Nitroxyl-Mediated Peroxide Modification of Polypropylene

Benjamin Rhys Jones

A thesis submitted to the Department of Chemical Engineering In conformity with the requirements for the degree of Master of Applied Science

Queen’s University Kingston, Ontario, Canada (May 2021)

The thermal stability of alkoxyamine derivatives of polypropylene (PP) is examined through studies of model hydrocarbon systems and atactic-PP substrates. Primary, secondary and tertiary alkoxyamines are integral to modern techniques for modifying the architecture of linear PP materials, and detailed knowledge of their stability at polymer processing temperatures is needed to further understand the underlying principles of nitroxyl-based formulations.

GC analysis of alkoxyamines derived from 2,4-dimethylpentane + TEMPO show that the tertiary regioisomer is susceptible to disproportionation to alkene + HOTEMPO at temperatures as low as 140 oC. Extension of these thermolysis experiments to atactic PP-g-HOTEMPO mirrored the model compound results, with appreciable extents of nitroxyl generation observed over a 20 min timescale at 180 oC. This reaction extent is not expected to dramatically alter the performance of nitroxyl-based formulations that modify the structure and composition of PP homopolymers.

The yield of H-atom abstraction by peroxide-derived alkoxy radicals is another fundamental reaction that underlies polyolefin modifications. Indirect measures based on analysis of peroxide byproducts are supplemented with a fluorescence spectroscopy method that quantifies polymer- bound alkoxyamines. H-atom transfer from high density polyethylene (HDPE) and poly(ethylene-co-propylene) (EPR) was gained by using napthoyloxy-TEMPO, whose fluorophore supports a highly sensitive analytical method. This technique is extended to studies of reactive HDPE/EPR blending to ascertain the extent of peroxide migration from one phase to the other during the compounding process. The data indicates that mass transfer of the peroxide is competitive with the rate of peroxide breakdown, causing both polymers to incur structure modification irrespective of how the initiator is charged to the blend components.

Acknowledgements I would like to thank Dr. Scott Parent for his expertise and guidance through this new field of study for me. I would like to thank my lab mates: Mitch Sadler, Kyle Ford and Matt Hawrylow for making my time in the lab more enjoyable. I’d also like to thank all the other staff and students in the department who made these times memorable. Special shoutout to Andrew

Sellathurai for keeping me sane during the COVID-19 pandemic.

Lastly, I would like to thank my partner, Kylie, and my family, who were always there to support me. To my mother, who always encouraged me to continue my education, while only encouraging my brother, Craig, to eat more bacon. This is for you.

Table of Contents Abstract…………………………………………………………………………………………….i

Acknowledgements………………………………………………………………………………..ii

List of Figures…………………………………………………………………………………….vi

List of Tables…………………………………………………………………………………….vii

List of Schemes…………………………………………………………………………………viii

List of Equations………………………………………………………………………………….ix

List of Abbreviations……………………………………………………………………………...x

Chapter 1 Introduction………………………………………………………………………….…1

1.1 Polypropylene Degradation………..……………………………………………….....1

1.2 H-Atom Abstraction by Peroxide Initiators...………………………………………....2

1.3 Nitroxyl Trapping of Alkyl Radicals………………………..………………………...4

1.4 Alkoxyamine Instability………………………………..……………………………...7

1.5 Model Compounds…………………………………………………..………………...8

Chapter 2 Experimental Methods………………………………………………………………..10

2.1 Materials………………………………………………………………..…………....10

2.2 Instrumentation and Analysis……………………………..…………………………11

2.3 Synthesis of 1-methoxy-2,2,6,6-tetramethyl-1-piperidine (Me-TEMPO)…………...12

2.4 Synthesis of (2,2,6,6,-tetramethylpiperidine-N-oxyl)-2,4-dimethylpentane regioisomers (DMP-TEMPO isomers)……………………………………...…………...12

2.5 Synthesis of 1-(2,2,6,6-tetramethylpiperdine-N-oxyl)-cyclohexane (cyclohexane-g-TEMPO)...... 13

2.6 Thermolysis of cyclohexane and DMP alkoxyamines…………………………...…..14

iii

2.7 Synthesis of PPatactic-g-HOTEMPO…………………………………………..……...14

2.8 Thermolysis of PPatactic-g-HOTEMPO…………………………………………….....14

2.9 Synthesis of Model Compounds for AE Determinations……………………………15

2.10 Synthesis of 4-(1-Naphthoyloxy)-2,2,6,6-tetramethylpiperdine-1-oxyl (N- TEMPO)……………………………………………………………………………….....15

2.11 Synthesis of Dodecyloxy-(4-(1-Naphthoyloxy)-2,2,6,6-tetramethylpiperdine) (Dodecane-NTEMPO)……………………………………………………………….…..16

2.12 Synthesis of Polymer-g-NTEMPO samples (EPR/HDPE-g-NTEMPO)………...…17

2.13 Synthesis of EPR/HDPE-g-NTEMPO Polymer Mix……………………………….18

Chapter 3 –Tertiary Alkoxyamine Disproportionation…………………………………………..19

3.1 Introduction…………………………………………………………………..………19

3.2 Results and Discussion………………………………………………………..……..20

3.3 Conclusion………………………………………………………………………..….26

Chapter 4- Fluorescence Determination of Macroradical Yields in Polyolefin Melt Blends……………………………………………………………………………………………28

4.1 Introduction………………………………………………………………………..…28

4.2 Results and Discussion…………………………………………………………..…..31

4.2.1 Abstraction Efficiency Validations………………………………….…..…31

4.2.2 Independent HDPE and EPR Modifications……………………………….33

4.2.3 HDPE + EPR Blend Modifications…………………………….……….....38

4.3 Conclusion………………………………………………………………………..….41

Chapter 5 Future Work…………………………………………………………………………..42

5.1 Tertiary Alkoxyamine Instability…………………………..………………………………...42

5.2 Peroxide Migration Trials………………………..…………………………………………..42

References………………………………………………………………………………………..42

Appendix…………………………………………………………………………………………49

Appendix A 1H NMR spectrum of dodecane-g-NTEMPO……………………………...49

Appendix B 1H NMR of PP-g-TEMPO………………………………………………….50

List of Figures

Figure 1. GC signal integrations from nitroxyl liberated by Cyclohexane-g-NTEMPO and DMP-g-NTEMPO at 140oC………………………………………………………………...……22

Figure 2. GC signal integrations from nitroxyl liberated by Cyclohexane-g-NTEMPO and DMP-g-NTEMPO after 20 min …………………………………………………………………23

Figure 3 H-atom abstraction efficiency measurements acquired with different analytical techniques………………………………………………………………………………………..33

Figure 4 Calibrations of dodecane-g-NTEMPO in cyclohexane (295nm excitation, 280-580nm emission at 20oc) and dodecane-g-NTEMPO in xylenes (295nm excitation, 315-430nm at 125oc) ……………………………………………………………………………………………35

Figure 5. Comparison of dodecane-g-NTEMPO and EPR-g-NTEMPO fluorescence spectra in cyclohexane (20oC, 295 excitation wavelength)..……………………………………………..36

Figure 6 Comparison of HDPE-g-NTEMPO and dodecane-g-NTEMPO fluorescence spectra in xylenes (125oC, 295 excitation wavelength)..………………………………………………...37

Figure 7. EPR-g-NTEMPO and HDPE-g-NTEMPO alkoxyamine yields……………………...38

Figure 8. Fluorescence spectra for EPR-g-NTEMPO isolated from the compounded blend and its parent material …………………………………………………………………………...40

Figure 9. Fluorescence spectra for HDPE-g-NTEMPO isolated from the compounded blend and its parent material……………………………………………………………………………40

List of Tables

Table 1. Amount of nitroxyl liberated from PP-g-HOTEMPO at 140oC…………………….....25

Table 2. Amount of nitroxyl liberated from PP-g-HOTEMPO after 20 min …………………...25

Table 3. AE of DCP acting upon various model hydrocarbons at 160oC……………………….32

vii

List of Schemes

Scheme 1. Production and -scission of tertiary macroradicals …………………………………1 Scheme 2. Peroxide thermolysis and hydrogen abstraction……………………………………….2 Scheme 3. Nitroxyl trapping of alkyl radicals produced by peroxide initiator……………………5 Scheme 4. Disproportionation products of 4-TEMPO-2,4,6-trimethyl heptane…………………..8 Scheme 5. Model alkoxyamine studies…………………………………………………………..20 Scheme 6: Cumyloxyl scission and H-atom abstraction products of AE determinations……….29 Scheme 7. Structure of NTEMPO……………………………………………………………….30

viii

List of Equations

Equation 1. Abstraction Efficiency………………………………………………………………..3 Equation 2. Trapping Ratio………………………………………………………………………..5 Equation 3. Induction time provided by a nitroxyl formulation…………………………………..6

List of Abbreviations 1o – primary 2o - secondary 3o – tertiary AE – abstraction efficiency BDE – bond dissociation energy

CDCl3 – deuterated chloroform DCP – dicumyl peroxide DMP – 2,4-dimethylpentane DMP-TEMPO – (2,2,6,6-tetramethylpiperidine) 2,4-dimethylpentane isomers DMSO – dimethylsulfoxide EPR – ethylene propylene rubber

. FeSO4 7H2O – iron (II) sulfate heptahydrate g – gram GC – gas chromatography

H2O – water

H2O2 – hydrogen peroxide

H2 – hydrogen HOTEMPO – 4-hydroxy-2,2,6,6-tetramethylpiperdine-N-oxyl HDPE – high density polyethylene k – rate constant L130 – 2,5-di(tert-butylperoxy)-2,5-dimethyl-3-hexyne L231 – 1,1-bis(tert-butylperoxy)-3,3,5-trimethylcyclohexane Me-TEMPO - 1-methoxy-2,2,6,6-tetramethyl-1-piperidine MW – molecular weight

N2 – nitrogen

NaSO4 – sodium sulfate NMR – nuclear magnetic resonance

NTEMPO – 4-(1-Naphthoyloxy)-2,2,6,6-tetramethylpiperdine-1-oxyl Min – minutes MS – mass spectroscopy PE – polyethylene PIB – polyisobutylene PP – polypropylene ppm – parts per million t1/2 – initiator half lives tind – induction time TEMPO – 2,2,6,6-tetramethylpiperidine-N-oxyl

TMIO - 1,1,3,3-Tetramethylisoindolin-2-yloxy TPE – Thermoplastic Elastomer TPV – Thermoplastic Vulcanizate VTES – vinyl triethoxy silane

Chapter 1

Introduction

1.1 Polypropylene Degradation The mechanism by which polypropylene degrades when heated in the melt state with peroxides is well established. [1][2][3]. H-atom abstraction by peroxide-derived radical intermediates from the polymer yields primary, secondary and tertiary alkyl macroradicals, the latter of which are susceptible to -scission, resulting in a loss of molecular weight (Scheme 1). Given that H-atom transfer from the tertiary position is preferred, [4] reductions in polypropylene melt viscosity can be extensive, particularly when the reaction is conducted above the 160°C melting temperature of a propylene homopolymer. [5] Furthermore, macroradical scission yields a terminal olefin and a secondary alkyl radical, which facilitates continued modification of chain architecture by radical chemistry. [6]

Scheme 1. Production and -scission of tertiary macroradicals [2]

Tertiary macroradical cleavage is the dominant molecular weight altering process in polypropylene, overcoming radical termination reactions such as combination and disproportionation. This effect is exploited by some polymer producers to narrow the molecular weight distribution of polydisperse polypropylene by “controlled degradation”. However, in many cases, degradation is an undesired outcome, such as in the radical-mediated grafting of

1 functional monomers such as maleic anhydride and vinyl trialkoxysilanes. Therefore, there is continued interest in radical chemistry that functionalizes polypropylene without incurring severe losses in molecular weight. [7]

1.2 H-atom Abstraction by Peroxide Initiators Organic peroxides are widely used to initiate solvent-free, melt-state reactions of commodity polyolefins. Selecting a particular initiator is usually based on the desired reaction temperature, which must be higher than the polymer melting point. In all cases, initiator activation occurs by thermolysis to yield alkoxy radicals, which can either abstract an H-atom from the polyolefin, or cleave to a methyl radical and ketone. [8] Scheme 2 illustrates peroxide thermolysis and possible pathways, where R1 is generally a phenyl, methyl or tertiary alkyl substituent.

Scheme 2. Peroxide thermolysis and hydrogen abstraction

As illustrated in Scheme 2, the fate of peroxide-derived alkoxy radicals includes H-atom abstraction and fragmentation, with the proportion of each being sensitive to reaction temperature and polyolefin structure [9]. It does not include combination reactions with alkyl radicals to form the corresponding ethers [10] since these terminations are kinetically uncompetitive [11]. Given that the objective of a polyolefin modification is to produce macroradical intermediates, direct H-atom transfer to alkoxy radicals is highly desirable. The

2 methyl radical produced by alkoxy radical cleavage is a much less potent H-atom abstractor, preferring less productive, alternate pathways such as monomer addition [12].

Since abstraction by alkoxy radicals is not quantitative, H-atom transfer yield is a key variable in polyolefin modification process development. The abstraction efficiency, defined as the fraction of alkoxy radicals that abstract an H-atom, is a simple measure of this reaction yield [9]. It is derived from measurement of the relative amounts of alcohol and ketone generated by peroxide thermolysis, as described in equation 1.

k [RH] AE = k[RH] +k

Equation 1. Abstraction efficiency

Abstraction efficiencies have been reported for different peroxides acting upon a range of polymers and associated model compounds [4] [13] [14] [15]. In general, t-butoxyl is more efficient than cumyloxyl, and polymers bearing sterically unencumbered C-H bonds with low homolytic bond dissociation energies are the most reactive. Some studies include reports of include the regioselectivity of H-atom transfer for hydrocarbons bearing different types of C-H bonds. Garrett et al. [4] determined through a model compound of PP that the cumyloxy radical preferentially abstracts the hydrogen in the tertiary position over the primary or secondary

3 positions (3o: 64%, 2o: 8% , 1o: 28%). This is problematic, as discussed in Section 1.1, since tertiary radicals are prone to -scission.

1.3 Nitroxyl Trapping of Alkyl Radicals Nitroxyls are “stable” radicals that quench alkyl radicals by combination to produce isolable alkoxyamines. [16] This occurs at the diffusion limit of reaction velocity, with bimolecular rate constants on the order of 107-109 M-1s-1 [17]. The most popular application of this chemistry involves the control of styrenic and acrylic radical populations during polymerization processes, wherein a dynamic equilibrium is established between radical combination and alkoxyamine thermolysis, such that a small, pseudo steady-state radical population is sustained [18]. By maintaining very small radical concentrations, these controlled radical polymerizations become pseudo-living systems that are capable of producing low polydispersity products and block copolymer structures.

Radical trapping by nitroxyls is also used in antioxidant formulations, wherein quenching alkyl radical intermediates interrupts the air oxidation cycle. These “chain breaking acceptor” applications require the alkoxyamine functionality to be stable, in contrast to the controlled radical polymerizations described above. This concept has been extended to peroxide-initiated crosslinking reactions of ethylene-rich thermoplastics, wherein alkyl macroradicals that are generated in the early stages of initiator thermolysis are trapped by nitroxyl (Scheme 3), thereby preventing crosslinking until the polymer melt assumes the desired shape of a mold cavity [19].

In these cases, nitroxyls are classified as scorch protectants, since they delay the onset of thermoset formation.

Scheme 3. Nitroxyl trapping of alkyl radicals produced by peroxide initiator

Formulating nitroxyl-mediated polyolefin modifications is mostly concerned with amount of nitroxyl relative to the number of radicals introduced by the initiator. This is known as the trapping ratio (TR, Equation 2), which accounts for the fact that an initiator produces two radicals for every peroxide bond in the molecule. A TR value of 0 denotes a nitroxyl-free reaction, while a TR of 1 provides enough nitroxyl to quench all the radicals generated by the initiator. This interpretation of TR assumes that alkoxyamine formation is irreversible and stoichiometric, which may not apply to systems that suffer from alkoxyamines instability [20].

moles of nitroxyl TR = moles of radical produced by initiator

Equation 2. Trapping ratio

The speed by which nitroxyls trap alkyl macroradicals and methyl radicals results in an induction period wherein alkoxyamine production is the sole reaction outcome. This induction period (tind) is a simple function of the peroxide thermolysis rate constant (kd) and the nitroxyl trapping ratio, as described by equation 3 [18]. This equation assumes that the reaction temperature is constant, that the nitroxyl does not affect peroxide thermolysis, and that the alkoxyamine products are stable over the timescale of the polymer modification process.

1 푡 =− ln [1 − TR] 푘

Equation 3. Induction time provided by a nitroxyl formulation

The extension of nitroxyl chemistry to polypropylene was a key advance, in that it allowed this scission-prone polyolefin to be crosslinked into lightly branched and thermoset architectures.

This was accomplished using nitroxyls bearing polymerizable functionality, such as 4- acryloyloxy-2,2,6,6,-tetramethylpiperidin-N-oxyl (AOTEMPO) and 4-vinylbenzoic-2,2,6,6- tetramethylpiperidin-N-oxyl (VBTEMPO) [21] [22]. Macroradical trapping by these nitroxyls transform the material into a macromonomer without changing its molecular weight distribution.

Beyond the induction period associated with the formulation trapping ratio, peroxide-initiated oligomerization of these acrylic or styrenic groups crosslinks the material to produce the desired chain architecture.

These polypropylene modifications generate primary, secondary and tertiary alkoxyamines at reaction temperatures well in excess of the material’s melting point of 160 oC. Maximum efficiency can only be realized if all three alkoxyamines are stable over the course of the modification reaction, since crosslinks are comprised of this functional group. Therefore, insight

6 into the high-temperature stability of these 1o, 2o and 3o alkoxyamines of TEMPO is needed to complete our understanding of the technology.

1.4 Alkoxyamine Stability Alkoxyamine instability can occur through two decomposition pathways, thermolysis to a carbon-centred radical + nitroxyl, and disproportionation to give an olefin + hydroxylamine.

Thermolysis rates can be measured through nitroxyl exchange experiments, wherein the alkoxyamine is heated in the presence of a different nitroxyl, and the progress toward mixed alkoxyamines is monitored. Disproportionation rates can be determined by monitoring the loss of alkoxyamine to the corresponding olefin byproducts.

While studies of styrenic and acrylic based systems are many, due to their applicability to controlled radical polymerization [20] [23] [24], reports of the aliphatic alkoxyamines of present interest are few. Scott et al. [25] studied exchange and disproportionation yields for a range of aliphatic and allylic alkoxyamines at elevated temperature, and observed that TEMPO-heptane was stable over extended periods, suggesting that 2o alkoxyamines are stable under polymer modification conditions.

A limited study by Ozols et al. [20] on a tertiary alkoxyamine, 4-TEMPO-2,4,6-trimethyl heptane in DMSO, showed 85% disproportionation of the starting material to hydroxylamine and olefin when it was held at 100 oC for 60 min (Scheme 4). Note that the hydroxylamine resulting from 3o alkoxyamine disproportionation is readily oxidized in air back to its nitroxyl [26]. This has potential implications for a PP modification process, since a significant alkoxyamine disproportionation process could regenerate TEMPO, thereby lengthening induction times and affecting polymer modification yields.

Scheme 4. Disproportionation products of 4-TEMPO-2,4,6-trimethyl heptane [23]

1.5 Model Compounds Precise characterization of small changes to the chemical structure of polyolefins is complicated by the low functional group content of the derivatives, and their low solubility in organic solvents. Model compounds that contain the reactive functionality of a polymer are well suited to these studies, since reaction products can be separated from unmodified material and subjected to a wide range of analytical techniques, including 1H NMR spectroscopy and mass spectrometry.

As such, model systems have figured prominently in studies of alkoxyamine products of nitroxyl radical trapping experiments [4] [25] [20] [23] [27].

However, interpreting model compound data requires an appreciation of key differences between small molecules and macromolecules. By virtue of their high molar mass, polymers have a very low proportion of chain ends compared to model systems. For example, the use of a model such as 2,4-dimethylpentane to simulate PP is complicated somewhat by a higher proportion of methyl groups, which can affect measurements such as abstraction efficiency [4]. Furthermore,

8 some researchers have reported higher H-atom abstraction yields from certain model compounds, which are presumed to be due to polymer chain entanglement effects that are not present in the model system [28].

The physical and chemical properties of the medium can also differ between polymer and model reactions. Not only is a polymer melt more viscous, but its functional group content is relatively small. If a polar functional group is concentrated in the model compound experiment, then the dielectric constant of the solution may differ from that of the polymer melt, which may factor into the observed reactivity [12]. In the case of -scission of alkoxy radicals, higher polarity solvents lead to increased rates, due to stabilization of the transition state to the incipient carbonyl compound [9].

Chapter 2

Experimental Methods

2.1 Materials

The following materials were used as received from Sigma-Aldrich (Oakville, Ontario): 2,4- dimethylpentane (99%), 1,1-bis(tert-butylperoxy)-3,3,5-trimethylcyclohexane (L231, 92%), dicumyl peroxide (DCP, 99.0+%), pentane (99%), ethyl acetate (99.5%), pyridine (99.0+), 1- naphthoyl chloride (97%), cyclohexane (99+%), dimethyl sulfoxide (DMSO, 99.8%), hydrogen peroxide (H2O2, 50%), acetophenone (99%), sodium chloride (99.0%), sodium bicarbonate

(99.7%), 2-phenylpropan-2-ol (cumyl alcohol, 97%), dodecane (>99%), 2,2,6,6- tetramethylpiperidine-N-oxyl (TEMPO, 98%), 4-hydroxy-2,2,6,6-tetramethylpiperdine-N-oxyl

. (HOTEMPO, 97%) and iron (II) heptahydrate (FeSO4 7H2O, +99%). Acetone (>99.8%), hydrochloric acid (1N), sodium sulfate anhydrous (99%) and hexanes (95%) were used as received from Fisher Scientific (Ottawa, Ontario). Atactic Polypropylene (PPatactic, MW =

10,000g/mol) was used as received from Scientific Polymer Products Inc (Ontario, New York).

High density polyethylene (HDPE, SCLAIR 2710) was used as received by Nova Chemicals.

Hydrogen (H2, 99.999%), nitrogen (N2, 99.999%) and compressed air were used as received from Praxair (Kingston, Ontario). Ethylene propylene rubber (EPR, 60% ethylene content) was received from Scientific Polymer Products Inc. (Ontario, New York) and was purified via dissolution-precipitation and dried under vacuum.

2.2 Instrumentation and Analysis.

Analysis of tertiary alkoxyamine thermal instability was conducted using a Varian CP-3800 gas chromatograph equipped with a Chrompack CP8771 silica column (30 m x 0.25 mm x 8CB 1

μm) using 2.0 mL/min of hydrogen as the carrier gas. Injector temperature was set to 275 oC, while the detector temperature was held at 275 oC. The oven temperature was set to 35 oC and initially held for 5 min, ramped to 100 oC at 5 oC/min, held for 5 min, ramped to 150 oC at 10 oC/min, held for 5 min, ramped to 200 oC at 15 oC/min, held for 2min, ramped to 280 oC at 20 oC/min and held for 5min. 1H NMR spectra were produced using a Bruker AC-400 spectrometer.

The solvent used was CDCl3, with chemical shifts reported relative to chloroform (7.24 ppm).

1 For PPatactic-g-TEMPO samples, the polymer was dissolved in CDCl3 and H NMR spectra were acquired with 128 scans and a delay time of 5 sec.

Abstraction efficiency values were determined using a Varian CP-3800 gas chromatograph equipped with a Chrompack CP8771 silica column (30 m x 0.25 mm x 8CB 1 μm) using 1.8 mL/min of hydrogen as the carrier gas. Injector and detector temperatures were set to be held at

275 oC. The oven temperature was set to 100 oC and initially held for 2min, ramped to 150 oC at

20 oC/min, held for 2 min, ramped to 195 oC at 10 oC/min, held for 5 min, ramped to 220 oC at 3 oC/min, held for 2 min, ramped to 280 oC at 25 oC/min, held for 2 min.

Fluorescence analysis was performed using a Photon Technology International

Spectrofluorometer with a Xenon arc lamp as a source with a digital excitation monochromator.

EPR samples were dissolved overnight at room temperature in cyclohexane. HDPE samples were dissolved in xylenes at reflux then quickly transferred to the heated cuvette holder (60 oC).

Model compounds were dissolved in either cyclohexane or hot xylenes to match the polyolefin samples. Felix32 software was used to take the fluorescence spectra.

2.3 Synthesis of 1-methoxy-2,2,6,6-tetramethyl-1-piperidine (Me-TEMPO)

Me-TEMPO was prepared as described previously [29] with minor modifications. To a solution

. of DMSO (10 mL), TEMPO (1.003 g, 6.42 mmol), and FeSO4 7H2O (3.104 g, 11.65 mmol), a solution comprised of 30% H2O2 (2.0 mL, 19.40 mmol) and DMSO (20 mL) was added dropwise over 1 hr. The reaction proceeded at room temperature under N2 for 1.5 hr. Following this, the reaction mixture was diluted in H2O (350 mL) and extracted with hexanes (4 x 50 mL).

The organic layer was washed with H2O (6 x 100 mL) and subsequently dried with a brine wash

(1 x 100 mL) and Na2SO4. The solution was concentrated, removing hexanes by rotary evaporation to yield a clear yellow liquid (0.55 g). The product was not purified further, as the hexanes would not interfere in its GC analysis and the product has a low boiling point as well.

1 Product was confirmed via H NMR analysis in CDCl3 with the appearance of a sharp singlet at

3.60 ppm, which was assumed to be the methyl group attached to the oxygen of the TEMPO

(29% purity).

2.4 Synthesis of (2,2,6,6,-tetramethylpiperidine-N-oxyl)-2,4-dimethylpentane regioisomers

(DMP-TEMPO isomers)

L-231 (92% purity) was added to acetone to create a stock solution. This was performed to combat the high viscosity of the initiator, allowing for higher precision when weighing out the material for reactions. L-231 in acetone (675 uL of 67.4 mg/mL, 0.14 mmol) was added to a 1 mL glass vial and acetone was removed by evaporation. 2,4-dimethylpentane (1.67 g, 1.67 mmol) was added to the vial as well as (0.046 g, 0.294 mmol) TEMPO. The clear orange

12 solution was transferred to a 10 mL stainless steel reaction vessel along with a stir bar. The vessel was pressurized to 350 psi with nitrogen, stirred for 5 min and then degassed. This degassing process was repeated three times to ensure the oxygen content in solution was minimal. The reaction proceeded with the vessel being placed in an oil bath at 125 oC for 115 min (6 x t1/2 L231) under pressure. The resulting solution was a clear, faint yellow colour. GC analysis was performed on the solution as is, with no purification or concentration of the products. This was to ensure that low boiling point by-products could also be determined by GC.

Naphthalene was added to the reaction mixture as an external standard.

2.5 Synthesis of 1-(2,2,6,6-tetramethylpiperdine-N-oxyl)-cyclohexane (cyclohexane-g-

TEMPO)

L-231 was added to acetone to create a stock solution. This was performed to combat the high viscosity of the initiator, allowing for higher precision when weighing out the material for reactions. (675 uL of 67.4 mg/mL, 0.138 mmol) L231 (92% purity) in acetone was added to a 1 mL glass vial and acetone was removed by evaporation. Cyclohexane (1.67 g, 1.98 mmol) was added to the vial as well as TEMPO (0.050 g, 0.320 mmol). The clear orange solution was transferred to a 10 mL stainless steel reaction vessel along with a stir bar. The vessel was pressurized to 350 psi with nitrogen, stirred for 5 min and then degassed. This degassing process was repeated three times to ensure the oxygen content in solution was minimal. The reaction

o proceeded with the vessel being placed in an oil bath at 125 C for 115 min (6 x t1/2 L231) under pressure. The resulting solution was a clear, faint yellow colour. GC analysis was performed on the solution as is, with no purification or concentration of the products. Naphthalene was added to the reaction mixture as an external standard.

2.6 Thermolysis of cyclohexane and DMP alkoxyamines

To a 10 mL stainless steel reactor vessel, 1.0 g of the unpurified reaction mixture for DMP-

TEMPO or cyclohexane-g-TEMPO was added, as well as a stir bar. The vessel was pressurized to 350 psi with N2 gas, stirred for 5 min and subsequently degassed. This degassing process was repeated three times to ensure the oxygen content in solution was minimal. The vessel was then placed in an oil bath heated to the prescribed reaction temperature (140 – 180oC) and allowed to run for the pre-determined reaction time (20 – 120 min). The reaction mixture was analysed via

GC, with naphthalene being used as an external standard.

2.7 Synthesis of PPatactic-g-HOTEMPO

To a 100 mL round bottom flask, 9.20 g of PPatactic was added with a stir bar. The viscous material was melted at 115 oC for 15 mins and then allowed to cool to room temperature. L231

(0.38 g, 1.25mmol) and HOTEMPO (0.39 g, 2.51 mmol) were added to the round bottom flask and purged with N2 three times. The deoxygenated reaction mixture was then placed in an oil

o bath at 115 C and the reaction proceeded for 5 hours (5 x t1/2 L231). The initiator decomposition by-products, Me-TEMPO and any free HOTEMPO was removed by dissolution precipitation in xylenes/acetone. This was filtered and dried in vacuo overnight.

2.8 Thermolysis of PPatactic-g-HOTEMPO

In a 100 mL round bottom flask, 6.0 g PPatactic was added with a stir bar. The vessel was gassed and degassed three times with N2. The vessel was subsequently placed in an oil bath set to its appropriate reaction temperature (140, 160, 180 oC) for the predetermined reaction time (20, 40,

60 min). The reaction was allowed to cool down to room temperature before 20 mL of acetone

14 was added to dissolve any of the free HOTEMPO. 1H NMR characterization of the olefin content of the product was acquired.

2.9 Synthesis of Model Compounds for AE Determinations

In a 10 mL stainless steel reactor vessel, 2.00 g of the model compound (pentane, cyclohexane,

DMP) was added along with DCP (0.040 g, 0.148 mmol) and TEMPO (0.046 g, 0.294 mmol).

The reaction vessel was pressurized to 350 psi with N2 gas, stirred for 5 min and subsequently degassed. This degassing process was repeated three times to ensure the oxygen content in solution was minimal. The reactor vessel was placed in an oil bath set to the appropriate reaction temperature (160 oC) and ran for 5 peroxide half lives (27.5 min). The reaction was subsequently allowed to cool to room temperature before being decanted. GC analysis was performed on the unpurified reaction, with consideration on the acetophenone and cumyl alcohol ratios.

2.10 Synthesis of 4-(1-Naphthoyloxy)-2,2,6,6-tetramethylpiperdine-1-oxyl (N-TEMPO)

NTEMPO was prepared and purified according to the method of Jones et al [30] with minor modifications. Pyridine was dried via molecular sieves overnight to remove moisture from the solvent before use. In a 25mL round bottom flask, HOTEMPO (0.766 g, 4.445 mmol) was added to 5 mL of dry pyridine and allowed to dissolve at room temperature. An addition funnel was quickly placed on the round bottom flask and to it, 1-naphthoyl chloride (1.03 g, 5.40 mmol) in 1 mL of dry pyridine was added, producing a sharp yellow solution. Nitrogen was purged through the system and the 1-naphthoyl chloride mixture was added dropwise to the HOTEMPO solution. Once everything was added, the reaction was allowed to run for 18 hr at room temperature. This yielded an orange solution with white precipitate, to which 1 mL of cold

15 distilled water was added and allowed to dissolve over 30 min. The solution was then poured over 50 mL of distilled ice water. The solution was then transferred into a 125 mL separatory funnel and extracted in ethyl acetate (3x25 mL), worked up with 1M hydrochloric acid (3x15 mL), neutralized with saturated sodium bicarbonate solution (3x20 mL), washed with distilled water (2x20 mL), dried with saturated brine solution (2x20 mL) and then dried further with sodium sulfate, yielding a deep orange solution. Solvent was removed in vacuo to yield orange crystals. Recrystallization was performed in ethyl acetate to yield vibrant orange needles (1.10 g,

3.260 mmol, 61% yield). Purity was determined by melting point (reported: 101-102 oC, observed 101 oC)

2.11 Synthesis of Dodecyloxy-(4-(1-Naphthoyloxy)-2,2,6,6-tetramethylpiperdine)

(Dodecane-NTEMPO)

In a 250 mL round bottom flask, HOTEMPO (0.919 g, 5.35 mmol) was added to 50 mL of dodecane. This reaction was heated to 150 oC before an addition funnel was attached containing

DCP (0.915 g, 3.38 mmol) in 15mL of dodecane. The DCP solution was added in four separate additions every two peroxide half lives (30 min). Once all additions had been made, the reaction proceeded an additional two hours at 150 oC. The product was isolated by Kugelrohr distillation, removing the more volatile solvent and DCP thermolysis and abstraction byproducts, yielding a light yellow-brown oil. This was purified via column chromatography (5% hexanes : 95% ethyl acetate) before proceeding to the next step. Dodecane-HOTEMPO alkoxyamine (0.89 g, 2.61 mmol) was then added to 5 mL of dry pyridine in a 25 mL round bottom flask. An addition funnel was added containing 1-naphthoyl chloride (1.37 g, 7.20 mmol) in 2.5 mL of dry pyridine.

The system was purged with nitrogen, followed by dropwise addition of the 1-naphthoyl chloride

16 solution. The reaction was allowed to proceed for 17 hr at room temperature, yielding a deep yellow solution with white precipitate, to which 0.5 mL of distilled water was added and allowed to proceed an additional 30 min. The solution was then poured over 50mL of distilled ice water, then extracted in ethyl acetate (2 x 20 mL), acidified with 1M hydrochloric acid (3 x 15 mL), neutralized with saturated sodium bicarbonate solution (3 x 20 mL), washed with distilled water

(2 x 20 mL) and then dried with brine (2 x 20 mL), followed by sodium sulfate. This yielded a

1 deep orange-brown oil (0.80 g, 1.58 mmol, 23% yield). H NMR: 0.88 (m, 6H, CH3-CH2-), 1.17

– 1.31 (m, 27H, O-N-C(CH3)2, Ali-H), 1.78 (m, 4H, O-N-C(CH3)2-CH2 and O-CH-CH2), 2.05

(m, 2H, N-C(CH3)2-CH2), 3.77 (m, 1H, CH-O), 5.40 (m, 1H, (O=CO)-CH)), 7.56 (m, 3H, aromatic), 7.87 (d, 1H, aromatic), 8.01 (d, 1H, aromatic), 8.16 (d, 1H, aromatic), 8.90 (d, 1H, aromatic)

2.12 Synthesis of EPR-g-NTEMPO and HDPE-g-NTEMPO

In an 80 mL beaker, 2.50 g of finely ground polymer (either EPR or HDPE) and was solution- coated with an acetone solution (750 uL) containing DCP (6.25 mg, 23.10 umol) and NTEMPO

(34.0 mg, 101.0 umol). The acetone was allowed to completely evaporate prior to compounding.

The coated polymer was charged as 2 equal samples ~1.25 g to a preheated DSM 5cc Twin

Screw Compounder at 100 rpm and mixed for 5 peroxide half lives at the specified reaction temperature. The first sample extruded was discarded, as it was slightly contaminated with the previous sample that had been extruded from the instrument. Prior to fluorescence analysis, samples were purified from residual nitroxyl and initiator byproducts through dissolution/precipitation, filtered and dried overnight under vacuum.

2.13 Synthesis of EPR-g-NTEMPO / HDPE-g-NTEMPO Blend

In a 50 mL beaker, finely chopped EPR (2.00 g), was coated with an acetone (750 uL) solution containing NTEMPO (35.7 mg, 10.06 umol) and DCP (10.8 mg, 4.00 umol). After the acetone was allowed to evaporate, the mixture was compounded in a DSM Micro 5cc Twin Screw

Compounder for 10 min at 100 oC to create a masterbatch. In a separate 50mL beaker, finely ground HDPE (2.00 g) was coated with an acetone (500 uL) solution containing NTEMPO (18.9 mg, 5.60 umol). After acetone removal, this mixture was masterbatched in the microcompounder at 140 oC for 10 min.

The EPR and HDPE masterbatches (0.5 g each) were charged to the hopper of the microcompounder and mixed at 160 oC at 100 RPM for 30 min (5 DCP half-lives).

The EPR fraction of the reacted blend was dissolved away from the mixture blend by soaking in cyclohexane (5mL) at 40 oC for 72 hr. The resulting cyclohexane solution of EPR-g-NTEMPO was isolated from undissolved HPDE-g-NTEMPO by filtration. The EPR-g-NTEMPO was recovered from solution by precipitation from acetone, dried under vacuum, and dissolved in cyclohexane for fluorescence analysis. HDPE-g-NTEMPO isolated by the cyclohexane extraction process was purified via dissolution/precipitation, dried under vacuum, and dissolved in hot xylenes for fluorescence analysis.

Chapter 3

Tertiary Alkoxyamine Disproportionation

3.1 Introduction

Nitroxyl-based formulations that are designed to provide scorch protection and/or polyolefin functionalization require their alkoxyamine products to be stable over the time course of the peroxide modification. Although reversible thermolysis + recombination is not a particular concern, since the alkoxyamine remains intact, disproportionation to hydroxylamine + olefin is undesirable, as it cleaves the pendant groups that establish polymer crosslinking.

An early study by Scott et al. [25] showed that secondary alkoxyamines derived from TEMPO do not undergo thermolysis or disproportionation when held at 160 oC over reasonable time frames, only demonstrating some evidence of degradation after 5 hr. This data provides reassurance that that losses in secondary alkoxyamines (and, by extension primary alkoxyamines) will not be significant during standard polyolefin modifications.

The first report of tertiary alkoxyamine instability was given by Garrett et al. [4] during a study of H-atom abstraction from 2,4-dimethylpentane, a model compound of polypropylene, in which a correction was required to account for progressive alkoxyamine disproportionation at 160 oC.

Ozols et al. [20] continued this line of inquiry by studying the stability of 4-(2,2,6,6- tetramethylpiperdine)-2,4,6-trimethyl heptane in DMSO at 100 oC. 1H NMR analysis revealed

85% loss of the tertiary alkoxyamine to disproportionation after 60 min. Given that the polarity of DMSO is inconsistent with that of a PP melt,[8] Ozols extended the study to TEMPO-based alkoxyamines of atactic polypropylene, which showed a 26% decline in alkoxyamine content after 30 min at 165 oC, and a 30% decline after 30 min at 180 oC.

These preliminary reports suggest that tertiary alkoxyamine instability may influence polypropylene modification processes, thereby warranting further study. Therefore, investigation of disproportionation rates and yields were continued in a series of model compound and atactic-

PP experiments in an effort to better understand the importance of this side reaction.

3.2 Results and Discussion

The experimental program started with model hydrocarbons that are amenable to precise analytical characterizations, in particular gas chromatography. Initial studies utilized cyclohexane as a model of secondary alkoxyamine reactivity. These experiments were designed to confirm the stability of this functionality, as well as to develop confidence in the experimental methods. Since it produces only one alkoxyamine isomer and yields only cyclohexene as a disproportionation product, its analytical studies are relatively simple.

Cyclohexane model studies

ONR 2 ROOR D + .ONR 2 125 oC air . + HONR2 ONR2 Cyclohexane-g-TEMPO

ONR 2 Dimethylpentane model studies

ROOR + .ONR 2 125 oC

ONR 2

D or ONR 2

DMP-g-TEMPO air + HONR .ONR 2 2

Scheme 5. Model alkoxyamine studies

Cyclohexane-g-TEMPO was prepared by reaction of a 0.95 trapping ratio mixture with L-231 as a peroxide initiator (Scheme 5). The product was analyzed to determine the amounts of Me-

TEMPO and cyclohexane-TEMPO as well as to confirm that no residual nitroxyl was present in the sample.

Heating cyclohexane-g-TEMPO under a pressurized nitrogen atmosphere showed no evidence of

TEMPO or cyclohexene over 2 hours at 140 oC. Similarly, no evidence of disproportionation products was observed after 20 min at 160 oC, and only trace amounts of TEMPO, near the detection limit of the GC, were seen after 20 min at 180 oC. These results are consistent with the literature, [29] confirming that secondary alkoxyamines and, by extension, primary alkoxyamines, are robust at these temperatures over practical reaction times.

Extending this methodology to 2,4-dimethylpentane provided insight into tertiary alkoxyamine stability. The focus remained on monitoring the concentration of TEMPO that is liberated from

DMP-g-TEMPO by disproportionation. Heating the model compound to 140 oC for 60 min transformed the sample from a pale-yellow solution to a deeper orange that is consistent with liberated nitroxyl. This change was not observed in the cyclohexane-g-TEMPO trials, and confirmation of the presence of free TEMPO was confirmed by GC analysis (Figure 1). The initial concentration of TEMPO was near the detection limit of the GC, but rose significantly with increasing heating time, increasing tenfold over a 60 min period.

40.0 35.9 35.0

30.0

25.0 21.6

20.0

14.3

GC integration GC 15.0

10.0

5.0 2.9 3.5 0.0 0.0 0.0 0.0 0.0 0.0 0 20 40 60 80 100 120 Dimethylpentane Reaction Time (mins) Cyclohexane

Figure 1. GC signal integrations from nitroxyl liberated by Cyclohexane-g-TEMPO and DMP-g-TEMPO at 140 oC

Figure 2 provides GC integrations of TEMPO released by model alkoxyamines after heating for

20 min at 140, 160, and 180 oC. Whereas cyclohexane-g-TEMPO showed only marginal degradation at the highest temperature, DMP-g-TEMPO released a substantial amount of nitroxyl under equivalent conditions.

35.0

30.0 28.8

25.0

20.0 17.0

15.0 GC integration GC

10.0

5.0 3.5 1.0 0.0 0.0 0.0 135 145 155 165 175 185 Dimethylpentane Temperature (C) Cyclohexane

Figure 2. GC signal integrations from nitroxyl liberated by Cyclohexane-g-TEMPO and DMP-g-TEMPO after 20 min.

Attempts to close material balances by quantifying the hydrocarbon-based products of alkoxyamine disproportionation met with limited success. Although signals in the elution time range of DMP-g-TEMPO were observed, definitive structural assignments and GC response factors could not be made without authentic samples of each alkoxyamine regioisomer. Although two pronounced signals remained consistent with reaction time while one declined significantly, these changes cannot be assigned unambiguously to the disproportionation reactions of interest.

A final series of experiments were performed on an alkoxyamine derivative of low molecular weight, atactic PP. Graft-modification of the polyolefin was accomplished in the same manner as the model compounds, using HOTEMPO as opposed to TEMPO to limit the potential for losses of nitroxyl to sublimation during the polymer grafting process. Heating the resulting PP-g-

HOTEMPO under various conditions was followed by extraction of the product with acetone and subsequent GC analysis of the solution. The objective was to quantify the amount of nitroxyl liberated by the polymer.

Data acquired from the 140 oC series of experiments showed evidence of progressive nitroxyl release over a 60 min period (Table 1), as was consistent with the DMP trials. Furthermore, increasing the exposure temperature raised nitroxyl yields substantially, as demonstrated by the temperature series data listed in Table 2. Note that 1H NMR analysis of the residual polymer product was done in an attempt to observe olefinic products of alkoxyamine disproportionation.

1H NMR spectra of the products (Appendix 1) showed no discernable increase in olefin content beyond what is originally present in the atactic-PP starting material. Indeed, the rather large amount of olefinic unsaturation in the polymer complicated attempts to detect additional vinylidene resonances in the disproportionated material.

Table 1. Amount of nitroxyl liberated from PP-g-HOTEMPO at 140 oC

Time (min) Moles of

HOTEMPO/g of PP

0 -

20 4.90E-06

40 1.12E-05

60 1.92E-05

Table 2. Amount of nitroxyl liberated from PP-g-HOTEMPO after 20 min

Temperature (oC) Mol HOTEMPO/g of PP

140 4.90E-06

160 1.04E-05

180 3.17E-05

Having demonstrated the susceptibility of 3o alkoxyamine to disproportionation, the question remains as to the impact of this instability on PP graft modification. Note that the lower limit of any melt state process is the 160 oC melting point of the polymer, but temperatures in excess of

180 oC are often applied to reduce the polymer viscosity. Over a 20 min timescale at this temperature, the 1o and 2o alkoxyamines are stable while the 3o alkoxyamine degrades relatively slowly. Garrett et al. [4] reported a 64% yield of 3o alkoxyamine by cumyloxyl acting on DMP at 160 oC. Based on this value, nearly two-thirds of the alkoxyamine functionality in PP-g-

TEMPO may be susceptible to disproportionation, with the remaining one-third being robust under melt processing conditions.

Note that polyolefin modifications, including formulations involving functional nitroxyls, are designed to be completed as quickly as possible. Peroxide initiators are selected on the basis of their half-lives at the processing temperature, with the goal of providing enough time to mix the reagents throughout the polymer melt, but also minimizing the required residence time. In reactive extrusion, residence times can be as little as one minute, while reactive compression molding cycles may be only slightly longer. As such, the disproportionation rates observed in this study suggest that instability is not a dominating influence on PP functionalization yields.

Indeed, independent rheometry studies by Molloy, Ozols and Bodley have shown that TEMPO based PP formulations provide induction periods that are well represented by the standard trapping ratio equation [21] [22]. Given that alkoxyamine disproportionation releases hydroxylamine that is rapidly oxidized to nitroxyl, a significant reaction extent would lengthen induction periods beyond predicted values. Taken together, these rheometry studies and the instability results acquired in the present work are consistent in their finding that disproportionation is undesirable, but not a yield-controlling process.

3.3 Conclusion

The thermal instability of the tertiary alkoxyamine within PP-based alkoxyamines has been confirmed by studies of model hydrocarbon and polymer systems. The results suggest that tertiary alkoxyamine disproportionation has the potential to negatively affect PP graft

26 modification yields but can be avoided by minimizing reaction temperature and times to conventional polymer processing conditions.

Chapter 4

Fluorescence Determination of Macroradical Yields in Polyolefin Melt Blends

4.1 Introduction

Solvent-free, peroxide-initiated modifications of polyolefins is used widely to alter chain architectures, [31] introduce pendant functional groups [32] and enhance polymer blend and/or adhesive properties. [33] H-atom abstraction from the polymer by peroxide-derived radicals is fundamental to these processes, since all they require alkyl macroradical intermediates to support the desired addition, scission or combination reactions.

Insight into the H-atom donor reactivity of a polymer can be gained by measurement of a peroxide’s abstraction efficiency (AE, Scheme 6), defined as the fraction of alkoxy radicals that react via H-atom transfer rather than cleave to methyl radical + ketone.[34] This property is readily accessible for dicumyl peroxide, which yields cumyl alcohol and acetophenone from H- atom abstraction and cumyloxyl fragmentation reactions, respectively.[4] These products have similar boiling points and can be quantified by gas chromatography. AE measurements on other peroxides can be complicated by differences in the structure and volatility of their byproducts, which makes it difficult to isolate them and determine their concentration.

DCP thermolysis D . Ph OO Ph 2 Ph O

Cumyloxyl scission / H-atom abstraction O . . ONR2 + CH Me-ONR . Ph 3 2 Ph O R-H . . ONR2 Ph OH + R R-ONR2

R-H = pentane, 2,4-dimethylpentane cyclohexane, dodecane, HPDE . ONR = TEMPO, NTEMPO 2 Scheme 6: Cumyloxyl scission and H-atom abstraction products of AE determinations

An alternate means of measuring AE involves alkyl radical trapping by nitroxyl and quantification of the resulting alkoxyamines. Dupont et al. [29] attempted to measure the AE of

DCP for dodecane at various temperatures by trapping radicals with TEMPO and measuring the concentrations of cumyl alcohol, acetophenone, dodecane-g-TEMPO and Me-TEMPO using gas chromatography. AE calculations based on peroxide byproducts (cumyl alcohol and acetophenone) and TEMPO products (dodecane-g-TEMPO and Me-TEMPO) were in good agreement, providing confidence in the TEMPO-based method.

Direct analysis of alkoxyamines in polyolefin derivatives is much more difficult than in model hydrocarbons such as dodecane, given that the alkoxyamine concentration generated by a typical polyolefin modification process is on the order of micromoles per gram. Since polymer-bound alkoxyamine cannot be separated from the polymers that are not chemically modified, such a low functional group content presents severe analytical chemistry challenges. While conventional absorption and resonance spectroscopy is not sensitive enough for this application, fluorescence emission spectroscopy has provided detection limits that could be sufficient. [35]

Trapping alkyl macroradicals using a nitroxyl bearing functionality such as naphthoyloxy group

(Scheme 2) yields a polymer-bound fluorophore whose concentration can be quantified by fluorescence spectroscopy. This approach has been examined by the Parent group through a MSc thesis produced by John Dupont [29], who quantified H-atom abstraction yields of DCP acting on an HDPE melt. The fluorescent alkoxyamine method produced AE values consistent with precise GC measurements of cumyl alcohol and acetophenone, providing confidence that the technique can be extended to polyolefin blends.

Scheme 7. Structure of NTEMPO

Peroxide-initiated radical chemistry is used widely to produce thermoplastic vulcanizates

(TPV’s) comprised of a continuous thermoplastic phase and dispersed phase of crosslinked rubber.[36] Polyolefin-based TPVs are commonplace, wherein the thermoplastic component is made from polyethylene or PP, and the vulcanized elastomer is derived from a rubbery terpolymer such as EPDM.[37] These materials are produced through a reactive blending process known as dynamic vulcanization, [38] in which blend components are mixed in the melt state while the peroxide is activated. The resulting TPV morphology is a function of the initial viscosity of each polymer, the relative amounts of each material, the shear stresses imposed by the mixer, and the dynamics of polymer modification. [39]

An ideal dynamic vulcanization process would crosslink the elastomer phase while leaving the thermoplastic unaffected.[40] [41] However, migration of the peroxide between the blend components results in the chemical modification of the polyolefin. In the case of an ethylene- rich thermoplastics, this serves to crosslink the matrix, whereas in the case of PP-based TPVs, curative migration produces a degraded matrix. One strategy to mitigate against these thermoplastic modifications is to premix all of the peroxide into the elastomer, then compound the resulting masterbatch with polyolefin in the dynamic vulcanization process. The success of this technique depends on whether mass transfer of the peroxide occurs appreciably over the timescale of initiator decomposition.

This chapter describes an extension of fluorescent NTEMPO chemistry to the study of PE + EPR blends. The objective was to demonstrate the basic principles of the experimental method while providing preliminary insights into peroxide transport between phases during dynamic vulcanization. It begins with a validation of earlier experimental abstraction efficiency reports before progressing to single-phase polyolefin modifications and then to two-phase mixtures that are relevant to TPV production.

4.2 Results and Discussion

4.2.1 Abstraction efficiency validations

Preliminary experiments were performed on model compounds with published AE values to validate the method. These trials were performed with dicumyl peroxide (DCP) at 160 oC under a nitrogen atmosphere. The data presented in Table 3 confirm that the GC technique developed in this work yields results that are consistent with published reports.

Table 3. AE of DCP acting upon model hydrocarbons at 160 oC

AE (Experimental) AE (Literature) [2] [36]

Pentane 0.41 0.45

DMP 0.36 0.34

Cyclohexane 0.56 0.53

Having validated the GC method for abstraction efficiency determinations, the study shifted to previous work conducted by John Dupont on the dodecane system.[29] He measured AE values with two techniques; through GC analysis of cumyl alcohol and acetophenone, as described above, and using GC analysis of dodecyl-g-TEMPO and methyl-TEMPO alkoxyamines.

Whereas Dupont’s alkoxyamine calculations matched those published by Garrett et al., his peroxide-derived calculations did not. Based on the results obtained in this study, it appears that

Dupont’s latter results were in error. Figure 3 shows that AE values for DCP acting upon dodecane fall into the range of 0.46-0.62 at reaction temperatures between 140-170 oC.

Moreover, these model hydrocarbon results are consistent with Dupont’s fluorescent alkoxyamine experiments on HDPE, providing confidence in the polymer analysis technique.

[29]

100 Dodecane: GC Analysis (this work) Dodecane: Nitroxyl Analysis 75 HDPE: Fluorescence Analysis

25 DCP Abstraction Efficiency (%)

0 130 150 170 190 o Reaction Temperature ( C)

Figure 3: H-atom abstraction efficiency measurements acquired with different analytical techniques

4.2.2 Independent HDPE and EPR Modifications

Polyolefin experiments started with studies of single polymers, as opposed to blends, in order to define the H-atom transfer reactivity of each material. The choice of HDPE as a thermoplastic component was based on Dupont’s success in applying NTEMPO trapping to this material [29].

A random copolymer of ethylene and propylene (EPR) was selected as the elastomer component based on its saturated polymer backbone and room-temperature solubility in organic solvents.

The latter property is essential to subsequent studies of polyolefin blends, since the ability to isolate each polymer by dissolving under different conditions is central to a peroxide migration study.

Quantifying the fluorescent alkoxyamine content of a polymer derivative requires calibration of the instrument with an appropriate standard. In the present work, dodecane-g-NTEMPO was deemed to be suitable based on the structural similarity of its naphthoyl ester functionality. [42]

Note that Singer et al. [43] [44] reported that nitroxyls diminish fluorescence intensity by quenching the excited states of aromatic groups. Therefore, residual nitroxyl must be removed from calibration standards and polymer modification products prior to fluorescence analyses.

The 1H NMR spectrum of dodecane-g-NTEMPO (APPENDIX 1) showed the sharp, well- defined peaks that are expected of nitroxyl-free material. [45]

Instrument calibrations were derived from dodecane-g-NTEMPO solutions using conditions to match polymer analyses. For EPR-g-NTEMPO, samples were dissolved in cyclohexane and analyzed at room temperature, whereas for HDPE-g-NTEMPO, samples were dissolved in hot xylenes and analyzed at 125 oC. Note that temperature affects fluorescence intensity and, as such, analysis temperatures must be known and controlled. [46] The calibrations illustrated in

Figure 4 illustrate the impressive sensitivity of the fluorescence analysis, with micromolar solutions producing strong peak intensities that are well in excess of the detection threshold.

Note that xylenes demonstrates fluorescence at an emission wavelength of ~290nm, while dodecane-NTEMPO has a peak around 350 nm. This led to an overlap in emission peaks in this solvent that were handled by the calibration. [47]

7.5E+06 DD-NTEMPO in Cyclohexane (20C) DD-NTEMPO in Xylenes (120C) 6.0E+06

4.5E+06 PeakArea 3.0E+06

1.5E+06

0.0E+00 0.0E+00 1.5E-05 3.0E-05 4.5E-05 Concentration (mol/L)

Figure 4. Calibrations of dodecane-g-NTEMPO in cyclohexane (295 nm excitation, 280-580 nm emission at 20 oC) and dodecane-g-NTEMPO in xylenes (295 nm excitation, 315-430 nm at 125 oC)

HDPE-g-NTEMPO and EPR-g-NTEMPO were produced using identical DCP + NTEMPO formulations at 150, 160, and 170 oC, and then purified by dissolution-precipitation to remove residual NTEMPO, Me-NTEMPO and peroxide byproducts. The fluorescence emission spectra of EPR-g-NTEMPO and its calibration standard, dodecane-g-NTEMPO, showed similar emission peak shapes, while cyclohexane contributed no signal intensity (Figure 5).

300000

250000 DD-NTEMPO

EPR-NTEMPO 200000 Cyclohexane

150000 Counts (1/s) Counts

100000

50000

0 315 365 415 465 515 565 Wavelength (nm)

Figure 5. Comparison of dodecane-g-NTEMPO and EPR-g-NTEMPO fluorescence spectra in cyclohexane (20 oC, 295 nm excitation wavelength)

Similarly, the fluorescence spectra of HDPE-g-NTEMPO and dodecane-g-NTEMPO are in good agreement (Figure 6). A narrower emission window was recorded for these samples to minimize the analysis time. The instrument cuvette can be maintained at 60 oC, while HDPE-g-NTEMPO is soluble in xylenes above 120 oC. Therefore, the analysis solution temperature was monitored with a thermocouple to ensure that spectra were acquired in the 125-129 oC range, and the analyses were performed in triplicate to minimize uncertainty in the alkoxyamine concentration data.

250000

DD-NTEMPO

200000 HDPE-NTEMPO

Xylenes

150000

Counts (1/s) Counts 100000

50000

0 315 340 365 390 415 Wavelength (nm)

Figure 6. Comparison of HDPE-g-NTEMPO and dodecane-g-NTEMPO fluorescence spectra in xylenes (125 oC, 295 nm excitation wavelength)

Figure 7 presents a plot of the alkoxyamine contents of HDPE-g-NTEMPO and EPR-g-

NTEMPO as a function of reaction temperature. Note that Garrett et al. [4] reported substantially higher AE values for DCP acting upon HPDE compared to their EPR copolymer.

This is consistent with the present work, in that alkoxyamine yields were, on average, 24% greater for HDPE than for the EPR used in this study. Moreover, the yield data shows little sensitivity to temperature, which is also consistent with prior reports.

8.0E-06

6.0E-06

4.0E-06

2.0E-06 HDPE-g-NTEMPO EPR-g-NTEMPO Bound Bound NItroxyl Concentration (mole/g)

0.0E+00 140 150 160 170 180 o Reaction Temperature ( C)

Figure 7. EPR-g-NTEMPO and HDPE-g-NTEMPO alkoxyamine yields

4.2.3 HDPE + EPR Blend Modifications

Having established confidence in the fluorescence-based alkoxyamine method, the study progressed to a study of reactive HDPE + EPR blending. The experiment involved preparing an

EPR + DCP + NTEMPO masterbatch and an HDPE + NTEMPO masterbatch. The trapping ratio of the EPR masterbatch was 1.23, while the amount of NTEMPO charged to the HDPE masterbatch was sufficient to produce a trapping ratio of 0.75 relative to the amount of DCP loaded to EPR. This formulation ensured ensure that macroradicals produced in the EPR phase and in the HPDE phase (due to DCP migration) would be quenched by available nitroxyl.

The two masterbatches were compounded in a 50:50 wt:wt ratio at 160 oC for 30 min, which corresponds to five initiator half-lives. EPR-g-NTEMPO was isolated from the reacted blend by solvent extraction with cyclohexane and precipitation from excess acetone. HDPE-g-NTEMPO was retained as a solid from the cyclohexane extraction process. A material balance showed the masses of each phase to be within 11% of expected values. Both materials were purified to remove Me-NTEMPO, residual NTEMPO and peroxide byproducts, and subjected to fluorescence analysis of their polymer-bound alkoxyamine content.

Figures 8 and 9 show fluorescence emission spectra for isolated blend components as well as their unmodified parent materials. The post-reaction alkoxyamine content of EPR-g-NTEMPO post reaction was 2.76 umol/g, while that of HDPE-g-NTEMPO was 2.50 umol/g. Given that the initiator was premixed into EPR prior to reactive blending, the large amount of polymer-bound alkoxyamine within EPR-g-NTEMPO is expected. However, the HDPE phase contained no peroxide prior to the reactive blending process, and yet the product contained a substantial amount of alkoxyamine. This could only have occurred by migration of initiator from the EPR phase to the HDPE phase during compounding, where thermolysis ultimately produced macroradicals that were quenched by NTEMPO.

1.E+05

8.E+04 Neat EPR

EPR-g-NTEMPO 6.E+04

4.E+04 Counts Counts (1/s)

2.E+04

0.E+00 320 345 370 395 420 445 470 Wavelength (nm) Figure 8. Fluorescence spectra for EPR-g-NTEMPO isolated from the compounded blend and its parent material.

8.E+04 Neat HDPE HDPE-g-NTEMPO 6.E+04

4.E+04 Counts Counts (1/s) 2.E+04

0.E+00 315 340 365 390 415 Wavelength (nm) Figure 9. Fluorescence spectra for HDPE-g-NTEMPO isolated from the compounded blend and its parent material.

Note that the alkoxyamine contents reflect the concentration of macroradicals generated in a given phase, not the amount of initiator that produced them. The overall amount of DCP

40 activated in each phase can be calculated from the alkoxyamine concentration and the abstraction efficiencies of the peroxide on each polyolefin. The pure polymer yield data plotted in Figure 7 show that HDPE is 24% more reactive toward cumyloxy radicals than EPR. Therefore, the ratio of DCP that acted in each of the blend phases was EPR:HDPE = 1.45:1. This confirms a preference for DCP activation in the initiator-masterbatched EPR phase, but also indicates a substantial extent of DCP migration to the HDPE blend component. If the TPV formulation contained no NTEMPO, one would expect a substantial increase in the HDPE phase viscosity as a result of DCP-initiated crosslinking.

4.3 Conclusions

Trapping of alkyl macroradicals with NTEMPO produces polymer-bound alkoxyamine whose concentration can be quantified by fluorescence spectroscopy. This analytical technique can be applied to single polymers to quantify their H-atom donor reactivity toward an oxygen-centred radical, which is of central importance to the chemical modification of polyolefins. The methodology can be extended to polyolefin TPV formulations, wherein macroradical yields in blend components provide insight into initiator migration during dynamic vulcanization.

Chapter 5

Future Work

5.1 Tertiary Alkoxyamine Instability

Synthesis of a tertiary alkoxyamine, devoid of primary and secondary regioisomers, should be synthesized and its disproportionation sensitivity quantified by GC analysis. Rates of thermolysis could be determined through the decrease of alkoxyamine peak, the increase of an alkene peak, as well as the increase of liberated TEMPO.

5.2 Peroxide Migration Trials

Understanding peroxide migration during dynamic vulcanization is important, especially when the continuous phase is a scission-prone polymer. A comprehensive study of migration extents as a function of reaction temperature/time and blend component properties such as viscosity and abstraction efficiency could prove useful for designing and optimizing thermoplastic vulcanizates properties.

References

[1] P. Hudec, L. Obdrzalek, “Change of Molecular Weights at Peroxide Initiated Degradation of PP,” Die Angewandte Makromolekulare Chemie, vol 89, pp. 41-45, 1980.

[2] C. Tzoganakis, Y. Tangm J. Vlachopoulos, A.E. Hamielec, “Controlled Degradation of Polypropylene: A Comprehensive Experimental and Theoretical Investigation,” Polym. Plast. Technol. Eng., vol. 28, pp. 319-350, 1989.

[3] F. Berzin, B. Vergnes, S.V. Canevarolo, A.V. Machado, J.A. Covas, “Evolution of the Peroxide Inducted Degradation of Polypropylene Along a Twin-Screw Extruder: Experimental Data and Theoretical Predictions,” Journal of Applied Polymer Science, vol. 99, pp. 2082-2090, 2006.

[4] G. Garret, E. Mueller, D. Pratt, J.S. Parent, “Reactivity of Polyolefins Toward Cumyloxy Radical: Yields and Regioselectivity of Hydrogen Atom Transfer,” Macromolecules, vol 47, no. 2, pp. 544-551. 2014.

[5] M. F. Diop, J. Torkelson, “Maleic Anhydride Functionalization of Polypropylene with Supressed Molecular Weight Reduction via Solid-State Shear Pulverization,” Polymer, vol. 54, pp. 4143-4154, 2013.

[6] M. Ratzsch, M. Arnold, E. Borsig, H. Bucka, N. Reichelt, “Radical Reactions on Polypropylene in the Solid State,” Progress in Polymer Science, vol 27 (7), pp. 1195-1282, 2002.

[7] G. Hulse, R. Kerstring, D. Warfel, “Chemistry of Dicumyl Peroxide Induced Crosslinking of Linear Polyethylene,” J. Polym. Sci. Polym. Chem., vol. 19, pp. 655-667, 1981.

[8] I. Chodak, D. Bakos, V. Mihalov, “Mechanism of the Reaction of Cumyloxyl Radicals in a Mixture of Hydrocarbons,” J. Am. Chem. Soc., vol. 10, pp. 1457-1459, 1980.

[9] D. Avila, C.E. Brown, K.U. Ingold, J. Lusztyk, “Solvent Effects on the Competitive B- Scission and Hydrogen Atom Abstraction Reactions of the Cumyloxyl Radical. Resolution of a Long-Standing Problem,” J. Am. Chem. Soc., vol. 115, pp. 466-470, 1993.

[10] K. Russell, “Free Radical Graft Polymerization and Copolymerization at Higher Temperatures,” Progress in Polymer Science, vol. 27, no 6, pp. 1007-1038, 2001

[11] E. Niki, Y. Kamiya, “Reactivities of Polystyrene and Polypropylene Toward Tert-Butoxy Radical,” Journal of Organic Chemistry, vol. 38, pp.1403-1406, 1973

[12] Ingold, K. U. In Free Radicals; Kochi, J. K., Ed.; Wiley: New York, 1973; Vol. 1, p 92

[13] E. Rizzardo, D. Solomon, “A New Method for Investigating the Mechanism of Initiation of Radical Polymerization,” Polym. Bull, vol. 1(8), pp. 529-534, 1979.

[14] J.D. Van Drumpt H.H. Oosterwijk, “Kinetics and Mechanism of the Thermal Reaction Between Tert-Butyl Perbenzoate and n-alkanes: A Model System for the Crosslinking of Polyethylene,” J. Polym. Sci: Polym. Chem. Ed, vol. 14(6), pp. 1495-1511. 1976.

[15] P.D. Rowe, D.K. Thomas, “The Thermal Decomposition of Dicumyl Peroxide in Polyethylene Glycol and Polypropylene Glycol,” Journal of Applied Polymer Science, vol. 7(2), pp. 461-468, 1963.

[16] D. K. Hyslop and J. S. Parent, "Dynamics and Yields of AOTEMPO-mediated Polyolefin Cross-linking," Polymer, vol. 54, pp. 84-89, 2013.

[17] J. Chateauneuf, J. Lusztyk, K.U. Ingold, “Absolute Rate Constants for the Reaction of Some Carbon-Centred Radicals with 2,2,6,6-tetramethyl-1-piperidinoxyl,” J Org Chem, vol.53, pp. 1629-1632, 1988.

[18] D. K. Hyslop and J. S. Parent, "Functionalized Nitroxyls for Use in Delayed-Onset Polyolefin Cross-Linking," Macromolecules, vol. 45, no. 20, pp. 8147-8154, 2012.

[19] A. Beckwith, V. Bowry, M. O’Leary, G. Moad, E. Rizzardo, D.H. Solomon, “Kinetic Data for Coupling of Primary Alkyl Radicals with a Stable Nitroxide,” Journal of the Chemical Society, Chemical Communications,” vol 13, pp. 1003-1004, 1986.

[20] K.E. Ozols, “Polypropylene Thermosets by Functional Nitroxyl-Mediated Radical Crosslinking” Masters of Applied Science, Queen’s University, 2017.

[21] K.E. Ozols, B.M. Molloy, J.S. Parent, “Polypropylene Thermosets by VBTEMPO-Mediated Radical Crosslinking,” Polymer, vol. 123, pp. 211-218, 2017.

[22] M.W. Bodley, J.S. Parent, “AOTEMPO-Mediated Dynamic Vulcanization: Synthesis of Impact Modified Polypropylene,” Polymer Engineering and Science, vol. 58, pp. 1999-2007, 2017.

[23] K. Ohno, Y. Tsujii and T. Fukuda, "Mechanism and Kinetics of Nitroxide-Controlled Free Radical Polymerization. Thermal Decomposition of 2.2.6.6-Tetramethyl-1- polystyroxypiperidines," Macromolecules, vol. 30, pp. 2503-2506, 1997.

[24] P. Engel, S. Duan and G. Arhancet, “Thermolysis of a Tertiary Alkoxyamine. Recombination and Disproportionation of a a-Phenethyl/Diethyl Nitoxyl Radical Pairs”, J. Org. Chem, vol 62, pp. 3537-3541, 1997.

[25] M.E. Scott, “Polymer Modification by Nitroxyl-Mediated Radical Chemistry” Masters of Applied Science Dissertation, Queen’s University, 2002.

[26] W.B. Qin, Q. Chang, Y.H. Bao, N. Wang, Z.W. Chen, L.X. Liu, “Metal-Free Catalyzed Oxidative Trimerization of Indoles by Using TEMPO in Air: A Biomimetic Approach to 2-1H- indol-3-yl)-2,3’-biindolin3-ones,” Organic and Biomolecular Chemistry, vol. 10, pp. 8814-8821, 2012.

[27] G. Ananchenko, H. Fischer, “Decomposition of Model Alkoxyamines in Simple and Polymerizing Systems. I. 2,2,6,6,-tetramethylpiperidinyl-N-oxyl-based Compounds,” Journal of Polymer Science: Polymer Chemistry, vol 39(20), pp. 3604-3621, 2001.

[28] E. Niki, Y. Kamiya, “Reactivities of Polystyrene and Polypropylene Toward tert-butoxy Radical”, J. Org. Chem, vol 38 (7), pp. 1403-1406, 1973.

[29] J.A.C. Dupont, “Initiation Efficiency in Radical-Mediated Polymer Modification,” Masters of Applied Science Dissertation, Queen’s University, 2003.

[30] M. Jones, G. Moad, E. Rizzardo, D.H. Soloman, “The Philicity of tert-Butoxy Radicals. What Factors Are Important in Determining the Rate and Regiosceficity of tert-Butoxy Radical Addition to Olefins?” Journal of Organic Chemistry, vol. 54, pp 1607-1611. 1989.

[31] J.A. Manson, L.H. Spearing, “Polymer Blends and Composites” Plenum, New York. 1976.

[32] C. Tzoganakis, J. Vlachopoulos, A.E. Hamielec, “Effect of Molecular Weight Distribution on the Rheological and Mechanical Properties of Polypropylene,” Polymer Engineering, Vol 29, no. 6, pp 390-396, 1989.

[33] M.A. Chem, X.M Zhang, R. Huang, X.B Lu, “Mechanism of Adhesion Promotion Between Aluminium Sheet and Polypropylene with Maleic Anhydride-grated Polypropylene by Aminopropyltriethoxy Silane,” Surface and Interface Analysis, Vol. 40, no. 8, pp 1209-1218, 2008.

[34] I. Chodak, D. Bakos, “Collection of Czechoslovak Chemical Community” vol. 43, pp 2574- 2577, 1976. [35] X.F. Yang, X.Q. Guo, “Fe(II)-EDTA Chelate-Induced Aromatic Hydroxylation of Terephthalate as a New Method For The Evaluation of Hydroxyl Radical-Scavenging Ability,” Analyst, vol. 126, pp 1800-1804, 2001.

[36] A. Thitithammawong, C. Nakason, K. Sahakaro, J.W.M. Noordermeer, “NR/PP Thermoplastic Vulcanizates: Selection of Optimal Peroxide Type and Concentration in Relation to Mixing Conditions,” Journal of Applied Polymer Science, vol. 106, pp. 2204-2209, 2007.

[37] S.K. Henning, R. Costin, “Fundamentals of Curing Elastomers with Peroxides and Coagents” Rubber World, vol. 233, pp. 28-35, 2006.

[38] S. Abdou-Sabet, R.C. Puydak, C.P. Rader, “Dynamically Vulcanized Thermoplastic Elastomers,” Rubber Chemistry and Technology, vol. 69, pp. 476-494, 1996.

[39] D. Pizele, V. Kalkis, R. M. Meri, T. Ivanova, and J. Zicans, “On the Mechanical and Thermomechanical Properties of Low-Density Polyethylene/ethylene-α-octene Copolymer Blends,” Mechanics of Composite Materials, vol. 44, pp. 191–196, 2008.

[40] K. Naskar, “Dynamically vulcanized PP/EPDM thermoplastic elastomers. Exploring novel routes for crosslinking with peroxides,” University of Twente, The Netherlands, 2004.

[41] R. R. Babu, N. K. Singha, and K. Naskar, “Interrelationships of morphology, thermal and mechanical properties in uncrosslinked and dynamically crosslinked PP/EOC and PP/EPDM blends,” Express Polymer Letters, vol. 4, no. 4, pp. 197–209, 2010.

[42] N.V. Blough, D.J. Simpson, “Chemically Mediated Fluorescence Yield Switching in Nitroxide-Fluorophore Adducts: Optical Sensors of Radical/Redox Reactions,” Journal of American Chemical Society, vol. 110, pp. 1915-1917, 1988.

[43] J.A. Green, L.A. Singer, “Di-Tert Butyl Nitroxide as a Convenient Probe for Excited Singlet States,” Journal of American Chemical Society, vol. 96, pp 2730-2733, 1974.

[44] C.L. Kwan, S. Atik, L.A. Singer, “An Electron Spin Resonance Study of the Association of a Surfactant Nitroxyl Radical with a Cationic Micelle Using Spin-Intensity Measurements and Hyperfine Structure Analyses,” Journal of American Chemical Society, vol. 100, pp. 4783-4786, 1978.

[45] C.P. Slichter, “Principles of Magnetic Resonance, 1st edition,” Springer, Berlin, 1989.

[46] S.M. Hanagodimath, B. Siddlingeshwar, J. Thipperudrappa, “Fluorescence-Quenching Studies and Temperature Dependence of Fluorescence Quantum Yield, Decay Time and Intersystem Crossing Activation Energy of TPB,” Journal of Luminescence, vol. 129, pp. 335- 339, 2009.

[47] M.F.S. Khan, J. Wu, B. Liu, C. Cheng, M. Akbar, Y. Chai, A. Memon, “Fluorescence and Photophysical Properties of Xylene Isomers in Water: With Experimental and Theoretical Approaches,” Royal Society Open Science, vol. 5, pp. 171719-171719, 2018.

Appendix

Appendix A 1H NMR spectrum of dodecane-g-NTEMPO

Appendix B 1H NMR of PP-g-TEMPO