The Pennsylvania State University. The Graduate School SYNTHESIS AND CHARACTERIZATION OF LONG CHAIN BRANCHED

The Pennsylvania State University The Graduate School Department of Materials Science and Engineering SYNTHESIS AND CHARACTERIZATION OF LONG CHAIN BRANCHED ISOTACTIC POLYPROPYLENE VIA A METALLOCENE CATALYST AND T-REAGENT A Thesis in Materials Science and Engineering by Justin August Langston 2007 Justin August Langston Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy December 2007

The thesis of Justin Langston was reviewed and approved* by the following: Tze-Chiang Chung Professor of Materials Science and Engineering Thesis Advisor Chair of Committee Ralph H. Colby Professor of Materials Science and Engineering Ronald Hedden Assistant Professor of Materials Science and Engineering Ayusman Sen Professor of Chemistry Gary L. Messing Distinguished Professor of Ceramic Science and Engineering Head of the Department of Materials Science and Engineering *Signatures are on file in the Graduate School

ABSTRACT ii Long chain branched isotactic polypropylene (LCBPP) was synthesized via the combination of a metallocene catalyst, p-(3-butenyl)styrene (T-reagent), and hydrogen. T-reagent, in the presence of hydrogen, simultaneously served as comonomer and chain transfer agent, resulting in a high molecular weight, branched polypropylene. The preparation of LCBPP by the addition of T-reagent was a one-step procedure in which the metallocene catalyst remained highly reactive. To understand the structure-property relationships, a series of LCBPP was prepared with similar weight-average molecular weights of about 250,000 g/mol and branch densities ranging from 0 (linear ipp) to 3.3 branches per 10,000 carbon. SEC equipped with triple detectors revealed the presence of high molecular weight branches whose density depended on the concentration of T- reagent, hydrogen, and propylene. Melt properties were examined by small amplitude oscillatory shear and extensional flow measurements. The zero-shear viscosities of LCBPP displayed a systematic increase as branch density increased despite similar molecular weights. Strain hardening was observed in extensional flow of LCBPP. Because of these melt properties, the LCBPP and other branched polypropylenes displayed the ability to retain orientation after high temperature (T > T m ) deformations.

TABLE OF CONTENTS iii LIST OF FIGURES...v LIST OF TABLES...x ACKNOWLEDGEMENTS...xi Chapter 1 Introduction to Branched Polyolefins...1 1.1 Introduction...1 1.2 Branched Polyethylene...3 1.2.1 Low Density Polyethylene...4 1.2.2 Linear Low Density Polyethylene...5 1.2.3 Long Chain Branched Polyethylene...7 1.2.4 Well-Defined Branched Polyethylene...10 1.3 Branched Polypropylene...11 1.3.1 Macromonomer Approach...12 1.3.2 Non-Conjugate α,ω-diene Approach...16 1.3.3 Post-polymerization Chemical Modification Approach...21 1.3.4 Coupling Approach...27 1.4 Conclusions...28 Chapter 2 Characterization of Long Chain Branched Polymers...30 2.1 Introduction...30 2.2 Nuclear Magnetic Resonance Spectroscopy...30 2.3 Size Exclusion Chromatography with Triple Detectors...33 2.4 Rheological Techniques...43 2.5 Conclusions...55 Chapter 3 Synthesis of Long Chain Branched Polypropylene via the T-reagent Approach...57 3.1 Introduction...57 3.2 Experimental...60 3.2.1 Instrumentation and Materials...60 3.2.2 Synthesis of p-(3-butenyl)styrene (T-reagent)...61 3.2.3 Copolymerization of Propylene and 4-phenylbut-1-ene...62 3.2.4 Synthesis of Long Chain Branched Isotactic Polypropylene (LCBPP)...62 3.2.5 Hydrogenation of LCBPP 100...63 3.2.6 Size Exclusion Chromatography...63 3.2.7 Small Amplitude Dynamic Oscillatory Shear...64 3.3 Synthesis of LCBPP...65

3.4 Bulk Polymerizations of Propylene...74 3.5 Catalyst Effect...76 3.6 Hydrogen Effect...84 3.7 Conclusions...86 Chapter 4 Rheological Characterization of LCBPP...88 4.1 Introduction...88 4.2 Experimental...89 4.2.1 Small Amplitude Oscillatory Shear...89 4.2.2 Extensional Flow Measurements...89 4.3 Small Amplitude Oscillatory Shear...90 4.4 Extensional Flow Measurements...99 4.5 Conclusions...101 Chapter 5 Application of LCBPP as a Dielectric Material...103 5.1 Introduction...103 5.2 Experimental...106 5.2.1 Materials...106 5.2.2 Preparation of Films...107 5.2.3 X-Ray Diffraction...107 5.2.4 Breakdown Strength...108 5.3 Degree of Crystallinity...108 5.4 Orientation...112 5.5 Breakdown Strength...119 5.6 Conclusions...122 Chapter 6 Conclusions and Suggestions...124 6.1 Summary of Present Work...124 6.2 Final Conclusions...126 6.3 Potential Applications...127 6.4 Suggested Future Work...129 6.4.1 Synthesis of Long Chain Branched Polypropylene...129 6.4.1.1 Macromonomer Incorporation...129 6.4.1.2 Chain-transfer to Polymer Containing Pendant Styrene Groups...129 6.4.1.3 Compatible Catalysts...130 6.4.1.4 T-reagent Design...131 6.4.2 Rheological Measurements of Long Chain Branched Polypropylene...131 Bibliography...133 iv

LIST OF FIGURES v Figure 1-1: Examples of possible branched structures. (A) linear chain, (B) long chain comb branching branches, (C) star branching, (D) long chain random branching, and (E) short chain branching...2 Figure 1-2: Representations of polyethylene backbone structures for various commercially important products...4 Figure 1-3: Constrained geometry catalyst (CGC) 20....6 Figure 1-4: Comonomer distributions as measured by temperature rising elution fraction chromatography (TREF) 18...6 Figure 1-5: Scheme for the preparation of LCBPE via macromonomer incorporation...7 Figure 1-6: Complex viscosities of LCBPE prepared at various pressures. Pressures in bar are indicated in the legend before the sample identification. The zero-shear viscosities of comparable linear polyethylenes are also included 24...9 Figure 1-7: Complex viscosities of various polyethylenes at 190 C. Sample A5, Mw = 115,000 g/mol, MWD 2.8 and B2, Mw 153,000 g/mol, MWD 3.6. Sample C3, Mw 117,000 g/mol, MWD 3.6 and D1, Mw 114,000 g/mol, MWD 2.2. Reference material Ref-1, Mw 380,000 g/mol, MWD 2.1 and Ref- 2, Mw 100,000 g/mol, MWD 3.8. 23...10 Figure 1-8: Generalized scheme for the formation of LCBPP by the incorporation of macromonomer...12 Figure 1-10: The log-log plot of the radius of gyration, R g, as a function of weight -average molecular weight as measured by SEC-MALLS (C) 33. A linear PP is plotted with C for comparison. Additionally, the differential weight fraction (A) and the branching index, R g R, (B) are shown....15 g. br g, l Figure 1-11: General scheme for the formation of LCBPP from the copolymerization on α,ω-diene and propylene...16 Figure 1-12: Molecular weight distributions (MWD) of LCBPP prepared by the copolymerization of 1,9-decadiene and propylene 47, where PP1 was a propylene homopolymerization and PP2 through PP5 were copolymerizations of propylene and diene. The concentration of 1,9-decadiene increased from 0.177 to 3.54 mmol/l from PP2 to PP5...17 M

Figure 1-13: Complex viscosities of LCBPP at 190 C prepared by the compolymerization of 1,9-decadiene and propylene 47, where PP1 through PP4 are the same samples shown in Figure 1-12...18 Figure 1-14: Dynamic moduli and complex viscosity of a LCBPP at 200 C prepared by the copolymerization of 1,9-decadiene and propylene 46. This data is representative of the gel-like behavior of LCBPP prepared containing higher concentrations of diene comonomer...19 Figure 1-15: Elongation stress growth coefficients of a LCBPP at 180 C prepared by the copolymerization of 1,9-decadiene and propylene (PP-B) 50, where the shear stress growth coefficients, multiplied by 3, are shown as solid lines and the Hencky strain rates are indicated. PP-A and PP-C are a linear isotactic polypropylene and a LCBPP prepared by electron beam irradiation, respectively...21 Figure 1-16: Mechanism for the formation of LCBPP via reactive extrusion in the presence of a polyfunctional monomer and an iniferter 53....23 Figure 1-17: The effect of peroxide and polyfunctional monomer concentrations on the melt flow index at 230 C (Data from reference 52). The polyfunctional monomer was pentaerythritol triacrylate (PETA) and the peroxide was 2,5-dimethyl-2,5(t-butylperoxy)hexane peroxide with concentrations shown in the legend 52. The melt flow index of the starting material was 17 g/10min...24 Figure 1-18: Dynamic moduli for LCBPP at 180 C prepared by reactive extrusion utilizing an iniferter, tetraethyl thiuram disulfide. The concentration of polyfunctional monomer increased from 6 to 12 g/kg in M1 to M4. PP is the linear material before chemical modification 53....25 Figure 1-19: MWD of LCBPP prepared by electron beam irradiation. The irradiation doses in kgy are indicated by the number behind PP 55....26 Figure 1-20: Complex viscosities of LCBPP at 180 C prepared by electron beam irradiation. The irradiation doses in kgy are indicated by the number behind PP 55....27 Figure 1-21: The synthetic scheme for the formation of LCBPP by coupling of functionalized polypropylene 61...28 Figure 2-1: An expanded 13 C NMR (bottom) and DEPT NMR (top) spectra of a LCBPP from the incorporation of macromonomer prepared in situ. The peaks marked with an asterisk at 44.88, 44.74, 44.08, and 31.74 ppm are peaks assigned to branch points 33....32 vi

Figure 2-2: Chemical shift assignments for 13 C NMR experiments. (I) Macromonomer insertion after 1,2-insertion. (II) Macromonomer insertion after 2,1-misinsertion. (III) Macromonomer insertion after a 1,3- misinsertion 33...32 Figure 2-3: Calculated behavior for the radius of gyration as a function of molecular weight for a M b of 25,000 g/mol by Eq. 2.4 and Eq. 2.6 71. The solid line represents a linear polymer while the dashed line represents the trifunctional branching and the dotted line represents the tetrafunctional branching....36 Figure 2-4: Mark-Houwink plot of LCBPE obtained from high temperature SEC with triple detection. LCB density increases from HDB1 to HDB4 74...40 Figure 2-5: Mark-Houwink plot of SCBPE obtained by high temperature SEC with triple detection. SCB frequency increases from LDB1 to LDB3 74....41 Figure 2-6: Mark-Houwink plot of LCBPP obtained by high temperature SEC with triple detection. PP-A is a linear polypropylene, PP-B is a LCBPP prepared by diene copolymerization, and PP-C is a LCBPP prepared by postpolymerization chemical modification 50...42 Figure 2-8: The dynamic moduli of a 210,000 g/mol linear polybutadiene before (line) and after (hollow symbols) hydrogenation at 130 C 29...46 Figure 2-9: The dynamic moduli of linear (filled) and star (open) hydrogenated polybutadienes at 190 C 29. The molecular weight of the arm was 174,000 g/mol for 3HPB-2, 41,700 g/mol for 3PHB-1, and 80,300 g/mol for 4PHB-1....49 Figure 2-10: Complex viscosity of a linear (PEL), star (PES), and comb (PEC) hydrogenated polybutadiene at 190 C 30....51 Figure 2-11: Diagram of the extensional viscosity fixture. One cylinder rotates around its own axis while rotating around the axis of the other cylinder...53 Figure 2-12: Diagram depicting the deformation of the sample at several Hencky strains in an extensional viscosity fixture (reproduced from TA Instruments literature The ARES-EVF: Option for Measuring Extensional Viscosity of Polymer Melts by Frank.)...53 Figure 2-13: The elongational behavior of a LCBPE and LLDPE at 150 C 85....54 Figure 2-14: The strain hardening behavior of a LCBPP observed during elongational flow 60...55 vii

Figure 3-1: Generalized scheme for the preparation of a chain-end functionalized polypropylene through the use of a functionalized styrenic chain-transfer agent. The substituent R could be any number of protected functional groups including -OH, -Cl, or NH 2...57 Figure 3-2: Scheme for chain-transfer to a styrenic chain transfer agent in the presence of hydrogen...59 Figure 3-3: Hypothesized scheme to prepare macromonomers containing the reactive allyl end-group through a functionalized chain transfer agent....60 Figure 3-4: Scheme for the synthesis of LCBPP via the T-reagent approach....67 Figure 3-5: Representative 1 H NMR and assignments of LCBPP prepared by the T-reagent approach....69 Figure 3-7: Mark-Houwink plots of LCBPP obtained from SEC with triple detection. Full distributions are shown in Figure 3-6...73 Figure 3-8: Mark-Houwink plots of LCBPP prepared by slurry and bulk polymerizations...75 Figure 3-9: Two metallocenes were used to polymerize LCBPP. Dimethylsilanediylbis(2-methyl-4-phenylindenyl)]zirconium dichloride (A)(left) and dimethylsilane(2-methyl-benzoindenyl)(2-methyl-4-phenyl-4- H-azulenyl)hafnium dichloride (B)(right)....77 Figure 3-11: Complex viscosity at 190 C of LCBPP prepared by catalyst A...82 Figure 3-12: Complex viscosity at 190 C of LCBPP prepared using catalyst B....82 Figure 3-13: Dynamic storage modulus at 190 C of LCBPP prepared by catalyst A....83 Figure 3-14: Dynamic storage modulus at 190 C of LCBPP prepared by catalyst B....83 Figure 4-1: Master curves of storage and loss moduli at 190 C. (A) Linear PP1, (B) LCBPP4...91 Figure 4-2: Storage moduli of several LCBPP at 190 C...92 Figure 4-4: Complex viscosities for various isotactic polypropylenes at 190 C...95 Figure 4-5: Relationship of zero-shear viscosity at 190 C and weight-average molecular weight for linear and branched isotactic polypropylenes. The linear fit of the linear polymers is described by Eq. 4.3....97 viii

Figure 4-6: Extensional stress growth functions at various strain rates for LCBPP4 at 180 C....100 Figure 4-7: Extensional viscosity versus strain rate for LCBPP4 at 180 C. Solid line corresponds to 3 times the zero-shear viscosity obtained from smallamplitude oscillatory shear...101 Figure 5-1: Diffraction pattern of the ExxonMobil capacitor grade PP....110 Figure 5-3: 2D WAXS pattern for the Exxon Mobil capacitor grade polypropylene, PP 4342C2, before uniaxial orientation (left) and after uniaxial orientation (right)...114 Figure 5-4: WAXS patterns from three different material planes of a polypropylene film with uniaxial orientation....114 Figure 5-5: χ dependence of the 040 crystallographic plane for ExxonMobil PP4342C2 before and after uniaxial orientation at 145 C...115 Figure 5-6: χ dependence of the 040 crystallographic planes for various uniaxial oriented polypropylenes. All films were stretched at 145 C...116 Figure 5-7: ExxonMobil PP 4342C2 before (left) and after (right) uniaxial orientation at 170 C followed by crystallization at 100 C...117 Figure 5-10: Chi dependence of several LCBPP after uniaxial orientation at 160 C (T>T m ) followed by crystallization at 20 C...119 Figure 5-11: Weibull analysis of the electrical breakdown of polypropylene films at room temperature...121 Figure 6-1: SEM of the cell morphology of polypropylene foams prepared from linear polypropylene (left) and branched polypropylenes (middle and right) 6...128 Figure 6-2: Shear creep compliance of a polymer melt...132 ix

LIST OF TABLES x Table 3-2: Comparison of slurry and bulk polymerizations....75 Table 3-3: Comparison of T-reagent/H 2 behavior in various catalyst systems. Slurry polymerizations of propylene in the presence of T-reagent and H 2...79 Table 4-1: Rheological results for LCBPP....96 Table 5-2: Degrees of crystallinity for several isotactic polypropylenes. Samples were thick films prepared by compression molding...111 Table 5-3: Orientation of several polypropylene materials stretched at 145 C...116

ACKNOWLEDGEMENTS xi There are many individuals and groups which deserve acknowledgement for professional or personal support they may have provided. I would like to thank my office mates, specifically Ziqi Liang whose many interesting conversions I will remember. I would like to thank all of my lab mates within the Chung group; most notably Han Hong, Usama Kandil, and Zhiming Wang who provided me with many useful pieces of wisdom and intelligence. Drew Poche, Fumihiko Shimizu, Toru Suzuki, and Masuru Aoki deserve acknowledgement for their help in measuring and analyzing many samples of mine with techniques which were unavailable at Penn State. I would like to thank my committee members for their attention despite my many grammatical and spelling mistakes. A special thanks to Professor Ralph Colby for his rheological guidance and to his students, Brian Erwin and Arnov Ghosh, for their experimental training. A special thanks to my thesis adviser, Professor Mike Chung, to whom I am indebted for his advice, guidance, and support in my perusal of a post-graduate degree. Finally, I would like to extend an extraordinary thanks to my family. Without their encouragement and support none of this would have been possible.

Chapter 1 Introduction to Branched Polyolefins 1.1 Introduction Polymers can be prepared by numerous synthetic approaches providing structures which are linear, branched, or crosslinked 1. Linear polymers are the most commonly described and produced structure by polymerizations and have well described properties. Linear polymer chains are produced by connecting monomers in one continuous end-toend segment forming a more or less one dimensional structure. Branched or crosslinked structures are often obtained by altering the conditions of a linear polymerization. Such conditions lead to monomers which are connected in a two dimensional fashion by connections that are end-to-end and end-to-side. Branched and crosslinked structures are usually desired for their properties which can differ from linear polymers drastically. Branched polymers can have multiply structures (see Figure 1-1 for examples), but are generally described by chain segments which are attached to the side of a main chain. Branches may be randomly distributed throughout the main chain or adhere to a regular pattern. Branching may be contained to only the backbone (comb polymer) or exist throughout the entire polymer. The length of a branch may be short or long. Long chain branches, defined here as branches containing segments longer that the critical molecular weight for entanglements, are of significant interest due to the effect they have on a material s properties. Crosslinking occurs when polymer chains are connected to

2 each other through chemical links created at locations other than the chain-end or through the connection of polymer chains at the chain-end using multifunctional linkers. Crosslinking can occur during the polymerization or post-polymerization through chemical modification. Light crosslinking of materials is often used to obtain elastic materials such as rubbers, while high degrees of crosslinking lead to rigid thermosets. A D C B E Figure 1-1: Examples of possible branched structures. (A) linear chain, (B) long chain comb branching branches, (C) star branching, (D) long chain random branching, and (E) short chain branching. Long chain branched (LCB) polymers have value in processing techniques which demand high melt strength, including thermoforming 2-4, film blowing 4, 5, extrusion coating 6, 7, and blow molding 8, 9. Within the polyolefin family, polyethylene has received much attention in respect to the many types of possible branching, while polypropylene has gained much less attention due to the difficulties in preparing branched polypropylene, particularly those with well defined structure. Difficulties in creating

3 branched polypropylene arise from the limited chemistry to produce highly isotactic polypropylenes. These chemistries include Ziegler-Natta and metallocene catalysis which produce highly linear and highly stereospecific polymers. Although these catalysts produce polypropylene with desirable properties, they typically cannot produce a branched structure. As a result, the use of polypropylene has been limited in some applications. Despite the difficulties in producing a branched polypropylene, the commercial importance of such a material has lead to many attempts at preparing long chain branched polypropylene. These attempts, including macromonomer, diene, coupling, and post-polymerization chemical modification approaches, have produced LCBPP with varying degrees of success. 1.2 Branched Polyethylene Polyethylene (PE) has a huge market consisting of materials with applications based in containers, packaging-films, and appliances. The shear number and breadth of applications make PE and copolymers of ethylene the most used polymer today. The breadth of applications for PE and its copolymers relies in part on the flexibility of the material and processing properties obtained by altering the structure and type of branching. The major structures of commercial polyethylene resins are shown in Figure 1-2 and will be discussed in the following sections.

4 HDPE LDPE LLDPE Figure 1-2: Representations of polyethylene backbone structures for various commercially important products. 1.2.1 Low Density Polyethylene The most common example of branching in PE is low density polyethylene (LDPE). LDPE is prepared by high temperature and pressure free radical polymerizations of ethylene. During the polymerization the reactive site of a growing polymer may chain transfer via intra- or intermolecular mechanisms. These processes lead to the creation of an active site at which propagation will produce a branched polymer. Intramolecular chain transfer is often referred to as back-biting and results in the formation of short chain branches 10. Intermolecular chain transfer results in the formation of long chain branches. LDPE prepared by high pressure, free radical polymerizations has a wide distribution of both short and long chain branching resulting from a mixture of intra- and

5 intermolecular chain transfer. The alteration of polymerization conditions allows for the modification of branching distribution and length and can subsequently be used to prepare materials with tailored properties. The number of short chain branches affects the crystallinity of the LDPE. LDPE containing short chain branches has a maximum crystallinity of about 60%. The long chain branches, which are present in a much lower concentration, affect the melt properties and processing behavior of LDPE 11, 12. The analysis of short chain branching can be determined relatively easily by nuclear magnetic resonance (NMR) 13-15 and Fourier transform infrared (FTIR) spectroscopy techniques 16, 17, but the analysis of long chain branches remains a more difficult property to measure. 1.2.2 Linear Low Density Polyethylene New technologies developed by Dow Chemical Company and ExxonMobil in single-site metallocene catalysis have lead to advances in LLDPE synthesis. Catalysts designed for the incorporation of α-olefins have lead to Dow s constrained geometry catalysts (CGC) (Figure 1-3) which increased the reactivity of the catalyst towards α- olefin comonomers and decreased the heterogeneity in the comonomer distribution (Figure 1-4) 18. LLDPE produced using CGC catalyst contains long chain branches which are produced in situ through macromonomer incorporation. During the copolymerization the catalyst is constantly undergoing propagation and termination mechanisms. Termination mechanisms such as chain-transfer to monomer or beta-hydride elimination lead to the formation of unsaturated chain-ends. CGC catalysts have the capability of

6 incorporating these high molecular weight chain-end unsaturated polymers to form longchain branches. LLDPE containing LCB has been shown to have several processing advantages 19. Figure 1-3: Constrained geometry catalyst (CGC) 20. Figure 1-4: Comonomer distributions as measured by temperature rising elution fraction chromatography (TREF) 18.

1.2.3 Long Chain Branched Polyethylene 7 As depicted in Figure 1-5, LCBPE can be prepared using CGC catalysts. Homopolymerizations of ethylene with CGC catalysts lead to the formation of macromonomers which can be incorporated into the backbone of a polymer chain. The incorporation of polyethylene macromonomer depends on the catalyst system and polymerization conditions 21-27. The catalyst system must be able to both produce and incorporate vinyl-terminated polyethylene (macromonomer). Conditions are most favorable for the incorporation of macromonomer when the concentration of polymer is high and the concentration of ethylene is low. Production of vinyl-terminated macromonomers fluctuates with processing conditions such as pressure and temperature. CGC catalyst ethylene macromonomer + CGC catalyst LCBPP Figure 1-5: Scheme for the preparation of LCBPE via macromonomer incorporation. The effects of long chain branches produced by single-site metallocene catalysts have been observed through NMR and rheological results 23-28. The rheological properties of the prepared LCBPE could be altered by changing the polymerization conditions. Lower densities of branching were observed at higher pressures of ethylene. Branching

8 density could be increased by lowering the pressure of ethylene 23, 24, 26, 27. As shown in Figure 1-6, the zero-shear viscosities increased abruptly despite only a slight increase in MW, suggesting the formation of long chain branches. These rheological results were supported by 13 C NMR measurements which suggested the presence of branches longer than six carbons. Additionally, the effect of the catalyst was observed by comparing two single-site metallocenes. The metallocene which prepared a polyethylene rich in chainend vinyl groups resulted in high zero-shear viscosities and high activation energies suggesting long chain branches. The other metallocene, which prepared a polyethylene with fewer chain-end vinyls, lacked the distinct change in rheological properties and instead showed a much milder effect (Figure 1-7). In Figure 1-7, Samples A5 and B2 were prepared with a single-site metallocene catalyst which preferred the formation of vinyl chain-ends. The complex viscosities of these samples displayed strong shearthinning and lacked a frequency independent region. Samples C3 and D1 were prepared with a single-site metallocene catalyst which prepared polymers with fewer vinyl chainends. These samples displayed a less dramatic change in rheological properties than A5 and B2. Reference samples Ref-1 and Ref-2 were a HDPE and LLDPE prepared by Zielger-Natta polymerization, respectively.

Figure 1-6: Complex viscosities of LCBPE prepared at various pressures. Pressures in bar are indicated in the legend before the sample identification. The zero-shear viscosities of comparable linear polyethylenes are also included 24. 9

10 Figure 1-7: Complex viscosities of various polyethylenes at 190 C. Sample A5, Mw = 115,000 g/mol, MWD 2.8 and B2, Mw 153,000 g/mol, MWD 3.6. Sample C3, Mw 117,000 g/mol, MWD 3.6 and D1, Mw 114,000 g/mol, MWD 2.2. Reference material Ref-1, Mw 380,000 g/mol, MWD 2.1 and Ref-2, Mw 100,000 g/mol, MWD 3.8. 23. 1.2.4 Well-Defined Branched Polyethylene There is a strong desire to prepare polymers with well-defined branched structures, such that basic structure-property relationships can be studied. Well-defined polyethylene has been prepared through the post-polymerization hydrogenation of polybutadiene 29-31. Polybutadiene prepared by anionic polymerization is typically linear in structure, but through the use of coupling agents and careful synthetic techniques a wide range of branched structures can be obtained. Despite the difficult nature of these

11 methods, branched polybutadiene has been popular for preparing monodisperse, welldefined polymers whose structure-property relationships can be clearly studied. The results of these studies will be discussed further in Chapter 2. 1.3 Branched Polypropylene Isotactic polypropylene (ipp) is one of the most commercially important polymer materials today and is second in consumption only to polyethylene. High melting temperature, high tensile strength, and chemical resistance make isotactic polypropylene a popular choice in applications such as films, containers, and molded parts. The only limitation to isotactic polypropylene is the limited availability of high melt strength materials for demanding applications. The low melt strength of isotactic polypropylene arises from the nature of current catalyst technologies which are limited to Zielger-Natta and metallocene catalysts. These catalyst technologies produce a predominantly linear structure. Low melt strength has been attributed to the processing short-comings of isotactic polypropylene. For example, undesirable sagging in thermoforming was related to the low melt strength of isotactic polypropylene 32. As will be discussed further, the melt strength of polypropylene can be improved by the addition of a small amount of long chain branches, but unlike polyethylene there exists no convenient polymerization technique to prepare isotactic polypropylene with long chain branches. Therefore, many research programs have attempted to discover a synthetic scheme to provide long chain branching. These attempts, including

macromonomer, diene, coupling, and post-polymerization chemical modification approaches, have produced LCBPP with varying degrees of success. 12 1.3.1 Macromonomer Approach Direct synthesis of long chain branched isotactic polypropylenes (LCBPP) has been limited. The incorporation of polypropylene macromonomers has been studied utilizing various techniques. In situ formation of polypropylene macromonomers has been studied by the use of single-site metallocene catalysts 33. These techniques attempt to select a catalyst or a set of conditions capable of preparing polymer chains with chain-end unsaturation. The unsaturated chain-ends may be further incorporated into another polymer chain and are called macromers or macromonomers (Figure 1-8). macromonomer Figure 1-8: Generalized scheme for the formation of LCBPP by the incorporation of macromonomer. The reactivity of the unsaturated chain-end depends on the mechanism of chain transfer. Many polypropylene polymerizations propagate through 1,2-insertions of propylene monomer and β-hydrogen elimination is typically the major mode of termination. Termination via β-hydrogen elimination results in a sterically congested

13 vinylidene chain-end (Figure 1-9) 34, 35. Other commonly observed chain-ends are those of the 2-butenyl and isobutenyl structures. These chain-ends are sterically congested as well. A sterically accessible chain-end is available through β-methyl elimination (Figure 1-9) 36, 37. This allyl chain-end is the most desirable structure for macromonomers, but termination via β-methyl elimination is less favorable for most metallocenes capable of isotactic polypropylene polymerizations 36. Select metallocenes favor β-methyl elimination 36 and have been used in the preparation of LCBPP 33, 38-40. Figure 1-9: Pathways for the formation of unsaturated chain-ends during the metallocene mediated polymerization of propylene. Weng et al. used an isospecific metallocene activated with methyaluminoxane (MAO) at low concentrations of propylene to prepare LCBPP 33. Low concentrations of propylene increased the rate at which unsaturated chain-ends were produced and

14 incorporated. Slight changes in temperature (5 C) and solvent also had an effect on the amount of long chain branching. Decreases in temperature lead to polymerizations which produced fewer unsaturated chain-ends and resulted in higher molecular weight polymer. Polymers prepared at higher temperatures contained higher degrees of branching, but the increased number of regio defects resulted in lower melting temperatures. An increase in unsaturated chain-ends and LCBs was observed after a change in solvent from hexanes to toluene. Conditions that are favorable for the preparation of a desired macromonomer (low molecular weight, high percentage of allyl chain-ends, and good solubility in the reaction medium) are not favorable for the preparation of high molecular weight LCBPP 36,41. A catalyst which may produce polypropylene with a high concentration of unsaturated chain ends is usually unable to also produce high molecular weight branched polymers. Therefore, two polymerizations are often used in consecutive reactions 38,42. One polymerization is selected to produce lower molecular weight polymer with a high concentration of macromonomers containing allyl chain-ends. The other is selected to successfully incorporate the macromonomer into a high molecular weight polymer. The efficiency of these polymerizations was low despite efforts to produce low molecular weight macromonomers containing the more reactive allyl chain-end. Conversion of macromonomer remained below 50% for macromonomers of only 840 g/mol and decreased as the molecular weight of the branched PP was increased 38. Even under more desirable conditions, with soluble macromonomer prepared in situ, conversions remained low 43,44. Low conversions of macromonomers can prove to be an undesirable occurrence

15 in the preparation of LCBPP. Low molecular weight macromonomers remaining in the sample can negatively affect many of the properties of the polymer. The characterization of the branched structure of LCBPP prepared by in situ formation or multi-step polymerizations has been limited. Size exclusion chromatography (SEC) was used to measure the molecular weight distributions of LCBPP. The MWD remained narrow with only a slight increase in MWD at high incorporation of macromonomer 33, 38. The radius of gyration, R g, was measured using SEC with a multiangle laser light scattering (MALLS) detector (Figure 1-10) 33. The log-log plot of R g and molecular weight reveals the presence of high molecular weight branching. B A C Figure 1-10: The log-log plot of the radius of gyration, R g, as a function of weight - average molecular weight as measured by SEC-MALLS (C) 33. A linear PP is plotted with C for comparison. Additionally, the differential weight fraction (A) and the branching index, R g R, (B) are shown. g. br g, l M

1.3.2 Non-Conjugated α,ω-diene Approach 16 Methods have been developed to produce long chain branches via non-conjugated diene comonomers 45-47. The general scheme of the non-conjugated diene approach is represented in Figure 1-11. One of the allyl functionalities in the non-conjugated diene comonomer may be incorporated by 1,2-insertion into the metallocene mediated polymerization of propylene. The incorporated diene comonomer provides a reactive functional group along the backbone of polypropylene. Further incorporation of this functional group creates a branch point with tetrafunctional structure. Figure 1-11: General scheme for the formation of LCBPP from the copolymerization on α,ω-diene and propylene. Compared to the multi-step processes of some macromonomer approaches, the non-conjugated diene approach is significantly more straightforward. The selection of catalyst and non-conjugated diene are based upon compatibility. Certain catalyst and diene combinations have shown a tendency to form cyclic structures 48, 49 which are undesirable byproducts of the polymerization. In general, longer α,ω-dienes such as 1,7- octadiene and 1,9-decadiene have shown lower tendencies to cyclize. The concentration

17 of diene is another important condition which affects the materials prepared. Lower concentrations of diene will produce polymers with low levels of branching which perhaps may not be enough to affect the rheological properties. High concentrations of diene can lead to insoluble gels 46, 47. LCBPP created by the diene approach exhibited broadened molecular weight distributions (MWD) with a polydispersity index as large as 14.3 observed for a higher concentration of 1,9-decadiene 47. These increases in MWD were observed in combination with the formation of a high molecular weight tail in the distribution (Figure 1-12) 46, 47. A sharp change in distribution was observed as the concentration of diene was increased. The abrupt change from narrow to broad suggests controlling the breadth of the molecular weight distribution would be difficult. At the lowest concentration of diene, PP2, the presence of diene was unobservable by 13 C techniques. Gels were observed at higher concentrations of α,ω-diene. Figure 1-12: Molecular weight distributions (MWD) of LCBPP prepared by the copolymerization of 1,9-decadiene and propylene 47, where PP1 was a propylene homopolymerization and PP2 through PP5 were copolymerizations of propylene and diene. The concentration of 1,9-decadiene increased from 0.177 to 3.54 mmol/l from PP2 to PP5.

18 Small amplitude oscillatory shear measurements revealed the LCBPP prepared via the diene approach had increased zero-shear viscosities as shown in Figure 1-13 46, 47. Small amounts of incorporated diene significantly increased the zero-shear viscosity and enhanced the shear-thinning behavior of the polymer melt. Complex viscosities of LCBPP containing observable amounts of diene had complex viscosities which did not display Newtonian behavior at the lowest frequencies measured. The abrupt change in the behavior of the complex viscosity corresponds with the observed behavior of the molecular weight distribution, further suggesting the difficulties which would arise when attempting to adjust the rheological properties of the branched polymer. Figure 1-13: Complex viscosities of LCBPP at 190 C prepared by the compolymerization of 1,9-decadiene and propylene 47, where PP1 through PP4 are the same samples shown in Figure 1-12. The storage modulus, G, and loss modulus, G, of LCBPP prepared by the diene approach showed an abrupt transition from viscoelastic liquid to viscoelastic solid as the concentration of diene was increased 46, 47. At low concentrations of diene there was a clear crossover of G and G at high frequencies and a transition towards Newtonian behavior at low frequencies where G was much greater than G. As the concentration of

19 diene was increased, the crossover of G and G shifted to lower frequencies until a crossover was no longer observed and G remained greater than G for all frequencies observed (Figure 1-14). This transition could be caused by the early stages of gelation in the LCBPP obtained from higher concentrations of diene. Increased zero-shear viscosities and enhanced shear-thinning can be desirable properties, but the extremely elastic nature of materials containing a high concentration of diene comonomer should be avoided for most processing applications. Figure 1-14: Dynamic moduli and complex viscosity of a LCBPP at 200 C prepared by the copolymerization of 1,9-decadiene and propylene 46. This data is representative of the gel-like behavior of LCBPP prepared containing higher concentrations of diene comonomer. It has also been shown that these branched polymers display strain hardening behavior under elongational flows 50. Along with small amplitude oscillatory shear

20 measurements, Sugimoto et al measured the elongational flow behavior of LCBPP prepared by the copolymerization of 1,9-decadiene and propylene. The resulting LCBPP had a low concentration of incorporated diene which was undetectable by 13 C NMR experiments 50, similar to that of previously described PP2. The elongational stress growth coefficient was measured at 180 C for several Hencky strain rates (Figure 1-15). At all of the strain rates studied, strain hardening was observed as an elongational stress growth + coefficient, η ( t, & ε ), which increased sharply at long times to values far exceeding three E times the shear stress growth coefficient, η + (t).

21 Figure 1-15: Elongation stress growth coefficients of a LCBPP at 180 C prepared by the copolymerization of 1,9-decadiene and propylene (PP-B) 50, where the shear stress growth coefficients, multiplied by 3, are shown as solid lines and the Hencky strain rates are indicated. PP-A and PP-C are a linear isotactic polypropylene and a LCBPP prepared by electron beam irradiation, respectively. 1.3.3 Post-polymerization Chemical Modification Approach Difficulties in the synthesis of branched polypropylene have led to increased interest in chemical modification of linear polypropylene. Among the most successful have been reactive extrusion 51-53 and electron beam irradiation 54, 55. These techniques degrade the polypropylene in a controlled fashion in order to create branches. Chemical

22 modification of isotactic polypropylene by reactive extrusion utilizes a high temperature process which promotes chemical degradation (Figure 1-16). High temperature peroxide is added to the polymer melt, which begins to degrade the polymer through hydrogen abstraction to form tertiary and secondary alkyl macroradicals. Tertiary macroradicals lead to degradation via chain scission while secondary macroradicals lead to branching and crosslinking through combination with other macroradicals 56, 57. A polyfunctional monomer, such as pentaerythritol triacrylate or trimethylol propane triacrylate (TMPTA) in, can be added to the reactive mixture to stabilize tertiary macroradicals by forming the more stable secondary macroradicals 51, 52. Iniferters, such as tetramethyl thiuram disulfide in, may also be added to the reactive mixture to further stabilize tertiary macroradicals 53.

23 Figure 1-16: Mechanism for the formation of LCBPP via reactive extrusion in the presence of a polyfunctional monomer and an iniferter 53. The concentrations of peroxide and polyfunctional monomer affect the properties of the prepared LCBPP (Figure 1-17). An increase in peroxide concentration without the addition of polyfunctional monomer resulted in the degradation of PP observed by an increase the melt flow index and a decrease in viscosity. As polyfunctional monomer was added the melt flow index gradually decreased to values lower than the original linear polypropylene 52. An increase in the concentration of the polyfunctional monomer eventually resulted in an increase in insoluble gels formed by crosslinking 52.

24 MFI g/10min 70 65 60 55 50 45 40 35 30 25 20 15 10 5 0 Initial PP 200 PPM 600 PPM 1000 PPM 0 1 2 3 4 5 6 wt% PETA Figure 1-17: The effect of peroxide and polyfunctional monomer concentrations on the melt flow index at 230 C (Data from reference 52). The polyfunctional monomer was pentaerythritol triacrylate (PETA) and the peroxide was 2,5-dimethyl-2,5(tbutylperoxy)hexane peroxide with concentrations shown in the legend 52. The melt flow index of the starting material was 17 g/10min. Small amplitude oscillatory shear performed at 180 C is shown in Figure 1-18. At the lowest concentration of polyfunctional comonomer (M1) some degradation was observed by a horizontal shift of the dynamic moduli which correlates to shorter relaxation times. The ability of the iniferter to stabilize the tertiary free radicals and prevent β-scission, through a reversible reaction with the polymer macroradical, depends greatly on the concentration of polyfunctional monomer. LCBPP prepared with iniferter displayed a systematic increase (after the initial decrease) in relaxation times with an increase in polyfunctional monomer concentration. The increase in relaxation times suggested the formation of long chain branches.

25 Figure 1-18: Dynamic moduli for LCBPP at 180 C prepared by reactive extrusion utilizing an iniferter, tetraethyl thiuram disulfide. The concentration of polyfunctional monomer increased from 6 to 12 g/kg in M1 to M4. PP is the linear material before chemical modification 53. Electron beam irradiation techniques are similar to those of reactive extrusion save the fact that the radicals are being generated through the use of an electron beam. The properties of the polymer can be controlled by the length and type of irradiation to which the polymer is exposed 54. Polyfunctional monomers are added to the polymer melt as with the reactive extrusion techniques. Different materials could be obtained by altering the structure and type of polyfunctional monomer. Bifunctional monomers were observed to increase the melt strength by as much as 7 times that of the original

26 polypropylene, while polyfunctional monomers resulted in gel contents of 2 to 5% with marginal increases in melt strength 54. Samples irradiated in the presences of bi- and polyfunctional monomers displayed a reduction in both MFR 54 and M 55 w (Figure 1-19). Unlike previously discussed methods, the MWD of LCBPP via irradiation lead to narrower MWD suggesting the occurrence of degradation 55. Figure 1-19: MWD of LCBPP prepared by electron beam irradiation. The irradiation doses in kgy are indicated by the number behind PP 55. For these two techniques the role of degradation becomes dominant. Any improvement in melt strength comes at the cost of a reduction in MWD, molecular weight, and zero-shear viscosity (Figure 1-20) 54, 55. Both of these techniques have previously been used to prepare branched polymers with broadened relaxation spectra and strain hardening behavior 58-60 which are associated with high melt strength. Unfortunately, due to the complex nature of the prepared materials consisting of many structures it is hard to characterize structure-properties of electron beam irradiated branched polypropylene beyond simple branch density observations.

27 Figure 1-20: Complex viscosities of LCBPP at 180 C prepared by electron beam irradiation. The irradiation doses in kgy are indicated by the number behind PP 55. 1.3.4 Coupling Approach The preparation of a relatively well-defined LCBPP structure via the coupling of two functionalized polypropylenes was developed by the Chung group at Penn State. A comb structure can be created by coupling together graft functionalized copolymers and chain-end functionalized polymers. These functionalities are chosen such that coupling between the two polymers results in the formation of a thermally stable imide bond (Figure 1-21). LCBPP has been prepared in this fashion by combining isotactic polypropylene-graft-maleic anhydride and amine-terminated isotactic polypropylene 61.

28 Figure 1-21: The synthetic scheme for the formation of LCBPP by coupling of functionalized polypropylene 61. Similar reactions have been reported for graft copolymers of propylene and ethylene-propylene copolymer 62, but this method relies on a degradative process to produce maleic anhydride-terminated isotactic polypropylene. The Penn State method, described by Lu et al. 61, prepared amine-terminated isotactic polypropylenes from the chemical modification of maleic anhydride-terminated polypropylene with diamine reagent. These maleic anhydride-terminated polymers were prepared from a boranecontaining precursor prepared 63, 64 by hydroboration 65 of chain-end unsaturated polypropylene. These two techniques allowed for preparation of relatively well-defined branches through multiply post-polymerization chemical modifications. The prepared LCBPP had branching ratios, g = [ η] [ η], which depended on both the molecular br l M weight and frequency of grafted branches 61. 1.4 Conclusions All of the previously mentioned procedures to produce LCBPP have advantages and disadvantages. The incorporation of macromonomers would be an extremely

29 straightforward technique if only a catalyst/condition combination was available to incorporate and prepare in situ high molecular weight isotactic macromonomer with a reactive chain-end. In reality, attempts to increase the reactivity of the macromonomer often lead to low molecular weight, low tacticity material. The diene approach is useful for preparing LCBPP with low levels of branching, but the possibility of crosslinking at higher diene concentrations may limit its use in some applications. Reactive extrusion and electron beam irradiation unfortunately degrade the starting material and despite increases in melt strength, the increased melt flow index and decreased shear viscosities may be undesirable. Coupling methods, while producing relatively well-defined structures, require multi-step reactions. As will be discussed in Chapter 3, a one-pot method to synthesize LCBPP via the metallocene mediated polymerization of propylene with T-reagent has been reported by our group 66. This unique procedure involving both copolymerization and chain transfer mechanisms allowed for the formation of tri-functional branch points. Results suggest the presence of high molecular weight branches without gels or broadened MWDs. Additionally, this approach maintains the ability to control the molecular weight without complicated multi-step procedures. In a more detailed study 67, the effect of branches on rheological properties was systematically examined with high molecular weight LCBPP produced via this method. These results provided further evidence of high molecular weight, sparse branching. This reference is effectively a synopsis of this dissertation, particularly Chapters 3 and 4.

Chapter 2 Characterization of Long Chain Branched Polymers 2.1 Introduction Branching modifies many of the solution and melt properties of a polymer. These modifications in properties are responsible for the practical importance of the materials and require careful characterization. An understanding of how branch structure affects these properties has been developed from careful studies of well-defined branched materials. Methods to obtain qualitative and quantitative measurement of the branching structure have been developed. Most often a successful observation of a polymer s branching structure results from the use of multiple complimentary techniques. 2.2 Nuclear Magnetic Resonance Spectroscopy Nuclear magnetic resonance (NMR) spectroscopy has been routinely used to measure branching in polymers where the chemical shift of a branch point differs from that of the main chain. 13 C NMR has been used to measure the morphology of polyethylene containing short and long chain branches. 13 C NMR was first used by Randall who measured the spectra obtained from a variety of ethylene copolymers and LDPE 13. Armed with the information that alkane 13 C chemical shifts were sensitive to the neighboring carbon atoms up to five carbons away 68, 69, Randall was able to show the

31 length of branching can be determined for branches containing less than five carbons. After a branch length of five carbons the chemical shift became indistinguishable from a five carbon branch. The chemical shifts and methods of Randall were used to observe the short and long chain branch densities (branches greater than six carbons). However, there was wide concern over the apparent lack of agreement between the many experiments of the time. These differences were explained by the intricacies involved in the experimental conditions of 13 C NMR and the wide range of LDPE resins available. Axelson et al. studied a wide variety of LDPE resins and concluded that careful selection of experimental conditions and utilization of the detailed information for each specific branch type could lead to the ability to obtain short and long chain distributions 70. Additionally, it was shown that quantitative measurements obtained from 13 C NMR were in agreement with measurements obtained by other methods. For polypropylene, the investigation of branching by 13 C NMR has been much less popular. Shiona et al. produced branched polypropylene containing atactic branches and identified the presence of a branched structure using a combination of 1 H and 13 C NMR techniques 38. Weng et al. utilized 13 C NMR experiments to obtain quantitative amounts of branching in metallocene polypropylene and correlated those with results obtained from other measurements 33. The LCBPP 13 C NMR spectrum was observed to contain four new chemical shifts which were not observed in the spectrum of linear polypropylene (Figure 2-1). Macromonomer insertion was assumed to prevail following the 1,2-insertion of propylene (structure I Figure 2-2) as there was no evidence of a favored macromonomer insertion after 2,1- or 3,1-misinsertions (structures II and III in

32 Figure 2-2) and concentrations of misinsertions were low for the metallocene used. The chemical shifts at 44.88, 44.74, and 44.08 ppm were assigned to methylene (-CH 2 -) carbons adjacent to the methine carbon at the branch point. From these assignments and quantitative 13 C NMR experiments Weng et al. 33 determined branch point densities for LCBPP prepared by the incorporation of macromonomer prepared in situ. Figure 2-1: An expanded 13 C NMR (bottom) and DEPT NMR (top) spectra of a LCBPP from the incorporation of macromonomer prepared in situ. The peaks marked with an asterisk at 44.88, 44.74, 44.08, and 31.74 ppm are peaks assigned to branch points 33. Figure 2-2: Chemical shift assignments for 13 C NMR experiments. (I) Macromonomer insertion after 1,2-insertion. (II) Macromonomer insertion after 2,1-misinsertion. (III) Macromonomer insertion after a 1,3-misinsertion 33.

33 The inability to measure the length of a branch past five or six carbons is a clear disadvantage of 13 C NMR experiments. The length or molecular weight of the branch plays an important role in determining the effect a branch may have on material properties. Branches need to be on the order of 2,000 g/mol before they alter the rheology, while unfortunately 13 C NMR can only determine branches on the order of 100 g/mol. 2.3 Size Exclusion Chromatography with Triple Detectors Size exclusion chromatography (SEC) is a routine liquid chromatographic technique within the synthetic polymers field. In SEC, macromolecules are separated based upon their size in solution. Separation is obtained by carrying the dissolved polymer through a column by means of a steady flow of solvent. Polymer chains of different size experience paths of varying tortuousity depending on the distribution of pore sizes in the packing material. SEC is routinely used for obtaining molecular weight information such as the number average molecular weight (M n ), weight average molecular weight (M w ), and the molecular weight distribution (MWD). A conventional differential refractometer placed after the separation allows for the collection of a size distribution curve. Calibration of the SEC system with known standards allows for the size distribution curve to be converted into a molecular weight distribution curve. Molecular weight averages and the polydispersity index can be calculated from the molecular weight distribution curve as follows in Eq. 2.1, Eq. 2.2, and Eq. 2.3.

nim i wi hi M n = = = 2.1 n i i i ( w M ) ( h M ) i i 34 M w = 2 nim i wi M i = = n M w i i i hi M h i i 2.2 M w PDI = 2.3 M n Here w i and n i are the weight and number of molecules of molecular weight M i, respectively. M n is defined as the mass of the sample in grams, the sum of w i or n i M i, divided by the total number of chains present, n, which is the sum of all n i. The size distribution obtained from the differential refractometer is a weight concentration and M n can be calculated from the height of the size distribution curve, h i, assuming that h i is proportional to the concentration of polymer and that M i is sampled over equal volume increments, i. M w can be calculated in a similar fashion from h i and M i. Calibration of the size distribution curve and subsequent calculations of molecular weight are not valid for polymers containing branches. Branches reduce the size of a polymer in solution relative to a linear polymer of the same molecular weight. Thus the calibration curve constructed from linear polymer standards differs from that of a branched polymer. In 1949, Zimm and Stockmayer described a method for determining the branch density by fractionation of a sample by molecular weight and subsequently measuring the radii of gyration, R g, or intrinsic viscosities, [η], of each fraction 71. It was shown that for randomly branched tri- and tetra-functional polymers, the branching ration, g, could be

35 predicted by Eq. 2.4 and Eq. 2.5 where m is the average number of branches per molecule. <g 3 (m)> and <g 4 (m)> are the branching ratios for tri- and tetra-functional branching, respectively, defined as the ratio between the mean squared radii of a branched and linear polymer of the same molecular weight, M. The average branches per molecule, m, can be defined as the ration between M and M b, where M b is the numberaverage molecular weight between two branch points in a randomly branched polymer (Eq. 2.6). 0.5 0. 5 [(1 + m 7) + 4m ] g = π 2.4 av 3( m) 9 g 0.5 0. 5 [(1 + m 6) + 4m ] 4 ( m) 3 av = π 2.5 The separation capability of SEC combined with detectors to determine R g or [η] should allow for the determination of the branching frequency in randomly branched polymers using the equations provided by Zimm and Stockmayer (Figure 2-3) 71. Recent advances in detector capabilities allow for in-line measurements of R g and [η] using light scattering and viscometer detectors. M m = 2.6 M b

36 10 R g 1 0.1 1000 10000 100000 1000000 M Figure 2-3: Calculated behavior for the radius of gyration as a function of molecular weight for a M b of 25,000 g/mol by Eq. 2.4 and Eq. 2.6 71. The solid line represents a linear polymer while the dashed line represents the trifunctional branching and the dotted line represents the tetrafunctional branching. Light scattering detectors have gained popularity due to their capability of measuring absolute molecular weights without the need for the preparation of extensive calibration curves. Rayleigh (static) light scattering can be used to determine the molecular weight of a polymer in solution so long as several general considerations are followed. Static light scattering requires the solvent have a refractive index significantly different than the polymer such that excess scattering caused by the polymer in solution can be detected at some angle θ to the incident beam. Secondly, one must be prepared to measure the dn/dc of a polymer/solvent combination or have reference to a previous measurement. For most polymer/solvent combinations the requirement of refractive index

37 is of little concern and many literature values of dn/dc for such polymer/solvent combinations exist. A typical static light scattering detector simultaneously measures the intensity of the incident beam generated by a polarized, monochromatic laser and the intensity of scattering at several angles θ to the incident beam. From these measurements a weight average molecular weight can be determined from Eq. 2.7 where the concentration of polymer, c, is determined by an in-line concentration detector such as a differential refractometer and K is an optical constant containing dn/dc. K is defined by Eq. 2.8, where λ is the wavelength of the scattered light, n is the refractive index of the medium, and N Av is Avogadro s number. Eq. 2.7 is only valid for very low concentrations, such as those frequently used in size exclusion chromatography. Molecular weights calculated by this manner are more reliable at high molecular weights (> 10 4 g/mol). R θ = KcM 2.7 K 2 4π n = 4 λ N 2 Av dn dc 2 2.8 A differential viscometer detector capable of measuring the specific viscosity of the fractionated polymer has been developed 72, 73. The set-up of the differential viscometer allows for the measurement of specific viscosity by monitoring the change in pressure across a number of capillaries through Eq. 2.9 where P in is the pressure at the entrance to the detector and ΔP is the differential pressure calculated by monitoring the flow of solvent and solution through two identical capillaries. Through use of a

38 concentration detector, such as a differential refractometer, the intrinsic viscosity, [η], can be obtained from Eq. 2.9 and Eq. 2.10. 4ΔP η sp = 2.9 P 2ΔP in η sp [ η] = lim 2.10 c 0 c Measurements of branching ratios, g and g (Eq. 2.11 and Eq. 2.12), have been accomplished by triple detection SEC, a technique utilizing the combination of differential refractometer, light scattering, and differential viscometer detectors with SEC separation. In triple detection SEC, each fraction of separated polymer in solution is subjected to three types of detection. A light scattering detector measures the weight average molecular weight of each fraction and the viscosity detector measures the intrinsic viscosity. A differential refractometer measures the weight concentration of each fraction which is employed in both calculations of viscosity and molecular weight. Rg, Branched g = 2.11 R g, Linear [ η] g = [ η] Branched = Linear g ε 2.12 By plotting log[η] versus logm, a so called Mark-Houwink plot can be obtained. For linear polymers this log-log plot should reveal a single slope corresponding to the Mark-Houwink equation (Eq. 2.13) where K and a are the Mark-Houwink constants and vary with changes in solvent, temperature, and polymer type.

a [η ] = KM 2.13 39 When measuring a branched polymer a Mark-Houwink plot should reveal the subtle differences in R h. This is due to the fact that the product of M[η] is directly proportional to R 3 h and will decrease as the branch density of the polymer increases, as describe by the Fox-Flory equation (Eq. 2.14). [ ] R h 3 η 2.14 M A measure of the branching density can be obtained by measuring the branching ratio, g, of each fraction. g must be converted to g by use of ε g = g' from which the average number of branches per molecule, m, can be obtained from the appropriate Zimm Stockmayer equation (Eq. 2.4 or Eq. 2.6). For each fraction of molecular weight, M, the number of branch points per 10,000 carbons, LCB/10 4 C, can be calculated from m and the molecular weight by Eq. 2.15. Finally, the average number of branch points per 10,000 carbons is calculated using the weight concentration obtained from the differential refractometer (Eq. 2.16). LCB mm 010,000 = 4 2.15 10 C M M i i LCB 4 10 C = LCB wi 4 10 C w i M i = LCB hi 4 10 C h i M i 2.16 This manner of long chain branch detection has been shown a successful means of evaluating polyethylene 28, 74 and other branched polymers 75. For branched polyethylene

40 Wood-Adams et al. 74 measured the structure of branched materials both qualitatively and quantitatively. Qualitatively they noticed that a double logarithmic plot of intrinsic viscosity and molecular weight of a long chain branched polyethylene resulted in a plot similar to that which Zimm and Stockmayer described for a randomly branched polymer (Figure 2-4). At low molecular weights the plot intersected and followed the behavior observed as with linear polymers, but at higher molecular weights deviated from the linear sample. Short chain branched polyethylene was also evaluated by triple detection SEC and was observed to behave differently than long chain branched samples (Figure 2-5). The Mark-Houwink plots of short chain branched samples were of similar slope to that of the linear polyethylene, but were shifted. These results suggested the ability to discriminate between short chain and long chain branching. Additionally, quantitative measurements of the short chain and long chain branching content were made using Eq. 2.4, Eq. 2.15, and Eq. 2.16. The resulting branch densities were in agreement with other measurements for both short and long chain branching. Figure 2-4: Mark-Houwink plot of LCBPE obtained from high temperature SEC with triple detection. LCB density increases from HDB1 to HDB4 74.

41 Figure 2-5: Mark-Houwink plot of SCBPE obtained by high temperature SEC with triple detection. SCB frequency increases from LDB1 to LDB3 74. Gabriel et al. 28 measured the branching structure of long chain branched polyethylene by SEC combined with multi-angle laser light scattering (MALLS) 28. Although the experimental procedure was different than that of Wood-Adams et al., quantification of branching was also based upon the Zimm and Stockmayer equations and the ability to measure the difference in R g between linear and branched polymers. Qualitatively the results of Gabriel et al. were similar to those of Wood-Adams et al. The slope of the branched polyethylene deviated from the slope of the linear polymer at high molecular weights. Deviation became more apparent as the branch density of long chain branches was increased. However, Gabriel et al. observed differences in the quantitative measurements of branch density. They were able to observe similar trends of increasing or decreasing branch density, but SEC-MALLS measurements resulted in higher values of branching than that obtained from 13 C NMR experiments.

42 The characterization of branched polypropylene by triple detection SEC has been limited. Sugimoto et al. measured several branched polypropylenes using a dual detection SEC combining a differential refractometer and a multi-angle laser light scattering detector 50. Despite the lack of a viscometer detector they observed branching in polypropylene prepared from the copolymerization of alpha-omega diene and by electron beam irradiation techniques. The two samples showed similar behavior to that observed with polyethylene and described by Zimm and Stockmayer (Figure 2-6). Calculations of long chain branch density were made from measurements of R g and suggested that both branched polypropylenes have branch densities which decreased slightly as molecular weight increased (Figure 2-7). This behavior was not described by Zimm and Stockmayer nor was it observed for LCB polyethylene in previous studies. The averaged branch densities were not calculated or compared with other measurements of branching density. Figure 2-6: Mark-Houwink plot of LCBPP obtained by high temperature SEC with triple detection. PP-A is a linear polypropylene, PP-B is a LCBPP prepared by diene copolymerization, and PP-C is a LCBPP prepared by post-polymerization chemical modification 50.

43 Figure 2-7: Degree of branching calculated from SEC results for LCBPP made by diene copolymerization (PP-B) and post-polymerization chemical modification (PP-C) 50. For other polymers, such as polystyrene and poly(methyl methacrylate), SEC with triple detectors was used to determine the branching density of several polymers 75. Scorah et al. studied polystyrene and poly(methyl methacrylate) produced using a tetrafunctional peroxide initiator. Both MALLS and LALLS detectors were used to determine M w and R g of the polymer fractions. Branching was detected by SEC in polystyrene, but not in the case of PMMA, despite other measurements suggesting the presence of branching in both. 2.4 Rheological Techniques Of significant interest is to study the flow of the LCBPP melts so possible processing benefits can be identified. The melt flow of linear and branched polymers is qualitatively and quantitatively different; therefore, rheology measurements are

44 extremely helpful in determining the presence and nature of branches contained within a material. Among rheological techniques small amplitude oscillatory shear is one of the most convenient and allows for measuring a material s response over a wide range of time scales. By applying a sinusoidal shear strain, γ(t), (Eq. 2.17) one can obtain a stress, σ(t), which is also sinusoidal with the same frequency as the applied shear strain, but leads the strain by a phase angle, δ (Eq. 2.18). This phase angle can vary from 0 to π/2. γ ( t) = γ 0 sin( ωt) 2.17 σ ( t ) = σ 0 sin( ωt + δ ) 2.18 At a phase angle of 0, the stress is in phase with the strain and the material is a Hookean solid which can be described by Hooke s law (Eq. 2.19). At a phase angle of π/2, the stress is in phase with the strain rate and the material is a Newtownian liquid (Eq. 2.20). The linear response of viscoelastic materials falls between these two extremes. σ ( t) = Gγ 0 sin( ωt) 2.19 σ ( t) = ηγ 0ω cos( ωt) 2.20 The stress can be separated into two components that oscillate at the same frequency, one in phase with the strain and the other in phase with the strain rate (Eq. 2.21). [ G ( ω)sin( ωt) + G ( ω)cos( ω )] σ ( t) = γ 0 t 2.21

45 The storage, G, and loss moduli, G, can be related to the amplitude of the stress and strain and the phase angle at a given frequency (Eq. 2.22 and Eq. 2.23). The loss tangent, tanδ, can then be calculated as the ratio of G and G (Eq. 2.24). σ 0 G = cos δ 2.22 γ 0 σ 0 G = sin δ 2.23 γ 0 The terminal (low frequency) response of a viscoelastic liquid is dominated by the loss modulus. The stress is nearly in phase with the shear rate and the loss modulus is much greater than the storage modulus. At low frequencies, the loss modulus becomes proportional to the frequency with the viscosity as the proportionality constant (Eq. 2.25). The storage modulus of a viscoelastic liquid at low frequency is proportional to the frequency squared and provides a measure of the stored elastic energy described as the steady state compliance, J eq (Eq. 2.26). G tan δ = 2.24 G G ( ω) η = lim 2.25 ω 0 ω 1 G ( ω) lim 2 0 2 2.26 η ω ω J eq = An example of the linear viscoelastic response of a nearly monodisperse, hydrogenated polybutadiene is shown in Figure 2-8. The material has a low frequency (terminal) response which behaves as described in Eq. 2.25 and Eq. 2.26, 2 G ~ ω and

46 G ~ ω. These terminal regions can be extrapolated to obtain an intersection from which the relaxation time can be obtained by taking the reciprocal of the frequencies at which they cross Eq. 2.27. Figure 2-8: The dynamic moduli of a 210,000 g/mol linear polybutadiene before (line) and after (hollow symbols) hydrogenation at 130 C 29. ω x 1 1 = ηj τ = eq 2.27 Arrhenius activation energies of flow can be obtained by measuring the terminal response of a material at a minimum of two temperatures which are significantly higher (> 100 C) than glass transition temperature. Using Eq. 2.28, the Arrhenius activation energy of flow, E a, can be determined by measuring the zero-shear viscosity, η 0, at more than one temperature. E a can alternatively be measured by the frequency scale shift

factors, a T, obtained from the time-temperature superposition of the oscillatory shear data obtained at several temperatures. 47 η η E a 1 = exp R T 0, T = at 1 0, T 0 0 T 2.28 It has been know for some time that melts of polymers containing long chain branches behave in a different fashion than linear polymers. Additionally, the effect that large scale structure can have on the properties of a polymer melt seems to be universal among polymer chemistries. Because of this universality, long chain branching has been studied quite extensively 76. From this extensive research, it has been well shown that the flow behavior of polymeric materials is extremely sensitive to large-scale structure. The length, frequency, and position of branches can drastically affect the melt behavior 77. Early on it was noted that it was important to create well-defined polymers so that the effect of branch length, frequency, and position could be analyzed. Some of the first examples studied the effect of branching on viscoelastic response by preparing long chain branched polybutadiene 78 and polstyrene 79, 80 via anionic polymerizations. Kraus and Gruver 78 were among the first to note the melt viscosity s dependence on the molecular weight of the branches. A decrease in melt viscosity was observed at branch molecular weights near or below the entanglement molecular weight. Above the entanglement molecular weight the melt viscosities of the branched polymers increased abruptly to values several orders of magnitude greater than linear samples. Kraus and Gruver 78 suggested that the intermolecular entanglements of the branched polymers were the cause

of these increases melt viscosities. At low shear stresses these entanglements act as permanent constraints due preventing chain slippage 78. Rheological studies of monodisperse, branched polyethylene have shown branched polymers of varying architecture displayed relaxation times, zero-shear viscosities, and Arrhenius activation energies for flow which were sensitive to the density of branching, the length of the branches, and molecular weight of the polymer backbone 29, 30,81,82. Raju et al. 29 prepared well-defined star polyethylene by hydrogenation of polybutadiene prepared by anionic polymerization. The star hydrogenated polybutadienes (HPB) were compared to linear HPBs with similar molecular weight. Oscillatory shear experiments were performed at 130 and 190 C and the moduli of both the linear and star HPBs displayed the expected low frequency limiting behavior of G ω and G 2 ω. G of the star HPB showed a broadened transition to Newtownian behavior resulting in a broadened relaxation spectrum and longer relaxation times (Figure 2-9). Additionally, it was observed that all star PBs obeyed time-temperature superposition while the star HPBs did not. For HPBs, the activation energy of flow resembled that of the linear polymer at high frequencies, but steadily increased as the frequency decreased. Despite the wide spread universal effect of long chain branching on most polymers it seemed that HPBs and polyethylene differed slightly as compared to other polymers. 48

49 Figure 2-9: The dynamic moduli of linear (filled) and star (open) hydrogenated polybutadienes at 190 C 29. The molecular weight of the arm was 174,000 g/mol for 3HPB-2, 41,700 g/mol for 3PHB-1, and 80,300 g/mol for 4PHB-1. Raju et al. 29 also noted that the zero-shear viscosities of the HPBs were considerably higher than those of their linear counterparts and for long-arm star HPBs were not measurable due to the lack of limiting behavior in G at the highest temperature studied. Activation energies of flow were calculated from the temperature dependence of the zero-shear viscosities and showed a strong dependence on the arm molecular weight of the stars. More recent work was completed by Lohse et al. on HPBs systematically created to offer insight into the effects of branch length, frequency, and position 30. Many types of branched structures, ranging from symmetric stars to combs, were prepared by anionic

50 polymerizations. The addition of branches caused polymer melts to shear thin more strongly despite the nature of the branching (Figure 2-10). Branch length proved to be an important factor as the magnitude of increase in zero-shear viscosity and activation energy was directly related to the length of the branch. Melts containing shorter branches, although different from the linear polymers, showed less dramatic changes than those containing the long branches. Branch density proved to be less straightforward than branch length. A star polymer containing only one branch should similar increases in zero-shear viscosity as a comb polymer containing many branches. In Figure 2-10, a star polymer containing one branch of 5,000 g/mol and backbone molecular weight of 100,000 g/mol is compared with a comb polymer of similar backbone and branch molecular weight. The two materials have similar zero-shear viscosities despite the comb polymer containing 30 branches. Results such as these demand caution when attempting to use single values such as zero-shear viscosity to evaluate the density of branching.

51 Figure 2-10: Complex viscosity of a linear (PEL), star (PES), and comb (PEC) hydrogenated polybutadiene at 190 C 30. 50, 55, 58- Similar results have been obtained for long chain branched polypropylenes 60. Hingmann and Marczinke 58 measured the rheological response of three polypropylene melts, two of which were branched polypropylene. As expected, universal branched behavior was described for the chemically modified polypropylenes. The molecular weight dependence of the zero-shear viscosities was stronger than that observed for the linear polymers. Similarly, Kurzbeck et al. 60 compared a linear polypropylene with a branched isotactic polypropylene prepared by chemical modification through electron beam irradiation. The branched polypropylene was observed to have a higher Arrhenius activation energy of flow than the linear polypropylene. Auhl et al 55 observed that branched polypropylenes showed a broadened transition to Newtonian (terminal) behavior in oscillatory shear than linear polypropylene, but did not observe increased shear thinning.

Also of rheological importance is the affect of long chain branches on the extensional flow of polymers. Extensional rheology is the study of a material s response to elongational or stretching deformations. A stretched material undergoes high degrees of molecular orientation; therefore, extensional flows are described as strong flows. These extensional flows are extremely sensitive to crystallinity, entanglements, and long chain branching, making them ideal for rheological measurements to analyze the same. Additionally, many melt flows and deformations in polymer processing can be related to extensional flows. In a typical simple extensional flow measurement, the sample is deformed at a constant Hencky strain rate, ε&. The Hencky strain, ε H 52, is calculated from the ratio of the current length of the sample, L(t), and the initial length of the sample, L 0, by Eq. 2.29. The time derivative of the Hencky strain is the Hencky strain rate, ε& H = ε H t, and from now on simply referred to as ε&. L( t) ε H = ln 2.29 L 0 The extensional viscosity is the ratio of the steady-state extensional stress and the Hencky strain rate (Eq. 2.30). A Sentmanat extensional rheometer or extensional viscosity fixture can be used to measure the extensional stress or the extensional viscosity using an ARES or similar strain-controlled rotational rheometer. These attachments are based upon the Meissner uniaxial extensional rheometer 83, but can be confined within the environmental control chambers of the rheometer. With these attachments, the sample is placed between two cylinders of radius, r. The sample is uniaxially stretched by rotating

53 one cylinder around the other at a rotational rate, Ω (Figure 2-11 and Figure 2-12). By this method, the sample is equally stretched in both directions. The Hencky strain rate and the transient extensional viscosity can be calculated from the measured torque, T(t), and the cylinder rotation rate, Ω, (Eq. 2.31 and Eq. 2.32), respectively. Figure 2-11: Diagram of the extensional viscosity fixture. One cylinder rotates around its own axis while rotating around the axis of the other cylinder. Figure 2-12: Diagram depicting the deformation of the sample at several Hencky strains in an extensional viscosity fixture (reproduced from TA Instruments literature The ARES-EVF: Option for Measuring Extensional Viscosity of Polymer Melts by Frank.) σ η = E E & ε 2.30 2ΩR ε& = 2.31 L 0

+ η ( t) = E T ( t) 2R & εa( t) 2.32 54 It has been described that LCB containing polymers effectively increase the shear thinning behavior, but for elongational flows it has been shown that polymers containing more than two branches should result in strain hardening 30. This behavior is most clearly observed for the case of polyethylene (Figure 2-13) 84-87. Under elongational melt flows branched polyethylene strain-hardens while linear polyethylene with similar zero-shear viscosity does not. For branched materials, there exists a region of response before the onset of strain hardening that follows the generalization of Trouton s rule, which states that the extensional stress growth coefficient is three times that of the shear stress growth coefficient 88. Linear and branched polypropylene have been examined in a similar manner 50, 58, 60, 87. Polypropylene containing long chain branches strain-hardened while a linear polypropylene did not (Figure 2-14). Figure 2-13: The elongational behavior of a LCBPE and LLDPE at 150 C 85.

55 Figure 2-14: The strain hardening behavior of a LCBPP observed during elongational flow 60. 2.5 Conclusions Characterization of long chain branching is difficult at best. There are a number of techniques which allow for the observation of the nature of branching. NMR techniques allow for the detection of branch points and in the case of polyethylene the determination of short chain branching contents. Long chain branches are difficult to analyze completely as NMR cannot differentiate branches longer than six carbons. Additionally, the sensitivity of 13 C NMR measurements requires long time experiments that may be limited by the detection limits.

56 Other techniques such as size exclusion chromatography (SEC) and rheological experiments may provide more detailed information about the nature of a material s branching structure. SEC equipped with triple detection capabilities can detect subtle differences in a material s structure through changes in the hydrodynamic volume. These changes can be used to calculate a branch density from equations derived by Zimm and Stockmayer for random branching. Although these measurements can detect the difference between MWD effects, short-chain branching, and long chain branching, they sometime have difficulties in detecting low levels of branching. Rheological experiments, on the other hand, are capable of measuring even the slightest change in a material s structure. Shear and extensional viscosities are extremely sensitive to the presence of branching. Ideally, all three of these techniques should be used in combination to fully characterize the sample as possible.

Chapter 3 Synthesis of Long Chain Branched Polypropylene via the T-reagent Approach 3.1 Introduction The approach to prepare LCBPP via the polymerization of propylene in the presence of T-reagent and hydrogen described here originates from the results of previous works within the Chung group at Penn State. Many of these works have focused on the preparation of functionalized polyolefins 89. Perhaps the most important development in regards to LCBPP research was that of styrene-based chain transfer agents which allow for the functionalization of the chain-end (Figure 3-1) 90-93. + R Metallocene/MAO hydrogen PP R Figure 3-1: Generalized scheme for the preparation of a chain-end functionalized polypropylene through the use of a functionalized styrenic chain-transfer agent. The substituent R could be any number of protected functional groups including -OH, -Cl, or NH 2. Many have attempted to polymerize other important vinyl monomers with single-site metallocene catalysts with varying success 94-97. Styrene, which is commonly polymerized via free radical polymerizations to obtain atactic polystyrene, has been successfully polymerized by metallocene catalysts. These metallocene polymerizations

58 offer routes to producing sydiotactic polystyrene 95, 96 which may become an important engineering material in the future. During the search for a suitable catalyst to polymerize styrene, some ansa-metallocenes, specifically zirconocenes, were discovered to have low or no activity in the polymerization of styrene 98. Copolymerization of propylene and styrene with ansa-zirconocenes also proved ineffective 99. The addition of p-methylstyrene to the polymerization of propylene dramatically lowered the catalyst activity and reduced the molecular weight of the produced polymers 99. A 1,2-insertion of propylene is electronically and sterically favored for ansa-metallocenes, but for these same catalysts a 2,1-insertion of styrene is favored due to the cationic nature of the metallocene active site. The 2,1-insertion of styrene leads to an active-site which is blocked by the bulky nature of the ligands and the styrene. Further incorporation of propylene via a 1,2- insertion would lead to an even more congested active site. The nearly dormant site following 2,1-insertion of styrene was utilized as a chain-transfer agent 90, 92. A small amount of hydrogen added to the polymerization can complete the chain transfer after a dormant catalyst site is formed following the insertion of styrene (Figure 3-2). The process regenerates the metallocene catalyst and results in a chain-end containing the styrene group. From this process, chain-end functionalized polypropylene was prepared by adding functionality to the chain transfer agent 92. These chain-end functionalized polypropylenes have been shown to be effective agents in the preparation of modified organosilicates for nanocomposite applications 93, 100.

59 PP p-methylstyrene (PMS) Chain Transfer 1 H 2 PP-L n M Regenerated Catalyst H-L n M Chain Transfer 2 Me 2 Si(2-Me-4-PhInd) 2 ZrCl 2 PP PP-t-PMS Figure 3-2: Scheme for chain-transfer to a styrenic chain transfer agent in the presence of hydrogen. As was stated previously in Chapter 1, the significant limitation in preparing LCBPP through a macromonomer approach is the apparent complexity in preparing high molecular weight macromonomers which contain the reactive allyl end-group. Chain transfer via a specially designed agent should result in the formation of allyl end groups (Figure 3-3). By this approach it was theorized that a high percentage of chain ends would contain allyl end groups and LCB would occur in situ as, in the case of LCBPE.

60 + p-(3-butenyl)styrene T-reagent (BST) Metallocene/MAO hydrogen Reactive Macromonomer Figure 3-3: Hypothesized scheme to prepare macromonomers containing the reactive allyl end-group through a functionalized chain transfer agent. 3.2 Experimental 3.2.1 Instrumentation and Materials All 1 H NMR spectra were recorded on a Bruker AM 300 instrument in 1,1,2,2- tetrachloroethane-d 2 at 110 C. The melting temperatures of the polymers were measured by differential scanning calorimetry (DSC) using a TA Q100. Samples were crystallized from a 200 C melt at 20 C/min. The melting transitions were recorded after from 40 to 200 C with a heating rate of 20 C/min. All oxygen and moisture sensitive manipulations were carried out inside an argon-filled Vacuum Atmosphere dry box. Toluene was refluxed over metallic sodium with benzophenone and distilled under nitrogen atmosphere prior to use. p-toluenesulfohydrazide (97%) was recrystallized from methanol. 4-Vinylbenzyl chloride (90%) was vacuum distilled over calcium hydride. Polymerization-grade propylene was purified by passage through a column of 4Å molecular sieves and calcium

61 carbonate. Allyl magnesium bromide (1.0 M solution in diethyl ether ), diethyl ether (anhydrous), methylaluminoxane (MAO) (10 wt% solution in toluene), 1,1,2,2-tetrachloroethane-d 2 (99.6% D), and 1,3,5-trimethyl-2,4,6-tris(3,5-di-tert-butyl-4- hydroxybenzyl)benzene (99%) were used as received. Rac-dimethylsilanediylbis(2- methyl-4-phenylindenyl) zirconium dichloride was prepared by previously reported methods 101. 3.2.2 Synthesis of p-(3-butenyl)styrene (T-reagent) To a dry 500 ml round bottom flask equipped with addition funnel, condenser, and magnetic stir bar was transferred 200 ml (0.2 moles) of allyl magnesium bromide solution. 20 ml (0.14 moles) of vinylbenzyl chloride diluted with 50 ml of diethyl ether was added dropwise to the allyl magnesium bromide at ice bath temperatures. After complete addition, the mixture was warmed to room temperature and stirred for 12 hours. Slowly, 200 ml of distilled water was added to the mixture. The aqueous layer was separated and washed three times with diethyl ether. Organic solvent was removed under vacuum. Obtained crude product was dried with calcium hydride and distilled under vacuum before use. Yield: 20.5 g, 93 %. 1 H NMR (CDCl 3, 300 MHz): δ 7.1 7.5 (m, 4H, aromatic H), δ 6.7 (m, 1H, aromatic-ch=c), δ 5.9 (m, 1H, C-CH=C), δ 5.7 (d, 1H, aromatic-c=ch), δ 5.3 (d, 1H, aromatic-c=ch), δ 5.1 (m, 2H, C-C=CH 2 ), δ 2.7 (t, 2H, aromatic-ch 2 ), δ 2.4 (m, 2H, CH 2 ).

3.2.3 Copolymerization of Propylene and 4-phenylbut-1-ene 62 100 ml of toluene and 1 ml of MAO solution were added to a dry 300 ml Parr stainless steal autoclave equipped with a mechanical stirrer. The autoclave was then charged with 40 psi of propylene gas, heated to 45 C, and charged with 1 psi of hydrogen. Propylene gas was used to charge the reaction vessel to a total pressure of 140 psi. 1.1 x 10-7 moles of catalyst diluted in 1 ml of toluene and 4-phenylbut-1-ene were then injected under propylene pressure. The reaction vessel was stirred and maintained at 55 C and 140 psi for 30 minutes. The reaction was terminated by the addition of methanol and cooled to room temperature. The product was washed with acidic methanol and dried under vacuum at room temperature. 1 H NMR (1,1,2,2-tetrachloroethane-d 2, 110 C, 300 MHz): δ 7.1 7.4 (m, aromatic H), δ 2.7 (m, aromatic-ch 2 -) δ 1.7 (m, -CH-), δ 1.3 (m, -CH 2 -), δ 0.9 (m, CH 3 ). 3.2.4 Synthesis of Long Chain Branched Isotactic Polypropylene (LCBPP) 100 ml of toluene and 1 ml of MAO solution were added to a dry 300 ml Parr stainless steal autoclave equipped with a mechanical stirrer. The autoclave was then charged with 40 psi of propylene gas, heated to 45 C, and charged with 1 psi of hydrogen. Propylene gas was used to charge the reaction vessel to a total pressure of 140 psi. 1.1 x 10-7 moles of rac-me 2 Si[2-Me-4-Ph(Ind)] 2 ZrCl 2 diluted in 1 ml of toluene and p-(3-butenyl)styrene were then injected under propylene pressure. The reaction vessel was stirred and maintained at 55 C and 140 psi for 30 minutes. The reaction was terminated by the addition of methanol and cooled. The product was washed with acidic

63 methanol, rinsed with methanol, and dried under vacuum at room temperature. 1 H NMR (1,1,2,2-tetrachloroethane-d2, 110 C, 300 MHz): δ 7.3 7.4 (d, aromatic H), δ 7.1 7.2 (d, aromatic), δ 7.1 (s, aromatic, symmetric), δ 6.7 (m, aromatic-ch=c), δ 5.7 (d, aromatic-c=ch), δ 5.3 (d, aromatic-c=ch), δ 2.7 (t, 2H, aromatic-ch 2 ), δ 1.7 (m, -CH-), δ 1.3 (m, -CH 2 -), δ 0.9 (m, CH 3 ). 3.2.5 Hydrogenation of LCBPP 102 To a 250 ml round bottom flask equipped with condenser and magnetic stir bar was added 1 g of LCBPP, 0.2 g of p-toluenesulfohydrazide, 0.08 g of tripropylamine, and 100 ml of toluene. The mixture was brought to a reflux and stirred for 3 hours. The mixture was then cooled and washed with methanol. The precipitated polymer was filtered, extracted with methanol for 12 hours, and dried in a vacuum oven at 80 C. Complete hydrogenation of styrene functionality was observed as the disappearance of the 1H-NMR peaks at 6.8, 5.7 and 5.2 ppm and the increased intensity of the singlet peak at 7.1 ppm associated with the para-substituted aromatic ring. 3.2.6 Size Exclusion Chromatography Absolute molecular weight, molecular weight distribution, and viscosity were obtained by SEC with triple detectors by Drew Poche at the Dow Chemical Company. The measurements were made at 160 C using 1,2,4-trichlorobenzene. The chromatographic system consisted of a Polymer Labs PL 220 high temperature

64 chromatograph equipped with a Precision Detectors 2-angle laser light scattering detector Model 2040, a Viscotek model 210R viscometer, and a differential refractive index detector. The 15-degree angle of the light scattering detector was used for calculation purposes. Data collection was performed using Viscotek TriSEC software version 3 and a 4-channel Viscotek Data Manager DM400. The system was equipped with an on-line solvent degassing device from Polymer Laboratories. The chromatographic columns used were three PLgel MIXED-B columns obtained from Polymer Laboratories. The samples were prepared at a concentration of 0.1 g of polymer in 50 ml of solvent. The samples were stirred gently at 160 C for 2.5 h. The injection volume was 100 ml and the flow rate was 1.00 ml/min. The system was calibrated with narrow molecular weight polystyrene standards and the detector constants were determined using a Dow Chemical Co. broad polystyrene standard. The refractive index increment, dn/dc, was calculated from the calibrated DRI detector as 0.104 ml/g. Molecular weights for the isotactic polypropylene samples were calculated from the intrinsic viscosity detector using the following Mark-Houwink parameters; K = 1.90x10-4 dl/g and a = 0.725 established for linear isotactic polypropylene from a polystyrene calibration 103. 3.2.7 Small Amplitude Dynamic Oscillatory Shear Rheological measurements were completed on a strain-controlled Rheometric Scientific ARES rheometer using a heated stream of nitrogen gas for temperature control. All measurements were completed under nitrogen atmosphere and

65 sample stability was monitored by replicate measurements. 25 mm parallel plates with a gap slightly exceeding 1 mm were used. Samples were formed into 25 mm discs by vacuum-assisted compression molding. Strain sweeps at frequencies of 1 and 100 rad/s were performed to assure measurements were within the region of linear response. Frequency sweeps were performed at five temperatures between 170 and 190 C in a frequency range of 0.01 to 100 rad/s. 3.3 Synthesis of LCBPP The hypothesized scheme to produce reactive macromonomers through a functionalized chain transfer agent, proposed in Figure 3-3, becomes much more complicated. Along with the styrenic chain transfer agent portion of p-(3-butenyl)styrene, the metallocene should be reactive towards the allyl functional group. This reactivity should lead to a much more complicated reaction where copolymerization and chaintransfer can take place at a number of sites. Figure 3-4 illustrates the updated reaction scheme for the formation of LCBPP polymers via a metallocene-mediated propylene polymerization, using isospecific rac- Me 2 Si[2-Me-4-PhInd] 2 ZrCl 2 /MAO catalyst, in the presence of T-reagent, p-(3- butenyl)styrene (BSt), and a small amount of hydrogen. T-reagent (II) serves as both comonomer and chain transfer agent 90, 92. Copolymerization of the but-3-ene moiety of BSt via 1,2-insertion results in the formation of a copolymer (III) with pendant styrene functionality. On the other hand, incorporation of the styrene moiety of BSt via 2,1- insertion results in a dormant catalyst site (IV) due to steric hindrance from the bulky

66 aromatic ring and confined catalyst site 90. Hydrogen can revive the dormant catalyst site via a subsequent chain-transfer reaction to produce a macromonomer with the reactive allyl group at the chain-end. Theoretically, branches in LCBPP (VI) can be formed when a growing polymer chain either incorporates the allyl chain-end of a macromonomer (IV) or chain transfers to the pendant styrene of a copolymer (III). As discussed later, the predominant route for the formation of branches involves the pendant styrene of the copolymer (III), which serves as the chain transfer agent to form the LCBPP (VI) in the presence of hydrogen.

Figure 3-4: Scheme for the synthesis of LCBPP via the T-reagent approach. 67

68 Slurry polymerizations of propylene were conducted with varying amounts of T-reagent. The catalyst maintained high activity throughout all the concentrations of T-reagent due to an effective chain transfer reaction to styrene and then hydrogen. Catalyst activities of LCBPP1 and LCBPP2 were comparable to both the homopolymerization of propylene and the copolymerization of propylene with 4-phenylbut-1-ene. With higher T-reagent concentrations, LCBPP3 and LCBPP4, the catalyst activity was proportionally reduced, which may be associated with the slow down in chain transfer reaction to T-reagent/H 2. The chemical structures of the obtained materials were examined by 1 H NMR and a representative spectrum is displayed in Figure 3-5. The structures of copolymer and macromonomer could be identified by their unique chemical shifts. Copolymer containing pendant styrene functionality was identified by three vinyl protons with chemical shifts at 6.8, 5.7, and 5.2 ppm, while macromonomer was observed by three allyl protons at 5.9 and 5.1 ppm. Branch points created by the combination of copolymerization and chain transfer of/to T-reagent were identified by comparing the aromatic protons with a chemical shift of 7.1 ppm with the amount of macromonomer observed. Macromonomer only accounted for a minor amount of the aromatic protons at 7.1 ppm leaving branch points as the major contributor. Low levels or lack of macromonomer structure suggested either the macromonomers with allyl-terminated chain-end were incorporated into LCBPP at high conversions, or incorporation of the T- reagent highly preferred insertion of the allyl moiety. Unfortunately, 1 H NMR is unable to distinguish between the two in the final product. The latter should be the preferred

explanation, because the incorporation of highly crystalline, high molecular weight PP macromonomer should be very difficult under these heterogeneous conditions. 69 Figure 3-5: Representative 1 H NMR and assignments of LCBPP prepared by the T- reagent approach.

70 Table 3-1 summarizes the results of several polymerizations using various amounts of T-reagent. The branch point density was calculated from 1 H NMR spectra. The branch point density was proportional to the feed concentration of T-reagent and increased from LCBPP1 to LCBPP4. It is interesting to note that on average 43% of the incorporated T-reagent was a branch point, independent of T-reagent concentration and reaction time. LCBPP4 was reacted twice as long as LCBPP3 but contained similar ratios between the amounts of pendant styrene and branch points observed. These results may imply that copolymerized T-reagent was immediately involved in a chain transfer reaction to form a branch point before the polymer precipitated from solution and the incorporated BSt units became inaccessible. Melting temperature, T m, of LCBPP samples, examined by DSC, showed a gradual reduction of T m with increase of branch density (Table 3-1).

71 Table 3-1: Comparison of linear and LCBPP polymers prepared by rac-me 2 Si[2-Me-4-Ph(Ind)]Zr 2 Cl 2 /MAO mediated propylene polymerizations. b b c Sample [BSt] Catalyst M n,vis M w,vis M n,lalls M w,lalls F BSt [η] d Branch T m (mol/l) Activity a g e (g/mol) (g/mol) (g/mol) (g/mol) (mol%) (dl/g) Density f ( C) PP1 0.000 210 166,900 443,400 160,400 429,900 0.00 2.283 0.99 0.0 158 PP2 g 0.013 h 190 103,700 255,000 95,300 247,800 0.15 i 1.542 1.00 0.0 155 LCBPP1 0.007 225 91,200 237,600 93,900 248,600 0.07 1.485 0.96 1.0 157 LCBPP2 0.013 190 91,200 239,900 91,200 239,500 0.13 1.412 0.91 1.8 155 LCBPP3 0.020 100 95,500 237,600 93,900 246,800 0.15 1.385 0.92 2.2 153 LCBPP4 0.030 64 87,100 238,170 86,900 255,500 0.21 1.370 0.87 3.3 153 a. catalyst activity: [kg of PP/(mmol of Zr h)] b. measured by SEC viscosity detector c. mol% of incorporated BSt d. weight average intrinsic viscosity obtained from SEC viscosity detector e. g = [η] B /[ η] L, where [η] B is the weight-average intrinsic viscosity and [η] L is calculated from the Mark-Houwink parameters f. 1 H NMR branch density: [branch points/10,000 carbon] g. linear copolymer of propylene and 4-phenylbut-1-ene h. concentration of 4-phenylbut-1-ene i. mol% of incorporated 4-phenylbut-1-ene

72 Figure 3-6 shows molecular weight distribution curves of linear polypropylene and LCBPP samples, which were examined by SEC with triple detection. The weightaverage molecular weights determined by light scattering, M w,lalls, are displayed in Table 3-1. Compared with ipp homopolymer (PP1), M w,lalls decreased with the addition of comonomer (PP2) and initially with the addition of BSt (LCBPP1). Further addition of BSt increased M w,lalls only slightly. Hence, all four LCBPPs have a M w,lalls of about 250,000 g/mol. The polydispersity index, M w /M n, of all the resulting polymers remained between 2.5 and 3. differential weight fraction 1.2 1.0 0.8 0.6 0.4 0.2 LCBPP1 LCBPP2 LCBPP3 LCBPP4 PP2 0.0 10 3 10 4 10 5 10 6 10 7 molecular weight (g/mol) Figure 3-6: Molecular weight distribution of linear and LCBPP. Figure 3-7 shows the Mark-Houwink plots of the LCBPPs with respect to a linear standard. The linear standard behaved in a fashion described by the Mark- Houwink relation where K and a can be obtained from the slope and intercept of the

73 Mark-Houwink plot. The linear isotactic polypropylene had a single slope of a = 0.725. The slopes of the Mark-Houwink plot for the LCBPP deviated from the linear standard. The deviation from linear behavior was subtle at low branch point density, but became more apparent as branch point density was increased. The largest change in slope to a = 0.28 was observed for LCBPP4 containing the highest content of branch points. The effective molecular weight, M x, between long branches can be obtained from locating the intersection of linear and branched behavior. For all samples list in Table 3-1, M x was ~500,000 g/mol which are several times larger than M w. These values of M x suggested most of the branches existed in the high molecular weight fraction of the polymer. 1.2 10 1.0 intrinsic viscosity (dl/g) 1 LCBPP1 LCBPP2 LCBPP3 LCBPP4 Linear standard 0.8 0.6 0.4 differential weight fraction 0.2 0.1 10 5 10 6 10 7 molecular weight (g/mol) 0.0 Figure 3-7: Mark-Houwink plots of LCBPP obtained from SEC with triple detection. Full distributions are shown in Figure 3-6.

3.4 Bulk Polymerizations of Propylene 74 LCBPP was produced via bulk polymerizations of propylene with T-reagent and hydrogen. Bulk polymerizations utilizing liquid propylene were completed in an attempt to produce LCBPP with weight average molecular weights greater 300,000 g/mol. Molecular weight is directly proportional to monomer concentration for many metallocenes and Ziegler-Natta catalyst 104. The Mark-Houwink plot of two LCBPPs is shown in Figure 3-8. LCBPP prepared with bulk polymerizations were observed to contain higher degrees of branching despite the introduction of similar concentrations of T-reagent. The LCBPP prepared using a slurry polymerization only displayed an onset of branching at very high molecular weights. As shown in Table 3-2, the branching density calculated from 1 H NMR for this LCBPP differs greatly from that of the branching density calculated from SEC. On the other hand, LCBPP prepared from a bulk polymerization of propylene contained far more branching and results from NMR and SEC were closer in agreement. The polydispersity index of the bulk polymerization was much higher than that of the slurry polymerization. The bulk polymerization conditions allowed for the preparation of high molecular weight branched material. Additionally, the bulk polymerization and high molecular weight material apparently allowed for the preparation of a highly branched LCBPP.

75 1 Linear standard Bulk polymerization Slurry polymerization 10 differential weight fraction 1 intrinsic viscosity (dl/g) 0.1 0 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 molecular weight (by LALLS) (g/mol) Figure 3-8: Mark-Houwink plots of LCBPP prepared by slurry and bulk polymerizations. Table 3-2: Comparison of slurry and bulk polymerizations. Branch Branch M w Density Density Sample (g/mol) PDI (NMR) (SEC) Slurry polymerization Bulk polymerization 255,000 2.7 3.3 0.18 380,000 8.9 3.8 2.3

3.5 Catalyst Effect 76 The selection of catalyst is an important factor in the preparation of LCBPP via the T-reagent approach. High molecular weight isotactic polymer is desirable. The active site of the catalyst should be confined so as to facilitate a dormant active site after the 2,1-insertion of styrene. Incorporation of propylene after the 2,1-insertion of styrene leads to copolymerization of the styrene-moiety and should result in a crosslinked polymer product. The copolymerization of the but-1-ene moiety via 1,2-insertion is preferred. Small amounts of T-reagent incorporated at high conversion should be able to produce the desired LCBPP. Three catalyst systems have been studied in an attempt to understand how the catalyst structure may effect the formation of LCBPP. A popular single-site ansametallocene, dimethylsilanediylbis(2-methyl-4-phenylindenyl)]zirconium dichloride (A), is shown in Figure 3-9 and was used as a baseline for performance. A has all of the desired properties of single-site metallocene for the preparation of LCBPP and has been shown to be an effective catalyst system in the preparation of chain-end functionalized isotactic polypropylene 90, 92. The active site of catalyst A promotes chain-transfer via 2,1- insertion of the styrene moiety while preventing the copolymerization of the styrene. Catalyst A can incorporate alpha-olefins. Another single-site metallocene, dimethylsilane(2-methyl-benzoindenyl)(2-methyl-4-phenyl-4-h-azulenyl)hafnium dichloride (B), was used to compare to catalyst A. Catalyst B is a C1 symmetric ansahalfnocene as compared to the C2 symmetric zirconocene of catalyst A. Despite C1 symmetry the ligand structure of catalyst B promotes the stereo-specific incorporation of

77 propylene to produce high molecular weight isotactic polypropylene. The hafnium metal center and confined active site should promote the formation of a dormant site following 2,1-insertion of styrene. Finally, a simple Ziegler-Natta catalyst, TiCl 3 AA/AlEt 2 Cl (catalyst C), was used as a comparison to the metallocenes. Ziegler-Natta catalysts are well known for their ability to copolymerization propylene and styrene 99 and should not be effective in the preparation of LCBPP. Cl Cl Zr Si Cl Cl Hf Si Figure 3-9: Two metallocenes were used to polymerize LCBPP. Dimethylsilanediylbis(2- methyl-4-phenylindenyl)]zirconium dichloride (A)(left) and dimethylsilane(2-methylbenzoindenyl)(2-methyl-4-phenyl-4-h-azulenyl)hafnium dichloride (B)(right). Table 3-3 summarizes the results of slurry polymerizations using the three catalysts. The incorporation of the T-reagent was monitored by 1 H NMR. The metallocene polymerizations resulted in a polymer product which included copolymerization of the but-1-ene moiety (III) and branch points (VI) as the two major products. Macromonomer was not detected by 1 H NMR in the final product. The ratio between the copolymer and the branch points was about 1 to 1 for catalyst A. Catalyst B prepared a polymer which contained more copolymer (III) than branch points (VI). The

78 polymer produced from the polymerization with catalyst C displayed a distinct difference in that no branch points were obtained. Instead the presence of copolymerized styrene was observed by 1 H NMR for catalyst C.

79 Table 3-3: Comparison of T-reagent/H 2 behavior in various catalyst systems. Slurry polymerizations of propylene in the presence of T-reagent and H 2 Sample Catalyst Activity T-reagent Conversion Structure (III) Structure (IV) Structure (VI) M w,lalls M w /M n T m (wt%) (mol%) (mol%) (mol%) (g/mol) ( C) Catalyst A, linear Catalyst A, LCBPP Catalyst B, linear Catalyst B, LCBPP 210,000 a n/a n/a n/a n/a 430,000 2.7 158 100,000 a 8 0.09 n.o. 0.07 247,000 2.6 153 16,000 a n/a n/a n/a n/a 243,000 2.8 157 13,500 a 4 0.07 n.o. 0.04 174,000 2.9 153 Catalyst C, linear 180 b n/a n/a n/a n/a 380,000 8.2 161 Catalyst C, 200 b 1 0.03 0.03 n/a 410,000 8.9 158 LCBPP a. catalyst activity = [kg PP/(mol of metallocene)/(h)]. b. catalyst activity = [g PP/(g of TiCl3 AA)/(h)]

The conversion of the T-reagent was determined from the 1 H NMR results and it was clear catalyst C lacked the reactivity desired. Only 1 wt% of the T-reagent was incorporated into the polymer. Catalysts A and B had conversions of 8 and 4 wt%, respectively. This suggests that high concentrations of T-reagent would be required to produce polymers containing comparable amounts of T-reagent as obtained with other catalysts. In Figure 3-10, SEC with triple detection was used to analyze the polymers prepared by catalysts A, B, and C. The molecular weight distributions (MWD) of polymers produced by catalysts A and B remained rather narrow (M w /M n < 3.0) while the MWD of catalyst C became slightly broadened. Sparse high molecular weight branching was observed for catalyst A as was observed by a deviation of the Mark-Houwink plot at high molecular weights from a linear standard. Branching was not observed in SEC experiments for the polymer prepared by catalyst B or C.

81 intrinsic viscosity (dl/g) 10 Catalyst A B C Linear stadard 1.0 0.8 0.6 0.4 0.2 differential weight fraction 1 0.0 10 5 10 6 10 7 molecular weight (by LALLS) (g/mol) Figure 3-10: Mark-Houwink plots of LCBPP prepared using a variety of catalyst. Small amplitude oscillatory shear was used to complement the results of SEC. Figure 3-11 and Figure 3-12 show the complex viscosities of LCBPP produced by catalysts A and B, respectively. There was observed an increase in zero-shear viscosity as the branch density increased suggesting the presence of high molecular weight branches. The increase in zero-shear viscosity for catalyst B contradicts the results obtained from SEC intrinsic viscosity. The dynamic storage moduli (Figure 3-13 and Figure 3-14) also suggested the presence of high molecular weight branches in LCBPP prepared by catalyst A and B. As the branch density was increased, the storage modulus shifted to lower frequencies suggesting longer relaxation times. The effect of branching on the storage modulus was more dramatic for catalyst A, but the effect was still observable for catalyst B. The rheological results suggested the formation of branching in both metallocene catalysts, despite the apparent lack of branching observed in SEC.

82 10000 complex viscosity (Pa-s) 1000 100 linear isotactic polypropylene 1.0 branches per 10,000 carbon 1.8 branches per 10,000 carbon 3.3 branches per 10,000 carbon 0.1 1 10 100 frequency (rad/s) Figure 3-11: Complex viscosity at 190 C of LCBPP prepared by catalyst A. 1000 complex viscosity (Pa-s) 100 linear isotactic polypropylene 0.8 branches per 10,000 carbon 1.0 branches per 10,000 carbon 1.9 branches per 10,000 carbon 1 10 100 frequency (rad/s) Figure 3-12: Complex viscosity at 190 C of LCBPP prepared using catalyst B.

83 10000 linear isotactic polpropylene 2.2 branches per 10,000 carbon 3.3 branches per 10,000 carbon storage modulus (Pa-s) 1000 100 10 0.1 1 10 100 frequency (rad/s) Figure 3-13: Dynamic storage modulus at 190 C of LCBPP prepared by catalyst A. 10000 linear isotactic polypropylene 1.0 branches per 10,000 carbon 1.9 branches per 10,000 carbon storage modulus (Pa-s) 1000 100 10 1 0.1 1 10 100 frequency (rad/s) Figure 3-14: Dynamic storage modulus at 190 C of LCBPP prepared by catalyst B.

84 Zielger-Natta catalysts appear to be unpractical for use in the preparation of LCBPP by a T-reagent approach. The catalyst produced polymers containing low levels of T-reagent and failed to prevent the copolymerization of the styrene moiety. Increasing the concentration of T-reagent could lead to the formation of crosslinked materials. Single site ansa-bridged zirconocenes and halfnocenes with a broad range of ligands should be acceptable for the preparation of LCBPP by the T-reagent approach. Here two catalysts with different ligands and metal center were used to prepare LCBPP. Differences in catalyst structure lead to differences in the prepared LCBPP and required different optimal conditions in the polymerization. 3.6 Hydrogen Effect Another significant concern in this scheme is the direct chain transfer reaction to hydrogen, which produces linear PP polymer with a reduced molecular weight. It is important to choose a catalyst which is less sensitive to hydrogen and to apply only the required amount of hydrogen to prepare LCBPP. As depicted in Table 3-4, hydrogen was a necessary component in the process of chain transfer to T-reagent during rac-me 2 Si[2- Me-4-Ph(Ind)] 2 ZrCl 2 /MAO mediated propylene polymerization. In PPH1, the catalyst activity was almost completely retarded by the addition of a small amount of BSt. 2,1- insertion of the styrene moiety created an inactive site which could no longer incorporate propylene monomer 90. With the addition of a small amount of hydrogen (1 psi) in PPH2, the catalyst activity returned to expected levels and the polymer formed was high molecular weight. With an increased amount of hydrogen (PPH3, 10 psi), the molecular

weight of the resulting polymer was reduced. At this high hydrogen concentration, direct chain transfer to hydrogen became a significant side reaction. Therefore, for the preparation of LCBPPs it is important to use only minimum amounts of hydrogen, such that chain transfer via hydrogen does not compete with chain transfer to T-reagent. Table 3-4: Experimental results of rac-me 2 Si[2-Me-4-Ph(Ind)]Zr 2 Cl 2 /MAO mediated propylene polymerizations in the presence of T-reagent (0.007 mol/l) and various hydrogen pressures. Sample Hydrogen Catalyst Activity (psi) [kg of PP/(mmol of Zr h)] M w,lalls (g/mol) PPH1 0 20 n/a PPH2 1 225 232,500 PPH3 10 330 45,000 85 Figure 3-15 depicts the Mark-Houwink plots for two LCBPPs and a linear ipp. The two LCBPPs were prepared with 2 and 10 psi of hydrogen. The M x and a of the two LCBPPs remained similar despite a reduction of M w by 42% for the sample prepared with 10 psi of hydrogen. These results suggested that the structure of the branched material was unaffected by the increased concentration of hydrogen, namely, M x remained essentially unchanged. Although the structure of the branched material was unaffected, the higher concentration of hydrogen decreased the concentration of branched material. The sample prepared with 2 psi hydrogen contained 17 branches per 10,000 carbons while the sample prepared with 10 psi had 14 branches per 10,000 carbons. 1 H NMR further clarified the effect of hydrogen on the samples. 56 % of the incorporated BSt produced a branch point for the 2 psi LCBPP, while only 49% produced branch points in

the 10 psi LCBPP. At high molecular weights, direct transfer to hydrogen competed with chain transfer to pendant styrene groups resulting in the reduction of branch points. 86 10 2 psi hydrogen 10 psi hydrogen Linear standard 0.8 intrinsic viscosity (dl/g) 1 0.6 0.4 0.2 differential weight fraction 0.0 0.1 10 4 10 5 10 6 10 7 molecular weight (by LALLS) (g/mol) Figure 3-15: Mark-Houwink plots of LCBPP. 3.7 Conclusions LCBPP was prepared by the metallocene mediated polymerization of propylene in the presence of p-(3-butenyl)styrene and hydrogen. This approach, dubbed the T-reagent approach, prepared polymers which were observed to contain branching by 1 H NMR, SEC, and oscillatory shear. In comparison to previously discussed approaches (in Chapter 1) to prepare LCBPP the T-reagent approach has several advantages. T-reagent effectively acts as both a comonomer and a chain transfer agent with hydrogen. This

87 differs from the α,ω-diene approach which has two copolymerizable groups. Two comonomer units resulted in a lightly crosslinked structure in the α,ω-diene approach where it was important to minimize the amount of diene incorporated. The T-reagent is designed such that only one functional group is copolymerizable. This design ensures crosslinking does not occur from incorporation of both reactive sites. The experimental results confirmed the T-reagent approach did not facilitate the formation of crosslinked polymer; however, LCBPP contained pendant styrene groups which upon processing in the melt lead to the formation of crosslinked material. These pendant styrene groups were stable under high temperature solution conditions and could be chemically modified to obtain a functionalized isotactic polypropylene and/or melt processable polymer. In the T-reagent approach, branched material was prepared in a one step process. During the reaction the catalyst remained highly active. The conditions of this reaction were similar to those which may be used to produce high molecular weight, high melting temperature homopolymers. This is an advantage over approaches such as the macromonomer approaches which relied on varying conditions and catalysts. The T-reagent approach produced polymers whose properties could be adjusted by altering the concentration of T-reagent present during the polymerization of propylene. SEC with triple detection techniques suggested the presence of high molecular weight branching and oscillatory shear measurements confirmed the presence of branching. A more involved study of the LCBPP s rheological properties follows in Chapter 4.

Chapter 4 Rheological Characterization of LCBPP 4.1 Introduction Of significant interest was to study the flow of the LCBPP melts so possible processing benefits could be identified. As discussed in Chapter 2, the melt flow of linear and branched polymers is qualitatively and quantitatively different; therefore, rheology measurements are extremely helpful in determining the presence and nature of branches contained within a material. Rheological studies of monodisperse, branched polyethylene have shown branched polymers of varying architecture displayed relaxation times, zeroshear viscosities, and activation energies for flow which were sensitive to the density of branching, the length of the branches, and molecular weight of the polymer backbone. Also of rheological importance is the affect of long chain branches on the extensional flow of polymers. Under elongational melt flows LDPE strain-hardens while HDPE with similar zero-shear viscosity does not. Linear and branched polypropylenes have been examined in a similar manner. Polypropylene containing long chain branches strain-hardened while a linear polypropylene did not. These differences in rheology make branched polymers beneficial for some polymer processing operations, such as blow molding and film blowing.

4.2 Experimental 89 4.2.1 Small Amplitude Dynamic Oscillatory Shear Rheological measurements were obtained on a strain-controlled Rheometric Scientific ARES rheometer using a heated stream of nitrogen gas for temperature control. All measurements were completed under nitrogen atmosphere and sample stability was monitored by replicate measurements. 25 mm parallel plates with gap slightly exceeding 1 mm were used. Samples were formed into 25 mm discs by vacuum-assisted compression molding. Strain sweeps at frequencies of 1 and 100 rad/s were performed to assure measurements were within the linear response region. Frequency sweeps were performed at five temperatures between 170 and 190 C in a frequency range of 0.01 to 100 rad/s. 4.2.2 Extensional Flow Measurements Extensional flow measurements were completed by Mitsubishi Chemical Company. Measurements were made on a strain-controlled TA instruments ARES rheometer equipped with the extensional viscosity fixture (EVF). Polymer samples were prepared by first gently pressing the polymer into a sheet at 200 C. This sheet was cut and pressed a second time at 10 kg/cm 2 and 200 C for 10 minutes, producing a sheet 0.7 mm thick. Samples were cut to the size of 1.0 cm x 1.8 cm and loaded onto the EVF. Tensile stress growth was measured at strain rates varying from 0.005 to 5 s -1 at 180 C.

4.3 Small Amplitude Oscillatory Shear 90 Oscillatory shear measurements were obtained at five temperatures between 170 and 190 C. Figure 4-1-A shows the master curves of storage and loss modulus obtained by time-temperature superposition for PP1 (See Chapter 3 and Table 3-1 for polymer descriptions). The linear sample showed characteristic behavior of a viscoelastic material in the Newtonian (terminal) region. At higher frequencies a crossover of the storage and loss moduli, G and G, was observed. At lower frequencies G and G became proportional to ω 2 and ω, respectively. Branched polymers, represented by LCBPP4 in Figure 4-1-B, had longer relaxation times and the proportionality of G and G to ω 2 and ω was not observed at the lowest frequencies measured. These results were consistent with the SEC results which suggested significant branching in the high molecular weight region of the distribution.

91 1000000 100000 A G' (filled) & G" (hollow) [Pa] 10000 1000 100 10 1 0.1 1 10 100 ωa T [rad/s] 1000000 G' (filled) & G" (hollow) [Pa] 100000 10000 1000 100 10 B 1 0.1 1 10 100 ωa T [rad/s] Figure 4-1: Master curves of storage and loss moduli at 190 C. (A) Linear PP1, (B) LCBPP4.

92 Figure 4-2 shows the storage moduli for LCBPP with various branch densities and PP2. Linear PP2 with similar molecular weight to that of the LCBPP, behaved as expected with the slope of G approaching 2 at low frequency. However, the LCBPP were observed to have broadened storage moduli with slopes only approaching values of 1.5 at the lowest frequencies measured, indicating the relaxation times are longer than their linear counterparts. 100000 10000 G' [Pa] 1000 100 10 PP2 LCBPP1 LCBPP2 LCBPP3 LCBPP4 0.1 1 10 100 ω [rad/s] Figure 4-2: Storage moduli of several LCBPP at 190 C Additional evidence of a branched structure was observed by the failure of timetemperature superposition for long chain branched samples. Time-temperature superposition worked well for LCBPP1 and LCBPP2 with low branch point densities, but sample LCBPP3 and specifically sample LCBPP4 showed poor time-temperature

93 superposition. Figure 4-3 shows an attempted superposition of LCBPP4 data at five temperatures. The loss tangent, tan(δ), demonstrates unambiguously the poor superposition, as no vertical scale shifting is allowed for tan(δ), as it is a ratio of moduli. 5 4 190C 185C 180C 175C 170C tan(δ) 3 2 1 0.01 0.1 1 10 100 ωa T [rad/s] Figure 4-3: Attempted time-temperature superposition of LCBPP4. Data were superimposed at high frequency. Reference temperature is 190 C. Failure of time-temperature superposition is characteristic of a thermorheologically complex material. Most commonly this behavior has been attributed to branched polyethylene and has been suggested to depend strongly on the temperature coefficient of chain dimensions 105. Polymers with sufficient branch density and large negative temperature coefficients of chain dimensions are anticipated to be thermorheologically complex. Atactic polypropylene has a temperature coefficient of chain dimensions of -0.1 x 10-3 K -1 106, while polyethylene has a temperature coefficient

94 of chain dimensions of -1.05 x 10-3 K -1 107. The temperature coefficient of chain dimensions for polypropylene is negative, but small in comparison to the coefficient for polyethylene. Failure of time-temperature superposition in LCBPP was less dramatic than what has been reported for branched polyethylene, which could possibly be due to the difference in magnitude of the temperature dependence of chain dimensions. Therefore the observed thermorheological behavior for the branched polypropylenes is in agreement with the hypothesis of Graessley 105. Additionally, these results suggested LCBPP4 contained high molecular weight branches of significant branch density. The complex viscosities of the LCBPP and PP2 are presented in Figure 4-4. The complex viscosity, η *, and particularly zero-shear viscosity, η 0, are extremely sensitive to branching. The molecular weight of PP2 was similar to those of the LCBPP, but the zeroshear viscosities of the LCBPP were considerably higher than linear PP2 (Table 4-1). Additionally LCBPP had broader transitions from shear-thinning behavior at higher frequencies to Newtonian behavior at low frequencies, consistent with broad molecular weight branched materials. Newtonian behavior was difficult to obtain in LCBPP as is clearly illustrated by sample LCBPP4 s lack of a frequency independent viscosity within the range measured. This behavior can be attributed to the long relaxation times of branched polymers.

95 10000 PP2 LCBPP1 LCBPP2 LCBPP3 LCBPP4 η* [Pas] 1000 0.1 1 10 100 ω [rad/s] Figure 4-4: Complex viscosities for various isotactic polypropylenes at 190 C.

96 Table 4-1: Rheological results for LCBPP. Sample M n,lalls M w,lalls Branch b η 0,190 C E a M b (g/mol) (g/mol) Density a (Pa s) (kj/mol) (g/mol) PP1 160,400 429,900 0.0 17,250 39 linear PP2 c 95,300 247,800 0.0 1,030 40 linear LCBPP1 93,900 248,600 1.0 3,200 48 246,000 LCBPP2 91,200 239,500 1.8 3,890 49 232,000 LCBPP3 93,900 246,800 2.2 5,700 50 235,000 LCBPP4 86,900 255,500 3.3 9,030 61 238,200 a. Branch density = branches per 10,000 carbon b. Average molecular weight between branches (M b ) determined from Janzen-Colbly method. c. Copolymer of propylene and 4-phenylbutene Figure 4-5 shows the log-log plot of zero-shear viscosity and weight average molecular weight. The zero-shear viscosity of a linear polymer is expected to follow Eq. 4.1 and Eq. 4.2 where M c is the critical molecular weight for entanglements or approximately twice the molecular weight between entanglements, M e. The five linear isotactic polypropylenes followed the expected behavior for entangled melts 108 and could be described by Eq. 4.3, but the LCBPP deviated from this behavior. For LCBPP the zero-shear viscosity was more sensitive to molecular weight. The zero-shear viscosity of LCBPP4 was nearly three times the zero-shear viscosity of a linear polymer of the same M w.

97 4.5 4.0 3.5 Linear Branched log 10 (η 0 [Pa-s]) 3.0 2.5 2.0 1.5 1.0 0.5 4.6 4.8 5.0 5.2 5.4 5.6 log 10 (M w [g/mol]) Figure 4-5: Relationship of zero-shear viscosity at 190 C and weight-average molecular weight for linear and branched isotactic polypropylenes. The linear fit of the linear polymers is described by Eq. 4.3. 1 0 M w ( w c η for M < M ) 4.1 3.4 η 0 M w for ( M w > M c ) 4.2 16 3.56 1.77x M w η 0 = 10 (Pa s, 190 C) 4.3 Utilizing the measured zero-shear viscosities and absolute weight-averaged molecular weights of the LCBPP, the average molecular weight between branches, M b, was calculated using a model proposed by Janzen and Colby 109. M b was obtained from Eq. 4.4 and Eq. 4.5, with M c = 2M e = 13640 g/mol, M Kuhn = 187.8 g/mol, B = 6, and A = 1.02 x 10-5 calculated from fitting several linear polymers to Eq. 4.6. Table 4-1 contains the results for M b of the LCBPPs. The calculated M b slightly decreased with increasing

branch density. An M b of slightly less than M w suggested sparse branching with high molecular weight branches, which agreed with SEC results. 98 2.4 s / γ M b M w η0 = AM b 1 + M c M 4.4 c 3 9 M b s / γ = max 1, + B ln 4.5 2 8 90M Kuhn 2.4 M w η 0 = AM w 1+ 4.6 M c The Arrhenius activation energy for flow, E a, can be determined using the frequency scale shift factors obtained from the time-temperature superposition of the oscillatory shear data at several temperatures from Eq. 4.7. Arrhenius activation energies for flow are listed in Table 4-1. Even LCBPP1 with the lowest levels of branching showed a significant increase in flow activation energy. The systematic increase in Arrhenius activation energy for flow with branch density can be attributed to the long chain branched structure 77. LCBPP4 with the highest branch density had a flow activation energy higher than the linear standards and branched polypropylenes prepared by electron beam irradiation (48 kj/mol) or diene comonomer (42 kj/mol) methods 50. E a 1 1 a = T exp 4.7 R T T0

4.4 Extensional Flow Measurements 99 An examination of the extensional flow characteristics of LCBPP4 was performed using an ARES rheometer equipped with the extensional viscosity fixture. The extensional stress growth function, η + E (t,ε& ), is shown in Figure 4-6 at various Hencky extension rates, ε&, for LCBPP4. At each extension rate there existed a range of deformation which perfectly tracked the linear viscoelastic response. Additionally, at all extension rates studied, with perhaps the exception of ε& = 0.005 s -1, strain hardening was observed as a sharp increase of η + E (t,ε& ) above the values at ε& = 0.005 s -1. Such strain hardening is believed to be important for polymer processing operations requiring high melt strength, such as fiber spinning and film blowing.

100 0.5 0.2 0.08 0.03 100000. ε = 5 1/s 2 1 0.01 η E + (t,ε) [Pa.s] 10000 0.005 1000 0.01 0.1 1 10 100 time [s] Figure 4-6: Extensional stress growth functions at various strain rates for LCBPP4 at 180 C. Extensional viscosities, η E (ε& ), were obtained from the maximum values of the stress growth functions at each ε& and are plotted against ε& in Figure 4-7. Trouton s rule predicts η ( & ε ) = 3η ( & γ ) 88. At the lowest strain rate the melt behaved as a linear E viscoelastic liquid and the extensional viscosity was three times the zero-shear viscosity within experimental uncertainties. At higher strain rates the extensional viscosity increased with increasing strain rate and reached a maximum value of η E (ε& ) = 11.5*3η 0 at a Hencky rate of 0.08 s -1. At Hencky rates above 0.08 s -1 the extensional viscosity

101 decreased with increasing strain rate, possibly because the sample failed before reaching steady state. The extensional flow properties of polymer melts were extremely sensitive to the presence of branches, with results similar to those obtained for LCBPP4 having been reported 84. 1000000 η E (ε) [Pa.s]. 100000 3η 0 10000 1E-3 0.01 0.1 1 10. ε [s -1 ] Figure 4-7: Extensional viscosity versus strain rate for LCBPP4 at 180 C. Solid line corresponds to 3 times the zero-shear viscosity obtained from small-amplitude oscillatory shear. 4.5 Conclusions LCBPP materials, produced by the combination of T-reagent and metallocene catalyst, were further confirmed to contain high molecular weight branches. The melt behavior observed was similar to previously reported results. Small amplitude oscillatory shear showed control of branch point density allowed for the synthesis of LCBPPs with varying zero-shear viscosities, relaxation times, and Arrhenius activation energies for

102 flow. LCBPP was observed to behave as a strain hardening material under extensional flow. Notable differences were observed in the behavior of LCBPP as campared to previously discussed approaches. The lack of a low frequency region dominated by the storage modulus, as has been reported for the diene approach 46, 47, suggested the formation of trifunctional branch points which prevent the formation of crosslinks. Shear viscosities increased with the addition of branches in contrast to chemical modification approaches which caused a decrease in shear viscosity due to degradation. The observed rheological properties of LCBPP make it an idea material for applications which require high melt strength. The strain hardening observed in LCBPP is useful for polymer processing operations requiring stability in extension such as film blowing and film casting.

Chapter 5 Application of LCBPP as a Dielectric Material 5.1 Introduction Energy storage is an essential element to efficient energy usage, but has long been scientifically challenging 110. Batteries and capacitors are two of the best known energy storage devices, but each fall short of expectations for many applications. Batteries have high energy densities and low power densities, while capacitors 111, 112 have low energy densities and high power densities. It is desirable to increase the energy density of capacitors, which is governed by the dielectric material. In the past decade, metallized polymer film capacitors 113, 114, using semi-crystalline thermoplastic polymers as the dielectric materials, have attracted a great deal of attention due to their desirable properties, such as light weight, low cost, and excellent processability for production of thin films with a large surface areas. Advanced capacitors using biaxial oriented 115, 116 polypropylene (BOPP) thin film displayed high dielectric strength and selfhealing 117, 118 ; however, these metallized polymer film capacitors generally suffer from low energy density (0.5-2.0 J/cm 3 ) due to their low relative dielectric constant (k < 3). There is significant interest in developing dielectric materials for high energy storage capacitors. Current goals set by the Office of Naval Research for capacitors with energy densities of 30 J/cm 3 exceed the current limits (~ 2 J/cm 3 ) of polypropylene-based capacitors. In order to increase the energy storage there exist two main strategies; to

104 increase the dielectric constant or increase the breakdown strength of a dielectric material 119. Energy stored on a capacitor, U, is described by Eq. 5.1, where ε 0 is the permittivity of free space, k is the relative dielectric constant of dielectric material, and E is the applied electrical field. An increase in k or E would increase the energy stored on a capacitor, but an increase in dielectric constant is usually combined with an increase in dielectric loss and an increase in the electric field leads to dielectric layer instability. U 2 = 1 ke 2ε 0 5.1 The instability of the dielectric layer can be expressed by terms of a dielectric strength or a breakdown strength which occurs at a field, E b. The breakdown strength at which a dielectric material fails is affected by numerous factors. It has been shown by several studies that breakdown strength is directly related to material structure. Properties such as purity, crystallinity, morphology, and orientation of the crystalline domain are important factors effecting the breakdown strength of a material 120-123. The breakdown strength of a solid is determined by polymer structure. The structure of a polymeric material is rarely uniform and the breakdown strength does not result from the average of these structures, but instead is a property of the weakest structure. Amorphous domains located throughout the structure of semi-crystalline polymers constitute the weakest section. Lack of orientation leads to low density and molecular scale voids within the amorphous domain. Free volume theory states that these voids lead to instability in the amorphous domains of the polymer film 124, 125. Although theories and predictions of breakdown strength are difficult to prove experimentally there are no arguments that

105 factors such as purity, crystallinity, and orientation effect the breakdown strength of polymeric dielectric materials 120, 122, 123, 126-128. Impurities in a polymeric material can originate from a wide range of sources. Small amounts of catalyst, initiator fragments, solvents, or even the chain-ends of polymer can be considered impurities. After the formation of crystallites these impurities are often located at the boundaries of crystalline material or within the amorphous domains. Impurities can negatively effect the thermal dissipation of the dielectric material. Impurities can cause a local increase in temperature leading to an instability. The location of these impurities further weakens the already weak domain leading to low breakdown strengths. Orientation of the crystalline domain has been observed to increase the breakdown strength of polymeric materials. Orientation of the amorphous domains should coincide with the orientation of the crystalline domains. Increasing the density of the amorphous domain through orientation of the amorphous domain may decrease the volume of free space leading to higher breakdown strengths. It may be more effective to replace E with E b in Eq. 5.1 when attempting to increase the energy storage capabilities of a dielectric material. The breakdown strength usually defines the maximum electrical field which can be applied to the capacitor. Depending on the application, the capacitor may only be run at a fraction of the breakdown strength to ensure a long lifetime. The processing benefits of LCBPP have been discussed in the previous chapter. These processing benefits have been shown to improve the processing capabilities of polypropylene and other polymers for various processing techniques. Additionally, these

processing benefits may allow for the preparation of polypropylene films with improved properties for use in dielectric applications. 106 5.2 Experimental 5.2.1 Materials The materials used in this chapter were obtained from ExxonMobil and Borealis. PP 4342C2, obtained from ExxonMobil, is an ultra clean polypropylene resin designed for capacitor applications. Special care has been employed to remove catalyst residues, volatiles, gels, or dust particles from the resin. Daploy WB135HMS, obtained from Borealis, is a high melt strength isotactic polypropylene designed for use in applications requiring drawability of the polymer melt. These materials were dried in a vacuum oven at 80 C for 12 hours prior to use. The LCBPP materials used in the analysis of breakdown strength were purified prior to measurements. Samples were dissolved in xylene at 150 C and crystallized from solution at room temperature. The precipitated material was collected by filtration and washed with acetone and methanol. The procedure of dissolution, crystallization, and filtration was repeated an additional two times before the samples were dried in a vacuum oven at 80 C for 12 hours.

5.2.2 Preparation of Films 107 Polypropylene films were prepared using a Carver Inc. hydraulic press equipped with 15 x 15 heated plates. For films of thickness greater than 100 micrometers, the prescribed amount of dried polymer sample was placed into a picture frame mold (ASTM D4703-03) and heated at 190 C for 5 minutes. The sample was then pressed to a pressure of 2,500 pounds at 190 C for an additional 5 minutes. The films were cooled at room temperature in the molds following the press. For films of thickness less than 100 micrometers, the prescribed amount of dried polymer, preferably pressed into thicker sheet, was placed between two telfon sheets and heated at 190 C for 5 minutes. The sample was then pressed to a pressure of 2,500 to 10,000 pounds at 190 C for an additional 5 to 15 minutes. The films were cooled at room temperature between the Teflon sheets following the press. 5.2.3 X-Ray Diffraction Diffraction patterns were obtained on a Phillips MPD diffractometer equipped with a Cu Kα source. Patterns were collected from 2θ = 3 to 30 at an interval of Δ(2θ) = 0.02. Data analysis was preformed using Jade software. 2D wide-angle x-ray scattering (WAXS) patterns were collected on a Rigaku D/MAX RAPID II diffractometer equipped with a Cu source and a curved imaging plate. Transmission patterns were collected using a 240 second exposure with a 0.3 mm pinhole collimator. Data analysis was preformed using a combination of Rigaku AreaMax and Jade software.

5.2.4 Breakdown Strength 108 The breakdown strength of the prepared films was measured using an automated polarization measurement system. Thin, circular gold electrodes with a 1 mm diameter were sputtered onto the polypropylene films. Increasing electrical fields at 50 MV/m steps were applied to the films at 10 Hz until a breakdown event was witnessed. 5.3 Degree of Crystallinity The crystallinities of branched polymers are typically lower than those of their linear analogues. Branch points and an increased number of chain ends act as impurities in formation of the crystal structure, reducing the perfection of crystals and decreasing the volume fraction of crystalline material. The amplitude of the effect on crystallinity from the branches depends on the structure and density of the branches. LDPE containing several short chain branches per 1,000 carbons has a significantly reduced melting temperature and a lower degree of crystallinity, while LCBPE containing only a few branches per 10,000 carbons only has a very slight decrease in melting temperature and crystallinity. Table 5-1 shows the effect of the T-reagent on the melting temperature (T m ) of several LCBPP. The T m of the homopolymer, PP1, prepared by a Me 2 Si[2-Me-4- Ph(Ind)]Zr 2 Cl 2 /MAO catalyzed polymerization was 158 C. The addition of a small amount of comonomer in sample PP2 reduced the T m to 155 C. A similar decrease in T m was observed for the addition of the same amount of T-reagent in LCBPP2. For LCBPP4

containing 3.3 branches per 10,000 carbons the T m of the material only decreased to 153 C. Table 5-1: Melting temperatures (T m ) of linear polypropylenes and LCBPP. Sample Branch T m,dsc Density a ( C) PP1 0 158 PP2 0 155 LCBPP1 1 157 LCBPP2 1.8 155 LCBPP3 2.2 153 LCBPP4 3.3 153 a. Branch density measured by 1 H NMR, branches per 10,0000 carbons 109 Figure 5-1 and Figure 5-2 show the diffraction patterns of the Exxon Mobil capacitor grade PP and a LCBPP containing 2.3 branches per 10,000 carbons. Both samples were prepared by compression molding as described in the experimental section of this chapter. The peak locations of patterns indicate the α crystalline structure of isotactic polypropylene. From these patterns, the degree of crystallinity was determined by comparing the integrated areas of the amorphous and crystalline domains obtained in the diffraction pattern Eq. 5.2 129. The degree of crystallinity was similar for all three materials (Table 5-2). The Exxon Mobil capacitor grade polypropylene was observed to have the highest degree of crystallinity. The commercial branched polypropylene and the LCBPP contained only a slightly lower amount of crystalline domain despite their lower melting temperatures. These observations are consistent with thinner crystal lamellae as a result of the additional branch point impurities. Degrees of crystallinity were determined from thermograms to confirm the results obtained from XRD. The X C,DSC was determined using Eq. 5.3 where ΔH is the measured enthalpy of fusion and ΔH u is the calculated

110 enthalpy of fusion for a perfect crystal 130. The value of ΔH u used was 209 J/g 130, 131. The degrees of crystallinity measured by DSC agreed with those obtain from XRD. The degree of crystallinity of the LCBPP was only slightly lower than that of a commercially available capacitor grade isotactic polypropylene. 2000 1500 Counts 1000 500 0 5 10 15 20 25 30 35 40 2θ Figure 5-1: Diffraction pattern of the ExxonMobil capacitor grade PP.

111 1200 1000 800 Counts 600 400 200 0 10 20 30 40 2θ Figure 5-2: Diffraction pattern of a LCBPP containing 2.3 branches per 10,000 carbons. X Acrystalline, = 5.2 C XRD ( A + A ) amorphous crystalline X ΔH, = 5.3 ΔH C DSC u Table 5-2: Degrees of crystallinity for several isotactic polypropylenes. Samples were thick films prepared by compression molding. Sample X C,XRD X C,DSC T m ( C) PP 4342C2 0.71 0.65 160 Daploy WB135HMS 0.70 0.65 159 LCBPP 0.68 0.63 153

112 The addition of a small amount of long chain branches decreased the melting temperature only slightly and did not change the crystalline structure of the produced films. The addition of long chain branches by the T-reagent approach did not decrease the degree of crystallinity. These are important factors for the consideration of LCBPP as a dielectric material as the melting temperature and degree of crystallinity have been shown to be important factors in regards to the breakdown strength. 5.4 Orientation The orientation of crystallites and polymer chains within a polymer sheet or film affects physical properties of the material. The orientation of crystal lamellae in the direction of a uniaxial deformation leads to a polymer film that is stronger in the direction of the deformation. Biaxial orientation of a polymer films leads to a film in which the crystal lamellae are oriented in the plane of the film. Biaxial-oriented polypropylene (BOPP) film is used in a variety of applications. BOPP films have gained popularity in packaging applications due to good strength, stiffness, optical properties, and moisture barrier properties. BOPP films have also become the popular dielectric material in power capacitors due to their ability to prepare thin films of high degrees of orientation with few irregularities 123. The orientation of polypropylene films is a required step in the production of capacitors. Unoriented and uniaxially oriented polypropylene films have breakdown strengths significantly lower than BOPP 123. A comparison between two polypropylene films of similar quality led to breakdown strengths which were four times

113 greater for a BOPP than an unoriented film. It has been hypothesized that this enhancement in dielectric strength can be related to the improved uniformity of the BOPP. The orientation of polypropylene films was obtained by stretching the sample mounted between to supports, one of which could be moved by adjusting a knob. Temperature control was maintained by the use of a high temperature oil bath. Crystallization of the oriented films was completed by quickly transferring the stretch films to a second oil bath of the appropriate temperature. Films were first uniaxially oriented at temperatures slightly below their melting temperatures. At these temperatures the deformation of the spherulitic morphology results in the formation of a fibriller crystalline structure. Figure 5-3 shows the 2D diffraction pattern for the PP 4342C2 film prepared in this manner. Before deformation the diffracted intensity is independent of χ, indicating crystal lamellae are randomly orientated about the polymer films. After orientation, the intensity of the 040 reflection becomes concentrated at χ = 90 degrees with respect to the direction of the deformation (MD). This angle of χ indicates perpendicular orientation of the b-axis of the monoclinic unit cell of isotactic polypropylene for the 040 plane at 17 degrees 2θ. The polymer chain axis (c-axis) is located perpendicular to the b-axis and should be oriented in the direction of deformation. This can be further illustrated by measuring the 2D diffraction pattern through a second or third plane of the orientated film (Figure 5-4). The combined diffraction patterns indicate that the c-axis should be oriented parallel to the direction of deformation.

114 MD Figure 5-3: 2D WAXS pattern for the Exxon Mobil capacitor grade polypropylene, PP 4342C2, before uniaxial orientation (left) and after uniaxial orientation (right). MD MD TD TD N N Figure 5-4: WAXS patterns from three different material planes of a polypropylene film with uniaxial orientation. The degree of orientation can be estimated by integrating the intensity of the scattering signal as a function of χ for any angle 2θ. The breadth of this peak, quantified as a full-width at half-maximum (FWHM), relates to the degree of orientation for that crystallographic plane. Higher degrees in orientation lead to a narrow peak closely concentrated about an angle χ. It is important to note that all measurements of orientation have been made on samples with high draw ratios. At these high draw ratios, the

115 observed orientation has reached a maximum and FWHM cannot be increased by further deformation. The χ dependence of the 040 plane for the ExxonMobil material is shown in Figure 5-5. Before uniaxial orientation the intensity of the 040 diffraction signal was uniformly distributed among all angles of χ, indicating a randomly oriented b-axis. After uniaxial orientation the 040 plane became concentrated about equator (90 = χ), indicating an orientation of the b-axis perpendicular to the deformation direction. counts (Au) 5500 5000 4500 4000 3500 3000 2500 2000 1500 1000 500 0 Before uniaxial orientation After uniaxial orientation -500 0 30 60 90 120 150 180 χ Figure 5-5: χ dependence of the 040 crystallographic plane for ExxonMobil PP4342C2 before and after uniaxial orientation at 145 C. Three other materials were uniaxially orientated in the same fashion as PP 4342C2 and their χ dependence is shown in Figure 5-6. The results of these orientations, quantified by the resulting FWHM of the 040 crystallographic planes, are displayed in Table 5-3. The value of (1/FWHM)x100 has been used to simplify the comparison of orientations. Uniaxial orientation at temperatures below T m of all four of the materials

116 resulted in orientation which looked very similar in nature. At these temperatures orientation of the crystallites is obtained through the deformation of spherulites into a fibrillar structure 132. For condition such as these branching contained in the material has no impact on the final orientation of the material. 7500 5000 2500 LCBPP Counts (Au) 12500 10000 7500 5000 2500 3000 2500 2000 1500 1000 500 4000 3000 2000 1000 0 30 60 90 120 150 180 Metallocene PP 0 30 60 90 120 150 180 WB135HMS 0 30 60 90 120 150 180 PP4342C2 0 30 60 90 120 150 180 χ Figure 5-6: χ dependence of the 040 crystallographic planes for various uniaxial oriented polypropylenes. All films were stretched at 145 C. Table 5-3: Orientation of several polypropylene materials stretched at 145 C Sample 1/(FWHM)x100 PP4342C2 10 Daploy WB135HMS 11 Metallocene PP 11 LCBPP 10

117 The polymer melt is deformed over a wide range of temperatures for many processing techniques designed to impart orientation. It is necessary to stretch the material at a temperature below or near the T m of the polymer in order to maintain high degrees of orientation. At temperatures above T m the created orientation can relax rather quickly to random orientation. Figure 5-7 shows the 2D WAXS patterns of a PP4342C2 film which had been uniaxially stretched at 170 C. The high temperature process was unable to impart a significant degree of orientation. Figure 5-8 shows the 2D WAXS patterns of a Daploy WB135HMS film stretched at 170 C. The high temperature process was observed to impart orientation in the branched material. The same could be said for LCBPP, which for varying branch densities and molecular weights showed similar behavior (Figure 5-9 and Figure 5-10). The long chain branches contained within these materials significantly increases the relaxation times of the melt and prevents the obtained orientation from relaxing. MD Figure 5-7: ExxonMobil PP 4342C2 before (left) and after (right) uniaxial orientation at 170 C followed by crystallization at 100 C.

118 MD Figure 5-8: Borealis WB135HMS PP film before (left) and after (right) uniaxial orientation at 170 C followed by crystallization at 100 C. MD Figure 5-9: LCBPP film before (left) and after (right) uniaxial orientation at 170 C followed by crystallization at 100 C.

119 3500 3000 2500 2000 1500 1000 500 0 Mw = 246.8 kg/mol, 2.2 branches per 10,000 carbon counts 0 3500 50 100 150 3000 2500 2000 1500 1000 500 0 Mw = 248.6 kg/mol, 1.0 branches per 10,000 carbon 0 3500 50 100 150 3000 2500 2000 1500 1000 500 0 Mw = 174 kg/mol, 1.1 branches per 10,000 carbon 0 30 60 90 120 150 180 χ Figure 5-10: Chi dependence of several LCBPP after uniaxial orientation at 160 C (T>T m ) followed by crystallization at 20 C. 5.5 Breakdown Strength The breakdown strength of a dielectric material is a statistical event and therefore can be analyzed using Weibull statistics. In Weibull analysis, series of samples are tested for the field at which they breakdown and ranked according to the field at which they failed. From this analysis the breakdown strength of a material can be described as a probability of failure. As the field applied to the capacitor is increased the likelihood of a breakdown event increases. Weibull plots of the breakdown strengths are pictured in Figure 5-11 for four different materials. The breakdown strength of the linear

120 metallocene polypropylene prepared was observed as the lowest with a Weibull characteristic breakdown strength of 230 MV/m. The number of low field breakdowns observed in the linear metallocene films was much higher than any other sample. The crystallinity, melting temperature, and molecular weight of the unoriented linear metallocene polypropylene were lower than PP4342C2 (330 MV/m) and WB135HMS (390 MV/m), but were similar to the LCBPP (415 MV/m). The lower breakdown strength of the linear metallocene polypropylene was likely due to a combination of these values. The breakdown strengths of the branched polypropylenes were higher than the capacitor grade polypropylene, PP4342C2, despite lower crystallinities and melting temperatures. Higher breakdown strengths are likely due to small amounts of orientation in the films. In many cases, orientation was witnessed in branched films after melt compression as can be observed in the unoriented LCBPP film in Figure 5-9 whose χ dependence is plotted in Figure 5-12.

121 2.0 1.5 1.0 ln(ln(1/(1-medianrange))) 0.5 0.0-0.5-1.0-1.5-2.0-2.5-3.0-3.5 Metallocene PP PP 4342C2 Daploy WB135HMS LCBPP 0 100 200 300 400 500 600 breakdown field (MV/m) Figure 5-11: Weibull analysis of the electrical breakdown of polypropylene films at room temperature.

122 11000 10000 9000 counts 8000 7000 6000 Mw = 380 k/gmol, 2.3 branches per 10,000 carbon 0 50 100 150 200 250 300 350 χ Figure 5-12: Chi dependence of a LCBPP film containing orientation prepared by compression molding. 5.6 Conclusions Polypropylene containing long chain branches should be an acceptable and advantageous material for use as a dielectric material in high energy density capacitors. The processing benefits related with extensional deformations observed in films making processes are described in Chapter 4. In this chapter, long chain branch containing polypropylenes showed potential to produce films with high degrees of orientation at high processing temperatures. Branched polypropylene had higher breakdown strengths than linear polymers despite lower melting temperatures and crystallinities. This increase

in breakdown strength could be related to slight degrees of orientation in the branch containing polypropylene. 123

Chapter 6 Conclusions and Suggestions 6.1 Summary of Present Work Long chain branched polypropylene (LCBPP) could be directly synthesized from the metallocene-mediated polymerization of propylene in the presence of T-reagent and hydrogen. During the polymerization, chain transfer to the styrene moiety was possible. Hydrogen was necessary to complete the chain transfer mechanism, but too much hydrogen could compete with chain transfer to the T-reagent. Direct chain transfer to hydrogen is undesired as it prevented the formation of LCBPP. The T-reagent can also be copolymerized. Copolymerization of the allyl-like portion of the T-reagent was observed as pendant styrene groups after the polymerizations. These groups typically accounted for roughly half of the T-reagent incorporated. Macromonomer was not observed after the polymerizations despite other indications of chain-transfer. The mechanistic scheme for the formation of LCBPP via the T-reagent approach has been discussed in Chapter 1. Two major pathways to the formation of LCBPP were described; the direct chain transfer to T-reagent to obtain macromonomer with allyl-like chain end followed by the incorporation of the macromonomer and copolymerization of the T-reagent through the allyl-like moiety to obtain a copolymer followed by chain transfer to the copolymer. From the observed structures and lack of observed macromonomer there exist two major possibilities. Firstly, that direct chain transfer to the

125 T-reagent prepared an allyl-terminated polypropylene macromonomer which was extremely reactive. Secondly, that copolymerization of the allyl-like moiety was preferred over the direct chain-transfer. The first possibility should be the less plausible of the two methods proposed. There have been several attempts at preparing polypropylene macromonomers with allyl chain ends 33, 38-41, but high conversion of the macromonomer has not be observed in any these reports. Therefore, the second possibility is considered a more likely pathway. The prepared LCBPP were analyzed by size exclusion chromatography (SEC) with triple detection capabilities. Triple detection SEC confirmed the presence of long chain branches for LCBPP prepared using the metallocene rac-me 2 Si[2-Me-4-Ph- Ind] 2 ZrCl 2, but was unable to detect any branching in materials produced from Mitsubishi Chemical Company s C1 metallocene catalyst. The observed high molecular weight branching was observed to increase with the increase in T-reagent concentration and length of the reaction. Calculated branch densities were much lower than those observed through 1 H NMR. Small amplitude oscillatory shear was used to analyze the LCBPP melt behavior. LCBPP prepared by rac-me 2 Si[2-Me-4-Ph-Ind] 2 ZrCl 2 and Mitsubishi Chemical Company s C1 metallocene catalyst both displayed rheological behavior consistent with high molecular weight sparse branching. The measured complex viscosities were observed to shear thin more intensely than linear polymers of a similar molecular weight. The zero-shear viscosities were more sensitive to changes in molecular weight and depended on the branch density of the LCBPP. The relaxation times of LCBPP were longer than linear polymers of similar molecular weight. These results contradicted those

126 obtained from SEC with triple detection for the LCBPP prepared using Mitsubishi Chemical Company s C1 metallocene catalyst. Additional oscillatory shear results were typically of long chain branching. The LCBPP was observed as having high Arrhenius activation energies for flow. LCBPP with high branch densities was thermorheologically complex, as time-temperature superposition began to fail at higher concentrations of branching. Strain-hardening was observed during the extensional deformation of a LCBPP melt. Strain-hardening is typically not observed for linear polypropylenes and results from the presence of long chain branches or an extremely high molecular weight tail. SEC results showed that the MWD of the prepared LCBPP remained rather narrow; therefore, the observed strain hardening arises from the presence of high molecular weight branching. 6.2 Final Conclusions A novel approach to the preparation of long chain branched polypropylene has been developed. T-reagent combined with an appropriate metallocene catalyst has allowed for the formation of branches through a complicated pathway which involves both copolymerization and chain transfer. The T-reagent approach has shown promise for a convenient route to the preparation of long chain branched polypropylene. In contrast to other approaches, the T-regent approach is a one-pot process that produces high molecular weight polymers through a non-degenerative process. Although special care

127 must be taken to avoid crosslinked material, gels are generally avoided during and after the polymerization. The resulting polymers have shown common signs of branching in numerous characterization techniques. The rheological results suggested that LCBPP materials should be applicable for processing techniques which demand high melt strength. Low relaxations times and high zero shear viscosities should be beneficial for applications such as thermoforming and blow molding, while strain hardening should be desirable for processes involving elongational melt flows. 6.3 Potential Applications Long chain branched (LCB) polymers have value in processing techniques which demand high melt strength, including thermoforming 2-4, film blowing 4, 5, extrusion foaming 6, 7, and blow molding processes 8, 9. The addition of a small percentage of LCBPP to a traditional polypropylene resin can improve the behavior of the melt during these processes. Polypropylene foams are becoming more popular, but are considered difficult to prepare do to their low melt strengths. During the extrusion foaming process, polypropylene has a tendency to produce foam with an open cell structure. This open cell structure is undesired. LCBPP which strain hardens can improve the quality of foam by promoting a closed cell structure 6, 7. In Figure 6-1, the cell morphology of several foams is shown for comparison. The LCBPP, prepared by reactive extrusion, allowed for the

preparation of a closed cell structure which was rather uniform. In comparison, foam prepared from the linear polypropylene was non-uniform and contained many open cells. 128 Figure 6-1: SEM of the cell morphology of polypropylene foams prepared from linear polypropylene (left) and branched polypropylenes (middle and right) 6. Extrusion foaming and other processing techniques have been studied preciously and the advantages of using a branched material are well known. A more novel area of application is that of dielectrics or electrical applications. Polypropylene containing long chain branches should be an acceptable and advantageous material for use as a dielectric material in high energy density capacitors. LCBPP showed potential to produce films with high degrees of orientation at processing temperatures at or above the melt temperature of polypropylene. Branched polypropylene had higher breakdown strengths than linear polymers despite lower melting temperatures and crystallinities. This increase in breakdown strength could be related to slight degrees of orientation in the branch containing polypropylene.