Homogeneous Catalysis of Nickel Hydride Complexes Bearing a Bis (Phosphinite) Pincer Ligand

Home , Boranes, Dihydrogen complex, Heptene, Titanium hydride, Transition metal hydride

Homogeneous Catalysis of Nickel Hydride Complexes Bearing a Bis(phosphinite) Pincer Ligand

A Dissertation Submitted to the Graduate School of the University of Cincinnati in Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY (Ph.D.)

in the Department of Chemistry of McMicken College of Arts and Sciences by

Sumit Chakraborty

Master of Science (M. S.), Chemistry Indian Institute of Technology, Kanpur, India, 2007

Bachelor of Science (B. S.), Chemistry Hooghly Mohsin College, Burdwan University, India, 2005

Dissertation Advisor: Hairong Guan, Ph.D.

Homogeneous Catalysis of Nickel Hydride Complexes Bearing a Bis(phosphinite) Pincer Ligand

ABSTRACT

This dissertation is focused on the synthesis of well-defined nickel hydride complexes bearing a bis(phosphinite) pincer ligand (commonly known as a POCOP ligand) and utilities of these metal complexes in varieties of useful organic transformations such as reduction of carbonyl functionalities present in aldehydes, ketones and CO2, dehydrogenation of formic acid, isomerization of terminal olefins, and cyanomethylation of aldehydes.

Aldehydes insert cleanly and selectively into the Ni-H bonds of (POCOP)NiH complexes to form nickel alkoxide complexes. These nickel alkoxide compounds further react with phenylsilane to regenerate (POCOP)NiH and produce silyl ethers. Based on these observations, an efficient and chemoselective hydrosilylation process has been developed utilizing nickel hydrides as catalysts. The process is highly compatible with varieties of functional groups in aldehydes. A nickel hydride complex with smaller substituents on the POCOP ligand proves to be more effective hydrosilylation catalyst. In case of ketones, partial hydrosilylation occurs.

(POCOP)NiH complexes also react with CO2 to produce nickel formate complexes.

When stoichiometric amounts of boranes are used, nickel hydrides are cleanly reformed. The use of catalytic amounts of nickel hydrides and excess of boranes reduces CO2 to the corresponding methanol derivatives. The initial reduction products can be further hydrolyzed to yield methanol. The overall transformation is comprised of three catalytic cycles; in each cycle the formal oxidation state of CO2 is reduced by 2, followed by the consumption of one equivalent of borane. A catalyst with more bulky substituents on the phosphorus atoms of pincer ligand is a more effective catalyst than those containing smaller substituents. This

i phenomenon has been rationalized by invoking more favorable dihydridoborate adduct formation between less bulky nickel hydrides and boranes. One of such dihydridoborate adduct has been successfully isolated and its influence on the catalytic CO2 reduction has been demonstrated.

(POCOP)NiH complexes have been found be active catalysts for the decomposition of formic acid to release dihydrogen. When the kinetics the reaction is monitored by in-situ IR spectroscopy, a unique sigmoidal pattern is observed. Several mechanistic possibilities such as heterogeneity, cooperativity, and formation of other reactive intermediates have been experimentally tested to understand the unusual kinetic behavior.

A mechanistic investigation pertaining to NiH-catalyzed isomerization of terminal olefins such as 1-butene has been carried out. Two major pathways for olefin isomerization reaction, namely (a) metal hydride addition-elimination and (b) π-allyl metal hydride have been ruled out on the basis of reactivity studies of branched nickel butyl complex and labeling experiments. An

H-atom transfer pathway has been proposed based on the radical inhibitor studies. In addition to the olefin isomerization, the branched nickel butyl complex isomerizes to the linear nickel butyl complex possibly via a unique intramolecular pathway.

Nickel hydrides efficiently catalyze the cyanomethylation of aldehydes to produce β- hydroxy nitriles. This reaction does not require an external base which is mandatory for other related systems. The (POCOP)Ni-CH2CN complex, a potential intermediate in this reaction, has been isolated and characterized. This compound reacts directly with benzaldehyde to produce the β-hydroxy nitrile. The nickel alkoxide complexes that result from the insertion reaction between nickel hydrides and aldehydes, act as internal bases in this system. Varieties of aldehydes including base-sensitive aldehydes have been successfully coupled with acetonitrile using nickel hydride catalysts.

iii

Homogeneous Catalysis of Nickel Hydride Complexes Bearing a Bis(phosphinite) Pincer Ligand

TABLE OF CONTENTS

Chapter 1: Introduction

1.1 Reduction of Carbonyl Functionalities with First-Row Transition Metals 2 1.2 Catalysis Involving the Insertion of Carbonyl Substrates into an M-H Bond 6 1.3 Our Mechanistic Hypothesis 10 1.4 Nickel Hydrides in Stoichiometric and Catalytic Studies 12 1.5 Goals of the Research Project 15

Chapter 2: Catalytic Hydrosilylation of Aldehydes and Ketones

2.1 Introduction 17 2.2 Synthesis of Nickel Pincer Hydride Complexes 20 2.3 Stoichiometric Reduction of Aldehydes and Ketones with Nickel Hydride Complexes 24 2.4 Catalytic Hydrosilylation of Aldehydes and Ketones with Nickel Hydride Complexes. 28 2.5 Conclusions 35 2.6 Experimental Section 35

Chapter 3: Catalytic Reduction of CO2 to Methanol with Simple Boranes

3.1 Introduction 44

3.2 CO2 Insertion into Ni-H Bonds of Nickel Pincer Hydride Complexes 47 3.3 Regeneration of Nickel Hydrides from Nickel Formate Complexes 51

3.4 Catalytic Reduction of CO2 with Catecholborane 54

3.5 Comparative Studies of CO2 Reduction with Different Nickel Catalysts and Boranes 57 3.6 Reactions of Nickel Hydride Complexes with Boranes 61

3.6.1 Reactions with BH3•THF 64 3.6.2 Reactions with 9-BBN 73 3.6.3 Reactions with HBcat 79 3.7 Rationalization of Catalytic Activities 84 3.8 Conclusions 85 3.9 Experimental Section 85

Chapter 4: Catalytic Decomposition of Formic Acid to Release Dihydrogen

4.1 Introduction 103

4.2 Stoichiometric Steps Related to Ni-Catalyzed Dehydrogenation of HCO2H 108

4.3 Catalytic Decomposition of HCO2H 109

4.4 Kinetic Studies of Catalytic Decomposition of HCO2H 111 4.4.1 Heterogeneity 112 4.4.2 Metal-Metal Cooperativity 113 4.4.3 Formation of Other Reactive Intermediates 114 4.4.4 Dual Processes Catalyzed by Nickel Hydrides 115 4.5 Conclusions 117 4.6 Experimental Section 118

Chapter 5: Mechanistic Insight of Nickel-Catalyzed Olefin Isomerization Reaction

5.1 Introduction 123 5.2 Catalytic Isomerization of 1-Butene to 2-Butene 124 5.3 Metal Hydride Addition-Elimination Pathway 125 5.4 π-Allyl Metal Hydride Pathway 130 5.5 Hydrogen Atom Transfer Pathway 132 5.6 Isomerization of Branched Nickel Alkyl Complex to Linear Nickel Alkyl Complex 135 5.7 Conclusions 139 5.8 Experimental Section 139

Chapter 6: Catalytic Cyanomethylation of Aldehydes

6.1 Introduction 146 6.2 Transition Metal-Catalyzed Cyanomethylation of Aldehydes and Ketones 147 6.3 Catalytic Cyanomethylation of Aldehydes with Nickel Hydride Complexes 149 6.4 Synthesis of Nickel Cyanomethyl Complex 152 6.5 Reactivities of Nickel Cyanomethyl Complex 153 6.6 Conclusions 158 6.7 Experimental Section 159

Homogeneous Catalysis of Nickel Hydride Complexes Bearing a Bis(phosphinite) Pincer Ligand

LIST OF FIGURES

Chapter 1: Introduction

Figure 1 Iron-Based Achiral and Chiral Catalysts for the Reduction of Carbonyl Groups 4

Chapter 2: Catalytic Hydrosilylation of Aldehydes and Ketones

t Figure 1 X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]NiCl (2b) 21 t Figure 2 X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]NiH (3b) 23

Chapter 3: Catalytic Reduction of CO2 to Methanol with Simple Boranes

i Figure 1 X-ray Crystal Structure of [2,6-( Pr2PO)2C6H3]NiOC(O)H (9a) 49 t Figure 2 X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]NiOC(O)H (9b) 50 c Figure 3 X-ray Crystal Structure of [2,6-( Pe2PO)2C6H3]NiOC(O)H (9c) 50 Figure 4 1H NMR for the Reduction of 9b-13C with HBcat 53 Figure 5 13C{1H} NMR for the Reduction of 9b-13C with HBcat 53 Figure 6 Possible Products from the Reaction of a Metal Hydride Complex with a Borane 62 i 2 Figure 7 X-ray Crystal Structure of [2,6-( Pr2PO)2C6H3]Ni(η -BH4) (10a) 69 t 2 Figure 8 X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]Ni(η -BH4) (10b) 70 c 2 Figure 9 X-ray Crystal Structure of [2,6-( Pe2PO)2C6H3]Ni(η -BH4) (10c) 71 c Figure 10 X-ray Crystal Structure of [2,6-( Pe2PO)2C6H3]Ni(µ-H)2[BC8H14] (11c) 77 Figure 11 Variable-Temperature 1H NMR Spectra of the Mixture of 3b and HBcat 80 Figure 12 Variable-Temperature 1H NMR spectra of the 1 : 1 Mixture of 3a and HBcat 82 Figure 13 Temperature-Dependent 11B NMR Study of a Mixture of 3a and HBcat 83

vii

Chapter 4: Catalytic Decomposition of Formic Acid to Release Dihydrogen

Figure 1 Selected Examples of Homogeneous Catalysts for the Decomposition of Formic Acid 105

Figure 2 Kinetics of Ni-Catalyzed Dehydrogenation of HCO2H 112 Figure 3 Structures of Potential Reactive Intermediates 114 Figure 4 Successive-Addition Kinetics of Ni-Catalyzed Dehydrogenation

of HCO2H 115

Chapter 5: Mechanistic Insight of Nickel-Catalyzed Olefin Isomerization Reaction

i sec Figure 1 X-ray Crystal Structure of [2,6-( Pr2PO)2C6H3]Ni(Bu ) (14a) 127

viii

Homogeneous Catalysis of Nickel Hydride Complexes Bearing a Bis(phosphinite) Pincer Ligand

LIST OF TABLES

Chapter 2: Catalytic Hydrosilylation of Aldehydes and Ketones

i Table 1 Stoichiometric Reactivity of [2,6-( Pr2PO)2C6H3]Ni(OCH2Ph) (4a) toward Different Silanes 28 Table 2 Comparing the Catalytic Activities of 3a-c in Hydrosilylation Reaction 29 Table 3 Hydrosilylation of Aldehydes Catalyzed by Nickel Pincer Hydride 31 Table 4 Crystal Data and Refinement Parameters for 2b and 3b 42

Chapter 3: Catalytic Reduction of CO2 to Methanol with Simple Boranes

Table 1 Catalytic Activity of Nickel Hydride Complexes in the Reduction of CO2 60 Table 2 Selected Infrared Frequencies (cm-1) of 10a-c (in Solid State) 68 Table 3 Crystal Data Collection and Refinement Parameters of 9a-c 101 Table 4 Crystal Data Collection and Refinement Parameters of 10a-c and 11c 101

Chapter 4: Catalytic Decomposition of Formic Acid to Release Dihydrogen

Table 1 Comparing Catalytic Activity of 9a-c in the Closed System 110 Table 2 Comparing Catalytic Activity of 9a-c in the Open System 110

Chapter 5: Mechanistic Insight of Nickel-Catalyzed Olefin Isomerization Reaction

Table 1 Crystal Data and Refinement Parameters for 14a 144

Chapter 6: Catalytic Cyanomethylation of Aldehydes

Table 1 Comparing Catalytic Activity of 3a-c in Cyanomethylation Reaction 149 Table 2 Catalytic Cyanomethylation of Aldehydes with 3a 151

ACKNOWLEDGEMENTS

First, I would like to thank my advisor Professor Hairong Guan. Over the years, I have learnt a large amount of chemistry from Hairong. Being the first PhD student of Hairong, I learned most of my experimental skills directly from him. He taught me not only how to make a specific reaction work, but also more broadly how to turn a specific reaction into a complete research project. His enthusiasm and thoughtful suggestions, particularly when my projects were not cooperating, were essential to my success. I also thank him for always treating me with respect and care and being extremely patient with me. Besides science, I thank Hairong for teaching me baseball rules, discussing politics, and constantly pushing me to get the Ohio

Driver’s License (so far I have cleared by written test, and attempt to learn actual driving is underway!). Hypothetically, if I had to go back and do my PhD once again, I would choose

Hairong over others any day.

I was privileged to have Professor William B. Connick and Professor Allan R.

Pinhas on my dissertation committee. Their continuous suggestions and critical opinions helped me complete my research projects. I would also like to thank all of the other chemistry faculty members that I have come in contact with during graduate school.

My graduate research was greatly aided by two staff scientists. Dr. Jeanette A.

Krause has taught me a great amount of crystallography and I believe she is a good friend of mine. Also, Dr. Keyang Ding has always answered my NMR-related questions readily and cheerfully.

I have learned a lot from my fellow lab mates in the Guan group, and would like to thank all past and present members for making my experience in lab enjoyable. In particular, I like to thank our previous postdoctoral fellow, Dr. Jie Zhang, who I collaborated with on some of my projects. Present group members: Sanjeewa Rodrigo, Papri Bhattacharya, Anubendu

Adhikary, Arundhoti Chakraborty (my wife), Gleason Wilson and Nadeesha Wellala also helped me get through the day with entertaining conversations and occasional jokes. I will miss all of them. I wish them all the best for their graduate studies. Special thanks go to Sanjeewa, Jie and

Hairong for dropping me and picking me up from CVG airport whenever I travelled to some places.

I was fortunate to supervise some terrific undergraduate (Nick Huff, Emma Senour,

Yogi Patel, Jinal Patel, Ochea Johnson, James Ohaeri, Michael Gibson, and Chao Zhu) and high school students (Jeff Reynolds and Jonisha Baggett). I thank them for their patience and willingness to work on some of my suggestions. I wish them all the best for their future endeavors.

Lastly, I would like to thank my parents, brother, sister and my wife. All that I have accomplished in my life is the result of their continuous love and support. My wife, Arundhoti, has offered me the best life I can ever imagine. I thank her for the encouragement, motivation and support she has given me over the years.

Preface

Parts of this thesis have been adapted from articles co-written by the author. The following articles were reproduced in part with permission from the American Chemical Society,

Royal Society of Chemistry, and Elsevier:

(1) Chakraborty, S.; Krause, J. A.; Guan, H. "Hydrosilylation of Aldehydes and Ketones Catalyzed by Nickel PCP-Pincer Hydride Complexes." Organometallics 2009, 28, 582- 586.

(2) Chakraborty, S.; Guan, H. "First-Row Transition Metal Catalyzed Reduction of Carbonyl Functionality: A Mechanistic Perspective." Dalton Trans. 2010, 39, 7427-7436.

(3) Chakraborty, S.; Zhang, J.; Krause, J. A.; Guan, H. "An Efficient Nickel Catalyst for the Reduction of Carbon Dioxide with a Borane." J. Am. Chem. Soc. 2010, 132, 8872-8873.

(4) Chakraborty, S.; Patel, Y. J.; Krause, J. A.; Guan, H. "Catalytic Properties of Nickel Bis(phosphinite) Pincer Complexes in the Reduction of CO2 to Methanol Derivatives."Polyhedron 2012, 32, 30-34.

(5) Chakraborty, S.; Zhang, J.; Patel, Y. J.; Krause, J. A.; Guan, H. "Pincer-Ligated Nickel Hydridoborate Complexes: the Dormant Species in Catalytic Reduction of Carbon Dioxide with Boranes."Inorg. Chem. 2012, 51, in press.

xii

Chapter 1 Introduction

1.1 Reduction of Carbonyl Functionalities with First-Row Transition Metals

Development of efficient homogeneous catalysis using first-row transition metals such as iron and nickel is highly desirable because of their low cost and high abundance when compared to the precious second and third-row transition metals. However, the rich chemistry of second and third-row metals cannot be simply extrapolated to their first-row congeners. The reactivities of first-row transition metal compounds are often accompanied with paramagnetism, weak metal-carbon bonds, facile change of oxidation states, and radical intermediates.1 In one hand, these properties are unique to these metals; on the other hand, they often pose great challenges for the rational design of well-defined catalytic systems. As a result, very limited amount of information is known in the literature about the kinetics and thermodynamics of bond-making and bond-breaking steps involving first-row transition metal complexes.

One of the important reactions in homogeneous catalysis is to reduce an unsaturated chemical bond using a transition metal catalyst. For the reduction of carbonyl groups in particular, the formation of metal hydrides, the key catalytic intermediates, is not favorable for first-row transition metals. This is in part due to their weaker MH bonds than second and third- row transition metal complexes.2 Even if they are formed, first-row metal hydrides tend to have relatively lower kinetic and thermodynamic hydricities. In spite of these obstacles, some progress has been made in first-row transition metal-catalyzed reduction of carbonyl functionalities (Figure 1). Chirik and coworkers have conducted their investigation on bis(imido)pyridine iron complexes as highly efficient catalysts for the hydrosilylation of

1 Cotton, F. A.; Wilkinson, G.; Murillo, C. A.; Bochmann, M. Advanced Inorganic Chemistry, 6th ed.; John Wiley & Sons, Inc: New York, 1999; pp 775-854. 2 Uddin, J.; Morales, C. M.; Maynard, J. H.; Landis, C. R. Organometallics 2006, 25, 5566.

3,4 aldehyde and ketone. More practical catalytic methods using the combination of Fe(OAc)2 and a phosphine ligand5 or chelating diamine ligand6 have been reported by the Beller and Nishiyama groups. Asymmetric versions of these catalysts have also been demonstrated by the groups of

Chirik,7 Beller,8 and Gade (Figure 1).9 Although these transformations are highly promising, mechanistic details are limited. Our group has recently reported pincer ligand-supported iron hydride complexes as efficient hydrosilylation catalysts.10 In this system, the hydride moiety remains as a spectator ligand during the catalysis, but its high trans-directing ability helps create a vacant coordination site for substrate binding.

3 Bart, S. C.; Lobkovsky, E.; Chirik, P. J. J. Am. Chem. Soc. 2004, 126, 13794. 4 Tondreau, A. M.; Lobkovsky, E.; Chirik, P. J. Org. Lett. 2008, 10, 2789. 5 Shaikh, N. S.; Junge, K.; Beller, M. Org. Lett. 2007, 9, 5429. 6 Nishiyama, H.; Furuta, A. Chem. Commun. 2007, 760. 7 Tondreau, A. M.; Darmon, J. M.; Wile, B. M.; Floyd, S. K.; Lobkovsky, E.; Chirik, P. J. Organometallics 2009, 28, 3928. 8 Shaikh, N. S.; Enthaler, S.; Junge, K.; Beller, M. Angew. Chem. Int. Ed. 2008, 47, 2497. 9 Langlotz, B. K.; Wadepohl, H.; Gade, L. H. Angew. Chem. Int. Ed. 2008, 47, 4670. 10 Bhattacharya, P., Krause, J. A.; Guan, H. Organometallics 2011, 30, 4720. 3

Figure 1. Iron-Based Achiral and Chiral Catalysts for the Reduction of Carbonyl Groups

Iron-based metal-ligand bifunctional catalysis provides another strategy to reduce carbonyl moieties. Casey and Guan have demonstrated that a hydroxycyclopentadienyl iron dicarbonyl hydride, first reported by Knölker,11 catalyzes efficient and chemoselecitve hydrogenation of ketones (Scheme 1).12 With aromatic aldehydes as the substrates for stoichiometric reduction reactions, cyclopentadienone Fe(0) alcohol complexes have been successfully synthesized, and one of the compounds has also been studied by X-ray crystallography (Scheme 1).13

11 Knölker, H.-J.; Baum, E.; Goesmann, H.; Klauss, R. Angew. Chem. Int. Ed. 1999, 38, 2064. 12 Casey, C. P.; Guan, H. J. Am. Chem. Soc. 2007, 129, 5816. 13 Casey, C. P.; Guan, H. J. Am. Chem. Soc. 2009, 131, 2499. 4

Complexes based on other first-row transition metals have also been explored as catalysts for the reduction of carbonyl functionalities. For example, Iyer and Varghese have shown that

14 NiCl2(PPh3)2 can catalyze transfer hydrogenation of ketones. They have proposed a dihydrido intermediate in this process; however, no experimental evidence has been provided to support the existence of such an intermediate. Manganese complexes such as (5-1H-

6 hydronaphthalene)Mn(CO)3 and [( -naphthalene)Mn(CO)3]BF4 have exhibited catalytic activity for the hydrosilylation of ketones.15 It has been suggested that the hapticity change (53 and

64) of the ligand leaves a vacant coordination site for -complexation of the silane.

14 Iyer, S.; Varghese, J. P. J. Chem. Soc., Chem. Commun. 1995, 465. 15 (a) Son, S. U.; Paik, S.-J.; Lee, I. S.; Lee, Y.-A.; Chung, Y. K.; Seok, W. K.; Lee, H. N. Organometallics 1999, 18, 4114. (b) Son, S. U.; Paik, S.-J.; Chung, Y. K. J. Mol. Catal. A: Chem. 2000, 151, 87. 5

1.2 Catalysis Involving the Insertion of Carbonyl Substrates into an M-H Bond

From the mechanistic point of view, insertion of carbonyl group into a metal-hydrogen bond, followed by the cleavage of metal-oxygen bond with silanes, H2 or other reducing sources would produce the reduced product catalytically (Scheme 2). Compared to the well-studied alkene insertion into metal hydrides, mechanistic insights of the insertion of C=O into a metal- hydrogen bond are surprisingly rare, particularly for the first-row transition metals. It is often assumed that a vacant coordination site is required for the insertion, and a metathesis-type process (or an oxidative addition-reductive elimination sequence) is responsible for the regeneration of the metal hydride.

This type of reaction pattern is well precedented, especially in titanium-catalyzed reduction of aldehydes and ketones. One of the early reports by the Sato group has shown that the reaction of Grignard reagents with carbonyl substrates in the presence of a catalytic amount of Cp2TiCl2 affords predominantly the reduction products, rather than the expected alkyl addition products (Scheme 3).16 These results have been explained by the insertion of carbonyl groups into a Ti(III) hydride species, which is presumably generated from the reduction of Ti(IV) to Ti(III) followed by a -hydride elimination step. A similar

16 Sato, F.; Jinbo, T.; Sato, M. Tetrahedron Lett. 1980, 21, 2171. 6 mechanism has been proposed for the hydrosilylation of ketones catalyzed by Cp 2TiPh2, except that the regeneration of titanium hydride is accomplished by a silane. 17

The development of chiral titanocene complexes has further advanced the field and led to one of the synthetically useful protocols for the asymmetric hydrosilylation of ketone.18 For the reactions of aromatic ketones, alcohol products with ee’s as high as 99% have been obtained. The most enantioselective system involves (R,R)-(EBTHI)TiF2 (EBTHI

= ethylenebis(tetrahydroindenyl)) as a precatalyst and MeOH as an additive.18b Buchwald has attributed the high enantioselectivity to the favorable transition state with less steric interactions between the aromatic ring and the tetrahydroindenyl ring.18a,b This hypothesis is in accordance with Lauher and Hoffmann’s molecular orbital analysis on the insertion of

17 Nakano, T.; Nagai, Y. Chem. Lett. 1988, 481. 18 (a) Carter, M. B.; Schiøtt, B.; Gutiérrez, A.; Buchwald, S. L. J. Am. Chem. Soc. 1994, 116, 11667. (b) Yun, J.; Buchwald, S. L. J. Am. Chem. Soc. 1999, 121, 5640. (c) Halterman, R. L.; Ramsey, T. M.; Chen, Z. J. Org. Chem. 1994, 59, 2642. (d) Halterman, R. L.; Ramsey, T. M.; Chen, Z. J. Org. Chem. 1994, 59, 2642. (e) Xin, S.; Harrod, J. F. Can. J. Chem. 1995, 73, 999. (f) Rahimian, K.; Harrod, J. F. Inorg. Chim. Acta 1998, 270, 330. (g) Beagley, P.; Davies, P.; Adams, H.; White, C. Can. J. Chem. 2001, 79, 731. 7

19 alkenes into a Cp2TiH fragment. The intermediacy of the Ti(III) hydride in the catalytic cycle is still hard to verify desipte the fact that (EBTHI)TiH has been independently synthesized and crystallographically characterized as a dimer with the hydride ligands bridging two Ti centers.20 The chemical complexity of the actual catalytic systems, including various reactions between titanocene complexes and silanes,21 has not yet ruled out silyl

Ti(IV) hydrides, bimetallic titanium hydrides, or Ti(III) silyl compounds as the possible intermediates.18a-d

Late transition metal hydride complexes can also be used to catalyze the reduction of carbonyl functionalities. Bianchini et al. reported transfer hydrogenation of ,-unsaturated ketones catalyzed by a well-defined iron(II) cis-hydride 2-dihydrogen complex (Scheme

4).22 The chemoselectivity is highly sensitive to the structures of the substrates, and simple ketones are often resistant to the reduction conditions. Nevertheless, this research group has demonstrated that the dissociation of the dihydrogen ligand creates the necessary vacant coordination site for the binding of ketones and carbonyl insertion. Alkoxide ligand exchange with 2-propanol followed by -hydride elimination regenerates the coordinatively unsaturated iron hydride for the reduction of another ketone molecule. This mechanism is consistent with the observation that both H2 and N2 inhibit the catalytic reactions.

19 Lauher, J. W.; Hoffmann, R. J. Am. Chem. Soc. 1976, 98, 1729. 20 Xin, S.; Harrod, J. F.; Samuel, E. J. Am. Chem. Soc. 1994, 116, 11562. 21 (a) Samuel, E.; Harrod, J. F. J. Am. Chem. Soc. 1984, 106, 1859. (b) Aitken, C. T.; Harrod, J. F.; Samuel, E. J. Am. Chem. Soc. 1986, 108, 4059. (c) Harrod, J. F.; Yun, S. S. Organometallics 1987, 6, 1381. (d) Harrod, J. F.; Yun, S. S. Organometallics 1987, 6, 1381. (e) Harrod, J. F.; Ziegler, T.; Tschinke, V. Organometallics 1990, 9, 897. 22 Bianchini, C.; Farnetti, E.; Graziani, M.; Peruzzini, M.; Polo, A. Organometallics 1993, 12, 3753. 8

More recent examples of iron-catalyzed transfer hydrogenation of ketones have been reported by the Beller group (Scheme 5).23 The structure of the active catalyst is unclear at the moment, although experiments with iPrOD seem to support a major reaction pathway involving a monohydride intermediate.24 A related transfer hydrogenation system catalyzed

i 25 by [Fe3(CO)12]/porphyrin/ PrONa may follow a similar mechanism.

The Mindiola group recently reported a nickel catalyst system for the hydrosilylation of aldehydes and ketones.26 In this case, a transient nickel hydride complex from the

t reduction of a nickel bromide by KO Bu/Et3SiH, is the active catalyst (Scheme 6).

Presumably, this dimeric hydride dissociates to a monomeric hydride in solution, providing the vacant coordination site for the migratory insertion of organic carbonyls. The cleavage of the newly formed NiO bond by a silane releases the product and closes the catalytic cycle. This chemistry is reminiscent of Stryker’s [(Ph3P)CuH]6-catalyzed hydrosilylation

23 Enthaler, S.; Hagemann, B.; Erre, G.; Junge, K.; Beller, M. Chem. Asian J. 2006, 1, 598. 24 The reaction scheme was reproduced from Ref. 23. The minor product might be PhC(D)(Me)(OD) rather than PhC(D)(Me)(OH) as the latter is expected to undergo a rapid H/D exchange with the solvent iPrOD. 25 Enthaler, S.; Erre, G.; Tse, M. K.; Junge, K.; Beller, M. Tetrahedron Lett. 2006, 47, 8095. 26 Tran, B. L.; Pink, M.; Mindiola, D. J. Organometallics 2009, 28, 2234. 9 and hydrogenation of ketones.27 More recently, new efforts focusing on smaller copper hydride aggregates as well as on asymmetric reduction reactions have been made by Nolan28 and Lipshutz.29

1.3 Our Mechanistic Hypothesis

If a metal center does not possess a vacant coordination site, metal hydrides may have higher stability. However, when an appropriate ligand is chosen, the hydricity of the metal- hydrogen bond could be enhanced, so that direct hydride transfer to the polarized C=O bond may occur, leading to the formation of the reduced carbonyl species (Scheme 7). In order to increase the hydride donor ability of a metal hydride complex, a strong trans-influencing ligand needs to

27 (a) Mahoney, W. S.; Stryker, J. M. J. Am. Chem. Soc. 1989, 111, 8818. (b) Chen, J.-X.; Daeuble, J. F.; Brestensky, D. M.; Stryker, J. M. Tetrahedron 2000, 56, 2153. 28 (a) Kaur, H.; Zinn, F. K.; Stevens, E. D.; Nolan, S. P. Organometallics 2004, 23, 1157. (b) Díez-González, S.; Kaur, H.; Zinn, F. K.; Stevens, E. D.; Nolan, S. P. J. Org. Chem. 2005, 70, 4784. (c) Díez-González, S.; Scott, N. M.; Nolan, S. P. Organometallics 2006, 25, 2355. (d) Díez-González, S.; Stevens, E. D.; Scott, N. M.; Petersen, J. L.; Nolan, S. P. Chem. Eur. J. 2008, 14, 158. (e) Díez-González, S.; Nolan, S. P. Aldrichimica Acta 2008, 41, 43. (f) Díez-González, S.; Nolan, S. P. Acc. Chem. Res. 2008, 41, 349. 29 (a) Lipshutz, B. H.; Lower, A.; Noson, K. Org. Lett. 2002, 4, 4045. (b) Lipshutz, B. H.; Noson, K.; Chrisman, W.; Lower, A. J. Am. Chem. Soc. 2003, 125, 8779. (c) Lipshutz, B. H.; Noson, K.; Chrisman, W.; Lower, A. J. Am. Chem. Soc. 2003, 125, 8779. 10 be installed opposite to the hydride moiety. Arene-based ligands are known to serve as good trans-directing groups.30 In addition, such a ligand set also offers great flexibility in terms of steric and electronic modifications. Moreover, metal complexes with pincer ligands have shown numerous unique applications in organic synthesis, polymerization, and molecular sensing.31 I aimed to synthesize well-defined square-planar nickel hydrides bearing a bis(phosphinite) pincer ligand (Scheme 7, ligand shown in red). This specific ligand set was chosen for a number of reasons. First, this type of ligands has been demonstrated to form robust metal complexes.32 The resulting compounds have also shown excellent reactivities in Pd-catalyzed cross-coupling reactions33 and Ir-catalyzed alkane dehydrogenation reactions.32 It was anticipated that the phenyl backbone would exert a strong trans-influence and render the nickel hydride sufficiently hydridic. In addition, the synthesis of the pincer ligands involves relatively inexpensive materials. I focused on the square-planar molecules as they are considered as coordinatively

2 “saturated” compounds. The sterically accessible trajectory contains an occupied dz orbital,

2 2 whereas the unoccupied dx -y orbital is blocked by the ligand.

30 Inorganic Chemistry, 4th edition by Garry L. Miesslar and Donald A. Tarr. 31 (a) Gossage, R. A.; van de Kuil, L. A.; van Koten, G. Acc. Chem. Res. 1998, 31, 423-431. (b) Albrecht, M.; van Koten, G. Angew. Chem. Int. Ed. 2001, 40, 3750. (c) Singleton, J. T. Tetrahedron 2003, 59, 1837. (d) van der Boom, M. E.; Milstein, D. Chem. Rev. 2003, 103, 1759. (e) Liang, L.-C. Coord. Chem. Rev. 2006, 250, 1152. (f) Nishiyama, H. Chem. Soc. Rev. 2007, 36, 1133. 32 Göttker-Schnetmann, I.; White, P.; Brookhart, M. J. Am. Chem. Soc. 2004, 126, 1804. 33 (a) Bedford, R. B.; Draper, S. M.; Scully, P. N.; Welch, S. L. New J. Chem. 2000, 24, 745. (b) Morales-Morales, D.; Grause, C.; Kasaoka, K.; Redón, R.; Cramer, R. E.; Jensen, C. M. Inorg. Chim. Acta 2000, 300, 958. 11

1.4 Nickel Hydrides in Stoichiometric and Catalytic Studies

Nickel hydride complexes are important in the research areas of homogeneous catalysis, coordination chemistry, and enzymatic reaction mechanisms. They are often postulated as key intermediates in a variety of nickel-catalyzed organic transformations such as isomerization of olefins34 and enynes,35 dimerization or oligomerization of olefins,36 and reduction of unsaturated double bonds37 (Scheme 8, one representative example is shown). These hypothesized nickel hydrides are usually too reactive to allow direct observation of reaction intermediates or thorough investigation of reaction mechanisms. Therefore, details of their existence during catalytic processes largely rely on computational studies.

34 (a) Gosser, L. W.; Parshall, G. W. Tetrahedron Lett. 1971, 12, 2555. (b) Bontempelli, G.; Fiorani, M.; Daniele, S.; Schiavon, G. J. Mol. Catal. 1987, 40, 9. (c) Raje, A. P.; Datta, R. J. Mol. Catal. 1992, 72, 97. (d) Frauenrath, H.; Brethauer, D.; Reim, S.; Maurer, M.; Raabe, G. Angew. Chem. Int. Ed. 2001, 40, 177. (e) Cuperly, D.; Petrignet, J.; Crévisy, C.; Grée, R. Chem. Eur. J. 2006, 12, 3261. 35 Tekavec, T. N.; Louie, J. Tetrahedron 2008, 64, 6870. 36 (a) Maruya, K.-i.; Mizoroki, T.; Ozaki, A. Bull. Chem. Soc. Jpn. 1972, 45, 2255. (b) Peuckert, M.; Keim, W. Organometallics 1983, 2, 594. (c) Müller, U.; Keim, W.; Krüger, C.; Betz, P. Angew. Chem. Int. Ed. Engl. 1989, 28, 1011. (d) Keim, W. Angew. Chem. Int. Ed. Engl. 1990, 29, 235. (e) Bertozzi, S.; Iannello, C.; Barretta, G. U.; Vitulli, G.; Lazzaroni, R.; Salvadori, P. J. Mol. Catal. 1992, 77, 1. (f) Fan, L.; Krzywicki, A.; Somogyvari, A.; Ziegler, T. Inorg. Chem. 1994, 33, 5287. (g) Fan, L.; Krzywicki, A.; Somogyvari, A.; Ziegler, T. Inorg. Chem. 1996, 35, 4003. (h) Brown, J. M.; Hughes, G. D. Inorg. Chim. Acta 1996, 252, 229. (i) Wiencko, H. L.; Kogut, E.; Warren, T. H. Inorg. Chim. Acta 2003, 345, 199. (j) Kogut, E.; Zeller, A.; Warren, T. H.; Strassner, T. J. Am. Chem. Soc. 2004, 126, 11984. (k) Wang, K.; Patil, A. O.; Zushma, S.; McConnachie, J. M. J. Inorg. Biochem. 2007, 101, 1883. 37 (a) Brunet, J. J.; Mordenti, L.; Loubinoux, B.; Caubere, P. Tetrahedron Lett. 1977, 18, 1069. (b) Brunet, J. J.; Mordenti, L.; Caubere, P. J. Org. Chem. 1978, 43, 4804. (c) Chow, Y. L.; Li, H. Can. J. Chem. 1986, 64, 2229. (d) Sakai, M.; Hirano, N.; Harada, F.; Sakakibara, Y.; Uchino, N. Bull. Chem. Soc. Jpn. 1987, 60, 2923. (e) Chow, Y. L.; Li, H.; Yang, M. S. J. Chem. Soc., Perkin Trans. 2 1990, 17. 12

In contrast, a number of discrete and stable nickel hydride complexes have been prepared and used in many useful stoichiometric transformations such as C-S, C-C and C-H bond activation reactions38 (Scheme 9). Other fundamental studies related to well-defined nickel hydrides39 include the hydride donor ability,40 protonation reactions,41 and etc. However, very

38 (a) Vicic, D. A.; Jones, W. D. J. Am. Chem. Soc. 1997, 119, 10855. (b) Vicic, D. A.; Jones, W. D. Organometallics 1998, 17, 3411. (c) Edelbach, B. L.; Vicic, D. A.; Lachicotte, R. J.; Jones, W. D. Organometallics 1998, 17, 4784. (d) Vicic, D. A.; Jones, W. D. J. Am. Chem. Soc. 1999, 121, 7606. (e) Garcia, J. J.; Jones, W. D. Organometallics 2000, 19, 5544. (f) Garcia, J. J.; Brunkan, N. M.; Jones, W. D. J. Am. Chem. Soc. 2002, 124, 9547. (g) Brunkan, N. M.; Brestensky, D. M.; Jones, W. D. J. Am. Chem. Soc. 2004, 126, 3627. (h) Garcia, J. J.; Arévalo, A.; Brunkan, N. M.; Jones, W. D. Organometallics 2004, 23, 3997. (i) Ateşin, T. A.; Li, T.; Lachaize, S.; Brennessel, W. W.; Garcia, J. J.; Jones, W. D. J. Am. Chem. Soc. 2007, 129, 7562. (j) Swartz, B. D.; Reinartz, N. M.; Brennessel, W. W.; Garcia, J. J.; Jones, W. D. J. Am. Chem. Soc. 2008, 130, 8548. 39 (a) Srivastava, S. C.; Bigorgne, M. J. Organomet. Chem. 1969, 18, P30. (b) Green, M. L. H.; Saito, T. Chem. Commun. 1969, 208. (c) Schunn, R. A. Inorg. Chem. 1970, 9, 394. (d) Tolman, C. A. J. Am. Chem. Soc. 1970, 92, 4217. (e) Nesmeyanov, A. N.; Isaeva, L. S.; Lorens, L. N. J. Organomet. Chem. 1977, 129, 421. (f) Rigo, P.; Bressan, M.; Basato, M. Inorg. Chem. 1979, 18, 860. (g) Darensbourg, D. J.; Darensbourg, M. Y.; Goh, L. Y.; Ludvig, M.; Wiegreffe, P. J. Am. Chem. Soc. 1987, 109, 7539. (h) Chen, W.; Shimada, S.; Tanaka, M.; Kobayashi, Y.; Saigo, K. J. Am. Chem. Soc. 2004, 126, 8072. (i) Ozerov, O. V.; Guo, C.; Fan, L.; Foxman, B. M. Organometallics 2004, 23, 5573. (j) Clement, N. D.; Cavell, K. J.; Jones, C.; Elsevier, C. J. Angew. Chem. Int. Ed. 2004, 43, 1277. (k) Liang, L.-C.; Chien, P.-S.; Huang, Y.-L. J. Am. Chem. Soc. 2006, 128, 15562. (l) She, L.; Li, X.; Sun, H.; Ding, J.; Frey, M.; Klein, H.-F. Organometallics 2007, 26, 566. (m) Liang, L.-C.; Chien, P.-S.; Lee, P.- 13 few of these complexes are catalytically competent; of the known well-defined catalytic systems, nickel hydride complexes are solely used as catalysts for olefin isomerization or for oligomerization.42 This limitation of utilizing nickel hydrides in catalysis has further prompted us to prepare bis(phosphinite) pincer ligand-supported of nickel hydride complexes and to explore their potentials as relatively inexpensive metal catalysts for various organic reactions, particularly the reduction of carbonyl functionalities.

Y. Organometallics 2008, 27, 3082. (n) Vicic, D. A.; Anderson, T. J.; Cowan, J. A.; Schultz, A. J. J. Am. Chem. Soc. 2004, 126, 8132. (o) Tyree, W. S.; Vicic, D. A.; Piccoli, P. M. B.; Schultz, A. J. Inorg. Chem. 2006, 45, 8853. 40 (a) Miedaner, A.; DuBois, D. L.; Curtis, C. J.; Haltiwanger, R. C. Organometallics 1993, 12, 299. (b) Berning, D. E.; Noll, B. C.; DuBois, D. L. J. Am. Chem. Soc. 1999, 121, 11432. (c) Berning, D. E.; Miedaner, A.; Curtis, C. J.; Noll, B. C.; DuBois, M. C. R.; DuBois, D. L. Organometallics 2001, 20, 1832. (d) Curtis, C. J.; Miedaner, A.; Ellis, W. W.; DuBois, D. L. J. Am. Chem. Soc. 2002, 124, 1918. (e) Curtis, C. J.; Miedaner, A.; Ciancanelli, R.; Ellis, W. W.; Noll, B. C.; DuBois, M. R.; DuBois, D. L. Inorg. Chem. 2003, 42, 216. (f) Curtis, C. J.; Miedaner, A.; Raebiger, J. W.; DuBois, D. L. Organometallics 2004, 23, 511. (g) Fraze, K.; Wilson, A. D.; Appel, A. M.; DuBois, M. R.; DuBois, D. L. Organometallics 2007, 26, 3918. (h) Nimlos, M. R.; Chang, C. H.; Curtis, C. J.; Miedaner, A.; Pilath, H. M.; DuBois, D. L. Organometallics 2008, 27, 2715. 41 (a) James, T. L.; Cai, L.; Muetterties, M. C.; Holm, R. H. Inorg. Chem. 1996, 35, 4148. (b) Wilson, A. D.; Shoemaker, R. K.; Miedaner, A.; Muckerman, J. T.; DuBois, D. L.; DuBois, M. R. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 6951. 42 (a) Tolman, C. A. J. Am. Chem. Soc. 1970, 92, 6777. (b) Green, M. L. H.; Munakata, H. J. Chem. Soc., Dalton Trans. 1974, 269.

1.5 Goals of the Research Project

The goals of my dissertation research are to synthesize well-defined nickel hydride complexes bearing a bis(phosphinite) pincer ligand and to utilize them in varieties of useful organic transformations. My initial focus was placed on the reduction of carbonyl functionalities such as those in aldehydes, ketones and CO2. As the research progressed, I also discovered that these nickel hydride complexes can be used as catalysts in dehydrogenation of formic acid, isomerization of terminal olefins, and cyanomethylation of aldehydes.

Chapter 2 Catalytic Hydrosilylation of Aldehydes and Ketones

2.1 Introduction

Reduction of carbonyl functionalities, specifically those in aldehydes and ketones is an important transformation in organic synthesis.1 Use of stoichiometric quantities of main-group hydrides such as NaBH4 and LiAlH4 for the reduction of aldehydes and ketones to alcohols is a common practice, but of concern due to large amounts of waste produced particularly when these reactions are performed on large scales. Reduction of carbonyl group followed by the protection of the alcohol moiety would be attractive from a synthetic point of view. In this context, catalytic hydrosilylation reaction is important because both the reduction of carbonyl groups and the protection of the resulting alcohol with silyl moieties can be accomplished in a single step.

Traditionally, precious or heavy metals such as Re,2 Rh,3 Ru,4 Ir5 have been mostly employed for the catalytic hydrosilylation of carbonyl functionalities. However, the cost associated with these metals as well as the concern of their disposal have prompted the search for hydrosilylation catalysts based on inexpensive transition metals such as Ti,6 Cu,7 Fe8 and Zn.9 In recent years,

1 (a) Smith, M. B.; March, J. March’s Advanced Organic Chemistry; Wiley-Interscience: New York, 2001. (b) Ohkuma, T.; Noyori, R. In Comprehensive Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer: Berlin, 1999. (c) Nishiyama, H.; Itoh, K. In Catalytic Asymmetric Synthesis; Ojima, I., Ed.; Wiley-VCH: New York, 2000. 2 (a) Nolin, K. A.; Krumper, J. R.; Pluth, M. D.; Bergman, R. G.; Toste, F. D. J. Am. Chem. Soc. 2007, 129, 14684. (b) Du, G.; Abu-Omar, M. M. Organometallics 2006, 25, 4920. (c) Ison, E. A.; Trivedi, E. R.; Corbin, R. A.; Abu- Omar, M. M. J. Am. Chem. Soc. 2005, 127, 15374. 3 (a) Gade, L. H.; Cesar, V.; Bellemin-Laponnaz, S. Angew. Chem. Int. Ed. 2004, 43, 1014. (b) Nishiyama, H.; Sakaguchi, H.; Nakamura, T.; Horihata, M.; Kondo, M.; Itoh, K. Organometallics 1989, 8, 846. (c) Sawamura, M.; Ryoichi, K.; Ito, Y. Angew. Chem. Int. Ed. 1994, 33, 111. (d) Tao, B.; Fu, G. C. Angew. Chem. Int. Ed. 2002, 41, 3892. 4 (a) Zhu, G.; Terry, M.; Zhang, X. J. Organomet. Chem. 1997, 547, 97. (b) Nishibayashi, Y.; Takei, I.; Uemura, S. Organometallics 1998, 17, 3420. 5 (a) Chianese, A. R.; Crabtree, R. H. Organometallics 2005, 24, 3066. (b) Ojima, I.; Nihonyanagi, M.; Nagai, Y. J. Chem. Soc., Chem. Commun. 1972, 938a. 6 (a) Carter, M. B.; Schiott, B.; Guiterrez, A.; Buchwald, S. L. J. Am. Chem. Soc. 1994, 116, 11667. (b) Yun, J.; Buchwald, S. L. J. Am. Chem. Soc. 1999, 121, 5640. 7 (a) Deutsch, C.; Krause, N.; Lipshutz, B. H. Chem. Rev. 2008, 108, 2916. (b) Lipshutz, B. H.; Lower, A.; Noson, K. Org. Lett. 2002, 4, 4045. (c) Lipshutz, B. H.; Frieman, B. A. Angew. Chem. Int. Ed. 2005, 44, 6345. (d) Lipshutz, B. H.; Noson, K.; Chrisman, W.; Lower, A. J. Am. Chem. Soc. 2003, 125, 8779. (e) Diez-Gonzalez, S.; Scott, N. M.; Nolan, S. P. Organometallics 2006, 25, 2355. (f) Kaur, H.; Zinn, F. K.; Stevens, E. D.; Nolan, S. P. Organometallics 2004, 23, 1157.

Cu(I), in the form of (IPr)CuOtBu (IPr: N,N′-bis(2,6-diisopropylphenyl)imidazol-2-ylidene), has been shown to act as a precatalyst for the hydrosilylation of carbonyl groups in the presence of various silanes (Scheme 1).7e,f The active catalyst for this transformation has been postulated as a copper hydride species. Sadighi and coworkers have recently isolated this important intermediate, a dimeric form of [(IPr)CuH], through the reaction between (IPr)Cu(OtBu) and

10 (EtO)3SiH.

8 (a) Shaikh, N. S.; Enthaler, S.; Junge, K.; Beller, M. Angew. Chem. Int. Ed. 2008, 47, 2497. (b)Shaikh, N. S.; Junge, K.; Beller, M. Org. Lett. 2007, 9, 5429. (c) Langlotz, B. K.; Wadepohl, H.; Gade, L. H. Angew. Chem. Int. Ed. 2008, 47, 4670. (d) Tondreau, A. M.; Lobkovsky, E.; Chirik, P. J. Org. Lett. 2008, 10, 2789. (e) Furuta, A.; Nishiyama, H. Tetrahedron Lett. 2008, 49, 110. (f) Nishiyama, H.; Furuta, A. Chem. Commun. 2007, 760. (g) Bhattacharya, P., Krause, J. A.; Guan, H. Organometallics 2011, 30, 4720. 9 (a) Mimoun, H.; Laumer, J. Y. D. S.; Giannini, L.; Scopelliti, R.; Floriani, C. J. Am. Chem. Soc. 1999, 121, 6158. (b) Gerard, S.; Pressel, Y.; Riant, O. Tetrahedron: Asymmetry 2005, 16, 1889. (c) Bette, V.; Mortreux, A.; Savoia, D.; Carpentier, J.-F. Adv. Synth. Catal. 2005, 347, 289. 10 Mankad, N. P.; Laitar, D. S.; Sadighi, J. P. Organometallics 2004, 23, 3369.

Although there are precedents for Ni-catalyzed hydrosilylation reactions involving substrates with less polarized C=C or C≡C bonds (Scheme 2),11,12 analogous nickel catalysts for the hydrosilylation of C=O bonds are exceedingly scarce in the literature.13,14 This surprising dearth of nickel systems for catalytic carbonyl reduction has prompted us to investigate hydrosilylation chemistry using well-defined nickel pincer hydride complexes.

11 (a) Bennett, E. W.; Orenski, P. J. J. Organomet. Chem. 1971, 28, 137. (b) Kiso, Y.; Kumada, M.; Maeda, K.; Sumitani, K.; Tamao, K. J. Organomet. Chem. 1973, 50, 311. (c) Kiso, Y.; Tamao, K.; Kumada, M. J. Organomet. Chem. 1974, 76, 95. (d) Kumada, M.; Kiso, Y.; Umeno, M. J. Chem. Soc. D 1970, 611. (e) Lappert, M. F.; Nile, T. A.; Takahashi, S. J. Organomet. Chem. 1974, 72, 425. (f) Yamamoto, K.; Hayashi, T.; Uramoto, Y.; Ito, R.; Kumada, M. J. Organomet. Chem. 1976, 118, 331. (g) Yamamoto, K.; Uramoto, Y.; Kumada, M. J. Organomet. Chem. 1971, 31, C9. (h) Hyder, I.; Jimenez-Tenorio, M.; Puerta, M. C.; Valerga, P. Dalton Trans. 2007, 3000. 12 (a) Chaulagain, M. R.; Mahandru, G. M.; Montgomery, J. Tetrahedron 2006, 62, 7560. (b) Tamao, K.; Miyake, N.; Kiso, Y.; Kumada, M. J. Am. Chem. Soc. 1975, 97, 5603. (c) Tamao, K.; Kobayashi, K.; Ito, Y. J. Am. Chem. Soc. 1989, 111, 6478. (d) Lappert, M. F.; Nile, T. A. J. Organomet. Chem. 1975, 102, 543. (e) Boudjouk, P.; Choi, S.-B.; Hauck, B. J.; Rajkumar, A. B. Tetrahedron Lett. 1998, 39, 3951. 13 Díez-González, S.; Nolan, S. P. Org. Prep. Proced. Int. 2007, 39, 523. 14 (a) Frainnet, E.; Martel-Siegfried, V.; Brousse, E.; Dedier, J. J. Organomet. Chem. 1975, 85, 297. (b) Frainnet, E.; Bourhis, R.; Simonin, F.; Moulines, F. J. Organomet. Chem. 1976, 105, 17. (c) Lee, S. J.; Kim, T. Y.; Park, M. K.; Han, B. H. Bull. Korean Chem. Soc. 1996, 17, 1082. (d) Fontaine, F.-G.; Nguyen, R.-V.; Zargarian, D. Can. J. Chem. 2003, 81, 1299. (e) Irrgang, T.; Schareina, T.; Kempe, R. J. Mol. Catal. A: Chem. 2006, 257, 48. (f) Kong, Y. K.; Kim, J.; Choi, S.; Choi, S.-B. Tetrahedron Lett. 2007, 48, 2033. (g) Tran, B. L.; Pink, M.; Mindiola, D. J. Organometallics 2009, 28, 2234.

2.2 Synthesis of Nickel Pincer Hydride Complexes

To prepare the desired nickel hydride complexes, a three-step synthetic route was designed as outlined in scheme 3. First, bis(phosphinite) pincer ligands were synthesized from a cheap starting material resorcinol and corresponding chlorophosphines in the presence of a base

(eq 1). These ligands are air-sensitive materials. Ligands 1a,15 1b16 and 1d17 have been previously reported, and 1c was prepared in a similar fashion.

Nickel pincer chlorides 2a16 and 2d17 were prepared according to literature protocols, through cyclometallation of 1a and 1d with anhydrous NiCl2 (eq 2). Nickel chlorides 2b and 2c were synthesized similarly and characterized by 1H NMR, 31P{1H} NMR, and elemental analysis. Complex 2b was also characterized by X-ray crystallography (Figure 1).

Alternatively, nickel pincer chloride complexes could be prepared by refluxing resorcinol,

15 Göttker-Schnetmann, I.; White, P.; Brookhart, M. J. Am. Chem. Soc. 2004, 126, 1804. 16 Pandarus, V.; Zargarian, D. Organometallics 2007, 26, 4321. 17 Gómez-Benitez, V.; Baldovino-Panteleón, O.; Herrera-Á Ivarez, C.; Toscano, R. A.; Morales-Morales, D. Tetrahedron Lett. 2006, 47, 5059.

20 chlorophosphines, NEt3, and NiCl2 in a single reaction vessel in THF (eq 3). This alternative route is more convenient for the preparation of 2a-d and avoids the separation of air-sensitive ligands.18

t Figure 1. X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]NiCl (2b) (50% probability level).

Selected bond lengths (Å) and angles (deg): NiCl 2.2046(7), NiC1 1.887(2), NiP1

18 Chakraborty, S.; Patel, Y. J.; Krause, J. A.; Guan, H. Polyhedron 2012, 32, 30.

2.1912(7), NiP2 2.1868(7), P1NiP2 164.18(3), C1NiCl 179.43(7), P1NiC1 82.20(7),

P2NiC1 82.00(7).

Reaction of 2a-c with excess of LiAlH4 at room temperature generated the desired nickel pincer hydride complexes 3a-c as orange/yellow solids in good isolated yields (eq 4). The 1H

NMR spectra of 3a-c in C6D6 revealed characteristic hydride resonances as triplets at  7.89

(JH-P = 55.2 Hz),  7.96 (JH-P = 53.2 Hz),  7.88 (JH-P = 54.4 Hz) respectively. The hydride ligand in 3b was also located by X-ray diffraction of its single crystal (Figure 2). Attempted synthesis of 3d (R = Ph) from the reduction of 2d with LiAlH4, NaBH4, or LiEt3BH gave unidentified products. A possible reason that 3d did not form may come from the fact that the hydride moiety in 3d does not have enough steric protection by the phenyl groups as compared to the more bulky alkyl groups in 3a-c. More exposed hydride moiety is possibly more susceptible to decomposition reactions with other reagents or impurities in the reaction mixture.

Moreover, previous group members have shown that exposed pincer backbone (P-O bonds) of

2d is available for the attack by bases and other nucleophiles before desired reaction takes place at the metal center.19

19 Zhang, J.; Medley, C. M.; Krause, J. A.; Guan, H. Organometallics 2010, 29, 6393.

t Figure 2. X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]NiH (3b) (50% probability level).

Selected bond lengths (Å) and angles (deg): NiH 1.37(3), NiC1 1.892(3), NiP1 or NiP1A

2.1160(5), P1NiP1A 166.26(3), C1NiH 180.000(9), P1NiC1 or P1ANiC1 83.132(17).

Alternatively, complexes 3a-c were generated by the treatment of 2a-c with Na[NMe2-

BH3] (eq 5). Only stoichiometric quantity of Na[NMe2-BH3] is needed to generate the nickel hydrides and the reaction time is shortened. The yields are comparable to the method stated earlier. More importantly, this method is safer when the hazardous waste disposal generated from the LiAlH4 route is taken into account. When 2d was treated with Na[NMe2-BH3], a deep red precipitate immediately precipitated out of the toluene solution. However, this precipitate turned black when solvent was removed under vacuum. 1H NMR of the black residue indicated no sign of nickel hydride species.

2.3 Stoichiometric Reduction of Aldehydes and Ketones with Nickel Hydride Complexes

After the synthesis of nickel hydrides, I set out to investigate their reactivities toward aldehydes and ketones. When one equivalent of benzaldehyde was mixed with 3a, rapid insertion reaction took place at room temperature. 1H NMR spectroscopy showed the formation of nickel benzyloxide 4a; the intensity of a new resonance at  4.88 (singlet) increased as the intensities of both hydride 3a ( 7.90) and PhCHO ( 9.62) decreased (Scheme 4). The insertion reaction was complete within 30 min and provided 4a in > 95% NMR yield. To support the proposed structure, complex 4a was independently generated from metathesis reaction between 2a and PhCH2ONa (Scheme 4), although isolation of compound 4a in an analytically pure form was unsuccessful after repeated trials. Nevertheless, the stoichiometric reaction shown here has set a precedent of insertion of an organic carbonyl group into a nickel hydrogen bond.20

20 For examples of C=O insertion in other M-H bonds, see: (a) Weinert, C. S.; Fanwick, P. E.; Rothwell, I. P. Organometallics 2005, 24, 5759. (b) van der Zeijden, A. A. H.; Berke, H. Helv. Chim. Acta 1992, 75, 513. (c) Furno, F.; Fox, T.; Schmalle, H. W.; Berke, H. Organometallics 2000, 19, 3620. (d) Liang, F.; Jacobsen, H.; Schmalle, H. W.; Fox, T.; Berke, H. Organometallics 2000, 19, 1950. (e) Liang, F.; Schmalle, H. W.; Fox, T.; Berke, H. Organometallics 2003, 22, 3382. (f) Zhao, Y.; Schmalle, H. W.; Fox, T.; Blacque, O.; Berke, H. Dalton Trans. 2006, 73. (g) Cugny, J.; Schmalle, H. W.; Fox, T.; Blacque, O.; Alfonso, M.; Berke, H. Eur. J. Inorg. Chem. 2006, 540. (h) Du, G.; Fanwick, P. E.; Abu-Omar, M. M. J. Am. Chem. Soc. 2007, 129, 5180. (i) Baratta, W.; Ballico, M.; Esposito, G.; Rigo, P. Chem. Eur. J. 2008, 14, 5588. (j) van Leeuwen, P. W. N. M.; Roobeek, C. F.; Orpen, A. G. Organometallics 1990, 9, 2179.

The reverse step of aldehyde insertion, namely -hydride elimination of a metal alkoxide, is more commonly observed for late transition metals and in fact often used to synthesize metal hydride complexes.21 To probe the reversibility of the aldehyde insertion step, 4a was treated with p-tolualdehyde and in a separate experiment 5a was treated with benzaldehyde. The formation of the second nickel alkoxide complex in either reaction would indicate that the aldehyde insertion step is reversible. However, only the starting nickel complex was observed even after a long period of time, implying that the insertion is irreversible (Scheme 5).

Surprisingly, when complex 3b was treated with one equivalent of benzaldehyde at room temperature, no insertion product was observed by 1H and 31P{1H} NMR. Heating the reaction mixture to 60oC did not result in any new species (eq 6). This behavior is attributed to the steric hindrance between the two reactants in the transition state. It is reasonable to assume that the tBu groups in 3b pose a much greater steric congestion near the hydride moiety than the iPr and

21 Bryndza, H. E.; Tam, W. Chem. Rev. 1988, 88, 1163.

25 cPe groups in 3a and 3c respectively. Another explanation would be that β-hydride elimination of the resulting primary alkoxide complex 4b becomes much favorable when more bulky tBu groups are present. In order to probe the possibility of favorable β-hydride elimination, 3b-D was mixed with PhCHO in protio-benzene and the reaction was monitored by both 2H and

31P{1H}NMR. No deuterium exchange was observed either at room temperature or elevated temperature after 48 h (eq 7). This result strongly suggests no interaction between 3b and

PhCHO.

As anticipated, ketones were less reactive than PhCHO as the resulting secondary alkoxides are more likely to undergo -hydride elimination than primary alkoxides.20 When a solution of 3a in C6D6 was treated with an equimolar amount of PhCOCH3 at room temperature for 24 h, only 17 mol% of the hydride was converted to the insertion product (Scheme 6).

Reaction with PhCOPh under the similar condition did not yield any alkoxide species. The slow reaction with ketones might also be a result of increased steric congestion in the transition state.

To complete a potential catalytic cycle for the hydrosilylation of aldehydes, complex 4a was reacted with various silanes to regenerate hydride 3a (eq 8). Both PhSiH3 and Ph2SiH2 were identified as suitable silyl reagents for the nickel benzyloxide; complete regeneration of 3a and release of the silyl ether product were seen within a few minutes. Other silanes such as triethoxysilane and poly(methylhydrosiloxane) (PMHS) regenerated hydride 3a after relatively longer reaction times (> 1 h). Et3SiH showed no reaction with complex 4a possibly because of steric reasons (Table 1).

Table 1. Stoichiometric Reactivity of 4a toward Different Silanes

2.4 Catalytic Hydrosilylation of Aldehydes and Ketones with Nickel Hydride Complexes

After establishing the protocols of aldehyde insertion and hydride regeneration, I set out to investigate the catalytic activity of hydrides 3a-c for the hydrosilylation of benzaldehyde

(Scheme 7, Table 2). With PhSiH3 or Ph2SiH2 as the silyl reagent, hydride 3a catalyzed the hydrosilylation reaction efficiently; the reaction was complete within two hours at room temperature with catalyst loading as low as 0.2 mol%. Consistent with the stoichiometric experiments, catalytic reactions with triethoxysilane and PMHS were much slower than those with PhSiH3 and Ph2SiH2. A control experiment in the absence of the nickel hydride 3a showed no significant hydrosilylation even at 60C for 5 days, indicating that the reaction is indeed catalyzed by nickel. Based on the steric argument, I anticipated that 3c would be a more active catalyst because cyclopentyl groups are smaller than isopropyl groups and the insertion of

PhCHO into 3c should be more favorable. However, complex 3c was found to be less effective than 3a under similar reaction conditions. It is likely that complex 3c undergoes partial degradation or side reactions (vide infra). Interestingly, hydride 3b also slowly catalyzed the

28 hydrosilylation reaction, despite the fact that the stoichiometric reaction of 3b and PhCHO did not yield nickel alkoxide species.

Table 2. Comparing the Catalytic Activities of 3a-c in Hydrosilylation Reaction

The scope of this catalytic system was studied using 0.2 mol% of 3a in the presence of a slight excess of PhSiH3, and the reduction products were isolated as alcohols following basic hydrolysis of the silyl ethers (Scheme 8, Table 3). As shown in Table 3, the hydrosilylation reaction was tolerant to many functional groups including OMe (entry 2), NMe2 (entry 3), Cl

(entry 4), NO2 (entry 5), and CN (entry 10). It is difficult to rationalize the electronic effect of substituents on the relative hydrosilylation rates since both the aldehyde with an electron- donating group (entry 2) and the one with an electron-withdrawing group (entry 4) are less reactive than the unsubstituted benzaldehyde (entry 1). An explanation of these results will require more detailed kinetic studies on individual steps of the potential catalytic cycle for each substrate. For ,-unsaturated aldehydes (entry 9 and 12), only the 1,2-addition products were obtained. These results are in contrast to other Ni-catalyzed22 or Cu-catalyzed23 hydrosilylation

22 Kim, S. O.; Rhee, S.; Lee, S. H. Bull. Korean Chem. Soc. 1999, 20, 773. 23 Jurkauskas, V.; Sadighi, J. P.; Buchwald, S. L. Org. Lett. 2003, 5, 2417.

29 of ,-unsaturated carbonyl compounds where only the 1,4-addition products were isolated.

The high chemoselectivity for carbonyl group over C=C was also observed for aldehydes with isolated C=C bonds (entry 13). In addition to substituted benzaldehydes, other aromatic (entries

6-8) and aliphatic (entries 11-13) aldehydes were viable substrates for hydrosilylation. In each reaction studied, no side products such as enoxysilanes and disiloxanes were observed, making our nickel system superior to other nickel systems reported in the literature.14a,b An aldehyde bearing a cyano group on the ortho position (entry 10) gave a mixture of two products: the expected 2-cyanobenzyl alcohol 6 and a lactone 7. Compound 7 resulted from cyclization of 6 under the hydrolysis conditions. Ketones were much less reactive catalytically. Only partial hydrosilylation was observed for acetophenone (18%), cyclohexanone (60%), and benzophenone

(6%), even at an elevated temperature (70C, 24 h) using a higher catalyst loading (1 mol%).

Table 3. Hydrosilylation of Aldehydes Catalyzed by Nickel Pincer Hydride

Based on the stoichiometric studies, a catalytic cycle was proposed for the hydrosilylation reactions (Scheme 9). The first step of this cycle involves the insertion of carbonyl group into the nickel hydride, followed by the cleavage of Ni-O bond by the silane

(alternatively, this could happen by an oxidative addition-reductive elimination sequence). A similar mechanism has also been proposed by Nolan in his Cu(I)/carbene-catalyzed hydrosilylation reactions.7e,f In the case of our hydrosilylation system, the carbonyl insertion step might proceed via a direct H- transfer pathway. In other words, no precoordination is required as the nickel center is coordinatively saturated and hydride moiety is positioned opposite to the good trans-influencing phenyl backbone.

Alternatively, nickel hydrides might activate the silanes prior to their interaction with the carbonyl groups. This would resemble the well-known Chalk-Harrod mechanism for alkene hydrosilylation.24 This type of mechanism also has been proposed for Rh-catalyzed hydrosilylation of carbonyl substrates.25 In addition, oxidative addition of silanes or Si-H –

24 Chalk, A. J.; Harrod, J. F. J. Am. Chem. Soc. 1965, 87, 16. 25 Ojima, I.; Kogure, T.; Kumagai, M.; Horiuchi, S.; Sato, T. J. Organomet. Chem. 1976, 122, 83.

32 bond coordination to a nickel center has been reported.26 To probe the possibility of this silane- activation mechanism, 3a was mixed with phenylsilane. A new triplet resonance was observed in

1 3 the H NMR spectrum at 4.62 ppm ( JP-H = 9.9 Hz) and it grew as the resonances corresponding to 3a and PhSiH3 (SiH resonance appear at 4.22 ppm) disappeared. A new singlet phosphorous resonance at 200 ppm was also observed in the 31P{1H}NMR. This new species (8) is assigned to be the nickel-silyl complex27 (Scheme 10).

In order to understand the reactivity of 8 and its possible involvement in the catalytic hydrosilylation reaction, this complex was treated with one equivalent of benzaldehyde. No reaction was evidenced either at room temperature or at elevated temperature. This result suggests that compound 8 is not a catalytically relevant species. The interaction of silane with nickel hydride was also established by performing a reaction of 3a-D with phenylsilane where deuterium was rapidly exchanged from nickel to silane as observed by 2H NMR (eq 9). Even though these nickel silyl species is not involved in the catalysis, their formation during the catalysis might decrease the concentration of the active nickel hydride complexes. Complex 3c

26 Steinke, T.; Gemel, C.; Cokoja, M.; Winter, M.; Fischer, R. A. Angew. Chem. Int. Ed. 2004, 43, 2299. 27 (a) Iluc, V. M.; Hillhouse, G. L. Tetrahedron 2006, 62, 7577. (b) Shimada, S.; Rao, M. L. N.; Tanaka, M. Organometallics 1999, 18, 292. (c) Avent, A. G.; Gehrhus, B.; Hitchcock, P. B.; Lappert, M. F.; Maciejewski, H. J. Organomet. Chem. 2003, 686, 292. (d) Shimada, S.; Rao, M. L. N.; Hayashi, T.; Tanaka, M. Angew. Chem., Int. Ed. 2001, 40, 213. (e) Bierchenk, T. R.; Guerra, M. A.; Juhlke, T. J.; Larson, S. B.; Lagow, R. J. J. Am. Chem. Soc. 1987, 109, 4855. (f) Nlate, S.; Herdtweek, E.; Fischer, R. A. Angew. Chem. Int. Ed. 1996, 35, 1861. (g) Maciejewski, H.; Marciniec, B.; Kownacki, I. J. Organomet. Chem. 2000, 597, 175. (h) Gehrhus, B.; Hitchcock, P. B.; Lappert, M. F.; Maciejewski, H. Organometallics 1998, 17, 5599. (i) Kang, Y.; Lee, J.; Kong, Y. K.; Kang, S. O.; Ko, J. Chem. Commun. 1998, 2343. (j) Schmedake, T. A.; Haaf, M.; Paradise, B. J.; Powell, D.; West, R. Organometallics 2000, 19, 3263. (k) Adhikari, D.; Pink, M.; Mindiola, D. J. Organometallics 2009, 28, 2072.

33 containing small cPe groups possibly undergo this side reaction more favorably than 3a which has relatively more bulky iPr groups. This might explain the lower catalytic activity of 3c compared to 3a.

Although 3a reacts with both aldehyde and phenylsilane, aldehyde insertion was found to be much faster than silane activation. To fully understand the mechanism, detailed kinetic studies of the individual steps (carbonyl insertion, regeneration, and silane activation) needs to be carried out. More mechanistic insights could be obtained by investigating the electronic influence on the turnover-limiting step of the catalytic cycle. Preliminary kinetic studies performed on the benzaldehyde insertion into 3a under pseudo first-order reaction conditions revealed that the reaction finishes within 30 min at -15oC with a half-life of approximately seven minutes (Scheme 11). The kinetic trace was obtained by monitoring the intensity of NiH resonance (triplet, δ -7.90 ppm) with respect to the hexamethylbenzene resonance (internal standard, singlet, δ 2.12 ppm). Even at -15oC, the reaction is too fast to allow for reliable kinetic measurements. However, this problem can possibly be solved in the future either by performing this reaction at even lower temperature or by using a stopped-flow apparatus.

2.5 Conclusions

I have synthesized nickel POCOP-pincer hydride complexes as hydrosilylation catalysts for aldehydes and ketones and have tested the hypothesis that strong trans influence of the pincer ligand would facilitate hydride transfer from nickel to carbon. Facile insertion of aldedydes into nickel hydrides has been observed. Silanes such as PhSiH3 interact with the nickel alkoxide species to reform the nickel hydrides. A series of aldehydes can be selectively reduced to alcohols catalyzed by the nickel hydrides. This catalytic hydrosilylation process is highly chemoselective for C=O over C=C and no silicon-based side products are produced. Partial hydrosilylation has been observed for ketones. Complex containing bulky tBu groups on the phosphorous atoms has proved be inferior catalyst compared to catalysts with smaller groups such as iPr and cPe groups. Lower activity of cPe- containing nickel hydride can be rationalized either by its propensity to undergo partial degradation or by a more favorable reaction with silane owing to its smaller size.

2.6 Experimental Section

General Comments. All the air-sensitive compounds were prepared and handled under an argon atmosphere using standard Schlenk and inert-atmosphere box techniques. All aldehyde and ketone substrates were purchased from commercial sources and were used without further purification. Dry and oxygen-free solvents were collected from an

Innovative Technology Solvent Purification System and used throughout all the experiments.

1 Toluene-d8 and C6D6 were distilled from Na and benzophenone under argon. Both H NMR and 31P{1H} NMR spectra were recorded on a Bruker Avance-400MHz NMR spectrometer.

2H NMR spectra were recorded on a Bruker Avance-500MHz NMR spectrometer. Chemical

35 shift values in 1H NMR spectra were referenced internally to the residual solvent resonances

31 1 (δ 7.26 for CDCl3, δ 7.15 for C6D6, and δ 2.09 for toluene-d8). The P { H} NMR spectra were referenced to an external 85% H3PO4 sample (δ 0). Column chromatography was performed with silica gel and solvents of commercial grade. All isolated alcohol products were known compounds and characterized by 1H NMR spectroscopy. The NMR data obtained for all alcohol products were consistent with the literature values.28 1,3- i 16 t 15 17 ( Pr2PO)2C6H4 (1a), 1,3-( Bu2PO)2C6H4 (1b), 1,3-(Ph2PO)2C6H4 (1d), [2,6- i 16 17 ( Pr2PO)2C6H3]NiCl (2a), and [2,6-(Ph2PO)2C6H3]NiCl (2d) were prepared as described in the literature.

Synthesis of Na[H3B-NMe2]. Under an argon atmosphere, a THF solution of H3B-

NHMe2 (100 mg, 1.7 mmol) was slowly added to a suspension of NaH (52 mg, 2.04 mmol) in THF at 0oC. The resulting solution was refluxed for 2 h and after that evaporation of the

1 solvent yielded the product as a white powder (101 mg, 74%). H NMR (400 MHz, THF-d8,

1 δ): 1.29 (bq, Na[H3B-NMe2], JB-H = 85.4 Hz, 3H), 2.08 (s, Na[H3B-NMe2], 6H).

c Synthesis of 2,6-( Pe2PO)2C6H4 (1c). Under an argon atmosphere

resorcinol (1.1g, 10 mmol) and NaH (504 mg, 21 mmol) were refluxed for 2 h in 20 mL of THF. To this mixture, a 20 mL THF solution of chlorodicyclopentyl phosphine

(4 mL, 21 mmol) were slowly added at 0oC. The resulting mixture was stirred at room temperature for 24 h and then the solvent was removed in vacuo to obtain a white residue. 50 mL of pentane was added to this white residue and the colorless liquid was separated by cannula filtration. Evaporation of this liquid yielded the product as a colorless oil (3.5g,

28 Shaikh, N. S.; Junge, K.; Beller, M. Org. Lett. 2007, 9, 5429.

1 79%). H NMR (400 MHz, C6D6, ): 1.41 (m, CH2, 8H), 1.54 (m, CH2, 8H), 1.60 (m, CH2,

3 8H), 1.83 (m, CH2, 8H), 2.04 (m, PCH, 4H), 7.00 (d, ArH, JH-H = 7.2 Hz, 2H), 7.06 (t, ArH,

3 31 1 JH-H = 7.6 Hz, 1H), 7.46 (s, ArH, 1H). P{ H} NMR (162 MHz, C6D6):  139.23 (s).

t Synthesis of [2,6-( Bu2PO)2C6H3]NiCl (2b). Under an argon

atmosphere 50 mL of toluene was added to a mixture of 1b (1.20 g, 3.0 mmol) and anhydrous NiCl2 (389 mg, 3.0 mmol), giving an orange suspension. While the solution was refluxed for 18 h, a brown precipitate formed, which was removed by filtration after the mixture was cooled to room temperature. The volume of the orange filtrate was reduced to 5 mL, and then Et2O was added to cause precipitation. The product was collected by filtration, washed with

1 Et2O, and dried under vacuum to give an orange/green powder 2b (725 mg, 50% yield). H

NMR (400 MHz, CDCl3, ): 1.49 (virtual triplet, PC(CH3)3, JP-H = 6.8 Hz, 36H), 6.38 (d, ArH,

31 1 JH-H = 8.0 Hz, 2H), 6.92 (t, ArH, JH-H = 8.0 Hz, 1H). P{ H} NMR (162 MHz, CDCl3, ):

187.20 (s). Anal. Calcd for C22H39ClO2P2Ni: C, 53.75; H, 8.00; Cl, 7.21. Found: C, 54.11; H,

8.09; Cl, 7.25.

c Synthesis of [2,6-( Pe2PO)2C6H3]NiCl (2c). Under an argon

atmosphere 50 mL of toluene was added to a mixture of 1c (4.46 g, 10 mmol) and anhydrous NiCl2 (1.56 g, 12 mmol), giving an orange suspension. While the solution was refluxed for 18 h a brown precipitate formed, which was removed by filtration after the mixture was cooled to room temperature. The volume of the orange filtrate was reduced to 5 mL, and then Et2O was added to cause precipitation. The product was collected by filtration, washed with

1 Et2O, and dried under vacuum to give an orange/yellow powder 2c (4.7 g, 83% yield). H NMR

(400 MHz, CDCl3, ): 1.38-1.43 (m, CH2, 8H), 1.64-1.67 (m, CH2, 8H), 1.79-1.82 (m, CH2, 8H),

3 1.93-1.97 (m, CH2, 8H), 2.41-2.49 (m, PCH, 4H), 6.77 (d, ArH, JH-H = 7.6 Hz, 2H), 7.03 (t,

3 31 1 ArH, JH-H = 7.6 Hz, 1H). P{ H} NMR (162 MHz, CDCl3, ): 175.35 (s). Anal. Calcd for

C26H39ClO2P2Ni: C, 57.86; H, 7.28; Cl, 6.57. Found: C, 57.67; H, 7.25; Cl, 6.46.

General Procedure for the One-Pot Synthesis of 2a-d. Under an argon atmosphere, 60 mL of THF was added to a Schlenk flask containing resorcinol (1.10 g, 10 mmol), NEt3 (2.81 mL, 20 mmol), anhydrous NiCl2 (1.30 g, 10 mmol), and chlorophosphines (20 mmol). The resulting mixture was refluxed for 12-24 h to give a deep brown solution along with a purple/green precipitate, which was filtered through a short column packed with silica. The filtrate was concentrated by reducing the volume under vacuum to 5 mL, followed by the addition of anhydrous diethyl ether (ca. 10 mL) until a precipitate formed. The solid was collected by gravity filtration and dried under vacuum to give an orange/yellow powder.

i Synthesis of [2,6-( Pr2PO)2C6H3]NiH (3a). Method A: Under an argon

atmosphere the suspension of LiAlH4 (872 mg, 23 mmol) and 2a (500 mg, 1.15 mmol) in 60 mL of toluene was stirred at room temperature for 24 h. The resulting mixture was filtered through a short plug of Celite to give a yellow solution. After the solvent was evaporated under vacuum, the desired hydride 3a was isolated as an orange/yellow crystalline solid (405 mg,

88% yield). Method B: To a solution of 2a (100 mg, 0.23 mmol) in 10 mL of toluene was added

o Na[NMe2-BH3] (18.6 mg, 0.23 mmol) under an argon atmosphere at -78 C and the resulting orange-yellow solution was gradually warmed to room temperature. The solution was stirred for three hours at room temperature. The resulting mixture was passed through a dry pad of celite

38 and filtrate was concentrated in vacuo to obtain the desired product (360 mg, 78%). 1H NMR

(400 MHz, toluene-d8, ): 7.90 (t, NiH, JP-H = 55.2 Hz, 1H), 1.08-1.17 (m, PCH(CH3)2, 24H),

2.05-2.11 (m, PCH(CH3)2, 4H), 6.74 (d, ArH, JH-H = 8.0 Hz, 2H), 6.97 (t, ArH, JH-H = 8.0 Hz,

31 1 1H). P{ H} NMR (162 MHz, toluene-d8, ): 206.56 (s). Anal. Calcd for C18H32O2P2Ni: C,

53.90; H, 8.04. Found: C, 53.88; H, 8.17.

t Synthesis of [2,6-( Bu2PO)2C6H3]NiH (3b). This complex was

prepared in 80% (Method A) and 91% (Method B) isolated yield by procedures

1 similar to that were used for 3a. H NMR (400 MHz, C6D6, ): 7.96 (t, NiH, JP-H = 53.2 Hz,

1H), 1.30 (virtual triplet, PC(CH3)3, JP-H = 6.8 Hz, 36H), 6.85 (d, ArH, JH-H = 7.6 Hz, 2H), 7.02 (t,

31 1 ArH, JH-H = 7.6 Hz, 1H). P{ H} NMR (162 MHz, C6D6, ): 219.35 (s). Anal. Calcd for

C22H40O2P2Ni: C, 57.80; H, 8.82. Found: C, 57.65; H, 8.85.

c Synthesis of [2,6-( Pe2PO)2C6H3]NiH (3c). This complex was

prepared in 71% (Method A) and 77% (Method B) isolated yield by procedures

1 2 similar to that were used for 3a. H NMR (400 MHz, C6D6, ): 7.88 (t, NiH, JP-H = 54.4 Hz,

1H), 1.36-1.40 (m, CH2, 8H), 1.60-1.63 (m, CH2, 8H), 1.77-1.79 (m, CH2, 8H), 1.91-1.98 (m,

3 3 CH2, 8H), 2.31-2.39 (m, PCH, 4H), 6.87 (d, ArH, JH-H = 7.6 Hz, 2H), 7.05 (t, ArH, JH-H = 7.6

31 1 Hz, 1H). P{ H} NMR (162 MHz, C6D6, ): 199.66 (s). Anal. Calcd for C26H40O2P2Ni: C,

61.81; H, 7.98. Found: C, 61.12; H, 8.10.

i Synthesis of [2,6-( Pr2PO)2C6H3]NiOCH2Ph (4a). Method A: To

a solution of 3a (40 mg, 0.10 mmol) in 5 mL of toluene was added degassed benzaldehyde (10 µL, 0.10 mmol) under an argon atmosphere and the resulting mixture was stirred at room temperature for 1 h. Evaporating the solvent under vacuum yielded orange

1 oil and the NMR spectra of the crude product were recorded. H NMR (400 MHz, toluene-d8,

): 1.18-1.29 (m, PCH(CH3)2, 12H), 1.32-1.41 (m, PCH(CH3)2, 12H), 2.09-2.15 (m,

PCH(CH3)2, 4H), 4.88 (s, OCH2Ph, 2H), 6.50 (d, ArH, JH-H = 8.0 Hz, 2H), 6.84 (t, ArH, JH-H =

31 1 8.0 Hz, 1H), 7.00-7.56 (m, CH2Ph, 5H). P{ H} NMR (162 MHz, toluene-d8, ): 176.59 (s).

Attempts to further purify 4a via recystallization led to decomposition of the product. Method

B: Under an argon atmosphere, the suspension of 2a (87 mg, 0.2 mmol) and sodium benzyloxide

(39 mg, 0.3 mmol) in 10 mL of toluene was stirred at room temperature for 24 h. The resulting mixture was passed through a short plug of Celite to remove NaCl and the filtrate was concentrated under vacuum. Again, an orange oil was obtained and its NMR spectra matched the data above, although further purification of 4a led to decomposition of the product.

Stoichiometric Reaction of 3a with Phenylsilane. In a J. Young NMR tube, 3a (10.0 mg, 25 μmol) and phenylsilane (3.1 µL, 25 μmol) were mixed in ca. 0.6 mL of C6D6. Formation

1 31 1 1 of 8 was monitored by H and P{ H}NMR spectroscopy. H NMR (400 MHz, C6D6, ): 1.06-

1.15 (m, PCH(CH3)2, 24H), 2.12-2.19 (m, PCH(CH3)2, 4H), 4.62 (t, NiSiH2Ph, JP-H = 9.9 Hz,

31 1 2H), 6.68 (d, ArH, JH-H = 8.0 Hz, 2H), 6.92 (t, ArH, JH-H = 8.0 Hz, 1H). P{ H} NMR (162

MHz, toluene-d8, ): 200.18 (s).

Stoichiometric Reaction of 8 with PhCHO. In a screw-cap NMR tube, 3a (10.0 mg, 25

μmol) and phenylsilane (3.1 µL, 25 μmol) were mixed in ca. 0.6 mL of C6D6. After complete

40 generation of 8, PhCHO (2.5 µL, 25 μmol) was added to the mixture via a 10 µL-syringe and the reaction was monitored by 1H and 31P{1H} NMR spectroscopy.

General Procedures for Hydrosilylation. To a flame-dried Schlenk flask was added a solution of nickel 3a (8.0 mg, 20 µmol) in toluene (6 ml) and an aldehyde substrate (10 mmol) under an argon atmosphere. The resulting mixture was stirred at room temperature for 5-10 minutes, after which PhSiH3 (1.48 ml, 12 mmol) was added via a gas tight syringe. The reaction was stirred at room temperature or at a higher temperature until there was no aldehyde left

(monitored by withdrawing aliquot and analyzing its 1H NMR spectrum). The reaction was then quenched by 10% aqueous solution of NaOH (about 10 mL) with vigorous stirring for more than

12 h. The solution containing the alcohol product was extracted with diethyl ether three times, dried over anhydrous Na2SO4, and concentrated under vacuum. The desired alcohol was further purified by flash column chromatography.

X-Ray Diffraction Studies. Single crystals of 2b were obtained from a saturated

o solution of the complex in CH2Cl2 at 0 C. X-ray quality crystals of 3b were grown by layering

CH3OH on a saturated THF solution of the hydride at 35C and slowly allowing it diffuse.

Crystal data and refinement parameters for 2b and 3b are given in table 4. A block-like yellow crystal of 2b, approximate dimensions 0.12 x 0.11 x 0.10 mm, was cut from large rectangular plates. A pale yellow plate-like crystal of 3b with approximate dimensions 0.20 x 0.17 x 0.05 mm, was selected. For X-ray examination and data collection, the crystals were mounted in a cryo-loop with paratone-N and transferred immediately to the goniostat bathed in a cold stream.

Intensity data for both complexes were collected at 150K on a standard Bruker

SMART6000CCD diffractometer using graphite-monochromated Cu Kα radiation, λ=1.54178Å.

A series of 5-s data frames measured at 0.3o increments of ω were collected to calculate a unit

41 cell. For data collection frames were measured for a duration of 5-s for 2b and 3-s for 3b at 0.3o intervals of ω with a maximum 2θ value of ~135o. The data frames were processed using the program SAINT. The data were corrected for decay, Lorentz and polarization effects as well as absorption and beam corrections based on the multi-scan technique. The structures were solved by a combination of direct methods in SHELXTL and the difference Fourier technique and refined by full-matrix least squares on F2. Non-hydrogen atoms were refined with anisotropic displacement parameters with the exception of the minor component of 3b. The t-butyl groups show typical disorder; a disorder model is presented for C9 and C10 of 2b and C10 and C12 of

3b. The hydride atom in 3b was located directly from the difference map and the position refined. The remaining H-atoms were either located or calculated and subsequently treated with a riding model. The H-atom isotropic displacement parameters were defined as a*Ueq (a=1.5 for methyl and 1.2 for all others).

Table 4. Crystal Data and Refinement Parameters for 2b and 3b

2b 3b

empirical formula C22H39O2P2ClNi C22H40O2P2Ni crystal system triclinic Monoclinic space group P-1 P2/n

a, Å 8.3879(2) 11.0751(2) b, Å 11.7974(2) 9.2096(2) c, Å 13.3800(3) 12.9521(2) , deg 99.589(1) 90 , deg 96.029(1) 111.550(1) , deg 104.354(1) 90 Volume, Å3 1249.84(5) 1228.73(4) Z 2 2 no. of data collected 10165 10080

no. of unique data, Rint 4245, 0.0247 2207, 0.0240 R1, wR2 (I > 2(I)) 0.0374, 0.0958 0.0360, 0.0923 R1, wR2 (all data) 0.0476, 0.1024 0.0385, 0.0943

Chapter 3

Catalytic Reduction of CO2 to Methanol with Simple Boranes

3.1 Introduction

The global demand for energy is on the rise as a result of population and economic growth. Currently, the majority of energy usage involves the combustion of nonrenewable fossil

1,2 fuels, which leads to enormous increase in CO2 emission to the atmosphere. One of the approaches to reduce the amount of CO2 is the conversion of CO2 back to fuels or value-added

3 materials. Catalytic reduction of CO2 is therefore considered as a potential solution to the CO2 problem. Transition metal-catalyzed homogeneous hydrogenation of CO2 often leads to either formic acid that is thermodynamically stabilized by a base,4 or CO via the reverse process of the

5 water-gas shift reaction. Transforming CO2 into methane, the most reduced form of carbon, under homogeneous conditions can be accomplished using silanes as the reducing reagents.6

Reducing CO2 to methanol would be even more desirable for the advantages of transporting a liquid fuel rather than a gas. The product methanol is both a potential gasoline replacement7 and a precursor for the synthesis of various important chemicals such as ethylene and propylene.8

1 Annual Energy Review 2009; U.S. Energy Information Administration: Washington, DC, 2010; Tables 12.2-12.4, pp 349-353. 2 Lewis, N. S.; Nocera, D. G. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 15729. 3 (a) Marks, T. J.; et al. Chem. Rev. 2001, 101, 953. (b) Louie, J. Curr. Org. Chem. 2005, 9, 605. (c) Sakakura, T.; Choi, J.-C.; Yasuda, H. Chem. Rev. 2007, 107, 2365. (d) Darensbourg, D. J. Chem. Rev. 2007, 107, 2388. (e) Aresta, M.; Dibenedetto, A. Dalton Trans. 2007, 2975. 4 (a) Jessop, P. G.; Ikariya, T.; Noyori, R. Chem. Rev. 1995, 95, 259. (b) Leitner, W. Angew. Chem. Int. Ed. 1995, 34, 2207. (c) Jessop, P. G.; Joó, F.; Tai, C.-C. Coord. Chem. Rev. 2004, 248, 2425. 5 (a) Ford, P. C.; Trabuco, E.; Mdleleni, M. M. Water Gas Shift Reactions Homogeneous. In Encyclopedia of Catalysis, 1st ed; Horváth, I. T., Ed.; Wiley Interscience: Hoboken, NJ, 2003; Vol. 6, pp 651-658. (b) Esswein, A. J.; Nocera, D. G. Chem. Rev. 2007, 107, 4022. 6 Matsuo, T.; Kawaguchi, H. J. Am. Chem. Soc. 2006, 128, 12362. 7 Olah, G. A.; Goeppert, A.; Prakash, G. K. S. Beyond Oil and Gas: The Methanol Economy; Wiley-VCH: Weinheim, Germany, 2006. 8 (a) Dai, W.; Wang, X.; Wu, G.; Guan, N.; Hunger, M.; Li, L. ACS Catal. 2011, 1, 292. (b) Cui, Z.-M.; Liu, Q.; Song, W.-G.; Wan, L.-J. Angew. Chem. Int. Ed. 2006, 45, 6512. 44

Previously both homogeneous and heterogeneous catalysts have been employed to reduce

CO2. Hydrogenation of CO2 to methanol has been performed using Cu-based heterogeneous catalysts.9,10 However, these systems suffer from high operating temperatures (200-250oC).11 In addition, tuning the reactivities of heterogeneous catalysts are challenging because of limited mechanistic details.11 For these reasons, recent work has aimed to develop homogeneous catalysts for the conversion of CO2 to methanol under milder conditions.

12 One effective strategy is to use silanes to convert CO2 to methanol derivatives.

Catalytic hydrosilylation of CO2 to methoxysilyl species is feasible with Ir(CN)(CO)(dppe)

(dppe = 1,2-bis(diphenylphosphino)ethane) at 40oC, albeit with limited turnover numbers.12a

More efficient hydrosilylation reactions are catalyzed by N-heterocyclic carbenes (metal-free) at ambient temperature with turnover frequencies (TOF) as high as 25.5 h-1 (based on SiH), and methanol is produced from the hydrolysis of the initial reduction products (Scheme 1).12b

9 (a) Ushikoshi, K.; Mori, K.; Watanabe, T.; Takeuchi, M.; Saito, M. Stud. Surf. Sci. Catal. 1998, 114, 357. (b) Saito, M. Catal. Surv. Jpn. 1998, 175. (c) Grabow, L. C.; Mavrikakis, M. ACS Catal. 2011, 1, 365. 10 (a) Tominaga, K.; Sasaki, Y.; Kawai, M.; Watanabe, T.; Saito, M. J. Chem. Soc., Chem. Commun. 1993, 629. (b) Tominaga, K.-I.; Sasaki, Y.; Saito, M.; Hagihara, K.; Watanabe, T. J. Mol. Catal. 1994, 89, 51. 11 Joo, F. Activation of Carbon Dioxide. In Physical Inorganic Chemistry: Reactions, Processes, and Applications; Wiley-VCH: Weinheim, Germany, 2007; pp 259. 12 (a) Eisenschmid, T. C.; Eisenberg, R. Organometallics 1989, 8, 1822-1824. (b) Riduan, S. N.; Zhang, Y.; Ying, J. Y. Angew. Chem. Int. Ed. 2009, 48, 3322-3325. (c) Huang, F.; Lu, G.; Zhao, L.; Li, H.; Wang, Z.-X. J. Am. Chem. Soc. 2010, 132, 12388. 45

Recent development in frustrated Lewis acid-base pairs (FLPs) chemistry13 has led to

14 alternative strategies for the reduction of CO2 to the methoxide level, provided with either H2

15 or H3NBH3 as a hydrogen source. The driving force of these reactions is the formation of strong CH3OB and CH3OAl bonds for the FLP systems involving 2,2,6,6-

14 15 tetramethylpiperidine/B(C6F5)3 and PMes3/AlX3 (Mes = 2,4,6-C6H2Me3; X = Cl or Br), respectively. However, stoichiometric amounts of FLPs are required as the cleavage of these highly stable chemical bonds by cheaper hydrogen source such as H2 is difficult.

Very recently, The Sanford group has shown catalytic hydrogenation of CO2 to methanol through “cascade catalysis” (Scheme 2).16 Although this result is very promising, precious metal- based catalysts have to be used and in addition there is still much room for improvement in terms of turnover number.

As shown in the previous chapter, hydrosilylation of aldehydes and ketones can be accomplished using nickel hydride catalysts, I then decided to seek the possibility of catalytic reduction of CO2, a more challenging substrate, using analogous nickel hydride complexes.

Although the long-term objective of this project is to use H2 as the reducing agent, my immediate focus has been placed on studying CO2 reduction with silanes and boranes.

13 Stephan, D. W. Dalton Trans. 2009, 3129. 14 Ashley, A. E.; Thompson, A. L.; O’Hare, D. Angew. Chem. Int. Ed. 2009, 48, 9839–9843. 15 Ménard, G.; Stephan, D. W. J. Am. Chem. Soc. 2010, 132, 1796. 16 Huff, C. A.; Sanford, M. S. J. Am. Chem. Soc. 2011, 133, 18122. 46

3.2 CO2 Insertion into Ni-H Bonds of Nickel Pincer Hydride Complexes

Insertion of CO2 into a metal-hydrogen bond constitutes a critical step in many transition

4 metal-catalyzed reduction of CO2. Despite the high thermodynamic stability of CO2, we found

17a that the reaction of POCOP-pincer nickel hydrides 3a-c with CO2 proceeded rapidly at room temperature to give nickel formate complexes 9a-c as sole products (eq 1).17b Related trans-

18 t 19 (Cy3P)2Ni(H)(Ph) and [2,6-( Bu2PCH2)2C6H3]PdH show similar reaction patterns with CO2.

t In contrast, the interaction between CO2 and [( Bu2PCH2SiMe2)2N]NiH, a PNP-pincer nickel hydride, gives rise to a hydridonickel cyanate complex through an N/O transposition.20

17 (a) Chakraborty, S.; Krause, J. A.; Guan, H. Organometallics 2009, 28, 582. (b) Chakraborty, S.; Zhang, J.; Krause, J. A.; Guan, H. J. Am. Chem. Soc. 2010, 132, 8872. 18 Darensbourg, D. J.; Darensbourg, M. Y.; Goh, L. Y.; Ludvig, M.; Wiegreffe, P. J. Am. Chem. Soc. 1987, 109, 7539. 19 Johansson, R.; Wendt, O. F. Organometallics 2007, 26, 2426. 20 Laird, M. F.; Pink, M.; Tsvetkov, N. P.; Fan, H.; Caulton, K. G. Dalton Trans. 2009, 1283. 47

Complexes 9a-c were characterized by NMR and IR spectroscopy, elemental analysis, and X-ray crystallography. Of particular interest is the orientation of the formato groups in their crystal structures (Figures 1-3). Both 9a and 9c show a relatively small dihedral angle between the least-squares plane defined by P1, P2, C1, Ni atoms and the OC(O)H plane [11.3(5) in 9a;

16.8(5) and 6.6(7) in 9c]. To accommodate the "in plane" conformation, the iPr and cPe groups are steered away from the OC(O)H moiety to avoid the steric clash (Figures 1 and 3). In contrast, 9b contains a "perpendicular" formato group in the solid-state structure with a much larger dihedral angle [73.8(3)] between the P1-P2-C1-Ni plane and the OC(O)H plane (Figure

2). However, in the solution the rotation of the NiO3 bond is expected to be rapid. Restricted bond rotation favoring the "in plane" conformer of 9 would render the two phosphorus nuclei inequivalent. I found that within the temperature range of 23 C to 50 C, the 31P {1H} NMR spectra of 9a and 9c in toluene-d8 showed singlet resonances with no significant broadening. X- ray analysis of the single crystals of 9b (Figure 2) revealed a distorted square planar geometry for Ni with the O3 atom displaced 0.41 Å out of the least squares plane that is defined by P1, C1,

Ni, and P2 atoms; however, the long through-space distance NiO4 (3.271(3) Å) would argue

48 against a potential 2-coordination mode for the formato group. The bond lengths of C23O3

(1.250(4) Å) and C23O4 (1.236(4) Å) are fairly close due to a significant degree of electron density delocalization within the formate moiety,21 but slightly shorter than the CO bond

22 lengths (1.27 Å) in HCO2Na.

i Figure 1. X-ray Crystal Structure of [2,6-( Pr2PO)2C6H3]NiOC(O)H (9a) (50% probability level). Hydrogen atoms (except the formate hydrogen) are omitted for clarity. Selected bond lengths (Å) and angles (deg): NiC1, 1.883(2); NiP1, 2.1641(7); NiP2, 2.1927(7); NiO3,

1.9135(17); C23O3,1.279(3); C23O4, 1.222(3); P1NiP2, 164.31(3); O3C23O4, 127.5(3).

21 Grove, D. M.; van Koten, G.; Ubbels, H. J. C.; Zoet, R.; Spek, A. L. Organometallics 1984, 3, 1003. 22 Zachariasen, W. H. J. Am. Chem. Soc. 1940, 62, 1011. 49

t Figure 2. X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]NiOC(O)H (9b) (50% probability level). Hydrogen atoms (except the formate hydrogen) are omitted for clarity. Selected bond lengths (Å) and angles (deg): NiC1, 1.886(2); NiP1, 2.2040(7); NiP2, 2.1817(8); NiO3,

1.920(2); C23O3,1.250(4); C23O4, 1.236(4); P1NiP2, 164.02(3); O3C23O4, 128.0(3).

Through-space distance (Å): NiO4, 3.271(3).

c Figure 3. X-ray Crystal Structure of [2,6-( Pe2PO)2C6H3]NiOC(O)H (9c) (50% probability level). Hydrogen atoms (except the formate hydrogen) are omitted for clarity. Two independent molecules in the crystal lattice were found. Selected bond lengths (Å) and angles (deg):

Ni1C1A, 1.873(2); Ni1P1A, 2.1637(11); Ni1P2A, 2.1953(11); NiO3A, 1.910(3); 50

C27AO3A, 1.280(4); C27AO4A, 1.246(4); P1ANi1P2A, 162.84(3); O3AC27AO4A,

127.9(3).

The CO2 insertion in this system was found to be reversible. This was supported by the following two experimental observations: (i) mixing 13C-labeled nickel formate (9b-13C) with ~

1 atm of CO2 at room temperature resulted in the formation of non-labeled 9b (eq 2), and (ii) heating the solid sample of 9b under vacuum at 60oC for 5 h generated 3b in a nearly quantitative yield (eq 3).

3.3 Regeneration of Nickel Hydrides from Nickel Formate Complexes

Reforming 3a-c from 9a-c would close a potential catalytic cycle for the reduction of

CO2. To test this hypothesis, the reactions between 9b and various reducing reagents were investigated at room temperature. My initial attempt to cleave the Ni-O bonds in nickel formates with silanes proved to be unsuccessful. When one equivalent of silane was mixed with one

51 equivalent of 9b, no net reaction was observed at room temperature (Scheme 3). However, treatment of 9b with stoichiometric amount of catecholborane (HBcat) resulted in the rapid regeneration of 3b along with a new boron species which is assigned to catBOC(O)H (Scheme

3). Interestingly, when HBcat was employed in a large excess, a new resonance appeared at 3.34 ppm (singlet) in the 1H NMR spectrum of the reaction, consistent with the formation of a methanol derivative. To further understand this process, a series of NMR experiments involving

13 13 13 CO2 were carried out. When 9b- C (generated in situ from 3b and CO2) was treated with 4

1 1 equivalents of HBcat, the corresponding H NMR spectrum exhibited a doublet ( JC-H = 145.2

Hz) at 3.34 ppm (Figure 4). When this sample solution was spiked with regular methanol, the methyl resonance of non-labeled methanol derivative appeared right at the center of the doublet.

The same reaction monitored by 13C NMR spectroscopy revealed a singlet at 53.6 ppm in the 1H-

1 1 decoupled spectrum (Figure 5) and a quartet ( JC-H = 145.2 Hz) in the H-coupled spectrum, confirming the presence of a methoxy group. Thus, the reduction product was identified as

13 23 CH3OBcat, the spectra of which are consistent with those reported in the literature.

23 Povie, G.; Villa, G.; Ford, L.; Pozzi, D.; Schiesser, C. H.; Renaud, P. Chem. Commun. 2010, 46, 803.

13CH OBcat   3

  (ii)

(iii)

H13COONi(tBuPCP) (i)

9.4 9.2 9.0 8.8 8.6 8.4 8.2 8.0 4.0 3.8 3.6 3.4 3.2 3.0 2.8 ppm

1 13 13 1 Figure 4. (i) H NMR spectra of 9b- C (with residual CO2) in C6D6 (bottom) (ii) H NMR

13 13 1 spectra 9b- C (with residual CO2) with added HBcat (4 equiv) in C6D6 (iii) H NMR sample from (ii) spiked with CH3OH (only the methanol peak is shown)

13 tBu 13 H COONi( PCP)/catBOBcat CH3OBcat H13COOBcat 13CO (ii) 2 (iii) H13COONi(tBuPCP)

13 CO2 (i)

13 1 13 13 13 Figure 5. (i) C{ H} NMR spectra of 9b- C (with residual CO2) in C6D6 (ii) 9b- C and

13 1 13 CO2 (residual) with added HBcat (4 equiv) in C6D6 (iii) H coupled C NMR of sample (ii)

(only the methanol resonance is shown)

To further support the structure of the methanol derivative, it was synthesized independently from catBCl and NaOCH3 (eq 4). The product of this reaction, B-methoxy catecholborane (catBOCH3), matches well with the reduced product of CO2. The by-product of the CO2 reduction process was catBOBcat, which resulted in a downfield shift of the resonances

13 13 of 9b- C (Figure 4 and 5), presumably through a Lewis acid-base interaction. CH3OBcat

13 proved to be hygroscopic; it absorbed moisture quickly from the air and generated CH3OH, which was confirmed by mass spectroscopy.

3.4 Catalytic Reduction of CO2 with Catecholborane

It was also discovered that the reduction of CO2 with HBcat was catalytic in nickel. The mixture of HBcat and 3b in a ratio of 500 : 1 under 1 atm of CO2 generated CH3OBcat with 495 turnovers (based on BH) in 1 h, while the other boron product (catBOBcat) precipitated from the C6D6 solution (Scheme 4). A control experiment showed no reaction in the absence of 3b, confirming that the reduction was indeed catalyzed by the nickel hydride. The initial reduction products were hydrolyzed in large excess of water and vacuum distillation of the resulting mixture yielded 0.28 M methanol in C6D6 (or 61% yield based on HBcat). When HBcat was depleted in the solution, the nickel species rested as 9bcatBOBcat, and further addition of

HBcat reinitiated the reaction with no loss of catalytic activity. In fact, the independently synthesized 9b exhibited comparable catalytic efficiency as 3b, which is an advantage in our catalytic system as compound 9b is air and moisture stable.

Proposed catalytic cycles consistent with our experimental observations are shown in

Scheme 5. The catalytic reduction of CO2 begins with a reversible insertion of CO2 into a NiH bond. The subsequent cleavage of NiO bond with HBcat regenerates 3b and releases

H(O)COBcat, which is reduced to formaldehyde by another HBcat. Arguably, the reaction of boryl formate (catBOC(O)H) and HBcat could be catalyzed by nickel. However, I could not experimentally verify this possibility because of the fast nature of this catalytic reduction process

55 and the challenge of isolating the boryl formate complex. Computational study performed on this system provided additional mechanistic insights.24 Density functional theory (DFT) studies suggested that formation of formaldehyde from boryl formate is a nickel-catalyzed multistep process (Scheme 6). The uncatalyzed reaction has a kinetic barrier of 40 kcal/mol which is too high to be operative at room temperature. On the other hand, involvement of nickel hydride lowers the barrier to 27.6 kcal/mol.

The final reduced product, catBOCH3, is formed in the next catalytic cycle which is analogous to the nickel-catalyzed hydrosilylation of aldehydes (discussed in Chapter 2).17a To test the mechanistic hypothesis that formaldehyde is involved in this catalytic process, a reaction of paraformaldehyde and 3b was performed in the presence of HBcat. The formation of catBOCH3 as the end product was confirmed (Scheme 7).

24 Huang, F.; Zhang, C.; Jiang, J.; Wang, Z.-X.; Guan, H. Inorg. Chem. 2011, 50, 3816. 56

3.5 Comparative Studies of CO2 Reduction with Different Nickel Catalysts and Boranes

DFT calculations of our catalytic system24 have implied that the steric environment around the nickel center is critical to the efficiency of the catalysis. The choice of reductant is also crucial. The computed free-energy profile of the entire process initially suggested to us that the reaction of nickel hydride with H(O)COBcat (Scheme 6) would cross the highest kinetic barrier. It was therefore anticipated that replacing the bulky tBu groups in the nickel pincer complex with smaller alkyl groups such as iPr and cPe would accelerate the turnover-limiting step, leading to a more efficient catalysis. To validate these computational results, I focused my attention to employ nickel catalysts with different size of alkyl groups on the phosphorus donor atoms, and to investigating the scope of reducing reagents.

To understand how the substituents on the phosphorous donor atoms influence the reactivity of 9a-c, the NMR-scale reactions of these complexes with three equivalents of HBcat were investigated (eq 5). Under the same conditions (same concentration and temperature), the reaction of 9b was the fastest with quantitative formation of CH3OBcat within 10 minutes.

Counterintuitive to the notion that a less sterically hindered nickel center would be more susceptible to the attack of HBcat and the subsequent insertion reactions, the reaction of 9a took

45 min to complete. Interestingly, the reaction of 9c finished within 10 minutes; however, the

NMR yield of CH3OBcat was merely 74%, implying a partial decomposition of the nickel complex 9c (or 3c, which is the intermediate during the reaction).25

To expand the scope of reducing reagents that can be used for the reduction of CO2, I focused on the stoichiometric reaction of 9b with other boranes and silanes under similar reaction conditions (Scheme 8). [3.3.1]-9-borabicyclononane (9-BBN) behaves similarly to

HBcat, reducing the nickel formate complex to 3b and a B-methoxy-9-BBN species within 1 h.

In contrast, even with a large excess of pinacolborane (HBpin), only one equivalent of borane was consumed to yield 3b and presumably H(O)COBpin. While this type of boryl formate species is too unstable to be isolated, spectroscopic evidence to support its formation includes 1H

NMR resonances at 8.07 ppm (singlet) and 0.88 ppm (singlet) with an expected integration ratio of 1 : 12. It is worth mentioning again that PhSiH3 does not regenerate 3b when mixed with 9b at room temperature. According to the computational studies, the kinetic reaction barrier with

PhSiH3 has found to be 42.6 kcal/mol.

25 Chakraborty, S.; Patel, Y. J.; Krause, J. A.; Guan, H. Polyhedron 2012, 32, 30. 58

In a comparative study, catalytic activity of 3a-c in the reduction of CO2 was then carried out using different boranes (Table 1). In a typical catalytic study, borane and 3a-c (with a ratio of

100 : 1) were mixed in C6D6 and then stirred under 1 atm of CO2 (eq 6). Hexamethylbenzene was used as an NMR internal standard and the progress of the reaction was monitored by 1H

NMR spectroscopy. When 3b was used as the catalyst, the reaction took only 45 min and 1 hour to reach the maximum TON of 100 (based on BH, Table 1, entry 2 and 5) for catecholborane and 9-BBN respectively. The less sterically bulky nickel hydrides 3a and 3c proved to be less effective catalysts. The same reaction catalyzed by 3a required 2 h for HBcat (entry 1) and 1.5 h for 9-BBN (entry 4) to complete. The catalytic reaction with 3c with HBcat did not go to completion within 12 h, giving a TON of 30 (entry 3). Interestingly, when 9-BBN was used for this reaction, the reaction was finished within 4 hour with the maximum TON (entry 6). Use of

BH3•THF as a reducing source did not result in any reduction product (entry 7-9). I also attempted to carry out catalytic hydrosilylation of CO2 using PhSiH3 as the reducing reagent and

3b as the catalyst (entry 10). No reduced product was obtained at room temperature even after

48 h. This result is consistent with the stoichiometric reaction shown in Scheme 3 as well as the

DFT studies.

Table 1. Catalytic Activity of Nickel Hydride Complexes in the Reduction of CO2

bTurnover numbers (TON) were calculated based on B-H bonds, cHexamethyldisilane was used as an internal standard because the resonance of hexamethylbenzene overlaps with those of 3c

Although the DFT studies predicted that complex 3c ought to be the best catalyst among the three nickel hydrides, experimental results showed that it is the least effective catalyst. This discrepancy between experimental outcomes and computational prediction may be reconciled by invoking off-the-catalytic-loop adducts generated from nickel hydrides and boranes (highlighted in red in Scheme 9 using HBcat as the representative borane). Nickel hydrides with smaller substituents may interact with boranes more favorably, resulting in a lower concentration of the catalytically active species. If this is the case, the structures of these adducts, the extent to which they form, and the rates at which they release the nickel hydrides back to the catalytic cycles are all important to better understand the catalytic system.

3.6 Reactions of Nickel Hydride Complexes with Boranes

The literature has suggested many possible products for the reaction of a transition metal hydride complex with a borane (Figure 6). If the metal center possesses a vacant coordination site, activation of borane B–H bond leads to the formation of a -borane complex (structure

A),26,27 or a boryl dihydride complex (structure B, endo or exo isomer)28 as the oxidative addition product. Abstraction of H– by a Lewis acidic boron center gives rise to a metal complex bearing either a bidentate (structure C)29 or a monodentate hydridoborate ligand (structure D).30

Evolution of H2 with concomitant formation of a metal-boryl species (structure E) is another possibility.31 For the reactions involving HBcat, degradation of the borane by metal hydrides to

– 32 [B(cat)2] has been previously observed.

H H BR'2 BR'2 H H LnM LnM LnM BR'2 LnM BR'2 LnM BR'2 H H H H A B C D E

Figure 6. Possible Products from the Reaction of a Metal Hydride Complex with a Borane.

26 (a) Lachaize, S.; Essalah, K.; Montiel-Palma, V.; Vendier, L.; Chaudret, B.; Barthelat, J.-C.; Sabo-Etienne, S. Organometallics 2005, 24, 2935. (b) Denney, M. C.; Pons, V.; Hebden, T. J.; Heinekey, D. M.; Goldberg, K. I. J. Am. Chem. Soc. 2006, 128, 12048. (c) Hebden, T. J.; Denney, M. C.; Pons, V.; Piccoli, P. M. B.; Koetzle, T. F.; Schultz, A. J.; Kaminsky, W.; Goldberg, K. I.; Heinekey, D. M. J. Am. Chem. Soc. 2008, 130, 10812. 27 (a) Alcaraz, G.; Sabo-Etienne, S. Coord. Chem. Rev. 2008, 252, 2395. (b) Lin, Z. Struc. Bond. 2008, 130, 123. 28 (a) Baker, R. T.; Ovenall, D. W.; Calabrese, J. C.; Westcott, S. A.; Taylor, N. J.; Williams, I. D.; Marder, T. B. J. Am. Chem. Soc. 1990, 112, 9399. (b) Hartwig, J. F.; De Gala, S. R. J. Am. Chem. Soc. 1994, 116, 3661. (c) Lantero, D. R.; Motry, D. H.; Ward, D. L.; Smith, M. R., III. J. Am. Chem. Soc. 1994, 116, 10811. (d) Lantero, D. R.; Miller, S. L.; Cho, J.-Y.; Ward, D. L.; Smith, M. R., III. Organometallics 1999, 18, 235. (e) Ohki, Y.; Hatanaka, T.; Tatsumi, K. J. Am. Chem. Soc. 2008, 130, 17174. 29 Lantero, D. R.; Ward, D. L.; Smith, M. R., III. J. Am. Chem. Soc. 1997, 119, 9699. (b) Hascall, T.; Bridgewater, B. M.; Parkin, G. Polyhedron 2000, 19, 1063. (c) Antiñolo, A.; Carrillo-Hermosilla, F.; Fernández-Baeza, J.; García-Yuste, S.; Otero, A.; Rodríguez, A. M.; Sánchez-Prada, J.; Villaseñor, E.; Gelabert, R.; Moreno, M.; Lluch, J. M.; Lledós, A. Organometallics 2000, 19, 3654. (d) Liu, X.-Y.; Bouherour, S.; Jacobsen, H.; Schmalle, H. W.; Berke, H. Inorg. Chim. Acta 2002, 330, 250. (e) Essalah, K.; Barthelat, J.-C.; Montiel, V.; Lachaize, S.; Donnadieu, B.; Chaudret, B.; Sabo-Etienne, S. J. Organomet. Chem. 2003, 680, 182. (f) Evans, W. J.; Lorenz, S. E.; Ziller, J. W. Chem. Commun. 2007, 4662. (g) Bontemps, S.; Vendier, L.; Sabo-Etienne, S. Angew. Chem. Int. Ed. 2012, 51, 1671. 30 Rossin, A.; Peruzzini, M.; Zanobini, F. Dalton Trans. 2011, 40, 4447. 31 (a) Burgess, K.; van der Donk, W. A.; Westcott, S. A.; Marder, T. B.; Baker, R. T.; Calabrese, J. C. J. Am. Chem. Soc. 1992, 114, 9350. (b) Adhikari, D.; Huffman, J. C.; Mindiola, D. J. Chem. Commun. 2007, 4489. (c) Zhu, Y.; Chen, C.-H.; Fafard, C. M.; Foxman, B. M.; Ozerov, O. V. Inorg. Chem. 2011, 50, 7980. 32 Westcott, S. A.; Blom, H. P.; Marder, T. B.; Baker, R. T.; Calabrese, J. C. Inorg. Chem. 1993, 32, 2175. (b) Westcott, S. A.; Marder, T. B.; Baker, R. T.; Harlow, R. L.; Calabrese, J. C.; Lam, K. C.; Lin, Z. Polyhedron 2004, 32, 2665. (c) Knizek, J.; Nöth, H. Eur. J. Inorg. Chem. 2011, 1888.

Prior to my work, there have been two reports pertaining to the reaction of a pincer- ligated nickel hydride complex and a borane. To demonstrate that a Ni–BR'2 moiety is capable of transferring boron to organic molecules, the Mindiola group has prepared a PNP-pincer nickel boryl complex from the corresponding nickel hydride and HBcat (eq 7).31b In a comparative study of the hydride affinity of BH3 and BF3, Peruzzini and co-workers have recently investigated the interaction between a PCP-pincer nickel hydride and BH3•THF, which yields a nickel borohydride complex as suggested by the NMR spectroscopy (eq 8).30 On the basis of

1 2 DFT analyses, they have proposed a ground-state η -BH4 structure, along with the η -BH4 structure being the transition state that is responsible for the exchange between the bridging and terminal hydrogens. However, isolation of this nickel borohydride complex in the solid-state form has not been accomplished, mainly due to the facile dissociation of BH3 at temperatures that are convenient for work-up.30

To understand the nature and role of the interaction between nickel hydrides and boranes in our catalytic process, I decided to investigate the structure-reactivity relationship between these species. By varying the alkyl substituents on the phosphorus donor atoms of the pincer ligand, I wish to address how the size of the substituents impacts the reactivity of the nickel hydride complexes toward different boranes, as well as its relation with the efficiency of the nickel hydrides in catalyzing the reduction of CO2 with boranes. For this study, boranes such as

BH3•THF, 9-BBN, and HBcat were chosen to react with different nickel hydrides.

3.6.1 Reactions with BH3•THF

The (CO) stretching frequencies of (tBuPOCOP)Ir(CO)33 [1949 cm-1] and

(tBuPCP)Ir(CO)34 [1928 cm-1] seem to support the notion that the POCOP-pincer ligand makes the metal center more electron deficient.35 However, as suggested by Krogh-Jespersen and

Goldman, the difference in (CO) values may reflect electrostatic effects rather than the extent of

Ir to CO π-back-donation.36 They have also pointed out that comparing the electronic difference between analogous PCP- and POCOP-pincer complexes is not simple because the influence of electron-withdrawing oxygen atoms on the phosphorus donors may be offset by the increased

OC(aryl) -donation.37 Nevertheless, electrochemistry studies of (iPrPOCOP)NiBr and

(iPrPCP)NiBr carried out by Zargarian and coworkers have also suggested that the POCOP-pincer

33 Göttker-Schnetmann, I.; White, P. S.; Brookhart, M. Organometallics 2004, 23, 1766. 34 Krogh-Jespersen, K.; Czerw, M.; Zhu, K.; Singh, B.; Kanzelberger, M.; Darji, N.; Achord, P. D.; Renkema, K. B.; Goldman, A. S. J. Am. Chem. Soc. 2002, 124, 10797. 35 Kuklin, S. A.; Sheloumov, A. M.; Dolgushin, F. M.; Ezernitskaya, M. G.; Peregudov, A. S.; Petrovskii, P. V.; Koridze, A. A. Organometallics 2006, 25, 5466. 36 Goldman, A. S.; Krogh-Jespersen, K. J. Am. Chem. Soc. 1996, 118, 12159. 37 Zhu, K.; Achord, P. D.; Zhang, X.; Krogh-Jespersen, K.; Goldman, A. S. J. Am. Chem. Soc. 2004, 126, 13044. 64 complex is more electron deficient.38 From the steric point of view, the O-linkages in a

(RPOCOP)M system often impose a more obtuse P–M–P angle than the related (RPCP)M system, and consequently the metal center supported by a POCOP-pincer ligand is sterically more accessible.39 This is certainly true for nickel hydride complexes; the P–Ni–P angle of 3b17a

[166.26(3)] has been found to be smaller than that of (tBuPCP)NiH [173.54(3)].40 I therefore suspected that the more electrophilic and more exposed nickel center of 3b would interact with

tBu BH3•THF more strongly than ( PCP)NiH, allowing us to isolate the product and fully characterize it.

Guided by this hypothesis, a toluene solution of 3b was treated with a slight excess of

BH3•THF at 0C, after which an immediate color change from yellow to orange was noticed.

Removal of the volatiles under vacuum did not appear to reverse the reaction either, evidenced by the persistent orange color throughout the evaporation. The 1H NMR spectrum of the isolated material in C6D6 displays a broad quartet at –0.45 ppm with an intensity ratio of approximately

1 : 1 : 1 : 1, indicative of B–H coupling. The resonance integrates as four hydrogens with respect to one pincer unit. The boron resonance of the compound appears at –35.9 ppm as a quintet in 11B NMR, and as a singlet in 11B{1H} NMR. The NMR data suggest that the adduct of

3b and BH3•THF is a typical borohydride complex. Related complexes 10a and 10c were synthesized in good yield from hydrides 3a and 3c, respectively (eq 9).

38 (a) Pandarus, V.; Zargarian, D. Chem. Commun. 2007, 978. (b) Castonguay, A.; Spasyuk, D. M.; Madern, N.; Beauchamp, A. L.; Zargarian, D. Organometallics 2009, 28, 2134. 39 Choi, J.; MacArthur, A. H. R.; Brookhart, M.; Goldman, A. S. Chem. Rev. 2011, 111, 1761. 40 Boro, B. J.; Duesler, E. N.; Goldberg, K. I.; Kemp, R. A. Inorg. Chem. 2009, 48, 5081. 65

To further support the structural assignment, 10a-c were synthesized through an

– – alternative route involving the substitution of Cl by BH4 from nickel chloride complexes 2a-c

(eq 10). The substitution reaction of 2a or 2c at room temperature is completed within 12 h; however, for the sterically more crowded compound 2b, a prolonged reaction time of three days is needed. In any case, the NMR spectra of the product are identical to those obtained from the hydride route. The success of both synthetic methods suggests that the POCOP-pincer nickel borohydride complexes are thermally stable. At room temperature, they are remarkably stable; both solid and solution samples of these complexes can be handled in air without noticeable

tBu decomposition. In strong contrast to this result, the reaction of ( PCP)NiCl with NaBH4 have

tBu tBu been previously shown to generate ( PCP)NiH as the isolable product, with ( PCP)Ni(BH4) being the transient intermediate that is only observed in solution by NMR spectroscopy.30

– The coordination mode of BH4 was established by IR spectroscopy. Attenuated total reflectance (ATR) IR spectra of the solid samples of 10a-c show two strong bands within the

-1 range of 2370-2420 cm , which are attributed to terminal hydrogen-boron stretch [(B-Ht)]. The bridging boron-hydrogen stretching bands [(B-Hb)] are very broad but located in the region

1790-2070 cm-1. Compounds 10a and 10c also have a strong band at 1144 cm-1 and 1122 cm-1, respectively, which are assigned as the BH2 deformation mode. However, such a band is absent in 10b, or more likely overlapped with the bands associated with the pincer framework. The IR spectra of deuterium-labeled 10a-c (synthesized from 2a-c and NaBD4) exhibit the predicted terminal deuterium-boron stretching bands from the known (B-H)/(B-D) ratio of around

1.35.41 All these data, as summarized in Table 2, are consistent with transition metal complexes

2 – 42,43 bearing a bidentate or  -BH4 ligand. Transmission-IR spectra of 10a-c in CH2Cl2 reproduced the major IR features with bands being slightly shifted from those observed for the solid samples, suggesting that the 2-coordination mode is also the ground-state structure in solution. Nevertheless, fluxionality of terminal and bridging hydrogens observed in the 1H NMR spectra suggests that these hydrogens are rapidly exchanged on the NMR time scale, possibly

1 3 through an intermediate involving an  - or  -BH4.

41 Corey, E. J.; Cooper, N. J.; Canning, W. M.; Lipscomb, W. N.; Koetzle, T. F. Inorg. Chem. 1982, 21, 192. 42 Marks, T. J.; Kolb, J. R. Chem. Rev. 1977, 77, 263. 43 Besora, M.; Lledós, A. Struct. Bond. 2008, 130, 149. 67

Table 2. Selected Infrared Frequencies (cm-1) of 10a-c (in Solid State)

Abbreviations for the intensities: s, strong; m, medium; w, weak; sh, shoulder; br, broad.

Among the known nickel borohydride complexes, very few have been characterized by X-

44-48 – 44 ray crystallography. The BH4 ligand in both trans-NiH(BH4)(PCy3)2 [Cy = cyclohexyl]

45 2 and (triphos)Ni(BH4) [triphos = MeC(CH2PPh2)3] coordinates to nickel in an  fashion,

46 whereas the same ligand in Tp*Ni(BH4) [Tp* = hydrotris(3,5-dimethylpyrazolyl)borate] adopts

3 47 an  coordination mode. A recent study of Ni(cyclam)(BH4)2 [cyclam = 1,4,8,11-

1 tetraazacyclotetradecane] has shown two isomeric structures: one contains two trans  -BH4

2 – ligands, while the other isomer is ionic with an  -BH4 ligand and a BH4 counterion. For a

– 48 dinuclear Ni(II) complex, BH4 bridges two nickel centers as a 1,3 ligand. In our case, X-ray

2 structure determination of 10a-c reveals a  -BH4 ligand (Figures 7-9), which is consistent with

44 Saito, T.; Nakajima, M.; Kobayashi, A.; Sasaki, Y. J. Chem. Soc., Dalton Trans. 1978, 482. 45 Desrochers, P. J.; LeLievre, S.; Johnson, S. R.; Lamb, T. B.; Phelps, A. L.; Cramer, S. P. Inorg. Chem. 2003, 42, 7945. 46 Kandiah, M.; McGrady, G. S.; Decken, A.; Sirsch, P. Inorg. Chem. 2005, 44, 8650. 47 Churchard, A. J.; Cyranski, M. K.; Dobrzycki, L.; Budzianowski, A.; Grochala, W. Energy Environ. Sci. 2010, 3, 1973. 48 Journaux, Y.; Lozan, V.; Klingele, J.; Kersting, B. Chem. Commun. 2006, 83. 68 the IR studies described above. The B–H lengths and H–B–H angles agree with a tetrahedral geometry for the boron, when relatively large errors in locating the positions of hydrogen atoms by X-ray diffraction are considered. Compared to the reported Ni–H length of 1.37(3) Å for

3b17a, the Ni–H lengths in 10b [1.78(3) Å and 1.85(3) Å] are substantially elongated. The Ni…B

2 44 distances in 10a-c [2.180(3)-2.214(3) Å] are similar to those in trans-NiH( -BH4)(PCy3)2

2 45 2 + – 47 [2.201(8) Å], (triphos)Ni( -BH4) [2.24 Å] and [Ni(cyclam)( -BH4)] [BH4] [2.202(6) Å],

3 46 but significantly longer than those in Tp*Ni( -BH4) [2.048(5) Å] as well as Mindiola’s PNP- pincer nickel boryl complex [1.9091(18) Å].31b

i 2 Figure 7. X-ray Crystal Structure of [2,6-( Pr2PO)2C6H3]Ni( -BH4) (10a) (50% probability level). Hydrogen atoms (except those attached to boron) are omitted for clarity. (Two independent molecules were found in the crystalline lattice; only one is shown here). Selected bond lengths (Å) and bond angles (deg): Ni1H 1.77(4) and 1.87(4), B1H (bridging) 1.19(5) and 1.10(5), B1H (terminal) 1.16(5) and 1.11(5), Ni1C1A 1.898(4), Ni1P1A 2.1514(13),

Ni1P2A 2.1490(13), Ni1…B1 2.189(3), P1ANi1P2A 163.08(5), C1ANi1P1A 81.28(13),

C1ANi1P2A 81.98(13), HNi1H 63(2), HB1H 117(3).

t 2 Figure 8. X-ray Crystal Structure of [2,6-( Bu2PO)2C6H3]Ni( -BH4) (10b) (50% probability level). Hydrogen atoms (except those attached to boron) are omitted for clarity. Selected bond lengths (Å) and angles (deg): NiH 1.85(3) and 1.78(3), B1H (bridging) 1.22(3) and 1.17(3),

B1H (terminal) 1.04(3) and 1.06(3), NiC1 1.901(2), NiP1 2.2046(7), NiP2 2.2025(7),

Ni…B1 2.214(3), P1NiP2 163.03(3), C1NiP1 81.65(8), C1NiP2 81.44(8), HNiH

65.2(14), HBH 109(2).

c 2 Figure 9. X-ray Crystal Structure of [2,6-( Pe2PO)2C6H3]Ni( -BH4) (10c) (50% probability level). Hydrogen atoms (except those attached to boron) are omitted for clarity (One of the cyclopentyl rings shows some disorder; a two-component model is presented for C24/C25 with

55% major occupancy). Selected bond lengths (Å) and angles (deg): NiH 1.86(3) and 1.60(3),

B1H (bridging) 1.14(3) and 1.28(3), B1H (terminal) 1.12(3) and 1.17(3), NiC1 1.892(2),

NiP1 2.1475(6), NiP2 2.1473(6), Ni…B1 2.180(3), P1NiP2 161.49(3), C1NiP1 81.96(7),

C1NiP2 81.84(7), HNiH 67.0(13), HBH 112.6(19).

Although 10a-c are incredibly robust at room temperature, they do lose BH3 under forcing conditions (60oC), particularly in the case of 10b with a relatively crowded nickel center.

For instance, heating a solid sample of 10b under a dynamic vacuum at 60ºC for 24 h produces

3b with a 17% conversion (Scheme 10). In strong contrast, 10a and 10c remain intact under the same conditions. An alternative way to release BH3 from a borohydride complex is by adding a trapping reagent such as triethylamine (NEt3) to form a stable BH3•NEt3 adduct (Scheme 10).

Such a reaction with 10b at 60ºC yields 3b cleanly and nearly quantitatively within 4 h. The

71 decomplexation of BH3 from 10a and 10c, on the other hand, are much more sluggish. After 24 h at 60ºC, approximately 40% of 10a and 10c are converted to the corresponding nickel hydride complexes. In addition, side products start to form, as indicated by 31P{1H} NMR spectroscopy.

Previous group member’s work on related pincer complexes have suggested that the P–O bonds in less bulky pincer complexes are more vulnerable to attack by a base,49 resulting in breakdown of the pincer backbone. A possible explanation for a much faster and cleaner reaction with 10b

t could be that the bulky Bu groups not only facilitate the loss of BH3, but also shield the P–O bonds preventing potential decomposition reactions.

Because 10a-c can dissociate to form nickel hydrides and a borane (B2H6) with or without an external reagent, I sought to explore the possibility of reducing CO2 with these borohydride complexes following mechanistic steps analogous to those outlined in Scheme 9.

When exposed to 1 atm of CO2 at room temperature, none of the borohydride complexes showed any reaction. However, when the mixture was warmed up to 60ºC, 10b was fully converted to a

49 Zhang, J.; Medley, C. M.; Krause, J. A.; Guan, H. Organometallics 2010, 29, 6393. 72 nickel formate complex 9b over a period of 8 h (eq 11). The 1H NMR spectrum of the reaction mixture also revealed several resonances in the range of 3.33-3.62 ppm, suggesting that CO2 is reduced to (CH3O)3B and other methoxyboryl species. Under the same conditions, 10a and 10c are much less reactive toward CO2, producing only 5-8% of the corresponding nickel formate complexes even after 48 h. The amounts of methoxyboryl species are negligible in both cases and could not be observed by 1H NMR. The results here follow the same trend observed in the

BH3 decomplexation experiments. A pincer ligand with smaller phosphorus substituents allows

– nickel to bind to BH4 more tightly, leading to less efficient release of BH3 for subsequent reactions.

3.6.2 Reactions with 9-BBN

As shown in Scheme 8, in the presence of 9-BBN (exists as a dimer), the formate moiety of 9b can be reduced to B-methoxy-9-BBN. Because 3b is cleanly produced in this reaction and because 3b can react with CO2 to reform 9b, 9-BBN could be a suitable reagent for the catalytic reduction of CO2, only if its interaction with 3b does not play a detrimental role. This prompted me to investigate the reactions of nickel hydride complexes 3a-c with 9-BBN. I first focused on the reactivity of 3b, which showed no color change upon mixing with a half-equivalent of 9-

1 31 1 BBN dimer in C6D6. Both H NMR and P{ H} NMR spectra of the mixture confirm no net

73 reaction. However, the deuterium in 3b-D exchanges readily with the BH proton of 9-BBN dimer, indicating some degree of interaction between the two species.

The sterically less demanding hydride 3a behaves drastically differently. Its yellow

31 1 solution in C6D6 turns deep red once mixed with a half-equivalent of 9-BBN dimer. The P{ H} spectrum shows two singlet resonances at 206.55 ppm and 200.80 ppm in a ratio of 22 : 78. The minor peak corresponds to the unreacted 3a while the major one suggests a new species 11a being formed. The most characteristic hydrogen resonance of 11a is a broad singlet at –2.22 ppm (1/2 = 148 Hz at 22ºC), which integrates as two hydrogens per pincer molecule.

Replacing the NMR solvent with toluene-d8 allowed me to examine the reaction in a wider temperature range. From –50oC to 40oC, the ratio between 3a and 11a does not vary but the hydride resonance of 11a becomes much sharper at lower temperature. Above 40oC, both hydride and phosphorus resonances of 3a and 11a are significantly broadened and eventually coalesce at about 70oC. The room-temperature 11B NMR spectrum shows a singlet at 27.8 ppm for the unreacted 9-BBN and a singlet at –10.9 ppm presumably for 11a. The negative chemical shift is consistent with a four-coordinate boron center, and 11a is therefore best described as a dihydridoborate complex. The fact that 3a, 11a, and 9-BBN coexist in solution with no change in ratios over time suggests that the reaction is under an equilibrium condition (eq 12). The Keq value at 22oC was determined to be 13.1 ± 4.5 M-1/2 (the average of three experiments) from

31P{1H} NMR integrations and the initial concentrations of 3a and 9-BBN. The entropy loss due to the formation of the adduct 11a is presumably compensated by the entropy gained from the dissociation of 9-BBN dimer, which explains the virtual insensitivity of the equilibrium constant toward temperature change.

It was reasoned, from the steric consideration, that 3c might have a slight advantage over

3a in promoting the formation of the related dihydridoborate complex. Indeed, the reaction of 3c

31 1 with 9-BBN in toluene-d8 generates 11c as the major species with only 5% (based on P{ H}

NMR) of unreacted 3c. The hydride resonance of 11c appears at –1.87 ppm as a broad singlet

(1/2 = 107 Hz at 22ºC) and the phosphorus and boron resonances are found at 187.72 ppm and

–11.7 ppm in 31P{1H} and 11B NMR spectra, respectively. One interesting observation made during the reaction of 3c, however, was that after several hours, a small amount (< 5%) of a second new species was produced, as first suggested by a proton resonance centered at 4.14 ppm as a doublet of doublet of triplets (J = 323.6, 15.2, and 3.6 Hz). The minor product did not grow significantly when the reaction was monitored for longer time. However, under catalytic conditions, this side reaction could cause some serious damage to the active nickel hydride catalyst. On the basis of the large one-bond coupling constant of 323.6 Hz and the splitting pattern, I hypothesized that 9-BBN could also attack the more exposed pincer backbone to

c c release ( Pe)2PH from the nickel pincer complex as a borane adduct ( Pe)2PH•(9-BBN). This was

1 c confirmed by examining the H NMR spectrum of the 2 : 1 mixture of ( Pe)2PH and 9-BBN

31 11 dimer in toluene-d8. In addition, this adduct exhibits a P resonance at –6.80 ppm and a B resonance at –18.1 ppm, both of which are present in the corresponding NMR spectra of the

3c/9-BBN reaction. Despite the side reaction as well as the unreacted hydride 3c, the IR 75 spectrum of the reaction mixture (in toluene) shows no bands in the (B-Ht) region, but two very

-1 -1 weak bands at 1943 cm and 1857 cm , which can be assigned to B-Hb stretches. Therefore, the dihydridoborate ligand in the major species 11c most likely adopts a 2-coordination mode, in a

– similar fashion as the BH4 ligand in 10a-c.

Fortunately, X-ray quality single crystals were successfully grown from a toluene solution of 11c (generated in situ) layered with pentane at –30oC. When the crystals were left dry, their deep red color faded immediately to yellow, implying the formation of the nickel hydride 3c due to the dissociation of 9-BBN. Extra care was thus taken to ensure that the crystals made contact with a small amount of mother liquid, and then were quickly covered with

Paratone oil prior to X-ray diffraction at low temperature (–123oC). As shown in Figure 10, dihydridoborate acts as a bidentate ligand. The Ni…B distance of 2.207(3) Å is comparable to those in 10a-c, but much longer than the sum of Ni and B covalent radii of 2.08 Å. Disregarding

c the Pe groups, the structure would belong to a C2v point group. The deviation from planarity for the plane generated by P1, C1, P2, Ni, B1, C27, and C31 atoms is only 0.0171Å. If Y is defined as the centroid of C27 and C31 atoms, it is almost linear with Ni and B1 with a NiB1Y angle of 179.48o. These structural parameters provide further evidence supporting the 2-coordination mode.

c Figure 10. X-ray Crystal Structure of [2,6-( Pe2PO)2C6H3]Ni(-H)2[B(C8H14)] (11c) (50% probability level). Hydrogen atoms (except those attached to boron) are omitted for clarity (One of the cyclopentyl rings shows some disorder; a two-component model is presented for C25 with

70% major occupancy). Selected bond lengths (Å) and angles (deg): NiH 1.71(3) and 1.77(2),

B1H both 1.20(3), NiC1 1.901(2), NiP1 2.1570(8), NiP2 2.1586(8), Ni…B1 2.207(3),

P1NiP2 163.33(3), C1NiP1 81.41(8), C1NiP2 81.96(8), HNiH 65(1), HBH 104(2).

At room temperature, a complete conversion from 11c to 3c is accomplished under a dynamic vacuum in just 1 h. The use of any reagents that trap either 9-BBN or 3c should also favor the similar equilibrium mentioned in equation 12 to the backward direction. As expected, addition of one equivalent of NEt3 to a solution of 11c in C6D6 produces 3c and 9-BBN•NEt3 in

90% conversion (Scheme 11). Additionally, 11c readily reacts with CO2 (~1 atm), as suggested by an immediate color change from red to yellow. In view of rapid insertion of CO2 into 3c, the

77 obtained nickel species was initially postulated as the insertion product, nickel formate complex

9c. However, the 1H NMR spectrum of the reaction mixture shows a new peak at 8.62 ppm, which is shifted from 8.39 ppm for the formate resonance of 9c. The 31P{1H} NMR spectrum also displays a more downfield shifted resonance (175.24 ppm) than the one for 9c (173.35 ppm). Possibly, the new species is an adduct of 9c and 9-BBN through a Lewis acid-base interaction. An alternative structure involves both a hydride and a formate bridging nickel and boron centers. In any event, 9c slowly forms over a period of 8 h, with a concurrent formation of

B-methoxy-9-BBN and a borate ester (Scheme 11) which supports the relevance of these adducts in the catalytic reduction of CO2.

3.6.3 Reactions with HBcat

Compared to BH3 and 9-BBN, HBcat has a less electrophilic boron center due to the donation of electrons from neighboring oxygen atoms. As a result, it is less favorable for HBcat to abstract H– from a nickel hydride complex to produce a dihydridoborate complex. When a solution of 3b in C6D6 was treated with 1 equiv of HBcat at room temperature, no color change was observed. Judging from the 31P{1H} and 11B NMR spectra, there is no net reaction between the two species. Interestingly, the 1H NMR spectrum reveals the absence of NiH and BH resonances, suggesting a rapid exchange process involving these two hydrogens. All other resonances of 3b and HBcat remain the same as those for individual reagents. Variable-

1 temperature experiments in toluene-d8 provided some insights. The H NMR spectra of the solution were recorded in the temperature range from −70oC to 65oC and the representative ones were shown in Figure 11. A broad resonance near −3.0 ppm was present at room temperature or higher, but absent at lower temperatures. Growing crystals of the product at −30oC was unsuccessful. Attempted synthesis of the compound on a preparative scale fully recovered 3b when the reaction solution was concentrated under vacuum (HBcat has a boiling point of 50oC at

50 mmHg). These studies suggest that 3b and HBcat may reversibly form an adduct with a hydride ligand bridging both Ni and B. To provide further support for the exchange reaction, a reaction of 3b-D and HBcat (approximately 1 : 1) was carried out in C6D6 (eq 13). It is interesting to note that the 31P{1H} NMR spectrum of this mixture resembles 3b rather than 3b-

D; the 31P resonance of 3b appears at 219.35 ppm as a singlet while the 31P resonance of 3b-D exhibits at 219.39 ppm as a 1 : 1 : 1 triplet due to isotopic shift and 31P-2H coupling. Evaporation of the volatiles under vacuum followed by redissolution of the residue in C6D6 shows 88% of the

79 nickel species containing protium, which further supports the exchange of hydrogens between 3b and HBcat.

Figure 11. Variable-Temperature 1H NMR Spectra of the Mixture of 3b and HBcat in Toluene- d8.

The first indication of a net reaction between 3a and HBcat was the change of color from yellow to orange when one equiv of HBcat was added to the solution of 3a in C6D6. Although the 31P{1H} NMR spectrum only shows the resonance of 3a, the 1H NMR spectrum contains a slightly broad peak at –1.90 ppm (1/2 = 28 Hz at 22ºC), a chemical shift similar to the one observed in the 3a/9-BBN reaction. This hydride resonance (for the same reaction in toluene-d8) shifts downfield and becomes broader as the temperature decreases (Figure 12). This can be explained by equilibrium (eq 14) that is fast on the NMR time scale. The lower temperature shifts the equilibrium more to the 12a side, resulting in a significant change in chemical shift.

The line broadening reflects the deceased rates of the forward and backward reactions; however, it is reasonable to assume that even at –70C the reactions remain very fast. Variable- temperature 11B NMR experiments provide additional evidence supporting the proposed equilibrium. At room temperature, only one resonance is observed at 27.9 ppm, which corresponds to HBcat. At –60C, a broad new resonance presumably for 12a emerges at 13.7 ppm while the 11B resonance of HBcat is substantially broadened and its intensity is decreased

(Figure 13). The upfield shifted 11B resonance is more consistent with a dihydridoborate complex. In contrast, the 11B resonance of a -borane complex27,28 or a metal boryl complex31

(other possible structures of the adduct) is typically downfield from that of free borane.

70 oC

50oC

30oC

10oC

22oC

40oC

o 60 C

0 -3 ppm

Figure 12. Variable-Temperature 1H NMR Spectra of the 1 : 1 Mixture of 3a and HBcat in

Toluene-d8.

o -60 C

Figure 13. Temperature-Dependent 11B NMR Study of a Mixture of 3a and HBcat in

Toluene-d 8 o 22 C

Analogously, 3c and HBcat are in rapid equilibrium with a dihydridoborate complex.

The NMR data are similar to those described above for 3a, except that more side reactions occur

1 in this case. A small amount of H2 was detected by H NMR spectroscopy, implying that a reaction related to eq 7 is perhaps taking place. In addition, a set of broad multiplets in the 3.8-

c 11 5.0 ppm range is likely due to the formation of ( Pe)2PH•HBcat. B NMR spectroscopy also

– reveals peaks at 22.3 ppm and 14.7 ppm, which are assigned to B2(cat)3 and [B(cat)2] based on literature values.32a Collectively, these results once again highlight the vulnerability of the P–O bonds49 in the presence of a hydride donor if they are not sterically protected.

3.7 Rationalization of Catalytic Activities

The abovementioned interactions of nickel hydride complexes with boranes are expected to have a significant influence on the catalytic reduction of CO2. As the size of the substituents on the phosphorus donors decreases (from 3b to 3a to 3c), the required time to reach the maximum TON becomes longer (shown in Table 1). The relative catalytic activity of the hydride complexes correlates with their relative reactivity observed in the stoichiometric reactions with boranes. Complex 3b, which is the best catalyst of the series, does not form any adduct with either 9-BBN or HBcat. In contrast, complex 3a reacts with both HBcat and 9-BBN more favorably than 3b to form dihydridoborate complexes, as demonstrated in equations 12 and

14. Although this process is reversible and the release of 3a from these hydridoborate complexes is reasonably fast, the equilibrium significantly reduces the steady-state concentration of the active nickel hydride species, resulting in slower catalysis. In the case of 3c, the analogous equilibrium is even more favorable toward the catalytically dormant nickel hydridoborate species, possibly because of reduced congestion at the nickel center. The observation of a

c phosphine-borane adduct ( Pe)2PH•(9-BBN) in the stoichiometric reaction suggests that the decomposition of catalyst 3c by 9-BBN through the cleavage of P–O bonds may play an additional role in reducing the concentration of 3c during catalysis. The same type of catalyst- decomposition pathway of 3c possibly becomes more pronounced for HBcat which led to incomplete catalytic reaction (Table 1, entry 3). Compared to 9-BBN, HBcat has a weaker interaction with the nickel hydride complexes. Related dihydridoborate complexes do form when complexes 3a and 3c are mixed with HBcat, but at room temperature they exist in such a small quantity that they should not drastically impact the catalysis. When BH3•THF was employed as reducing reagent, active nickel hydride complexes formed robust nickel

84 borohydride complexes under the catalytic conditions. As demonstrated in the stoichiometric studies reversing these reactions is only possible at higher temperatures if there is a trapping agent available, or under a dynamic vacuum (only for 3b). As catalytic reactions with BH3•THF were performed at ambient temperature and no trapping reagent was used, no reduction product was observed.

3.8 Conclusions

Efficient nickel catalysts have been developed to convert CO2 to methanol derivatives using boranes as the reducing reagents. The mechanism consists of three catalytic cycles. In each catalytic cycle, the oxidation state of CO2 is reduced by 2, followed by consumption of one equivalent of borane. Formaldehyde has been experimentally tested as an intermediate and supported by DFT analysis. It has been observed that complexes with more bulky phosphorous substituents are more effective catalysts than the ones with small substituents. This difference in catalytic activity has been rationalized by proposing off-the-catalytic-loop adducts of nickel hydrides and boranes. NMR studies and X-ray structure determination have confirmed dihydridoborate complexes as the adducts of nickel hydrides and boranes. When these complexes form favorably, they become the dormant species of the catalytic reduction of CO2.

3.9 Experimental Section

General Experimental Information. Unless otherwise noted, all the organometallic compounds were prepared and handled under an argon atmosphere using standard Schlenk and inert-atmosphere box techniques. Dry and oxygen-free solvents (toluene and THF) were collected from an Innovative Technology solvent purification system and used throughout the

85 experiments. C6D6 and toluene-d8 were distilled from Na and benzophenone under an argon atmosphere. Other solvents (pentane, THF and CD2Cl2) were used as received from commercial sources. Catecholborane was purified by vacuum distillation prior to use, and other boranes were purchased from Sigma-Aldrich and used without further purification. 1H, 13C{1H}, and

31P{1H} NMR spectra were recorded on a Bruker Avance-400MHz spectrometer. 11B NMR spectra were recorded on a Bruker AMX-400MHz wide-bore spectrometer. 2H NMR spectra were recorded in a Bruker Avance-500MHz instrument. Chemical shift values in 1H and 13C

NMR spectra were referenced internally to the residual solvent resonances. 31P and 11B NMR spectra were referenced externally to 85% H3PO4 (0 ppm) and BF3•Et2O (0 ppm), respectively.

Infrared spectra were recorded on a Thermo Scientific Nicolet 6700 FT-IR spectrometer equipped with smart orbit diamond attenuated total reflectance (ATR) accessory. Compounds

2a-c50,17a,25 and 3a-c17a,25 were prepared according to the literature procedures.

i Synthesis of [2,6-( Pr2PO)2C6H3]NiOC(O)H (9a). Under a carbon

dioxide atmosphere (1 atm), the solution of 3a (200 mg, 0.50 mmol) in 20 mL of toluene was stirred at room temperature for 1h. The solvent was evaporated under vacuum and the nickel formate 9a was isolated as an orange/yellow crystalline solid (183 mg, 82%

1 yield). H NMR (400 MHz, C6D6, ): 1.16-1.21 (m, PCH(CH3)2, 12H), 1.30-1.36 (m,

3 PCH(CH3)2, 12H), 2.25-2.30 (m, PCH(CH3)2, 4H), 6.53 (d, ArH, JH-H = 7.6 Hz, 2H), 6.85 (t,

3 31 1 ArH, JH-H = 7.6 Hz, 1H), 8.34 (s, NiOCOH, 1H). P{ H} NMR (162 MHz, C6D6, ): 183.29

50 Pandarus, V.; Zargarian, D. Organometallics 2007, 26, 4321. 86

-1 (s). IR (in toluene, cm ): 1605 (C=O). Anal. Calcd for C19H32O4P2Ni: C, 51.27; H, 7.25.

Found: C, 51.49; H, 7.19.

t Synthesis of [2,6-( Bu2PO)2C6H3]NiOC(O)H (9b). Under a

carbon dioxide atmosphere (1 atm), the solution of 3b (137 mg, 0.30 mmol) in 20 mL of toluene was stirred at room temperature for 30 min. The solvent was evaporated under vacuum and the nickel formate 9b was isolated as an orange/yellow crystalline

1 3 solid (126 mg, 84% yield). H NMR (400 MHz, C6D6, δ): 1.40 (virtual triplet, PC(CH3)3, JP-H =

3 3 7.2 Hz, 36H), 6.48 (d, ArH, JH-H = 8.0 Hz, 2H), 6.84 (t, ArH, JH-H = 8.0 Hz, 1H), 8.41 (t,

4 13 1 2 NiOCHO, JP-H = 2.2 Hz, 1H). C{ H} NMR (101 MHz, C6D6, δ): 27.98 (t, C(CH3)3, JP-C = 3.2

1 2 Hz), 39.40 (t, C(CH3)3, JP-C = 6.6 Hz), 105.79 (t, ArC, JP-C = 5.8 Hz), 125.10, 129.31, 168.02,

31 1 -1 170.57. P{ H} NMR (162 MHz, C6D6, δ): 186.61 (s). IR (in toluene, cm ): 1604 (C=O). Anal.

Calcd for C23H40O4P2Ni: C, 55.12; H, 8.04. Found: C, 55.09; H, 8.05.

c Synthesis of [2,6-( Pe2PO)2C6H3]NiOC(O)H (9c). Under a carbon

dioxide atmosphere (1 atm), the solution of 3c (100 mg, 0.20 mmol) in 20 mL of toluene was stirred at room temperature for 30 min. The solvent was evaporated under vacuum and the nickel formate 9c was isolated as an orange/yellow crystalline solid (85 mg,

1 77% yield). H NMR (400 MHz, C6D6, ): 1.38-1.42 (m, CH2, 8H), 1.60-1.76 (m, CH2, 12H),

3 2.00-2.10 (m, CH2, 12H), 2.55-2.60 (m, PCH, 4H), 6.56 (d, ArH, JH-H = 7.6 Hz, 2H), 6.86 (t,

3 31 1 ArH, JH-H = 7.6 Hz, 1H), 8.39 (s, NiOCHO, 1H). P{ H} NMR (162 MHz, C6D6, ): 173.31

-1 (s). IR (in toluene, cm ): 1605 (C=O). Anal. Calcd for C27H40P2O4Ni: C, 59.04; H, 7.34.

Found: C, 59.06; H, 7.48.

i Synthesis of [2,6-( Pr2PO)2C6H3]Ni(BH4) (10a). Method A from 3a:

Under an argon atmosphere, a 1.0 M solution of BH3•THF in THF (0.33 mL,

0.33 mmol) was added slowly to a Schlenk flask containing a chilled solution (0 oC) of 3a (100 mg, 0.25 mmol) in 10 mL of toluene. The color of the solution changed immediately from yellow to orange. After stirring at room temperature for 30 min, the reaction mixture was exposed to air and filtered through a pad of Celite. Removal of the volatiles under vacuum afforded an orange solid, which was further purified by washing with pentane (3 mL × 3) and then dying under vacuum (86 mg, 85% yield). Method B from 2a: Under an argon atmosphere,

10 mL of toluene was added to a Schlenk flask containing 2a (100 mg, 0.23 mmol) and NaBH4

(9.0 mg, 0.24 mmol). The reaction mixture was stirred at room temperature for 12 hours, resulting in a yellow-orange solution, which was filtered through a pad of Celite (under air) and evaporated to dryness. The obtained orange solid was washed with pentane (3 mL × 3) and dried

1 under vacuum to afford the pure product (89 mg, 90% yield). H NMR (400 MHz, C6D6, δ):

0.71 (br q, NiBH4, 4H), 1.05-1.10 (m, PCH(CH3)2, 12H), 1.23-1.29 (m, PCH(CH3)2, 12H),

3 3 2.07-2.14 (m, PCH(CH3)2, 4H), 6.61 (d, ArH, JH-H = 8.0 Hz, 2H), 6.85 (t, ArH, JH-H = 8.0 Hz,

1 1 1H). H NMR (400 MHz, CD2Cl2, δ): 1.39 (br q, JH-B = 79 Hz, NiBH4, 4H), 1.24-1.36 (m,

3 PCH(CH3)2, 24H), 2.29-2.36 (m, PCH(CH3)2, 4H), 6.42 (d, ArH, JH-H = 8.0 Hz, 2H), 6.91 (t,

3 13 1 ArH, JH-H = 8.0 Hz, 1H). C{ H} NMR (101 MHz, C6D6, δ): 16.60 (s, CH3), 16.89 (s, CH3),

28.35 (t, JC-P = 11.9 Hz, PCH), 105.49 (t, JC-P = 6.4 Hz, ArC), 168.75 (t, JC-P = 9.6 Hz, ArC);

13 1 other resonances were obscured by the solvent peaks. C{ H} NMR (101 MHz, CD2Cl2, δ):

16.90 (CH3), 17.15 (t, JC-P = 2.5 Hz, CH3), 28.65 (t, JC-P = 11.9 Hz, PC(CH3)3), 105.34 (t, JC-P =

6.5 Hz, ArC), 128.08 (s, ArC), 129.94 (t, JC-P = 21.3 Hz, ArC), 168.68 (t, JC-P = 9.5 Hz, ArC).

31 1 31 1 P{ H} NMR (162 MHz, C6D6, δ): 199.43 (s). P{ H} NMR (162 MHz, CD2Cl2, δ): 199.29

11 1 11 1 (s). B NMR (128 MHz, C6D6, δ): 33.5 (quin, JB-H = 63.9 Hz). B{ H} NMR (128 MHz,

-1 C6D6, δ): 33.5 (s). Selected data from ATR-IR (solid, cm ): 2406 (s), 2382 (s), 2274 (w), 1947

-1 (br), 1144 (s). Selected data from transmission-IR (CH2Cl2, cm ): 2398 (s), 2380 (s), 2273 (w),

1954 (br), 1143 (s). Anal. Calcd for C18H35O2BP2Ni: C, 52.10; H, 8.50. Found: C, 52.30; H,

8.47.

t Synthesis of [2,6-( Bu2PO)2C6H3]Ni(BH4) (10b). This compound was

prepared in 88% yield (via Method A) and 71% yield (via Method B) by

1 procedures similar to those used for 10a. H NMR (400 MHz, C6D6, δ): 0.45 (br q, NiBH4, 4H),

3 3 3 1.37 (vt, JH-P = 7.0 Hz, CH3, 36H), 6.56 (d, ArH, JH-H = 8.0 Hz, 2H), 6.85 (t, ArH, JH-H = 8.0

1 1 3 Hz, 1H). H NMR (400 MHz, CD2Cl2, δ): 1.14 (br q, JH-B = 80 Hz, NiBH4, 4H), 1.43 (vt, JH-P

3 3 = 7.0 Hz, CH3, 36H), 6.41 (d, ArH, JH-H = 8.0 Hz, 2H), 6.90 (t, ArH, JH-H = 8.0 Hz, 1H).

13 1 C{ H} NMR (101 MHz, C6D6, δ): 27.95 (t, JC-P = 2.7 Hz, CH3), 40.06 (t, JC-P = 7.6 Hz,

PC(CH3)3), 105.14 (t, JC-P = 6.1 Hz, ArC), 169.47 (t, JC-P = 8.8 Hz, ArC); other resonances were

13 1 obscured by the solvent peaks. C{ H} NMR (101 MHz, CD2Cl2, δ): 28.19 (s, CH3), 40.44 (t,

JC-P = 7.7 Hz, PC(CH3)3), 105.02 (t, JC-P = 6.1 Hz, ArC), 127.65 (t, JC-P = 20.5 Hz, ArC), 127.76

31 1 31 1 (s, ArC), 169.50 (t, JC-P = 8.7 Hz, ArC). P{ H} NMR (162 MHz, C6D6, δ): 200.15 (s). P{ H}

11 1 NMR (162 MHz, CD2Cl2, δ): 200.03 (s). B NMR (128 MHz, C6D6, δ): 35.9 (quin, JB-H = 89

11 1 - 65.4 Hz). B{ H} NMR (128 MHz, C6D6, δ): 35.9 (s). Selected data from ATR-IR (solid, cm

1 -1 ): 2412 (s), 2397 (s), 1961 (br). Selected data from transmission-IR (CH2Cl2, cm ): 2412 (s),

2397 (s), 1955 (br). Anal. Calcd for C22H43O2BP2Ni: C, 56.10; H, 9.20. Found: C, 55.88; H,

9.15.

c Synthesis of [2,6-( Pe2PO)2C6H3]Ni(BH4) (10c). This compound was

prepared in 83% yield (via Method A) and 79% yield (via Method B) by

1 procedures similar to those used for 10a. H NMR (400 MHz, C6D6, δ): 0.63 (br d, NiBH4,

3 4H), 1.37-2.10 (m, CH2, 32H), 2.34-2.38 (m, PCH, 4H), 6.64 (d, ArH, JH-H = 7.2 Hz, 2H), 6.87

3 1 (t, ArH, JH-H = 7.2 Hz, 1H). H NMR (400 MHz, CD2Cl2, δ): 1.39 (br q, NiBH4, 4H), 1.61-

3 1.77 (m, CH2, 32H), 2.41-2.46 (m, PCH, 4H), 6.39 (d, ArH, JH-H = 8.0 Hz, 2H), 6.89 (t, ArH,

3 13 1 JH-H = 8.0 Hz, 1H). C{ H} NMR (101 MHz, C6D6, δ): 26.57 (t, JC-P = 4.4 Hz, CH3), 26.99 (t,

JC-P = 3.6 Hz, CH3), 27.88 (t, JC-P = 3.0 Hz, CH3), 28.32 (CH3), 39.55 (t, JC-P = 13.0 Hz, PCH),

105.58 (t, JC-P = 6.5 Hz, ArC), 130.60 (t, JC-P = 21.5 Hz, ArC), 168.53 (t, JC-P = 9.9 Hz, ArC); one

13 1 resonance was obscured by the solvent peaks. C{ H} NMR (101 MHz, CD2Cl2, δ): 26.83 (t,

JC-P = 4.5 Hz, CH3), 27.23 (t, JC-P = 3.7 Hz, CH3), 28.01 (t, JC-P = 3.2 Hz, CH3), 28.52 (CH3),

39.65 (t, JC-P = 13.1 Hz, PCH), 105.35 (t, JC-P = 6.4 Hz, ArC), 127.96 (s, ArC), 130.33 (t, JC-P =

31 1 21.3 Hz, ArC), 168.38 (t, JC-P = 9.7 Hz, ArC). P{ H} NMR (162 MHz, C6D6, δ): 190.42 (s).

31 1 11 P{ H} NMR (162 MHz, CD2Cl2, δ): 190.43 (s). B NMR (128 MHz, C6D6, δ): 32.2 (br).

11 1 -1 B{ H} NMR (128 MHz, C6D6, δ): 32.2 (s). Selected data from ATR-IR (solid, cm ): 2387

(s), 2370 (s), 2068 (br), 1886 (w), 1799 (br), 1122(s). Selected data from transmission-IR

-1 (CH2Cl2, cm ): 2381 (br s), 2075 (br), 1885 (br), 1792 (br), 1123(s). Anal. Calcd for

C26H43O2BP2Ni: C, 60.16; H, 8.35. Found: C, 60.02; H, 8.51.

c Synthesis of [2,6-( Pe2PO)2C6H3]Ni(BC8H16) (11c). Under an

argon atmosphere, 9-BBN dimer (13.3 mg, 54.5 µmol) and 3c (50 mg, 99

µmol) were mixed with ~1 mL of toluene in a scintillation vial. After standing at room temperature for 1 h, the resulting red solution was carefully layered with about same volume of pentane and then kept in a –30ºC freezer for at least 12 h. Red crystals obtained were suitable for X-ray crystallographic study. The NMR sample was prepared by mixing a 1 : 2 ratio of 9-

BBN dimer and 3c in ~0.6 mL of C6D6. About 5% of the total nickel species was identified as unreacted 3c. When the solution was kept for several hours, a new species started to form and

c 1 was identified as ( Pe)2PH•(9-BBN). Characterization data for 11c: H NMR (400 MHz, C6D6,

δ): 1.80 (br s, Ni(H)2B, 2H), 0.91 (br, BCH, 2H), 1.36-2.35 (m, CH and CH2, 48H), 6.60 (d,

3 3 31 1 ArH, JH-H = 8.0 Hz, 2H), 6.83 (t, ArH, JH-H = 8.0 Hz, 1H). P{ H} NMR (162 MHz, C6D6, δ):

11 11 1 188.13 (s). B NMR (128 MHz, C6D6, δ): 11.4 (br s). B{ H} NMR (128 MHz, C6D6, δ):

11.3 (s). Selected data from transmission-IR (toluene, cm-1): 1943 (w), 1857 (w).

13 13 Exchange Reaction between 9b- C and CO2. In a J. Young NMR tube, 9b- C was

13 prepared in situ by mixing a toluene-d8 solution (ca. 0.6 mL) of 3b (4.6 mg, 10 μmol) with CO2

(1 atm) at room temperature for 30 min. The 1H NMR spectrum of the solution confirmed complete conversion of 3b to 9b-13C with a characteristic 13C-labelled formate resonance (8.42

1 4 ppm, doublet of triplets, JC-H = 194.4 Hz, JP-H = 2.4 Hz). The NMR tube was cooled by liquid

N2 and the gas inside the tube was removed under vacuum. The NMR tube was then disconnected from the vacuum source, warmed to room temperature, and backfilled with 1 atm 91

1 of CO2. After 12 h of mixing at room temperature, its H NMR spectrum showed that 36% of the

4 nickel formate contained the non-labeled formate carbon (8.42 ppm, triplet, JP-H = 2.4 Hz).

Stoichiometric Reduction of 9b with Catecholborane (HBcat). In a glove box, an

NMR tube equipped with a Teflon screw-cap was charged with 9b (12.5 mg, 25 μmol, independently synthesized from 3b and HCO2H) and 0.6 mL of C6D6. Catecholborane (2.7 μL,

25 μmol) was then added via syringe and the reaction was immediately monitored by both 1H

NMR and 31P{1H} NMR spectroscopy. The major resonances were assigned to those for 3b and

1 2 HCOOBcat. H NMR (400 MHz, C6D6, δ): −7.96 (t, NiH, JP-H = 53.2 Hz, 1H), 1.30 (virtual

3 3 triplet, PC(CH3)3 of 3b, JP-H = 6.8 Hz, 36H), 6.85 (d, Ar of 3b, JH-H = 7.6 Hz, 2H), 6.98 (t, Ar of

3 31 1 3b, JH-H = 7.6 Hz, 1H), 8.68 (bs, HCOOBcat, 1H). P{ H} NMR (162 MHz, C6D6, δ): 219.37

(PC(CH3)3 of 3b). When more than one equivalent of catecholborane was added, a new set of

1 1 resonances appeared in H NMR that were assigned to those for CH3OBcat. H NMR (400 MHz,

C6D6, δ): 3.34 (s, CatBOCH3, 3H), 6.71-6.78 (m, Ar), 6.88-7.00 (m, Ar).

Attempted Synthesis of HCOOBcat. Under an argon atmosphere ClBcat (0.31 g, 2 mmol) and sodium formate (0.14 g, 2 mmol) were mixed in 30 mL of THF. The mixture was refluxed for 24 h, after which the solvent was removed under vacuum. The residue was extracted with toluene to give a clear solution, and the solvent was removed again under vacuum to afford a colorless oil. 1H NMR spectroscopy revealed a complicated mixture of boryl species, with no evidence for the formation of HCOOBcat.

Independent Synthesis of CH3OBcat. Under an argon atmosphere ClBcat (1.00 g, 6.48 mmol) and sodium methoxide (0.385 g, 7.13 mmol) were mixed in 30 mL of THF. The mixture was refluxed for 24 h, cooled to room temperature, and passed through a pad of Celite.

Evaporating the solvent under vacuum gave a colorless viscous oil (0.62 g, 64% yield). 1H NMR

11 (400 MHz, C6D6, δ): 3.35 (s, CH3OBcat, 3H), 6.71-6.73 (m, Ar, 2H), 6.88-6.90 (m, Ar, 2H). B

NMR (128 MHz, C6D6, δ): 23.4 (broad singlet). These data is consistent with the values reported in the literature.23

Reduction of 9b-13C with HBcat. In a J. Young NMR tube, 9b-13C was prepared in situ

13 by mixing a C6D6 solution (ca. 0.6 mL) of 3b (11.4 mg, 25 μmol) with CO2 (1 atm) at room

1 1 13 1 temperature for 30 min. The H NMR (8.42 ppm, doublet, JC-H = 194.4 Hz) and C{ H} NMR

(168.3 ppm) spectra of the solution confirmed complete conversion of 3b to 9b-13C with residual

13 13 1 CO2 present (125.3 ppm in the C{ H} NMR spectrum). The NMR tube was brought into a glove box, and HBcat (10.8 μL, 100 μmol) was added via syringe and the reaction was further

1 monitored by NMR spectroscopy. H NMR (400 MHz, C6D6, δ): 1.14 (virtual triplet, PC(CH3)3,

3 13 1 JP-H = 6.8 Hz, 36H), 3.34 (d, CH3OBcat, JC-H = 145.2 Hz, 3H), 6.32 (d, Ar of 9b-

13 3 13 3 C•catBOBcat, JH-H = 7.6 Hz, 2H), 6.73 (t, Ar of 9b- C•catBOBcat, JH-H = 7.6 Hz, 1H), 6.73-

13 tBu 1 7.01 (m, Ar of boryl species), 8.45 (d, H COONi( PCP)•catBOBcat, JC-H = 218.0 Hz, 1H),

13 1 13 1 8.67 (d, H COOBcat, JC-H = 214.0 Hz, 1H). C{ H} NMR (101 MHz, C6D6, δ): 53.6

13 13 13 ( CH3OBcat), 125.3 (residual CO2), 162.1 (H COOBcat), 175.3 (d,

13 tBu 1 13 H COONi( PCP)•catBOBcat). H-coupled C NMR (101 MHz, C6D6, δ): 53.6 (q,

13 1 13 13 1 CH3OBcat, JC-H = 145.2 Hz), 125.3 (residual CO2), 162.1 (d, H COOBcat, JC-H = 214.0

13 tBu 1 31 1 Hz), 175.3 (H COONi( PCP)•catBOBcat, JC-H = 218.0 Hz). P{ H} NMR (162 MHz, C6D6,

δ): 188.73 (9b-13C•catBOBcat). 11B NMR spectrum did not yield any well-resolved peaks. The volatiles from the above reaction mixture were vacuum transferred into a flask and its 1H NMR

13 1 spectrum was recorded, which confirmed the formation of CH3OH. H NMR (400 MHz, C6D6,

13 1 13 δ): 2.98 (d, CH3OH, JC-H = 140.4 Hz, 3H), 5.23 (broad singlet, CH3OH, 1H). For mass

13 spectral analysis, CH3OH was derivatized to 2-methoxy-1-methyl pyridinium salt by reacting it

93 with 2-fluoro-1-methylpyridinium p-toluenesulfonate in the presence of triethylamine. HRMS

12 13 (ESI) calcd (found) for [ C6 CH10NO]+ 125.07904 (125.07902).

Catalytic Hydroboration of CO2 with 3b. Under an argon atmosphere a flame-dried 50 mL Schlenk flask was charged with 3b (5.0 mg, 11 μmol), catecholborane (586 μL, 5.50 mmol), and C6D6 (4 mL). To this solution, hexamethylbenzene (3.6 mg, 22 μmol) was added as an internal standard. The mixture was degassed by a freeze-pump-thaw cycle and placed under 1 atm of CO2. After stirring for 1 h at room temperature, a white precipitate was formed and allowed to settle at the bottom of the flask. A small portion of the clear liquid layer (ca. 0.6 mL) was taken into a J. Young NMR tube under an argon atmosphere. The 1H NMR spectrum of the solution showed that catecholborane was consumed completely. Turnover number (TON) was calculated as 495 (based on B−H) by comparing the NMR integration of the CH3OBcat methyl resonance (3.34 ppm) with that of the internal standard (2.10 ppm). The white precipitate was filtered off, washed with a small amount of pentane, and dried under vacuum (181 mg, 39% yield). It was identified as catBOBcat that had been previously synthesized by Nöth and co-

51 1 11 workers. H NMR (400 MHz, C6D6, δ): 6.71-6.73 (m, Ar, 4H), 6.88-6.91 (m, Ar, 4H). B

NMR (128 MHz, C6D6, δ): 22.4 (broad singlet). The single crystals were obtained by slow evaporation of a saturated solution of catBOBcat in pentane. A low resolution (Cu Kα at 2- theta

40 deg) structure determination confirmed the known catBOBcat structure.51

Hydrolysis of CH3OBcat to Generate Methanol. After the above catalytic reaction was complete, distilled water (300 μL, 16.5 mmol) was added via syringe and the resulting mixture was stirred at room temperature under an argon atmosphere for 24 h. The volatiles were vacuum transferred (60 oC) into a Schlenk tube, from which 600 μL of the solution was withdrawn using

51 Lang, A.; Knizek, J.; Nöth, H.; Schur, S.; Thomann, M. Z. Anorg. Allg. Chem. 1997, 623, 901. 94 a gas-tight microliter syringe and mixed with hexamethylbenzene (3.6 mg, 22 μmol) as the internal standard in an NMR tube. The concentration of methanol was calculated as 0.28 M by comparing the NMR integration of the CH3OH methyl resonance (3.08 ppm) with that of the internal standard (2.10 ppm). Using HBcat as the limiting reagent as well as the balanced equation in Scheme 4, the yield for CH3OH was calculated as 61%.

Catalytic Hydroboration of CO2 with 9b. In a glove box, a J. Young NMR tube was charged with 9b (5.0 mg, 10 μmol), catecholborane (32 μL, 300 μmol), and C6D6 (0.8 mL).

Hexamethylbenzene (1.6 mg, 10 μmol) was added as an internal standard. The resulting mixture was degassed by freezing the sample with liquid N2 followed by the evacuation under vacuum.

While the solution was allowed to thaw at room temperature, the NMR tube was charged with 1 atm of CO2. The constant gas-liquid mixing was accomplished by rotating the NMR tube, and the progress of the reaction was monitored by 1H NMR. The reaction was complete after 1h at room temperature, and the TON was calculated as 30. A similar reaction using 3b as the catalyst

(identical catalyst concentration) gave the same TON after 1 h at room temperature.

NMR Studies of Potential Catalyst Degradation. A similar procedure as above was used to check the catalytic activity of 3b (11 μmol in 0.8 mL of C6D6, with 50 equiv. of HBcat and 10 equiv. of hexamethylbenzene). The reaction was complete in 1 h with a TON of 39. 1H

NMR spectrum showed that no HBcat was left and the nickel species existed as 9b•catBOBcat.

Another 50 equiv. of HBcat was added into the J. Young NMR tube, followed by the addition of more CO2 (1 atm). Again reaction was complete in 1 h at room temperature with a cumulative

TON of 78, suggesting that the catalytic reaction was reinitiated with no sign of catalyst degradation.

Hydroboration of Paraformaldehyde with 3b (Stoichiometric Reaction). In a J.

Young NMR tube, 3b (10.0 mg, 22 μmol) and paraformaldehyde (0.7 mg, 22 μmol based on

o monomer) were mixed in ca. 0.6 mL of C6D6. The resulting solution was heated at 60 C for 1h to ensure that paraformaldehyde dissolved completely. 1H NMR spectrum showed approximately

10 resonances in the chemical shift range of 4.57-5.32 ppm and almost no aldehyde resonances.

HBcat (2.4 μL, 22 μmol) was then added at room temperature and the 1H NMR spectrum was recorded after 10 min. All the resonances in 4.57-5.32 ppm disappeared, while a new singlet appeared at 3.34 ppm that is consistent with the formation of CH3OBcat.

Lewis Acid-Base Interaction between 9b and catBOBcat. In a J. Young NMR tube, 9b was prepared in situ by mixing a solution of 3b (5.0 mg, 11 μmol) in ca. 0.6 mL C6D6 with CO2

(1 atm) at room temperature for 30 min. An equimolar amount of catBOBcat (2.8 mg, 11 μmol)

1 1 was added and the H NMR spectrum was recorded. H NMR (400 MHz, C6D6, δ): 1.15 (virtual

3 3 triplet, PC(CH3)3, JP-H = 7.2 Hz, 36H), 6.33 (d, Ar, JH-H = 7.6 Hz, 2H), 6.71-6.78 (m, Ar, 5H),

6.96-6.99 (m, Ar, 4H), 8.45 (s, HCOONi, 1H). These resonances were shifted significantly compared to those of free 9b.

Stoichiometric Reaction of 9 with HBcat. A screw-cap NMR tube was charged with 9a or 9b (25 μmol), hexamethylbenzene (internal standard, 1.0 mg, 6.1 μmol), and 0.6 mL of C6D6.

To this solution, HBcat (8.1 μL, 75 μmol) was added via a microliter syringe and the reaction was monitored by 1H and 31P{1H} NMR spectroscopy. The NMR yield was calculated by comparing the integration of CH3OBcat methyl resonance (3.34 ppm) with that of the internal standard (2.12 ppm). Because the 1H NMR resonances of 9c, overlap with that of hexamethylbenzene, for the study of 9c, hexamethyldisilane (0.07 ppm) was used as the NMR

96 internal standard. Control experiments showed that neither hexamethylbenzene nor hexamethyldisilane interfered with the reactions.

Stoichiometric Reaction of 9b with Boranes. A screw-cap NMR tube was charged with

9b (12.5 mg, 25 μmol), hexamethylbenzene (internal standard, 2.0 mg, 12.5 μmol), and 0.6 mL of C6D6. To this solution, 9-BBN or HBpin (75 μmol) was added via a microliter syringe and the reaction was monitored by 1H NMR and 31P{1H} spectroscopy. After 1h reaction with 9-BBN,

3b was the only nickel species observed by 31P{1H} spectroscopy. The boron-containing products were identified as the known B-methoxy-9-BBN and a borate ester. The reaction of

HBpin led to the quantitative formation of 3b and HCOOBpin within 15 min.

Attempted Stoichiometric Reaction of 9b with PhSiH3. A screw-cap NMR tube was charged with 9b (12.5 mg, 25 μmol), PhSiH3 (3.2 μL, 25 μmol), and 0.6 mL of C6D6. The progress of the reaction was monitored by 1H and 31P{1H} NMR spectroscopy. No appreciable reaction was observed within 16 h.

Attempted Catalytic Hydrosilylation of CO2 with 3b. Under an argon atmosphere, a flame-dried 50 mL Schlenk flask was charged with 3b (12.5 mg, 25 μmol), PhSiH3 (953 μL, 7.50 mmol), and 2 mL of C6D6. The mixture was degassed by a freeze-pump-thaw cycle and placed under 1 atm of CO2. After 24 h, a small portion of the clear liquid layer (ca. 0.6 mL) was transferred into a J. Young NMR tube under an argon atmosphere. The 1H NMR spectrum of the solution showed the unreacted PhSiH3 with no evidence of forming a methanol derivative.

c Phosphine-borane adduct ( Pe)2PH•(9-BBN) generated in situ. The NMR sample was

c 1 prepared by mixing a 2 : 1 ratio of ( Pe)2PH and 9-BBN dimer in ~0.6 mL of toluene-d6. H

1 3 NMR (400 MHz, C7D8, δ): 1.23-2.32 (m, CH and CH2, 32H), 4.14 (ddt, JP-H = 323.6 Hz, JH-H =

31 1 15.2 Hz, JH-H = 3.6 Hz, P–H, 1H); BH was not located. P{ H} NMR (162 MHz, C7D8, δ):

11 6.80 (s). B NMR (128 MHz, C7D8, δ): 18.1 (s).

Release of BH3 from 10a-c under the vacuum. A solid sample of a nickel borohydride complex (25 mol) in a J. Young NMR tube was heated at 60 oC under a dynamic vacuum for

24 h. The residue was dissolved in ca. 0.6 mL of benzene-d6 for NMR analysis. The percentage conversion for each reaction was calculated by comparing the integration of the produced nickel hydride species with that of unreacted borohydride complex from the 31P NMR spectrum.

Reactions of 10a-c with triethylamine. A screw cap NMR tube was charged with a solution of a nickel borohydride complex (12 mol) in 0.6 mL of benzene-d6. Triethylamine

(1.7 L, 12 mol) was added via a microliter syringe and the NMR tube was heated by a 60 ºC oil bath. The progress of the reaction was monitored by 1H and 31P{1H} NMR spectroscopy.

The percentage conversion for each reaction was calculated by comparing the integration of the produced nickel hydride species with that of unreacted borohydride complex from the 31P NMR spectrum.

Reactions of 10a-c with CO2. In a glove box, a nickel borohydride complex (25 mol) and 0.6 mL of benzene-d6 were transferred into a J. Young NMR tube. The resulting solution was degassed by performing three freeze-pump-thaw cycles and then placed under 1 atm of CO2.

The NMR tube was heated by a 60 ºC oil bath, and the progress of the reaction was monitored by

1H and 31P{1H} NMR spectroscopy. The percentage conversion for each reaction was calculated by comparing the integration of the produced nickel formate complex with that of unreacted borohydride complex from the 31P NMR spectrum.

Reaction of 11c with triethylamine. In a J. Young NMR tube, complex 11c was generated in situ by mixing 3c (10.0 mg, 19.8 mol) and 9-BBN dimer (2.4 mg, 9.9 mol) in 0.6 98 mL of benzene-d6 at room temperature for 1 h. To this solution, triethylamine (2.8 L, 19.7

mol) was added via a microliter syringe. A complete conversion to 3c was accomplished within 30 min, as confirmed by 1H and 31P NMR spectroscopy.

Reaction of 11c with CO2. In a J. Young NMR tube, complex 11c was generated in situ by mixing 3c (10.0 mg, 19.8 mol) and 9-BBN dimer (2.4 mg, 9.9 mol) in 0.6 mL of benzene- d6 at room temperature for 1 h. The resulting solution was degassed by performing three freeze- pump-thaw cycles and then placed under 1 atm of CO2. The progress of the reaction was monitored by 1H and 31P NMR spectroscopy. After 8 h, 9c was the only nickel species observed by 31P{1H} spectroscopy. The boron-containing products were identified as the known B- methoxy-9-BBN52 and a borate ester.53

Catalytic Reduction of CO2 with Boranes (BH3•THF, 9-BBN, and HBcat). Under an argon atmosphere, a flame-dried 50 mL Schlenk flask was charged with a nickel hydride complex (25 mol), a borane (272 μL, 2.50 mmol), and 2 mL of C6D6. To this solution, hexamethylbenzene (10.0 mg, 62.5 μmol) was added as an internal standard except for the reactions involving 3c. To avoid the overlap of proton resonances between hexamethylbenzene and 3c, hexamethyldisilane was used as the internal standard. The mixture was degassed by a freeze-pump-thaw cycle and placed under 1 atm of CO2. At different time intervals, a small portion of the clear liquid layer (ca. 0.6 mL) was withdrawn from the flask and transferred into a

J. Young NMR tube under an argon atmosphere. From the 1H NMR spectrum of the aliquot,

52 Kramer, G. W.; Brown, H. C. J. Organomet. Chem. 1974, 73, 1. 53 Köster, R.; Tsay, Y.-H.; Krüger, C.; Serwatowski, J.; Chem. Ber. 1986, 119, 1174.

99 turnover number (TON) was calculated by comparing the integration of CH3OBR'2 methyl resonance (near 3.3 ppm) with that of the internal standard (2.12 ppm).

X-ray Structure Determinations. X-ray quality single crystals of 9a and 9c were obtained by cooling a saturated pentane solution of the complex to 30 oC whereas suitable crystals of 9b were grown by slow evaporation of a concentrated solution of the complex in

C6D6. Two independent molecules of were located in the crystal lattice of 9c. Single crystals of

10a-c were grown from pentane at 30 oC. Single crystals of 11c were grown from toluene- pentane at 30 oC. Crystal data collection and refinement parameters are summarized in Tables

4 and 5. Intensity data were collected at 150K on a Bruker SMART6000 CCD diffractometer using graphite-monochromated Cu Kα radiation, λ = 1.54178Å. The data frames were processed using the program SAINT. The data were corrected for decay, Lorentz, and polarization effects as well as absorption and beam corrections based on the multi-scan technique. The structures were solved by a combination of direct methods in SHELXTL and the difference Fourier technique and refined by full-matrix least-squares procedures. Non-hydrogen atoms were refined with anisotropic displacement parameters. H-atoms attached to the boron were located directly from the difference map and their positions refined. All remaining H-atom positions were calculated and treated with a riding model. None of the structures has solvent present in the crystalline lattice. The cyclopentyl rings in 10c and 11c show typical disorder; a two-component model is presented for C24/C25 in 10c (major occupancy is 55%) and C25 in 11c (major occupancy is 70%). Compound 10a crystallized as two independent molecules in the crystalline lattice.

100

Table 4. Crystal Data Collection and Refinement Parameters of 9a-c

9a 9b 9c

empirical formula C19H32O4P2Ni C23H40O4P2Ni C27H40O4P2Ni crystal system triclinic triclinic triclinic space group P-1 P-1 P-1 a, Å 8.1207(2) 8.3696(2) 8.6624(2) b, Å 10.0328(2) 11.8615(3) 14.8388(2) c, Å 13.7821(3) 13.5708(3) 20.8500(3) , deg 99.432(1) 100.375(1) 83.670(10) , deg 99.873(1) 97.124(1) 87.601(1) , deg 93.002(1) 104.482(1) 89.795(1) Volume, Å3 1087.54(4) 1262.74(5) 2661.38(8) Z 2 2 4 no. of data collected 9221 10790 22268 no. of unique data, Rint 3743, 0.0256 4319, 0.0312 9101, 0.0536 R1, wR2 (I > 2(I)) 0.0352, 0.0894 0.0396, 0.1014 0.0518, 0.1227 R1, wR2 (all data) 0.0447, 0.0954 0.0509, 0.1087 0.0852, 0.1394

Table 5. Crystal Data Collection and Refinement Parameters of 10a-c and 11c

10a 10b 10c 11c empirical formula C18H35O2BP2Ni C22H43O2BP2Ni C26H43O2BP2Ni C34H55O2BP2Ni crystal system monoclinic triclinic triclinic monoclinic space group P21/c P-1 P-1 P21/c a, Å 12.9681(4) 8.2870(2) 8.7254(1) 15.7800(7) b, Å 23.5433(8) 12.1490(3) 10.5060(2) 13.0913(6) c, Å 14.6846(5) 13.4267(3) 14.5596(2) 17.1942(8) , deg 90 100.274(1) 81.742(1) 90 , deg 100.352(1) 95.653(1) 86.572(1) 112.518(2) , deg 90 105.074(2) 83.280(1) 90 Volume, Å3 4410.4(3) 1269.32(5) 1310.50(3) 3281.2(3) Z 8 2 2 4 no. of data collected 33917 11009 11353 22642 no. of unique data, Rint 6774, 0.0987 4390, 0.0327 4546, 0.0244 5826, 0.0571 R1, wR2 (I > 2(I)) 0.0498, 0.1124 0.0410, 0.1045 0.0389, 0.0999 0.0470, 0.1179 R1, wR2 (all data) 0.0844, 0.1286 0.0522, 0.1119 0.0452, 0.1043 0.0728, 0.1291

101

Chapter 4 Catalytic Decomposition of Formic Acid to Release Dihydrogen

102

4.1 Introduction

Dihydrogen represents an important alternative energy feedstock for both environmental and economic reasons, and when combined with fuel-cell technology, very efficient energy conversion can be achieved.1 Although there are several advantages of using dihydrogen over fossil fuels, the actual use of hydrogen as a transportation fuel is limited mainly because of storage and delivery problems. Conventional hydrogen-storage methods, such as high-pressure gas containers and cryogenic liquid/gas containers, have weight and safety issues.2

Consequently, a great deal of research is being performed to develop new materials, such as metal hydrides2,3 and carbon nanostructures4 that have the potentials to store and release hydrogen effectively. Among the different hydrogen-storage materials, formic acid (4.4 wt% of hydrogen), as well as metal formates, have recently received considerable attention.5 Formic acid is one of the major products formed in biomass processing such as fermentation, pyrolysis and can undergo selective decomposition to release dihydrogen and carbon dioxide only in the presence of a catalyst.6 In general, formic acid and formates can be decomposed via two pathways: (a) dehydrogenation and (b) dehydration (Scheme 1). The dehydration pathway generates CO, which can poison fuel cells. Therefore, this pathway must be suppressed for the subsequent conversion of hydrogen into electrical energy.

1 Steele, B. C. H.; Heinzel, A. Nature 2001, 414, 345. 2 Züttel, A.; Schlapbach, L. Nature 2001, 414, 353. 3 (a) Filinchuk, Y.; Chernyshov, D.; Nevidomskyy, A.; Dmitriev, V. Angew. Chem. Int. Ed. 2008, 47, 529. (b) SchJth, F.; Bogdanović, B.; Felderhoff, M. Chem. Commun. 2004, 2249. 4 (a) Liu, C.; Cheng, H.-M.; J. Phys. D 2005, 38, R231; b) Binger, U.; Züttel, W.; Appl. Phys. A 2001, 72, 147. 5 (a) Enthaler, S. ChemSusChem 2008, 1, 801. (b) Joó, F. ChemSusChem 2008, 1, 805. 6 (a) Reunanen, J.; Oinas, P.; Nissinen, T. A process for recovery of formic acid. PCT Int. Appl., 2009. (b) Kruse, A.; Gawlik, A. Ind. Eng. Chem. Res. 2003, 42, 267. (c) Hayes, D. J.; Fitzpatrick, S.; Hayes, M. H. B.; Ross, J. R. H. in Biorefineries-Industrial Processes and Products; Kamm, B.; Gruber, P. R., Kamm, M., Ed.; Wiley-VCH, Weinheim, 2006; p 139.

103

Several heterogeneous7 and homogeneous8,9 catalysts (Figure 1) have been reported for the release of hydrogen from formic acid. Puddephat and coworkers have investigated the binuclear ruthenium phosphine complex [Ru2(µ-CO)(CO)4(dppm)2] (dppm =

10 diphenylphosphinomethane) for selective hydrogen generation from HCO2H. Fukuzimi and coworkers have used [Rh(Cp*)(bipy)(H2O)][SO4] (bipy = bipyridine) for the hydrogen production from aqueous formic acid solutions. They have demonstrated that the decomposition

11 of HCO2H occurs via metal formate and hydride intermediates. Recently, this group has shown that heterobinuclear Ir-Ru complexes are highly efficient catalysts for hydrogen release in an aqueous solution under ambient conditions.11b Himeda and coworkers have unveiled iridium complexes for hydrogen generation from formic acid/sodium formate in aqueous solution.12

Beller and Laurenczy have independently demonstrated that H2 production is also possible under

7 (a) Rienäcker, G.; Mueller, H. Z. Anorg. Allg. Chem. 1968, 357, 255. (b) Garcia-Verdugo, E.; Liu, Z.; Ramirez, E.; Garcia-Serna, J.; Fraga-Dubreuil, J.; Hyde, J. R.; Hamley, P. A.; Poliakoff, M. Green Chem. 2006, 8, 359. (c) Hyde, J. R.; Poliakoff, M. Chem. Commun. 2004, 1482. (d) Hyde, J. R.; Walsh, B.; Singh, J.; Poliakoff, M. Green Chem. 2005, 7, 357. (e) Wiener, H.; Sasson, Y.; Blum, J. J. Mol. Catal. 1986, 35, 277. (f) Zhou, X.; Huang, Y.; Xing, W.; Liu, C.; Liao, J.; Lu, T. Chem. Commun. 2008, 3540. (g) Ojeda, M.; Iglesia, E. Angew. Chem. Int. Ed. 2009, 48, 4800. (h) Kilic¸, E. Ö ; Koparal, A. S.; Ögütveren, Ü . B. Fuel Proc. Technol. 2009, 90, 158. (i) Sun, B.; Smirniotis, P. G. Catal. Today 2003, 88, 49. (j) Kakuta, S.; Toshiyuki, A. Appl. Mater. Interfaces 2009, 1, 2707. 8 (a) Johnson, C. T.; Morris, D. J.; Wills, M. Chem. Soc. Rev. 2010, 39, 81. (b) Loges, B.; Boddien, A.; Gartner, F.; Junge, H.; Beller, M. Top. Catal. 2010, 53, 902. 9 (a) Coffey, R. S. Chem. Commun. 1967, 18, 923. (b) Forster, D.; Beck, G. R. Chem. Commun. 1971, 994, 1072. (c) Laine, R. M.; Rinker, R. G.; Ford, P. C. J. Am. Chem. Soc. 1977, 99, 252. (d) Yoshida, T.; Ueda, Y.; Otsuka, S. J. Am. Chem. Soc. 1978, 100, 3941. (e) Strauss, S. H.; Whitmire, K. H.; Shriver, D. F. J. Organomet. Chem. 1979, 174, C59. (f) Paonessa, R. S.; Trogler, W. C. J. Am. Chem. Soc. 1982, 104, 3529. (g) King, R. B.; Bhattacharyya, N. K. Inorg. Chim. Acta 1995, 237, 65. (h) Man, M. L.; Zhou, Z.; Ng, S. M.; Lau, C. P. Dalton Trans. 2003, 3727. 10 (a) Gao, Y.; Kuncheria, J.; Yap, G. P. A.; Puddephatt, R. J. Chem. Commun. 1998, 2365. (b) Gao, Y.; Kuncheria, J. K.; Jenkins, H. A.; Puddephatt, R. J.; Yap, G. P. A. J. Chem. Soc., Dalton Trans. 2000, 3212. (c) Shin, J. H.; Churchill, D. G.; Parkin, G. J. Organomet. Chem. 2002, 642, 9. 11 (a) Fukuzumi, S.; Kobayashi, T.; Suenobu, T. ChemSusChem 2008, 1, 827. (b) Fukuzumi, S.; Kobayashi, T.; Suenobu, T. J. Am. Chem. Soc. 2010, 132, 1496. 12 Himeda, Y. Green Chem. 2009, 11, 2018.

104 milder reaction conditions using ruthenium phosphine complexes.13,14 These catalytic systems require the use of amines as additives and the catalyst activity is largely dependent on the nature and concentration of amine.15 The Beller group has shown that a combination of N,N- dimethylhexylamine, [RuCl2(benzene)2]2, and 1,2-bis(diphenylphosphino)ethane (dppe) is a highly active catalyst that operates at room temperature with a TOF of 900 h-1.16 Most remarkably, when this reaction is carried out continuously, an overall turnover number (TON) of

260,000 can be achieved. The same research group has also demonstrated that ruthenium phosphine catalyst systems can be triggered by irradiation with visible light.17

Figure 1. Selected Examples of Homogeneous Catalysts for the Decomposition of Formic acid

13 (a) Fellay, C.; Yan, N.; Dyson, P. J.; Laurenczy, G. Chem. Eur. J. 2009, 15, 3752. (b) Gan, W.; Dyson, P. J.; Laurenczy, G. React. Kinet. Catal. Lett. 2009, 89, 205. (c) Loges, B.; Boddien, A.; Junge, H.; Beller, M. Angew. Chem. Int. Ed. 2008, 47, 3962. (d) Fellay, C.; Dyson, P. J.; Laurenczy, G. Angew. Chem. Int. Ed. 2008, 47, 3966. 14 Boddien, A.; Loges, B.; Junge, H.; Beller, M. ChemSusChem 2008, 1, 751. 15 Junge, H.; Boddien, A.; Capitta, F.; Loges, B.; Noyes, J. R.; Gladiali, S.; Beller, M. Tetrahedron Lett. 2009, 50, 1603. 16 Boddien, A.; Loges, B.; Junge, H.; Gärtner, F.; Noyes, J. R.; Beller, M. Adv. Synth. Catal. 2009, 351, 2517. 17 Loges, B.; Boddien, A.; Noyes, J. R.; Baumann, W.; Junge, H.; Beller, M. Chem. Commun. 2009, 28, 4185.

105

One of the long-term goals in catalysis is the replacement of precious metal-based catalysts with those derived from nonprecious metals including iron, nickel, and copper while maintaining similar or achieving better level of efficiency. Thus far, however, very few inexpensive metal-based catalysts have been reported for the dehydrogenation of formic acid.

Very recently, the Beller group has reported the first iron-catalysts capable of generating hydrogen from formic acid under ambient conditions.18 Based on IR and DFT studies, they have

- proposed [HFe(CO)3(PPh3)] as the active iron catalyst for the dehydrogenation of formic acid.

They have observed that the addition of terpyridine or phenanthroline ligand prevents the rapid dissociation of the active catalyst by formation of temporarily stabilized iron complexes. The generation of the active catalyst from the precatalyst is photo-triggered. They have also developed a remarkably active catalytic system involving Fe(BF4)2.6H2O catalyst for the same purpose.19 In the presence of propylene carbonate that is an environmentally benign reagent, with no further additive or base, TOFs up to 9425 h-1 and a TON of more than 92,000 have been achieved at 80oC. This is the most active catalyst for the dehydrogenation of formic acid to date.

Based on in-situ NMR investigation, kinetics studies, and DFT calculations, they have proposed a reaction pathway as outlined in Scheme 2. In this mechanism, iron hydride and iron formate complexes have been suggested as key intermediates.

18 Boddien, A.; Loges, B.; Gartner, F.; Torborg, C.; Fumino, K.; Junge, H.; Ludwig, R.; Beller, M. J. Am. Chem. Soc. 2010, 132, 8924. 19 Boddien, A.; Mellman, D.; Gartner, F.; Jackstell, R.; Junge, H.; Dyson, P. J.; Laurenczy, G.; Ludwig, R.; Beller, M. Science 2011, 333, 1733.

106

Inspired by these studies, I hypothesized that having both the nickel hydride20 and nickel formate21,22 characterized and knowing they can be interconverted, I could verify their potential involvement in catalytic dehydrogenation of formic acid. Therefore, I intended to investigate the possibility of catalytic decomposition of formic acid with nickel hydride complexes.

20 Chakraborty, S.; Krause, J. A.; Guan, H. Organometallics 2009, 28, 582. 21 Chakraborty, S.; Zhang, J.; Krause, J. A.; Guan, H. J. Am. Chem. Soc. 2010, 132, 8872. 22 Chakraborty, S.; Patel, Y. J.; Krause, J. A.; Guan, H. Polyhedron 2012, 32, 30.

107

4.2 Stoichiometric Steps Related to Ni-Catalyzed Dehydrogenation of HCO2H

When nickel pincer hydride complexes 3a-c were reacted with one equivalent of formic acid, rapid generation of nickel formate complexes 9a-c was observed with the concomitant release of hydrogen gas (eq 1). As shown in Chapter 3, it was demonstrated that nickel hydrides

3a-c can be reformed from the nickel formates 9a-c when heated under a dynamic vacuum (eq

2).

If these two stoichiometric steps are combined, a potential catalytic cycle for the dehydrogenation of formic acid can be envisioned (Scheme 3). If the reaction follows this proposed cycle, the benefits of our catalytic system would be three-fold: (i) unlike other catalytic systems no base or additives is required, (ii) air-stable nickel formate complexes can be used as catalysts, which will give a much better handle to understand the reaction mechanism, and (iii) catalysts based on inexpensive metal will be employed for the dehydrogenation of formic acid.

108

4.3 Catalytic Decomposition of HCO2H

When the decomposition of formic acid was performed in toluene-d8 using 9a-c as

1 13 1 catalysts (10 mol%), both H2 and CO2 were detected in the solution by H and C{ H} NMR spectroscopy. In the 1H NMR, dihydrogen resonance was located at 4.46 ppm as a singlet and in

13 1 C{ H} NMR the resonance for CO2 was observed at 125.3 ppm. Complete consumption of formic acid dissolved in a sealed NMR tube required 7-12 hours depending on different catalysts

(Scheme 4, Table 1). However, when same reactions were performed in an open vessel (exposed to air), decomposition of formic acid was finished in much shorter period of time (Scheme 5,

Table 2). The difference in catalytic activity of closed and open systems can be rationalized if the insertion of CO2 to nickel hydrides, the reverse step of CO2 elimination, is taken into account. From deinsertion point of view, it is thermodynamically uphill; however, the overall decomposition of HCO2H is driven by the protonation of the nickel hydride intermediate. In a closed system, the buildup of CO2 would make the insertion of CO2 competitive with the protonation step.

109

Table 1. Comparing Catalytic Activity of 9a-c in the Closed System

Table 2. Comparing Catalytic Activity of 9a-c in the Open System

110

4.4 Kinetic Studies of Catalytic Decomposition of HCO2H

The protonation of basic Ni-H moiety with formic acid (step 1, Scheme 3) is a fast reaction at room temperature. On the other hand, the elimination of CO2 from nickel formate complexes (step 2, Scheme 3) occurs rather slowly,23 indicating this step to be the rate-limiting step during the catalysis. In order to gain better understanding of the overall kinetics, disappearance of formic acid was monitored as a function of time by in-situ IR spectroscopy.

When 5 mol% of 9b was employed as the catalyst in toluene, the dehydrogenation process followed a sigmoidal decay pattern (Figure 2). Catalysts 9a and 9c gave similar results. This is unique to our system as exponential decay has been commonly observed for other HCO2H- dehydrogenation systems. Sigmoidal kinetics often suggests one of the following mechanistic possibilities: (a) heterogeneity, (b) metal-metal cooperativity, and (c) formation of other more reactive species during catalysis. The following sections describe how I explored these mechanistic scenarios.

23 Previous group member has determined the rates of this deinsertion step in the range of 10-4 - 10-3 s-1 at 60 oC. Zhang, J.; Guan, H. Unpublished work

111

0.12

0.1

0.08

0.06 Absorbance

0.04

0.02

0 0 50 100 150 200 250 300 Time (Min)

Figure 2. Kinetics of Ni-Catalyzed Dehydrogenation of HCO2H (the Absorbance is for the C=O

-1 Stretching of HCO2H at 1726 cm )

4.4.1 Heterogeneity

A possibility of generating heterogeneous Ni nanoparticles under the acidic reaction conditions was first investigated by mercury poisoning experiment (Scheme 6). If nanoparticles are produced, they will form amalgam with Hg, thereby slowing down the catalysis. To test this hypothesis, 300 equivalents of elemental mercury (with respect to Ni) were added to a mixture of

9b and formic acid at 80oC and the kinetics was monitored by IR spectroscopy. The required time to complete the decomposition of HCO2H was found to be the same as compared to the one without added Hg. This result strongly supports the non-heterogeneous nature of this catalytic system.

112

4.4.2 Metal-Metal Cooperativity

Cooperativity between two metal centers often leads to sigmoidal kinetic behavior and they are commonly observed for enzymatic reactions in biological processes. In our case, a possible cooperation might exist between the two nickel species, nickel hydride and nickel formate, both of which are involved in the catalytic process. To test this possibility, equimolar

1 31 1 amounts of 3b and 9b were mixed in C6D6 and their reaction was monitored by H and P{ H}

NMR spectroscopy. However, no net reaction between these two species was observed either at room temperature or at 80oC, suggesting no cooperativity between them (Scheme 7). When 3b-

D was mixed with unlabelled 9b, H/D exchange was not observed as judged from 1H and

31P{1H}NMR (Scheme 8), which further argues against the cooperative mechanism.

113

4.4.3 Formation of Other Reactive Intermediates

The sudden increase in the catalytic reactivity could also be explained by proposing the formation of intermediates that are more reactive than the starting nickel formate complex.

Figure 3 shows some possible structures of these intermediates. In order to test this hypothesis, successive-addition experiment with formic acid was performed. If these intermediate species are indeed responsible for the sigmoidal kinetics and their concentration in the solution remain high toward the end of the dehydrogenation, a different decay profile is anticipated when more formic acid is added to the reaction vessel. However, similar sigmoid kinetic trace was obtained when the second batch of formic acid was injected to the reaction system (Figure 4). This outcome suggests that after first run either the intermediate species is converted back to the starting nickel formate complex or no additional reactive species is generated at all.

Figure 3. Structures of Potential Reactive Intermediates

114

0.08

0.07

0.06

0.05

0.04 Absorbance 0.03

0.02

0.01

0 0 20 40 60 80 100 120 140 160 180 200 Time (Min)

Figure 4. Successive-Addition Kinetics of Ni-Catalyzed Dehydrogenation of HCO2H (the

-1 Absorbance is for the C=O Stretching of HCO2H at 1726 cm )

4.4.4 Dual Processes Catalyzed by Nickel Hydrides

If nickel hydrides catalyze the decomposition of formic acid via more than one pathway and if these pathways have comparable rates, unique kinetic behavior might result for the overall process. In addition to the mechanism proposed in Scheme 3, nickel hydride might catalyze the decomposition of formic acid via a separate pathway that does not involve nickel formate species

(Scheme 9, Process II). In this mechanism the release of H2 and CO2 from HCO2H occurs in a single step, which is facilitated by the nickel hydride. As both nickel hydride and formic acid are

115 involved in this mechanism, an exponential decay is expected for this pathway with steady-state concentration of the nickel hydride. On the other hand, a linear decay is anticipated for the stepwise mechanism (Process I) involving nickel formate, considering fast protonation step and slow CO2 deinsertion. By first approximation, the sigmoidal pattern shown in Figure 2 can be viewed as a nearly linear decay for the first 120 minutes, followed by an exponential decay.

The reaction between 3b and DCOOH in a 2 : 1 ratio would distinguish between Process

I and the dual process (Scheme 10). If only the catalytic cycle in Process I operates, we expect to have 3b and 9b-D in 1 : 1 ratio. However, if both cycles shown in Scheme 9 operate simultaneously, the products should contain 3b-D in addition to 3b and 9b-D. This labeling experiment should be carried out in the future to verify this mechanism.

116

4.5 Conclusions

Both nickel hydrides and nickel formates catalyze the dehydrogenation of formic acid to release H2 and CO2. The simplified mechanism of this process consists of two steps: (a) protonation of nickel hydrides by HCO2H to form nickel formate species and (b) deinsertion of

CO2 from nickel formates to regenerate the active nickel hydride complex. Both stoichiometric steps have been investigated. Kinetic studies of the catalytic reaction have revealed a sigmoidal decay of the formic acid. Different mechanistic possibilities have been explored experimentally in order to understand the unique kinetic behavior. Heterogeneous and cooperative mechanisms have been ruled out. Successive-addition experiment has shown a similar kinetic trace after the formic acid is replenished, which argues against the buildup of more active species during the catalysis.

117

4.6 Experimental Section

General Comments. Unless otherwise noted, all the organometallic compounds were prepared and handled under an argon atmosphere using standard Schlenk and inert-atmosphere box techniques. Dry and oxygen-free toluene was collected from an Innovative Technology solvent purification system and used throughout the experiments. C6D6 and toluene-d8 were distilled from Na and benzophenone under an argon atmosphere. HCO2H (90%) were purchased from Sigma-Aldrich and used without further purification. 1H and 31P{1H} NMR spectra were recorded on a Bruker Avance-400MHz spectrometer and 2H NMR was recorded in a Bruker

Avance-500MHz instrument. Chemical shift values in 1H spectra were referenced internally to

31 the residual solvent resonances. P spectra were referenced externally to 85% H3PO4 (0 ppm).

Nickel hydrides 3a-c20,22 were prepared according to the literature procedure.

i Synthesis of [2,6-( Pr2PO)2C6H3]NiOC(O)H (9a). Under an argon atmosphere, HCOOH (90 wt% in H2O, 40 μL, 0.94 mmol) was added dropwise to the solution of

3a (200 mg, 0.50 mmol) in 10 mL of toluene. After the hydrogen evolution ceased (within minutes), the resulting mixture was placed under vacuum for 1 h and the product was isolated as a bright yellow solid (155 mg, 75% yield). The characterization data are the same as those in

Chapter 3; there this compound was prepared from the insertion of CO2 into 3a.

118

t Synthesis of [2,6-( Bu2PO)2C6H3]NiOC(O)H (9b). Under an argon atmosphere, HCOOH (90 wt% in H2O, 35 μL, 0.82 mmol) was added dropwise to the solution of

3b (200 mg, 0.44 mmol) in 10 mL of toluene. After the hydrogen evolution ceased (within minutes), the resulting mixture was placed under vacuum for 1 h and the product was isolated as a bright yellow solid (160 mg, 72% yield). The characterization data are the same as those in

Chapter 3; there this compound was prepared from the insertion of CO2 into 3b.

c Synthesis of [2,6-( Pe2PO)2C6H3]NiOC(O)H (9c). Under an argon atmosphere, HCOOH (90 wt% in H2O, 35 μL, 0.82 mmol) was added dropwise to the solution of

3c (200 mg, 0.44 mmol) in 10 mL of toluene. After the hydrogen evolution ceased (within minutes), the resulting mixture was placed under vacuum for 1 h and the product was isolated as a bright yellow solid (150 mg, 77% yield). The characterization data are the same as those in

Chapter 3; there this compound was prepared from the insertion of CO2 into 3c.

General Procedure for the Regeneration of 3a-c from 9a-c. A solid sample of a nickel formate complex (25 mol) in a J. Young NMR tube was heated at 60oC under a dynamic vacuum for 5 h. The residue was dissolved in ca. 0.6 mL of C6D6 for NMR analysis. The percentage conversion for each reaction was calculated by comparing the integration of the produced nickel hydride species with that of unreacted formate complex from the 31P{1H} NMR spectrum.

119

General Procedure for the Catalytic Decomposition of Formic Acid. Inside a J.

Young NMR tube, a nickel formate complex (25 mol), HCO2H (9 L, 250 mol) were

o dissolved in ca.0.6 mL of toluene-d8 and heated the mixture to 80 C using an oil-bath. The disappearance of the resonances related to formic acid was monitored by 1H NMR spectroscopy at different time intervals.

General Procedure for Kinetic Studies. A V-shaped two-necked 10 mL flask fitted with a condenser was charged with nickel formate (25 mol), HCO2H (18 L, 500 mol), and 4 mL of toluene. The IR probe was connected to the other neck and the flask was placed in an oil- bath preheated to 80oC. The kinetic plot was obtained by monitoring the disappearance of the

C=O stretching frequency of formic acid (1726 cm-1) as a function of time.

Mercury Poisoning Experiment. A V-shaped two-necked 10 mL flask fitted with a condenser was charged with nickel formate (25 mol), HCO2H (18 L, 500 mol), elemental mercury (1.5 g, 7.5 mmol) and 4 mL of toluene. The IR probe was connected to the other neck and the flask was placed in an oil-bath preheated to 80oC. The kinetic plot was obtained by monitoring the disappearance of the C=O stretching frequency of formic acid (1726 cm-1) as a function of time.

Stoichiometric Reaction of 3b with 9b. Inside a J. Young NMR tube, 3b (11 mg, 25

mol) and 9b (12 mg, 25 mol) were dissolved in ca. 0.6 mL of C6D6 and the reaction was monitored first at room temperature and then at 80oC by 1H and 31P{1H} NMR spectroscopy.

Stoichiometric Reaction of 3b-D with 9b. Inside a J. Young NMR tube, 3b-D (11 mg,

25 mol) and 9b (12 mg, 25 mol) were dissolved in ca. 0.6 mL of C6D6 and the reaction was monitored first at room temperature and then at 80oC by 1H and 31P{1H} NMR spectroscopy.

120

Successive Addition Experiment. A V-shaped two-necked 10 mL flask fitted with a condenser was charged with nickel formate (25 mol), HCO2H (18 L, 500 mol), and 4 mL of toluene. The IR probe was connected to the other neck and the flask was placed in an oil-bath preheated to 80oC. The kinetic plot was obtained by monitoring the disappearance of the C=O stretching frequency of formic acid (1726 cm-1) as a function of time. After consumption of all the HCO2H, the system was recharged with the same amount of formic acid (18 L) and kinetics was monitored again for the second run.

121

Chapter 5 Mechanistic Insight of Nickel-Catalyzed Olefin Isomerization Reaction

122

5.1 Introduction

Olefins are an important class of compounds in organic synthesis and indispensable precursors for the industrial synthesis of polymers.1,2 In target-directed synthesis, they can be readily transformed into molecules with a wide variety of other functionalities, and they serve as substrates for the formation of new C-C double bonds via metathesis reactions.2,3 Despite the fact that many synthetic methods have been established for the introduction of C-C double bonds into carbon chains, the methodologies are not always straightforward, and often lack generality for the substrates. One potentially useful way of positioning the unsaturation of a molecule is to migrate a pre-existing C-C double bond. Therefore, the controlled migration of a C=C bond along a carbon chain, or the isomerization of an olefin is of high importance.4

Depending upon the particular catalyst involved, isomerization reactions can proceed via either a metal hydride addition-elimination mechanism or by a mechanism involving a π-allyl metal hydride intermediate. One of the fundamental differences between the two mechanisms is that the metal hydride addition-elimination mechanism often leads to the exchange of hydrogens at all positions while the π-allyl metal hydride mechanism results in a selective 1,3 hydrogen

1 (a) Kricheldorf, H. R.; Nuyken, O.; Swift, G. In Handbook of Polymer Synthesis, 2nd ed.; Marcel Dekker: New York, 2005. 2 (a) Evans, D. A.; Trotter, B. W.; Cote, B.; Coleman, P. J.; Dias, L. C.; Tyler, A. N. Angew. Chem. Int. Ed. 1997, 36, 2744. (b) Evans, D. A.; Fitch, D. M.; Smith, T. E.; Cee, V. J. Am. Chem. Soc. 2000, 122, 10033. (c) Ley, S. V.; Brown, D. S.; Clase, J. A.; Fairbanks, A. J.; Lennon, I. C.; Osborn, H. M. I.; Stokes, E. S. E.; Wadsworth, D. J. J. Chem. Soc., Perkin Trans. 1 1998, 2259. (d) Evans, D. A.; Carter, P. H.; Carreira, E. M.; Charette, A. B.; Prunet, J. A.; Lautens, M. J. Am. Chem. Soc. 1999, 121, 7540. (e) Fleming, I.; Barbero, A.; Walter, D. Chem. Rev. 1997, 97, 2063. (f) Nicolaou, K. C.; He, Y.; Vourloumis, D.; Vallberg, H.; Roschangar, F.; Sarabia, F.; Ninkovic, S.; Yang, Z.; Trujillo, J. I. J. Am. Chem. Soc. 1997, 119, 7960. 3 (a) Larock, R. C. In Comprehensive Organic Transformations: A Guide to Functional Group Preparations, 2nd ed.; VCH: New York, 1999; (b) Mackenzie, K. In The Chemistry of Alkenes; Patai, S., Ed.; Wiley: New York, 1964. For a review on the Wittig reaction, see: (c) Maryanoff, B. E.; Reitz, A. B. Chem. Rev. 1989, 89, 863. For a review on olefin metathesis: (d) Grubbs, R. H. Tetrahedron 2004, 60, 7117. (e) Handbook of Metathesis; Grubbs, R. H., Ed.; Wiley-VCH: Weinheim, Germany, 2003; Vols. 1-3. 4 For reviews of transition metal catalysis of olefin isomerization, see: (a) Davies, N. R. Reu. Pure Appl. Chem., 1967, 17, 83. (b) Bird, C. W. “Transition Metal Intermediates in Organic Synthesis,” Academic Press, New York, N. Y., 1967, pp 67. (c) Orchin, M. Adran. Catal. Relur. Subj., 1966, 16, 1. Tolman C. A. in “Transition Metal Hydrides,” Vol. 1, E. L. Muetterties, Ed., Marcel Dekker, New York, N. Y.,1971, Chapter 6. 123 shift. For isomerization to occur via the metal hydride addition-elimination pathway, generally a cocatalyst such as hydrogen or an acid is necessary, presumably because the cocatalyst reacts with the metal complex to produce a metal hydride intermediate. However, if the isomerization is carried out with a metal hydride itself, these additives are not required. Some representative catalyst systems that isomerize olefins via the metal hydride addition-elimination mechanism are

5a 5b [(C2H4)2RhC1]2 and Ni[P(OEt)3]4, both of which require an acid as a cocatalyst, and

6 PtC12(Ph3P)2-SnCl2, which requires H2 as a cocatalyst. Isomerization via a π-allyl metal hydride pathway is relatively less common and one of the most well-known examples is the

7 olefin isomerization reactions catalyzed by Fe(CO)5.

I hypothesized that nickel pincer hydride complexes could be used as catalysts for olefin isomerization reactions. No additives such as acids or dihydrogen are needed in these reactions as metal-hydrogen bond is preinstalled in these molecules. For this study, I chose 1-butene as my olefin substrate because only two isomerized products (cis-2-butene and trans-2-butene) can be formed, which simplifies NMR analyses.

5.2 Catalytic Isomerization of 1-Butene to 2-Butene

When one equivalent of 3a was treated with nine equivalents of 1-butene (gaseous) and the mixture was heated to 60oC, migration of the terminal C=C bond of 1-butene to the internal position was observed (Scheme 1). After four days, 91% of 1-butene was converted to 2-butene.

The ratio of cis and trans isomers of 2-butene was calculated to be 29 : 62 from 1H NMR integrations. This ratio closely matches the thermodynamic ratio (1 : 2) of these two isomers.

5 (a) Cramer, R. J. Am. Chem, Soc. 1966, 88, 2272. (b) Tolman, C. A. J. Am. Chem. Soc. 1972, 94, 2994. 6 Adams, R. W.; Batley, G. E.; Bailar, Jr., J. C. J. Am. Chem. Soc. 1968, 90, 6051. 7 (a) Emerson G. F.; Pettit, R. J. Am. Chem. Soc. 1962, 84, 4591. (b) Hendrix, W. T.; Cowherd, F. G.; von Rosenberg, J. L. Chem. Commun. 1968, 97. 124

31P{1H}NMR spectroscopy suggested that 3a was the major nickel species after four days.

Nickel alkyl complexes 13a (7%) and 14a (9%) were also formed as minor species during this reaction.

Even though the aforementioned reaction is catalyzed by nickel hydride, it is too slow to have any practical use. However, understanding the reaction mechanism of this process can guide us to improve the catalytic efficiency. Therefore, I have focused my attention to the mechanistic study of this nickel-catalyzed olefin isomerization reaction. A number of reaction pathways have been considered. For the rest of this chapter, I will discuss these mechanisms in separate sections and provide experimental evidence to either rule out or support the particular mechanism.

5.3 Metal Hydride Addition-Elimination Pathway

An olefin can insert into an M-H bond in two different ways (Scheme 2): (i) 1,2-insertion which leads to a linear metal alkyl complex that is both kinetically and thermodynamically favorable and (ii) 2,1-insertion which generates a branched metal alkyl species. In order to shift a terminal C=C bond to an internal C=C bond via an addition-elimination pathway, one has to

125 propose the formation of a branched metal alkyl complex, followed by β-H elimination from this complex. On the other hand, β-H elimination from the linear isomer would result the starting terminal olefin and metal hydride.

To investigate this type of pathway in our nickel-catalyzed olefin isomerization reaction, both the linear and branched nickel butyl complexes were synthesized and their possibility to undergo β-H elimination was assessed. Compounds 13a and 14a were prepared by the reaction of corresponding nickel pincer chloride 2a with nBuLi and secBuLi, respectively (Scheme 3).

Both were characterized by 1H, 31P{1H}, 13C{1H} NMR spectroscopy and elemental analysis. X- ray quality single crystals for 14a were grown from a saturated solution of this compound in pentane at -30oC (Figure 1). Some of the resonances for alkyl groups of these complexes are overlapped in the 1H NMR spectra with the iPr resonances of the pincer ligand. In the 31P{1H}

NMR, singlet phosphorous resonances were observed for both 13a (188.03 ppm) and 14a

(184.45 ppm), indicating magnetically equivalent phosphorous atoms in the solution. The ipso carbon of alkyl moities in 13a and 14a appear at 0.56 ppm and 14.08 ppm respectively as triplets in the 13C{1H} NMR because of the coupling with two equivalent phosphorous atoms. To assign carbon resonances of butyl moieties in the 13C{1H} NMR spectra, both DEPT90 and

126

DEPT135 NMR experiments were carried out. Outcomes of these experiments further support the structural assignments for compound 13a and 14a.

i sec Figure 1. X-ray Crystal Structure of [2,6-( Pr2PO)2C6H3]Ni(Bu ) (14a) (50% probability level).

The secBu group shows disorder in this molecule. Selected bond lengths (Å) and angels (deg):

NiC1 1.916(3), NiC19 2.009(3), NiP1 2.1334(7), NiP2 2.1571(8), C1NiC19 175.36(13),

C1NiP1 81.62(8), C1NiP2 81.52(8), P1NiP2 163.13(3).

127

After successful isolation of the linear and branched isomers of nickel butyl complexes, their propensity to undergo β-H elimination was investigated. Complexes 13a and 14a were

o dissolved in toluene-d8 separately and these solutions were heated at 100 C for several days. No evidence of β-H elimination was observed as 3a was not detected at all when these reactions were monitored by 1H and 31P{1H} NMR spectroscopy (Scheme 4). The remarkable thermal stability of 14a against β-H elimination suggested that it is not likely to be the intermediate in the isomerization reaction. Although no elimination was observed, complex 14a isomerized to 13a.

After seven days, 30% of the branched isomer converted to the linear isomer as observed in the

31P{1H}NMR spectroscopy (vide infra).

One can, however, argue that β-H elimination in Scheme 4 is still possible, but the reverse step, the insertion of olefin, is more favorable thermodynamically. If it is indeed the case, complex 14a should also catalyze the olefin isomerization reaction. However, no isomerization was observed when the reaction was monitored for 4 days at 60oC (Scheme 5).

128

To further probe the possibility that 3a may be produced from 14a through β-H elimination, a set of experiments were conducted in the presence of a different alkene. For this purpose, 14a was reacted with ethylene (Scheme 6). An olefin exchange product would suggest the intermediacy of 3a in this reaction. However, experimental results showed no exchange between secondary butyl moiety of 14a and ethyl moiety of 15a. It was anticipated that ethylene being a smaller alkene, it would react with 3a faster than the more bulky 1-butene. In fact, when one equivalent of 3a was treated with a mixture of ethylene and 1-butene (3 : 10), selective formation of nickel ethyl complex 15a (17% conversion) was observed in the 31P{1H}

NMR spectrum (Scheme 7).

129

Therefore, the key observations: (i) branched nickel alkyl isomer (14a) does not undergo

β-H elimination, (ii) the same complex does not catalyze 1-butene isomerization, and (iii) it does not exchange olefin with ethylene, all suggest that olefin isomerization in this system does not occur via the commonly proposed addition-elimination pathway.

5.4 π-Allyl Metal Hydride Pathway

π-allyl metal hydride mechanism (Scheme 8) can also be used to explain olefin isomerization reactions although it is much less common than the metal hydride addition- elimination mechanism. The distinctive feature of the reaction under this specific mechanism involves a 1,3 hydrogen atom shift,8 whereas in case of addition-elimination pathway hydrogen atoms at all positions will be exchanged.

Although the π-allyl metal hydride mechanism has often been suggested to account for olefin isomerization by transition metal complexes without hydride ligands, there are in fact very few definitive cases in which this mechanism is operating. Harrod and Chalk explained the 1,3- hydrogen shifts in the isomerization of l-heptene-3-D2 catalyzed by bis-

(benzonitrile)dichloropalladium(II) by invoking a π-allyl palladium hydride intermediate.9 The isomerization of allyl alcohol to propionaldehyde catalyzed by iron pentacarbonyl is perhaps the

8 Casey, C. P.; Cyr, C. R. J. Am. Chem. Soc. 1973, 95, 2248. 9 Harrod J. F.; Chalk, A. J. J. Am. Chem. Soc. 1966, 88, 3491. 130 best known example of an isomerization which proceeds by a π-allyl metal hydride mechanism.7

The deuterium labeling studies by Hendrix, demonstrated a net 1,3 hydrogen shift in the conversion of allyl alcohol to propionaldehyde.7b

After ruling out the metal hydride addition-elimination pathway, I decided to investigate the possibility of π-allyl metal hydride type mechanism in our system. If 1,3-hydrogen shift takes place during the reaction, it can be detected by selectively labeling the olefin substrate with deuterium. Therefore, a reaction of 3a with 3,3,3-D3-propene was carried out in protio-benzene at 60oC (Scheme 9). No deuteration at the C2 carbon atom of propene would strongly support a

π-allyl type pathway. However, when the products of this reaction mixture was analyzed by 2H

NMR spectroscopy, all the carbon atoms of propene was found to contain deuterium atom. In addition to the organic products, small amount of 3a-D was also observed in the reaction mixture. Quantification of deuterium on each carbon atom of propene and analysis of percentage of different deuterium-containing species in the mixture should be performed in the future to obtain useful mechanistic information. In any case, deuterium incorporation at the C2 carbon atom of propene argues against the formation of π-allyl metal hydride intermediate.

131

5.5 Hydrogen Atom Transfer Pathway

Metal hydride complexes are capable of transferring hydrogen atoms to olefins. The

Norton group has demonstrated this type of reactions utilizing metal hydrides with a relatively

10a small M-H bond dissociation energy. They have shown that Cp(CO)3CrH and

10b (dppb)(CO)4VH, with weak Cr-H (61.5 kcal/mol) and V-H (54.9 kcal/mol) bonds, can promote radical cyclization reaction with α-vinyl esters (Scheme 10). It is interesting to note that these reactions are often accompanied with a minor isomerized side product. Even though H- atom transfer pathway has rarely been proposed for olefin isomerization reactions, it is not impossible in our nickel hydride-catalyzed olefin isomerization reaction.

To test this mechanistic hypothesis, catalytic reaction between 3a and 1-butene was performed in the presence of radical inhibitors. When one equivalent of (2,2,6,6- tetramethylpiperidin-1-yl) oxyl (TEMPO) was mixed with four equivalents of 3a and ten equivalents of 1-butene in C6D6, no olefin isomerization was observed (Scheme 11). This result seems to support the involvement of radical intermediates. Control experiment between

10 (a) Smith, D. M.; Pulling, M. E.; Norton, J. R. J. Am. Chem. Soc. 2007, 129, 770. (b) Choi, J.; Pulling, M. E.; Smith, D. M.; Norton, J. R. J. Am. Chem. Soc. 2008, 130, 4250.

132

TEMPO and 3a in a 1 : 1 ratio showed complete disappearance of 3a within 48 hours at 60oC.

However, complete loss of isomerization reactivity in the presence of more than stoichiometric amount of 3a (4 equivalents) most likely indicates H-atom transfer pathway in our case. Similar results were obtained when galvinoxyl was used as a radical inhibitor instead of TEMPO.

Radical clock reagents have often been used to verify radical pathways. When vinylcyclopropane, known to generate a radical clock when added an H-atom, was mixed with

1 3a in C6D6, H NMR spectra showed shift in the resonances of both reagents at room temperature (Scheme 12). However, no new resonance grew in when the mixture was heated to

60oC for several days, and all the resonances become significantly broadened. When the organic component was separated from the metallic residue by vacuum transfer, unreacted vinylcyclopropane was recovered. This result suggests no net reaction between these two species.

133

Nevertheless, based on the radical inhibitor experiments, an H-atom transfer pathway is proposed at this moment for the nickel-catalyzed olefin isomerization reaction (Scheme 13).

This mechanism is also consistent with the deuterium scrambling result shown in Scheme 9. It is likely that in the case of vinylcyclopropane, the H-atom transfer is negligible with the sterically hindered cylopropyl group.

134

5.6 Isomerization of Branched Nickel Alkyl Complex to Linear Nickel Alkyl Complex

It was mentioned in Section 5.3 of this chapter that when a C6D6 solution of 14a was placed at 60oC, 30% of this molecule isomerized to the linear isomer 13a in several days

(Scheme 14). A deeper understanding of the process is required in order to understand the olefin isomerization mechanism. When 13a was subjected to same reaction conditions, 14a did not form indicating 13a is thermodynamically more stable than 14a. One of the common pathways for this alkyl isomerization is to first undergo β-H elimination to form metal hydride and terminal olefin, followed by insertion of terminal olefin into the M-H bond in a 1,2-fashion

(Scheme 15). However, this is not likely to operate in our case because 14a does not undergo β-

H elimination as demonstrated in Scheme 4.

135

Next, a homolytic cleavage of Ni-C bond of 14a was considered. In this mechanism, after the homolytic cleavage, the secondary butyl radical can undergo a 1,3 hydrogen atom shift to produce a primary butyl radical that can combine with the nickel radical to produce 13a

(Scheme 16). However, if this pathway is operating, isomerization of olefin would be a competitive reaction as both of these two isomerization processes would go through the same intermediate (considering a radical pathway for olefin isomerization). Moreover, based on the fact that olefin isomerization is relatively faster than nickel alkyl isomerization, it was expected that if homolytic cleavage happens the resulting species would preferentially undergo alkene isomerization compared to 1,3 hydrogen atom shift. Therefore, this mechanism is inconsistent with the result shown in Scheme 5.

136

I then considered a concerted mechanism which can be viewed as an intramolecular σ- bond metathesis type process (Scheme 17). This is rather an unusual reaction pathway, but might operate in our system. Under this mechanism, a γ-carbon atom in the Ni-bound-alkyl species is a necessity. Therefore, this pathway can be indirectly verified by investigating the isomerization of a nickel alkyl complex without a γ-carbon atom. Guided by this hypothesis, I planned to synthesize nickel isopropyl complex 16a that lacks the required γ-carbon. 16a was prepared from a reaction of 2a with iPrMgCl (Scheme 18). This complex was characterized by multinuclear NMR techniques and elemental analysis. The linear isomer 17a was also prepared and characterized similarly.

When 16a was heated at 60oC for seven days, no isomerization to its linear isomer (17a) was observed (Scheme 18). If the presence of the γ-carbon is not required, it is anticipated that

16a should undergo alkyl isomerization more readily than 14a because the smaller iPr alkyl chain should favor potential transition states. On the other hand, no isomerization of 16a strongly suggested the importance of having a γ-carbon atom in the alkyl moiety for the transformation.

137

Another experiment could be designed to probe this concerted mechanism (Scheme 19).

For this purpose, I planned to synthesize compound 17a-13C(α) with the labeled carbon atom bound to the nickel center. A possible synthetic route for 17a-13C(α) is given in Scheme 20. If this concerted mechanism is occurring in this system, the carbon-labeling in 17a-13C(α) should exchange with the unlabeled γ-carbon atom in this molecule to form 17a-13C(γ). This experiment shall be performed in the future.

138

5.7 Conclusions

Isomerization of 1-butene to cis and trans-2-butene has been carried out using nickel pincer hydride catalyst. Even though the catalytic process is not efficient, investigation of the mechanism has revealed a unique reaction pathway. Nickel butyl complexes, the potential intermediates of the isomerization, have been synthesized and shown that they do not undergo β-

H elimination reaction. Branched nickel butyl complex does not catalyze olefin isomerization reaction. On the basis of these results, a metal-hydride addition-elimination pathway has been ruled out. Labeling studies have also argued against a π-allyl metal hydride type mechanism.

On the other hand, an H-atom transfer mechanism is a more plausible reaction pathway based on the fact that use of radical inhibitors completely stopped the isomerization process. In addition to the butene isomerization, branched nickel butyl complex also isomerizes to its linear analogue.

A unique intramolecular pathway has been proposed for this isomerization and importance of having a γ-carbon atom in the alkyl moiety has been demonstrated.

5.8 Experimental Section

31P{1H} NMR spectra were recorded on a Bruker Avance-400MHz spectrometer. 2H NMR was recorded on a Bruker Avance-500MHz spectrometer. Chemical shift values in 1H and 13C spectra were referenced internally to the residual solvent resonances. 31P spectra were referenced 139

11 11 externally to 85% H3PO4 (0 ppm). Compounds 2a and 3a were prepared according to the literature procedure.

i n Synthesis of [2,6-( Pr2PO)2C6H3]Ni(Bu ) (13a). Under an argon

atmosphere, a 1.6 M solution of nBuLi in hexane (0.29 mL, 0.46 mmol) was added slowly to a Schlenk flask containing a cold solution (-78oC) of 2a (100 mg, 0.23 mmol) in

20 mL of toluene. The color of the solution changed from orange yellow to light yellow upon gradually warming up to room temperature over a period of 15 minutes. The solution was subsequently stirred at room temperature for 4 hours and then the solvent was removed under vacuum. The resulting yellow residue was treated with pentane and then filtered under argon through a pad of Celite to obtain a clear yellow solution. Removal of the solvent under vacuum afforded a yellow powder, which was further purified by washing with methanol (1 mL × 2) and drying under vacuum (90 mg, 86% yield). The compound was stored inside glovebox. 1H NMR

n n (400 MHz, C6D6, δ): 0.93 (br, NiBu , 2H), 1.28 (br, PCH(CH3)2, 24H), 1.73 (br, NiBu , 5H), 2.30

3 3 (br, PCH(CH3)2, 4H), 6.77 (d, JH-H = 8 Hz, ArH, 2H), 6.98 (t, JH-H = 7.2 Hz, ArH, 1H) . Other resonances of butyl moiety were overlapped with the resonance at 1.28 ppm. 13C{1H} NMR

(101 MHz, C6D6, δ): 0.56 (t, JC-P = 17.2 Hz, NiCH2CH2CH2CH3), 14.25 (s, NiCH2CH2CH2CH3,

1 supported by DEPT 135), 17.15 (s, PCH(CH3)2), 17.82 (s, PCH(CH3)2), 28.48 (t, JC-P = 11.1 Hz,

PCH(CH3)2), 30.26 (s, NiCH2CH2CH2CH3, supported by DEPT135), 36.29 (s,

31 1 NiCH2CH2CH2CH3, supported by DEPT135), 104.61 (s), 167.99 (t, JC-P = 10.4 Hz). P{ H}

NMR (162 MHz, C6D6, δ): 188.03 (s). Anal. Calcd for C22H40O2P2Ni: C, 57.80; H, 8.82.

Found: C, 57.67; H, 8.80.

11 Chakraborty, S.; Krause, J. A.; Guan, H. Organometallics 2009, 28, 582. 140

i sec Synthesis of [2,6-( Pr2PO)2C6H3]Ni(Bu ) (14a). Under an argon atmosphere, a 1.4 M solution of secBuLi in cyclohexane (0.66 mL, 0.92 mmol) was added slowly to a Schlenk flask containing a cold solution (-78oC) of 2a (200 mg, 0.46 mmol) in 20 mL of toluene. The color of the solution changed from orange yellow to orange upon gradually warming up to room temperature. The solution was subsequently stirred at room temperature for 4 hours and the solvent was removed under vacuum. The resulting yellow residue was then treated with pentane and then filtered under argon through a pad of Celite to obtain a clear yellow solution. Removal of the solvent under vacuum and drying under vacuum afforded a yellow powder (77 mg, 73% yield). The compound was stored inside a glove box in an amber vial at -30 oC. 1H NMR (400

sec sec MHz, C6D6, δ): 1.17 (br, PCH(CH3)2, 24H), 1.61 (br, NiBu , 2H), 2.03 (br, NiBu , 2H), 2.14

3 3 (br, PCH(CH3)2, 4H), 6.76 (d, JH-H = 7.6 Hz, ArH, 2H), 6.99 (t, JH-H = 7.6 Hz, ArH, 1H) . Other resonances of sec-butyl moiety were overlapped with the resonance iPr resonances. 13C{1H}

2 NMR (101 MHz, C6D6, δ): 14.25 (t, JC-P = 16.7 Hz, NiCH(CH3)(CH2CH3), supported by DEPT

90), 16.24 (s, NiCH(CH3)(CH2CH3), supported by DEPT 135), 16.99 (s, PCH(CH3)2), 17.34 (s,

3 PCH(CH3)2), 18.09 (s, PCH(CH3)2), 18.50 (s, PCH(CH3)2), 23.96 (t, JC-P = 3.0 Hz,

1 NiCH(CH3)(CH2CH3), supported by DEPT135), 28.70 (t, JC-P = 10.1 Hz, PCH(CH3)2), 29.01 (t,

1 3 JC-P = 10.7 Hz, PCH(CH3)2), 36.79 (t, JC-P = 4.5 Hz, NiCH(CH3)(CH2CH3), supported by

31 1 DEPT135), 104.43 (t, JC-P = 5.1 Hz), 128.60 (s), 138.96 (s), 167.73 (t, JC-P = 10.1 Hz). P{ H}

NMR (162 MHz, C6D6, δ): 184.45 (s). Anal. Calcd for C22H40O2P2Ni: C, 57.80; H, 8.82.

Found: C, 57.83; H, 8.91.

141

i i Synthesis of [2,6-( Pr2PO)2C6H3]Ni(Pr ) (16a). Under an argon atmosphere, a

2 M solution of iPrMgCl in diethyl ether (0.23 mL, 0.46 mmol) was added slowly to a Schlenk flask containing a cold solution (-78oC) of 2a (100 mg, 0.23 mmol) in 20 mL of toluene. The color of the solution changed from orange yellow to light yellow upon gradually warming up to room temperature. The solution was subsequently stirred at room temperature for 4 hours and then the solvent was removed under vacuum. The resulting yellow residue was treated with pentane and then filtered under argon through a pad of Celite to obtain a clear yellow solution. Removal of the solvent under vacuum and drying under vacuum afforded a

1 yellow powder (88 mg, 86% yield). H NMR (400 MHz, C6D6, δ): 1.01 (m, NiCH(CH3)2, 1H),

3 1.17 (m, PCH(CH3)2, 24H), 1.57 (s, NiCH(CH3)2, 6H), 2.13 (m, PCH(CH3)2, 4H), 6.75 (d, JH-H

3 13 1 = 8 Hz, ArH, 2H), 6.98 (t, JH-H = 8 Hz, ArH, 1H). C{ H} NMR (101 MHz, C6D6, δ): 5.52 (t,

2 JC-P = 17.1 Hz, NiCH(CH3)2) supported by DEPT90), 17.16 (s, PCH(CH3)2), 18.35 (s,

1 3 PCH(CH3)2), 28.83 (t, JC-P = 11.1 Hz, PCH(CH3)2), 29.07 (t, JC-P = 3.8 Hz, NiCH(CH3)2, supported by DEPT135), 104.46 (t, JC-P = 5.8 Hz), 128.56 (s), 138.99 (t, JC-P = 19.3 Hz), 167.74

31 1 (t, JC-P = 10.1 Hz). P{ H} NMR (162 MHz, C6D6, δ): 184.79 (s). Anal. Calcd for

C21H38O2P2Ni: C, 56.91; H, 8.64. Found: C, 56.70; H, 8.58.

i n Synthesis of [2,6-( Pr2PO)2C6H3]Ni(Pr ) (17a). Under an argon atmosphere,

a 2 M solution of nPrMgCl in diethyl ether (0.23 mL, 0.46 mmol) was added slowly to a Schlenk flask containing a cold solution (-78oC) of 2a (100 mg, 0.23 mmol) in 20 mL of toluene. The color of the solution changed from orangish yellow to light yellow upon gradually warming up to room temperature. The solution was stirred further at room temperature for 4 hours and the solvent was removed under vacuum. The resulting yellow

142 residue was treated with pentane and then filtered under argon through a pad of Celite to obtain a clear yellow solution. Removal of the solvent under vacuum and drying under vacuum afforded

1 a yellow powder (73 mg, 72% yield). H NMR (400 MHz, C6D6, δ): 0.81 (m, NiCH2CH2CH3,

3 2H), 1.15 (m, PCH(CH3)2, 24H), 1.27 (t, JH-H = 8 Hz, NiCH2CH2CH3, 3H), 1.66 (m,

3 3 NiCH2CH2CH3, 2H), 2.15 (m, PCH(CH3)2, 4H), 6.77 (d, JH-H = 8 Hz, ArH, 2H), 6.98 (t, JH-H =

13 1 2 8 Hz, ArH, 1H). C{ H} NMR (101 MHz, C6D6, δ): 4.57 (t, JC-P = 17.4 Hz, NiCH2CH2CH3, supported by DEPT135), 17.14 (s, PCH(CH3)2), 17.79 (s, PCH(CH3)2), 21.99 (s, NiCH2CH2CH3, supported by DEPT135), 27.44 (s, NiCH2CH2CH3, supported by DEPT135), 104.58 (t, JC-P = 5.1

31 1 Hz), 128.45 (s), 139.60 (t, JC-P = 18.7 Hz), 167.99 (t, JC-P = 10.1 Hz). P{ H} NMR (162 MHz,

C6D6, δ): 188.20 (s). Anal. Calcd for C21H38O2P2Ni: C, 56.91; H, 8.64. Found: C, 56.65; H,

8.47.

X-Ray Diffraction Study. Single crystals of 14a were obtained from a saturated solution of the complex in pentane at -30oC. Crystal data and refinement parameters for 14a are given in

Table 1. For X-ray examination and data collection, the crystals were mounted in a cryo-loop with paratone-N and transferred immediately to the goniostat bathed in a cold stream. Intensity data for both complexes were collected at 150K on a standard Bruker SMART6000CCD diffractometer using graphite-monochromated Cu Kα radiation, λ=1.54178Å. A series of 5-s data frames measured at 0.3o increments of ω were collected to calculate a unit cell. The data frames were processed using the program SAINT. The data were corrected for decay, Lorentz and polarization effects as well as absorption and beam corrections based on the multi-scan technique. The structures were solved by a combination of direct methods in SHELXTL and the

143 difference Fourier technique and refined by full-matrix least squares on F2. The secondary butyl group in 14a shows disorder. A disorder model is presented for C21 in this molecule.

Table 1. Crystal Data and Refinement Parameters for 14a

14a

empirical formula C22H40O2P2Ni crystal system monoclinic space group P2 /c 1 a, Å 17.2100(15) b, Å 9.1151(8) c, Å 16.6147(15) , deg 90 , deg 112.3090(10) , deg 90 Volume, Å3 2411.3(4) Z 4 no. of data collected 26364

no. of unique data, Rint 4970, 0.0673 R1, wR2 (I > 2(I)) 0.0387, 0.0862 R1, wR2 (all data) 0.0595, 0.0940

144

Chapter 6 Catalytic Cyanomethylation of Aldehydes

145

6.1 Introduction

β-hydroxy nitriles are useful intermediates for the synthesis of a variety of pharmaceutically important compounds such as 1,3-diamino alcohols.1 Among several methods of preparing β-hydroxy nitriles, the most common ones involve the reaction of 1,2-epoxides with nitriles in the presence of metal salts such as LiClO4/KCN, using lanthanide(III) alkoxides as catalysts,2 or with acetone cyanohydrin under mildly basic conditions.3 However, these approaches are often limited to simple aliphatic epoxides. The coupling of alkyl iodides, acrylonitriles and ketones in the presence of a Mn/Pb system can also produce β-hydroxy nitriles.4 However, the toxicity issues associated with the lead makes this method environmentally unsafe. β-hydroxy nitriles can also be formed by the treatment of an electron deficient alkene with mercury fulminate and LiBr.5 But, the toxicity of mercury, the explosive nature of mercury fulminate renders this method difficult and unsafe for large-scale synthesis.

Another route to synthesize β-hydroxy nitriles is deprotonation of acetonitrile by a strong base such as n-butyllithium or alkali amides followed by coupling with ketones and aldehydes.6

However, very low reaction temperatures (-80oC) are generally required to avoid many side reactions.

1 Fülöp, F.; Huber, I.; Bernáth, G.; Hönig, H.; Senger-Wasserthal, P. Synthesis 1991, 43. 2 Ohno, H.; Mori, A.; Inoue, S. Chem. Lett. 1993, 6, 975. 3 Mitchell, D.; Koenig, T. Tetrahedron Lett. 1992, 33, 3281. 4 Takai, K.; Ueda, T.; Ikeda, N.; Moriwake, T. J. Org. Chem. 1996, 61, 7990. 5 You, Z.; Lee, H. Tetrahedron Lett. 1996, 37, 1165. 6 Kaiser, E. W.; Hauser, C. R. J. Org. Chem. 1968, 33, 3402.

146

6.2 Transition Metal-Catalyzed Cyanomethylation of Aldehydes and Ketones

A complementary approach to produce β-hydroxy nitriles is the catalytic cyanomethylation of aldehydes and ketones under mild conditions. However, catalytic utilization of nitriles as nucleophiles has been mostly limited to β-cyano carbonyls (pKa = 13)7a and α-arylnitriles (pKa = 22)7b, which contain relatively acidic hydrogens. As a comparison, the pKa values (~31) for simple alkylnitriles are much higher.8 Consequently, the in-situ generation of nucleophiles from alkylnitriles requires highly basic conditions. Under this strongly basic condition, undesired side reactions become predominant which result in poor chemoselectivity.

Despite these obstacles, efficient catalytic cyanomethylation has been performed recently by using a proazaphosphatrane base9 or a metal tert-butoxide10 as the catalyst. The Shibasaki group has explored coordination to a Lewis acidic [CpRu(PPh3)(NCMe)2]PF6 complex as a way to activate acetonitrile towards deprotonation by a mild base.11 Recently, the Ozerov group has demonstrated that a PNP pincer ligand-derived nickel triflate complex can catalyze cyanomethylation of aldehydes in the presence of two equivalents of DBU (Scheme 1).12 Even though relatively milder reaction conditions have been employed, use of stoichiometric amount of external base is mandatory for these reactions. Mechanistic studies have suggested that at first acetonitrile can bind to the cationic (PNP)Ni fragment and thereby its protons become more acidic. Deprotonation of this more acidic proton by DBU and coupling of the resulting nucleophile with aldehydes produces the Ni-bound coupled product. The product can be released in the presence of acetonitrile to regenerate the active catalyst. Although the catalytic

7 (a) In DMSO (b) In CH3CN 8 Arseniyadis, S.; Kyler, K. S.; Watt, D. S. Org. React. 1984, 31, 1. 9 Review: Verkade, J. G.; Kisanga, P. Aldrichimica Acta 2004, 37, 3. 10(a) Suto, Y.; Kumagai, N.; Matsunaga, S.; Kanai, M.; Shibasaki, M. Org. Lett. 2003, 5, 3147. (b) Bunlaksananusorn, T.; Rodriguez, A. L.; Knochel, P. Chem. Commun. 2001, 745. 11 Kumagai, N.; Matsunaga S.; Shibasaki, M.; J. Am. Chem. Soc. 2004, 126, 13632. 12 Fan, L.; Ozerov, O. V. Chem. Commun. 2005, 4450.

147 cycle proposed by Ozerov shows the utilization of DBU as a cocatalyst, the need of excess amount of this base in the actual reaction has not been explained.

To improve the effectiveness of the cyanomethylation reaction of aldehydes and ketones, a mild “base-free” condition has to be developed. Generating an internal base that is capable of deprotonating acetonitrile would be a strategy to achieve this goal. It has been already demonstrated in chapter 2 that nickel pincer hydride complexes undergo insertion reaction with aldehydes. These reactions produce nickel alkoxide complexes. Use of an external base can be avoided if these nickel alkoxides are basic enough to deprotonate acetonitrile. Therefore, I intended to investigate catalytic cyanomethylation reaction of aldehydes and ketones with nickel pincer hydride complexes.

148

6.3 Catalytic Cyanomethylation of Aldehydes with Nickel Hydride Complexes

According to the plan, I tested nickel pincer hydride complexes as catalysts for the cyanomethylation of aldehydes and ketones (Scheme 2, Table 1). When 1 mol% of nickel pincer hydride complex 3a was used as the catalyst to perform the coupling reaction between benzaldehyde and acetonitrile, the reaction proceeded smoothly with the formation of the desired cyanomethylated product at room temperature. No external base was added into this reaction.

The reaction was finished in six hours and no dehydration side product was observed under this neutral pH condition. When 3c was tested as a catalyst, similar reactivity was observed.

However, 3b did not catalyze the cyanomethylation reaction. Inactivity of 3b could be rationalized by the fact that it does not undergo insertion reaction with benzaldehyde (see

Chapter 2).

Table 1. Comparing Catalytic Activity of 3a-c in Cyanomethylation Reaction

149

Next, the scope of substrates were investigated employing 3a as the catalyst (1 mol%)

(Scheme 3, Table 2). Clearly, aldehydes with electron-withdrawing groups (entry 2-4) reacted much faster than both unsubstituted aldehyde (entry 1) and aldehydes with electron-donating substituent (entry 5). Reactions were completed within few hours in case of aldehydes having electron-withdrawing substituent at the para position of the phenyl ring. In contrast, aldehydes containing electron-withdrawing substituents reacted much slowly and sometimes higher temperature was required. Functional groups such as NO2 and CF3 were intact under these reaction conditions. The reaction was found to be selective for aldehydes in the presence of ketones. In case of entry 6 only the aldehyde moiety underwent cyanomethylation. Base- sensitive aliphatic aldehydes such as heptaldehyde (entry 7) also reacted at room temperature to produce the corresponding β-hydroxy nitrile.

150

Table 2. Catalytic Cyanomethylation of Aldehydes with 3a

A possible catalytic pathway for this cyanomethylation reaction is outlined in Scheme 4.

The nickel cyanomethyl complex (18a) has been hypothesized as one of the potential intermediates in this process. To gain better understanding about this reaction, isolation of this potential intermediate was aimed and stoichiometric steps related to the hypothesized catalytic cycles were investigated by NMR studies.

151

6.4 Synthesis of Nickel Cyanomethyl Complex

The nickel cyanomethyl complex 18a was prepared following a two-step process. First, acetonitrile was deprotonated by nBuLi and then the resulting anion was reacted with nickel

- pincer chloride complex 2a (Scheme 5). Substitution of Cl ligand by [CH2CN] occurred rapidly at room temperature. Compound 18a has so far been characterized by multinuclear NMR and IR

1 spectroscopy. In the H NMR, the characteristic Ni-CH2CN resonance of 18a appeared as a

3 triplet ( JP-H = 8.8 Hz) at 0.81 ppm because of the coupling with two magnetically equivalent phosphorous atoms. The cyanomethyl carbon atom comes at a very high-field region (-23.9

13 1 2 ppm) in C{ H}NMR. This carbon appears as a triplet ( JP-C = 13.1 Hz) which strongly

152 supports that the cyanomethyl group is linked to the nickel center by the carbon atom rather than the nitrogen atom. A single resonance was observed for 18a at 190.85 ppm in the 31P{1H}NMR spectrum. In attenuated total reflection (ATR) IR spectroscopy, a strong band at 2184 cm-1 was observed for C≡N stretch.

6.5 Reactivities of Nickel Cyanomethyl Complex

To test the hypothesis that 18a is involved in the catalytic cycle, its reactivities were investigated. First, its reaction with aldehyde was examined. When 18a and PhCHO were mixed in a 1 : 1 ratio in C6D6, no reaction was observed at room temperature. Warming up the

o reaction mixture to 60 C did not result in any change (Scheme 6). C6D6 was chosen as the NMR solvent for the initial study due to the simplicity of the NMR spectrum; however it should be noted that the catalytic process was performed in CH3CN. Replacing C6D6 with CD3CN as the

NMR solvent did result in the formation of the β-hydroxy nitrile (Scheme 7).

153

One may anticipate that the reaction in Scheme 7 should yield 19a instead of 19a-D2. It is, however, interesting to notice that both the final products 18a-D2 and β-hydroxy nitrile compound contain deuterium atoms. This observation would suggest that 18a undergoes H/D exchange with CD3CN. To verify this hypothesis, 18a was dissolved in CD3CN at room temperature (Scheme 8). This experiment resulted in quantitative formation of 18a-D2 in one hour.

154

Generation of an intermediate 19a during the reaction of 18a and PhCHO (Scheme 7) was proposed but it was not observed by NMR spectroscopy. Perhaps, the deprotonation of acetonitrile by 19a is too fast to be detected by NMR. This challenge, however, can be overcome if a polar aprotic solvent such as THF-d8 is used instead of the polar protic acetonitrile

(Scheme 9). Independent synthesis of 19a should be performed to probe the properties of this molecule and validate our mechanistic hypothesis (Scheme 10).

155

Another important question that I wanted to address was how the nickel cyanomethyl complex formed from the nickel hydride precatalyst under the catalytic conditions. Indeed there are two possible pathways to generate 18a from 3a. Firstly, hydridic Ni-H can itself behave as a base to deprotonate acetonitrile with the concomitant release of dihydrogen. Secondly, nickel hydride can undergo insertion reaction with aldehydes, followed by deprotonation of acetonitrile with the help of basic Ni-bound alkoxide. Accordingly, both possibilities were investigated

(Scheme 11). When 3a was dissolved in CD3CN and mixed for several days at room temperature, formation of nickel cyanomethyl complex was not detected by 1H and 31P{1H}

NMR. This result is inconsistent with the first possibility. To test the second hypothesis, nickel benzyloxide complex 4a was prepared in situ by an insertion reaction of 3a with PhCHO (see

Chapter 2) in C6D6. Evaporation of the solvent, followed by redissolution in CD3CN rapidly

1 31 1 formed 18a-D2 as observed by H and P{ H} NMR spectroscopy. This observation not only demonstrates how 18a is generated from 3a but also verifies the hypothesis of nickel alkoxide complex acting as an internal base in our system.

156

On the basis of these NMR analyses, a more detailed mechanism for the catalytic cyanomethylation reaction has been proposed in Scheme 12. The first two steps of this mechanism illustrates the generation of active nickel cyanomethyl complex from staring nickel hydride precatalyst. The nickel alkoxide intermediate acts as an internal base in this reaction to form cyanomethyl complex. Then the nucleophilic methylene carbon of 18a attacks the electropositive center of aldehydes leading to the formation of Ni-bound coupled product. This nucleophilic attack is faster when an electron-withdrawing group is present in aldehydes compared to aldehydes bearing electron-donating groups. In fact, as shown in Table 2, electron- poor aldehydes react much faster than electron-rich aldehydes in our case. The carbon-carbon bond forming step is believed to be the rate-limiting step in this cyanomethylation reaction and this possibility should be tested by in-situ IR spectroscopy in the future. In the last step, again the basic Ni-bound alkoxide complex reacts with acetonitrile to regenerate the active nickel cyanomethyl complex and thereby releasing the β-hydroxy nitrile product.

157

6.5 Conclusions

Nickel hydride complexes efficiently catalyze the cyanomethylation reaction of aldehydes to produce β-hydroxy nitriles that are useful synthetic blocks. The reactions are performed under mild conditions without the use of an external base. Aldehydes with electron- withdrawing groups react faster than aldehydes with electron-donating groups. Base-sensitive aldehydes are also viable substrates for this reaction as no external base is being used. The nickel cyanomethyl complex, which is a potential intermediate of the reaction, has been synthesized independently. NMR study supports that this complex reacts with aldehydes in acetonitrile to form the β-hydroxy nitrile compound. The active nickel cyanomethyl species is

158 generated from nickel alkoxide complexes that results from the insertion of aldehydes to nickel hydrides, followed by the deprotonation of acetonitrile.

6.6 Experimental Section

General Comments. Unless otherwise noted, all the organometallic compounds were prepared and handled under an argon atmosphere using standard Schlenk and inert-atmosphere box techniques. Dry and oxygen-free toluene was collected from an Innovative Technology solvent purification system and used throughout the experiments. Off-the-shelf acetonitrile was used throughout the experiments. C6D6 was distilled from Na and benzophenone under an argon atmosphere. CD3CN and CD2Cl2 were purchased from Cambridge Isotope Laboratories and used without further purification. All aldehydes were vacuum distilled prior to use. 1H, 13C{1H}, and 31P{1H} NMR spectra were recorded on a Bruker Avance-400 MHz spectrometer. Chemical shift values in 1H and 13C spectra were referenced internally to the residual solvent resonances.

31 13 13 P spectra were referenced externally to 85% H3PO4 (0 ppm). Compounds 2a and 3a were prepared according to the literature procedure. All β-hydroxy nitrile compounds are known in the literature.14

13 Chakraborty, S.; Krause, J. A.; Guan, H. Organometallics 2009, 28, 582. 14 Pamies O.; Bӓckvall, J.-E. Adv. Synth. Catal. 2001, 343, 726.

159

i Synthesis of [2,6-( Pr2PO)2C6H3]Ni(CH2CN) (18a). CH3CN (17

μL, 0.31 mmol) was dissolved in 5 mL of toluene, followed by the addition of n-butyllithium (96 μL of 2.5 M solution in hexanes, 0.24 mmol). The solution was stirred at ambient temperature for 15 min, and then 2a (100 mg, 0.23 mmol) was added at 0oC. The color of the solution changed to light yellow gradually, and it was allowed to stir for 2 hours at room temperature. The volatiles were removed in vacuo, and the residue was redissolved in pentane.

The mixture was filtered through Celite, and the filtrate was collected. Evaporation of the

1 solvent gave a yellow solid (88 mg, 87%). H NMR (400 MHz, CD2Cl2, δ): 0.81 (br, NiCH2CN,

3 2H), 1.29 (m, PCH(CH3)2, 24H), 2.48 (m, PCH(CH3)2, 4H), 6.46 (d, JH-H = 8 Hz, 2H), 6.95 (t,

3 13 1 JH-H = 7.2 Hz, 1H). C{ H} NMR (101 MHz, CD2Cl2, δ): -23.91 (t, JC-P = 13.1 Hz, NiCH2CN),

17.13 (s, PCH(CH3)2), 18.04 (s, PCH(CH3)2), 28.57 (t, JC-P = 11.0 Hz, PCH(CH3)2), 104.99 (s,

ArC), 128.93 (s, NiCH2CN), 129.33 (s, ArC), 134.39 (t, JC-P = 19.7 Hz, ArC), 168.49 (t, JC-P =

31 1 -1 9.1 Hz, ArC) P{ H} NMR (162 MHz, CD2Cl2, δ): 190.85 (s). IR (ATR, cm ): 2184 (C≡N stretch).

General procedure for Catalytic Cyanomethylation of Aldehydes. To a flame-dried

10 mL vial, was added a solution of 3a (10 mg, 25 µmol) in acetonitrile (3 ml) and an aldehyde substrate (250 mmol) under an argon atmosphere. The resulting mixture was stirred at room temperature. After the reaction is completed, a small portion of the mixture was passed through a small pipet-column of Silica gel to separate the organic part from the metallic component.

Evaporation of the clear solution yielded the product. Conversions were calculated from the relative integration of starting aldehyde and the product.

160

Stoichiometric Reaction of 18a and PhCHO in C6D6. A J. Young NMR tube was charged with 18a (11 mg, 25 μmol), PhCHO (2.5 μL, 25 μmol), and 0.6 mL of C6D6. The progress of the reaction was monitored by 1H and 31P{1H} NMR spectroscopy. No appreciable reaction was observed within 48 h.

Stoichiometric Reaction of 18a and PhCHO in CD3CN. A J. Young NMR tube was charged with 18a (11 mg, 25 μmol), PhCHO (2.5 μL, 25 μmol), and 0.6 mL of CD3CN. The progress of the reaction was monitored by 1H and 31P{1H} NMR spectroscopy.

Exchange Interaction between 18a and CD3CN. Inside a J. Young NMR tube 18a (11 mg, 25 μmol) was dissolved in 0.6 mL of CD3CN. The progress of the reaction was monitored

1 31 1 by H and P{ H} NMR spectroscopy. The complex 18a-D2 was formed quantitatively after 1 hour.

Deprotonation of Acetonitrile by the Nickel Alkoxide Complex. Inside a J. Young

NMR tube, nickel benzyloxide complex (4a) was first prepared in-situ by mixing stoichiometric amount of 3a (10 mg, 25 μmol) and PhCHO (2.5 μL, 25 μmol). Once 4a was completely formed

(judged by 1H and 31P{1H} NMR spectroscopy), the solvent was removed under vacuum to obtain an orange oily residue of 4a. To this residue, 0.6 mL of CD3CN was added and the reaction mixture was analyzed by 1H and 31P{1H} NMR spectroscopy which clearly indicated the formation of 18a-D2.

Stoichiometric Reaction of 3a and CD3CN. Inside a J. Young NMR tube 3a (10 mg,

1 25 μmol) was dissolved in 0.6 mL of CD3CN. The progress of the reaction was monitored by H and 31P{1H} NMR spectroscopy. No reaction was observed within 48 h.

161