Group 4 Metal Complexes with Ferrocenyl Amidinates

Transcription

1 Group 4 Metal Complexes with Ferrocenyl Amidinates by Kanwarpal Multani A thesis submitted in conformity with the requirements for the degree of Master of Science Graduate Department of Chemistry University of Toronto Copyright by Kanwarpal Multani 2010

2 Group 4 Metal Complexes with Ferrocenyl Amidinates Abstract Kanwarpal Multani Master of Science Graduate Department of Chemistry University of Toronto 2010 Bis(amidinate) dichloride complexes of the type M(L) 2 Cl 2 (M=Zr, 2a; M=Ti, 2b; L=CyNC(C 5 H 5 FeC 5 H 4 )NCy) were synthesized by treating 2 equiv of ferrocenyl amidine, H(L), with M(NMe 2 ) 2 Cl 2 (M=Ti, Zr. 2THF). Half sandwich mono(amidinate) complexes, Cp ZrLCl 2 (Cp =Cp, 2c; Cp =Cp*, 2d), were prepared by the reaction of Cp ZrCl 3 with 1 equiv of Li(L). The dialkyl complexes, M(L) 2 Me 2 (M=Zr, 3a; M=Ti, 3b), CpZr(L)(CH 2 Ph) 2 (3c) and Cp*Zr(L)Me 2 (3d) were prepared by treatment of the dichloride complexes (2a- 2d) with an appropriate alkylating agent. The dichloride complexes (2a-2d) activated with MAO, and dialkyl complexes (3a-3d) activated with B(C 6 F 5 ) 3 and [Ph 3 C][B(C 6 F 5 ) 4 ] show low to moderate ethylene polymerization activities. Cyclic voltammetry studies on the metal complexes containing ferrocenyl amidinates reveals quasi reversible oxidation and reduction waves for the ferrocene/ferrocenium couple. ii

3 Acknowledgments I would like to acknowledge a number of people for their support over the past little while. First and foremost, I would to thank my research advisor Prof. Doug Stephan for his ideas, encouragement and enthusiasm. Thanks to all of the past and present Stephan group members with whom I have worked closely. Particularly, Dr. Alberto Ramos for teaching me basic laboratory skills when I initially started. A special thank you to Dr. Zach Heiden for his extensive help on cyclic voltammetry and solving crystal structures. Also, thanks to Meghan Dureen and Dr. Edwin Otten for editing this thesis. Lastly, I would like to thank my parents for all of their contributions. SABIC Corporation is acknowledged for providing the financial support. iii

4 Table of Contents Page Acknowledgments... iii Table of Contents... iv List of Tables... vi List of Figures... vii List of Schemes... ix List of Abbreviations and Symbols... x Compound Numbering Scheme... xi Chapter 1: Introduction Background Features of Successful Catalysts Catalyst Activity Polymer Properties Proposed Mechanism of Polymerization Activation Propagation Termination Ethylene Polymerization Catalysts Early Catalysts Metallocene Based Catalysts Post Metallocene Catalysts Amidinate Based Olefin Polymerization Catalysts Amidinates Modes of Coordination Synthesis of Metal-Amidinates Scope of Thesis Chapter 2: Synthesis, Characterization and Ethylene Polymerization Activity of Group 4 Ferrocenyl Amidinate Complexes Introduction iv

5 2.2 Results and Discussion Synthesis and Characterization Synthesis of Amidine and Amidinate Derivatives Synthesis of Group 4 Metal Dichloride Complexes Synthesis of Group 4 Metal Dialkyl Complexes Synthesis of Metal Monoalkyl Cationic Complexes Electrochemical Study of Dichloride and Dialkyl Complexes Polymerization Results Ethylene Polymerization using Dichloride Precatalysts Ethylene Polymerization using Dialkyl Precatalysts Summary Chapter 3: Experimental Details General Considerations Solvents Materials Instrumentation Nuclear Magnetic Resonance Spectroscopy (NMR) Electrochemistry Other Instrumentation Synthesis and Characterization Reagents and Starting Materials Organic and Organometallic Syntheses Polymerization Protocol Schlenk Line Polymerization Crystallography X-ray Data Collection and Reduction Structure Solution and Refinement References v

6 List of Tables Table 1.1. Ranking the effectiveness of the catalyst based on the activity Table 2.2. Selected bond distances (Å) and bond angles ( ) for [ i PrNC(Fc)N i Pr]H, 1b Table 2.3. Selected bond distances (Å) and bond angles ( ) for [CyNC(Fc)NCy]SiMe 3, 1c Table 2.4. Selected bond distances (Å) and bond angles ( ) for [CyNC(Fc)NCy] 2 ZrCl 2, 2a and [CyNC(Fc)NCy] 2 TiCl 2, 2b Table 2.5. Selected bond distances (Å) and bond angles ( ) for CpZr[CyNC(Fc)NCy]Cl 2, 2c, and for Cp*Zr[CyNC(Fc)NCy]Cl 2, 2d Table 2.6. Selected bond distances (Å) and bond angles ( ) for [CyNC(Fc)NCy] 2 ZrMe 2, 3a Table 2.7. Selected bond distances (Å) and bond angles ( ) for CpZr[CyNC(Fc)NCy](CH 2 Ph) 2, 3c Table 2.8. Selected bond distances (Å) and bond angles ( ) for [CyNHC(Fc)NHCy][B(C 6 F 5 ) 4 ] Table 2.9. Summary of the cyclic voltammetric data on the dichloride complexes, 2a-2d Table Summary of the cyclic voltammetric data on the dialkyl complexes, 3a, 3c- 3d Table Ethylene polymerization results of dichloride precursors using MAO as the cocatalyst Table Ethylene polymerization results of dialkyl precursors activated with B(C 6 F 5 ) 3 or [Ph 3 C][B(C 6 F 5 ) 4 ] Table 3.1. Table of crystallographic parameters for [ i PrNC(Fc)N i Pr]H (1b), [CyNC(Fc)NCy]SiMe 3 (1c) and [CyNC(Fc)NCy] 2 ZrCl 2 (2a) Table 3.2. Table of crystallographic parameters for [CyNC(Fc)NCy] 2 TiCl 2 (2b), CpZr[CyNC(Fc)NCy]Cl 2 (2c) and Cp*Zr[CyNC(Fc)NCy]Cl 2 (2d) Table 3.3. Table of crystallographic parameters for [CyNC(Fc)NCy] 2 ZrMe 2 (3a), CpZr[CyNC(Fc)NCy](CH 2 Ph) 2 (3c) and [CyNHC(Fc)NHCy][B(C 6 F 5 ) 4 ] (6) vi

7 List of Figures Figure 1.1. Representation of (a) isotactic polymers, (b) syndiotactic polymers and (c) atactic polymers Figure 1.2. Propagation step of polymerization involves olefin coordination followed by insertion... 6 Figure 1.3. Agostic interaction in propagation step of polymerization Figure 1.4. Several chain termination pathways. (a) β-hydride transfer to metal (b) β hydride transfer to monomer (c) β-methyl elimination (d) chain transfer to aluminum Figure 1.5. Examples of ansa-zirconocene catalysts controlling the steroselectivity of the resulting polymer Figure 1.6. General structure of constrained geometry catalysts Figure 1.7. General structure of (a) McConville s diamide catalysts (b) phosphinimide based catalysts (c) FI catalysts based on Group 4 salicylaldiminide complexes Figure 1.8. Active amidinate based olefin polymerization catalysts (a) bis(benzamidinate) complexes (b) acetamidinate complexes (c) guanidinate complexes (d) bis(iminophosphonamide) based complexes Figure 1.9. Resonance structures of amidinates Figure Ligands isoelectronic to amidinates: (a) guanidinates (b) carboxalates (c) triazenates Figure Commonly observed modes of coordination: (a) chelating bidentate (b) bridging (c) monodentate Figure 2.1. Active olefin polymerization catalysts containing ferrocene based ligands: (a) bis(amino)ferrocenyl ligand (b) ferrocenyl dimethylsilyl substituted zirconocenes (c) ferrocenyl substituted phosphinimine ligand Figure 2.2. Molecular structure of [ i PrNC(Fc)N i Pr]H, 1b and [CyNC(Fc)NCy]SiMe 3, 1c Figure H NMR spectrum of 2a from ppm shows splitting of substituted Cp protons and broadening of the cyclohexyl protons at low temperature Figure 2.4. Molecular structure of [CyNC(Fc)NCy] 2 ZrCl 2, 2a Figure 2.5. Molecular structure of [CyNC(Fc)NCy] 2 TiCl 2, 2b Figure 2.6. Molecular structure of CpZr[CyNC(Fc)NCy]Cl 2, 2c Figure 2.7. Molecular structure of Cp*Zr[CyNC(Fc)NCy]Cl 2, 2d Figure 2.8. Molecular structure of [CyNC(Fc)NCy] 2 ZrMe 2, 3a vii

8 Figure 2.9. Molecular structure of CpZr[CyNC(Fc)NCy](CH 2 Ph) 2, 3c Figure Molecular structure of [CyNHC(Fc)NHCy][B(C 6 F 5 ) 4 ] Figure Cyclic voltammogram of 2a Figure Cyclic voltammogram of 2c Figure Cyclic voltammogram of 3a Figure Cyclic voltammogram of 3c viii

9 List of Schemes Scheme 1.1. Activation of metallocene precatalysts by several routes to prepare the proposed active catalyst Scheme 1.2. Synthesis of N,N,N -tris(trimethylsilyl)benzamidine Scheme 1.3. Common routes for synthesis of metal amidinates. (a) elimination of trimethylsilyl chloride (b) amine elimination (c) carbodiimide insertion (d) salt metathesis Scheme 2.1. Synthesis of ferrocenyl amidine Scheme 2.3. Synthesis of ferrocenyl amidine with isopropyl substituents on the nitrogen Scheme 2.4. Synthesis of trimethylsilyl substituted ferrocenyl amidine, 1c Scheme 2.5. Synthesis of titanium and zirconium bis(amidinate) complexes, 2a and 2b Scheme 2.6. Rapid interconversion of 2a at room temperature Scheme 2.7. Synthesis of half sandwich mono(amidinate) zirconium complexes, 2c and 2d, via salt metathesis Scheme 2.8. Attempted synthesis of half sandwich mono(amidinate) titanium complexes via salt metathesis was unsuccessful Scheme 2.9. Synthesis of half sandwich zirconium mono(amidinate) dimethyl complex, which subsequently undergoes ligand redistribution Scheme Synthesis of cationic monomethyl zirconium bis(amidinate) complexes paired with MeB(C 6 F 5 ) - 3 or B(C 6 F 5 ) - 4 anions, 4a and 5a, respectively Scheme Synthesis of cationic monomethyl zirconium mono(amidinate) complex paired with MeB(C 6 F 5 ) - 3 or B(C 6 F 5 ) - 4 anions, 4d and 5d, respectively Scheme Synthesis of cationic zirconium mono(amidinate) complex, 4c, paired with MeB(C 6 F 5 ) - 3 anion and possible modes of benzyl coordination ix

10 List of Abbreviations and Symbols Ǻ δ Anal Calcd Bn br Cp Cp* Cy d DCM e.s.d Fc FcLi FcLi 2 Hz i-pr J m Me ml min mmol ORTEP Ph ppm s t-bu THF TMS Ǻngstrom, m chemical shift calculated elemental analysis benzyl broad cyclopentadienyl pentamethylcyclopentadienyl cyclohexyl doublet dichloromethane estimated standard deviation ferrocenyl monolithioferrocene dilithioferrocene Hertz isopropyl symbol for coupling constant multiplet methyl milliliter(s) minute(s) millimole(s) Oak Ridge thermal ellipsoid plot phenyl parts per million singlet tertiary butyl tetrahydrofuran trimethylsilyl x

11 Compound Numbering Scheme 1a 1b [CyNC(Fc)NCy]H [ i PrNC(Fc)N i Pr]H 1c [CyNC(Fc)NCy]SiMe 3 1d Li[CyNC(Fc)NCy] 2a [CyNC(Fc)NCy] 2 ZrCl 2 2b [CyNC(Fc)NCy] 2 TiCl 2 2c CpZr[CyNC(Fc)NCy]Cl 2 2d Cp*Zr[CyNC(Fc)NCy]Cl 2 3a [CyNC(Fc)NCy] 2 ZrMe 2 3b [CyNC(Fc)NCy] 2 TiMe 2 3c CpZr[CyNC(Fc)NCy](CH 2 Ph) 2 3d Cp*Zr[CyNC(Fc)NCy]Me 2 4a [{CyNC(Fc)NCy} 2 ZrMe][MeB(C 6 F 5 ) 3 ] 4b [{CyNC(Fc)NCy} 2 TiMe][MeB(C 6 F 5 ) 3 ] 4c [CpZr{CyNC(Fc)NCy}CH 2 Ph][PhCH 2 B(C 6 F 5 ) 3 ] 4d [Cp*Zr{CyNC(Fc)NCy}Me][MeB(C 6 F 5 ) 3 ] 5a [{CyNC(Fc)NCy} 2 ZrMe][B(C 6 F 5 ) 4 ] 5b [{CyNC(Fc)NCy} 2 TiMe][B(C 6 F 5 ) 4 ] 5c [CpZr{CyNC(Fc)NCy}CH 2 Ph][B(C 6 F 5 ) 4 ] 5d [Cp*Zr{CyNC(Fc)NCy}Me][B(C 6 F 5 ) 4 ] xi

12 1 Chapter 1 Introduction 1.1 Background Over the past several decades, plastic has slowly replaced alternative materials such as paper, glass and metal. Plastics are high molecular weight synthetic polymers, which exhibit a wide range of properties such as density, crystallinity, tensile strength and elasticity based on their microstructure. Small changes in the chemical makeup can lead to a significant difference in the properties of the resulting polymer. Many different polymers are commercially produced for a range of applications. Polyethylene is the most commonly used synthetic polymer, with an annual production reaching approximately 80 million metric tons, followed by polyvinylchloride and polypropylene. 1 It was originally discovered in 1933 by Imperial Chemical Industries in United Kingdom. These synthetic polymers can be classified into different categories based on physical properties such as density, molecular weight and extent of branching in the polymer. The two major classes of polyethylene are high density polyethylene (HDPE) and low density polyethylene (LDPE). HDPE is linear straight chain polyethylene with low degree of branching, whereas, LDPE has a high degree of short and long chain branching. Present, industrial and academic research is focused on developing new types of polymers with unique properties such as biodegradability, aesthetics and ease of processability. 2 Although these new polymers may revolutionize the polymer industry, polyethylene continues to be the most widely used polymer. As a result, an industrial interest still remains in the development and commercialization of new ethylene polymerization catalysts, and improving the catalytic activities of known catalysts.

13 Features of Successful Catalysts There are several criteria used to measure the effectiveness of polymerization catalysts such as activity, stereoregularity and molecular weight distribution of the resulting polymer. Very few catalysts make it to the commercial stage as there are additional requirements including the effectiveness of catalyst under industrial conditions and patent position Catalyst Activity Gibson developed a scale that can be used to rank the catalyst activity quantitatively (Table 1.1). 3 The catalyst activity or productivity can be calculated using the following formula (Eq. 1.1). polymer mass Activity = (Eq. 1.1) catalyst mass polymerization time monomer pressure In relevant literature, activities are most commonly reported in g. (mmol. hr. bar) -1 or g. (mmol. hr. atm) -1. There are several factors that contribute to the activity value during polymerization such as stirring rate, temperature, size of vessel, type of activator, solvent scavenger, solvent, and order of precatalyst and activator addition. Table 1.1. Ranking the effectiveness of the catalyst based on the activity. 3 Rating Activity, g. (mmol. hr. atm) -1 Very low < 1 Low 1 10 Moderate High Very high > Polymer Properties An important characteristic of the resulting polymer is the average molecular weight distribution determined through gel permeation chromatography (GPC). The molecular weight distribution curve shows the relative amounts of polymer of different

14 3 molecular weights in a given sample. The polydispersity index (PDI) is calculated as the ratio of the weight average molecular weight ( M w ) to the number average molecular weight ( M n ) and illustrates this distribution. NiM i M n = (Eq. 1.2) N i M w = N M i N M i 2 i i (Eq. 1.3) M w PDI = (Eq. 1.4) M n where N i is the total number of molecules with a molecular weight of M i. Single site catalysts produce a narrow molecular weight distribution with PDI of approximately 2. Stereoregularity of the resulting polymer is also an important consideration and the type of catalyst used for polymerization of higher olefins will define this feature. An isotactic polymer has substituents on the same side of the polymer, syndiotactic polymers have substituents on alternate sides, and in atactic polymers the substituents are placed randomly on the chain (Figure 1.1). Figure 1.1. Representation of (a) isotactic polymers, (b) syndiotactic polymers and (c) atactic polymers. 1.2 Proposed Mechanism of Polymerization The Cossee-Arlman mechanism is the commonly accepted mechanism for polymerization of olefins by Group 4 catalysts. 4,5 This process involves three main stages of activation, propagation and chain termination.

15 Activation The catalytically active species is proposed to be a cationic alkyl complex stabilized by ancillary ligands with a vacant coordination site for olefin binding. 6,7 This cationic species exist as a cation-anion pair, which is generated from dichloride or dialkyl precatalysts. Generation of this species requires an alkyl anion or halide abstraction by a strong Lewis acid, which is often a cocatalyst in polymerization. The cocatalyst is usually the source of the counteranion to the metal cation. Scheme 1.1. Activation of metallocene precatalysts by several routes to prepare the proposed active catalyst. There are several common routes for creating the cationic species. One method involves the use of methylaluminoxane (MAO) as the cocatalyst. MAO is proposed to methylate the dichloride precatalysts, followed by an abstraction of a methyl anion to produce a mono methyl cationic species (Scheme 1.1, a). 8,9 This cationic species is stabilized by the counteranion, [Me-MAO] -. MAO is synthesized through controlled hydrolysis of trimethylaluminum. Although its exact structure is unknown, it is agreed that it exists as a mixture of oligomeric structures with a general formula of [Me(Al)O] x. 10 This activator is used in large excess ( equivalents of precatalyst) and often

16 5 represents a major cost in production of polyolefins. MAO also functions as a solvent scavenger to remove any catalyst poison such as oxygen, water, or sulfur compounds. MAO is the most commonly used cocatalyst in the industry but other cocatalysts, such as perfluorinated substituted phenyl boranes and borates, have also emerged (Scheme 1.1, b-d). These borane activators follow the criteria required for highly active catalysts, mainly the formation of bulky counter anions and weakly coordinating pairs to allow olefin binding during polymerization. The borane activators are usually added in stoichiometric amounts relative to the precatalyst as opposed to MAO, which is added in large excess. The borate anions were found to be superior over MAO anions in the elucidation of mechanistic details of olefin polymerization. B(C 6 F 5 ) 3, a strong Lewis acid, can be used to abstract a methyl anion forming a MeB(C 6 F 5 ) - 3 anion (Scheme 1.1, b). Other commonly used activators include [Ph 3 C][B(C 6 F 5 ) 4 ] and [PhNHMe 2 ][B(C 6 F 5 ) 4 ] (Scheme 1.1, c-d). [Ph 3 C][B(C 6 F 5 ) 4 ] is usually preferred since it releases an unreactive hydrocarbon, whereas, [PhNHMe 2 ][B(C 6 F 5 ) 4 ] releases an amine, which may deactivate the catalyst through coordination to the metal center Propagation After the generation of catalytically active species, the second step of olefin polymerization involves coordination of the olefin to the metal center donating the π electron density to the Lewis acidic metal center (Figure 1.2). This step is followed by a 4-membered intermediate and olefin insertion into the metal-carbon bond of the growing chain. This results in a free coordination site and an alkyl chain bound to the metal center. The process repeats resulting in a growing polymeric alkyl chain. After each insertion, the alkyl chain and vacant coordination site alternate. Based on the ligand framework surrounding the metal center, the direction of olefin binding may be favoured differently at the alternating coordination sites and this phenomenon gives rise to

17 6 stereoselective polymers in higher olefins. Green and Brookhart modified this mechanism and proposed that the insertion step may be facilitated by an agostic interaction between the metal center and the hydrogen on the alpha carbon of the polymer chain (Figure 1.3). 11,12 Figure 1.2. Propagation step of polymerization involves olefin coordination followed by insertion. Figure 1.3. Agostic interaction in propagation step of polymerization Termination The last step of polymerization stops incorporation of new monomers and terminates the polymer chain (Figure 1.4). 13 One method is via a β-hydride elimination in which the hydride from the polymer chain is transferred onto the metal center resulting in a polymer chain with an olefin end (Figure 1.4, a). 14 The alkene insertion may take place restarting the propagation step to generate another polymer chain. An alternative termination pathway involves hydride transfer to an incoming bound olefin resulting in an alkyl chain bound to transition metal and polymer chain with olefin end group (Figure 1.4,

18 7 b). 15 After the polymer chain is ejected, growth of a new chain starts. β-hydride transfer to the monomer is the dominant pathway under normal experimental conditions and results in a new alkyl bound chain without the formation of metal hydride species. Figure 1.4. Several chain termination pathways. (a) β-hydride transfer to metal (b) β hydride transfer to monomer (c) β-methyl elimination (d) chain transfer to aluminum.

19 8 β-methyl elimination involves the transfer of a methyl group to the metal center (Figure 1.4, c). This termination method is quite rare but important in propylene polymerization using Cp* 2 M IV Cl 2 (M = Ti, Zr) complexes. 9 Chain termination by transalkylation to an aluminum center is also possible (Figure 1.4, d). This mechanism is highly dependent on Al:metal ratio and comonomer concentration. Lastly, chain transfer agents, such as molecular hydrogen, may be added to terminate the polymer chain and allowing specific control over polymer molecular weight. Other chain termination methods may involve irreversible deactivation of the catalyst. 1.3 Ethylene Polymerization Catalysts The commercial production of polyethylene began in the 1930 s using a free radical polymerization process. This method requires harsh conditions using high temperatures (approximately 300 C) and pressures (~2000 atm) producing highly branched low density polyethylene with limited applications. 16 Over the years, new catalytic processes have been developed that operate under mild conditions and broadened the range of polymers that are available in the market Early Catalysts In the 1950 s, Ziegler and Natta demonstrated that transition metal based catalysts could aid in polyethylene production. Ziegler discovered that a heterogeneous mixture of transition metal halides, such as TiCl 4, combined with alkyl aluminum based cocatalysts effectively produced high density polyethylene. 17 Around the same time, Natta showed that linear isotactic, crystalline polypropene could be produced using similar heterogeneous systems. These discoveries were a major breakthrough as stereoselective polymers could now be produced at lower temperatures (often C) and pressures (often 8-10 atm) compared to previous methods. 18 Later, Hogan and

20 9 Banks marked the commercial production of polyethylene at low pressures using CrO 3 /SiO 2 systems and this method still accounts for a majority of the worldwide polyethylene production. 19 Modern heterogeneous Ziegler-Natta catalysts with MgCl 2 supports have high activities for polypropene production showing excellent stereoselectivity for the formation of isotactic polymers Metallocene Based Catalysts With the success of Ziegler and Natta heterogeneous systems, many questions such as the exact role of the catalytic surface, the detailed mechanism of polymerization and the introduction of stereoregularity into the polymer chain were beginning to be asked. However, a fundamental problem with the heterogeneous catalysts was that they exhibited low activities due to the limited access to the active sites within the solid. An improvement to this system was the discovery of first homogenous catalyst by Natta and Breslow, who independently reported that titanium metallocenes (Cp 2 TiCl 2 ) along with a dialkyl aluminum chloride cocatalyst polymerized ethylene. 20,21 Although, these catalysts initially showed low activity, their homogeneous nature shed light on the mechanism and kinetics of olefin polymerization by Ziegler Natta catalysts. The success of metallocene catalysts is also due to the remarkable discovery by Kaminsky and Sinn, who reported that zirconium dimethyl metallocenes show a much higher activity when initiated with methylaluminoxane (MAO) instead of alkyl aluminum reagents. 22 The increased activity by MAO was observed for other systems (e.g. Cp 2 TiEtCl/AlEtCl 2 ) and previously inactive systems. 23,24 Moreover, these catalysts also produced polymers with narrow molecular weight distribution. Koppl outlines several advantages of homogenous metallocene catalysts. 25 The metallocene catalysts are more active than traditional heterogeneous Ziegler Natta catalysts. The increase in activity is attributed to their homogenous and uniform nature

21 10 compared to heterogeneous systems. Also, homogeneous systems contain a higher number of active sites relative to their mass since every metal is available to act as a catalyst. Homogeneous catalysts act as single site catalysts which aids in the production of highly uniform polymers with narrow molecular weight distributions. The catalysts can be easily altered to produce polymers with evenly distributed long chain and short chain branching on the polymer chain, features desirable for production of new polymers. The catalyst active site can be changed by modifying the cyclopentadienyl ligand system surrounding the metal center resulting in polymers with different molecular weights, branching and stereospecificity. An excellent control of stereospecificity achieved by metallocene catalysts is exemplified by the ansa-zirconocene catalysts. Activation of C 2 symmetric η 5 - indenylzirconcences and C s -symmetric η 5 -fluorenylzirconocenes with MAO, yields isotactic or syndiotactic polypropylene, respectively (Figure 1.5). 26,27 Hence, the changes in symmetry, steric and electronic properties of the Cp ring in addition to the locked orientation of the zirconocene result in dramatic changes in the microstructure of the polymer mainly the tacticity. Figure 1.5. Examples of ansa-zirconocene catalysts controlling the stereoselectivity of the resulting polymer.

22 Post Metallocene Catalysts New ligand designs that mimic the metallocene catalysts have been the major driving force of research in the past two decades. Okuda reported highly active olefin polymerization catalysts based on Group 4 complexes with ansa-bridged cyclopentadienyl amido ligand (Figure 1.6), commonly known as constrained geometry catalysts (CGCs). 28 CGCs have been studied extensively experimentally and theoretically and are patented by Dow Chemical Company and ExxonMobil Corporation. 29,30,31 In comparison to metallocenes, CGCs are more stable at high temperatures and show higher incorporation of α-olefin monomers in copolymerizations. The copolymerization is facilitated by the open nature of the catalyst and lower tendency of the growing polymeric chain to undergo chain transfer reactions. 32 The coordination sphere in CGC is less crowded considering that the Cp centroid -M-N bite angle is generally smaller by C compared to Cp centroid -M-Cp centroid in metallocenes. 33 Lastly, the polymers generated by CGC have narrow molecular weight distribution with long chain branches, which aids in processing of the polymers and is not observed in metallocene based catalysts. 3 Figure 1.6. General structure of constrained geometry catalysts. McConville and coworkers reported chelating diamide complexes of titanium (Figure 1.7, a), capable of living polymerization of α-olefins and also show high activity in 1-hexene polymerization catalysis. 34,35,36 Fenokishi-Imin Haiishi (FI) catalysts, based on unsymmetrical phenoxy-imine chelating ligand on titanium and zirconium complexes

23 12 (Figure 1.7, c), show high catalytic activity for polymerization of ethylene (20 times higher than Cp 2 ZrCl 2 ), 1-hexene or ethylene-propylene and living polymerization of ethylene. 37,38,39 Stephan and coworkers reported phosphinimide based titanium complexes (Figure 1.7, b) with high ethylene polymerization activities under industrially relevant conditions. 40,41,42 The success of the phosphinimide catalyst is due to its steric and electronic similarity to the Cp ligand. Figure 1.7. General structure of (a) McConville s diamide catalysts (b) phosphinimide based catalysts (c) FI catalysts based on Group 4 salicylaldiminide complexes Amidinate Based Olefin Polymerization Catalysts Group 4 complexes containing amidinates as ancillary ligands are active olefin polymerization catalysts and have been widely studied in the past (Figure 1.8). These complexes show moderate polymerization activities and excellent stereoselectivity for polymerization of higher olefins. The bis(benzamidinate) complexes (Figure 1.8, a) and half sandwich mono(benzamidinate) complexes have been reported to polymerize propylene and styrene with moderate activities. Specifically, the zirconium bis(benzamidinate) complexes polymerize propylene in a highly stereoregular manner, which can be modulated by propylene pressure to produce either atactic or isotactic polypropylene. 43 Sita and coworkers have extensively studied half sandwich acetamidinate based zirconium complexes (Figure 1.8, b), which promote stereospecific and living polymerization of 1-hexene forming isotactic, high molecular weight

24 13 polymers. 44 Guanidinate based complexes (Figure 1.8, c) are structurally similar to amidinates and some derivatives show high ethylene polymerization activity. 45 Lastly, Collins reported excellent ethylene polymerization activity of 1400 g. (mmol. hr. atm) -1 using bis(iminophosphonamide) based complexes (Figure 1.8, d). 46 Figure 1.8. Active amidinate based olefin polymerization catalysts (a) bis(benzamidinate) complexes (b) acetamidinate complexes (c) guanidinate complexes (d) bis(iminophosphonamide) based complexes. 1.4 Amidinates Amidinates are four electron donor, bidentate, anionic ligands with a general formula of RNC(R )NR (Figure 1.9) and coordinate to a metal center through the two nitrogen atoms. 47 These versatile ligands can be sterically and electronically modified by varying the substituents. The substituents on the nitrogen can be tuned to change the steric properties and control the mode of binding to a metal center. The substituent on the bridging carbon mainly affects the electronic properties of the ligand. Figure 1.9. Resonance structures of amidinates. Amidinates are closely related to other types of ligands such as guanidinates, where the substituent at the bridging carbon is a nitrogen donor (Figure 1.10).

25 14 Guanidinates are more electron donating than amidinates as the additional nitrogen can bear 1 or 2 substituents and also donate electron density through π bonding. Amidinates are isoelectronic to carboxylates but the additional substituents on the nitrogen in amidinates can be used to modify the steric and electronic properties. Figure Ligands isoelectronic to amidinates: (a) guanidinates (b) carboxalates (c) triazenates Modes of Coordination Amidinate complexes have been reported for most transition metals and some main group metals. There are several frequently observed modes of coordination to the metal center controlled by the substituents on the nitrogen and bridging carbon (Figure 1.11). The chelating metallacycle form is the most commonly observed mode followed by the bridging mode, which is found in dinuclear species with multiple metal-metal bonds. Sterically bulky substituents usually favour the chelating mode instead of the bridging mode since large substituents on the nitrogen and bridging carbon will position the lone pair of electrons on the nitrogen atoms closer together converging towards a single metal center. In contrast, small substituents will result in more parallel location of the lone pairs favouring the formation of the bridging mode. Lastly, the monodentate form is much rare but has been observed in complexes with sterically bulky substituents.

26 15 Figure Commonly observed modes of coordination: (a) chelating bidentate (b) bridging (c) monodentate Synthesis of Metal-Amidinates There are several common routes for the synthesis of metal amidinate complexes. Since, amidinates are salts of neutral amidines, few methods initially involve synthesis of the corresponding amidines. One common method introduced by Sanger involves the amidine synthesis by reacting benzonitrile with LiN(SiMe 3 ) 2 followed by trimethylsilyl chloride (Scheme 1.2). 48 Scheme 1.2. Synthesis of N,N,N -tris(trimethylsilyl)benzamidine. 48 Amidinate containing complexes can be synthesized from the corresponding amidines. Reacting trimethylsilyl substituted amidine with metal halide forms the amidinate complexes with the loss of trimethylsilyl chloride (Scheme 1.3, a). Second common method involves the reaction of protonated amidine with a metal amido species in an amine elimination reaction where the loss of secondary amine drives the reaction (Scheme 1.3, b). The success of the amidinate ligand is also due to its ease of synthesis using commercially available carbodiimides. Carbodiimide insertion into metal-nitrogen or metal-carbon bond occurs readily with organolithium reagents or metal complexes (Scheme 1.3, c). Lastly, deprotonation of amidines with alkali metals

27 16 followed by salt metathesis with metal halide precursors (Scheme 1.3, d) is also a common strategy for the synthesis of metal complexes. Due to the widespread success of the metallocene complexes as olefin polymerization catalysts, it is interesting to compare the amidinates to cyclopentadienyl ligands. The steric bulkiness of amidinates is ranked between Cp and Cp*. 49,50 In terms of electron donation to the metal center, amidinates are less electron donating compared to Cp derivatives. 51 Although electron rich ligands are desirable to stabilize the cationic metal center during polymerization, electron deficient bis(indenyl) based zirconium complexes have been reported for increased stereoselectivity in propylene polymerization. 52 Hence, amidinates are good alternatives to cyclopentadienyl ligands for use in olefin polymerization catalysts. Scheme 1.3. Common routes for synthesis of metal amidinates: (a) elimination of trimethylsilyl chloride (b) amine elimination (c) carbodiimide insertion (d) salt metathesis.

28 Scope of Thesis Group 4 complexes containing amidinates show low to moderate olefin polymerization activities and good stereoselectivity in polymerization of higher olefins. Amidinates can be sterically and electronically tuned by modifying the substituents on the nitrogen and the bridging carbon. In ferrocenyl amidinates, the placement of sterically bulky ferrocenyl substituent at the bridging carbon offers the potential for redox tunable polymerization catalysts and electronic cooperative interaction with iron center during polymerization. The work in this thesis includes synthesis and characterization of novel titanium and zirconium dichloride and dialkyl complexes containing ferrocenyl amidinates. Ethylene polymerization activity of these complexes was evaluated after activation with MAO, B(C 6 F 5 ) 3 and [Ph 3 C][B(C 6 F 5 ) 4 ] cocatalysts.

29 18 Chapter 2 Synthesis, Characterization and Ethylene Polymerization Activity of Group 4 Ferrocenyl Amidinate Complexes 2 Heading Introduction Derivatives of ferrocene have received enormous attention in the past as ligands in catalysis, polymeric materials and biomolecules. 53,54 The wide use of ferrocenes is primarily due to their high stability, low cost and versatile synthesis. Moreover, the steric bulk, rigid structure, and well behaved redox chemistry of ferrocene makes it an excellent substituent for ancillary ligands in olefin polymerization catalysts. Group 4 complexes containing ferrocene based ligands are active olefin polymerization catalysts and offer a number of advantages over conventional organic ligands (Figure 2.1). 55,56,57 Arnold and coworkers have prepared titanium and zirconium complexes with the bis(amino)ferrocenyl ligand (Figure 2.1, a), and were able to demonstrate high ethylene polymerization activity for the zirconium complex upon activation with [Ph 3 C][B(C 6 F 5 ) 4 ]. 58,59 Recently, Erker and coworkers have prepared bis(ferrocene-saliminato) group 4 metal complexes which show low to moderate activity upon activation with MAO. 60,61 Incorporation of ferrocene in the ligand backbone of a precatalyst allows one to apply the reversible redox properties of iron for the synthesis of redox tunable catalysts. 62 Modulation of the oxidation state of iron can potentially alter the electron density on the active metal center and affect the catalyst activity, comonomer incorporation and stereoregularity of the resulting polymer. Gibson has prepared titanium complexes containing ferrocenyl substituted bis(iminophenoxide) ligands and

30 19 reported significantly different conversions for ring-opening polymerization of rac-lactide by Fe III ferrocenium species in comparison to the corresponding Fe II species. 63 Figure 2.1. Active olefin polymerization catalysts containing ferrocene based ligands: (a) bis(amino)ferrocenyl ligand 58,59,64 (b) ferrocenyl dimethylsilyl substituted zirconocenes 65 (c) ferrocenyl substituted phosphinimine ligand. 66 Thus far, the study of redox tunable olefin polymerization catalysts has been limited. The Fe III ferrocenium derivative of the bis(amino)ferrocenyl zirconium complex (Figure 2.1, a) has been isolated and shown to be an active catalyst. 64 Recently, Gibson and coworkers have reported very similar ethylene polymerization activities by Fe II and their corresponding Fe III derivatives based on ferrocenyl substituted bis(imino)pyridyl complexes upon activation with MAO. 67 Heterobimetallic complexes containing the ferrocenyl fragment and group 4 metals also have the potential for electronic cooperative effects involving electron donation from the electron rich iron to the electrophilic cationic group 4 metal center during polymerization. 68 Heterobimetallic complexes containing group 4 metals and a ferrocenyl group, often as a ligand substituent linked by a carbon chain, have been shown to be active catalysts. 69 Mukaiyama and coworkers reported higher ethylene polymerization activity and higher stereoselectivity in cyclopolymerization of 1,5- hexadiene using ferrocenyldimethylsilyl substituted zirconocene precatalysts (Figure 2.1, b) compared to Cp 2 ZrCl 2. 65,68,70

31 20 Group 4 complexes containing amidinates as ancillary ligands are active catalysts for stereoselective olefin polymerization since amidinates are sterically and electronically similar to cyclopentadienyl ligands. Based on the preceding discussion, the introduction of a ferrocenyl group as an electron donating substituent at the central carbon of the amidinate backbone, allows the synthesis of novel ferrocene substituted amidinates. Moreover, in-plane orientation of the substituted Cp ring of ferrocene relative to the amidinate NCN plane may offer the possibility of additional electron donation by pi bonding. In this work, titanium and zirconium bis(amidinate) and half sandwich zirconium mono(amidinate) dichloride and dialkyl complexes were synthesized. The reactivity of the dialkyl complexes with Lewis acids was investigated. The metal complexes were evaluated as ethylene polymerization catalysts and the stability of the corresponding complexes containing Fe III derivatives was investigated using cyclic voltammetry. 2.2 Results and Discussion Synthesis and Characterization Synthesis of Amidine and Amidinate Derivatives N,N -dicyclohexyl ferrocenyl amidine (1a) was synthesized by a one-pot procedure reported by Arnold and Hagadorn. 71 A pentane solution of t BuLi was added to a THF/hexanes solution of ferrocene resulting primarily in the formation of monolithioferrocene (Scheme 2.1). Subsequently, this solution was treated with dicyclohexyl carbodiimide followed by hydrolysis with water. Sublimation of the crude reaction mixture was required to remove the unreacted ferrocene and recrystallization from hexanes was performed to selectively remove the ferrocene linked bis-amidine side product. 72

32 21 Scheme 2.1. Synthesis of ferrocenyl amidine. 71 The original procedure by Hagadorn and Arnold reported a large scale synthesis of 1a starting with 100 g of ferrocene but the reaction was scaled down to 30% and optimized for the yield. The conditions used for the lithiation of ferrocene were the major factor contributing to the conversion to 1a. Kagan and coworkers reported the effect of several conditions such as the solvent, type of organolithium reagent used for deprotonation, rate of reagent addition and reaction time on the relative ratio of ferrocene, monolithioferrocene (FcLi) and dilithioferrocene (FcLi 2 ) in solution. 73 Based on this work, conditions were selected to maximize FcLi formation and minimize the formation of undesirable FcLi 2. This resulted in less formation of ferrocene linked bis(amidine) and 1a was obtained in 61% isolated yield. In the previous one-pot procedure, the moderate yields are partly due to the decomposition of 1a during the sublimation at 65 C required to remove the unreacted ferrocene. Attempts to purify 1a by column chromatography were unsuccessful as the product decomposed on silica and alumina. Instead, the intermediate FcLi was selectively isolated as reported by Zanello and the sublimation was no longer required. 74 Similar to the previous method, the isolated FcLi was reacted with dicyclohexyl carbodiimide and hydrolyzed with water. This resulted in an overall yield of 85% starting

33 22 from FcLi and 68% yield starting from ferrocene, an improvement from the previous method. The synthesis of N,N -diisopropyl ferrocenyl amidine (1b) was performed by slightly modified procedure by starting from bromoferrocene. 1b was synthesized by treatment of bromoferrocene with n BuLi, subsequently with diisopropyl carbodiimide and followed by hydrolysis with water resulting in 70% isolated yield (Scheme 2.2). The 1 H NMR spectrum of 1b is consistent with the structure where the two triplet resonances at 3.93 ppm and 4.24 ppm are observed for the substituted Cp ring of ferrocene and a sharp singlet at 3.96 ppm for the unsubstituted Cp ring. Also, 1b shows two broad resonances and two doublets for the methine and methyl of isopropyl groups, respectively, consistent with unsymmetrical nature of the structure. Scheme 2.2. Synthesis of ferrocenyl amidine with isopropyl substituents on the nitrogen. In summary, three similar methods were used to prepare the ferrocenyl amidine derivatives, 1a and 1b. The first method was a one pot synthesis resulting in moderate yields. The second method resulted in slightly higher yields but involved isolation of the pyrophoric FcLi intermediate. The third method involved use of commercially available monosubstituted ferrocene source and in situ generation of FcLi. Ferrocenyl amidines 1a and 1b were easily synthesized since diisopropyl and dicyclohexyl carbodiimide readily undergo insertion into carbon-lithium bond in FcLi.

34 23 However, attempts to synthesize ferrocenyl amidine with a -SiMe 3 substituent on the nitrogen have been reported to be unsuccessful using similar strategy of bis(trimethylsilyl) carbodiimide insertion into FcLi due to the lower electrophilicity of the carbodiimide. 75,76 The cyclohexyl amidine derivative (1a) was used for the synthesis of metal complexes since the more bulky cyclohexyl group provides greater steric protection of the nitrogen atom in comparison to the isopropyl group. Protection of the nitrogen atom is an important requirement for a polymerization catalyst since it is susceptible to attack by Lewis acidic cocatalysts. Lithium ferrocenyl amidinate (1d) was prepared by deprotonating 1a with MeLi in hexanes by modified literature procedure. 71 The synthesis of titanium complexes can be achieved by Me 3 SiCl elimination instead of more common routes such as salt metathesis. Therefore, mixed -SiMe 3 and cyclohexyl substituted amidine (1c) was also synthesized by treatment of lithium ferrocenyl amidinate with trimethysilylchloride (Scheme 2.3). The 1 H NMR spectrum of 1c is consistent with the structure where the SiMe 3 resonance appears at 0.46 ppm. Scheme 2.3. Synthesis of trimethylsilyl substituted ferrocenyl amidine, 1c. A crystal structure determination resulted in the molecular structure of 1b depicted in Figure 2.2 and the selected bond distances and angles of 1b are shown in Table 2.2. The structure contains localized CN single and double bonds with a distance

35 24 of 1.362(2) Å and 1.283(2) Å for N1-C7 and N2-C7, respectively. The Cp plane of the ferrocene intersects the NCN plane at 41.1(3). The molecular structure of one of the two independent molecules in the asymmetric part of the unit cell of 1c and the selected bond distances and angles are shown in Figure 2.2 and Table 2.3, respectively. Similar to 1b, localized single and double CN bond distances are also observed. The Cp plane of the ferrocene intersects the NCN plane at 15.2(5). Figure 2.2. Molecular structure of [ i PrNC(Fc)N i Pr]H, 1b and [CyNC(Fc)NCy]SiMe 3, 1c. Thermal ellipsoids are shown at 50% probability. Hydrogen atoms have been omitted for clarity. Table 2.2. Selected bond distances (Å) and bond angles ( ) for [ i PrNC(Fc)N i Pr]H, 1b. N1-C (2) N1-C7-N (2) N2-C (2) N2-C7-C (2) C7-C (3) N1-C7-C (2) C5-N (2) C7-N2-C (2) C2-N (2) C7-N1-C (2)

36 25 Table 2.3. Selected bond distances (Å) and bond angles ( ) for [CyNC(Fc)NCy]SiMe 3, 1c. N3-C (2) N3-C1-N (1) N4-C (2) N3-C1-C (1) N3-C (2) N4-C1-C (1) N3-Si (1) C9-N4-C (2) N4-C (2) C2-N3-C (1) C1-C (2) C2-N3-Si (1) Si4-N3-C (1) Synthesis of Group 4 Metal Dichloride Complexes The synthesis of group 4 metal bis(amidinate) complexes was initially attempted using M(NR 2 ) 4 (M = Ti, Zr; R = Me, Et) precursors. Reacting Ti(NEt 2 ) 4 with 2 equiv of 1a resulted in no reaction at room temperature or at higher temperatures (80 C). However, 2 equiv of 1a reacted with Zr(NMe 2 ) 4 (THF) 2 at elevated temperatures (~80 C) but did not selectively result in the bis(amidinate) complex. The 1 H NMR spectrum is consistent with the formation of mono, bis- and tris- amidinate products. Zirconium complexes containing tris(amidinates) are known in the literature, even though the steric bulkiness may suggest otherwise. 77 Bis(amidinate) complexes were successfully prepared by treatment of M(NMe 2 ) 2 Cl 2 (M = Zr 2THF; M = Ti; Scheme 2.4) with 2.05 equiv of 1a after refluxing in toluene. The zirconium bis(amidinate) dichloride (2a) was isolated as an orange solid in high yields (85%) by trituration of the resulting reaction mixture with hexanes followed by subsequent washings with hexanes and ether. Hexane washings are required to completely remove the unreacted amidine, 1a. The titanium bis(amidinate) dichloride (2b) was isolated in 77% yield as a purple solid using the same workup procedure as for 2a. In this case, a color change from dark orange to purple was observed after refluxing overnight in toluene.

37 26 Scheme 2.4. Synthesis of titanium and zirconium bis(amidinate) complexes, 2a and 2b. The 1 H NMR spectra are consistent with the formation of 2a and 2b. The assignments of the NMR data were assisted by dept135 and HSQC experiments. Sharp resonances for the unsubstituted Cp ring of the ferrocene for both bis(amidinate) complexes are similar at ca ppm. These resonances are shifted downfield by 0.3 ppm compared to the protonated amidine, which is consistent with the decreased electron density in the ligand caused by withdrawal of electron density by the group 4 metal center. The resonances for substituted Cp rings and the alpha proton of the cyclohexyl ring are observed in the ppm region. Broad resonances are observed in ppm region for the axial and equatorial protons of the cyclohexyl groups. The NMR spectra of 2a and 2b show different fluxional behaviour in solution, a feature commonly observed with other bis(amidinate) complexes. 78,79,80 The C 2 symmetry of the bis(amidinate) complexes results in two types of nitrogens depending on their position (either cis or trans) relative to the chlorides. In the NMR spectra, the resonances observed for the cyclohexyl groups reflect the nitrogen environment. Specifically, the resonance for the alpha proton of the cyclohexyl group in 2a shows a multiplet at 4.12 ppm integrating to 4 protons, whereas, 2b shows two multiplets at 4.41 ppm and 3.99 ppm integrating to 2 protons each. In addition, the alpha carbon of the cyclohexyl group in the 13 C NMR spectra of 2a shows a single resonance at 57 ppm,

38 27 however, 2b shows two resonances at 61 ppm and 59 ppm. The two separate signals for alpha protons and alpha carbons in 2b are as expected and show that there are two unique cyclohexyl groups depending on their attachment to the cis or trans nitrogens. However, in 2a all of the cyclohexyl groups are equivalent, as each of the two amidinate ligands are interconverting at room temperature (Scheme 2.5). In fact, Xue and coworkers also reported this same difference in fluxional behavior based on the metal center in bis(amidinate) complexes of the type, M(NMe 2 ) 2 (CyNC(Me)NCy) 2 (M = Zr; Ti). 80 The size of the metal center has been found to influence this behaviour, as a smaller titanium center restricts this fluxionality, whereas the larger zirconium center allows the ligand to readily interconvert. Secondly, Richeson and coworkers have proposed that bulky substituents on the central carbon decreases the chance of this interconversion. 81 This trend is evident in the titanium bis(amidinate) complexes since Ti[ i PrNC(NMe 2 )N i Pr] 2 Cl 2 with a less bulky NMe 2 does show fluxional behaviour 82, whereas, 2b with a bulky ferrocene at the central carbon does not. The interconversion of the amidinate ligands in 2a may proceed through a dissociative pathway or an internal twist mechanism as observed for other bis- chelating octahedral complexes. The dissociative pathway has been observed but the Bailar-twist mechanism involving a trigonal prismatic intermediate has been commonly reported for the ligand interconversion in similar bis(amidinate) and bis(ketenimine) group 4 metal complexes based on the activation entropy and enthalpy calculations. 83,78,80,84

39 28 Scheme 2.5. Rapid interconversion of 2a at room temperature. An attempt was made to study this behaviour further by low temperature 1 H NMR. Upon cooling a CD 2 Cl 2 solution of 2a to 200K, the 1 H NMR spectrum reveals broadening of the alpha proton resonance and splitting of the resonance corresponding to the protons on the substituted Cp ring (Figure 2.3). The splitting is likely caused by the restricted rotation of the ferrocene moiety on the amidinate backbone at low temperatures and this was also observed for 2b. The 1 H NMR spectra do not show any major changes from 298K to 230K and broadening of the alpha proton resonance starts to occur noticeably at 230K. The kinetic parameters for the interconversion were not determined as the slow exchange limit could not be reached. 200K 230K 298K Figure H NMR spectrum of 2a from ppm shows splitting of substituted Cp protons and broadening of the cyclohexyl protons at low temperature.

40 29 The molecular structures of 2a and 2b are shown in Figure 2.4 and Figure 2.5, respectively, with the selected bond distances and angles in Table 2.4. The metal center is in a pseudo octahedral environment surrounded by 4 nitrogen atoms from the amidinate ligands and two chloride ligands. The bidentate amidinate ligands are positioned cis to each other. The bonding parameters within the two amidinate ligands are not significantly different. The metal center, two nitrogen atoms and the central carbon of the amidinate ligand lie in a single plane. The four nitrogens can be divided into two types based on their relative orientation to each other, cis-ncy (N1, N4) and trans-ncy (N2, N3) groups. In complex 2b, the Ti-N bond distance in cis-ncy groups (2.079(2) Å, 2.103(2) Å) are longer than trans-ncy groups (2.041(2) Å, 2.037(2) Å) due to the trans effect of two chlorides as observed in other group 4 bis(amidinate) complexes. 79 This trend is not evident for Zr-N bond distances in 2a. There are no significant differences in C alpha-cyclohexyl -N bond distance between cis-ncy and trans-ncy groups in both complexes. The amidinate bite angle in 2b (63.94(8), 63.76(8) ) is larger than the bite angle in 2a (60.02(8), 59.86(8) ). The charge delocalization and partial double bond character in the amidinate NCN backbone is evident by the C-N bond distances in 2a and 2b which range from 1.323(3) Å to 1.350(4) Å. The M-N-C alpha-cyclohexyl bond angle reflects the steric crowding around the metal center. The average M-N-C alpha-cyclohexyl angle in 2a is and is smaller than the average angle in [CyNC(Me)NCy] 2 ZrCl 2 (142.4 ). This shows that the bulky ferrocenyl group pushes the cyclohexyl groups closer to the metal center compared to the methyl group in acetamidinate complexes. The M-Cl bond distances in 2a (2.4311(7) Å, (9) Å) are longer than 2b (2.3069(8) Å, (7) Å). The Cl1-M-Cl2 bond angle in both complexes is similar (93.92(3) for 2a; 93.84(3) for 2b) and smaller compared to Cp 2 ZrCl 2 (97.1 ). 85 The Cl1-M-Cl2 angle in 2a is similar compared to [CyNC(Me)NCy] 2 ZrCl 2 (93.1(1) ) and

41 30 significantly different compared to [Me 3 SiNC(Ph)NSiMe 3 ] 2 ZrCl 2 ( ). 79,86 Richeson describes the bis(amidinate) complexes as pseudotetrahedral to compare them to structurally similar metallocene complexes. 79 In this description, two of the vertices are defined as the vectors that bisect the amidinate ligand at the central carbons and the other two vertices as the Zr-Cl vectors. 87 Using this description, C amidinate -M-C amidinate angle is similar in both complexes ( for 2a; for 2b) and significantly smaller compared to Cp centroid -Zr-Cp centroid angle of 134 in Cp 2 ZrCl 2. Figure 2.4. Molecular structure of [CyNC(Fc)NCy] 2 ZrCl 2, 2a. Thermal ellipsoids are shown at 50% probability. Hydrogen atoms have been omitted for clarity.

42 31 Figure 2.5. Molecular structure of [CyNC(Fc)NCy] 2 TiCl 2, 2b. Thermal ellipsoids are shown at 50% probability. Hydrogen atoms have been omitted for clarity. Table 2.4. Selected bond distances (Å) and bond angles ( ) for [CyNC(Fc)NCy] 2 ZrCl 2, 2a and [CyNC(Fc)NCy] 2 TiCl 2, 2b. 2a (M = Zr) 2b (M = Ti) M-Cl (7) (8) M-Cl (9) (7) M-N (2) 2.079(2) M-N (2) 2.041(2) M-N (2) 2.037(2) M-N (3) 2.103(2) C1-N (4) 1.329(3) C1-N (3) 1.347(3) C1-C (4) 1.484(4) C2-N (4) 1.363(3) C2-N (3) 1.323(3) C2-C (4) 1.475(4) N1-M-N (8) 63.94(8) N3-M-N (8) 63.76(8) Cl1-M-Cl (3) 93.84(3) N1-C1-N (24) (22) N3-C2-N (25) (22) M-N1-C (19) (16) M-N2-C (18) (16) M-N3-C (19) (16)

43 32 2a (M = Zr) 2b (M = Ti) M-N4-C (19) (17) Dihedral Angle: N1-C1-N2, Cp ferrocene 48.2(3) 47.9(2) N3-C2-N4, Cp ferrocene 38.2(3) 36.1(2) The zirconium mono(amidinate) half sandwich complexes were synthesized by salt metathesis between Cp ZrCl 3 (Cp = Cp; Cp*) and 1 equiv of 1d in THF solvent to afford Cp zirconium mono(amidinate) (2c) and Cp* zirconium mono(amidinate) (2d) dichloride complexes in 60% and 55% isolated yields, respectively (Scheme 2.6). It is noteworthy to mention that initial attempts for synthesis of these complexes using sodium amidinate salt, Na[CyNC(Fc)NCy], were unsuccessful. The two half sandwich complexes show similar solubilities except 2d shows a greater solubility in hexanes than 2c. The half sandwich mono(amidinate) complexes are known to undergo ligand flipping at the metal center at room temperature which was investigated with unsymmetrical amidinates by Sita and coworkers. 88 The 1 H NMR spectra of 2c and 2d show the resonances for the unsubstituted Cp ring of ferrocene at ca ppm, which is shifted downfield from the protonated amidine, 1d. Only one resonance for the alpha proton of the cyclohexyl ring is observed due to the C S symmetry of the complex. Furthermore, this resonance appears as a triplet of triplets due to difference in coupling between the neighbouring axial and equatorial protons on the cyclohexyl ring. The resonances for protons of the Cp ring and the methyl groups of Cp* ring bound to the zirconium are shifted downfield compared to Cp ZrCl 3.

44 33 Scheme 2.6. Synthesis of half sandwich mono(amidinate) zirconium complexes, 2c and 2d, via salt metathesis. The molecular structure of 2c and 2d are shown in Figure 2.6 and Figure 2.7, respectively, with the selected bond distances and angles in Table 2.5. Complex 2d contains two molecules in the asymmetric unit cell. Both complexes display a pseudotetrahedral geometry around the zirconium center. Complex 2c shows asymmetric binding to the metal center based on the different Zr1-N1 (2.224(1) Å) and Zr1-N2 (2.110(1) Å) bond distances. However, 2d shows symmetric binding since the Zr1-N1(2.215(3) Å) and Zr1-N2 (2.217(3) Å) bond distances are identical. The C6-N1 and C6-N2 bond distances in the amidinate backbone in 2c and 2d are indicative of partial double bond character and the amidinate ring forms a four membered nearly planar metallacycle. The amidinate bite angle in 2c (59.99(4) ) and 2d (60.18(9) ) are very similar. The ferrocene Cp plane is tilted (52.09(15) for 2c and 52.39(11) for 2d) relative to the amidinate NCN plane indicating no π conjugation.

45 34 Figure 2.6. Molecular structure of CpZr[CyNC(Fc)NCy]Cl 2, 2c. Thermal ellipsoids are shown at 50% probability. Hydrogen atoms have been omitted for clarity. Figure 2.7. Molecular structure of Cp*Zr[CyNC(Fc)NCy]Cl 2, 2d. Thermal ellipsoids are shown at 50% probability. Hydrogen atoms have been omitted for clarity.

46 35 Table 2.5. Selected bond distances (Å) and bond angles ( ) for CpZr[CyNC(Fc)NCy]Cl 2, 2c, and for Cp*Zr[CyNC(Fc)NCy]Cl 2, 2d. 2c (Cp = Cp) 2d (Cp = Cp*) Zr1-Cl (4) (9) Zr1-Cl (4) (9) Zr1-Cp centroid (8) 2.211(1) Zr1-N (1) 2.215(3) Zr1-N (1) 2.217(3) C6-C (2) 1.485(4) C6-N (2) 1.334(4) C6-N (2) 1.343(4) Cl1-Zr1-Cl (2) 89.80(3) N1-Zr1-N (4) 60.18(9) Cl2-Zr1-N (3) 91.56(7) Cl1-Zr1-N (3) 87.69(7) Zr1-N1-C (9) (20) Zr1-N2-C (10) (20) Zr1-N1-C (9) 93.97(19) Zr1-N2-C (8) 93.62(19) N1-C6-N (12) (27) Cp centroid -Zr1-N (4) (8) Cp centroid -Zr1-N (4) (8) Dihedral Angle: N1-C6-N2, Cp ferrocene 52.09(15) 52.39(11) The synthesis of similar titanium mono(amidinate) complexes was found to be unsuccessful using several different strategies. Salt metathesis of Cp TiCl 3 (Cp = Cp; Cp*) with Li[CyNC(Fc)NCy], Li[CyNC(Fc)NCy]. Et 2 O or Na[CyNC(Fc)NCy] did not result in a single major product after attempting at several different conditions (Scheme 2.7). The 1 H NMR spectrum of the reaction mixture with CpTiCl 3 shows two major peaks at 6.55 ppm and 6.10 ppm along with other minor peaks in this region corresponding to the protons of the Cp ring bound to titanium. Similarly, for Cp*TiCl 3 the reaction mixture shows two major peaks at 2.08 ppm and 1.94 ppm, and other minor peaks in this region for the methyl groups of the Cp* ring. The ferrocene Cp region from 4.0 to 4.5 ppm is unclear due to presence of overlapping multiplets. Based on these results, the reaction is not selective towards monosubstituted product and there is also formation of

47 36 bis(amidinate) half sandwich product. Further evidence is needed to confirm this hypothesis. An alternative synthesis was attempted by reacting the trimethylsilyl substituted amidine with Cp TiCl 3 to form the amidinate complex with trimethylsilyl chloride elimination but no reaction was observed at room temperature. Upon heating the reaction mixture, the 1 H NMR spectrum showed decomposition of amidine and a similar product distribution as previous attempt by salt metathesis. Scheme 2.7. Attempted synthesis of half sandwich mono(amidinate) titanium complexes via salt metathesis was unsuccessful Synthesis of Group 4 Metal Dialkyl Complexes The dialkyl complexes (L n MR 2, R = Me, CH 2 Ph) can be prepared from the dichloride complexes (L n MCl 2 ) by treatment with 2 equiv of alkyllithium or Grignard reagents. The zirconium bis(amidinate) dimethyl complex, 3a, was prepared by treatment of a toluene solution of 2a with 2 equiv of MeLi in 75% isolated yield. The 1 H and 13 C NMR spectra show the Zr-Me resonance at 1.12 ppm and 45 ppm, respectively. The resonance for the unsubstituted Cp ring of the ferrocenyl group in 3a is shifted upfield shift from 4.30 ppm to 4.15 ppm relative to 2a since methyl groups are more electron donating than chlorides, resulting in less electron withdrawal from the ligand by the group 4 metal center. The 1 H and 13 C NMR spectra show a single resonance for

48 37 alpha proton and alpha carbon of the cyclohexyl group indicating interconversion of the amidinate ligand as observed for the dichloride complex. The molecular structure of 3a is shown in Figure 2.8 with the selected bond distances and angles in Table 2.6. Complex 3a, structurally similar to 2a, shows a monomeric pseudo octahedral geometry, where the metal center, two nitrogen atoms of the amidinate ligand and the central carbon lie in a single plane. The four nitrogen atoms can be divided into two types based on their relative orientation to each other, cis- NCy (N1, N4) and trans-ncy (N2, N3) groups. The Zr-N bond distances in cis-ncy groups (2.246(5) Å, 2.235(5) Å) are shorter than trans-ncy groups (2.273(4) Å, 2.262(5) Å). The Zr-N bond distances in 3a are longer than the bond distances in 2a since methyl groups are more electron donating than chlorides. The C47-Zr-C48 bond angle in 3a (88.22(21) ) is smaller than the Cl1-Zr-Cl2 bond angle in 2a (93.84(3) ). The C47-Zr-C48 angle is also smaller compared to other dimethyl complexes such as [CyNC(Me)NCy] 2 ZrMe 2 (92.4(3) ) and Cp 2 ZrMe 2 (95.6 ). 79,89 The amidinate bite angle in 3a (58.49(16), 58.45(17) ) is smaller than the bite angle in 2a (60.02(8), 58.86(8) ). The charge delocalization and partial double bond character in the amidinate NCN backbone is evident by the average C-N bond distances of 1.33 Å in 3a.

49 38 Figure 2.8. Molecular structure of [CyNC(Fc)NCy] 2 ZrMe 2, 3a. Thermal ellipsoids are shown at 50% probability. Hydrogen atoms have been omitted for clarity. Table 2.6. Selected bond distances (Å) and bond angles ( ) for [CyNC(Fc)NCy] 2 ZrMe 2, 3a. Zr1-C (6) N1-Zr1-N (16) Zr1-C (6) N3- Zr1-N (17) Zr1-N (5) C47- Zr1-C (21) Zr1-N (4) N1-C1-N (49) Zr1-N (5) N3-C2-N (50) Zr1-N (5) C1-N (6) Zr1-N1-C (35) C1-N (6) Zr1-N2-C (33) C1-C (8) Zr1-N3-C (35) C2-N (7) Zr1-N4-C (36) C2-N (7) C2-C (8) Similarly, treatment of the half sandwich Cp* zirconium mono(amidinate) dichloride complex, 2d, with 2 equiv of MeLi in toluene afforded the dimethyl complex, 3d, in 71 % isolated yield. The NMR spectrum is consistent with the structure where the Zr-Me resonance appears at ppm and 47 ppm in the 1 H and 13 C NMR spectra, respectively. The resonances for the methyl groups of Cp* ring and unsubstituted Cp ring of the ferrocenyl group are shifted upfield compared to the dichloride complex. It is