulty can be attributed to the strength moisture. of the interaction between the alcohol (ROH) products and the LAs, which renders the LA·ROH adducts strongly acidic = < [11] Since the formalization of the concept within the last [cf. H2O·B(C6F5)3 ;pKa 8.4 (MeCN), 1(H2O, est.)]; decade, great attention has been focused on the development consequently, these adducts are fundamentally incompatible and study of frustrated Lewis pairs (FLPs): Lewis acid (LA) with the moderately strong N/P-centered LBs typically and base (LB) combinations that fail to form the classically incorporated into active FLP catalysts. Ultimately, turnover expected strong adduct, typically because it is sterically can only be achieved when such LBs are strictly excluded, due precluded.[1] The resulting combined reactivity has been to the necessarily highly Brønsted acidic media [for example, ! [10,12] found to lead to a range of novel bond activation reactions protonated ethers, pKa(H2O) 0]. that do not require the involvement of a transition metal Based on the above, we were motivated to investigate (TM).[2] Of particular interest has been the activation and FLPs based on heavier p-block LAs, which have thus far [13] cleavage of H2, which has allowed the development of the first attracted scant attention for use in FLP applications. general methodology for TM-free catalytic hydrogenation.[3] Specifically, our interest was drawn to stannylium ion + = [14] Computational investigations have suggested that the “R3Sn ”(R alkyl) LAs; these are isolobal with Ar3B

primary requirements for successful activation of H2 by an species commonly employed in FLP chemistry, and have been D = FLP are a sufficient cumulative LA/LB strength, and calculated to possess similar hydride ion affinities ( GHÀ [4] À1 À a suitable steric profile. One appealing aspect of FLP 65.83 and 64.95 kcalmol for nBu3Sn-H and [H-B(C6F5)3] chemistry is therefore the generality of the concept; indeed, respectively),[15] suggesting that they ought to demonstrate

FLP-type reactions have been identified for a broad spectrum comparable reactivity in FLP H2 activation and hydrogena- [2, 5] = of LAs and LBs. Nevertheless inspection of the literature tion reactions. Furthermore, C O reductions by R3SnH in reveals that, despite the apparent breadth of the field, protic media are well known to occur via ionic hydride investigations into TM-free FLP-catalyzed hydrogenation transfer.[16] Crucially, however, these LAs interact only much have focused overwhelmingly on a very narrow range of more weakly with hydroxylic species [for example, + = [17] (nBu3Sn·xH2O) ;pKa(H2O) 6.25]. [*] D. J. Scott, Dr. N. A. Phillips, J. S. Sapsford, A. C. Deacy, Manners et al. have previously investigated the use of + = Dr. M. J. Fuchter, Dr. A. E. Ashley nBu3SnOTf (an nBu3Sn equivalent; Tf CF3SO2)asaLA Department of Chemistry, Imperial College London partner in FLP chemistry,[13a] but reported that it was not

London, SW7 2AZ (UK) capable of activating H2 when combined with the strong E-mail: [email protected] amine base TMP (2,2,6,6-tetramethylpiperidine) at 508C, Homepage: http://www.ashleyresearchgroup.org.uk whereas the B(C6F5)3/TMP FLP readily cleaves H2, even at Supporting information and the ORCID identification number(s) for room temperature;[18] this result was attributed to the poorer the author(s) of this article can be found under http://dx.doi.org/10. electrophilicity of the Sn compound, and it is evident that the 1002/anie.201606639. Sn–OTf interaction is strong enough to substantially reduce 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. + the Lewis acidity of the nBu Sn fragment. KGaA. This is an open access article under the terms of the Creative 3 Commons Attribution License, which permits use, distribution and We envisioned that it should be possible to increase the reproduction in any medium, provided the original work is properly Lewis acidity, to the threshold necessary for favorable H2 cited. heterolysis, by simply increasing the size of the alkyl groups

14738 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2016, 55, 14738 –14742 Angewandte Communications Chemie on Sn, thereby increasing the degree of “internal frustra- Table 1: [1]OTf-catalyzed hydrogenation of imines. [19] + À tion” between the R3Sn and TfO moieties. To this end, we targeted the bulkier trialkylstannyl compound iPr3SnOTf ([1]OTf), which was readily prepared via reaction of excess iPrMgCl and SnCl4 to generate iPr4Sn, followed by facile protodealkylation with HOTf (Scheme 1). This straightfor- ward and inexpensive two-step procedure furnishes pure Entry[a] Substrate R R1 R2 Base t [h] Conversion [%][b] 1 2a HHtBu – 12 97 2 2b HMetBu – 16 85 3 2c HHPh–164 4 2c H H Ph Col 24 > 99 1 Scheme 1. Synthesis of [ ]OTf. 5 2d H Me Ph Col 32 96 6 2e HHTsCol8065 7 2f 4-Br H tBu Col 16 96 [1]OTf in good yield (42%, 2 steps), and can easily be [a] 10 bar refers to initial pressure at RT. [b] Conversions determined by performed on a multi-gram scale. [1]OTf is a white solid that 1H NMR spectroscopic analysis (see the SI). shows moderate solubility in polar halogenated solvents and its 119Sn{1H} spectrum shows a single broad resonance at d = D = 156 ppm ( v1/2 130 Hz, CDCl3). The high chemical shift is activation less favorable. Consistent with this interpretation, = [23] consistent with significant stannylium ion character, although addition of 2,4,6-collidine [Col; pKa(MeCN) 14.98] as an it is considerably upfield of the value reported for [nBu3Sn]- auxiliary base leads to a dramatic improvement in perfor- d = [CB11Me12]( 454 ppm), which displays the least coordi- mance (Table 1, entry 4), and also allows for reduction of the nated trialkylstannylium core to date.[20] Gutmann–Beckett related ketimine PhC(Me)=NPh (2d; Table 1, entry 5), and [21] = = Lewis acidity measurements support this conclusion, indi- even PhCH NTs (2e;Ts O2SC6H4Me, 4-toluenesulfonyl), cating increased electrophilicity in comparison with although the latter reaction is appreciably slower, presumably [22] = nBu3SnOTf, although still lower than B(C6F5)3 {AN 64.2 as the substrate is less basic still (Table 1, entry 6). Notably, = nBu3SnOTf; 68.0 [1]OTf; 78.1 B(C6F5)3}. [1]OTf has also been the bromoaryl imine 2falso undergoes efficient C N hydro- characterized by 1H, 13Cand19F NMR spectroscopy, MS and genation (Table 1, entry 7); no evidence of hydrodebromina- elemental analysis (see the Supporting Information (SI)). tion is observed during this reaction (no NMR resonances [24] Addition of DABCO (1,4-diazabicyclo[2.2.2]octane) to attributable to 2a/3a,[1]Br or [1]2), supporting the idea [1]OTf (1:1) leads to an upfield shift in the 119Sn{1H} that radical Sn species do not appear to be involved in this resonance (which remains similarly broad) to 39 ppm, con- reaction. Accordingly, we propose that hydrogenation occurs sistent with a donor–acceptor interaction. However, the via a polar mechanism analogous to that for related borane- 1 [1d,e,25] corresponding H NMR spectrum shows only a single reso- catalyzed systems: H2 activation by an FLP consisting of nance for the DABCO protons, suggesting rapid exchange [1]OTf/imine is followed by hydride transfer and release of between an adduct and FLP. Admission of H2 (4 bar) leads to amine at elevated temperature (Figure S15). This is further the appearance of resonances in the room temperature supported by the observation that pre-formed 2a·HOTf is 1H [5.12 ppm, SnH, 1J(119Sn/117Sn-1H) = 1471/1405 Hz; rapidly reduced by [1]H even at RT,[26] whereas the equivalent 10.93 ppm, NH]and119Sn{1H} (À46 ppm) NMR spectra, reactions with unprotonated 2a, either alone or in the that are consistent with formation of iPr3SnH ([1]H) and presence of [1]OTf, do not lead to significant reduction at 8 DABCO·HOTf, and hence H2 heterolysis by the N/Sn Lewis 120 C (see SI). Interestingly, there is evidence for autocatal- pair. Further, conclusive proof for H2 activation is provided ysis during the course of the reaction (16% conversion 119 1 by replacing H2 with D2, which causes the new Sn{ H} observed after 3 h, 60% after 6 h); comparable observations resonance to split into a triplet [1:1:1, 1J(119Sn-2H) = 226 Hz], have been made by Paradies et al. for imine hydrogenations 1 and the new resonances in the H NMR spectrum to be catalyzed by B(2,6-F2C6H3)3, and are attributed to the replaced by equivalent signals in the 2H spectrum. This increased basicity of the product amines, relative to the represents the first example of FLP H2 activation using a LA imine substrate, rendering H2 activation more favorable as based on Sn, or any p-block element beyond the 3rd row of the more product is formed.[25] periodic table. Following success in the hydrogenation of imines, we were

Having demonstrated H2 activation, our focus shifted to interested to see whether [1]OTf might also be capable of achieving catalytic hydrogenation using [1]OTf. Gratifyingly, mediating the hydrogenation of closely related carbonyl heating the archetypal FLP substrates PhCH=NtBu (2a)and compounds. Satisfyingly, when acetone (4a) is exposed to PhC(Me)=NtBu (2b) with 10 mol% [1]OTf to 1208C under reaction conditions similar to those used to hydrogenate 2c

H2 (10 bar) led to conversion to the respective amines (3aand catalytic conversion to 2-propanol (5a) is observed (Table 2, 3b; Table 1, entries 1 and 2). Conversely, the N-phenyl entry 1). Whilst the reaction at 1208C is somewhat slow, at analogue PhCH=NPh (2c) is reduced far less effectively 1808C near-quantitative conversion can be observed within (Table 1, entry 3), which is attributed to the reduced basicity 32 h (Table 2, entry 2). Significantly, no evidence of catalyst of both the imine and amine product, which makes H2 decomposition is observed in this homogeneous reaction,

Angew. Chem. Int. Ed. 2016, 55, 14738 –14742 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org 14739 Angewandte Communications Chemie

Table 2: [1]OTf-catalyzed hydrogenation of ketones and aldehydes.

Entry[a] Substrate R R1 Base t [h] Conversion [%][b] 1[c] 4a Me Me Col 96 78 2[d] 4a Me Me Col 32 97 3[d] 4b Ph Me Col 48 91[f] 4[d] 4c tBu H Col 48 79 [d] 4d 5 2,6-Cl2C6H3 H Col 32 91 6 4a Me Me Col 16 57 7 4a Me Me Lut 16 48 8 4a Me Me DABCO 16 14 9 4a Me Me [1]OiPr[e] 16 32 10[g] 4a Me Me Col 32 95 Scheme 2. a) Proposed mechanism for catalytic hydrogenation of 4a 1 = [a] 10 bar refers to initial pressure at RT. [b] Conversions determined by using [ ]OTf ([N] 2,4,6-collidine) and b) alternative H2 activation 1H NMR spectroscopic analysis (see the SI). [c] Reaction run at 1208C, using in situ generated [1]OiPr. repressurized after 48 h. [d] Repressurized at 16 h intervals. [e] Gener- ated in situ from [1]H and 4a (see the SI). [f] Based on consumption of 4b 5b 6 7 ; reaction produces in addition to and as side-products in a ca. to [1]OTf, indicative of SnÀO binding.[31] A proposed 74:18:8 molar ratio (see the SI). [g] Using undried reagents, solvent and subsequent HÀ transfer from [1]H to adduct {[1]·4a}OTf, to catalyst (see the SI). form [1]OiPr and regenerate [1]OTf, is supported by the observation that [1]H is capable of reducing 4a in the either by 1Hor119Sn{1H} NMR spectroscopy,[27] in comparison presence of [1]OTf even at RT, whereas no appreciable [1f,28] 8 with analogous FLP protocols mediated by B(C6F5)3. To conversion is observed in its absence either at RT or 120 C. the best of our knowledge this is the first example of Conversely, if [1]OTf is replaced by Col·HOTf, only slow [32] + a catalytically active FLP system capable of tolerating such release of H2 is observed at RT. In order for the final H conditions without degradation, and illustrates the impressive transfer step to occur efficiently it should be recognized that thermal stability of [1]OTf, which enables the use of more Col and [1]OiPr must be comparable in base strength and, forcing conditions in order to achieve an improved rate of therefore, it may be envisaged that once [1]OiPr is formed in

turnover. As well as 4a, other aliphatic and aromatic ketones the reaction mixture, it could also activate H2 in conjunction and aldehydes (4b—d) can be reduced under these conditions with [1]OTf (Scheme 2 b). In fact, catalytic hydrogenation can (Table 2, entries 3–5). In the case of acetophenone (4b), be observed by substituting Col with [1]OiPr (generated in 1H NMR spectroscopic analysis indicates formation of the situ from [1]H and 4a; Table 2, entry 9), thus demonstrating expected alcohol 5b, in addition to smaller quantities of its competence in this role. Even so, the reduced rate of styrene (6)anda-methylbenzyl ether (7). Similar side- turnover in this reaction indicates that the auxiliary base does reactions were observed in our previous attempts to reduce play a beneficial role beyond simply facilitating formation of [10b,12c] 4b using B(C6F5)3 in 1,4-dioxane, but in those cases this some initial [1]OiPr, presumably by rendering H2 activation led to severe reductions in conversion and rate of turnover. more favorable.[33] The ease and speed with which it was possible to apply this Clear tolerance of alcohol products suggested that these system to carbonyl hydrogenation stands in contrast to the reactions might also demonstrate appreciable moisture toler- extended period of development required before more ance.[10,12] Remarkably, when the hydrogenation of model conventional B-based FLPs were successfully used in this substrate 4a (chosen over an imine to avoid hydrolysis) was transformation.[10a,b] It is also noteworthy that the [1]OTf- prepared on the open bench, with non-anhydrous reagents catalyzed reaction can proceed using a rather conventional, and solvent, and using [1]OTf that had been exposed to air for moderately-strong, N-centered LB, which again contrasts 1 week, the reaction was observed to proceed without any with B-based systems and is consistent with less acidic adducts noticeable reduction in rate (Table 2, entry 10; details in SI). forming between the product alcohols (e.g. 5a) and [1]OTf. This is unprecedented in FLP catalysis, where even the most The choice of LB is important to the outcome of the tolerant of previously reported reactions have been dramat- [12] hydrogenation of 4a (Table 2, entries 6–8), with inferior ically slowed by adventitious H2O, and suggests a major results obtained using either a weaker or stronger base [2,6- advantage of using Sn-based LAs. = [29] lutidine (Lut), DABCO; pKa(MeCN) 14.13, 18.29]. Finally, we investigated the use of [1]OTf in the catalytic Given the low Brønsted basicity of 4a we propose hydrogenation of compounds containing other unsaturated a slightly different mechanism for its reduction than for functionalities; the heteroaromatic ring of acridine, and the 2a,[30] with the substrate activated by [1]+ rather than via H- C=C bonds in n-butyl acrylate and 1-piperidino-1-cyclohex- bonding to [Col-H]+ (Scheme 2a).[16b] Evidence for this comes ene could all be effectively reduced (yields 83–99%), further from the significantly upfield-shifted 119Sn{1H} NMR reso- demonstrating the versatility of this SnIV compound (Fig- nance (d = 92 ppm) observed upon addition of 4a (10 equiv.) ure S33).

14740 www.angewandte.org 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2016, 55, 14738 –14742 Angewandte Communications Chemie

In summary, we have demonstrated the use of readily Welch, M. Ullrich, Inorg. Chem. 2011, 50, 12338 – 12348; b) J. Paradies, Angew. Chem. Int. Ed. 2014, 53, 3552 – 3557; Angew. accessible and inexpensive iPr3SnOTf as a main-group LA catalyst for the hydrogenation of C=C, C=N and C=O bonds; Chem. 2014, 126, 3624 – 3629; c) L. J. Hounjet, D. W. Stephan, this constitutes only the second example of an FLP hydro- Org. Process Res. Dev. 2014, 18, 385 – 391. [4] T. A. Rokob, A. Hamza, I. Ppai, J. Am. Chem. Soc. 2009, 131, genation protocol utilizing a p-block LA not incorporating 10701 – 10710. boron, and the first such example shown to be applicable to [5] S. A. Weicker, D. W. Stephan, Bull. Chem. Soc. Jpn. 2015, 88, the reduction of a range of different functional groups. 1003 – 1016. Despite the ubiquity of Sn in industrial catalysis this also [6] A few examples of catalytic hydrogenation using TM-based represents, to the best of our knowledge, the first example of FLPs have also been reported. See: a) M. P. Boone, D. W. homogeneous catalytic hydrogenation using a Sn-based Stephan, J. Am. Chem. Soc. 2013, 135, 8508 – 8511; b) A. T. system of any kind.[34] Of particular interest is the ready Normand, C. G. Daniliuc, B. Wibbeling, G. Kehr, P. Le Gendre, applicability of this protocol to C=O bond hydrogenation, in G. Erker, J. Am. Chem. Soc. 2015, 137, 10796 – 10808; c) X. Xu, C. G. Daniliuc, G. Erker, J. Am. Chem. Soc. 2015, 137, 4550 – a reaction that displays an unparalleled level of H O 2 4557; d) N. S. Lambic, R. D. Sommer, E. A. Ison, J. Am. Chem. tolerance. This neatly demonstrates the value of pursuing Soc. 2016, 138, 4832 – 4842; e) S. R. Flynn, O. J. Metters, I. alternative FLP LAs, and can be jointly attributed to the Manners, D. F. Wass, Organometallics 2016, 35, 847 – 850. formation of weakly acidic LA·ROH adducts; a thermally [7] For catalytic hydrogenation using much weaker boranes paired + robust [iPr3Sn] core, allowing access to high reaction with strong bases, see: S. Mummadi, D. K. Unruh, J. Zhao, S. Li, temperatures; and the stability of the SnÀC bonds towards C. Krempner, J. Am. Chem. Soc. 2016, 138, 3286 – 3289. [8] a) T. vom Stein, M. Perz, R. Dobrovetsky, D. Winkelhaus, C. B. protolytic cleavage for example, by H2O. Clearly there is significant scope for variation of the triorganotin(IV) frame- Caputo, D. W. Stephan, Angew. Chem. Int. Ed. 2015, 54, 10178 – work in “R Sn+” species; investigations into how this affects 10182; Angew. Chem. 2015, 127, 10316 – 10320; imine hydro- 3 genation is also known using an Al-based LA or a pyridylidene their reactivity, functional group tolerance, and substrate as catalyst, but in both cases was reported not to proceed via an scope are currently underway. FLP mechanism: b) J. A. Hatnean, J. W. Thomson, P. A. Chase, D. W. Stephan, Chem. Commun. 2014, 50, 301 – 303; c) J. Auth, J. Padevet, P. Maulen, A. Pfaltz, Angew. Chem. Int. Ed. 2015, 54, Acknowledgements 9542 – 9545; Angew. Chem. 2015, 127, 9678 – 9681. [9] P. Spies, G. Erker, G. Kehr, K. Bergander, R. Frohlich, S. We thank GreenCatEng, Eli Lilly (Pharmacat consortium) Grimme, D. W. Stephan, Chem. Commun. 2007, 5072 – 5074. and the EPSRC for providing funding for a PhD studentship [10] a) T. Mahdi, D. W. Stephan, J. Am. Chem. Soc. 2014, 136, 15809 – 15812; b) D. J. Scott, M. J. Fuchter, A. E. Ashley, J. Am. Chem. (D.J.S.), and the Royal Society for a University Research Soc. 2014, 136, 15813 – 15816. Fellowship (A.E.A.). [11] C. Bergquist, B. M. Bridgewater, C. J. Harlan, J. R. Norton, R. A. Friesner, G. Parkin, J. Am. Chem. Soc. 2000, 122, 10581 – Keywords: catalysis · frustrated Lewis pairs · hydrogenation · 10590. stannylium · tin [12] a) T. Mahdi, D. W. Stephan, Angew. Chem. Int. Ed. 2015, 54, 8511 – 8514; Angew. Chem. 2015, 127, 8631 – 8634; b) . Gyç- How to cite: Angew. Chem. Int. Ed. 2016, 55, 14738–14742 mçre, M. Bakos, T. Fçldes, I. Ppai, A. Donjn, T. Sos, ACS Angew. Chem. 2016, 128, 14958–14962 Catal. 2015, 5, 5366 – 5372; c) D. J. Scott, T. R. Simmons, E. J. Lawrence, G. G. Wildgoose, M. J. Fuchter, A. E. Ashley, ACS Catal. 2015, 5, 5540 – 5544; d) V. Fasano, J. E. Radcliffe, M. J. [1] For key publications, see: a) G. C. Welch, R. R. San Juan, J. D. Ingleson, ACS Catal. 2016, 6, 1793 – 1798. Masuda, D. W. Stephan, Science 2006, 314, 1124 – 1126; b) G. C. [13] a) G. R. Whittell, E. I. Balmond, A. P. M. Robertson, S. K. Patra, Welch, D. W. Stephan, J. Am. Chem. Soc. 2007, 129, 1880 – 1881; M. F. Haddow, I. Manners, Eur. J. Inorg. Chem. 2010, 3967 – c) P. A. Chase, G. C. Welch, T. Jurca, D. W. Stephan, Angew. 3975; b) S. Freitag, J. Henning, H. Schubert, L. Wesemann, Chem. Int. Ed. 2007, 46, 8050 – 8053; Angew. Chem. 2007, 119, Angew. Chem. Int. Ed. 2013, 52, 5640 – 5643; Angew. Chem. 8196 – 8199; d) P. A. Chase, T. Jurca, D. W. Stephan, Chem. 2013, 125, 5750 – 5754; c) S. Freitag, K. M. Krebs, J. Henning, J. Commun. 2008, 1701 – 1703; e) D. Chen, J. Klankermayer, Hirdler, H. Schubert, L. Wesemann, Organometallics 2013, 32, Chem. Commun. 2008, 2130 – 2131; f) A. E. Ashley, A. L. 6785 – 6791; d) F. A. Tsao, L. Cao, S. Grimme, D. W. Stephan, J. Thompson, D. OHare, Angew. Chem. Int. Ed. 2009, 48, 9839 – 9843; Angew. Chem. 2009, 121, 10023 – 10027; g) K. Cherni- Am. Chem. Soc. 2015, 137, 13264 – 13267; e) Y. Yu, J. Li, W. Liu, chenko, . Madarsz, I. Ppai, M. Nieger, M. Leskel, T. Repo, Q. Ye, H. Zhu, Dalton Trans. 2016, 45, 6259 – 6268; f) K. M. Nat. Chem. 2013, 5, 718 – 723; h) M.-A. Lgar, M.-A. Courte- Krebs, J.-J. Maudrich, L. Wesemann, Dalton Trans. 2016, 45, 8081 – 8088. manche, . Rochette, F.-G. Fontaine, Science 2015, 349, 513 – + 516. [14] The use of “R3Sn ” LAs need not present excessive safety [2] For reviews of FLP chemistry, see: a) “Frustrated Lewis Pairs”: concerns. While certain organotin(IV) species are highly toxic = G. Erker, D. W. Stephan, Topics in Current Chemistry, Vols. I and (notably for R Me), this is highly dependent on the organic II, Springer, Berlin, 2013; b) D. W. Stephan, Acc. Chem. Res. skeleton. See, for example: T. Gadja, A. Jancs, Met. Ions Life 2015, 48, 306 – 316; c) D. W. Stephan, G. Erker, Angew. Chem. Sci. 2010, 7, 111 – 151, and references therein. Int. Ed. 2015, 54, 6400 – 6441; Angew. Chem. 2015, 127, 6498 – [15] Z. M. Heiden, A. P. Lathem, Organometallics 2015, 34, 1818 – 6541; d) D. W. Stephan, J. Am. Chem. Soc. 2015, 137, 10018 – 1827. 10032. [16] For example, see: a) K. Kamiura, M. Wada, Tetrahedron Lett. [3] For reviews of FLP-catalyzed hydrogenation, see: a) D. W. 1999, 40, 9059 – 9062; b) T. X. Yang, P. Four, F. Guib, G. Stephan, S. Greenberg, T. W. Graham, P. Chase, J. J. Hastie, Balavoince, New J. Chem. 1984, 8, 611 – 614, and references S. J. Geier, J. M. Farrell, C. C. Brown, Z. M. Heiden, G. C. therein.

Angew. Chem. Int. Ed. 2016, 55, 14738 –14742 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org 14741 Angewandte Communications Chemie

[17] C. G. Arnold, A. Weidenhaupt, M. M. David, S. R. Mller, S. B. spectrum, and a single broad resonance at around 30 ppm in the Haderlein, R. P. Schwarzenbach, Environ. Sci. Technol. 1997, 31, 119Sn{1H} spectrum. A comparable 119Sn{1H} signal can be 2596 – 2602. obtained by mixing [1]OTf, collidine and 5a (10 equiv) at the [18] V. Sumerin, F. Schulz, M. Nieger, M. Leskel, T. Repo, B. Rieger, same concentration in DCB. Angew. Chem. Int. Ed. 2008, 47, 6001 – 6003; Angew. Chem. [28] L. E. Longobardi, C. Tang, D. W. Stephan, Dalton Trans. 2014, 2008, 120, 6090 – 6092. 43, 15723 – 15726. [19] A similar strategy has been documented for silyium species: [29] a) I. Kaljurand, A. Ktt, L. Soovli, T. Rodima, V. Memets, I. a) B. Mathieu, L. Ghosez, Tetrahedron 2002, 58, 8219 – 8226; Leito, I. A. Koppel, J. Org. Chem. 2005, 70, 1019 – 1028; b) J. F. b) T. J. Herrington, B. J. Ward, L. R. Doyle, J. McDermott, Coatzee, G. R. Padmanabhan, J. Am. Chem. Soc. 1965, 87, 5005 – A. J. P. White, P. A. Hunt, A. E. Ashley, Chem. Commun. 2014, 5010. 50, 12753 – 12756. [30] H. J. Campbell, J. T. Edward, Can. J. Chem. 1960, 38, 2109 – 2116. [20] I. Zharov, B. T. King, Z. Havlas, A. Pardi, J. Michl, J. Am. Chem. [31] J. M. Blackwell, W. E. Piers, R. McDonald, J. Am. Chem. Soc. Soc. 2000, 122, 10253 – 10254. 2002, 124, 1295 – 1306. [21] a) U. Mayer, V. Gutmann, W. Gerger, Monatsh. Chem. 1975, 106, [32] If this reaction mixture is subsequently heated to 1208C very 1235 – 1257; b) M. A. Beckett, G. C. Strickland, J. R. Holland, slow formation of 5a is observed. In light of other results, this

K. S. Varma, Polymer 1996, 37, 4629 – 4631; c) G. C. Welch, L. reduction is most likely mediated by the [1]OTf formed upon H2 Cabrera, P. A. Chase, E. Hollink, J. D. Masuda, P. Wei, D. W. release from [1]H/Col·HOTf. Stephan, Dalton Trans. 2007, 3407 – 3414. [33] The reduced rates of turnover with 2,6-lutidine and DABCO

[22] A. E. Ashley, T. J. Herrington, G. G. Wildgoose, H. Zaher, A. L. relative to collidine can be attributed to less favorable H2 Thompson, N. H. Rees, T. Krmer, D. OHare, J. Am. Chem. Soc. activation, and H+ transfer, respectively (see the SI). Evidence 2011, 133, 14727 – 14740. that collidine is a slightly stronger base than [1]OiPr comes from [23] I. Kaljurand, A. Ktt, L. Soovli, T. Rodima, V. Memets, I. the observation that addition of collidine to a [1]OTf/5amixture Leito, I. A. Koppel, J. Org. Chem. 2005, 70, 1019 – 1028. leads to an upfield shift in the LA 1Hand119Sn resonances and [24] W. Kitching, H. A. Olszovy, G. M. Drew, Organometallics 1982, downfield shifts for the collidine 1H resonances, consistent with 1, 1244 – 1246. deprotonation. Similar shifts in the collidine resonances are [25] a) S. Tussing, L. Greb, S. Tamke, B. Schirmer, C. Muhle-Goll, B. observed during catalytic reactions (see the SI). Luy, J. Paradies, Chem. Eur. J. 2015, 21, 8056 – 8059; b) S. [34] We are aware of a single example of heterogeneous Sn-catalyzed Tussing, K. Kaupmees, J. Paradies, Chem. Eur. J. 2016, 22, 7422 – hydrogenation: S. Nishiyama, T. Kubota, K. Kimura, S. Tsuruya, 7426. M. Masai, J. Mol. Catal. A 1997, 120, L17 – 22. [26] Interestingly, while this reaction proceeds initially very rapidly at RT (reaching almost 50% conversion within minutes), the rate drops quickly (e.g. only 65% conv. after 17 h), presumably due to deprotonation of remaining 2a·HOTf by the more basic 3a. Received: July 8, 2016 [27] RT NMR spectroscopy on the final reaction mixture shows only Revised: August 5, 2016 one set of Sn-iPr resonances and no unassigned peaks in the 1H Published online: October 24, 2016

14742 www.angewandte.org 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2016, 55, 14738 –14742 Appendices

Appendix D – Reference summary of numbered structures

251

Appendices

252