Mechanistic Studies on Ruthenium-Catalyzed Hydrogen Transfer Reactions

Jenny B. Åberg ©Jenny B. Åberg, Stockholm 2009 Cover picture: Experimental set-up for in situ FT-IR measurements. Photography: Tomas Åberg

ISBN 978-91-7155-862-6

Printed in Sweden by US-AB, Stockholm 2009 Distributor: Department of Organic Chemistry, Stockholm University

”No man is an island” – John Donne –

Abstract

Mechanistic studies on three different ruthenium-based catalysts have been performed. The catalysts have in common that they have been employed in hydrogen transfer reactions involving alcohols and ketones, amines and imines or both. 5 Bäckvall’s catalyst, η -(Ph5C5)Ru(CO)2Cl, finds its application as racemization catalyst in dynamic kinetic resolution, where racemic alcohols are converted to enantiopure acetates in high yields. The mechanism of the racemization has been investigated and both alkoxide and alkoxyacyl intermediates have been characterized by NMR spectroscopy and in situ FT-IR measurements. The presence of acyl intermediates supports a mechanism via CO assistance. Substantial support for coordination of the substrate during the racemization cycle is provided, including exchange studies with both external and internal potential ketone traps. We also detected an unexpected alkoxycarbonyl complex from 5-hydroxy-1-hexene, which has the double bond coordinated to ruthenium. Shvo’s catalyst, [Ru2(CO)4(μ-H)(C4Ph4COHOCC4Ph4)] is a powerful catalyst for transfer hydrogenation as well as for dynamic kinetic resolution. The mechanism of this catalyst is still under debate, even though a great number of studies have been published during the past decade. In the present work, the mechanism of the reaction with imines has been investigated. Ex- change studies with both an external and an internal amine as potential traps have been performed and the results can be explained by a stepwise inner- sphere mechanism. However, if there is e.g. a solvent cage effect, the results can also be explained by an outer-sphere mechanism. We have found that there is no cage effect in the reduction of a ketone containing a potential internal amine trap. If the mechanism is outer-sphere, an explanation as to why the solvent cage effect is much stronger in the case of imines than ketones is needed. Noyori’s catalyst, [p-(Me2CH)C6H4Me]RuH(NH2CHPhCHPhNSO2C6H4- p-CH3), has successfully been used to produce chiral alcohols and amines via transfer hydrogenation. The present study shows that the mechanism for the reduction of imines is different from that of ketones and aldehydes. Acidic activation of the imine was found necessary and an ionic mechanism was proposed.

List of publications

This thesis is based on the following papers, which will be referred to by their Roman numerals I–VI.

I Racemization of Secondary Alcohols Catalyzed by Cyclopentadienylruthenium Complexes: Evidence for an Alkoxide Pathway by Fast β-Hydride Elimination– Readdition Martín-Matute, B.; Åberg, J. B.; Edin, M.; Bäckvall, J. -E. Chem. Eur. J. 2007, 13, 6063–6072.

II CO Assistance in Ligand Exchange of a Ruthenium Racemi- zation Catalyst: Identification of an Acyl Intermediate Åberg, J. B.; Nyhlén, J.; Martín-Matute, B.; Privalov, T.; Bäck- vall, J. -E. Submitted for publication

III Unexpected Formation of a Cyclopentadienylruthenium Alkoxycarbonyl Complex with a Coordinated C=C Bond Åberg, J. B.; Warner, M. C.; Bäckvall, J. -E. Manuscript

IV Mechanistic Study of Hydrogen Transfer to Imines from a Hydroxycyclopentadienyl Ruthenium Hydride. Experimental Support for a Mechanism Involving Coordination of Imine to Ruthenium Prior to Hydrogen Transfer Samec J. S. M.; Éll, A. H.; Åberg, J. B.; Privalov, T.; Eriksson, L.; Bäckvall, J. -E. J. Am. Chem. Soc. 2006, 128, 14293–14305.

V Investigation of a Possible Solvent Cage Effect in the Reduc- tion of 4-Aminocyclohexanone by a Hydroxycyclopentadienyl Ruthenium Hydride Åberg, J. B.; Bäckvall, J. -E. Chem. Eur. J. 2008, 14, 9169–9174.

VI Mechanistic Investigation on the Hydrogenation of Imines by [p-(Me2CH)C6H4Me]RuH(NH2CHPhCHPhNSO2C6H4-p- CH3). Experimental support for an Ionic Pathway Åberg, J. B.; Samec, J. S. M.; Bäckvall, J. -E. Chem. Commun. 2006, 26, 2771–2773.

Reprints were made with kind permission from the publishers.

Contents

Introduction ...... 13 1.1 Reaction mechanisms...... 13 1.2 Catalysis ...... 13 1.3 Mechanistic investigations...... 14 1.4 Hydrogen transfer ...... 14 1.4.1 Direct hydrogen transfer...... 15 1.4.2 The hydridic route ...... 15 1.4.3 The monohydridic vs. dihydridic route ...... 16 1.4.4 Inner-sphere vs. outer-sphere mechanism ...... 16 1.5 Chirality ...... 19 1.6 Dynamic kinetic resolution ...... 20 1.7 Content and objectives of this thesis...... 21

Bäckvall’s catalyst (Papers I–III)...... 22 2.1 Introduction ...... 22 2.2 Results and discussion ...... 25 2.2.1 New acyl intermediate...... 25 2.2.2 Detection of the acyl intermediate by NMR spectroscopy...... 27 2.2.3 Secondary alkoxide complex formation and ligand substitution reactions...... 28 2.2.4 Mechanism of the alkoxide substitution ...... 30 2.2.5 β-Hydride elimination and formation of ketone intermediates ...... 32 2.2.6 Kinetic isotope effects ...... 35 2.2.7 Unexpected formation of a cyclopentadienylruthenium alkoxycarbonyl complex with a coordinated C=C bond...... 36 2.3 Conclusions ...... 40

Shvo’s catalyst (Papers IV–V) ...... 41 3.1 Introduction ...... 41 3.1.1 Proposed mechanisms...... 41 3.1.1.1 Inner-sphere mechanisms...... 41 3.1.1.2 Outer-sphere mechanisms ...... 42 3.1.2 Previous mechanistic studies...... 43 3.1.2.1 Studies on kinetic isotope effects ...... 43 3.1.2.2 Electronic nature of the substrate ...... 46 3.1.2.3 Imine reduction in the presence of a potential external amine trap...... 46

3.1.2.4 Imine reduction in the presence of a potential internal amine trap...... 48 3.1.2.5 Need for further investigations...... 50 3.2 Results and discussion ...... 50 3.2.1 Imine reduction in the presence of a potential external amine trap...... 50 3.2.2 Imine reduction in the presence of a potential internal amine trap...... 52 3.2.3 Solvent cage effect...... 52 3.2.4 Mechanistic considerations...... 55 3.2.4.1 Imine reduction in the presence of a potential external amine trap...... 55 3.2.4.2 Imine reduction in the presence of a potential internal amine trap...... 56 3.2.4.3 Solvent cage effect ...... 57 3.3 Conclusions ...... 59

Noyori’s TsDPEN catalyst (Paper VI)...... 61 4.1 Introduction ...... 61 4.2 Results and discussion ...... 63 4.2.1 Mechanistic considerations...... 66 4.3 Conclusions ...... 67

Concluding remarks ...... 68

Acknowledgements ...... 69

References ...... 71

Abbreviations

Abbreviations and acronyms are in agreement with the standard of the sub- ject.1 Only non-standard and unconventional ones that appear in the thesis are listed here.

Cp cyclopentadienyl Cp* 1,2,3,4,5-pentamethylcyclopentadienyl DFT density functional theory DKR dynamic kinetic resolution ee enantiomeric excess k rate constant SNi internal nucleophilic substitution t1/2 half life TsDPEN tosyl-diphenyl-ethylenediamine

Introduction

1.1 Reaction mechanisms A reaction involves the redistribution of the connectivity of atoms within a chemical structure or between structures. One or more reactions constitute the transformation of one molecule into another and the individual steps involved in a reaction, when taken as a set, are called the reaction mechanism.2 The term mechanism is one of those terms frequently used by chem- ists that sometimes has different meanings to the speaker and listener. An- other definition is that it is “a step-by-step description of a chemical transformation at the molecular level”.3 Simply put, one can say that the reaction mechanism describes in detail how the reaction takes place.

1.2 Catalysis A catalyst is a substance that accelerates a reaction by lowering the activation energy without undergoing any net chemical change itself.4 Since the catalyst is not consumed in the reaction, it can be reused, and the use of catalysis has a large potential for cost effectiveness. In addition, catalysis is environmentally friendly, and it is mentioned as one of the 12 principles for green chemistry that were introduced 1998.5 These principles are much em- phasized today as a means to build a more sustainable future.6,7 Catalysis of asymmetric reactions is generating tremendous academic and industrial interest, even if it is not yet widely used in pharmaceutical produc- tion.8–10 Since the use of catalysis in the pharmaceutical industry is still lag- ging behind, there is a great need for new and better performing catalysts (i.e. catalysts with higher activity, better selectivity, productivity and stabil- ity, and technologies that are more robust and easier to implement). This demand for development does not only come from the pharmaceutical industry, but also from the fields of agricultural chemistry, fine chemicals and materials science.9

13 1.3 Mechanistic investigations There are many practical consequences of having a good understanding of the mechanism of a reaction. Once a mechanism has been delineated, a chemist can choose the appropriate reaction conditions or modify the reac- tants in order optimize the reaction. Therefore, the understanding of a reaction mechanism is often the most important tool a chemist can have for manipulating nature. This, in turn, can lead to improved chemical processes that are cheaper, greener, more selective etc. Understanding reaction mechanisms is even claimed to be a basic goal of the science of chemistry.2

1.4 Hydrogen transfer Hydrogen transfer is defined as “the reduction of multiple bonds with the aid of a hydrogen donor in the presence of a catalyst”.11 The catalyst transfers a hydride and a proton from the hydrogen donor DH2, e.g. an alcohol, to an acceptor A, e.g. a ketone, which is reduced (Scheme 1).

Scheme 1. Principle of hydrogen transfer (DH2 = hydrogen donor, A = hydrogen acceptor).

In the 1920s, Meerwein, Ponndorff and Verley independently found that carbonyl compounds were reduced in the presence of 2-propanol (Scheme 2).12 Later, Oppenauer found that secondary alcohols were oxidized in the presence of acetone (Scheme 2).13

Scheme 2. Meerwein-Ponndorff-Verley reduction and Oppenauer oxidation.

In the original hydrogen transfer reaction, Al(Oi-Pr)3 was employed to medi- ate the reaction. Later it was found that transition metals can catalyze hydrogen transfer. Since then, transfer hydrogenation has successfully been used for both imines and ketones.14,15 Transfer hydrogenation has the advantages of operational simplicity and environmental friendliness. Since no hydrogen pressure is used, no special equipment is required. In addition, no hazardous waste is produced, as in the case of stoichiometric reduction by borane re-

14 agents. When 2-propanol is used as hydrogen donor in transfer hydrogenation, the only side product formed is acetone, which is easily removed by distillation during workup. 2-Propanol and formic acid, both often used as hydrogen donors, are non-toxic and also inexpensive. One drawback with 2-propanol is that the reaction is reversible and high yields are therefore sometimes afforded at the expense of enantiomeric excess. This disadvan- tage can be overcome by the use of a formate, since the reaction in this case is irreversible, with loss of CO2. Transfer hydrogenation is well studied:16 two main pathways for its mechanism have been proposed depending on the type of metal used. Direct hydrogen transfer is proposed for main group elements whereas the hydridic route is considered to be the major pathway for transition metals.15

1.4.1 Direct hydrogen transfer In direct hydrogen transfer the hydrogen is transferred directly between the donor and acceptor without involvement of metal hydrides. The mechanism is thought to proceed through a cyclic six-membered transition state (Scheme 3). This mechanism was originally proposed for the Meerwein- 17 Ponndorf-Verley reduction, where Al(Oi-Pr)3 is used as catalyst.

M M O O M O O + O O H + 1 2 H 1 H R R R R1 2 R2 R Scheme 3. Direct hydrogen transfer proceeds through a six-membered cyclic transition state.

1.4.2 The hydridic route The hydridic route proceeds in a stepwise manner by way of a metal hydride, typically formed by β-hydride elimination from a donor such as 2-propanol (Scheme 4).18 The hydrogen is then transferred from the metal to the acceptor, e.g. a ketone. Transition metal catalysts are commonly operating via these kinds of mechanisms.

H [M]-cat. H O O O O + + H 1 2 1 H R R R R2 Donor Acceptor H via [M] H or [M] H Scheme 4. The hydridic route involves a metal hydride.

15 The hydridic route can be further divided into the monohydridic and the dihydridic route.

1.4.3 The monohydridic vs. dihydridic route In the monohydridic route only one of the hydrogens are transferred from the substrate to the metal (Scheme 5, path A), whereas in the dihydridic route two hydrogens are transferred to the metal (Scheme 5, path B). These mechanisms can be distinguished from each other by deuterium labeling,19,20 since the hydrogens loose their identity in the dihydridic route and are scrambled between the carbon and the oxygen (Scheme 5, path B). In the monohydridic route the hydrogens will keep their identity throughout the reaction (Scheme 5, path A).

Scheme 5. Monohydridic (path A) vs. dihydridic (path B) route in transfer hydrogenation.

1.4.4 Inner-sphere vs. outer-sphere mechanism A catalyst operating through the monohydride mechanism can operate in two different ways. Either a metal alkoxide is involved and the hydrogens are transferred in the inner-sphere of the metal, or the hydrogens are transferred in the outer-sphere of the metal, without involvement of a metal alkoxide. The main difference is that the alcohol is coordinated to the metal in the inner-sphere mechanism, and interacts without direct coordination to the metal in the outer-sphere mechanism. In both cases the hydride of the metal hydride migrates from the metal to the carbonyl carbon giving rise to the α- C-H. In the inner-sphere mechanism a coordinated carbonyl inserts into the metal hydride 1 via -coordination of the double bond (Scheme 6). Metal alkoxide 2 is formed, and after ligand exchange with 2-propanol, the product is released. Finally, metal hydride 1 is re-generated by β-hydride elimination.

Scheme 6. Catalytic cycle via a metal alkoxide (inner-sphere mechanism).

There is ample support for the alkoxide mechanism given in Scheme 621 and there are many examples of transition metal hydrides that are obtained from β-hydride elimination of the corresponding alkoxide complexes.22 In the outer-sphere mechanism, the reaction proceeds through the outer sphere of the metal without coordination of the alcohol to the metal. This kind of mechanism was first proposed by Noyori et al. for his TsDPEN catalyst 3 (Scheme 7).23 The alkoxide intermediate mechanism could be ruled out since the reaction could be performed without base.24

Scheme 7. Concerted outer-sphere mechanism via six-membered cyclic transition state A.

The mechanism is proposed to start with conversion of precatalyst 3 to coordinatively unsaturated ruthenium species 4 by base promoted elimination of HCl. The active ruthenium hydride 5 is then generated by concerted hydride and proton transfer from 2-propanol to 4, via a cyclic six-membered transition state A. The ketone is converted to a chiral alcohol in the same way, through simultaneous transfer of the hydride from ruthenium and the proton from the amine ligand of 5. Catalysts proposed to operate through this mechanism are often called metal-ligand bifunctional catalysts. The hydrogen transfer in an outer-sphere mechanism may proceed not only in a concerted manner (as for Noyori’s catalyst) but also in two discrete steps, where the protonation of the substrate precedes the hydride transfer to the metal. Norton and Bullock first proposed these types of mechanisms which are usually referred to as ionic mechanisms for different types of transition metal catalysts.25 The proposed mechanisms are slightly different, depending on the substrate. For imines, the catalytic cycle starts with formation of the active hydrogenation catalyst 6 by deprotonation of the metal dihydride 6a by the product amine or substrate imine (Scheme 8).

Scheme 8. Catalytic cycle of the ionic hydrogenation of imines.

The metal hydride is suggested to be delivered without prior coordination of the double bond to produce a free amine (or alcohol) and an unsaturated catalyst 6b. Hydrogen gas finally regenerates dihydride 6a and completes the cycle.

1.5 Chirality A chiral molecule is defined as “a molecule that cannot be superimposed on its mirror image” and two chiral molecules that are each other’s mirror im- ages are called enantiomers.26 Enantiomers have identical chemical and physical properties except when they are in a chiral environment and interact with other chiral molecules. This is why they usually have very different biological properties. For example (R)-carvone has the flavor of spearmint whereas (S)-carvone has the flavor of caraway (Figure 1).27

Figure 1. The two enantiomers of carvone have different flavors.

The importance of chirality is clear: about 70% of the new small molecule drugs the Food & Drug Administration approved in 2007 contained at least one chiral center.28 In addition, in nine of the top ten best selling drugs of 2004 the active ingredients were chiral.29 Out of these, five were used only as single enantiomers. The demand for enantiomerically pure compounds is increasing not only for use in pharmaceuticals and flavor and aroma chemicals but also for agricultural chemicals and specialty materials.10 There are many ways to prepare enantiopure compounds: methods based upon the chiral pool30 and resolution of racemates,31 which are the tradi- tional ones, and asymmetric synthesis.32 Resolution is often limited to a maximum yield of 50%, since the unwanted isomer is wasted, and methods based upon the chiral pool are limited to naturally occurring non-racemic compounds (e.g. sugars and amino acids).

1.6 Dynamic kinetic resolution Two of the ruthenium catalysts that have been investigated in this thesis have been used in combination with an enzyme for dynamic kinetic resolution (DKR). The role of the catalyst is to racemize the starting material, e.g. an alcohol. The enzyme reacts with only one of the enantiomers, and without the racemization catalyst the theoretical maximum yield is 50%. However, the enzyme in combination with the racemization catalyst is a powerful method that overcomes this limitation and leads to a possible yield of 100% (Scheme 9).33

Scheme 9. Principle of DKR of secondary alcohols.

1.7 Content and objectives of this thesis Mechanistic studies constitute a powerful tool in order to improve existing chemical processes and develop new ones. There is a constant need for better processes that are more efficient, robust and selective. In addition, high productivity and low cost is desired. Moreover, there is an increasing demand for sustainable and environmentally friendly processes, and in this respect catalysis has a great potential. In this thesis, mechanistic investigations have been performed on three different ruthenium-based catalysts. The catalysts have all been used for hydrogen transfer reactions involving alcohols and ketones, amines and imines or both.34 One of the catalysts has been used for asymmetric transfer hydrogenation and the other two for dynamic kinetic resolution. In these processes, enantiomerically pure alcohols and amines are obtained. Chirality is getting an increasing importance, and there is also a need for optimizing existing and developing new methods for the preparation of enantiomerically pure compounds. The objective of this work has been to gain a mechanistic understanding of three different but related catalysts, and thereby enabling improvement of the processes in which they are involved.

21 Bäckvall’s catalyst (Papers I–III)

2.1 Introduction Catalytic racemization of secondary alcohols mediated by transition metal complexes has attracted much attention in recent years.20,35,36 Ruthenium complexes 7–10 (Figure 2) are efficient racemization catalysts that have successfully been combined with an in situ enzyme-catalyzed kinetic resolution (KR) leading to deracemization of alcohols via dynamic kinetic resolution (DKR).35,37–41

PhO O Ph H H Ph N Ph Ph Ph Ph Ph Ph Ph Ph Ph Ph Ph Ph Ph Ph Ph H Ph Ru Ru Ph Ru Ph Ru Ph Ru Cl X Cl OC CO OC OC OC CO CO CO CO PPh3

7 8 9a X=Cl 10 9b X=Br 9c X=I Figure 2.

Our group has developed a DKR, where complexes 9a–b combined with a lipase and an acyl donor produce enantiopure (R)-acetates in high yields and excellent ee´s (Scheme 10).39–41 In addition, catalyst 9a has been used for dynamic kinetic asymmetric transformations (DYKAT) of 1,2-,42 1,3-,43 1,4-44 and 1,5-diols,45 leading to enantiomerically and diastereomerically pure diacetates.

OH OAc 9a (5 mol%), t-BuOK (5 mol%)

CALB, Na2CO3, toluene, rt, 3 h OAc rac-11 95% (>99% ee) Scheme 10. DKR of 1-phenylethanol (11) (CALB = Candida antarctica lipase B).

The proposed mechanism for racemization of sec-alcohols starts with activation of catalyst 9a by t-BuOK, leading to formation of tert-butoxide complex

22 12 (Scheme 11, step i). Ruthenium tert-butoxide complex 12 was characterized by 13C NMR spectroscopy already in 2004.39 It was shown to be an active intermediate by addition of (S)-1-phenylethanol ((S)-11), which was immediately racemized, as confirmed by GC analysis.39

Scheme 11. Proposed mechanism for racemization of secondary alcohols by 9a.

The racemization begins with an alkoxide exchange (Scheme 11, step ii) where the tert-butoxide ligand of complex 12 is exchanged by a sec-alcohol, resulting in complex 13. The alkoxide substitution was initially studied by the addition of tert-amyl alcohol to tert-butoxide complex 12 in toluene-d8 at 40 °C.40 Even at this low temperature tert-amyl alkoxide complex started to build up within 10 minutes, which indicates a very fast alkoxide substitution at room temperature. The alkoxide substitution is followed by β-hydride elimination and it was proposed that η3-keto-hydride species 14 is formed (Scheme 11, step iii).40 Reversible insertion can occur from either face of the prochiral ketone,

23 which produces a racemic alcohol (Scheme 11, step iv). The alcohol is then released after subsequent ligand exchange (Scheme 11, step v). The proposed mechanism has been supported by several experiments, e.g. racemization of (S)-1-phenylethanol ((S)-11) by η5-ruthenium hydride 15 in the presence of acetophenone.46 This reaction showed an induction period of two hours (Figure 3).40 From this experiment it is obvious that η5-hydride 15 can not be an abundant species in the catalytic process.

100

ee (%) 40

0 01234 Time (h)

Figure 3. Racemization of (S)-1-phenylethanol by Ru-hydride 15.

Further evidence was obtained by racemization of (S)-11 by ruthenium catalyst 9a in the presence of p-tolyl methyl ketone, which led to <1% of 1-(p- tolyl)-ethanol. If the newly generated ketone would not stay coordinated to ruthenium after -hydride elimination, a much higher degree of exchange would have been expected.40 Based upon these results, and since η5-ruthenium hydride 15 is not a likely intermediate in the catalytic process it was proposed that the substrate stays coordinated throughout the racemization cycle. Since a free coordination site is required for the β-hydride elimination to occur, an η5 → η3 ring slippage to give η3-keto-hydride complex 14 was proposed (Scheme 11, step iii).40 Recent computational studies in our group have shown that it is very likely that both the activation (Scheme 11, step i) and alkoxide exchange

24 (Scheme 11, step ii) take place via CO assistance.47 A CO group assists the ligand exchange by interacting with the incoming alkoxide/alcohol, e.g. an acyl intermediate B was computationally identified in the activation of ruthenium catalyst 9a (Figure 4).47

Figure 4. Computationally detected acyl intermediate B.

During the computational studies it was also found that η5 → η3 ring slip is endothermic by ca. 25 kcal/mol. The formation of η3-keto-hydride complex 14 has an activation energy of 42 kcal/mol. Therefore this pathway seems less likely. A pathway where racemization is enabled by loss of one CO group had an activation barrier of ca. 23 kcal/mol and seemed more likely.47

2.2 Results and discussion

2.2.1 New acyl intermediate We have observed the computationally detected acyl intermediate B (Scheme 12) by in situ FT-IR measurements and NMR spectroscopy.

Scheme 12. Acyl intermediate B was observed in the activation of catalyst 9a.

In our previous in situ FT-IR measurements, intermediate B was not detected.48 However, it was difficult to obtain conclusions from these experiments, since the experimental setup did not prevent fast decomposition of the intermediates. In our current experiments, we have used a proper adaptor

25 with NS-fitting for the probe, and run the reaction under constant argon flow or constant argon pressure. Under these conditions, decomposition was not a problem. Ruthenium chloride 9a has CO peaks at 2049 and 2001 cm-1 (symmetric and asymmetric stretch) and similarly, tert-butoxide complex 12 has two CO peaks 2021 and 1964 cm-1. Acyl intermediate B has two characteristic peaks at 1933 cm-1 (CO) and 1596 cm-1 (acyl, COOt-Bu).49 The shift towards lower wavenumber of the CO group (1933 cm-1) is consistent with the negative charge of the adduct.50 When the ruthenium center becomes more electron rich the metal-to-CO π* backbonding becomes more important, and the C=O bond becomes weaker. The CO stretching vibration of the acyl peak (1596 cm-1) correlates well with the stretching vibrations of similar structures in the literature.50–53 Intermediate B was formed within 15 s and its lifetime was <3 min at room temperature (Figure 5).

Figure 5. Surface view of in situ FT-IR measurements at rt shows acyl intermediate B.

The acyl peak of intermediate B was possible to detect only under very dry conditions (Figure 6).54 Intermediate B was formed rapidly even at low temperature, and at –78 °C it was formed within 3 min. At low temperature it was also stable for prolonged time; at –55 °C it was stable for at least 3 h.

Figure 6. FT-IR spectra before (black) and after (grey) the addition of t-BuOK at –55 °C. The acyl peak at 1596 cm-1 is clearly visible.

When the reaction mixture was allowed to reach room temperature, tert- butoxide complex 12 was immediately formed. At 0 °C the transformation of B into complex 12 was slow (t1/2 = 26 min), and at –20 °C it was even slower.

2.2.2 Detection of the acyl intermediate by NMR spectroscopy We attempted to study acyl intermediate B by NMR spectroscopy. However, this was complicated due to the low solubility of catalyst 9a in toluene and the low viscosity of toluene at low temperatures. The sample was prepared by pre-forming intermediate B at –78 °C and quickly transferring the solution containing B via a cannula to an NMR tube. The reaction mixture was then frozen (–196 °C) and allowed to thaw inside a probe of a pre-cooled NMR spectrometer. The problem of the low solubility, which made it difficult to get good signal to noise ratio of the CO peaks, was circumvented by 13CO labeling of ruthenium chloride 9a. Acyl intermediate B was detected at –32 °C (Scheme 13).

Scheme 13. 13C NMR shifts of catalyst 9a, acyl intermediate B, and tert-butoxide complex 12 in toluene-d8 at –32 °C.

The CO peaks of acyl intermediate B appear at δ 209.5 (CO) and 208.7 ppm (acyl). The peak of the CO group is shifted downfield. This can be compared to the chemical shift reported for other anionic complexes such as + – 55 [PPN] [Ru3(CO)11(CO2CH3)] (16). The CO groups of complex 16 appear 53 at δ 208.1 ppm in THF-d8. The acyl peak is appearing at unusually high chemical shift, compared to similar structures in the literature.51,53 For example, the acyl peak of ruthenium complex 16 appears at δ 189 ppm.53 How- ever, none of the previously known compounds is having a chloride ligand bound to ruthenium. We have calculated the 13C NMR shifts of intermediate B by density functional theory (DFT) calculations and found that the acyl group is indeed expected to end up much more downfield than in the previously known compounds.56 When the temperature was raised to 0 °C, acyl intermediate B was slowly transformed into tert-butoxide complex 12. At room temperature this transformation was even faster.

2.2.3 Secondary alkoxide complex formation and ligand substitution reactions The secondary alkoxide complex 13 was characterized by 1H and 13C NMR spectroscopy at – 61 °C (Scheme 14).

Scheme 14. Characterization of sec-alkoxide complex 13, structures labeled with 13C NMR shifts. tert-Butoxide complex 12 was pre-formed in an NMR tube in toluene-d8 at room temperature and alcohol 11 (1.1 equiv.) was added at –196 °C. The NMR tube was transferred to the –61 °C pre-cooled probe of the NMR spectrometer.

The fact that the CO groups show two different resonances in 13C NMR (δ 199.7 and 201.1 ppm, respectively) reflects their diastereotopic character and confirms that the chiral alkoxide ligand is bound to ruthenium. In 1H NMR, the coordination to ruthenium affects the chemical shift of the -CH proton, which appear at δ 5.07 ppm. This corresponds to a shift 0.3 ppm downfield compared to free alcohol 11 (δ 4.8 ppm). The 1H NMR spectrum of sec-alkoxide complex 13 also revealed that there is an equilibrium between 1-phenylethanol (11) and the alkoxide ligand. The interaction between the equilibrating alcohol and alkoxide was proposed to take place through hydrogen bonding (C, Figure 7).57

Ph Ph

Ph Ph Ph Ph Ru O OC H OC O

Ph C Figure 7. Ruthenium alkoxide 13 with hydrogen-bonded 1-phenylethanol 11 (C).

In addition to ruthenium sec-alkoxide complex 13, ruthenium alkoxide complexes of rac-1-(p-methoxyphenyl)ethanol (rac-17) and (S)-1-(p- bromophenyl)ethanol ((S)-18) were characterized at low temperature. These sec-alkoxide complexes also had diastereotopic CO groups, confirming that the chiral alkoxide is coordinated to ruthenium. All secondary alkoxide complexes were formed fast (the formation of 13 was observed by 1H NMR after less than 5 min at 61 ºC). The complex from (S)-18 was worked up and GC analysis showed complete racemization of the alcohol.

29 The alkoxide substitution was also studied by in situ FT-IR. The alkoxide exchange is very fast at room temperature, and no intermediate similar to acyl intermediate B could be observed (Figure 8, cf. Figure 5).

Figure 8. Surface view of in situ IR measurements at rt shows no intermediate in the alkoxide exchange.

At –78 °C the tert-butoxide ligand of complex 12 is not exchanged by phenylethanol for at least 1 h. However, at –50 °C the reaction slowly takes place (t1/2 = 15 min). The rate of disappearance of tert-butoxide complex 12 (2021 and 1964 cm-1) is the same as the rate of formation of sec-alkoxide complex 13 (2026 and 1971 cm-1) and within 2.5 h the reaction is complete. The concentration of this experiment with respect to ruthenium is 12 mM, which is lower (ca. 5 times) than in the NMR studies on the alkoxide exchange (Scheme 14), and therefore it is not surprising that the reaction is slower. No intermediate could be detected at low temperature. The failure to detect an intermediate in the alcohol-alkoxide exchange is in accordance with the low computed barrier of activation for this reaction via CO partici- pation (12 kcal/mol).47

2.2.4 Mechanism of the alkoxide substitution There are two principal mechanisms for ligand substitution reactions: (i) a dissociative pathway, and (ii) an associative pathway.58 A dissociative pathway was considered unlikely, since enantiopure acetates are produced under

30 DKR conditions (Scheme 10). Dissociation to free alkoxide would have led to chemical acylation and low ee. The observed hydrogen bonded species C (Figure 7) clearly indicates that the incoming alcohol interacts with the alkoxide through hydrogen bonding.57,59 Two possible mechanisms, of which one is associative, have been considered for the alkoxide exchange reaction (Scheme 15, path A and B).

Scheme 15. Possible mechanisms for the alkoxide substitution.

In both mechanisms the incoming alcohol interacts with sec-alkoxide complex 13 through hydrogen bonding (C). In the associative mechanism (Scheme 15, path A), the incoming alcohol coordinates to the ruthenium center of C via 5  3 ring slippage. In the other mechanism (Scheme 15, path B) the hydrogen bonded alkoxide on the ruthenium center dissociates with concomitant formation of a Ru-O bond with the incoming alcohol. This pathway is analogous to the SNi-type mechanism and ring slip is avoided. A recent computational study by our group has suggested that the alcohol- alkoxide exchange more likely takes place in one of two additional ways; either via CO assistance or loss of one carbonyl group.47 Loss of one CO group would enable simultaneous coordination of the alcohol and alkoxide and thereby facilitate the alkoxide exchange. The CO assisted mechanism would involve interaction between the incoming alcohol and one of the CO

31 groups of sec-alkoxide complex 13 (Scheme 16). This interaction makes it possible to exchange the alkoxide ligand without the need for ring slippage or dissociation of a CO molecule.60 A hydrogen bonded intermediate such as C (Figure 7) could also be involved in this mechanism. Intermediate E (Scheme 16) is similar to acyl intermediate B that was detected in the activation of catalyst 9a (Section 2.2.1–2.2.2). The calculated energy barrier between the alkoxide complexes 12 and 13, when the alkoxide substitution takes place via CO assistance, was 12 kcal/mol.47

Ph Ph Ph Ph Ph Ph OH OH Ph Ph ' Ph Ph Ph Ph R R R Ph Ru R Ph Ru O Ph Ru R' O O OC CO OC C H OC CO O O 13 13' R' E Scheme 16. Possible alcohol-alkoxide exchange via CO assistance.

2.2.5 β-Hydride elimination and formation of ketone intermediates Racemization of sec-alcohols must occur via β-hydride elimination and reversible insertion on either face of the prochiral ketone that is produced. β-Hydride elimination from ruthenium alkoxide complex 13 (Scheme 11, step iii) requires a free coordination site on the ruthenium in order to fulfill the 18-electron rule. We have proposed that the most likely pathway for 3 racemization with 9a and 9b involves the formation η -Ph5C5 hydride ketone ruthenium complex 14 from alkoxide complex 13 (Scheme 11, step iii).40 The free coordination site on ruthenium is then formed via η5 → η3 ring slippage of the cyclopentadienyl ring. Phenyl groups on Cp rings are known to favor η5 → η3 ring slip, since they stabilize the η3 intermediate by conjugation.61 However, this only holds for Ph groups that are coplanar with the Cp ring, enabling conjugation. The proposed ring slip is further supported by the recognition that the Cp analogue to 9a (that is the analogue with no phenyl groups) is totally inactive in the racemization of phenylethanol.62 In addition, ring slippage is characteristic for metal complexes with Cp* rings which also 63 have OR or NR2 groups bound to the metal. Other cyclopentadienyl ruthenium complexes that are η3-coordinated have also been characterized by NMR spectroscopy.64 An alternative (ionic) mechanism, in which the alkoxide dissociates followed by hydride transfer from non-coordinated alkoxide to ruthenium giv-

32 ing η5-ruthenium hydride 15 and free ketone, was discarded. Such a mechanism is highly unlikely since η5-hydride 15 needs an induction period of 2.5 h for racemization (Figure 3).40 Furthermore, the free ketone formed would undergo ketone exchange and this was not found when (S)-1- phenylethanol ((S)-11) was racemized in the presence of p-tolyl methyl ketone at room temperature; <1% 1-(p-tolyl)-ethanol was deteted.40 When the reaction was run at 80 °C a much higher degree of exchange product was detected, but this could be explained by an entropy effect. The minimal ketone exchange at room temperature supports the proposed mechanism with coordination of the ketone after β-hydride elimination and it was confirmed by additional experiments. For example, the racemization of a 1:1 mixture of (S)-1-phenylethanol ((S)-11) and (S)-1-phenylethanol-d4 ((S)-11-d4) did not lead to any deuterium scrambling (Scheme 17), which would be the result if exchange occurred.

Scheme 17. Racemization of a 1:1 mixture of (S)-11 and (S)-11-d4.

The lack of ketone exchange in the above experiments could in principle be explained by an alternative mechanism. Similar results would have been obtained if the newly generated ketone would not be coordinated to ruthenium but trapped within a solvent cage, and if diffusion from the solvent cage was slower than re-addition to ruthenium (Scheme 18).65

Scheme 18. Alternative mechanism with slow diffusion from a solvent cage.

33 To find out if such a mechanism is operating, we studied the racemization of substrate (S)-19 which contains a ketone moiety within the same molecule. If the ketone would be trapped in a solvent cage, the competing ketone would still be in the close proximity of ruthenium, and rac-21 would be detected as well as rac-19 (Scheme 19).

Ph Ph Ph O Ph Ph (S)-19 Ph Ph CD3 12 Ph Ru Ph Ph O Ru O t-BuOH OC CO OC OC H CD3 O

D3C O Ph O Ph Ph Ph

Ph Ph O Ph Ph Ph Ru Ph H Ru OC CD3 O CO cage OC CO 15 20 CD3 [Ru] [Ru] O

O OH OH

HO O O + CD CD3 CD3 3 rac-21 rac-19 rac-19 not observed only product observed Scheme 19. Racemization of (S)-19 eliminates a possible mechanism where the uncoordinated ketone is trapped within a solvent cage.

Racemization of (S)-19 (80% D) showed no deuterium scrambling after complete racemization, which took place within 30 min. This can only be explained if the ketone produced after β-hydride elimination from a ruthenium alkoxide stays coordinated to the ruthenium atom. Based upon these 3 observations we have proposed a mechanism involving η -Ph5C5 keto hydride ruthenium complex 14 (Scheme 11, step iii). A recent computational study by our group suggests an alternative mechanism, where the free coordination site is formed by loss of CO.47

34 2.2.6 Kinetic isotope effects In order to investigate if β-hydride elimination (Scheme 11, step iii) is the rate limiting step, we performed kinetic isotope effect measurements. Initial attempts to compare reaction rates by running racemization of the deuterated and non-deuterated substrates in parallel separate experiments were unsuc- cessful. The reaction is very sensitive to moisture, and therefore it was difficult to tell if small rate differences were real or only caused by different amounts of water present in the experimental setup. This uncertainty was eliminated in a competitive experiment, where a 1:1 mixture of (S)-11 and (S)-11--d1 was racemized in one pot. The reaction was followed by GC-MS and the amount of (R)-11 and (R)-11--d1 was measured at around 10% conversion. Similar studies were performed with two other secondary alcohols with different electronic properties (Table 1).

Table 1. Kinetic isotope effects for racemization of secondary alcohols by 9a.a

b c Entry Alcohol R kH/kD 1 (S)-11 H 1.27 ± 0.06 2 (S)-22 CF3 1.45 ± 0.13 3 (S)-17 OMe 1.08 ± 0.17 a 9a (0.1–0.5 mol%) was dissolved in toluene (0.5 mL) under an argon atmosphere. A solution of t-BuOK (0.25 M in THF, 0.3–1 mol%) was added. After 6 min, a solution of a 1:1 mixture of deuterated and non-deuterated alcohol in toluene (0.5 mL) was added. The reaction was quenched with dilute HCl when ca. 10% of the (R)- alcohol had been formed. The reaction mixtures were analyzed using GC-MS spectroscopy. b Stereochemistry refers to starting material. c Average of at least three measurements on each sample.

A small kinetic isotope effect was found for racemization of (S)-11 (kH/kD = 1.27  0.06, Table 1, entry 1). This observation indicates that β-hydride elimination might be the rate-determining step. For electron-deficient (S)-1- (p-trifluoromethyl)ethanol ((S)-22) the isotope effect is higher (kH/kD = 1.45  0.13, Table 1, entry 2), whereas for electron-rich (S)-1-(p- methoxyphenyl)ethanol ((S)-17) a negligible isotope effect was found (kH/kD = 1.08  0.17, Table 1, entry 3). These results show that β-hydride elimination is more rate-determining for electron-poor than for electron-rich alcohols. For electron-rich (S)-17, β-hydride elimination is no longer the slowest step. Since oxidation of electron-rich alcohols is more facile, this step occurs at a higher rate.

35 2.2.7 Unexpected formation of a cyclopentadienylruthenium alkoxycarbonyl complex with a coordinated C=C bond

Dynamic kinetic resolution of 5-hydroxy-1-hexene 23 (Figure 9) does not work well due to slow racemization.44 The reason for the slow racemization might be interference from the double bond with ruthenium. In order to test this hypothesis, we have studied the reaction of alcohol 23 and ruthenium catalyst 9a by NMR spectroscopy and in situ FT-IR measurements. In addition to this, we were hoping that this study would shed light on the racemization mechanism of sec-alkohols by 9a. The coordination of the double bond requires a free coordination site on ruthenium, as also β-hydride elimination does (Section 2.2.5). Therefore, this study might also give information on the mechanism.

5 3 1 6 4 2 23 Figure 9. 5-hydroxy-1-hexene (23). tert-Butoxide complex 12 was pre-formed and allowed to react with alcohol 23 in an NMR tube at room temperature. Two diastereomeric complexes, which according to 1H and 13C NMR spectroscopy have the double bond coordinated to the ruthenium, were observed. The ratio between the complexes was 1:1. Initially, three possible structures were considered (Figure 10); complexes with only one CO ligand (24a–b), complexes with an η3- coordinated cyclopentadienyl ligand (25a–b), and acyl complexes, similar to acyl intermediate B (26a–b, cf. Scheme 12).66 In the two diastereoisomers the methyl bound to the stereogenic α-carbon of the alcohol points either up or down. For simplicity, the diasteromers will be referred to as a and b.

Figure 10. Proposed structures of the alkene complexes of alcohol 23.

In the 1H NMR spectrum, the protons on the more substituted double bond carbon (H2) appeared at δ 4.52 ppm for a and at δ 4.28 ppm for b, respec-

36 tively. This corresponds to an upfield shift of 1.2–1.5 ppm as compared to free alcohol 23 (δ 5.74 ppm), and it is similar to reported structures in the literature.67 Similarly, the shifts of the corresponding carbons (C2) of the two diasteromers are shifted 53–55 ppm upfield, and appear at δ 85.6 (b) and 84.5 ppm (a). The terminal double bond protons (H1 and H1′) are shifted much more upfield and appear as a multiplet containing the protons for both a and b at δ 2.22–2.37 ppm. Likewise, the terminal carbons (C1) appear at δ 21.9 (a) and 19.4 ppm (b), respectively; they are shifted 93–95 ppm upfield. The -CH protons (H5) are shifted downfield, similar to the -CH proton of 1-phenylethanol (11) when it is coordinated to ruthenium in sec- alkoxide complex 13 (Section 2.2.3, Scheme 14). In the 13C NMR spectrum, the -carbon (C5) of both complexes are also shifted downfield, compared to free alcohol. This initially led us to believe that the oxygen of the alcohol was coordinated to ruthenium, and would seem to support structures 24 and 25. However, the -CH protons of 26 are also expected to shift downfield. β-Hydride elimination has been proposed to occur either via η5 → η3 ring slippage40,48 or loss of CO.47 Therefore, we were eager to find out whether the new alkene complexes (a and b) were having one or two remaining CO groups on ruthenium. The 13C NMR spectrum showed three peaks in a ratio 2:1:1 between δ 203–204 ppm (Figure 11). These results could be interpreted in different ways, but we were able to understand the appearance of the 13C NMR spectrum when [13CO] enriched tert-butoxide complex 12 (55% 13C) was used in the reaction with olefin alcohol 23. In this case, coupling 13 between the CO ligands (JCC = ca. 8.2 Hz) can be seen. In addition to these doublets the singlets from the 13CO groups having a neighboring 12CO group were observed. Overall this made the signals look like triplets (Figure 12). The coupling between two CO groups indicates that the alkene complexes have two CO groups each. The appearance of the spectrum could then be explained by the signal at lower field (δ 204.0) corresponding to two over- lapping CO groups.

203.950 203.549 203.480

204.0 ppm Figure 11. 13C NMR of the reaction between complex 12 and alcohol 23.

37 203.958 203.563 203.494

204.0 ppm Figure 12. 13C NMR of the reaction between [13CO] enriched complex 12 and alcohol 23.

2D NMR experiments confirmed that the new alkene complexes (a and b) were having two carbonyl groups each, and this supports that they are not formed via irreversible loss of CO. Thus, the HMBC spectrum68 showed cross peaks between the terminal double bond protons (H6) for both diastereomers and two CO groups each. The absence of free 13CO at δ 184.5 ppm, when the reaction was carried out with [13CO] labeled catalyst 9a, rules out structures 24a–b (Figure 10) as possible structures for alkene complexes a and b. Of the remaining possible structures, the acyl compound (26) is expected to be more stable than the η3-coordinated compound (25) (Figure 10). How- ever, the acyl peak typically appears at around δ ca. 190 ppm in 13C NMR.51,53 The cyclopentadienyl carbons of alkene complexes a and b give only one peak in 13C NMR at room temperature (δ 108.0 ppm for both a and b). If the cyclopentadienyl ring is η3-coordinated (as in 25), the five carbons of the ring are not equivalent, and would have to be in fast equilibrium in order to appear as a singlet. At lower temperature the peak is expected to broaden as the equilibrium is slowed down, and eventually split into 3 non- equivalent peaks in a ratio of 1:2:2. However, when the reaction mixture was cooled down, the peak was only split into two peaks in a 1:1 ratio (at –55 °C they appear at δ 107.5 and 107.6 ppm, respectively). Apparently, the cyclopentadienyl rings of both complexes are η5-coordinated and appear at the same chemical shift at room temperature, but at slightly different chemical shifts at lower temperatures. Therefore, structure 26 seems to bee the only possible structure for alkene complexes a and b. The 13C NMR shift of the acyl carbons (δ 203.50 and 203.56 ppm) is more downfield than similar compounds in the literature, similar to acyl complex B (Section 2.2.2).51,53 The reaction between 5-hydroxy-1-hexene (23) and tert-butoxide complex 12 was also studied by in situ FT-IR measurements. Since the formation of the alkene complexes (a and b) was slow, the interpretation of these measurements was complicated, due to the formation of decomposition

38 products.69 However, when alcohol 23 was added to tert-butoxide complex 12, we were able to observe the formation of two peaks at the same rate as the signal of complex 12 decreased. These peaks appeared at 1982 cm-1 (CO) and 1644 cm-1 (acyl). The peak at 1644 cm-1 appears in the region that is expected for a CO stretching vibration50–53 of an acyl compound (COOR), and additionally supports the alkoxycarbonyl structures 26a–b as the structures of the new alkene complexes (Figure 10). Alkoxycarbonyl complexes 26a–b could be formed via one of two possible routes (Scheme 20). Initially, sec-alkoxide complex 27 is formed from tert-butoxide complex 12 by alkoxide exchange (not shown in Scheme 20, cf. Scheme 11, step ii). The first possible pathway involves the formation of a free coordination site on ruthenium, either via η5 → η3 ring slippage40,48 or dissociation of a CO ligand47 (Scheme 20, step i). In either case, with 5-hydroxy-1-hexene (23), the reaction could proceed in two ways. Either β-hydride elimination (Scheme 20, step ii) occurs, which will lead to racemization of the alcohol (Scheme 20, step iii), or coordination of the double bond to ruthenium (Scheme 20, step iv). Complexes 26a–b would then be formed via a migration of the alkoxy moiety to the carbonyl (Scheme 20, step v). This migratory insertion would explain the easy formation of the large ring (26a–b).70 The second possible pathway for formation of 26a–b would be migration of the alkoxy group to CO in 27 (Scheme 20, step vi), followed by coordination of the double bond (Scheme 20, step vii). Unfortu- nately, no conclusions can be made about how the free coordination site, required for β-hydride elimination is formed (Scheme 20, step i).

Ln Ru racemization -hydride H iii elim. OC O Ph Ph Ln or ii Ph Ph Ru -CO Ln OC olefin Ph Ru O O Ru i H coord. OC CO OC iv O 27

vi 5 3 Ln = Cp, or Cp, CO v or rearrangement Ph Ph +CO Ph Ph Ph Ph Ph Ru Ph Ph OC vii C Ph Ru O O OC C O 26a-b O

Scheme 20. Possible routes for the formation of the diasteromers 26a and 26b.

39 2.3 Conclusions Our mechanistic study provides a detailed picture of the mechanism of racemization of secondary alcohols by pentaphenylcyclopentadienyl complex 9a. The racemization cycle begins with the activation of ruthenium chloride 9a by t-BuOK via CO assistance. t-BuOK is attacking one of the carbonyl ligands of catalyst 9a, and an acyl intermediate B is formed. This intermediate (B) is transformed into the active alkoxide complex 12. The racemization mechanism continues through ruthenium secondary alkoxide complexes, and the alkoxide exchange most probably takes place via hydrogen bonding and CO assistance. β-Hydride elimination from the alkoxide complex was pro- 3 posed to give a  -Ph5C5 hydride ketone complex 14. However, recent computational work has suggested that loss of one CO molecule is more likely than ring slippage.47 Further experimental work is needed to solve this issue. The ketone stays coordinated throughout the catalytic cycle, which was confirmed by several experiments. Racemization of a secondary alcohol bearing a ketone moiety within the same molecule rules out a mechanism in which the newly generated ketone is not coordinated but trapped within a solvent cage. Kinetic isotope effects indicate that β-hydride elimination might be the rate-determining step for electron-deficient alcohols. During our studies we have unexpectedly detected the formation of two alkoxycarbonyl complexes of 5-hydroxy-1-hexene. The formed complexes are diastereomers and have the double bond coordinated to ruthenium, and the oxygen bound to a carbonyl ligand.

40 Shvo’s catalyst (Papers IV–V)

3.1 Introduction In 1985 Shvo found that the dimeric catalyst 7 is an efficient hydrogenation catalyst for various substrates including ketones and aldehydes.71 At elevated temperature the catalyst breaks up into the two monomers 28 and F (Scheme 21).72 Whereas the former monomer (28) is able to hydrogenate a hydrogen acceptor, the latter monomer (F) can dehydrogenate a hydrogen donor. These processes interconvert 28 and F.

Scheme 21. Precatalyst 7 in equilibrium with its active monomers 28 and F.

Complex 7 has been shown to be active in the disproportionation73 of aldehydes to esters and in the hydrogenation of carbonyl compounds to alcohols.72,74 Our group has used catalyst 7 in Oppenauer-type oxidations of al- cohols75 and we and others have used it in combination with lipases for dynamic kinetic resolution (DKR) of primary76 and secondary35,38,77 alcohols and dynamic kinetic asymmetric transformation of diols.78 We have also reported that 7 can be used as a catalyst for the reduction of imines,79 oxidation of amines80 and for the racemization of amines.81,82 A 4-methoxy-phenyl substituted analogue of 7 has successfully been used for DKR of primary amines.82,83

3.1.1 Proposed mechanisms 3.1.1.1 Inner-sphere mechanisms Hydrogen transfer with the Shvo catalyst (7) is of mechanistic interest. Originally, a mechanism with Ru-alkoxide (and analogous Ru-amine) intermediates was proposed by Shvo. This mechanism is similar to the inner- sphere alkoxide mechanism discussed in the introduction (Scheme 6).73

41 Based on Shvo’s original proposal, an inner-sphere mechanism has been proposed by us to operate for the reduction of imines by 28 (Scheme 22).84 The hydrogen transfer (step ii) could be either stepwise or concerted.

Scheme 22. Proposed inner-sphere mechanism for hydrogenation of imines by 28.

In this mechanism the reaction of 28 with the imine starts with an imine- promoted ring slippage to give intermediate G (step i). This step may or may not be an equilibrium depending on the imine employed. The rate of the subsequent hydrogen transfer to form η2-complex H also depends on the nature of the imine; for N-aryl ketimines it is fast and for N-alkyl aldimines it is slow (step ii). Complex H finally rearranges to an η4-ruthenium amine complex (step iii).

3.1.1.2 Outer-sphere mechanisms An outer-sphere mechanism has been proposed by Casey for the reduction of both ketones (aldehydes) and imines by 28′ (the tolyl-derivative of 28, Scheme 23).85 This mechanism is similar to that proposed by Noyori for catalyst 3 (Scheme 7), discussed in the introduction.

Scheme 23. Proposed concerted outer-sphere mechanism for hydrogenation of ketones and imines by 28′.

42 A modified outer-sphere mechanism was later proposed by Casey to operate for imines (Scheme 24).65 In this mechanism the hydrogen transfer is still outer-sphere, but depending on whether the imine is electron-deficient or electron-rich the hydrogen transfer is rate limiting or not, respectively. In the latter case the amine coordination is rate-limiting.

Scheme 24. Modified outer-sphere mechanism for hydrogenation of imines by 28′.

3.1.2 Previous mechanistic studies The mechanism of transfer hydrogenation catalyzed by 7 involving ketones/alcohols and imines/amines has been studied by several groups and is still a matter of discussion. Both experimental studies20,65,84–93 and DFT cal- culations91,94,95 have been performed in order to elucidate the mechanism. In this thesis only the experimental work will be discussed. In 2001, it was shown by our group that the catalyst operates via a monohydride mechanism, since the deuterium is not scrambled when (S)-1-deuterio-1- 20 phenylethanol ((S)-11-α-d1) is racemized by 7. The following sections de- scribe some of the key experiments and how they can be interpreted.

3.1.2.1 Studies on kinetic isotope effects 3.1.2.1.1 Kinetic isotope effects in the case of ketones/alcohols Kinetic studies on the reduction of benzaldehyde by 28′ has been performed by Casey and co-workers, and the large combined isotope effect kRuHOH/kRuDOD = 3.6 together with individual isotope effects of 1.5 (RuD) and 2.2 (OD) is consistent with a concerted transfer of hydride and proton (Scheme 25).85 Based on these studies, an outer-sphere mechanism was proposed for both ketones and imines (Scheme 23, Section 3.1.1.2).

Scheme 25. Kinetic isotope effects of hydrogenation of benzaldehyde by 28′ show that the hydrogen transfer is concerted.

A mechanism involving simultaneous hydride and proton transfer was observed also by our group for the reverse reaction, the dehydrogenation of 2- (4-fluorophenylethyl)ethanol by F under catalytic conditions.88 The mechanism of reduction of ketones and oxidation of alcohols will not be further discussed in this thesis.

3.1.2.1.2 Kinetic isotope effects in the case of imines/amine Kinetic isotope effect studies performed by our group86,87 clearly show that the rate limiting step in the mechanism operating for imines and amines is different from that for ketones (aldehydes) and alcohols.85,89 In the stoichiometric hydrogenation of N-phenyl-1-(p-methoxyphenyl- ethylidene)amine (29) by ruthenium hydride 28 an isotope effect 86 kRuHOH/kRuDOD = 1.05 was observed (Scheme 26). This combined isotope effect is very small, which shows that the rate-determining step is not the hydrogen transfer, in sharp contrast to benzaldehyde (kRuHOH/kRuDOD = 3.6, Scheme 25).

Scheme 26. Kinetic isotope effect of hydrogenation of imine 29 by 28.

For the reverse reaction, the dehydrogenation of N-phenyl-1- phenylethylamine (30) by in situ generated ruthenium species F, a large isotope effect for the cleavage of the C-H bond was found by our group

44 87 (Scheme 27). This individual isotope effect kCHNH/kCDNH = 3.24, is equal (within experimental error) to the combined isotope effect kCHNH/kCDND = 3.26, and is consistent with a stepwise mechanism with rate determining cleavage of the C-H bond.

Scheme 27. Kinetic isotope effects for the dehydrogenation of amine 30 by 7 (2,6- diMeOBQ = 2,6-di-methoxy-benzoquinone).

Later, the isotope effects of hydrogenation by 28′ of a number of imines with different electronic properties were studied by the group of Casey.65 It was found that electron-deficient imines had individual isotope effects in agreement with the combined isotope effect, similar to benzaldehyde (Scheme 25). For these imines the hydrogen transfer therefore must be concerted. In addition it was found that electron-rich imines had small (N-aryl aldimines) isotope effects and even inverse (N-alkyl aldimines) individual isotope effects. The isotope effects could be explained by both an inner-84,86,87 and an outer-sphere65,89 mechanism.

3.1.2.1.3 Explanation in terms of an inner-sphere mechanism In an inner-sphere mechanism, the absence of a kinetic isotope effect in the hydrogenation of ketimine 29 by 7 (1.05) is explained by rate determining imine coordination concurrent with ring slippage (Scheme 22, step i). In this case, the hydrogen transfer occurs after the rate determining step and no isotope effect is expected. For electron deficient aldimines the hydrogen transfer (Scheme 22, step ii) is rate determining and the hydride and proton are transferred concertedly. For electron rich aldimines (N-alkyl aldimines) the η2 → η4 ring slippage is rate determining (Scheme 22, step iii). The significant CH/CD isotope effect in the dehydrogenation of amine 30 is explained by a stepwise hydrogen transfer, where the transition state of the rate determining step has little perturbation of the NH bond (Scheme 22, step ii).

3.1.2.1.4 Explanation in terms on an outer-sphere mechanism In an outer-sphere mechanism, the significant isotope effect of electron- deficient aldimines is explained by concerted, rate determining hydrogen transfer (Scheme 24, step i). The low isotope effect for ketimine 29 and the inverse isotope effects for N-alkyl aldimines which seem inconsistent with a concerted mechanism are explained in terms of subsequent rate determining

45 nitrogen coordination (Scheme 24, step ii). No distinction can then be made between a stepwise or concerted transfer of RuH and OH, since these transfers occur before the rate determining step. Comparable rates of hydrogen transfer and nitrogen coordination (Scheme 24, step i and ii) for N-aryl aldimines explain their negligible isotope effects. The significant CH/CD isotope effect in the dehydrogenation of amine 30 is explained by a transition state for the rate determining step with little perturbation of the NH bond.

3.1.2.2 Electronic nature of the substrate In catalytic hydrogen transfer reactions of imines and amines using Shvo’s catalyst 7, a correlation between the electronic nature of the substrate and the rate of the reaction was observed.79–81,87 In the case of imines, electron-rich N-alkyl imines reacted faster than the corresponding electron-deficient N- aryl imines. Stoichiometric reduction of imines to amines by the hydride monomer 28 leads to formation of a complex with the coordinated amine product.90 A correlation between the electronic properties of a number of ketimines, and the temperature at which the complexation between the imine and ruthenium hydride 28 begins, was found in a study performed by us. This correlation is similar to the correlation with the rate of transfer hydrogenation; complexation with electron-rich imines takes place at lower temperatures than electron-deficient imines (Paper IV). This could be explained in terms of both mechanisms. In an inner-sphere mechanism, the rate determining step for the electron rich ketimines is imine coordination concurrent with ring slippage (Scheme 22, step i). This step is expected to be faster for most electron-rich imines, explaining the higher rate of the reaction (Paper IV). In an outer-sphere mechanism, the increase in rate could be explained by a later transition state due to the higher basicity of the electron rich imine.91

3.1.2.3 Imine reduction in the presence of a potential external amine trap One way to differentiate between an outer- and inner-sphere mechanism is to have another amine, similar to the product amine, present in equimolar amounts. In an outer-sphere mechanism, in which the newly produced amine is not coordinated to ruthenium, the added amine would compete to associate with unsaturated ruthenium species F. This would lead to the formation of two different amine complexes (Scheme 28).

46 R3' HN R3 Ph Ph N Ph Ph R1' 2' O Ph OH Ph O R 1 2 Ph Ph R R H R3 Ph Ph Ph Ph N Ru H R3' Ru Ru OC N OC H 1 2 CO OC OC R R 3 CO R H R2' HN R1' 28 F R1 R2

Ph Ph O Ph Ph Ru H R3 OC N CO H R2 R1 Scheme 28. An external amine is expected to be trapped if the mechanism is outer- sphere.

In an inner-sphere mechanism, where the newly produced amine stays coordinated to ruthenium the added amine should not compete and only one amine complex would be formed (Scheme 29).

Scheme 29. No external amine trapping is expected if the mechanism is inner- sphere.

Both N-alkyl and N-aryl aldimines have been reduced by ruthenium hydride 28′ in the presence of alkyl- and aryl- primary amines by the group of Casey. It was found that the external amine was not incorporated.91 These results are difficult to explain by an outer-sphere mechanism unless the newly generated amine is strongly hydrogen bonded to or trapped within a solvent cage together with unsaturated ruthenium species F (X and Y, respectively) and coordination to ruthenium is faster than diffusion from X and Y (Scheme 30).96

Scheme 30. A solvent cage effect (Y) or strong hydrogen bonding (X) could prevent complex formation with an external amine even if the mechanism is outer-sphere.

3.1.2.4 Imine reduction in the presence of a potential internal amine trap To find out if the lack of exchange product when using an external amine trap (Section 3.1.2.3) is due to a cage effect, where the newly formed amine is trapped within a solvent cage together with ruthenium, one can use a substrate that has an internal amine trap. This was performed by the group of Casey, using 4-NH2-C6H4N=CHPh (31) in the reduction by 28′ (Scheme 31).

N Ph

Ph Ph Ph O Ph O Ph Ph Ph Ph Ph Ph OH NH2 H H Ru Bn Ru Ph N Ph 31 OC + OC NH Ru OC OC OC H OC toluene-d8 28' NH2 HN Bn 50% 50% Scheme 31. Reduction of imine 31, containing an internal amino group as potential trap, by 28′.

Hydrogenation of imine 31 gave a 1:1 mixture of the two possible amine complexes and this was taken as evidence for an outer-sphere mechanism.91 However, since the nitrogen atoms in 31 are connected with a benzene ring,

48 it is also possible that the internal exchange occurs via the π-system of the aromatic ring.97 Interestingly, when imine 31 was reduced by hydride 28′ in the presence of a potential external trap such as 4-isopropylaniline (32) or isopropylamine (33) no product from intermolecular trapping was observed.91 This was explained by a solvent cage effect where the diamine, which is produced by reduction of imine 31, is trapped within a solvent cage. The dissociation from the cage was proposed to be much slower than coordination to ruthenium. An explanation based upon strong hydrogen bonding can not account for these results, since hydrogen bond breakage is required in order to get intramolecular amine trapping (Scheme 32). Furthermore, the internal amino group of imine 31 is very similar to the external amine 32 and these are expected to form hydrogen bonds with about the same strength.

N Ph

Ph Ph Ph Ph Ph Ph Ph Ph 31 O H Bn O Bn O H OH N HN NH NH2 Ph Ph Ph Ph Ph Ph Ph Ph Ru Ru Ru Ru OC H CD Cl OC OC OC OC 2 2 CO CO CO 28' NH NH HN 2 2 Bn cage NH Bn 2 HN Ph Ph Ph O Ph O Ph Ph Ph Ph H 32 H Ru Bn NH2 Ru H OC N OC N OC OC Ph Ph O H NH Ph Ph NH2 Ru HN OC Bn CO 50% 50% Ph Ph O Ph Ph Ru H OC NH OC

not observed Scheme 32. If the mechanism is outer-sphere, a solvent cage effect is the only possible explanation as to why no amine 32 is trapped.

49 3.1.2.5 Need for further investigations The results presented so far (Section 3.1.2.1–3.1.2.4) cannot be used to satis- factorily distinguish between an outer-sphere and inner-sphere pathway. The kinetic isotope effects (Section 3.1.2.1) clearly show that the mechanism in imine reduction is different from that of alcohols (aldehydes). Since the rate- limiting step is not the concerted transfer of hydride and proton, the first outer-sphere mechanistic proposal (Scheme 23) does not adequately account for transfer hydrogenation of imines. However, the inner-sphere mechanism (Scheme 22) and the modified outer-sphere mechanism (Scheme 24) both can explain the results from these measurements. Therefore, other experiments that can distinguish between these mechanistic proposals are required. We have, among other things, used experiments with different amine traps and these results are discussed below.

3.2 Results and discussion

3.2.1 Imine reduction in the presence of a potential external amine trap The reduction of N-methyl-1-(phenyl-ethylidene)amine (34) in the presence of N-methyl-1-p-methoxyphenyl-ethylamine (35) as a potential external amine trap (cf. Section 3.1.2.3) has been studied. This reaction gave the two possible amine complexes in a ratio of 90:10 (Table 2).

50 Table 2. Reaction of 28 and imine 34 in the presence of external amine trap 35.a

Entry Atmosphere Equiv. of amine 35 Ratio of 36:37 1 Argon 1 90:10 2 Argon 0.5 90:10 3 Argon 2 90:10 4 H2 1 100:0 a The reactions were run by adding imine 34 (0.1 mL, 1 M solution in CD2Cl2, 0.1 mmol) and amine 35 (0.1 mL, 0.5/1/2 M solution in CD2Cl2, 0.05/0.1/0.2 mmol) to ruthenium hydride 28 (0.4 mL, 0.25 M solution in CD2Cl2, 0.1 mmol) at 196 C. The solutions were mixed at 78 C and analyzed by 1H NMR at 20 C for 20 min.

The ratio between the two amine complexes did not change with time at 20 °C over 2 h, nor did it change with the concentration of added amine (Table 2, entries 1–3). When the reaction was run under an H2 atmosphere, only the complex with the newly generated amine was formed (36, Table 2, entry 4). The predominant formation of 36, which is the complex between ruthenium and the newly generated amine, supports a mechanism in which the imine is coordinated during hydrogen transfer (Scheme 29). That complex 37 is formed, is seemingly inconsistent with an inner-sphere mechanism but could have several explanations, which are discussed in Section 3.2.4.1. The 10% formation of complex 37 could in principle be explained by a mechanism without coordination of the imine (Scheme 30). In this case the unexpected low amount could be explained by a cage effect or strong hydrogen bonding. However, the ratio between the two amine complexes would be expected to change with concentration of added amine in such a case, and this was not observed.

51 3.2.2 Imine reduction in the presence of a potential internal amine trap A substrate with a potential internal amine trap can be used to investigate if the reason why no or very little (10%) exchange with an external amine occurs is due to a cage effect or not. In order to avoid a possible exchange via an aromatic π-system, an experiment where the nitrogen atoms are connected by a saturated cyclohexane ring instead of a benzene ring, was designed (cf. Section 3.1.2.4). Hydrogenation of 4-(BnNH)-C6H9=NPh (38) by hydride 28 4 at 80 °C afforded only amine complex [2,3,4,5-Ph3(η - C4CO]Ru(CO)2NH(Ph)(C6H10-4-(NHCH2Ph)) (39), where the newly formed amino group is coordinated to ruthenium (Scheme 33).

Scheme 33. Exchange study using an internal amine trap.

4 At higher temperature 39 rearranges to the more stable [2,3,4,5-Ph3(η - C4CO]Ru(CO)2NH(CH2Ph)(C6H10-4-(NHPh)) (40). An analogue of imine 38 having the imine nitrogen 15N marked, was also prepared and the experiment was followed by 15N NMR. The failure to form 40 at low temperature is difficult to explain with the outer-sphere mechanism and only the mechanism where the imine is coordinated to ruthenium seems to be able to give an adequate explanation to these results.

3.2.3 Solvent cage effect

In the experiments with an added external amine trap no, or very little (10%), of this amine was trapped to form an additional ruthenium amine complex (Section 3.1.2.3 and 3.2.1). The failure of the added external amine to form a ruthenium amine complex supports an inner-sphere pathway and would seem to rule out the outer-sphere pathway. However, among other things, a solvent cage effect has been proposed for the outer-sphere pathway to account for these results.89,91,98 The explanation would then be that the

52 newly formed amine would only slowly dissociate out of the cage and subsequent coordination to the metal would be faster than dissociation (Scheme 30). In addition, the 1:1 formation of the possible amine complexes when 4-NH2-C6H4N=CHPh (31) was reduced by 28′ must be explained by a solvent cage effect (Section 3.1.2.4). In the present study a possible cage effect in the reduction of 4-aminocyclohexanone (41) by ruthenium hydride 28 was investigated. Re- duction of aminoketone 41 leads to the formation of 4-aminocyclohexanol (42) and the 16-electron complex F (Figure 13).

Figure 13.

Aminoketone 41 was chosen as substrate, since the alcohol moiety of 42 is not expected to form a stable coordination complex with ruthenium and therefore failure of trapping by an external amine in this case would occur only because of a cage effect (and not because of strong coordination to ruthenium).90 The coordinatively unsaturated complex F can produce three different products (Scheme 34): (i) the Shvo dimer (7), (ii) aminoalcohol complex 43, and (iii) aminoketone complex 44. The second product, aminoalcohol complex 43, is formed as a cis/trans mixture. The amino group of aminoalcohol 42 is acting as an internal amine trap and the amino group of unreacted aminoketone 41 is acting as an external amine trap. Hydride 28 is also acting as an external trap and the ratio of the different products will depend on the magnitude of the cage effect.

Bn HN

Ph Ph Ph Ph Ph O O O O Ph OH O H Ph Ph Ph Ph Ph Ph H Ph Ph 41 Ru Ru H Ph Ph Bn + Bn + Ph Ru o OC N OC N Ph H OC H -20 C OC OC Ph Ru Ru Ph OC CD2Cl2 OC CO OC CO 28 7 43 OH 44 O cis + trans Scheme 34. The products that are formed in the reduction of aminoketone 41 by hydride 28.

53 The formation of all the above mentioned products as well as the disappearance of hydride 28 was monitored by 1H NMR and the results are given in Figure 14.

100,0

90,0

80,0

70,0

60,0

50,0

40,0

30,0

20,0 Relative concentration Relative 10,0

0,0 0 50 100 150 time/min

Figure 14. Relative concentration of Ru-species over time in the reaction between Ru-hydride 28 and aminoketone 41 according to 1H NMR.99 Ru-hydride 28 (0.074 mmol) and aminoketone 41 (0.074 mmol) in CD2Cl2 were mixed at 196 ºC and heated to 20 °C. The reaction was followed by 1H NMR at 20 °C.

From these results it is obvious that there is no strong cage effect. With a strong cage effect aminoalcohol complex 43 would have dominated and no Shvo dimer 7 would have been formed. One could argue that the trapping of F by hydride 28 is much more efficient than the reaction of hydride 28 with the newly formed aminoalcohol 42. However, a control experiment showed that F reacts with N-cyclohexylbenzylamine and hydride 28 with comparable rates. In an attempt to quantify a possible small cage effect the expected ratio (cis+trans)-43:44 (assuming that there is no cage effect) was calculated and compared with the observed ratio (cis+trans)-43:44 (Figure 15).100 The observed ratio was slightly larger than the calculated ratio and this would seem to indicate that there is a small cage effect of about 1.2–1.7. However in the calculations of the ratio (cis+trans)-43:44 the possibility that aminoketone complex 44 is reduced by ruthenium hydride 28 to aminoalcohol complex 43 was not taken into account.101 When this was corrected for,102 the calculated ratio assuming no cage effect and the observed ratio were very similar (Figure 16). From these results it is clear that there is no measurable cage effect in the hydrogen transfer from hydride 28 to aminoketone 41.

54 1

0,9 1,00 0,8 0,90 0,7 0,80 0,6 0,70 0,60 0,5 0,50 0,4 0,40 Ratio 43:44 0,3 0,30 Ratio 43:44 43:44 Ratio

0,2 0,20 0,1 0,10 0 0,00 0 1020304050 0 1020304050 time (min) time (min) Figure 15. Actual (▪) and expected Figure 16. Corrected actual (▪) and (•) ratio between Ru-amine com- expected (•) ratio between Ru-amine plexes 43 and 44. complexes 43 and 44.

3.2.4 Mechanistic considerations 3.2.4.1 Imine reduction in the presence of a potential external amine trap The goal of the performed trapping experiments has been to be able to distinguish between the proposed inner- and outer-sphere mechanisms (Scheme 22 and Scheme 24, respectively). An inner-sphere mechanism readily explains the experiments with a potential external amine trap performed by the group of Casey (Section 3.1.2.3). No incorporation of external amine was found in the reduction of N-alkyl and N-aryl aldimines,91 which is expected with the inner-sphere mechanism, since the substrate is coordinated throughout the reaction. These results can be explained with an outer-sphere mechanism if there is a strong cage effect, i.e. that the newly generated amine is trapped within a solvent cage together with ruthenium (X, Scheme 30), and that diffusion from the solvent cage is slower than coordination to ruthenium. The newly formed amine could also be strongly hydrogen bonded to the cyclopentadienone ring (Y, Scheme 30). If breakage of the hydrogen bond is slower than coordination to ruthenium, this could also account for the absence of external trapping. Imine 34 has been reduced by ruthenium hydride 28 in the presence of the methoxyphenyl substituted amine 35 as a potential external trap (Section 3.2.1, Table 2). The 90% formation of complex 36 (which is the complex from the newly generated amine) can only be explained with an outer-sphere mechanism by strong hydrogen bonding (Y, Scheme 30) or if there is a strong cage effect (X, Scheme 30). That complex 37 is formed is seemingly inconsistent with an inner-sphere mechanism, but could have several explanations, e.g. it could be explained by generation of an unsaturated species such as F in the reaction. This was supported by the fact that if the reaction was run under H2, which would prevent formation of F and other unsaturated species, only complex 36 was formed. It is not clear how complex F would be formed, but loss of H2 is one possibility even though this normally re-

55 quires elevated temperatures.103 Amine exchange from the unsaturated η2-coordinated 16-electron complex H (Scheme 22) might be another explanation as to why exchange product 37 is formed. However, if the mechanism is associative,104 the ratio is expected to be dependent on the concentration of free amine, which is not the case. It has been suggested that the 10% formation of complex 37 might be a result of contamination with 10% acetophenone (the starting material of imine 34).98 In addition to the possibility of contamination, imine hydrolysis during the reaction would generate acetophenone in situ. Reduction of acetophenone by 28 would lead to the unsaturated complex E and 1-phenylethanol (11). In some of our experiments, 11 was indeed detected.105 However, due to low concentration and bad signal to noise ratio, which make it difficult to integrate the 1H NMR peaks accu- rately, it is not possible to say if this could account for all 10% of complex 37 or not.

3.2.4.2 Imine reduction in the presence of a potential internal amine trap Since an outer-sphere mechanism is dependent on a cage effect in order to be able to explain the results above, the experiments with a potential internal trap are highly interesting. The fact that the two possible amine complexes from the reduction of 4-NH2-C6H4N=CHPh (31) by 28′ performed by the group of Casey, were formed in a 50:50 ratio (Section 3.1.2.4), seems to support an outer-sphere mechanism.91 However, this exchange could also take place through the -system of the aromatic ring. That no external amine was incorporated when imine 31 was reduced in the presence of an external amine could only be explained by a solvent cage effect, since strong hydrogen bonding would prevent the formation of the intramolecular exchange product. An outer-sphere mechanism does not easily offer an explanation as to why there is no incorporation of the internal amine when 4-Bn-NH- C6H10=NPh (38) is reduced by 28 (Section 3.2.2, Scheme 33). Here the nitrogens are connected with a saturated cyclohexane ring. These results therefore provide support for an inner-sphere mechanism (Scheme 22). After paper IV was published, the 15N NMR technique was used to follow imine reduction in a similar experiment, performed by the group of Casey. In 15 these studies, 4-(Bn NH)-C6H9=NBn (45) was reduced by ruthenium hy- 98 dride 28′. When the reaction was carried out in CD2Cl2 91% of the amine complex from the newly generated amine (46-RuN,15N) and 9% of the complex where the 15N marked amine has been trapped by ruthenium (47- Ru15N,N) was found (Scheme 35).106 This is about the same ratio as the ratio of the complex from the reduced imine 34 and the complex resulting from trapping of the external amine 35 in our previous study (Table 2).

56 Bn N

Ph Ph Ph Ph Ph O O 15 Tol OH NH Ph Ph Ph Ph Bn H H Tol Ph 46 Ru Bn Ru Bn N 15N Ru o OC + OC OC H -20 C OC OC OC CD2Cl2

29' H15N HN Bn Bn 47 (cis+trans) 48 (cis+trans) 91 : 9 Scheme 35. Exchange study using an internal amine trap performed by the group of Casey.

The predominant formation of complex 46-RuN,15N with the amine originating from the imine would seem to favor the inner-sphere mechanism. How- ever, an outer-sphere mechanism could explain the results by a strong cage effect (X, Scheme 30) or by strong hydrogen bonding of the newly generated amine (Y, Scheme 30). The explanation for the difference between this experiment and the previous would be a difference in the competing rates for amine coordination and dissociation, either from the solvent cage or from the hydrogen bonded intermediate diamine (cf. Scheme 30). If the explanation is strong hydrogen bonding, the rate difference would depend on the nucleo- philicity of the coordinating nitrogens and the hydrogen bond strength. The hydrogen bond of the amine formed by reduction of imine 45 would be the weakest, and thereby allowing a faster dissociation. In addition, the amine part of imine 45 would be more nucleophilic, and thereby give more internal amine trapping. This would result in the formation of 9% of complex 47- Ru15N,N, as compared to imine 38, which gave no detectable internal amine trapping (Scheme 35 and Scheme 33, respectively). The very existence of complex 47-Ru15N,N would seem to favor an outer-sphere mechanism but the small amount of complex 47-Ru15N,N (ca. 10%) could have other explanations. For example amine exchange via intramolecular amine attack at the η2-coordinated unsaturated species H (cf. Scheme 22). Another possible explanation for the formation of complex 47-Ru15N,N is that it is an artifact of the measurements formed by rearrangement of 46-RuN,15N. This possibility was given serious consideration in the study, but was excluded due slow rearrangement at 0 °C.98

3.2.4.3 Solvent cage effect The reduction of 4-aminocyclohexanone 41 by ruthenium hydride 28 was studied in order to investigate if there is a solvent cage effect in this reaction

57 (Section 3.2.3). The ketone substrate was chosen since alcohols do not coor- dinate strongly to ruthenium.90 Failure to observe trapping products would therefore be caused by a cage effect only (and not because of coordination to ruthenium). Unreacted ketone was acting as an external amine trap and the expected ratio between the different amine complexes (i.e. if there was no cage effect) was almost the same as the actual ratio (Figure 16). According to the cage hypothesis, hydrogen bonded intermediate I would initially be formed (Scheme 36). No stable intermediate with the ruthenium bound alcohol is formed90 and the newly produced amine and unsaturated species F must dissociate from I to form solvent cage J in order for the trapping products to be formed. A preference for formation of 43 could therefore only be explained by a solvent cage effect.

Bn HN

Ph Ph Ph Ph Ph Ph O H O OH O O OH Ph Ph Ph Ph 41 Ph Ph Ru Ru Ru OC H CD Cl OC OC OC 2 2 CO CO 42 28 HN F HN Bn Bn cage I J 28 41 (i) (ii)(iii)

Ph Ph O O Ph Ph Ph H Ph O O Ph Ph Ph Ph Ph Ph Ph H Ph H Ru H Ru Ru Ru Ph Ph NBn OC NBn OC CO OC CO OC CO CO

7 43 44 cis + trans OH O Scheme 36. Considered solvent cage effect in the hydrogenation of aminoketone 41 by 28.

An alternative interpretation of these results is that the hydrogen bond to the alcohol in I is so strong that cage complex J is never formed (Scheme 36). Instead, trapping of F by hydride 28, aminoketone 41 and free aminoalcohol 42 would occur while the newly generated alcohol is still hydrogen bonded. If this is the case, a strong cage effect for amines can not be ruled out. How- ever, due to steric hindrance this alternative seems less likely.

58 To support an outer-sphere mechanism, a strong cage effect has been in- ferred as one explanation to the absence or very low (ca. 10%) amount of trapping products from external amines in imine reduction (Section 3.1.2.3 65,91,98 and 3.2.1). In the reduction of 4-NH2-C6H4N=CHPh (31) by 28′ in the presence of external amines a solvent cage effect is the only possible explanation as to why no external amine trapping was observed (Section 3.1.2.4). It has been shown that there is no cage effect in the hydrogenation of aminoketone 41 by hydride 28. If the mechanism is outer-sphere, an explanation as to why there is a strong solvent cage in the case of imines, when there is no cage effect in the case of ketones, is needed.

3.3 Conclusions Kinetic isotope experiments show that the mechanism for transfer hydrogenation of imines is different from that of alcohols (aldehydes). However the isotope effects can be explained by both an outer- and inner-sphere mechanism, which must therefore be differentiated by other experiments. Reduction of N-methyl(1-phenylethylidiene)amine (34) by hydride 28 in the presence of N-methyl-1-(4-methoxyphenyl)ethylamine (35) as a potential external amine trap gave 90–100% of the complex from the newly generated amine. If the mechanism is outer-sphere, these results could be explained if the newly generated amine is strongly hydrogen bonded to, or trapped within a solvent cage together with unsaturated ruthenium species F, thereby pre- venting coordination of the external amine. However, the ratio between the amine complexes is then expected to depend on the concentration of the added amine, which is not the case. If the mechanism is inner-sphere, the 10% formation of the complex from the external amine must have an explanation. This could be generation of unsaturated species F in the reaction, which is supported by the fact that only complex 36 was formed when the reaction was run under H2. Amine exchange from the unsaturated η2-coordinated 16-electron intermediate (H) might be another explanation as to why exchange product 37 is formed. In this case the ratio is also expected to depend on the concentration of added amine and therefore this seems less likely. A third possibility is experimental artifacts, such as the presence of ketone in the reaction mixture. 4-Bn-NH-C6H9=NPh (38), an imine with an internal potential amine trap within the same molecule, has been reduced by hydride 28 at low tempera- 4 ture. The only product observed was [2,3,4,5-Ph4(η - C4CO)](CO)2RuNH(Ph)(C6H10-4-NHBn) (39), where the ruthenium binds to the amine originating from the imine. This is consistent with a stepwise, inner-sphere mechanism, in which the substrate is coordinated to ruthenium 15 prior to hydrogen transfer. On the other hand, 4-(Bn NH)-C6H9=NBn (45)

59 has been reduced by hydride 28′ in CD2Cl2, and 9% of the ruthenium complex (47-Ru15N,N) with the amine that do not originate from the imine was detected. The latter experiment was performed by the group of Casey and supports an outer-sphere mechanism unless there is another explanation for the 9% formation of 47-Ru15N,N. However, the outer-sphere mechanism is dependent on a strong cage effect to explain the lack of intermolecular trapping when 4-NH2- C6H4N=CHPh (31) was reduced by 28′ in the presence of an external amine, an experiment that was also performed by the group of Casey. In order to find out if there is a cage effect the reduction of 4-aminocyclohexanone 41 by 28 was studied. In this reaction the product alcohol, having an internal amine moiety, would be trapped within a solvent cage. No preference for complexation of the internal amine was observed, and since alcohols do not form stable coordination complexes with ruthenium,90 failure of trapping product therefore must be caused only because of lack of a cage effect. If the mechanism is outer-sphere, an explanation as to why there is a strong solvent cage in the case of imines and not in the case of ketones is needed. In addition, an explanation as to why the hydrogen bond to the amino group that is formed when imines 38 and 45 are reduced is so much stronger than the hydrogen bond to the amine that is formed by reduction of imine 31 is needed.

60 Noyori’s TsDPEN catalyst (Paper VI)

4.1 Introduction There was a breakthrough in asymmetric catalysis when Noyori presented his chiral Ru(II)-TsDPEN complex 3 (Figure 17).

Figure 17. Noyori’s Ru-TsDPEN catalyst precursor 3, active catalyst 5 and reactive intermediate 4.

Catalyst 3 is one of the first efficient catalysts for transfer hydrogenation of ketones giving high yields and high enantioselectivities with low catalyst loading, and works with a range of aromatic carbonyl compounds.107,108 Catalyst 3 was also employed when Noyori et al. reported the first asymmetric transfer hydrogenation of imines, giving high yields with enantioselectivities of up to 97%.109 It has been observed that ketones and imines differ in their compatibility with hydrogen donors. In transfer hydrogenation of ketones catalyzed by 3, 2-propanol was originally used as hydrogen donor.107 The use of a mixture of formic acid and triethylamine improved both yields and enantioselectivities.108 For imines, 2-propanol does not work at all with Noyori’s catalyst 3, and the reaction is carried out with formic acid and triethyl amine (Scheme 37).109

OH (S,S)-3 (cat.), KOH (cat.) O (S,S)-3 (cat.) OH H H Ar R 2-propanol Ar R HCOOH/NEt3 Ar R

(S,S)-3 (cat.), KOH (cat.) R3 (S,S)-3 (cat.) R3 N HN H 2-propanol 1 2 HCOOH/NEt R R 3 R1 R2 Scheme 37. Transfer hydrogenation from 2-propanol to imines by 3 is not possible.

61 The mechanism for transfer hydrogenation of ketones has been studied by several groups, and is proposed to be concerted via a cyclic six-membered transition state A (Scheme 7, Figure 18).

R R2 1 H Ru X H N N Ts H Ph Ph X=O,NR A Figure 18. Proposed cyclic transition state A in transfer hydrogenation by 3.

Noyori and co-workers have isolated 4 and 5 and proven that they are the active species in catalysis involving ketones or aldehydes and alcohols,24 and the concerted process is supported by kinetic isotope measurements by Casey et al.110 When isopropyl alcohol was dehydrogenated by unsaturated ruthenium species 4 isotope effects of 1.79 for transfer of OH to N and 2.86 for transfer from CH to Ru were found. The isotope effect for doubly labeled material (kCHOH/kCDOD = 4.88) was within experimental error the product of the two individual isotope effects, and this is consistent with a concurrent transfer of hydrogen from oxygen to nitrogen and from carbon to ruthenium.111 The concerted pathway is also supported by calculations performed by the groups of Andersson,112 Noyori,113 and van Leeuwen.114 More recent calculations by the group of Meijer suggests that the proton source for the ketone in the reverse reaction could also be the solvent.115 While mechanistic studies have been extensive for ketones/alcohols, the studies on the corresponding reaction involving imines/amines are limited. It has merely been assumed that both classes of compounds follow the same mechanistic pathway.23 The calculations done by Noyori include a comparison of activation energies between ketones and imines for an oxygen derivative of 3. It was found that the activation barrier for imines was 4.9–10.9 kcal/mol higher than that of ketones (Figure 19).113

Figure 19. Comparison of calculated activation energies for transition states similar to A based upon an oxygen derivative of Noyori’s catalyst 3.

62 An interesting question is why 2-propanol is an appropriate hydrogen source for ketones but not for imines in the 3-catalyzed transfer hydrogenation. One reason could be product inhibition, where the amine produced could coordi- nate to ruthenium intermediate 4 to give a stable, unreactive Ru-amine complex. Another possibility is that the activation barrier is too high for imines and that formic acid is needed because it has a higher redox potential than 2-propanol. It might also be because the mechanism for imines simply is different from that of ketones.

4.2 Results and discussion When N-benzyl-1-phenylethylamine (48a) was mixed with unsaturated ruthenium species 4 no Ru-amine complex was detected (Scheme 38). This observation clearly shows that product inhibition is unlikely to explain why 2-propanol cannot be used as hydrogen donor in transfer hydrogenation of imines by 3.

Scheme 38. There is no reaction between amine 48a and ruthenium species 4 (2 equiv.)

No imine was detected on mixing amine 48a with 4, even with a large excess of amine (10 equiv.), which rules out an unfavorable equilibrium as the reason for lack of dehydrogenation products. When stoichiometric amounts of 5 were mixed with N-benzyl-(1- phenylethylidene)amine (49a) or N-methyl-1-(p-methoxyphenyl- ethylidene)amine (49b), no amine product was observed (Scheme 39). Even with 10 equiv. of imine no reaction between hydride 5 and the imine occurred after 12 hours.

63 R3 N H Ru + no reaction H2N N Ts R2 CD Cl RT Ph Ph R1 2 2, (S,S)-5 1-10 equiv. 49a:R1 =H,R2 =Me,R3 =Bn 49b:R1 =OMe,R2 =Me,R3 =Me Scheme 39. A mixture of 5 and imines 49 does not result in reaction or complex formation.

This reactivity is in sharp contrast to the reaction of ketones. Thus, treatment of 5 with a tenfold excess of acetone (Scheme 40) instantaneously gave the 16-electron species 4 and 2-propanol.24

Scheme 40. Ruthenium hydride 5 reacts rapidly with a ketone.

No Ru-amine complex could be detected in the reaction of 49a or 49b with 5 (Scheme 39). These results show that the reason for transfer hydrogenation of imines by 3 not working catalytically with 2-propanol cannot be only the different redox potentials of 2-propanol and formic acid as hydrogen donors. If this were the case the reaction between stoichiometric amounts of the active species 5 and an imine should proceed smoothly. Since formic acid/triethyl amine is used the in transfer hydrogenation of imines we argued that acidic activation is needed to overcome the activation barrier. This was supported by the observation that the addition of one equivalent of formic acid to a mixture of 5 and imine 49a or 49b immediately afforded the corresponding amine. To confirm that the formic acid was not working as hydrogen donor we performed the hydrogenation of imine 49a by 5 with different Brønstedt acids and one Lewis acid (Table 3).

64 Table 3.a

b c d Entry Acid pKa time/h Conversion (%) ee (%) 1 HCOOH 3.75 1 96 91 2 HCOOH 3.75 23 96 85 3 HBF4 0.5 1 >99 78 4 HBF4 0.5 23 >99 75 5 CF3COOH 0.52 1 39 69

6 CF3COOH 0.52 23 73 84 7 CH3COOH 4.76 1 39 94 8 CH3COOH 4.76 23 78 89 9 Sc(OTf)3 1 >99 80 10 Sc(OTf)3 23 >99 78 11 PhCOOH 4.2 23 26 82 12 Me3COOH 5.03 23 29 80 13 – – 31e – – a Unless otherwise noted, 35 µmol of imine 49a was mixed with acid (1 equiv.) followed by (S,S)-5 (2 equiv.) in CH2Cl2 under argon. The reaction mixture was worked up with 2 M NaOH, extracted with CH2Cl2, dried (Na2SO4) and purified by b bulb-to-bulb distillation. pKa values apply to dilute aqueous solutions and are taken from ref116 c Conversion determined by 1H NMR. d (S)-configuration, ee determined by GC analysis. e 4 equiv. of (S,S)-5 was used.

We found that most of these acids worked well to promote the hydrogenation. All yields and enantioselectivities are comparable to those of Noyori’s catalytic system (cf. 72% yield and 77% ee after 36 h).109 Formic acid as well as tetrafluoroboric acid gave excellent yields and high enantioselectivities in short reaction times (Table 3, entries 1 and 3). Acetic and trifluoroace- tic acid needed longer reaction times but still gave high yields and enantioselectivities (Table 3, entries 6 and 8). Scandium triflate, which has been used for activation of imines in reductive amination,117 also gave an excellent yield and high ee in the reaction between imine 49a and catalyst 5 (Table 3, entry 9). This further supports that the role of the acid is to activate the imine. Only the sterically hindered benzoic and pivalic acid gave moderate yields (Table 3, entries 11 and 12). The striking difference when no acid is used is still evident, in which case no amine was formed even after 31 hours (Table 3, entry 13). Encouraged by the finding that the reaction works with acidic activation we attempted to make catalytic transfer hydrogenation from 2-propanol to imine 49a catalyzed by 5 formed in situ (Table 4).

65 Table 4.a

Entry Acid Conversionb (%) 1 HCOOH 70 2 HBF4 – 3 CH3COOH – a [RuCl2(p-cymene)]2 (2 mg, 3 µmol) and (S,S)-TsDPEN (2 mg, 5 µmol) were stirred in 1 mL 2-propanol under argon for 15 min, followed by addition of 100 µl of a 0.1 M solution of KOH (10 µmol, in 2-propanol). After an additional 15 minutes 1mL of a 0.2 M solution of imine 49a (0.2 mmol, in 2-PrOH) premixed with acid (1 equiv.) was added. The reaction mixture was worked up after 2 days with 2 M b 1 NaOH, extracted with CH2Cl2 and dried (Na2SO4). Conversion determined by H NMR.

The reaction worked only with formic acid (Table 4, entry 1). Neither tetrafluoroboric acid nor acetic acid gave conversion to the amine (Table 4, entries 2 and 3). In the case of formic acid it is not clear if the hydride comes from 2-propanol or the formate. The reason for the failure of transfer hydrogenation in the case of tetrafluoroboric and acetic acid (Table 4, entries 1 and 2), could be that alkaline conditions are needed for the dehydrogenation of 2-propanol. Further investigations are needed to make certain conclusions.

4.2.1 Mechanistic considerations Based upon our results we propose an ionic mechanism, wherein the imine is protonated prior to hydrogen transfer (Scheme 41).

3 3 R N H+ R NH+ + H Ru 1 2 -H 1 2 R R R R H2N N Ts 49 R 5 R fast

3 + R3 + R NH H NH2 H H Ru + 1 2 R1 R2 -H R R HN N Ts 48 R R 4 Scheme 41. New proposed mechanism for transfer hydrogenation of imines by 5.

66 Another possibility is that ruthenium hydride 5 is protonated on the nitrogen attached to the tosyl group. However, this would lead to low enantioselectiv- ity, since a protonated nitrogen would not be expected to stay coordinated to ruthenium.118 Since transfer hydrogenations of imines by 3 gave high ee’s109 this mechanism seems unlikely. After paper VI was published, a study on transfer hydrogenation of an imine, catalyzed by a related Rh(III)-TsDPEN catalyst, was performed by the group of Blackmond.119 It was suggested that the true substrate in this case is the unprotonated imine. Recently, the transfer hydrogenation of imines by N-alkylated Ru-TsDPEN catalysts was studied by the group of Wills.120 An ionic mechanism, where the imine is initially protonated was suggested as a viable model for further investigations. More investigations are needed in order to elucidate the mechanism of imine reduction by Noyori’s catalyst 3. It would also be interesting to deter- mine how the actual proton and hydride transfer occur, e.g. if there is a simultaneous transfer of both hydrogens and if the imine is coordinated or not.

4.3 Conclusions We have shown that the previously reported concerted pathway23,113 (Scheme 7, Figure 18) does not operate for imines. The observation that transfer hydrogenation of imines catalyzed by 3 does not work with 2-propanol cannot be explained by product inhibition, since the unsaturated ruthenium species 4 does not react with N-benzyl-1-phenylethylamine (48a). Ruthenium hydride 5, which reacts quickly with ketones,24 does not react with stoichiometric amounts of N-benzyl-(1-phenylethylidene)amine (49a) or N-methyl-1-(p-methoxyphenyl-ethylidene)amine (49b). Acidic activation of the imine is required and therefore we propose an ionic mechanism, whereby the imine is activated by an acid prior to hydrogenation.

67 Concluding remarks

There is an increasing demand for more efficient and less expensive processes in chemistry. In addition, an environmental awareness is emerging. This creates a demand to develop better and greener methods for chemical transformations. Mechanistic studies of catalytic processes constitute a powerful approach for optimizing and developing new reactions. The goal of this work has been to create a better mechanistic understanding of how three different ruthenium based catalysts operate. The catalysts are useful for the enantioselective preparation of chiral alcohols, amines or derivatives of them via transfer hydrogenation, dynamic kinetic resolution or both. The mechanism for racemization of secondary alcohols by Bäckvall’s 5 catalyst, η -(Ph5C5)Ru(CO)2Cl, has been studied in detail. Support for a mechanism involving alkoxide intermediates and ligand exchange via CO assistance, is provided. In addition, it has been shown that the substrate is coordinated throughout the catalytic cycle. We have also provided support for a stepwise inner-sphere mechanism in the transfer hydrogenation of imines by Shvo’s catalyst, [Ru2(CO)4(μ-H)(C4Ph4COHOCC4Ph4)]. However, an outer-sphere mechanism can not be excluded in this case. Finally, we have shown that the mechanism of transfer hydrogenation of imines is different from that of ketones for Noyori’s well-known catalyst, [p-(Me2CH)C6H4Me]RuH(NH2CHPhCHPhNSO2C6H4-p-CH3). The fact that the mechanism had been assumed to be the same as for ketones further em- phasizes the need for mechanistic studies.

68 Acknowledgements

I am grateful for many people’s contributions, direct or indirect, which made this thesis possible. I especially wish to express my sincere gratitude to:

Prof. Jan-E. Bäckvall, for accepting me as a graduate student, sharing your broad knowledge about chemistry and for your guidance throughout this work.

My excellent co-workers, Assis. Prof. Belén Martín-Matute, Assis. Prof. Joseph S. M. Samec, Dr Michaela Vallin (née Edin), Ms. Madeleine C. War- ner, Mr Jonas Nyhlén, Assis. Prof. Timofei Privalov, Dr Alida H. Éll, and Dr. Ibrar Hussein. Thanks for great collaboration!

The students with whom I’ve had the pleasure to work, Ms Nazli Jalalian, Mr Juri Möbus and Ms Paola Lavalle.

All of the past and present members in the JEB group, for creating a great atmosphere and for keeping the ship floating.

All other co-workers at the department.

All of the technical staff at the department and in the chemical library, especially Ms. Kristina Romare for help with NMR measurements.

Assoc. Prof. Ulrika Nilsson for help with GC-MS measurements, Asiss. Prof. Ralf Torgrip for valuable discussions on statistics and Assis. Prof. Lars Eriksson for crystallography determination.

Assoc. Prof. Per I. Arvidsson, for accepting me as an undergraduate student during a 5 weeks research project, and for your much appreciated guidance at that time and later on, too.

The teachers in Uppsala, especially Curt-Eric Hagberg for always taking your time and answering my “small” questions, and Assoc. Prof. Adolf Gogoll, for introducing me to the fascinating world of nuclear magnetic resonance.

69 Prof. Carsten Bolm, for accepting me as a undergraduate student during my Diploma Work and for your guidance during that time.

All teachers that I’ve had, for giving me the joy of learning and sharing your knowledge. To mention a few: Kerstin Edmarker Andersson, Maja- Greta Vikar, Sverker Nyman, Håkan Gustavsson, Kjell Robertsson, Folke Johnson and Astrid Alnås-Widén.

Svenska Kemistsamfundet, C. F. Liljevalch J:ors stipendiefond, Berzelius Center EXSELENT and AstraZeneca’s resestipendium till minne av Nils Löfgren, for financial support in connection with conferences. L & E Ki- nanders stiftelse, for financial contribution to a computer.

Assis. Prof. Belén Martín-Matute, Asiss. Prof. Joseph S. M. Samec, and Dr Michaela Vallin (née Edin), for proofreading this thesis and Dr Eszter Borbas for linguistic improvements.

Tjejerna i Uppsala (Ida, Ann-Malin och Gerd), för allt roligt vi haft till- sammans; jag är jätteglad att ni är mina vänner!

Min underbara familj (mamma och pappa, Henrik, Linda, Noah och Sam), för stöd och uppmuntran genom åren.

Min kära syster Hanna, som delat allt så länge och som alltid kommer att ha en speciell plats i mitt liv.

Min älskade Tomas, för ovärderligt stöd och enormt tålamod, för att du är min bäste coach och för att du finns i mitt liv ♥

Gud, för Livet

70 References

(1) Org. Lett. 2009, 11, 24A–25A. (2) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic Chemistry; University Science Books: Sausalito, California, 2006; p 538. (3) Carroll, F. A. Perspectives on Structure and Mechanism in Organic Chem- istry; Brooks/Cole Publishing Company: Pacific Grove, California, 1998; p 316. (4) Atkins, P.; de Paula, J. Atkins' Physical Chemistry, 7th ed.; Oxford Univer- sity Press: New York, 2002; p 908. (5) Anastas, P. T.; Warner, J. C., In Green Chemistry: Theory & Practice, Oxford University Press: New York, 1998. (6) Ritter, S. K. Chem. Eng. News 2008, 86, 59–68. (7) Cabri, W. Catal. Today 2009, 140, 2–10. (8) a) Thayer, A. Chem. Eng. News 2007, 85, 11–19. b) Blaser, H.-U. Chem. Commun. 2003, 293–296. (9) Thayer, A. Chem. Eng. News 2005, 83, 40–48. (10) Rouhi, A. M. Chem. Eng. News 2004, 82, 47–62. (11) Zassinovich, G.; Mestroni, G.; Gladiali, S. Chem. Rev. 1992, 92, 1051– 1069. (12) a) Meerwein, H.; Schmidt, R. Liebigs. Ann. Chem. 1925, 444, 221–238. b) Ponndorf, W. Angew. Chem. 1926, 39, 138–143. c) Verley, A. Bull. Soc. Chim. 1925, 37, 537–542. (13) Oppenauer, R. V. Recl. Trav. Chim. Pays-Bas Belg. 1937, 56, 137–144. (14) a) Wills, M. In Modern Reduction Methods; Andersson, P. G.; Munslow, I. J. Eds.; Wiley-VCH: Weinheim, 2008; pp 271–296. b) Gladiali, S.; Taras, R. In Modern Reduction Methods; Andersson, P. G.; Munslow, I. J. Eds.; Wiley-VCH: Weinheim, 2008; pp 135–157. c) Baratta, W.; Rigo, P. Eur. J. Inorg. Chem. 2008, 2008, 4041–4053. d) Wu, X.; Mo, J.; Li, X.; Hyder, Z.; Xiao, J. Prog. Nat. Science 2008, 18, 639–652. e) Ikariya, T.; Blacker, A. J. Acc. Chem. Res. 2007, 40, 1300–1308. f) Morris, D. J.; Wills, M. Chim. Oggi Chem. Today 2007, 25, 11. g) Gladiali, S.; Alberico, E. Chem. Soc. Rev. 2006, 35, 226–236. h) Clapham, S. E.; Hadzovic, A.; Morris, R. H. Coord. Chem. Rev. 2004, 248, 2201–2237. i) Blaser, H.-U.; Malan, C.; Pu- gin, B.; Spindler, F.; Steiner, H.; Studer, M. Adv. Synth. Catal. 2003, 345, 103–151. (15) Palmer, M. J.; Wills, M. Tetrahedron: Asymmetry 1999, 10, 2045–2061. (16) a) Samec, J. S. M.; Bäckvall, J.-E.; Andersson, P. G.; Brandt, P. Chem. Soc. Rev. 2006, 35, 237–248. b) Bäckvall, J.-E. J. Organomet. Chem. 2002, 652, 105–111. c) Noyori, R.; Yamakawa, M.; Hashiguchi, S. J. Org. Chem. 2001, 66, 7931–7944.

71 (17) de Graauw, C. F.; Peters, J. A.; van Bekkum, H.; Huskens, J. Synthesis 1994, 1007–1017. (18) Sasson, Y.; Blum, J. J. Org. Chem. 1975, 40, 1887–1896. (19) Laxmi, Y. R. S.; Bäckvall, J.-E. Chem. Commun. 2000, 611–612. (20) Pamies, O.; Bäckvall, J.-E. Chem. Eur. J. 2001, 7, 5052–5058. (21) Gladiali, S.; Alberico, E. In Transition Metals for Organic Synthesis, 2nd ed.; Beller, M.; Bolm, C. Eds.; Wiley-VCH: Weinheim, 2004, pp 145–166. (22) a) Yi, C. S.; He, Z.; Guzei, I. A. Organometallics 2001, 20, 3641–3643. b) Ritter, J. C. M.; Bergman, R. G. J. Am. Chem. Soc. 1998, 120, 6826– 6827. c) Blum, O.; Milstein, D. J. Am. Chem. Soc. 1995, 117, 4582–4594. d) Bryndza, H. E.; Calabrese, J. C.; Marsi, M.; Roe, D. C.; Tam, W.; Ber- caw, J. E. J. Am. Chem. Soc. 1986, 108, 4805–4813. (23) Noyori, R.; Hashiguchi, S. Acc. Chem. Res. 1997, 30, 97–102. (24) Haack, K.-J.; Hashiguchi, S.; Fujii, A.; Ikariya, T.; Noyori, R. Angew. Chem. Int. Ed. Engl. 1997, 36, 285–288. (25) a) Guan, H.; Iimura, M.; Magee, M. P.; Norton, J. R.; Zhu, G. J. Am. Chem. Soc. 2005, 127, 7805–7814. b) Bullock, R. M. Chem. Eur. J. 2004, 10, 2366–2374. c) Magee, M. P.; Norton, J. R. J. Am. Chem. Soc. 2001, 123, 1778–1779. d) Schlaf, M.; Ghosh, P.; Fagan, P. J.; Hauptman, E.; Bullock, R. M. Angew. Chem. Int. Ed. 2001, 40, 3887–3890. (26) Atkins, P.; de Paula, J. Atkins' Physical Chemistry, 7th ed. Oxford Univer- sity Press: New York, 2002; p 462. (27) Cope, M. J. Anal. Proceed. 1993, 30, 498–500. (28) Thayer, A. Chem. Eng. News 2008, 86, 12–20. (29) Thayer, A. Chem. Eng. News 2005, 83, 49–53. (30) Hanessian, S. Pure Appl. Chem. 1993, 65, 1189–1204. (31) a) Fogassy, E.; Nogradi, M.; Kozma, D.; Egri, G.; Palovics, E.; Kiss, V. Org. Biomol. Chem. 2006, 4, 3011–3030. b) Faber, K. Chem. Eur. J. 2001, 7, 5004–5010. c) Kagan, H. B. Tetrahedron 2001, 57, 2449–2468. (32) Beck, G. Synlett 2002, 837–850. (33) a) Martín-Matute, B.; Bäckvall, J.-E. In Asymmetric Org. Synth. Enzymes; Gotor, V.; Alfonso, I.; Garcia-Urdiales, E. Eds.; Wiley-VCH: Weinheim, 2008, pp 89–113. b) Martín-Matute, B.; Bäckvall, J.-E. In Organic Synthe- sis with Enzymes in Non-Aqueous Media; Carrea, G.; Riva, S. Eds.; Wiley- VCH: Weinheim, 2008, pp 113–144. c) Pellissier, H. Tetrahedron 2008, 64, 1563–1601. d) Bäckvall, J.-E. In Asymmetric Synthesis, 2nd ed.; Christ- mann, M.; Bräse, S.; Seebach, D. Eds. Wiley-VCH: Weinheim, 2007, pp 171–175. e) Martín-Matute, B.; Bäckvall, J.-E. Curr. Opin. Chem. Biol. 2007, 11, 226–232. f) Ward, R. S. Tetrahedron: Asymmetry 1995, 6, 1475– 1490. (34) In paper IV I have chosen to mostly focus the discussion on those experiments where I have been involved. (35) Huerta, F. F.; Minidis, A. B. E.; Bäckvall, J.-E. Chem. Soc. Rev. 2001, 30, 321–331. (36) Ito, M.; Osaku, A.; Kitahara, S.; Hirakawa, M.; Ikariya, T. Tetrahedron Lett. 2003, 44, 7521–7523.

72 (37) a) Ko, S. B.; Baburaj, B.; Kim, M. J.; Park, J. J. Org. Chem. 2007, 72, 6860–6864. b) van As, B. A. C.; van Buijtenen, J.; Heise, A.; Broxterman, Q. B.; Verzijl, G. K. M.; Palmans, A. R. A.; Meijer, E. W. J. Am. Chem. Soc. 2005, 127, 9964–9965. c) Monsees, A.; Hoff, M.; Trauthwein, H. Tet- rahedron Lett. 2005, 46, 3403–3406. d) Verzijl, G. K. M.; de Vries, J. G.; Broxterman, Q. B. Tetrahedron: Asymmetry 2005, 16, 1603–1610. e) Choi, J. H.; Choi, Y. K.; Kim, Y. H.; Park, E. S.; Kim, E. J.; Kim, M.-J.; Park, J. J. Org. Chem. 2004, 69, 1972–1977. f) Pamies, O.; Bäckvall, J.-E. Trends Biotechnol. 2004, 22, 130–135. g) Pamies, O.; Bäckvall, J.-E. Curr. Opin. Biotechnol. 2003, 14, 407–413. h) Riermeier, T. H.; Gross, P.; Kim, M.-J.; Ahn, Y.; Park, J. Curr. Opin. Biotechnol. 2003, 14, 131. i) Choi, J. H.; Kim, Y. H.; Nam, S. H.; Shin, S. T.; Kim, M.-J.; Park, J. Angew. Chem. Int. Ed. 2002, 41, 2373–2376. j) Kim, M.-J.; Ahn, Y.; Park, J. Curr. Opin. Bio- technol. 2002, 13, 578–587. (38) Pamies, O.; Bäckvall, J.-E. Chem. Rev. 2003, 103, 3247–3261. (39) Martín-Matute, B.; Edin, M.; Bogár, K.; Bäckvall, J.-E. Angew. Chem. Int. Ed. 2004, 43, 6535–6539. (40) Martín-Matute, B.; Edin, M.; Bogár, K.; Kaynak, F. B.; Bäckvall, J.-E. J. Am. Chem. Soc. 2005, 127, 8817–8825. (41) a) Träff, A.; Bogár, K.; Warner, M.; Bäckvall, J.-E. Org. Lett. 2008, 10, 4807–4810. b) Bogár, K.; Bäckvall, J.-E. Tetrahedron Lett. 2007, 48, 5471–5474. c) Bogár, K.; Vidal, P. H.; Alcantara Leon, A. R.; Bäckvall, J.- E. Org. Lett. 2007, 9, 3401–3404. d) Bogár, K.; Martín-Matute, B.; Bäck- vall, J.-E. Beilstein J. Org. Chem. 2007, 3, 50. (42) Edin, M.; Martín-Matute, B.; Bäckvall, J.-E. Tetrahedron: Asymmetry 2006, 17, 708–715. (43) a) Martín-Matute, B.; Edin, M.; Bäckvall, J.-E. Chem. Eur. J. 2006, 12, 6053–6061. b) Olofsson, B.; Bogár, K.; Fransson, A.-B. L.; Bäckvall, J.-E. J. Org. Chem. 2006, 71, 8256–8260. c) Fransson, A.-B. L.; Xu, Y.; Leijon- dahl, K.; Bäckvall, J.-E. J. Org. Chem. 2006, 71, 6309–6316. (44) Borén, L.; Leijondahl, K.; Bäckvall, J.-E. Tetrahedron Lett., in press (Cor- rected Proof) DOI: 10.1016/j.tetlet.2009.1002.1079. (45) a) Leijondahl, K.; Borén, L.; Braun, R.; Bäckvall, J.-E. J. Org. Chem. 2009, 74, 1988–1993. b) Leijondahl, K.; Borén, L.; Braun, R.; Bäckvall, J.-E. Org. Lett. 2008, 10, 2027–2030. (46) Acetophenone is needed as a hydride acceptor, η5-hydride 15 cannot abstract the α-hydrogen from (S)-11, and thereby racemize it, without first donating its own hydride. (47) Nyhlén, J.; Privalov, T.; Bäckvall, J.-E. Chem. Eur. J., in press (Early View) DOI: 10.1002/chem.200900291. (48) Martín-Matute, B.; Åberg, J. B.; Edin, M.; Bäckvall, J.-E. Chem. Eur. J. 2007, 13, 6063–6072. (49) The reaction mixture containing intermediate B also showed characteristic peaks in the area of C-O stretching vibrations (Ot-Bu). However, the mixture possibly contains t-BuOH (and t-BuOK) and the peaks in this area were not easily assigned. (50) Gross, D. C.; Ford, P. C. J. Am. Chem. Soc. 1985, 107, 585–593.

73 (51) a) Dutta, B.; Scopelliti, R.; Severin, K. Organometallics 2008, 27, 423– 429. b) Suzuki, H.; Omori, H.; Mor-oka, Y. J. Organomet. Chem. 1987, 327, C47–C50. (52) a) Moreno, M. A.; Haukka, M.; Kallinen, M.; Pakkanen, T. A. Appl. Or- ganomet. Chem. 2006, 20, 51–69. b) Trautman, R. J.; Gross, D. C.; Ford, P. C. J. Am. Chem. Soc. 1985, 107, 2355–2362. (53) Taube, D. J.; Rokicki, A.; Anstock, M.; Ford, P. C. Inorg. Chem. 1987, 26, 526–530. -1 (54) H2O vapor in the system absorbs in the region of 1600 cm and a back- ground spectrum is automatically subtracted from the collected data. If the H2O vapors are not constant (which they seldom are) a peak in this region might not visible. Indeed, in our first experiments the system contained so much humidity that a lot of noise in this region made it impossible to detect the acyl peak. (55) [PPN]+ = [Bis(triphenylphosphoranylidene)ammonium]+ (56) The estimated 13C NMR shifts of intermediate B were δ 205.3 (acyl) and 211.4 (CO) ppm. For experimental details, see supporting information of paper II. (57) a) Baratta, W.; Bosco, M.; Chelucci, G.; Del Zotto, A.; Siega, K.; Toniutti, M.; Zangrando, E.; Rigo, P. Organometallics 2006, 25, 4611–4620. b) Osa- kada, K.; Ohshiro, K.; Yamamoto, A. Organometallics 1991, 10, 404–410. c) Kim, Y. J.; Osakada, K.; Takenaka, A.; Yamamoto, A. J. Am. Chem. Soc. 1990, 112, 1096–1104. d) Kegley, S. E.; Schaverien, C. J.; Freuden- berger, J. H.; Bergman, R. G.; Nolan, S. P.; Hoff, C. D. J. Am. Chem. Soc. 1987, 109, 6563–6565. (58) Crabtree, R. H. The Organometallic Chemistry of the Transition Metals, 4th ed. John Wiley & Sons: New York, 2005; p 546. (59) a) Fulton, J. R.; Holland, A. W.; Fox, D. J.; Bergman, R. G. Acc. Chem. Res. 2002, 35, 44–56. b) Bryndza, H. E.; Tam, W. Chem. Rev. 1988, 88, 1163–1188. (60) Casey, C. P.; Vosejpka, P. C.; Underiner, T. L.; Slough, G. A.; Gavney, J. A. J. Am. Chem. Soc. 1993, 115, 6680–6688. (61) Habib, A.; Tanke, R. S.; Holt, E. M.; Crabtree, R. H. Organometallics 1989, 8, 1225–1231. (62) Csjernyik, G.; Bogár, K.; Bäckvall, J.-E. Tetrahedron Lett. 2004, 45, 6799– 6802. (63) a) Bergman, R. G. Polyhedron 1995, 14, 3227–3237. b) Glueck, D. S.; Winslow, L. J. N.; Bergman, R. G. Organometallics 1991, 10, 1462–1479. c) Glueck, D. S.; Wu, J.; Hollander, F. J.; Bergman, R. G. J. Am. Chem. Soc. 1991, 113, 2041–2054. (64) Simanko, W.; Tesch, W.; Sapunov, V. N.; Mereiter, K.; Schmid, R.; Kirch- ner, K.; Coddington, J.; Wherland, S. Organometallics 1998, 17, 5674– 5688. (65) Casey, C. P.; Johnson, J. B. J. Am. Chem. Soc. 2005, 127, 1883–1894. (66) Two diastereoisomers are formed because there are two stereogenic cen- ters: the ruthenium and the α-carbon of the alcohol. (67) McWilliams, K. M.; Angelici, R. J. Organometallics 2007, 26, 5111–5118.

74 (68) An HMBC (Heteronuclear Multiple Bond Correlation) shows cross peaks for long range, through-bonds, couplings between protons and carbons. (69) Severe decomposition occurred after ca. 1 h and the reaction could not be followed until completion. (70) The easy formation of 7-memberd rings from 6-membered ring via 1,2- migration is known in a number of rearrangements, e.g. the Beckman rearrangement and the Baeyer-Villiger oxidation. (71) Blum, Y.; Czarkie, D.; Rahamim, Y.; Shvo, Y. Organometallics 1985, 4, 1459–1461. (72) Shvo, Y.; Czarkie, D.; Rahamim, Y.; Chodosh, D. F. J. Am. Chem. Soc. 1986, 108, 7400–7402. (73) Menashe, N.; Shvo, Y. Organometallics 1991, 10, 3885–3891. (74) a) Leijondahl, K.; Fransson, A.-B. L.; Bäckvall, J.-E. J. Org. Chem. 2006, 71, 8622–8625. b) Menashe, N.; Salant, E.; Shvo, Y. J. Organomet. Chem. 1996, 514, 97–102. (75) Almeida, M. L. S.; Beller, M.; Wang, G.-Z.; Bäckvall, J.-E. Chem. Eur. J. 1996, 2, 1533–1536. (76) Strübing, D.; Krumlinde, P.; Piera, J.; Bäckvall, J.-E. Adv. Synth. Catal. 2007, 349, 1577–1581. (77) a) Persson, B. A.; Larsson, A. L. E.; Le Ray, M.; Bäckvall, J.-E. J. Am. Chem. Soc. 1999, 121, 1645–1650. b) Larsson, A. L. E.; Persson, B. A.; Bäckvall, J.-E. Angew. Chem. Int. Ed. Engl. 1997, 36, 1211–1212. (78) a) Edin, M.; Steinreiber, J.; Bäckvall, J.-E. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 5761–5766. b) Martín-Matute, B.; Bäckvall, J.-E. J. Org. Chem. 2004, 69, 9191–9195. c) Kim, M.-J.; Choi, Y. K.; Choi, M. Y.; Kim, M. J.; Park, J. J. Org. Chem. 2001, 66, 4736–4738. d) Persson, B. A.; Huerta, F. F.; Bäckvall, J.-E. J. Org. Chem. 1999, 64, 5237–5240. (79) a) Samec, J. S. M.; Mony, L.; Bäckvall, J.-E. Can. J. Chem. 2005, 83, 909– 916. b) Samec, J. S. M.; Bäckvall, J.-E. Chem. Eur. J. 2002, 8, 2955–2961. (80) a) Samec, J. S. M.; Éll, A. H.; Bäckvall, J.-E. Chem. Eur. J. 2005, 11, 2327–2334. b) Ibrahem, I.; Samec, J. S. M.; Bäckvall, J.-E.; Córdova, A. Tetrahedron Lett. 2005, 46, 3965–3968. c) Éll, A. H.; Samec, J. S. M.; Brasse, C.; Bäckvall, J.-E. Chem. Commun. 2002, 1144–1145. (81) Pamies, O.; Éll, A. H.; Samec, J. S. M.; Hermanns, N.; Bäckvall, J.-E. Tetrahedron Lett. 2002, 43, 4699–4702. (82) Paetzold, J.; Bäckvall, J.-E. J. Am. Chem. Soc. 2005, 127, 17620–17621. (83) a) Thalén, Lisa K.; Zhao, D.; Sortais, J.-B.; Paetzold, J.; Hoben, C.; Bäck- vall, J.-E. Chem. Eur. J. 2009, 15, 3403–3410. b) Hoben, C. E.; Kanupp, L.; Bäckvall, J.-E. Tetrahedron Lett. 2008, 49, 977–979. (84) Samec, J. S. M.; Éll, A. H.; Åberg, J. B.; Privalov, T.; Eriksson, L.; Bäck- vall, J.-E. J. Am. Chem. Soc. 2006, 128, 14293–14305. (85) Casey, C. P.; Singer, S. W.; Powell, D. R.; Hayashi, R. K.; Kavana, M. J. Am. Chem. Soc. 2001, 123, 1090–1100. (86) Samec, J. S. M.; Éll, A. H.; Bäckvall, J.-E. Chem. Commun. 2004, 2748– 2749. (87) Éll, A. H.; Johnson, J. B.; Bäckvall, J.-E. Chem. Commun. 2003, 1652– 1653.

75 (88) Johnson, J. B.; Bäckvall, J.-E. J. Org. Chem. 2003, 68, 7681–7684. (89) Casey, C. P.; Johnson, J. B. Can. J. Chem. 2005, 83, 1339–1346. (90) Casey, C. P.; Bikzhanova, G. A.; Bäckvall, J.-E.; Johansson, L.; Park, J.; Kim, Y. H. Organometallics 2002, 21, 1955–1959. (91) Casey, C. P.; Bikzhanova, G. A.; Cui, Q.; Guzei, I. A. J. Am. Chem. Soc. 2005, 127, 14062–14071. (92) Casey, C. P.; Guan, H. J. Am. Chem. Soc. 2007, 129, 5816–5817. (93) Casey, C. P.; Johnson, J. B.; Singer, S. W.; Cui, Q. J. Am. Chem. Soc. 2005, 127, 3100–3109. (94) Privalov, T.; Samec, J. S. M.; Bäckvall, J.-E. Organometallics 2007, 26, 2840–2848. (95) a) Comas-Vives, A.; Ujaque, G.; Lledos, A. Organometallics 2008, 27, 4854–4863. b) Comas-Vives, A.; Ujaque, G.; Lledos, A. Organometallics 2007, 26, 4135–4144. (96) The strong hydrogen bonding in X sometimes has been referred to as a solvent cage effect. In order to minimize confusion the hydrogen bonding in X and the solvent cage Y will be referred to as two separate issues throughout this thesis. (97) DFT calculations by Privalov et al. have shown an energy minimum for a ruthenium complex with one phenyl of the tetraphenylcyclopentadienone coordinated to ruthenium. This suggests that ruthenium migration via the aromatic system might be possible. Ref 94. (98) Casey, C. P.; Clark, T. B.; Guzei, I. A. J. Am. Chem. Soc. 2007, 129, 11821–11827. (99) The initial amount of Ru dimer 7 was estimated to ca. 5% and all values have been corrected by subtraction of this amount. The figure shows the corrected values for [7]. (100) The expected ratio has been calculated from the ratio between aminoketone 41 and aminoalcohol 42 according to equation 1. This ratio changes during the reaction and if there is no cage effect, i.e. that they compete equally in coordination to Ru; the amount of each complex should be equal to the ac- cumulated concentration of the corresponding amine. [41] and [42] could not be obtained directly by integration in the NMR spectrum due to over- lapping peaks. Instead they were obtained indirectly by calculations.

t x 43  42 = dt (1) 44  41  t0  (101) A contol experiment, where cyclohexanone and aminoketone complex 44 in a 1:1 ratio was allowed to react with hydride 28, showed that they were reduced with approximately the same rate. (102) Since the rate of reduction of aminoketone 41 (d[41]/dt) and aminoketone complex 44 (d[44]/dt) is approximately the same, d[44]/dt was determined in each point, which affored d[41]/dt (d[41]/dt ≈ d[44]/dt). [44] was corrected by addition of this amount and [43] was corrected by subtraction of the difference between measured [44] and corrected [44] in each point. New actual and expected ratio 43:44 were then calculated.

76 (103) Casey and co-workers have recently studied how substrates such as alcohols, phosphines, and even water promote the H2 elimination from hydride 29. Ref 93. (104) An assocative mechanism would involve 16-electeron intermediates, whereas a dissociative mechanism would lead to a high energy 14-electron species. (105) Åberg, Jenny B.; Bäckvall, J.-E. unpublished results. 15 15 (106) When the reaction was run in toluene-d8 the ratio 46-RuN, N:47-Ru N,N was 83:17. (107) Hashiguchi, S.; Fujii, A.; Takehara, J.; Ikariya, T.; Noyori, R. J. Am. Chem. Soc. 1995, 117, 7562–7563. (108) Fujii, A.; Hashiguchi, S.; Uematsu, N.; Ikariya, T.; Noyori, R. J. Am. Chem. Soc. 1996, 118, 2521–2522. (109) Uematsu, N.; Fujii, A.; Hashiguchi, S.; Ikariya, T.; Noyori, R. J. Am. Chem. Soc. 1996, 118, 4916–4917. (110) Casey, C. P.; Johnson, J. B. J. Org. Chem. 2003, 68, 1998–2001. (111) If the product of the individual isotope effects is equal (or close) to the combined isotope effect a concerted process is expected. (112) Alonso, D. A.; Brandt, P.; Nordin, S. J. M.; Andersson, P. G. J. Am. Chem. Soc. 1999, 121, 9580–9588. (113) Yamakawa, M.; Ito, H.; Noyori, R. J. Am. Chem. Soc. 2000, 122, 1466– 1478. (114) Petra, D. G. I.; Reek, J. N. H.; Handgraaf, J.-W.; Meijer, E. J.; Dierkes, P.; Kamer, P. C. J.; Brussee, J.; Schoemaker, H. E.; van Leeuwen, P. W. N. M. Chem. Eur. J. 2000, 6, 2818–2829. (115) Handgraaf, J.-W.; Meijer, E. J. J. Am. Chem. Soc. 2007, 129, 3099–3103. (116) Lide, D. R. CRC Handbook of Chemistry and Physics, 86th ed.; CRC Press:, 2005. (117) Itoh, T.; Nagata, K.; Kurihara, A.; Miyazaki, M.; Ohsawa, A. Tetrahedron Lett. 2002, 43, 3105–3108. (118) a) Wu, X.; Liu, J.; Di Tommaso, D.; Iggo, Jonathan A.; Catlow, C. Rich- ard A.; Bacsa, J.; Xiao, J. Chem. Eur. J. 2008, 14, 7699–7715. b) Wu, X.; Li, X.; King, F.; Xiao, J. Angew. Chem. Int. Ed. 2005, 44, 3407–3411. (119) Blackmond, D. G.; Ropic, M.; Stefinovic, M. Org. Proc. Res. Dev. 2006, 10, 457–463. (120) Martins, J. E. D.; Clarkson, G. J.; Wills, M. Org. Lett. 2009, 11, 847–850.