Organocatalyzed Ring Opening Polymerization and the Design of Materials for Topological Trapping

ORGANOCATALYZED RING OPENING POLYMERIZATION AND THE DESIGN OF MATERIALS FOR TOPOLOGICAL TRAPPING

A DISSERTATION SUBMITTED TO THE DEPARTMENT OF CHEMISTRY AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

TYLER STUKENBROEKER JANUARY 2016

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/jd110qd1669

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Robert Waymouth, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Eric Kool

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Paul Wender

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost for Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii

Abstract

Ring-opening polymerization (ROP) is a highly useful route to well-defined, high molecular weight polymers. In recent years, strongly nucleophilic neutral bases such as

N-heterocyclic carbenes (NHCs) have been shown to catalyze ring opening polymerization with control over the topology of the polymer product. This mechanism has been termed Zwitterionic Ring Opening Polymerization (ZROP). Here, new catalysts and monomers have been explored for these processes.

Two types of organocatalysts were developed. A recently reported superbase, cyclopropenimine, was tested for activity with a number of monomers. It was discovered that this catalyst polymerizes lactide not through ZROP, but through anionic ROP initiated by a previously unreported mechanism. It was demonstrated that cyclopropenimines deprotonate lactide to form the corresponding enolate, which serves as the initiator and endgroup for the polymerization. Additionally, an electrophilic co- catalyst for lactide polymerization based on a cationic bis(imidazolium) structure was discovered.

Furthermore, a class of cyclic phosphotriesters was synthesized as a substrate for NHC-catalyzed polymerization. Mechanistic studies were consistent with the existing interpretation of ZROP. The topology of the polymer products was found to be highly impacted by slight changes in monomer structure. One of these monomers, 2-isoproxy-

1,3,2-dioxophospholane 2-oxide (iPP), yielded high molecular weight cyclic products. To further interrogate the polymer architecture, a crosslinked polyacrylate network was formed in-situ with the poly(iPP). In this way cyclic macromolecules were topologically trapped, preventing their extraction by solvent washing. The gel platform used for this trapping protocol was optimized and this process was explored as a way to sort polymers based on their topology. Finally, the properties of these gels, a unique form of interpenetrating network, were studied.

Acknowledgements

My time at Stanford coincided with an unprecedented period of Stanford football dominance (three Rose Bowls), wonderfully sunny weather (albeit due to an ecologically devastating drought), and three faculty members receiving the Nobel Prize in Chemistry.

Despite these auspicious concurrences, I am most thankful to have overlapped on the

Farm with many smart, kind, and fun individuals whom I am eager to acknowledge.

The research presented here was funded by multiple agencies. I was personally supported by a National Defense Science and Engineering Graduate Fellowship on behalf of the Army Research Office. The operational budget was provided by the

National Science Foundation via two grants, DMR-1001903 and DMR-1407658. Work done at the Molecular Foundry (LBNL) was supported by the US Department of Energy.

I was particularly fortunate to find Bob Waymouth’s lab when I arrived at

Stanford. Over the past five years Bob has allowed me the freedom to learn about organocatalysis and polymer physics and develop a project that incorporates both fields.

I am lucky to have had him as a scientific role model. If children eventually turn into their parents then graduate students turn into their advisors; recently I have caught myself dropping “Bob-isms” with alarming frequency.

I am grateful to my reading committee, Paul Wender and Eric Kool, for taking the time to read and critique various proposals and ideas. Undoubtedly their wisdom will prove invaluable as I move forward in my career. Other professors I have had the pleasure of working with include Yan Xia and Do Yoon who offered their insight regarding my projects. Tristan Lambert and especially his former student Jeff Bandar were incredibly patient with me on our collaborative project. Redouane Borsali was a cordial host when I visited France.

Many other people inside and outside of the chemistry department enabled the experiments and findings presented in this thesis. Steve Lynch, Theresa McLaughlin,

vii

Theresa Pick, and Jeff Tok all helped me to make important measurements. Without the assistance Roger Kuhn, Marie Herbert, Orlando Martinez, and most of all Dewi

Fernandez there is no way I could have kept all my requirements in order.

I will miss the students and post-docs in the Waymouth Lab. Naomi, Eric and

Jelena have been terrific desk-neighbors. Besides stimulating chemistry discussions, I have enjoyed having lunch, grabbing a beer, and going skiing with them. Kevin, Kate,

Tim and I share the bond of spending all five years together in lab and I couldn’t think of a better set of classmates to be paired with. Young is a wonderfully witty travel buddy.

James is my favorite Irishman and I will miss his astute observations of life in lab.

Xiangyi and Colin offered gracious help with polymerization questions and experiments on many occasions. Many postdocs deserve my gratitude, most recently Andrey, Chris,

Gregg, and Antonio. With Wilson, Weiwei, Liz, Katherine, Ben, and Rebecca around I am sure the lab will remain both entertaining and innovative. In other labs, Ben Elling,

Jingxian Li, and Tadanori Kurosawa have shared the agony and the ecstasy of maintaining GPCs with me.

Special acknowledgement is necessary for my friend and mentor Hayley Brown.

It is no exaggeration to say that everything I know about air-free technique, glove-box polymerizations, solvent drying, cyclic polymers, reading and writing manuscripts, giving group meetings, applying for fellowships, and much, much more came from Hayley. I am indebted to her for walking me through countless procedures and taking me under her wing when experiments weren’t working. Most of all, I enjoyed getting to know her as a friend both in lab, on camping trips, and during several drives up to Berkeley.

Stanford has offered me tremendous opportunities to teach and learn about science outside of my research. I have enjoyed tutoring athletes through the AARC program and visiting high schools with the chemistry outreach program. I had a blast

viii making the Goggles Optional show with a group of zany people who tolerated my brief stint as a podcasting personality.

Without my friends, I am sure I would have a very different experience in graduate school. There are far too many to list here, but I especially want to shout out the Reno Boys (Honk! Honk!) and the movie night crew (and the Moooovie goes to…) for all the fun memories. The brilliant and lovely Tali has probably spent the most time listening to me babble on about lab, so I thank her for that and much more. My good friends, near and far, have all been wonderfully supportive during my time here.

Finally, I would be nowhere without my family. My brother Wesley visited me in

California several times. I am constantly blown away hearing about his accomplishments. My parents, George and Susan, always made education a priority.

This included reading my essays, helping with my projects, checking my homework and teaching me to be a hard-working student. They made sure I went to the best high school, the college of my choosing, and don’t complain that my chemistry career seems to be taking me to progressively more distant locations. Thank you!

Table of Contents

Chapter 1: Topological Materials

1.1 Introduction 3

1.2 Topological Trapping 4

1.3 Slip-Link Gel 7

1.4 Summary 10

1.5 References 11

Chapter 2: Organocatalysts for Polymerization of Lactide

2.1 Introduction 15

2.2 Results and Discussion Bis-Imidazolium Salts as Electrophilic Co-Catalysts 2.2.1 17 for Lactide Polymerization Cyclopropenimines as Superbase Catalysts for Ring 2.2.2 21 Opening Polymerization 2.3 Experimental Procedures 35

2.4 References 44

Chapter 3: Polymerization of Phosphotriester Monomers

3.1 Introduction 49

3.2 Results and Discussion

3.2.1 Benzyl Phospholane Monomer 51

3.2.2 Isopropoxy Phospholane Monomer 60

3.2.3 Mechanistic Studies 63

3.3 Experimental Procedures 69

3.4 References 85

Chapter 4: Stimuli-Responsive Organogels

4.1 Introduction 91 4.2 Results and Discussion 4.2.1 Acid-Induced Transition from Organogel to Hydrogel 93 4.2.2 Reduction-Induced Degradation of Organogel 95 4.2.3 Organogels Synthesized from 1,2-Dithiolanes 98 4.3 Experimental Procedures 100 4.4 References 104

Chapter 5: Topological Trapping of Cyclic Polymers

5.1 Introduction 109

5.2 Results and Discussion

5.2.1 Poly(iPP) Trapping 110

5.2.1a Selective Polymer Retention Not a Molecular Weight Effect 117

5.2.2 Catch-and-Release Gels 120

5.2.3 Physical Properties of Topological Gels 126

5.2.4 Poly(valerolactone) Trapping 131

5.3 Experimental Procedures 136

5.4 References 142

xii

List of Figures

Fig. Description Page

1.1 Fraction of cyclic PDMS trapped by polyphenylene network 4

1.2 Three architectures for a slip-link gel 7

1.3 Macromonomer utilized by Tezuka with cyclic poly(THF) 8

1.4 Reversible bis(rotaxane) crosslinks designed by Takata 9

Evolution of molecular weight and polydispersity with conversion 2.1 18 of polylactide 1H NMR (d -DCM) of bis(imidazolium) 2•Br alone 2.2 2 19 and with 1 eq. of iPP

2.3 GPC of alcohol-initiated lactide polymerization 22

2.4 MALDI-MS of PLA synthesized with 4 and pyrenebutanol 23

2.5 Comparison of polymerization rates with and without and alcohol 25

2.6 Mark-Houwink plot of sample synthesized with cyclopropenimine 26

1H NMR (CDCl ) of polylactide obtained from cyclopropenimine 2.7 3 27 initiator after acylation 1 H NMR (CDCl3) of lactide and cyclopropenimine 4 (1:2 ratio) in 2.8 29 d6-benzene 1H NMR (CDCl ) of purified and unpurified polylactide obtained 2.9 3 30 from L-lactide with cyclopropenimine 4 13C NMR (CDCl ) of a polymerization initiated by methyl 2.10 3 31 isobutyrate enolate

1 2.11 H NMR (CDCl3) spectra of Table 2.3, entry 1 32

2.12 MALDI-MS analysis of PLA synthesized with BEMP 33

2-D NMR experiments (NOESY and HMQC) to identify proton 2.13 37 resonances of 2•Br

2.14 ESI-MS of PLA synthesized without alcohol initiator 40 MALDI-TOF of a polymerization initiated by methyl 2.15 42 isobutyrate enolate

1 3.1 H NMR (D2O) of poly(alkylene phosphate) 52

xiii

31 3.2 P NMR spectra of poly(BP) produced with Al(OMe)Et2 and NHC 54 Possible structures for the 31P NMR signals present in the 31P 3.3 55 NMR of poly(BP) Assignment of three 31P NMR signals of 3.4 55 poly(methoxyphospholane) by Penczek et al. 2D NMR 1H,31P HMBC spectrum of purified poly(BP) produced 3.5 56 with Al(OMe)Et2 Plot of the fraction of the central peak in the 31P NMR vs. the 3.6 58 Intrinsic Viscosity

3.7 MALDI-TOF MS spectrum of poly(iPP) synthesized with ZROP 62

X-ray crystal structure of 2-chloroethyl-(1,3-diisopropyl-4,5- 3.8 65 dimethyl-imidazolium)-phosphonate (4) 31P NMR (d -MeCN) of crude iPP ZROP polymerization at high 3.9 3 68 carbene concentration

3.10 MALDI-TOF spectra of poly(BP) sample synthesized with NHC 1 72

3.11 ESI-MS for imidazolium phosphonate 4 74

1 31 3.12 H and P NMR (CDCl3) for cyclic poly(iPP) 77 Representative GPC chromatogram for ZROP polymerization of 3.13 78 iPP with carbene 2

1 31 3.14 H and P NMR (CDCl3) for linear poly(iPP) 79 Representative GPC chromatogram for linear polymerization of 3.15 80 iPP with TU/DBU Heat flow from Differential Scanning Calorimetry of poly(iPP) 3.16 81 generated from carbene 2

4.1 FT-IR of HEMA-THP gel before and after deprotection 94

4.2 HEMA/BA gel made with DSDMA crosslinker 95

Gel made from HEMA and PVL-derived disulfide crosslinker with 4.3 97 PVL substrate swollen in DCM

4.4 Oxidation of disulfide network 99

4.5 Reduction of the “permanent” gel 99

A conceptual rendering of the entrapped cyclic polymers 5.1 109 in a polyacrylate network with PEG crosslinks

5.2 Image of poly(iPP)-infused gels 110

xiv

Weight % of cyclic poly(iPP), linear poly(iPP) or linear PDMAA 5.3 111 recovered from HEMA hydrogels Methyl methacrylate radically polymerized in the presence of 5.4 112 poly(iPP)

5.5 FT-IR of HEMA hydrogel 113

5.6 FT-IR of hydrogels subtracted spectra from Fig. 5.5 114

5.7 GPC traces of extracted cyclic poly(iPP) 116

5.8 “Catch and Release” gel strategy 120

GPC (RI detector) chromatograms of linear poly(iPP) before 5.9 121 (blue) and after (red) treatment with the gel dissolution conditions

5.10 Catch and release gels containing poly(iPP) 122

Solubility of network material formed by in-situ polymerization of 5.11 123 HEMA/BA with poly(iPP) GPC trace (RI detector) of material formed by in-situ 5.12 124 polymerization of HEMA/BA with linear poly(iPP) DSC scan of HEMA/BA gel with PEG crosslinker and cyclic 5.13 126 poly(iPP) substrate

5.14 Pictures of a gel used for rheological study 128

5.15 Oscillatory stress strain curves of gels 129

5.16 Frequency sweep experiment 130

5.17 Stress relaxation experiment 130

GPC (RI detector) trace of cyclic PVL substrate and the 5.18 132 extracted fraction from a HEMA:BA crosslinked gel

5.19 Picture of PVL gel dry and after extraction 134

5.20 DSC Scans of the PVL substrate and a PVL-embedded gel 135

Example of poly(iPP) quantification with NBu4PF6 internal 5.21 standard using 31P NMR. 139

1H NMR (CDCl ) of crude extracted material 5.22 3 140 containing poly(iPP), unreacted HEMA and DMF GPC chromatogram of extracted fraction from gel with PVL 5.23 140 substrate and PVL x-linker, overlayed with GPC traces of each

List of Tables

Table Description Page

Various experiments using the bis(imidazolium)/sparteine system 2.1 18 with lactide

2.2 ROP with cyclopropenimine 4 and 1-pyrenebutanol 22

2.3 Initiator-free ROP of lactide with cyclopropenimines 24

Representative polymerization of Benzyl Phospholane (BP) 3.1 52 monomer with NHC and organometallic catalysts Representative polymerizations of isopropoxy phospholane (iPP) 3.2 61 monomer

3.3 Polymerizations of iPP 83-84

4.1 HEMA gel deprotection results under TFA/MeCN conditions 94

Quantities of poly(iPP) extracted in gels containing a 50:50 (by 5.1 117 mass) mixture of linear and cyclic poly(iPP) Cumulative extracted fraction of poly(iPP) (%) from poly(HEMA) 5.2 119 gel with 16 mol % PEG crosslinker (Mn=3400) Fraction of cyclic poly(iPP) substrate recovered from generation II 5.3 123 gels by DCM extraction (24 h) and following gel degradation Samples of HEMA/BA gels with poly(iPP) substrates used for 5.4 128 rheological studies Extracted fraction of PVL from HEMA/BA gels with DSDMA 5.5 132 crosslinker Extracted fraction of PVL from HEMA/BA gels with α,ω- 5.6 133 bis(acrylate) poly(valerolactone) crosslinker

xvi

List of Schemes

Scheme Description Page

2.1 Synthesis of bis(imidazolium) salts 17 Lactide polymerization catalyzed by (-)-sparteine and 2.2 17 bis(imidazolium) 2•Br 2.3 Synthesis of 2-methyl bis(imidazolium) bromide 20 Cyclopropenimine catalysts (4-6) synthesized and monomers 2.4 21 screened 2.5 Proposed Enolate Initiation Mechanism 28

2.6 Synthesis of trimethylsilyl lactide 28

2.7 Lactide polymerization initiated by methyl isobutyrate enolate 31

2.8 Two polymerization mechanisms for lactide with organic catalysts 34

2.9 Lactide polymerization by cyclopropenimines 40 Zwitterionic Ring Opening Polymerization (ZROP) mechanism 3.1 49 shown for δ-Valerolactone with NHC catalyst Polymerization of benzyl phospholane with TBD and an alcohol 3.2 51 Initiator Comparison of the propagation pathways of the phosphorus- 3.3 53 based monomers versus lactones 3.4 Transesterification catalyzed by NHC 1 59 Generation of imidazolium chloroethoxy phosphonate and 3.5 64 proposed intermediate 3.6 Proposed mechanism for the ZROP of iPP 66 Hypothetical “Arbuzov Cyclization” pathway and standard ZROP 3.7 67 pathway 3.8 Polymerization of iPP with high concentration of carbene 67 Formation of dynamic hydrogel based on disulfide exchange of 4.1 92 asparagusic acid units Synthesis of α,ω-bis(acrylate) poly(valerolactone) with internal 4.2 96 disulfide 4.3 Triblock polymers tested for gelation 98

xvii

Abbreviations

AIBN Azobisisobutyronitrile BA Butyl Acrylate BP Benzyl Phospholane Da/kDa Dalton/kilodalton DBU 1,8-Diazabicycloundec-7-ene DCM Dichloromethane DMF Dimethylformamide DSC Differential Scanning Calorimetry DSDMA Bis(2-Methacryloxyethyl) Disulfide DTT Dithiothreitol ESI-MS Electron Spray Ionization Mass Spectrometry FT-IR Fourier Transform Infra-Red FW Formula Weight GPC Gel Permeation Chromatography HEMA 2-Hydroxyethyl Methacrylate iPP Isopropoxy Phospholane IV Intrinsic Viscosity LCST Lower Critical Solution Temperatures LS Light Scattering MALDI (or MALDI-TOF) Matrix Assisted Laser Desorption/Ionization (Time-of-Flight) MW Molecular Weight NHC N-Heterocyclic Carbene NMR Nuclear Magnetic Resonance PEG Poly(ethylene glycol) PDI Polydispersity PDMAA Poly(N,N-dimethylacrylamide) PLA Poly(lactic acid), also Polylactide PS Polystyrene PVL Poly(δ-valerolactone) RT Room Temperature TBD Triazabicyclodecene TEA Triethylamine TFA Triflouroacetic Acid THF Tetrahydrofuran THP Tetrahydropyran TU Thiourea ZROP Zwitterionic Ring Opening Polymerization

xviii

CHAPTER 1

TOPOLOGICAL MATERIALS

1.1 Introduction

Macromolecules derive many chemical properties, such as hydrogen bonding ability or acidity, from their constituent monomers. Other properties are strongly dependent on their molecular weight. Examples could include the modulus or viscosity of a material. However, two polymers having the same chemical composition and molecular weights can still be topologically distinct. For a single polymer chain, the simplest topological isomer is a cyclic. The topology of the polymer imparts its own properties, separate from the chemical monomer composition or the chain molecular weight. Cyclic polymers, for instance, are more compact than their linear counterparts, leading to lower viscosity, faster crystallization, and a variety of other distinct properties.1-5

Polymeric materials also derive their properties from these three parameters: intrinsic monomer chemical properties, molecular weight, and topology. A wide array of soft materials have been developed which exploit the first two aspects. For instance, functional polymers have been incorporated into gel networks to allow for biological activities such as drug delivery.6 On the other hand, the mechanical and rheological properties of the material are directly related to the molecular weight of the macromolecules from which they are formed. Comparatively little attention, however, has been directed toward utilizing topology to engineer materials. By using non-linear polymer components, networks and gels can be created which display novel physical properties. This chapter will introduce approaches to generating soft materials from one type of polymer topology—cyclics—and the resulting properties of these networks.

1.2 Topological Trapping

Semlyen first explored trapping cyclic polymers into a network material.7,8 Cycles formed by the ring-chain equilibrium of poly(dimethylsiloxane) (PDMS) were isolated via preparatory gel permeation chromatography to obtain PDMS cyclics of Mn=6840,

PDI=1.05. Separately, a linear poly(2,6-dimethyl-1,4-phenylene oxide) was functionalized with bromomethyl sidechains to allow for random crosslinking via bromine displacement by the phenol. A mixture of the two polymers, 31.7% PDMS by mass, was dissolved in toluene/chloroform and subsequently crosslinked to trap any threaded cyclic polymers. Samples were then washed for 24 h in chloroform to extract the non-trapped

PDMS cycles. It was found that the fraction of trapped polymers was strongly dependent upon the size of the PDMS cycle (Fig. 1.1).

Figure 1.1 Fraction of cyclic PDMS trapped by the polyphenylene network based on the degree of polymerization of the PDMS. The squares and asterik are experimental values. The dotted line represents the calculation from eq. 1. Figure reprinted with permission from Huang, W.; Frisch, H. L.; Hua, Y.; Semlyen, J. A. Journal of Polymer Science Part A: Polymer Chemistry 1990, 28, 1807. Copyright 1990 John Wiley and Sons.

The entrapped material contained two glass transitions, both of which were between that of the PDMS (-125 °C) and the crosslinked network (206 °C). Mechanical properties of the bulk networks with entrapped cyclics showed inconsistent trends. For example, the greatest change in elongation at break (%) occurred for small cycles

(DP=33), despite the fact that this sample had the lowest of cycles fraction retained.

Electron microscopy appeared to show microphase domains of 10-50 µm.

Semlyen and coworkers continued this line of research further with cyclic polyesters as the trapping substrate. This investigation was enabled by a polymer- supported synthetic method which allowed for the synthesis of cyclic polyesters of 176 and 238 skeletal bonds, comparable to the PDMS samples in earlier studies.9 These were trapped via a similar crosslinking and extraction protocol, and the observed trapped fractions were 27% and 50%, respectively, for the two molecular weights.10

Many efforts were made to model the trapping efficacy of these systems. The dashed line in Fig. 1.1 is based on a crude calculation proposed by Frisch and

Wasserman.11 They assumed a system consisting of two types of polymers, linears and cyclics. Based on spatial parameters, the probability that a cyclization event of the linear polymers forms a catenane with the existing cyclics, hypothetically trapping them, can then be estimated. The calculation assumes that all polymers are random, flexible coils

(Gaussian chains) and thus occupy a spherical volume with a radius that is a function of the number and length of constituent molecular segments. If the spheres of two chains overlap, then the probability of cyclization is equal to the fraction of linear polymer coil distributions which thread the cycle, a value they called β and estimated to be 0.5. After incorporating a function for overlap of the two types of random coils (spheres) and a correction factor for excluded volume, the final formula for trapping fraction (F) is:

퐹 = 〈푁12〉/훼푁2 where

훼훽푁 푁 4휋 1 푛 1/2 1 푛 1/2 3 1 〈푁 〉 = ( 1 2) { [ ( 1) 푏 + ( 2) 푏 ] − (푣 푛 + 푣 푛 )} 12 푉 3 2 2 1 2 2 2 2 1 1 2 2

1≡linear, 2≡cyclic; α= “extend of cyclization,” assumed to be 1 if the cyclic is

large enough to permit threading; N1, N2=number of linear and cyclic polymers,

respectively; n1, n2=number of statistical segments in the respective polymer

chains; b1, b2=length of the respective segments; V=volume

Further predictions of trapping fraction were conducted by Monte Carlo methods12 and later molecular modeling13 by Marks and DeBolt, respectively. These mechanisms utilized molecular parameters and randomly generated chains to create a static picture of the cyclic/network system and identify where cyclics had been threaded and trapped. Ultimately these models proved to be quite successful in matching the observed results of the PDMS trapping experiment. However, in all cases the free energy of mixing was neglected, a reasonable assumption for low-molecular weight cyclics.

Higher molecular weight systems have scarcely been explored for topological trapping. This is presumably due to the limited access to high molecular weight cyclic samples for experimental investigation. Additionally, computational study is considerably challenging if threading cross-section of the cyclic polymer must be calculated for a complex random coil (i.e. a numerical calculation for the 0.5 parameter used to generate eq. 1).

1.3 Slip-link Gel

A related type of system to entrapped cyclic polymers is the slip-link gel.14 Many of these systems were based on cyclic crown ethers, a common component of

“rotaxane” complexes.15 Several imaginative designs of these materials have been reported (Figure 1.2). Gibson copolymerized 32-crown-10 containing two pendant alcohol groups with a bis-isocyanate to create a polyurethane.16 The hydroxyl moieties

(on a separate crown ether or on an additional diol monomer) coordinated with crown ethers such that reaction with the isocyanate occurred “through” the crown ether. The slip-link topology was confirmed through 2D NMR spectroscopy. The materials themselves were amorphous and transparent, and contained a single glass transition.

This indicates that the materials were homogenous and the crosslinks were mobile to allow for complete stress relaxation.

Figure 1.2 Three different architectures for a slip-link gel. Reprinted from Kubo, M. In Topological Polymer Chemistry; WORLD SCIENTIFIC: 2013, p 199.

A slip-link gel based on pendant cyclics was synthesized by Zada.17,18 The cyclic monomer consisted of a cyclic PEG (DP=8) which contained an alkene bond in the ring.

Homopolymerization created a soluble polymer, as the macromonomer was too large to thread itself. However, copolymerization with styrene or MMA yielded an insoluble, swellable gel. Further investigations with this and related systems found that the

network material exhibited a single Tg, indicating a copolymer, yet it was higher than the value predicted by the Fox equation, indicative of restricted chain motion.

A similar system which utilized a much larger cyclic was developed by Tezuka.19

Here the cyclic polymer was ~300 atoms, an order of magnitude larger than the crown ether systems. Cyclic poly(THF) Mn=5200 Da was functionalized with a single pendant vinyl group. This was then copolymerized with methyl methacrylate to form an insoluble organogel product (Figure 1.3). Interestingly, when MMA was polymerized in the presence of non-functionalized cyclic poly(THF), no gelation occurred, indicating that the tethered PMMA chain formed on the poly(THF) was required for homogenizing the polymer solution and allowing threading.

Figure 1.3 Macromonomer utilized by Tezuka with cyclic poly(THF). Reprinted with permission from Oike, H.; Mouri, T.; Tezuka, Y. Macromolecules 2001, 34, 6229. Copyright 2001 American Chemical Society.

These examples utilized free radical polymerization to synthesize the gels. Kubo developed a slip-link gel based on the ring opening of polydimethyl siloxane.20 Two macromonomers based on D3 siloxane monomer were synthesized. The first contained a pendant cyclic polystyrene (Mn=2500). Anionic copolymerization with D4 monomer required the presence of a polar solvent for formation of the slip-link gel due to the incompatibility between PDMS and PS, despite the covalent linkages present in the macromonomer. By switching to a macromonomer D3 with a pendant cyclic PDMS

(Mn=2800), copolymerization in the bulk resulted in successful gelation. Consistent with

the Tezuka result, polymerization of D4 in the presence of non-functionalized cyclic

PDMS did not result in gel formation.

A class of topological materials based on bis(rotaxane) crosslinkers was developed by Takata.21 To a poly(crown ether) consisting of 24-crown-8 units, a bis(ammonium) molecule with internal disulfide was added. Addition of benzenethiol allowed the disulfide to be cleaved reversibly. In the cleaved form, the ammonium molecules were able to complex with crown ethers. When disulfides reformed, the polymer chains were crosslinked. The material formed an organogel in DMF.

Furthermore, addition of excess benzenethiol reduced the crosslinker and the original components were recovered.

1.4 Summary

These discoveries and many others22-24 represent a fascinating new field of soft materials that utilize macromolecule topology to realize new properties. However, limitations based on cyclic size, purity, or quantity have constrained the scope of these investigations. Furthermore, dynamic materials, which respond to external triggers, are limited to only a few examples. The following thesis expands our knowledge of the field in these and related areas. Chapter 2 discusses work towards new catalysts for cyclic polylactide, which, though ultimately unsuccessful, uncovered a novel initiation mechanism for organocatalytic lactide polymerization. Chapter 3 reviews efforts to synthesize novel cyclic phosphoester polymers of exceptionally high molecular weight.

Chapter 4 reports on the invention of various responsive organogel platforms. Finally,

Chapter 5 details trapping cyclic polymers and the properties of the resulting networks.

1.5 References

1. Beckham, H. W. In Complex Macromolecular Architectures; John Wiley & Sons

(Asia) Pte Ltd: 2011, p 791.

2. Burchard, W. In Cyclic Polymers; Semlyen, J. A., Ed.; Springer Netherlands: 1986, p

43.

3. Jia, Z. F.; Monteiro, M. J. J. Polym. Sci., Part A: Polym. Chem. 2012, 50, 2085.

4. Roovers, J. In Cyclic Polymers; Second ed.; Semlyen, J. A., Ed.; Kluwer Academic

Publishers: Dordrecht, 2000, p 347.

5. Cyclic Polymers; 2nd ed.; Semlyen, J. A., Ed.; Kluwer: Dordrecht, 2000.

6. Qiu, Y.; Park, K. Adv. Drug Deliv. Rev. 2012, 64, Supplement, 49.

7. Fyvie, T. J.; Frisch, H. L.; Semlyen, J. A.; Clarson, S. J.; Mark, J. E. Journal of

Polymer Science Part A: Polymer Chemistry 1987, 25, 2503.

8. Huang, W.; Frisch, H. L.; Hua, Y.; Semlyen, J. A. Journal of Polymer Science Part A:

Polymer Chemistry 1990, 28, 1807.

9. Wood, B. R.; Hodge, P.; Semlyen, J. A. Polymer 1993, 34, 3052.

10. Wood, B. R.; Joyce, S. J.; Scrivens, G.; Semlyen, J. A.; Hodge, P.; O'Dell, R.

Polymer 1993, 34, 3059.

11. Frisch, H. L.; Wasserman, E. Journal of the American Chemical Society 1961, 83,

3789.

12. DeBolt, L. C.; Mark, J. E. Macromolecules 1987, 20, 2369.

13. Joyce, S. J.; Hubbard, R. E.; Semlyen, J. A. European Polymer Journal 1993, 29,

305.

14. Kubo, M. In Topological Polymer Chemistry; WORLD SCIENTIFIC: 2013, p 199.

15. Mayumi, K.; Ito, K. Polymer 2010, 51, 959.

16. Gong, C.; Gibson, H. W. Journal of the American Chemical Society 1997, 119,

8585.

17. Zada, A.; Avny, Y.; Zilkha, A. European Polymer Journal 1999, 35, 1159.

18. Zada, A.; Avny, Y.; Zilkha, A. European Polymer Journal 2000, 36, 351.

19. Oike, H.; Mouri, T.; Tezuka, Y. Macromolecules 2001, 34, 6229.

20. Miki, K.; Inamoto, Y.; Inoue, S.; Uno, T.; Itoh, T.; Kubo, M. Journal of Polymer

Science Part A: Polymer Chemistry 2009, 47, 5882.

21. Oku, T.; Furusho, Y.; Takata, T. Angewandte Chemie 2004, 116, 984.

22. Kubo, M.; Matsuura, T.; Morimoto, H.; Uno, T.; Itoh, T. Journal of Polymer Science

Part A: Polymer Chemistry 2005, 43, 5032.

23. Okumura, Y.; Ito, K. Advanced Materials 2001, 13, 485.

24. Ito, K. Polym. J 2007, 39, 489.

CHAPTER 2

ORGANOCATALYSTS FOR POLYMERIZATION OF LACTIDE

Portions reprinted (adapted) with permission from

Stukenbroeker, T. S.; Bandar, J. S.; Zhang, X.; Lambert, T. H.; Waymouth, R. M. ACS Macro Letters 2015 4 (8), 853-856.

*Sections concerning cyclopropenimine catalysts were adapted from the above publication. 3-(butylimino)-N1,N1,N2,N2-tetracyclohexylcycloprop-1-ene-1,2-diamine hydrochloride were provided by J. S. Bandar and T. H. Lambert. Lactide polymerizations catalyzed by triazabicyclodecene were conducted by X. Zhang.

2.1 Introduction

Polylactide (PLA), also known as poly(lactic acid), is a common biodegradable material. The lactic acid starting material is most commonly derived from bacterial fermentation of starch or other polysaccharides, making it attractive as an alternative to petrochemical-derived materials.1 Lactic acid is converted to a polymer through several different methods depending on the molecular weight and stereocontrol required, which in turn depends on the application.2 For instance, the polymer may be either amorphous or semi-crystalline based on tacticity; this is an important parameter for various applications.3 High molecular weight, crystalline PLA biodegrades in the human body in months, making it suitable for bone or tissue implants.2 On the other hand, low molecular weight PLA and especially poly(lactide-co-glycolide) degrades more quickly, making it ideal for pharmaceutical delivery systems. Recently, PLA has been used for commercial disposable consumer plastic foodware items.1

Careful control of both the molecular weight and tacticity can be achieved by dimerizing lactic acid to lactide and employing ring opening polymerization. This process has been known since the 1950’s; since that time many different catalysts have been developed.4 These include organometallic complexes (coordination-insertion polymerization) and various compounds for cationic, anionic, or nucleophilic polymerization. The Waymouth lab has demonstrated a number of catalysts suitable for lactide polymerization5,6 including both electrophilic thioureas7 for monomer activation and various bases8 for nucleophilic ring-opening. These resulted in metal-free and extremely well-controlled polymerizations. However, there remains strong interest in developing catalysts that exhibit control over tacticity and molecular weight under mild and ambient reaction conditions.

Reported here are two investigations of lactide ring opening with different classes of compounds previously untested for ring-opening polymerization. The first, a dicationic bis(imidazolium), was an electrophile for activating the lactide monomer. It required the weakly basic co-catalyst sparteine. The second was a class of cyclopropenimine superbases, which showed polymerization of lactide through both an alcohol activation mechanism as well as an unexpected, enolate-initiated process.

2.2 Results and Discussion

2.2.1 Bis-Imidazolium Salts as Electrophilic Co-Catalysts for Lactide Polymerization

Scheme 2.1 Synthesis of bis(imidazolium) salts

Bis(imidazolium) 1 was synthesized by heating N-methyl imidazole with dibromomethane. Neither this compound (1•Br) nor the triflate salt (1•OTf), obtained following ion exchange, were soluble in THF or DCM, preventing 1 from being tested as a catalyst. To increase solubility in organic solvents, compound 2 was synthesized in a similar manner. The inclusion of butyl groups imparted solubility in DCM.

Scheme 2.2 Lactide polymerization catalyzed by (-)-sparteine and bis(imidazolium) 2•Br

Compound 2•Br was screened as a polymerization catalyst for lactide polymerization. With only an alcohol initiator, 2•Br did not polymerize lactide. However when a base co-catalyst, (-)-sparteine, was added polylactide was produced. After 24 h at room temperature, 55% of the lactide monomer had been converted to polymer. The molecular weight was not linear with conversion; the polymer grew to Mn=4000 in 6 h and then only to Mn=6200 after 24 h (Fig. 2.1). The associated increase in polydispersity suggested that chain scrambling was occurring. Control experiments showed no reaction 17

of lactide when any of the three components—bis(imidazolium), sparteine or benzyl alcohol—were excluded (Table 2.1). Analogous experiments with the bis(imidazolium) as a co-catalyst for the polymerization of valerolactone with DBU showed no reaction.

7,000 1.45 Mn Time Conv. M PDI 6,000 1.4 n 5,000 PDI 6 h 33.5% 4.0k 1.20 1.35 12 h 34.8% 5.0k 1.22

4,000 (Da)

1.3 18 h 44.3% 5.5k 1.30 PDI n 3,000 M 1.25 24 h 55.4% 6.2k 1.40 2,000 1,000 1.2 0 1.15 20% 40% 60% lactide conversion

Figure 2.1 Evolution of molecular weight and polydispersity with conversion of polylactide (conditions, see scheme 2.2)

Table 2.1 Various experiments using the bis(imidazolium)/sparteine system with lactide [lactide] [2•Br] [sparteine] [BnOH] Time Conv. M PDI (M) (M) (M) (M) (h) n 1.0 0.010 0.010 0.010 17 37% 4500 1.19

0.94 - 0.010 0.020 17 none - -

1.0 0.007 0.010 0.010 18 32% 4700 1.17

1.0 - 0.010 0.010 18 <1% - -

1.0 0.007 0.010 - 18 <2% - -

1.0 0.008 - 0.010 18 none - -

Polymerizations were conducted at R.T. in N2 glovebox in DCM solvent.

Binding experiments were conducted to determine the mechanism of activation for the bis(imidazolium). Compound 2•Br was combined with one equivalent of 2- isopropoxy-2-oxo-1,3,2-dioxaphospholane (iPP), a ring-opening polymerization substrate

of interest (see Chapter 3). Shifts in the proton NMR peaks of the bis(imidazolium) were observed. Identification of the resonances required NOESY and HMQC 2-D NMR experiments (see experimental section). The NMR studies showed downfield shifts of resonance ‘e’ and upfield shifts of ‘h’ and ‘g’ (Fig. 2.2) suggesting that the monomer was hydrogen bonding not via the proton on C2 (position ‘h’), but through the that on C5.9

This was unexpected; however there is literature precedence for “abnormal NHCs,”10,11 which bind to metals through C5.

1 Figure 2.2 H NMR shifts in d2-DCM of bis(imidazolium) 2•Br alone (top) and with 1 eq. of iPP (bottom), as well as the peak assignments.

To compare the behavior of the bis(imidazolium), 1-butylimidazole was observed in a 1:1 mixture with iPP. There were no detectable shifts in the backbone protons and only a minor downfield shift (ca. 0.2 ppm) in the 2-position (imidazolium) proton. This compound was an ineffective co-catalyst for the benzyl alcohol initiated sparteine polymerization of lactide. Finally, it was thought that monomer-catalyst interactions through the backbone protons could be tuned by the installation of functional groups on 19

the C2 carbon. Compound 3 was synthesized through a modified route (scheme 2.3).

However, 3•Br was not an effective catalyst for lactide polymerization for unknown reasons.

Scheme 2.3 Synthesis of 2-methyl bis(imidazolium) bromide

In conclusion, bis(imidazolium) salts were developed as electrophilic catalysts for ring-opening polymerization. Alkyl side-chains were required in order to impart solubility in organic solvents, a property that could be further tuned by selection of the counter-ion.

One compound was found to be an effective co-catalyst for the sparteine-catalyzed polymerization of lactide, though with uncertain control of the molecular weight and slow conversion compared with stronger bases. The mechanism of activation was hypothesized to be hydrogen bonding of the backbone protons to the carbonyl of lactide based on NMR experiments. With further development, this class of compounds could become promising alternatives to thioureas or squarene elecrophiles. It is worth noting that recent studies by other laboratories have embraced bis(thioureas) as a way of reducing catalyst loading in organocatalyzed ROP.12

2.2.2 Cyclopropenimines as Superbase Catalysts for Ring Opening Polymerization

A new class of nucleophile/base was also developed for lactide polymerization.

Lambert recently showed that bis(dialkylamino)-cyclopropenimines have comparable basicity to phosphazenes, are readily prepared in enantiomerically pure form, and are effective organocatalysts for enantioselective Michael and Mannich reactions of glycinate imines.13-15

Scheme 2.4 Cyclopropenimine catalysts (4-6) synthesized and monomers screened

Three achiral cyclopropenimines bearing N-alkyl substituents (Scheme 2.4) were prepared13-15 and tested for ring opening polymerization. Polymerizations of racemic lactide initiated with 1-pyrenebutanol in the presence of catalytic amounts of cyclopropenimine 4 proceeded rapidly with greater than 85% conversion in 30 seconds.

Molecular weights up to 13 kDa were readily obtained by controlling the monomer to alcohol ratio. The molecular weight distributions ranged from 1.2-1.4 and generally increased over the course of the reaction (Table 2.2 entries 4-6). Endgroup analysis by both 1H NMR and MALDI-TOF mass spectrometry showed clear evidence from pyrenebutanol endgroups (Fig. 2.4); this was further corroborated by gel permeation chromatography which showed a close correspondence between chromatograms with

UV (λ =327 nm) and RI detection (Fig. 2.3). 21

Table 2.2: ROP with cyclopropenimine 4 and 1-pyrenebutanola [ROH] conv. M b M c entry monomer time n n M /M /M (%) /Da (theor.) w n 1 rac-lactide 0.005 2 min 88 12800 25367 1.43 2 rac-lactide 0.020 2 min 98 8400 7056 1.44 3 rac-lactide 0.050 2 min 97 3600 2794 1.72 4 rac-lactide 0.010 0.5 min 89 11400 12816 1.34 5 rac-lactide 0.010 1 min 93 12400 13392 1.36 6 rac-lactide 0.010 2 min 93 12500 13392 1.48 7 rac-lactide 0.020 8 min 97 8880 6990 1.19 8 carbonatec 0.020 22 hrs 45 3200 2295 1.16 aConditions: 1.0 M monomer in dicholoromethane, 0.010 M catalyst 1, room temperature. Quenched with either 4-nitrophenol or benzoic acid. ROH=1-pyrene butanol. bdetermined by GPC vs. polystyrene standards. c theoretical molecular weight based on monomer:initiator and conversion. c carbonate=trimethylene carbonate

Refractive Index

UV (327nm)

0 5 10 15 20 25 30 Retention Time (min)

Figure 2.3. GPC of alcohol-initiated lactide polymerization (Table 1, entry 7).

Figure 2.4: MALDI-MS of PLA synthesized with 4 and pyrenebutanol (Table 1, entry 7).

The polymerization activities of the N-alkyl cyclopropenimines in the presence of alcohol initiators are higher than those of 4-dimethylaminopyridine (DMAP)16 or phosphines17 and comparable to those of N-heterocyclic carbenes (NHCs)18 and guanidines.19 Nevertheless, the molecular weight distributions observed (Table 2.2) are relatively broad and increased at longer reaction times. These data, coupled with the observations that polymeric ions corresponding to both odd and even lactic acid units were observed in the MALDI-TOF spectra (Figure 2.4) suggest that competitive transesterification reactions occur, leading to chain-scrambling and chain-transfer reactions.5,20

The deviation of the observed molecular weights from that predicted from the initial [M]0 / [ROH]0 at low initiator concentrations (Table 2.1, entry 1) led us to suspect

that the cyclopropenimines may be capable of acting as an initiator, as previously observed in polymerizations with the amidine 1,8-diazabicyclo-[5.4.0]undec-7-ene

(DBU).21,22 To test for competitive nucleophilic polymerization mechanisms by the cyclopropenimines, we investigated the ring-opening polymerization of lactide with cyclopropenimines 4-6 in the absence of alcohols (Table 2.3).

Table 2.3: Initiator-Free ROP of Lactides with Cyclopropeniminesa cat. conv. entry monomer catalyst conc. solvent time M b M /M (%) n w n (M)

1 rac-lactide 4 0.010 CH2Cl2 8 min 84 8390 1.42

2 rac-lactide 4 0.050 CH2Cl2 10 min 99 15300 1.57

c 3 rac-lactide 4 0.007 C6D6 2 days 99 70700 1.46

4 L-lactide 4 0.010 CH2Cl2 8 min 93 11500 1.30

5 L-lactide 5 0.010 CH2Cl2 8 min 90 17300 1.46

6 rac-lactide 6 0.020 CD2Cl2 20 min 98 13100 1.38

7 rac-lactide BEMP 0.010 CH2Cl2 8 min 65 8300 1.24

e 8 carbonate 4 0.010 CH2Cl2 22 hrs 3 - - a Conditions: 1.0 M monomer in solvent, room temperature. Quenched with either 4- nitrophenol or benzoic acid. b Determined by GPC vs. polystyrene standards. c Saturated monomer solution. d 0.35M monomer solution. e Carbonate=trimethylene carbonate.

conv. b conv. Mn

time Mn Mw/Mn time M /M (%) (%) /Da w n 0.5 min 55 10200 1.57 0.5 min 89 11400 1.34

1 min 69 11800 1.54 1 min 93 12400 1.36

NOalcohol with with alcohol 2 min 80 13500 1.58 2 min 93 12500 1.48

Figure 2.5: Comparison of polymerization rates with and without and alcohol. Conditions: [Catalyst 4]=0.01M, [rac lactide]=1.0M, alcohol initiated entries [1-pyrene butanol]=0.01M, room temperature in DCM.

Under these conditions, polymerization proceeded readily with rates only marginally slower than those observed in the presence of alcohol initiators. The molecular weights obtained ranged from Mn = 8000 - 70,000 Da and were observed to increase with increasing conversion, but exhibited little correlation with the initial [M]0 /

[I]0 ratio (where I = cyclopropenimine, Table 2.2, entry 2). Analysis of the resulting polymers by MALDI-TOF mass spectrometry revealed ions corresponding to exact multiples of lactide molecular weights. These data would be consistent with a cyclic polymer generated by a nucleophilic zwitterionic mechanism, but several lines of evidence indicate that a linear polymer is generated. Comparison of the dilute solution viscosities of a high molecular weight polylactide (PLA) prepared from the cyclopropenimine 4 and a known linear sample of PLA were similar (Fig. 2.6), implicating a linear topology for both samples.23 Furthermore, analysis of the purified polymer by 1H

NMR revealed two resonances indicative of polymer endgroups: one at δ 4.37 ppm

(CDCl3), diagnostic of a methine proton adjacent to a terminal hydroxyl group and another endgroup signal at δ 5.01 ppm (Fig. 2.7). These data are inconsistent with a nucleophilic zwitterionic mechanism, as observed for NHCs, amidines, and isothiorureas,22,24 but imply that, in the absence of alcohols, the cyclopropenimines initiate lactide polymerization by an alternate pathway.

Figure 2.6: Mark-Houwink Plot of sample synthesized with cyclopropenimine (Table 1, entry 3) compared against a known linear PLA. No significant difference in intrinsic viscosity was detected.

1 Figure 2.7. H NMR (CDCl3) of polylactide obtained from cyclopropenimine initiator before (top) and after (bottom) acylation with excess acetyl chloride and triethylamine.

As cyclopropenimines are potent bases (pKa of conjugate acid approx. 27 in

13 CH3CN), we reasoned that these superbases might deprotonate lactide to a lactide enolate, which subsequently initiates the polymerization of lactide (Scheme 2.5). Several recent reports have indicated that Zr, Zn, or Li enolates can initiate lactide polymerization.4,9,25-27

Scheme 2.5: Proposed Enolate Initiation Mechanism

The endgroup resonance observed in the 1H NMR spectrum at δ = 5.01 ppm in the 1H NMR spectra is consistent with that expected for a methine proton of an alkylated lactide.28 To contrast this with an O-acylated lactide, a lactide with a single trimethylsilyl group installed at one of the esters was prepared (Scheme 2.6). This silyl enolate was synthesized by deprotonating lactide with lithium diisopropylamide in the presence of excess trimethylsilyl chloride at low temperature. The methine proton of the non-silyated ester exhibits a resonance at δ 4.41 ppm. These data, as well as model studies described below, indicate that initiation involves acylation of lactide enolate at carbon rather than at oxygen.

Scheme 2.6 Synthesis of trimethylsilyl lactide

To assess if the cyclopropenimine is capable of deprotonating lactide, lactide was treated with cyclopropenimine 4 (1:2 ratio) in C6D6 at room temperature. The addition of 4 to lactide results in the rapid disappearance of the lactide methine resonance at δ = 3.67 ppm (Fig. 2.8), consistent with the deprotonation of lactide to the lactide enolate.

Figure 2.8: Lactide and cyclopropenimine 4 (1:2 ratio) in d6-benzene (R.T., 30 min). Note the methine signal from lactide (δ 3.7) disappears. Much smaller signals at appear δ 5.0 and δ 4.1, and are assigned to polymer and partially deprotonated lactide, respectively. The polymer peak (δ 5.0) was found to increase over the course of several hours.

Furthermore, epimerization of lactide monomers to form meso-lactide is observed during early stages of the polymerization reaction and polymerizations of L- lactide produce a small amount of atactic sequences, suggesting that reversible deprotonation occurs (Fig. 2.9).

1 Figure 2.9: H NMR (CDCl3) of purified (left) and unpurified (right) polylactide obtained from L-lactide with cyclopropenimine 4, enlarged to show methine region. Note the smaller downfield peaks indicating the presence of atactic sequences (δ 5.21-5.22) and the meso lactide peaks (δ 5.05-5.10) generated in the unpurified material.

To provide further evidence that enolates4,25-27 initiate the polymerization of lactide under these reaction conditions, the enolate of methyl isobutyrate was generated with LDA and utilized as an initiator for lactide polymerization (Scheme 2.7).29

Generation of the enolate of methyl isobutyrate with LDA in the presence of excess ester at -78 °C, followed by the addition of 25 equivalents of L-lactide resulted in the conversion of 69% of the lactide within 10 minutes to generate poly(lactide). Analysis of the MALDI-TOF mass spectra of the resulting polymer yielded ions corresponding to linear poly(lactide) with methyl isobutryate endgroups. Analysis of the 13C NMR of the resulting low molecular weight poly(lactide) (Mn ~ 2700 Da) was consistent with a 2,2- dimethyl methyl acetate endgroup resulting from acylation at the carbon center of the methyl isobutyrate enolate initiator (Fig. 2.10). These results are consistent with literature examples of carbon alkylation of methyl isobutyrate,30-33 and indicate that ester enolates initiate lactide polymerization by a Claisen-type condensation at the carbon of the initiating enolate.

Scheme 2.7: Lactide polymerization initiated by methyl isobutyrate enolate.

13 Figure 2.10: C NMR (CDCl3) of a lactide polymerization initiated by methyl isobutyrate enolate. Upper spectrum is enlarged to show smaller peaks.

The observation that trimethylene carbonate (TMC) polymerizes with cyclopropenimine 4 only in the presence of alcohol initiators is readily explained by the lack of enolizable protons of this carbonate monomer.

1 Figure 2.11: H NMR (CDCl3) spectra of Table 2.3, entry 1, enlarged to show observed endgroups. LA=lactide monomer

Figure 2.12: MALDI-MS analysis of PLA synthesized with BEMP (Table 2.3, entry 7).

As several superbases have been used as organocatalysts for ring-opening

5 polymerization, we sought to compare the behavior of the cyclopropenimines (pKa conjugate acid ~27) to another basic organocatalyst 2-tert-butylimino-2-diethylamino-1,3-

34 dimethylperhydro-1,3,2-diazaphosphorine (BEMP, pKa conjugate acid ~ 27.6 ). The ring-opening polymerization of lactide in the absence of alcohols with the phosphazene

BEMP (Table 2.3, Entry 7) was slightly slower but displayed similar polymer endgroup resonances (δ 5.01 and 4.37 ppm, see Fig. 2.11) and MALDI spectra (Fig. 2.12) to those of cyclopropenimines, indicative of an enolate-initiated polymerization mechanism.

In summary, superbases are a versatile class of catalysts for organocatalytic ring-opening polymerization reactions, whose polymerization behavior depends

sensitively on their basicity, nucleophilicity, and the presence or absence of alcohol initiators. In the presence of alcohol initiators, these superbases catalyze ring-opening polymerization by hydrogen-bond activation of initiating or propagating alcohols.

However, at low initiator (i.e. alcohol) concentrations, competitive reactions of the superbases with lactone monomers can lead to alternate mechanisms of initiation and polymerization. These competitive mechanisms can lead to a broadening of the molecular weight distributions, particularly at high [monomer]/[initiator] ratios.

Polymerization of lactones in the absence of alcohols can illuminate these alternative pathways and reveal a range of behaviors. N-heterocyclic carbenes,24,35 amidines,36 and isothioureas22 act as nucleophilic initiators, mediating zwitterionic ring-opening polymerizations to generate cyclic polyesters. Herein we demonstrated that cyclopropenimine and phosphazene superbases deprotonate lactide to generate enolates that initiate lactide polymerization.

Scheme 2.8 Two polymerization mechanisms for lactide with organic catalysts.

2.3 Experimental Procedures

General Considerations: All polymerizations were conducted in a dry nitrogen glovebox. Dichloromethane (DCM) used for polymerizations was stirred over calcium hydride and distilled. All chemicals were purchased from Sigma Aldrich unless otherwise specified. Polystyrene calibrated molecular weights were obtained on a

Viscotek GPCMax with two Waters columns (300 mm by 7.7 mm) in THF at 35°C at a flow rate of 1.0 mL/min and Viscotek S3580 refractive index detector. Monodisperse polystyrene calibrants ranged from Mp = 500 to 275,000. NMR data was collected on

300, 400 and 500 MHz Varian instruments.

MALDI-TOF samples were prepared by combining tetrahydrofuran (THF) solutions of dithranol (10 mg/mL), polymer (10 mg/mL) and sodium iodide (0.1 M) in a 10:10:1 ratio.

Mass spectra were obtained at either the Molecular Foundry, a part of Lawrence

Berkeley National Laboratory, on an Applied Biosystems Sciex TF4800 instrument or at the Stanford University Mass Spectrometry Center on a Bruker Microflex instrument.

Calibrants included peptide and protein standards. Expected masses of polymers were modeled using mMass software (http://www.mmass.org/ developed by Martin Strohalm).

Synthesis of 1,1’-methylenebis(3-methyl-1H-imidazol-3-ium) bromide (1•Br): 0.50 mL of 1-methylimidazole (FW=82.10, d=1.03) was added to 12.5 mL of p-xylene in glass pressure tube. 2.30 mL of dibromomethane (FW=173.83, d=2.47) were added. The vessel was flushed with N2, sealed, and heated to 200°C. A white precipitate formed in

10 min. After 20 h, the reaction had turned dark brown. The reaction was allowed to cool and then filtered. The filtrate was washed 3 times with hexanes. 0.78 g of light brown solid were recovered (37% yield). 1H NMR (DMSO): δ 3.92 (6H, s), 6.67 (2H, s),

7.81 (2H, s), 8.00 (2H, s), 9.42 (2H, s)

Ion exchange of 1•Br to 1•OTf 57.9 mg of compound 1•Br (FW=338.04) was dissolved in 3 mL methanol in a large vial with stirbar. 44 mg silver triflate was added and sonicated. A white suspension formed immediately. Stirred for 1 h at 40°C and then 16 h at RT. Product was filtered and 17.8 mg were recovered (22% yield). 1H NMR

(DMSO): δ 3.89 (6H, s), 6.61 (2H, s), 7.78 (2H, s), 7.93 (2H, s), 9.31 (2H, s)

Synthesis of 1,1’-methylenebis(3-butyl-1H-imidazol-3-ium) bromide (2•Br): 0.86 mL of 1-butylimidazole (FW=124.18, d=0.945) as added to 12.5 mL of p-xylene in glass pressure tube. 2.3 mL of dibromomethane (FW=173.83, d=2.47) were added. The vessel was flushed with N2, sealed, and heated to 100°C. A white precipitate formed in

10 minutes. After 16 h, the reaction had turned dark brown. The reaction was allowed to cool and then filtered. The filtrate was washed with toluene. 0.41 g (31% yield) of

1 light brown solid were recovered. H NMR (CD2Cl2): δ 0.49 (6H, t), 1.39 (4H, sex), 1.90

(4H, quint), 4.18 (4H, t), 7.22 (2H, s), 7.59 (2H, s), 9.45 (2H, s), 11.45 (2H, s). See 2D

NMR experiments below.

Figure 2.13 2-D NMR experiments (NOESY, top, and HMQC, bottom) to identify proton resonances of 2•Br

Synthesis of 1,1-methylenebis(3-hexyl-2-methyl imidazolium) bromide (3•Br) 4.10 g of 2-methyl-imidazole (Aldrich, FW=82.10) was mixed with 8.42 g (1.2 equivalents) of

1-bromohexane (Aldrich, FW=165.07, d=1.18). Refluxed at 120°C for 1 h. Heat removed, deionized water added. Several grams of sodium hydroxide added to precipitate a blackish product. Extracted with chloroform, dried with sodium sulfate.

Chloroform removed under vacuum. Vacuum distilled. Clear distillate appeared at bath

1 temperature 130°C, condensate temp 60°C, pressure 15 millitorr. H(CDCl3): δ 0.9 (3H,

37 t) δ 1.3 (6H, m) δ 1.7 (2H, quin) δ 2.0 (3H, s) δ 3.8 (2H, t) δ 6.8 (1H, s) δ 6.9 (1H, s) δ

7.2 (1H, s) Yield 1.86g (22.4%)

Product was then added to pressure vessel along with 0.47 mL (0.6 equivalents) of dibromomethane (Aldrich, FW=174.83, d=2.48) and heated to 45°C for three days. Pure white precipitate filtered off, washed with toluene. Yielded styrofoam-type material.

1 Dried under vacuum. Soluble in chloroform, DCM. Insoluble in THF. H(CDCl3): δ 0.9

(6H, t) δ 1.3 (12H, m) δ 1.8 (4H, quin) δ 3.0 (6H, s) δ 4.0 (4H, t) δ 7.2 (2H, s) δ 7.8 (2H, s) δ 8.4 (2H, s) Yield 0.86g (31%)

Linear Polymerization of δ-Valerolactone (VL) Using (-)-sparteine and bis- imidazolium as Co-catalysts 103 mg of VL (FW=100.12) was dissolved in DCM. 2.5 mg of 2•Br were added, followed by 1.4 mg sparteine and 0.6 mg BnOH. VL:

2•Br:sparteine:BnOH =185:1.1:1.1:1, [VL]0=1.55M. Reaction proceeded for 22 h then quenched and solvent removed under vacuum. NMR showed no conversion.

Linear Polymerization of Lactide Using (-)-sparteine and bis-imidazolium as Co- catalysts (Typical Procedure) 100 mg of rac-lactide was measured out and stirbar added. In a separate vial, 2•Br was combined with (-)-sparteine, benzyl alcohol, and

DCM. This was poured into the monomer to dissolve it and start polymerization.

Reaction was quenched with a drop of acetic acid and solvent was removed.

NMR investigation of 3•Br binding with various monomers The methylated bis- imidazolium salt 3•Br was mixed in J-young tubes with iPP or VL monomers in d2-DCM.

In both cases, the molar ratio of monomer to catalyst was approximately 5:1. The iPP sample showed no shift in the imidazolium protons. The VL showed very small

(<0.1ppm) downfield shifts in the backbone protons.

Synthesis of 3-(butylimino)-N1,N1,N2,N2-tetracyclohexylcycloprop-1-ene-1,2- diamine hydrochloride (4•HCl): The catalyst was synthesized according to the literature.13 Briefly, dicyclohexylamine (6.71 mL, 6 eq.) was added to tetrachlorocyclopropene (0.64 mL, 1 eq.) in 50 mL DCM and stirred for 4 h. A cloudy orange precipitate formed immediately. n-Butylamine (0.834 mL, 1.5 eq.) was added and the reaction mixture was stirred for an additional 11.5 h. The reaction was diluted with DCM and washed with 2.0M HCl. The organic layer was dried and volatiles were removed under vacuum. The remaining orange solid was recrystallized from

1 EtOAc/hexanes. Yield=1.06 g (37%). NMR: H (CDCl3): δ 12.1 (N-H), 3.8 (q, 2H), 3.3

(m, 4H), 2.0-1.2 (m, 44H), 0.9 (t, 3H).

Synthesis of N1,N1,N2,N2-tetracyclohexyl-3-(hexylimino)cycloprop-1-ene-1,2- diamine hydrochloride (5•HCl): Substance was prepared in the same manner as the n- butyl compound 1, above, except n-hexylamine was used in place of butylamine.

1 Yield=58%. NMR: H (CDCl3): δ 9.0 (N-H), 3.8 (q, 2H), 3.3 (m, 4H), 2.0-1.2 (m, 48H),

0.8 (t, 3H).

Air-Free Deprotonation of Cyclopropenimine Catalyst: Approximately 1 g of 4•HCl or

5•HCl was mixed with 1 eq. each sodium hydride and potassium tert-butoxide in 20 mL of THF and allowed to stir for 1 h. THF was removed under vacuum and pentane was added. The resulting suspension was filtered to yield a clear solution. Volatiles were removed under vacuum to give the free base in quantitative yield. In the case of 5, the product recrystallized from Et2O at -35°C to yield a white powder. The N-H signal was absent from the 1H NMR spectrum.

Scheme 2.9 Lactide polymerization by cyclopropenimines

Representative Polymerization of Lactide with 4: 100 mg (0.694 mmol) of lactide and 3.2 mg (0.012 mmol) of 1-pyrenebutanol were dissolved in DCM inside the glove box. In another vial, 4 was dissolved in DCM and added to the monomer with stirring, to bring the initial concentration of monomer to 1.0 M. For kinetic experiments, aliquots were taken at specific times by quenching with excess 4-nitrophenol or benzoic acid.

For polymerizations without an alcohol initiator, the pyrenebutanol was excluded.

Conversion was determined by integration of the lactide methine peak in the 1H NMR.

The crude mixture was purified by precipitation into MeOH and drying under vacuum.

Alternative solvents, catalysts, or monomers were used in the same manner.

Figure 2.14: ESI-MS of PLA synthesized without alcohol initiator (sample from Figure S6, 2 min entry, no alcohol).

Acylation of polylactide endgroups: Polymerization was carried out and quenched as described above. Excess acetyl chloride and triethylamine were then added to the reaction mixture and allowed to stir overnight. The resulting brown reaction solution was

40 concentrated under vacuum and precipitated into methanol. The purified polymer was dried under vacuum.

Lactide polymerization initiated by model enolate: 8.5 mg (0.083 mmol) of methyl isobutyrate was dissolved in 3 mL of THF and cooled to -78°C. 0.028 mmol of lithium diisopropylamide was added dropwise over 10 min via a 0.99 M solution in

THF/hexanes. 100 mg (0.694 mmol) of L-lactide were added rapidly via solution in 1 mL

THF. The solution was warmed to RT over 10 min and acetic acid was added to quench the reaction. The solvent was then removed under vacuum. 1H NMR analysis showed

69% polymerization of the lactide. The polymer was purified by precipitation in methanol and recovery via filtration. GPC analysis showed an approximate Mn=2700. ESI-MS

(acetonitrile) showed polymer peaks.

In a control experiment, 1.9 µL (0.014 mmol) of diisopropylamine were added to 4 mL of

THF and cooled to -78°C and a THF solution containing 50 mg (0.347 mmol) of lactide was added dropwise. The reaction was warmed to RT over 10 min. 1H NMR analysis showed no reaction of the lactide.

Figure 2.15: MALDI-TOF of a polymerization initiated by methyl isobutyrate enolate. The major peak corresponds to the ester-initiated polylactide.

Trapping of Lactide Enolate: 50 mg (0.35 mmol) of rac-lactide were dissolved with

188 mg (1.73 mmol) of trimethylsilyl choride in 4 mL of THF. The solution was cooled to

-78°C and 1 equivalent of lithium diisopropylamide was added dropwise via a 0.99 M solution in THF/hexanes. After 1 h, the solution was diluted with 20 mL of diethyl ether and filtered through Celite under nitrogen atmosphere. Volatiles were removed under vacuum and 70 mg of a slightly yellow liquid was obtained. 1H NMR showed 24 mol % of the isolated material was singly protected (one TMS group installed), 29 mol % was doubly protected (two TMS groups), and the remaining was unreacted lactide (no

1 epimerization detected). NMR: H (CDCl3): δ 4.68 (1H, q, J=6.8 Hz), 1.86 (3H, s), 1.55

13 (3H, d, J=6.8 Hz), 0.23 (9H, s) C (CDCl3): δ 166.35, 143.08, 115.60, 71.75, 47.57,

1 19.48, 0.30, H (C6D6): δ 3.383 (1H, q, J=6.7 Hz), 1.75 (3H, s), 1.25 (3H, d, J=6.7 Hz),

0.07 (9H, s).

One drop of concentrated HCl was shaken with the chloroform NMR sample from above.

1 After 45 minutes, the mixture was dried with MgSO4. H NMR showed 8% polylactide,

80% rac-lactide, and 12% meso lactide and complete disappearance of all peaks associated with singly- and doubly-TMS-protected lactide.

Formation of Lactide Enolate by Cyclopropenimine: 8.6 mg of L-lactide were dissolved in 500 µL of d6-benzene and added dropwise to a stirring solution of 56 mg of

4 in 100 µL of d6-benzene at room temperature. After addition, the solution was transferred to a J-Young tube inside the glovebox. See 2.8 for NMR spectra.

2.4 References

1. Lunt, J.; Bone, J. AATCC Review 2001, 1, 20.

2. Gupta, A. P.; Kumar, V. Eur. Polym. J. 2007, 43, 4053.

3. Sarasua, J. R.; Arraiza, A. L.; Balerdi, P.; Maiza, I. Polymer Engineering & Science

2005, 45, 745.

4. Dechy-Cabaret, O.; Martin-Vaca, B.; Bourissou, D. Chem. Rev. 2004, 104, 6147.

5. Kiesewetter, M. K.; Shin, E. J.; Hedrick, J. L.; Waymouth, R. M. Macromolecules

2010, 43, 2093.

6. Kamber, N.; Jeong, W.; Waymouth, R. M.; Pratt …, R. Chem. Rev 2007.

7. Dove, A. P.; Pratt, R. C.; Lohmeijer, B. G. G.; Waymouth, R. M.; Hedrick, J. L. J. Am.

Chem. Soc. 2005, 127, 13798.

8. Lohmeijer, B. G. G.; Pratt, R. C.; Leibfarth, F.; Logan, J. W.; Long, D. A.; Dove, A. P.;

Nederberg, F.; Choi, J.; Wade, C.; Waymouth, R. M.; Hedrick, J. L. Macromolecules

2006, 39, 8574.

9. Hibbert, F.; Emsley, J. In Adv. Phys. Org. Chem.; Bethell, D., Ed.; Academic Press:

1991; Vol. Volume 26, p 255.

10. Aldeco-Perez, E.; Rosenthal, A. J.; Donnadieu, B.; Parameswaran, P.; Frenking, G.;

Bertrand, G. Science 2009, 326, 556.

11. Lebel, H.; Janes, M. K.; Charette, A. B.; Nolan, S. P. J. Am. Chem. Soc. 2004, 126,

5046.

12. Spink, S. S.; Kazakov, O. I.; Kiesewetter, E. T.; Kiesewetter, M. K. Macromolecules

2015, 48, 6127.

13. Bandar, J. S.; Lambert, T. H. J. Am. Chem. Soc. 2012, 134, 5552.

14. Bandar, J. S.; Lambert, T. H. J. Am. Chem. Soc. 2013, 135, 11799.

15. Bandar, J. S.; Barthelme, A.; Mazori, A. Y.; Lambert, T. H. Chemical Science 2015,

6, 1537.

16. Nederberg, F.; Connor, E. F.; Möller, M.; Glauser, T.; Hedrick, J. L. Angewandte

Chemie International Edition 2001, 40, 2712.

17. Myers, M.; Connor, E. F.; Glauser, T.; Möck, A.; Nyce, G.; Hedrick, J. L. Journal of

Polymer Science Part A: Polymer Chemistry 2002, 40, 844.

18. Connor, E. F.; Nyce, G. W.; Myers, M.; Möck, A.; Hedrick, J. L. Journal of the

American Chemical Society 2002, 124, 914.

19. Pratt, R. C.; Lohmeijer, B. G. G.; Long, D. A.; Waymouth, R. M.; Hedrick, J. L.

Journal of the American Chemical Society 2006, 128, 4556.

20. Penczek, S.; Duda, A.; Libiszowski, J. Macromolecular Symposia 1998, 128, 241.

21. Brown, H. A.; De Crisci, A. G.; Hedrick, J. L.; Waymouth, R. M. ACS Macro Letters

2012, 1, 1113.

22. Zhang, X.; Waymouth, R. M. Acs Macro Letters 2014, 3, 1024.

23. Semlyen, J. A. In Cyclic Polymers; 2nd ed.; Semlyen, J. A., Ed.; Kluwer: Dordrecht,

2000, p 790.

24. Brown, H. A.; Waymouth, R. M. Acc. Chem. Res. 2013, 46, 2585.

25. Dobrzynski, P. J. Polym. Sci., Part A: Polym. Chem. 2004, 42, 1886.

26. Pastusiak, M.; Dobrzynski, P.; Kaczmarczyk, B.; Kasperczyk, J.; Smola, A. Polymer

2011, 52, 5255.

27. Bhaw-Luximon, A.; Jhurry, D.; Spassky, N.; Pensec, S.; Belleney, J. Polymer 2001,

42, 9651.

28. Trimaille, T.; Möller, M.; Gurny, R. J. Polym. Sci., Part A: Polym. Chem. 2004, 42,

4379.

29. Dong, S.; Parker, G. D.; Tei, T.; Paquette, L. A. Org. Lett. 2006, 8, 2429.

30. Wang, Z.; Ni, J.; Kuninobu, Y.; Kanai, M. Angew. Chem., Int. Ed. 2014, 53, 3496.

31. Nagendiran, A.; Verho, O.; Haller, C.; Johnston, E. V.; Bäckvall, J.-E. The Journal of

Organic Chemistry 2014, 79, 1399.

32. Friedrich, J.; Dörrich, S.; Berkefeld, A.; Kraft, P.; Tacke, R. Organometallics 2014,

33, 796.

33. Stoltz, B. M.; Bennett, N. B.; Duquette, D. C.; Goldberg, A. F. G.; Liu, Y.; Loewinger,

M. B.; Reeves, C. M. In Comprehensive Organic Synthesis II (Second Edition); Knochel,

P., Ed.; Elsevier: Amsterdam, 2014, p 1.

34. Schwesinger, R.; Schlemper, H. Angew. Chem. 1987, 99, 1212.

35. Jeong, W.; Shin, E. J.; Culkin, D. A.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem.

Soc. 2009, 131, 4884.

36. Brown, H. A.; De Crisci, A. G.; Hedrick, J. L.; Waymouth, R. M. ACS Macro Letters

2012, 1, 1113.

CHAPTER 3

POLYMERIZATION OF PHOSPHOTRIESTER MONOMERS

Portions reprinted (adapted) with permission from

Stukenbroeker, T. S.; Solis-Ibarra D.; and Waymouth, R. M. Macromolecules, 2014, 47 (23), pp 8224–8230

*Sections concerning the synthesis and polymerization of 2-isopropoxy-2-oxo-1,3,2- dioxaphospholane-based were adapted from the above publication. X-ray crystallography was performed and refined by Dr. Diego Solis-Ibarra.

3.1 Introduction

Organocatalytic ring-opening polymerization1-3 of cyclic monomers has proven a versatile strategy for the synthesis of well-defined poly(esters),2 poly(carbonates),2,4 poly(siloxanes)5-7 and poly(alkylene phosphates).8-10 Organic catalysts can mediate ring- opening by a variety of mechanisms, including acid-catalysis, H-bonding catalysis, general base catalysis and nucleophilic pathways.1,2 Nucleophiles can mediate the ring- opening polymerization of lactones or carbosiloxanes by a zwitterionic ring-opening polymerization mechanism to generate cyclic polyesters or cyclic poly(carbosiloxanes)

(Scheme 3.1).2,6,11 As part of our interest in extending zwitterionic12,13 ring-opening polymerization with other classes of monomers11 to generate cyclic polymers,14-18 we report herein the zwitterionic ring-opening polymerization of cyclic phosphate monomers to macrocyclic poly(alkylene phosphates) and their entrapment in crosslinked hydrogels.

Scheme 3.1 Zwitterionic Ring Opening Polymerization (ZROP) mechanism shown for δ-Valerolactone with NHC catalyst.

The ring-opening polymerization of cyclic phosphates with metal alkoxide catalysts as a strategy to generate poly(alkylene phosphates) was investigated extensively by Penzcek.19-21 Poly(alkylene phosphates), which share the phosphoester backbone of polynucleic acids, are biodegradable and biocompatible.22 These attributes have engendered considerable interest as biomedical materials.23-26 The organocatalytic ring-opening polymerization of cyclic phosphates was recently investigated by Iwasaki,8

Jérôme9, Wooley10,27, and others28 utilizing the guanidine 1,5,7-triazabicyclo[4.4.0]undec-

5-ene (TBD), the amidine 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) or a combination of

DBU and a thiourea as catalysts. For the generation of linear poly(alkylene phosphates) the DBU/TU system was observed to exhibit the highest level of control9 and, in analogy to that proposed for the ring-opening polymerization of lactones,2 was proposed to occur by a combination of H-bond activation of the cyclic phosphate by the thiourea and general-base activation of the alcohol by DBU.9

The zwitterionic ring-opening polymerization11 of cyclic phosphates in the absence of alcohol initiators would require a nucleophilic mechanism for phosphoester transesterification. Nucleophilic transesterification mechanisms at phosphorous centers are known and have been invoked for phosphoryl transfer mediated by enzymes such as the histidine kinases and nucleoside diphosphate kinases (NDPK).29 Herein we report that N-heterocyclic carbenes30 can mediate the zwitterionic ring-opening polymerization of 2-isopropoxy-2-oxo-1,3,2-dioxaphospholane-based (iPP) to generate cyclic poly(alkylene phosphates).

3.2 Results and Discussion

3.2.1 Benzyl Phospholane Monomer

A number of phospholane monomers bearing either alkoxy or aryloxy groups were synthesized. Benzyloxy phospholane (BP) and 4-methylbenzyloxy phospholane

(mBP) were synthesized via addition of benzyl alcohol or 4-methylbenzyl alcohol to 2- chloro-1,3,2-dioxaphospholane 2-oxide. The monomer was a white solid that sublimed under vacuum, allowing for rigorous purification. At room temperature in the glovebox it was stable for several weeks, though eventually degraded to oligomeric products.

Scheme 3.2 Polymerization of benzyl phospholane with TBD and an alcohol initiator. Subsequent hydrogenation yielded a water-soluble polymer.

The BP monomer polymerized with TBD and an alcohol initiator. The resulting amorphous polymer was treated with H2 and Pd/C. Following treatment, the resulting polymer product was insoluble in chloroform. NMR in D2O revealed the absence of all aryl peaks, indicating that the benzyl group had been entirely cleaved to yield a polyanionic polymer (Fig. 3.1).

1 Figure 3.1 H NMR (D2O) of poly(alkylene phosphate) in Scheme 3.2. Note the lack of aryl peaks indicating cleavage of the benzyl groups.

Table 3.1 Representative polymerization of Benzyl Phospholane (BP) monomer with NHC and organometallic catalysts a

cat. IV at [BP]0 time conv b b RH entry catalyst conc Mn Mw/Mn c MW=50 (M) (min) (%) (nm) b,d (M) kDa 1 1.0 1 0.023 4 96 59000 1.21 3.67 0.041 2 0.5 1 0.005 4 91 13700 1.31 1.85 0.034 3 1.0 1 0.010 4 90 23100 1.34 2.39 0.036

4 2.0 Al(OMe)Et2 0.040 16 h 99 72000 1.23 3.52 0.026

5 1.0 Al(OMe)Et2 0.020 16 h 91 27395 1.57 2.43 0.028 6e 1.0 1d 0.010 4 97 17700 1.14 1.94 0.032 a b Polymerizations carried out in THF at room temperature in N2 glove box determined by light scattering with dn/dc=0.066 M-1, c hydrodynamic radius, d intrinsic viscosity (IV) as determined by in-line viscometer; in some cases the log(IV) vs log(MW) relation was extrapolated to obtain IV at MW=50 kDa, e contained 0.030 M 1-pyrene butanol as alcohol initiator (theoretical Mn=6900 Da)

Polymerization with NHC 1 rapidly afforded polymer in both the presence and absence of an alcohol initiator. The reactions proceeded nearly to completion. Despite stringent air-free purification, these monomers showed significant batch-to-batch

52 variation. Furthermore, in the alcohol-free polymerizations, the molecular weight of the polymers obtained increased as the amount of carbene present was increased, contrary to the expected trend31 (Table 3.1, runs 1 vs. 3). This suggests that initiation by the carbene was not efficient.32 When an alcohol was present (Table 3.1, run 6), the molecular weight of the polymers obtained was higher than that predicted by the amount of initiator, again indicative of a slow or poorly-controlled initiation event. The modest polydispersities (1.1

To determine the topology of the poly(BP) obtained from NHC-catalyzed polymerization, samples were polymerized with an organometallic catalyst Al(OMe)Et2.

This catalyst operates via a coordination-insertion mechanism and is expected to produce linear chains with a methoxide group on the initiating chain end.33 These polymerization reactions also converted monomer in high yield to polymer. In this case the molecular weights were again positively correlated with the ratio of catalyst/initiator present, opposite of the expected trend, and higher than expected given the amount of initiator present.

Scheme 3.3 Comparison of the propagation pathways of the phosphorus-based monomers versus lactones. The lability of the –OR sidechain is key to favoring propagation vs. branching steps.

In all cases, the polymers were soluble in THF. Light scattering determined the absolute molecular weights of these polymers to be 10-100 kDa. However, these samples all eluted at times corresponding to Mn < 10 kDa by polystyrene standards.

This reflects the compact nature of the polymer chains, evident in the low hydrodynamic radii (Table 3.1, RH). While this could be due to collapsed chain conformations in THF, the high solubility led us to suspect branching may be occurring. Corroborating evidence for this hypothesis was provided by 31P NMR. Both crude and purified polymer samples showed multiple 31P signals. Multiple phosphorus environments suggested that there were branching points present in the polymer. These could be created by dissociation of a benzyloxy group during the polymerization which would result in chain transfer and a potential branch point. Polymerizations with Al(OMe)Et2 had particularly large secondary peaks. These were found to increase throughout the course of the reaction.

31 Figure 3.2 P NMR spectra of poly(BP) produced with Al(OMe)Et2 (left, Table 3.1, Run 4) and NHC 1 (right, Table 3.1, Run 1)

Figure 3.3 Possible structures for the peaks present in the 31P NMR of poly(BP).

Branch points have been observed with poly(phospholane) polymers by

Penczek.21 In that study, methyoxy phospholane was polymerized with triisopropoxyaluminum catalyst to yield polymers with Mn ranging from 6 kDa to 31 kDa.

The 31P NMR spectra of these polymers had three peaks at δ 0.7, -0.4 and -1.7, with proportions similar to those observed in the poly(BP) polymers. Based on analysis of the relative magnitudes of these three signals throughout the polymerization, they were assigned to four different phosphorus-centered structures, as shown in Figure 3.4.

Figure 3.4 Assignment of three 31P NMR peaks of poly(methoxy phospholane) by Penczek et al. Reprinted with permission from Penczek, S.; Libiszowski, J. Die Makromolekulare Chemie 1988, 189, 1765. Copyright 2003 John Wiley and Sons.

Despite the apparent similarity to the poly(BP) polymer spectrum, the same assignments were not consistent with a HMBC 2D NMR experiment which showed significant correlations between benzyl peaks and the downfield (left-hand) 31P signal at

δ 0.0, ruling out a branch point. It was surprisingly difficult to definitively assign structures to the side peaks, but it was clear that the central peak at δ -0.2 was a main- chain ring opened polymer given its prominence in the carbene-catalyzed polymerization and in low-conversion aliquots of the Al(OMe)Et2 reactions.

Figure 3.5 2D NMR 1H,31P HMBC spectrum of purified poly(BP) produced with 31 Al(OMe)Et2. The P peaks correspond with the benzyl (δ 3.0) and alkylene (δ 4.1) 1H signals to various degrees.

Therefore, despite uncertainty about the assignments of the outer peaks, the magnitude of the central peak was taken as a measure of the linearity or branched nature of the poly(BP). This was calculated as [integral of cental peak]/[sum of all three peak integrals]. To observe the chain topology, the viscosity of polymers was measured

56 using an in-line viscometer coupled to a GPC and light scattering detector. This allowed the construction of Mark-Houwink plots, which relate the intrinsic viscosity (IV) to the molecular weight of a polydisperse polymer sample. For a cyclic or branched polymer, the IV will be lower than that of a linear sample of the same molecular weight.34,35 When the relative magnitude of the central peak in the 31P NMR was plotted against the IV at

MW=50 kDa (for low molecular weight samples, this was a hypothetical IV obtained by extrapolating the linear log(IV) vs. log(MW) relationship), a correlation between the microstructural information in the 31P NMR and the topological information encompassed by the viscosity data was evident. The Al(OMe)Et2 catalyzed-polymerizations, which had large sidepeaks due to extensive transesterification, also had low viscosities of a

MW=50 kDa polymer chain, indicative a high degree of branching. The carbene-only catalyzed polymerizations had smaller sidepeaks and higher viscosities, characteristic of more linear chains. The carbene-catalyzed polymerization that contained an alcohol initiator showed intermediate values of both parameters. This suggests that the addition of an alcohol initiator promoted transesterification and branching.

Figure 3.6 Plot of the fraction of the central peak in the 31P NMR vs. the IV of a 50 kDa sample determined by in-line light scattering and viscometer on GPC instrument. All entries on Table 3.1 are included and the catalyst used is noted. The positive correlation indicates large peaks at δ 0.0 and -0.4, hypothesized to result from transesterification, are indicative of branching in the poly(BP) product.

To further investigate the carbene catalysts for transesterification of the phosphorus centers, tribenzyl phosphate was synthesized. This was added to one equivalent of trimethyl phosphate and a catalytic amount of NHC 1 was added (scheme

3.4). No reaction was observed over the course of 18 h. However, when the trimethyl phosphate was replaced with an alcohol, 1-pyrene butanol, the carbene promoted transesterification to dibenzyl (4-pyrenebutyl) phosphate. This is consistent with Nolan’s studies showing that NHCs are competent organocatalysts for tranesterification of phosphoesters.36

Scheme 3.4 Transesterification catalyzed by NHC 1. In the absence of a free alcohol, the NHC does not promote exchange.

In conclusion, benzyl phospholane was synthesized and could be rapidly polymerized by organometallic or organic catalysts. The resulting polymer could be deprotected to yield a water soluble polymer. However, different catalysts produced different polymer topologies by inducing varying amounts of polymer branching. This was observed in both the polymer microstructure and viscosity. Though the carbene- catalyzed polymerizations had relatively low levels of tranesterification, different endgroups were observed in the MALDI-TOF spectra (Figure 3.10). This, combined with the presence of small side-peaks in the 31P NMR, indicated that these polymerizations were not efficient routes to cyclic polyphosphates or polyphosphoesters. Therefore a phospholane-based monomer with a less-labile sidechain was explored in order to further suppress and hopefully eliminate the tranesterification reaction. By favoring propagation over tranesterification or chain transfer (Scheme 3.3), the polymerization can be controlled to a degree that would allow efficient production of low polydispersity linears or cyclic chains.

3.2.2 Isopropoxy Phospholane Monomer

The zwitterionic ring-opening polymerization of the isopropoxy phospholane iPP8 with N-heterocyclic carbenes (NHCs) 1,3,4,5-tetramethylimidazol-2-ylidene 1 and 1,3- diisopropyl-4,5-dimethylimidazol-2-ylidene 2 in the absence of alcohol initiators is rapid, reaching high conversion within minutes to afford clear, amorphous polymers (Table 3.2, runs 1-5, see also Table 3.3). The less nucleophilic carbene 3 (IMes) showed no conversion over the course of several hours (Table 3.2, entry 6). Analysis of the reaction mixtures by 1H NMR indicated conversion to polymer with no side products and no signals attributable to endgroups. MALDI-TOF MS of the poly(iPP) generated with carbene 2 (Table 3.2, entry 2) yielded ions corresponding to the sodium adducts of cyclic oligomers (Figure 3.8). Minor peaks indicated exchange of an isopropoxy group for a proton. No endgroup signals were observed in either the 1H or 31P NMR spectra.

The molecular weight of poly(iPP) synthesized via ZROP ranged from 50 to 200 kDa with Mw/Mn < 1.3. While the polydispersities are relatively narrow, the molecular weights are significantly higher than that predicted from the initial monomer to NHC ratio

([iPP]0/[NHC]0). The absence of observable endgroups by NMR spectroscopy and the

MALDI spectrum of Figure 3.7 indicate that the ring-opening of iPP generates cyclic structures.

Table 3.2 Representative polymerizations of isopropoxy phospholane (iPP) monomer a

cat. init. [iPP]0 catalyst Initiator solv. time conv b b entry conc conc. M M /M (M) (min) (%) n w n (M) (M) 1 1.0 2 0.020 - - THF 2 78 202,000 1.25

2 1.0 2 0.020 - - THF 2 89 113,000 1.05

3 0.7 2 0.008 - - PhMe 1 65 80,000 1.14

4 1.0 1 0.020 - - THF 10 27 55,000 1.19

5 1.0 1 0.010 BnOH 0.010 THF 12 90 23,000 1.08

6 1.0 3 0.028 - - THF 16 h 0 - -

7c 2.0 DBU/TU 0.10 EtOH 0.020 PhMe 165 82 16,000 1.10

8d neat TBD 1.0e BnOH 1.0e - 20 64 11,000 1.44 a carried out at ambient T, except where noted, DBU = 1,8-Diazabicycloundec-7-ene, TU = 1-(3,5-bis(trifluoromethyl)phenyl)-3-cyclohexylthiourea, TBD = Triazabicyclodecene b determined by light scattering, c reaction conducted at 0°C d became too viscous for stirring e mol percent relative to initial monomer concentration

Since the MALDI only samples lower molecular weight fractions, it is possible that the cyclic poly(iPP)s evident in this spectrum are not representative of the sample

(Mn=113,000 Da), but are a minor fraction of cyclic oligomers generated by intramolecular "backbiting"37 of the growing chains. As the topology of the higher molecular weight fractions can be investigated by comparison of the intrinsic viscosity of linear and cyclic chains,38 we sought to generate linear poly(iPP)s with comparable molecular weights to those of entries 1-4 (Table 3.2).

Figure 3.7: MALDI-TOF MS spectrum of poly(iPP) synthesized with ZROP (table 3.2, entry 2)

A series of linear poly(iPP)s were generated in the presence of alcohol initiators with N-heterocyclic carbene 1 (Table 3.2, entry 5), with DBU/TU (Table 3.2, entry 7), with

TBD (Table 3.2, entry 8), with Sn(Oct)2 or Al(OMe)Et2 (Table 3.3). Linear samples

9 generated with carbene 1 or DBU/TU afforded poly(iPP) with molecular weights Mw ≤

25,000 Da with monomodal and narrow molecular weight distributions (Mw/Mn ≤ 1.1).

However, efforts to generate linear poly(iPP)s with comparable molecular weights to those generated in the zwitterionic ring-opening polymerization (Table 3.2, entries 1-4) were unsuccessful, even with the DBU/TU catalyst system.9 Attempts to obtain higher molecular weight poly(iPP) with the DBU/TU catalyst system by increasing the monomer to initiator ratio gave bimodal molecular weight distributions, where the higher molecular weight fraction did not exhibit a UV-absorption indicative of the pyrene endgroup (Table

3.3, run G). It is likely that at very low alcohol concentrations (i.e. high M/I ratios), the amidine DBU can mediate a competitive nucleophilic zwitterionic polymerization to generate higher molecular weight cyclic (UV-inactive) chains, as recently observed for the ring-opening of lactide with DBU in the absence of alcohol initiators.39 Attempts to generate higher molecular weight linear poly(iPP) with Sn(Oct)2 or Al(OMe)Et2 afforded only low molecular weight polymers with broad molecular weight distributions (Mw/Mn =

1.37 - 2.57), which exhibited multiple resonances in the 31P NMR spectra, consistent with extensive P-OR transesterification.

3.2.3 Mechanistic Studies.

To provide evidence for a nucleophilic transesterification mechanism, model studies were carried out with NHC 2 and 2-chloro-2-oxo-1,3,2-dioxaphospholane. A cooled toluene solution of the chlorophospholane was treated with NHC 2 at -78 °C and slowly warmed to room temperature to afford a white cloudy solution. Isolation of the solid revealed 2-chloroethyl-(1,3-diisopropyl-4,5-dimethyl-imidazolium)-phosphonate 4 in

39% yield and recrystallization from CH2Cl2 afforded X-ray quality crystals (Figure 3.8).

The imidazolium phosphonate 4 was characterized by 1H, 13C, 31P NMR, ESI Mass

Spectrometry and single-crystal X-ray analysis.

Scheme 3.5 Generation of imidazolium chloroethoxy phosphonate and proposed intermediate.

The generation of the imidazolium phosphonate 4 indicates that nucleophilic attack at phosphorous by an NHC is a viable process and indirectly supports a nucleophilic mechanism for phosphoester transesterification (Scheme 3.5). We were unable to observe any intermediates in this process, but a reasonable mechanism for the formation of 4 is the nucleophilic displacement29 of chloride by the NHC, followed by attack of chloride on the phospholane ring in an Arbuzov reaction.40

The solid-state structure of 4 reveals a slightly distorted tetrahedral geometry at phosphorous with a P(1)-C(4) distance of 1.8568(12) Å, terminal P-O distances of

1.4762(9) Å (P(1)-O(3)) and 1.4814(9) Å (P(1)-O(2)), and P-OR distance of 1.6076(9) Å

(P(1)-O(3)). This structure is analogous to that of the NHC-oxophosphorane 2-

(diphenylphosphoryl)-1,3-dimesityl-4,5-dihydro-1H-imidazol-3-ium, recently reported by

Stephan.30 Related NHC P(III) and P(V) compounds are also known.32,41,42

Figure 3.8 X-ray crystal structure of the imidazolium phosphonate 4. Thermal ellipsoids are set to 50% probability. Selected bond distances and angles: P1-C4, 1.8568(12); P1- O3, 1.4762(9); P1-O2, 1.4814(9); P1-O1, 1.6076(9); O3-P1-O2, 122.01(5)°; O1-P1-C4, 100.82(5)°; O3-P1-C4, 106.69(5)°, O2-P1-C4, 107.83(5)°, O3-P1-O1, 106.68(5)°; O2- P1-O1, 110.71(5)°

Shown in Scheme 3.6 is a proposed mechanism for the Zwitterionic Ring

Opening Polymerization (ZROP) of iPP to generate cyclic poly(iPP). Nucleophilic attack of the NHC on iPP would generate a zwitterion that upon addition of further monomer would generate a macrozwitterion that could cyclize to generate the cyclic polymer. In analogy to other zwitterionic ring-opening polymerizations of lactones,11,43-46 it is likely that the initiation step is inefficient such that only a small fraction of carbenes transform to active zwitterions, which propagate rapidly.46 This is consistent with the observation that the molecular weights are much higher than that predicted based on the initial ratio of [iPP]0/[NHC]0. The fast rates and high molecular weights observed imply that propagation is rapid and faster than cyclization47 to liberate the cyclic poly(alkylene phosphate). The relatively narrow polydispersities are likely due to the inability of the liberated carbenes to re-initiate chains, as a consequence of the low initiation efficiency, especially at higher conversions.44,46 The fast rates for the ZROP of iPP with NHCs 2

65 and 3 impeded our initial attempts to investigate the kinetics or evolution of molecular weight with conversion, but further studies are underway to investigate the mechanism of these reactions.

Scheme 3.6: Proposed mechanism for the ZROP of iPP.

The crystal structure of compound 4 indicated that the backbone methylene groups in the monomer are somewhat electrophilic, given that the molecule is ring- opened by chloride. This observation raised questions as to the ultimate cyclization pathway for poly(iPP). If the propagating anion attacks at carbon rather than phosphorus, the resulting polymer would yield a NHC with bound phosphate.

Scheme 3.7 Hypothetical “Arbuzov Cyclization” pathway (top) and standard ZROP pathway (bottom). The former mechanism yields quenched zwitterion 5 as a byproduct.

To investigate whether this mechanism was active, a ZROP polymerization with a high concentration of NHC 2 was conducted (Scheme 3.8). The resulting crude reaction mixture was analyzed with 31P NMR (Fig. 3.15).

Scheme 3.8 Polymerization of iPP with high concentration of carbene

ts-6-78_mecn_p31

0.20 polymer/ oligomer 0.15 peaks

0.10 Normalized Intensity Normalized

0.05 expected region for peak from 5, based on spectra of 4

0 0.66 97.042.30

1 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10 -11 -12 Chemical Shift (ppm) 31 Figure 3.9 P NMR (d3-MeCN) of crude iPP ZROP polymerization at high carbene concentration. Note the absence of any carbene adduct. ESI-MS of the crude reaction mixture detected protonated carbene (m/z=181) but not 5 (m/z=303).

By comparison to the NMR spectra of 4, which showed signals at δ -8.8, it does not appear 5 is present. The protonated carbene is observed in ESI-MS. Combined with the MALDI-TOF (Fig. 3.8), which does not indicate the presence of a –(CH2)2-O-(CH2)2– linkage, this ruled out this alternative cyclization pathway.

In conclusion, the zwitterionic ring-opening polymerization of the cyclic phosphate iPP with NHCs in the absence of alcohols occurs rapidly to generate high molecular weight poly(iPP) with low polydispersity. The poly(alkylene phosphates) generated under these conditions can be entrapped in HEMA crosslinked gels (see chapter 5) implicating that a significant fraction of the high molecular weight poly(iPP) generated under these conditions consist of cyclic macromolecules.

3.3 Experimental Procedures

General Considerations: All polymerizations were conducted in a dry nitrogen glovebox. Tetrahydrofuran (THF) and toluene were distilled from sodium/benzophenone and stored under nitrogen. Dichloromethane (DCM) used for polymerizations was stirred over calcium hydride and distilled. All chemicals were purchased from Sigma

Aldrich unless otherwise specified. Polystyrene calibrated molecular weights were obtained on a Viscotek GPCMax with two Waters columns (300 mm by 7.7 mm) in THF at 35°C at a flow rate of 1 mL/min and Viscotek S3580 refractive index detector.

Monodisperse polystyrene calibrants ranged from Mp = 500 to 275,000. Light-scattering molecular weights were determined on Agilent 1260 infinity SEC pump with two Agilent polypore columns (300mm by 7.7mm) in DMF at 70°C at a flow rate of 0.6 mL/min and

Wyatt Heleos II 8-angle light scattering detector and Wyatt T-rex refractive index detector. NMR data was collected on 300, 400 and 500 MHz Varian instruments.

Synthetic Procedures: 1,3,4,5-tetramethylimidazol-2-ylidene (1) was prepared by synthesizing and deprotonating the imidazolium salt according to the literature.48 1,3- diisopropyl-4,5-dimethylimidazol-2-ylidene (2) was prepared by reducing 1,3-diisopropyl-

4,5-dimethyl-1H-imidazole-2(3H)-thione according to the literature.5 1,3-bis(2,4,6- trimethylphenyl)imidazol-2-ylidene (3) was prepared by deprotonating the 1,3-dimesityl-

1H-imidazol-3-ium chloride purchased from Strem according to the literature.49 1-(3,5- bis(trifluoromethyl)phenyl)-3-cyclohexylthiourea (TU) was synthesized according to the literature.50 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) and 1,5,7- triazabicyclo[4.4.0]undec-5-ene (TBD) were purchased and purified by distillation and sublimation, respectively.

Synthesis of 2-(benzyloxy)-1,3,2-dioxaphospholane 2-oxide (BP) 3.93 g of 2-chloro-

1,3,2-dioxaphospholane 2-oxide (TCI America, FW=147.48) was measured into Schlenk flask. 50 mL of dry THF and stirbar were added. 4.0 mL (1.05 eq) of dry triethylamine

(Aldrich, FW=101.19, d=0.725) and 2.86 mL (1.00 eq.) of anhydrous benzyl alcohol

(Aldrich, FW=108.14, d=1.04) were added sequentially. Heat evolution noted after 1 min. Placed in ice bath for 1 h. Total reaction time was 4 h. White precipitate

(triethylamine hydrochloride) was filtered in glovebox and discarded. THF was removed under vacuum, then product was redissolved in dry toluene and warmed with hands to form a nearly-saturated solution. Stored at -50°C overnight. White crystals formed.

Supernatant was decanted under flowing N2 and the resulting solid was dried under vacuum. The product was loaded into sublimation apparatus and put under vacuum with dry ice/IPA cold trap. (Product melted at approximately 60°C, prior to sublimation/distillation.) The temperature was gradually increased to 165°C and

1 sublimated material was recovered. NMR showed clean BP. H(CDCl3): δ 4.4 (4H m)

31 5.2 (2H d) 7.1-7.5 (5H m) P(CDCl3): δ 18.7. Yield: 3.24 g (55%) 2-(4- methylbenzyloxy)-1,3,2-dioxaphospholane 2-oxide (mBP) was synthesized in the same

1 manner using 4-methylbenzyl alcohol. H(CDCl3): δ 1.9 (3H s) 4.4 (4H m) 5.1 (2H d)

31 6.8 (2H d) 7.1 (2H d) P(CDCl3): δ 18.9. Yield: 52%

Polymerization of BP with TBD 66.0 mg of BP (FW=212.18) was combined with 1.7 mg of benzyl alcohol (FW=108.14) and 0.41 mL dry DCM. 3.5mg TBD (FW=139.2) was added to start the reaction. BP:BnOH = 20:1, TBD = 8 mol %, [BP]0 = 1M. After 30 minutes, the reaction was removed from the box and quenched with two drops of acetic acid. Solvent was removed under vacuum. 1H NMR indicated greater than 95% conversion, and the 31P contained three polymer peaks near 0.

Hydrogenation of poly(BP) 60 mg of poly(BP) was dissolved in approximately 12 mL of dry DCM, 1 mL of anhydrous methanol and 10 mg of Pd/C, (10% wt. Aldrich) was added with a stirbar. The schlenk flask underwent two brief freeze pump thaw cycles, refilling with H2 after each one. The reaction was left stirring with H2 balloon. After 24 h, the balloon was removed and the solution filtered to remove Pd/C. The remaining solution was clear. It was removed under vacuum to give a clear, sticky polymer. The polymer

1 was insoluble in CDCl3, but was soluble in D2O. The H NMR showed no aryl peaks.

The 31P spectra showed several peaks between δ 4 and -2.

Representative Polymerization of BP with NHC 300 mg of BP (FW=214.16) was dissolved in 1.40 g of THF. 4.0 mg of NHC 1 (FW=124.18) was added with stirring in a glovebox. After 4 min, excess 4-nitrophenol was added. [BP]0 = 1.0 M, [1] = 2.3 mol %

The reaction immediately turned yellow and was allowed to stir for 30 minutes. The solvent was removed under vacuum. 1H NMR showed 96% conversion. 1H NMR

31 (CDCl3) of polymer: δ 4.1 (4H m) 5.0 (2H m) 7.2-7.4 (5H br), see Fig 3.2 for P NMR spectra. GPC (LS) = Mn=59 kDa, Mw/Mn = 1.21. Polymerizations with alcohol initiator were conducted in a similar manner with the alcohol introduced prior to the NHC catalyst.

Figure 3.10 MALDI-TOF spectra of poly(BP) sample synthesized with NHC 1. Peak spacing of MW=214 Da is observed, indicating poly(BP) (monomer weight 214 Da). Multiple types of unidentified endgroups are present.

Representative Polymerization of BP with Al(OMe)Et2 200 mg of BP (FW=214.16) was dissolved in THF. 2.2 mg of Al(OMe)Et2 was washed into the reaction with THF to bring the total volume to 0.47 mL. [BP]0 = 2.0 M, [Al(OMe)Et2] = 2.0 mol % After 16 h, the reaction was quenched by exposure to air. 1H NMR showed 100% conversion.

GPC (LS) = Mn=72 kDa, Mw/Mn = 1.23

Synthesis of Tribenzyl Phosphate 1.0 mL (1.57 g, 1 eq.) of phosphorus trichloride was added dropwise to 25 mL of THF containing 3.92 mL (4.09 g, 3.3 eq.) of benzyl alcohol and 5.28 mL (3.83 g, 3.3 eq.) triethylamine at 0°C. After 30 min, the white precipitate

(triethylamine hydrochloride) was filtered off. Solvent removed and product purified via column chromatography (3:2 hexanes:EtOAc) to yield 1.75 g P(OBn)3. This was added directly to 1.1 eq of hydrogen peroxide (30% in H2O). Then vacuum was pulled

1 overnight and resulting material was recrystallized from EtOAc at -30°C. H(CDCl3): δ

31 5.0 (2H d) 7.2-7.4 (5H m) P(CDCl3): δ 0.22.

Transesterification of Model Phosphoesters 80 mg (1 eq., FW=590.60) of tribenzyl phosphate was dissolved in 0.3 mL of THF and 41 uL (1 eq.) of trimethyl phosphate

(FW=362.32, d=1.20) was added. 1.3 mg (0.08 eq.) of NHC 1 was added and the reaction was monitored via 31P NMR. After 18 h, no change occurred. In a separate experiment, 51.0 mg (1 eq.) of tribenzyl phosphate (FW=590.60) was dissolved in 0.5 mL THF with 23.4 mg (1 eq.) 1-pyrene butanol (FW=274.26). 0.7 mg (0.07 eq.) of NHC

1 was added and the reaction was monitored via 31P NMR. After 1 h, 31P integration indicated that 9.3% of the tribenzyl phosphate had been converted to dibenzyl (4- pyrenebutyl) phosphate (δ 4.4) and 2.7% had been converted to benzyl bis(4- pyrenelbutyl) phosphate (δ 3.0).

Synthesis of 2-isoproxy-1,3,2-dioxophospholane 2-oxide:8 2.54 g of 2-choloro-1,3,2- dioxophospholane 2-oxide (TCI America, FW=142.48) was dissolved in 35 mL of tetrahydrofuran. 2.74 mL (0.129 moles) of triethylamine (FW=101.2, d=0.726) was added dropwise at 0°C, followed by 1.40 mL (0.111 moles) of 2-propanol (FW=60.1, d=0.786). After eleven hours the reaction was filtered over celite and solvent removed.

The product distilled under vacuum at 140°C. Yield=1.76 g (59%). Spectra matched that

8 1 31 of literature. H NMR (CDCl3): δ 4.67 (m, 1H), 4.32 (m, 4H), 1.29 (d, 6H) P NMR

(CDCl3): δ 17.3 (s)

Synthesis of imidazolium phosphonate 4: 42.7 mg of 1,3-diisopropyl-4,5- dimethylimidazol-2-ylidene (FW=180.29) was dissolved in toluene and cooled in Schlenk flask to -78°C. A solution of 33.7 mg of 2-chloro-2-oxo-1,3,2-dioxaphospholane (TCI

America, FW=142.48) was added dropwise with stirring. Upon warming to RT a white

73 suspension formed. Solid portion was isolated and dried under vacuum. Yield= 29.9 mg

1 (39%) NMR and MS conducted in acetonitrile. H NMR (CD3CN): δ 1.5 (12H, d), 2.3

13 (6H, s), 3.7 (2H, t), 4.0 (2x, m), 6.4 (2H, br) C NMR (CD3CN): δ 10.4 (s), 20.7 (s), 22.1

31 (s), 44.6 (d JCP=7.6 Hz), 50.4 (s), 51.0 (s), 65.5 (d, JCP=6.1 Hz) P (CD3CN): δ -8.8 (s) m/z: 323.2, 281.1, 239.1 Product was dissolved in minimum amount of anhydrous DCM and sealed in a vial containing diethyl ether. After one week in -10°C freezer, large colorless crystals suitable for X-ray analysis were obtained.

Figure 3.11 ESI-MS for 2-chloroethyl-(1,3-diisopropyl-4,5-dimethyl-imidazolium)- phosphonate (4). Peak at 323.2 corresponds to M+H+, and peaks at 281.1 and 239.1 correspond to loss of one or both isopropyl groups.

Crystal structure determination A prism-like crystal was coated with Paratone-N oil, attached to a Kapton loop, and transferred to a Bruker D8 Venture diffractometer equipped with a Photon 100 CMOS detector. Frames were collected using ω and ψ

74 scans and the unit-cell parameters were refined against all data. Data were integrated and corrected for Lorentz and polarization effects using SAINT 8.27b, and were corrected for absorption effects using SADABS V2012.41 Space-group assignments were based upon systematic absences, E-statistics, agreement factors for equivalent reflections, and successful refinement of the structure. The structure was solved by direct methods and expanded through successive difference Fourier maps using

SHELXS-97. It was refined against all data using the SHELXTL-2013 software package.51,52 Hydrogen atoms were inserted at idealized positions and refined using a riding model with an isotropic thermal parameter 1.2 or 1.5 times that of the attached carbon. Thermal parameters for all non-hydrogen atoms were refined anisotropically.

Details regarding the data quality and a summary of the residual values of the refinement are shown here.

Crystallographic data for compound 4a Compound 4

Empirical formula C13H24ClN2O3P Formula weight, g mol–1 322.76 Temperature, K 100(2) Crystal system Monoclinic

Space group P21/c a, Å 8.437(1) b, Å 16.664(2) c, Å 11.784(1) , º 90 , º 91.11(1) , º 90 Volume, Å3 1656.4(3) Z 4 Density (calculated), g cm–3 1.294 Absorption coefficient, mm–1 0.34 F(000) 688 Crystal size, mm3 0.35 x 0.35 x 0.2 Theta range, º 2.1 to 27.9 11 ≤ h ≤ 10 Index ranges 21 ≤ k ≤ 21 15 ≤ l ≤ 13 Reflections collected/unique 32001 / 3932 Completeness to theta max, % 99.7 Max. and min. transmission 0.704 and 0.746 Data / restraints / parameters 3932 / 0 / 187 Goodness-of-fit on F2 1.05 R = 0.030 Final R indices [I>2sigma(I)]b 1 wR2 = 0. 039 R = 0.034 R indices (all data)b 1 wR2 = 0.078 Largest diff. peak and hole, e Å–3 0.38 and 0.24 aObtained with monochromated Mo Kα (λ = 0.71073 Å) radiation b 2 2 2 2 2 1/2 R1 = ||Fo| – | Fc||/|Fo|, wR2 = [w(Fo – Fc ) /(Fo ) ]

ZROP Polymerization iPP with NHC 2: 10.9 mg of 2 (FW=180.29) was dissolved in

3.02 mL of THF and added to 502 mg of iPP (FW=166.11) with stirring. [iPP]0:[2]=50:1.

[iPP]0=1M Quenched with approximately 30 mg 4-nitrophenol after 2 minutes. After 30 minutes, solvent removed under vacuum. NMR showed 77% conversion. GPC (LS):

Mn=144.3 kDa, Mw/Mn = 1.07. ZROP polymerization with carbenes 1 and 3 were conducted in a similar manner.

1 31 Figure 3.12 H and P NMR (CDCl3) for cyclic poly(iPP)

Cyclic poly(iPP) (RI detector)

0 10 20 30 40

Retention Time (min)

Figure 3.13 Representative GPC chromatogram for ZROP polymerization of iPP with carbene 2.

Polymerization of iPP with 1 and BnOH: 1.00g of iPP (FW=166.11) was added into a vial along with 6.5 mg of benzyl alcohol (FW=108.14) via a stock solution in THF. A solution of 7.8 mg of 1 (FW=124.2) in THF was added to the stirring monomer. The reaction was a clear rose color. After 12 minutes, approximately 20 mg of 4-nitrophenol were added to the reaction and allowed to stir for one hour. Solvent was removed under vacuum. NMR showed 90% conversion. GPC sample (LS): 23320, Mw/Mn = 1.08

Polymerization of iPP with EtOH, DBU/TU: 45.4 mg TU (FW=370.36) were added to

407 mg of iPP (FW=166.11) in 1.22 mL of toluene. 1.4 µL of ethanol (FW=46.07, d=0.789), followed by 18.6 mg DBU (FW=152.24). Reaction proceeded at 0°C for 4 hours, 20 min. Quenched with acetic acid outside of glovebox. NMR showed 82%

78 conversion. GPC sample (LS): Mn=16490, Mw/Mn = 1.10. Linear polymerization with sparteine was conducted by substituting (-)-sparteine for DBU in this procedure.

1 31 Figure 3.14: H and P NMR (CDCl3) for linear poly(iPP)

Linear poly(iPP) (RI detector)

0 10 20 30 40 retention time (min)

Figure 3.15 Representative GPC chromatogram for linear polymerization of iPP with TU/DBU. See Table 3.2 for polymerization conditions and results.

Polymerization of iPP with BnOH, TBD: 11.2 mg of benzyl alcohol was added to 1.66 g of iPP. 13.5 mg of TBD was added with stirring in a vial. After 20 minutes the clear reaction became too viscous for magnetic stirbar. NMR showed 64% conversion and the reaction was added to ether to precipitate the polymer. GPC sample (PS calibration): Mn=7416, Mw/Mn = 1.18

Purification poly(iPP): The solid polymer obtained from THF was redissolved in methylene chloride and precipitated twice in diethyl ether at 0°C. The sample was dissolved in methylene chloride and layered with water and one drop of triethylamine.

This was mixed vigorously with shaking and then separated via centrifuge. The organic layer was isolated and washed twice more with DI water and separated via centrifuge.

The organic layer was dried with sodium sulfate, filtered, and solvent removed under

80 vacuum to give a colorless, amorphous polymer. NMR showed monomer, catalyst(s)

1 and quenching agent had been removed. H NMR (CDCl3): δ 4.7 (oct, 1H), 4.2 (t, 4H),

31 1.3 (d, 6H) P (CDCl3): δ -1.1

Figure 3.16: Heat flow (exo up) from Differential Scanning Calorimetry of poly(iPP) generated from carbene 2 (Mn=90 kDa, Mw/Mn=1.09). Tg = -45°C.

Sample Preparation: MALDI-TOF MS: Polymer samples (for example, Table 3.3, entry

2) were analyzed by an Applied Biosystems Sciex TF4800 MALDI-TOF at the Molecular

Foundry, a part of Lawrence Berkeley National Laboratory. Samples were prepared by dissolving combining a 20 mg/mL solution of dithranol (Aldrich) with 10 mg/mL polymer and 0.1 M NaI (Aldrich) in a 10:10:1 ratio.

Polymerization of iPP with Sn(Oct)2: 301 mg of iPP was added to vial with septa.

Stock solutions of pyrene butanol and tin (II) octoate in THF were added to give appropriate concentrations. The reaction was heated to 40°C in an oil bath. NMR

81 showed 65% conversion with multiple 31P polymer peaks. GPC sample (PS calibration):

Mn=1442 PDI=1.37

Polymerization of iPP with Al(OMe)Et2: 402 mg of iPP and THF were added to a vial.

5.6mg of Al(OMe)Et2 was added via stock solution. After 1.5 hrs the reaction was removed from the glovebox and quenched with a drop of acetic acid. Solvent removed under vacuum and product dialyzed in MeOH. NMR showed 95% conversion with

31 multiple P polymer peaks. GPC sample (PS calibration): Mn=3949 PDI=1.71

The molecular weights reported in Table 3.3 were determined by light scattering for higher molecular weight samples, but the low refractive index increment (dn/dc = 0.030 mL/g in THF and 0.0167 mL/g in DMF) caused insufficient light scattering signal for low molecular weight samples. Lower molecular weight samples are reported relative to polystyrene standards in THF, which underestimated the actual weight by a factor of at least two.

Table 3.3 Polymerizations of iPP

Note: numbered runs appear in Table 3.2. The lettered runs are displayed exclusively in this chart. cat. init. tem [iPP] solven time quench. PDI M (LS) run cat conc. initiator conc. p conv. M (PS) n PDI (LS) t (min) agent n (PS) 0 (M) (M) ('C)

1 1.0 2 0.020 - - THF RT 2 4NP 78 - - 202400 1.25

A 1.0 2 0.020 - - THF RT 2 4NP 77 - - 144300 1.07

2 1.0 2 0.020 - - THF RT 2 4NP 89 16207 1.34 112742 1.05

C 0.5 2 0.010 - - THF RT 0.5 4NP 78 21152 1.16 78415 1.02

3 0.7 2 0.0077 - - PhMe RT 1 4NP 65 16055 1.12 80487 1.14

a D 1.0 2 0.020 PyBuOH 0.027 PhMe RT 2 CS2 92 2762 1.79 - -

4 1.0 1 0.020 - - THF RT 10 4NP 27 10980 1.38 54733 1.19

5 1.0 1 0.020 BnOH 0.010 THF RT 12 4NP 90 8178 1.32 23320 1.08

16 6 1.0 3 0.028 - - THF RT 4NP 0 - - - - hrs 0.10/0. E 2.0 DBU/TU EtOH 0.020 PhMe 0 180 PhCO2H 79 - - 23480 1.78 10 0.10/0. 7 2.0 DBU/TU EtOH 0.020 PhMe 0 165 PhCO2H 82 2529 1.43 16490 1.10 10 0.10/0. F 2.0 DBU/TU PyBuOH 0.020 PhMe 0 165 PhCO2H 90 3821 1.45 - - 10

b 0.10/0. G 2.0 DBU/TU PyBuOH 0.004 PhMe 0 180 PhCO2H 75 8194 1.23 - - 10 sparteine/ 0.05/0. H 1.0 PyBuOH 0.005 PhMe 0 30 AcOH 0 - - - - TU 05

8c neat TBD 1.0* BnOH 1.0* - RT 20 AcOH 64 7416 1.18 10841 1.44

12 Id 1.0 Sn(Oct) 0.025 PyBuOH 0.010 THF 40 air 65 1442 1.37 - - 2 hrs

d J 1.0 Al(OMe)Et2 0.020 - - THF RT 90 AcOH 95 3949 1.71 4988 2.57

*=mol percent relative to initial monomer concentration a: polymer precipitated from solution b: high MW peak UV inactive c: became too viscous for stirring d: multiple 31P NMR peaks in product abbreviations: DBU 1,8-Diazabicycloundec-7-ene TU=1- (3,5-bis(trifluoromethyl)phenyl)-3-cyclohexylthiourea TBD=Triazabicyclodecene 4NP=4-nitrophenol PS=relative to polystyrene standards in tetrahydrofuran LS=determined by light scattering

3.4 References

1. Kamber, N. E.; Jeong, W.; Waymouth, R. M.; Pratt, R. C.; Lohmeijer, B. G. G.;

Hedrick, J. L. Chem. Rev. 2007, 107, 5813.

2. Kiesewetter, M. K.; Shin, E. J.; Hedrick, J. L.; Waymouth, R. M. Macromolecules

2010, 43, 2093.

3. Fevre, M.; Pinaud, J.; Gnanou, Y.; Vignolle, J.; Taton, D. Chemical Society Reviews

2013, 42, 2142.

4. Tempelaar, S.; Mespouille, L.; Coulembier, O.; Dubois, P.; Dove, A. P. Chemical

Society Reviews 2013, 42, 1312.

5. Lohmeijer, B. G. G.; Dubois, G.; Leibfarth, F.; Pratt, R. C.; Nederberg, F.; Nelson, A.;

Waymouth, R. M.; Wade, C.; Hedrick, J. L. Organic Letters 2006, 8, 4683.

6. Brown, H. A.; Chang, Y. A.; Waymouth, R. M. Journal of the American Chemical

Society 2013, 135, 18738.

7. Rodriguez, M.; Marrot, S.; Kato, T.; Stérin, S.; Fleury, E.; Baceiredo, A. Journal of

Organometallic Chemistry 2007, 692, 705.

8. Iwasaki, Y.; Yamaguchi, E. Macromolecules 2010, 43, 2664.

9. Clément, B.; Grignard, B.; Koole, L.; Jérôme, C.; Lecomte, P. Macromolecules 2012,

45, 4476.

10. Zhang, S.; Wang, H.; Shen, Y.; Zhang, F.; Seetho, K.; Zou, J.; Taylor, J.-S. A.;

Dove, A. P.; Wooley, K. L. Macromolecules 2013, 46, 5141.

11. Brown, H. A.; Waymouth, R. M. Accounts of Chemical Research 2013, 46, 2585.

12. Johnston, D. S. Adv. Polym. Sci. 1982, 42, 51.

13. Suzuki, M. In Encyclopedia of Polymer Science and Technology; 4th ed.; John

Wiley & Sons, Inc.: New York, 2013.

14. Cyclic Polymers; 2nd ed.; Semlyen, J. A., Ed.; Kluwer: Dordrecht, 2000.

15. Hoskins, J. N.; Grayson, S. M. Polymer Chemistry 2011, 2, 289.

16. Topological Polymer Chemistry: Progress of cyclic polymers in synthesis, properties and functions; Tezuka, Y., Ed.; World Scientific Publishing: Singapore, 2013.

17. Jia, Z. F.; Monteiro, M. J. J. Polym. Sci., Part A: Polym. Chem. 2012, 50, 2085.

18. Kricheldorf, H. R. Journal of Polymer Science Part a-Polymer Chemistry 2010, 48,

251.

19. Klosinski, P.; Penczek, S. Macromolecules 1983, 16, 316.

20. Penczek, S.; Biela, T.; Klosinski, P.; Lapienis, G. Makromolekulare

Chemie.Macromolecular Symposia 1986, 6, 123.

21. Penczek, S.; Libiszowski, J. Die Makromolekulare Chemie 1988, 189, 1765.

22. Wang, Y.-C.; Yuan, Y.-Y.; Du, J.-Z.; Yang, X.-Z.; Wang, J. Macromolecular

Bioscience 2009, 9, 1154.

23. Du, J.-Z.; Du, X.-J.; Mao, C.-Q.; Wang, J. Journal of the American Chemical Society

2011, 133, 17560.

24. Zhao, Z.; Wang, J.; Mao, H. Q.; Leong, K. W. Adv. Drug Deliv. Rev. 2003, 55, 483.

25. Mao, H. Q.; Leong, K. W. In Advances in Genetics; Non-Viral Vectors for Gene

Therapy, Second Edition: Part 1; Leaf Huang, M.-C. H. a. E. W., Ed.; Academic Press:

2005; Vol. 53, p 275.

26. Wu, X. L.; Kim, J. H.; Koo, H.; Bae, S. M.; Shin, H.; Kim, M. S.; Lee, B.-H.; Park, R.-

W.; Kim, I.-S.; Choi, K.; Kwon, I. C.; Kim, K.; Lee, D. S. Bioconjugate chemistry 2010,

21, 208.

27. Zhang, S.; Li, A.; Zou, J.; Lin, L. Y.; Wooley, K. L. ACS Macro Letters 2012, 1, 328.

28. Steinbach, T.; Ritz, S.; Wurm, F. R. ACS Macro Letters 2014, 3, 244.

29. Admiraal, S. J.; Herschlag, D. Journal of the American Chemical Society 1999, 121,

5837.

30. Holthausen, M. H.; Mehta, M.; Stephan, D. W. Angewandte Chemie International

Edition 2014, 53, 6538.

31. Dove, A. P. In Handbook of Ring-Opening Polymerization; Wiley-VCH Verlag GmbH

& Co. KGaA: 2009, p 357.

32. Brown, H. A.; Xiong, S.; Medvedev, G. A.; Chang, Y. A.; Abu-Omar, M. M.;

Caruthers, J. M.; Waymouth, R. M. Macromolecules 2014, 47, 2955.

33. Duda, A.; Florjanczyk, Z.; Hofman, A.; Slomkowski, S.; Penczek, S. Macromolecules

1990, 23, 1640.

34. Graessley, W. W.; Mittelhauser, H. M. Journal of Polymer Science Part A-2:

Polymer Physics 1967, 5, 431.

35. Burchard, W. In Cyclic Polymers; Semlyen, J. A., Ed.; Springer Netherlands: 1986, p

43.

36. Singh, R.; Nolan, S. P. Chemical Communications 2005, 5456.

37. Penczek, S.; Biela, T.; Duda, A. Macromolecular Rapid Communications 2000, 21,

941.

38. Roovers, J. In Cyclic Polymers; Second ed.; Semlyen, J. A., Ed.; Kluwer Academic

Publishers: Dordrecht, 2000, p 347.

39. Brown, H. A.; De Crisci, A. G.; Hedrick, J. L.; Waymouth, R. M. ACS Macro Letters

2012, 1, 1113.

40. Bhattacharya, A. K.; Thyagarajan, G. Chem. Rev. 1981, 81, 415.

41. ; Bruker AXS Inc.: Madison, WI, 2007.

42. Abdellah, I.; Lepetit, C.; Canac, Y.; Duhayon, C.; Chauvin, R. Chemistry – A

European Journal 2010, 16, 13095.

43. Zhang, X.; Waymouth, R. M. ACS Macro Letters 2014, 3, 1024.

44. Brown, H. A.; Xiong, S. L.; Medvedev, G. A.; Chang, Y. A.; Abu-Omar, M. M.;

Caruthers, J. M.; Waymouth, R. M. Macromolecules 2014, 47, 2955.

45. Acharya, A. K.; Chang, Y. A.; Jones, G. O.; Rice, J. E.; Hedrick, J. L.; Horn, H. W.;

Waymouth, R. M. Journal of Physical Chemistry B 2014, 118, 6553.

46. Jeong, W.; Shin, E. J.; Culkin, D. A.; Hedrick, J. L.; Waymouth, R. M. Journal of the

American Chemical Society 2009, 131, 4884.

47. FTNT, Cyclization by an Arbuzov reaction at the imidazolium phosphonate terminus is also possible, but control experiments at high concentrations of NHC 2 did not provide any evidence for an isopropoxy imidazolium phosphonate analogous to 4.

48. Kamber, N. E.; Jeong, W.; Gonzalez, S.; Hedrick, J. L.; Waymouth, R. M.

Macromolecules 2009, 42, 1634.

49. Jeong, W.; Hedrick, J. L.; Waymouth, R. M. Journal of the American Chemical

Society 2007, 129, 8414.

50. Pratt, R. C.; Lohmeijer, B. G. G.; Long, D. A.; Lundberg, P. N. P.; Dove, A. P.; Li, H.;

Wade, C. G.; Waymouth, R. M.; Hedrick, J. L. Macromolecules 2006, 39, 7863.

51. Sheldrick, G. SHELXL-97, program for crystal structure refinement; University of

Göttingen, 1997.

52. Sheldrick, G. M. Acta Cryst. Sect. A 2008, 64, 112.

CHAPTER 4

STIMULI-RESPONSIVE ORGANOGELS

4.1 Introduction

Organogels, the analogue of hydrogels in organic solvent, are three dimensional network solids that contain an immobilized liquid solvent phase.1,2 Organogels are composed of either small molecule gelators which interact non-covalently or crosslinked polymers. Unsurprisingly, the nature of the solvent used can have a dramatic impact on gel formation and properties. Thermally responsive organogels have been synthesized based on LCST phase transition of certain polymers in organic solvents.3 Different stimuli have been utilized including mechanical force4, light5, or redox reactions6 in order to effect formation or dissolution.

Organogels have been used for a variety of applications including cosmetics and drug delivery7; however we were interested in developing them as platforms for trapping polymer substrates (see chapter 5). This application required the ability to (1) tune the hydrophilicity of the gel and (2) degrade it in a controlled manner. Custom polymer- derived organogels were designed for each of these objectives.

2-Hydroxyethyl methacrylate (HEMA) is a common monomer used to create hydrogels.8 Though poly(HEMA) is not soluble in water it will swell significantly.9 It was suspected that a crosslinked HEMA network could be protected with a simple tetrahydropyran (THP) group to preclude aqueous swelling. The use of this monomer,

HEMA-THP, in crosslinked particles has been reported in a patent.10 Switchable organo/hydro-gels have been reported before for small molecule gelators,11 but we sought to develop and characterize a switching method for our HEMA-THP gels.

To address the second objective, triggered degradation, disulfide bonds that could be reductively cleaved were utilized. Placement of these bonds within the network crosslinks would allow reductively-triggered degradation of the material. This strategy

91 has been employed by a number of groups to degrade hydrogels in water.12,13 We sought to transition this methodology to organogels for the desired application.

An additional attempt to achieve this goal was based around transitioning a class of dynamic hydrogels developed in the Waymouth lab14 to organogels. These gels are composed of a triblock polymer with pendant cyclic disulfides. Upon addition of a dithiol crosslinker, the polymer forms a gel at 10 wt. % in water. The mechanism of this gelation remains under investigation. It is thought, however, that the thiol initiates a cascade of disulfide ring-opening. Due to the dynamic nature of the disulfide bond, the material can deform under stress and exhibits self-healing. Attempts to reproduce the gelation and self-healing in non-aqueous solvents were only marginally successful.

Scheme 4.1 Formation of dynamic hydrogel based on disulfide exchange of asparagusic acid units. Addition of maleimide “caps” free thiols, preventing exchange. Adapted from reference 14.

4.2 Results and Discussion

4.2.1 Acid-Induced Transition from Organogel to Hydrogel

The THP-protected HEMA monomer polymerized readily under typical free- radical conditions. Poly(ethylene glycol) (PEG) was chosen as a crosslink material that would be compatible with both aqueous and organic solvents. Because many of the deprotection conditions employed for subsequent removal of the THP group were acidic, we sought to avoid ester linkages that might hydrolyze. Therefore acrylamide groups were used as reactive endgroups for the PEG crosslinker. With 0.7 mol % crosslinker (2 wt. % PEG) the protected gel was clear and rubbery. It swelled 324% by mass in THF and to a lesser degree in other organic solvents but did not absorb any water after soaking for several days.

Various deprotection conditions have been reported for the removal of THP groups,15 however the hydrophobicity of the protected gel prevented many of the required reagents from penetrating the material. For instance, gels swollen in ethanol and added directly to aqueous trifluoroacetic acid (TFA) solutions did result in deprotection of the gels. An ethanol-swollen gel soaked in TFA solution pH=1 for 12 h showed a 34% dry weight reduction in mass (theoretical mass loss for complete deprotection=39%) and a 62% mass swelling in water (non-protected HEMA gel swelling=77%). However the rate of the transformation was poorly reproducible due to varying ratios of water to ethanol present throughout the material. Thus primarily non- aqueous conditions were used: 90:10 MeCN:H2O with 1% or 0.1% (v/v) TFA.

Deprotection of the gel was monitored by measuring the weight loss, FT-IR, and the aqueous swelling. These were compared to a hydrogel sample prepared directly from HEMA to estimate the degree of deprotection. The deprotected gel remained homogenous and structurally intact but was much more brittle and less elastic than the

93 protected gel. FT-IR integration of the gels showed the appearance of a hydroxyl stretch after deprotection (Figure 4.1). Using the integration of this absorption peak and that of the ester carbonyl stretch as a reference, it appeared that the deprotection in 0.1%

TFA/MeCN proceeds in a zero-order fashion (Table 4.1).

Figure 4.1 FT-IR of HEMA-THP gel before (green) and after (blue) deprotection. The appearance of –OH stretch is visible at 3390 cm-1

Table 4.1 Deprotection results under TFA/MeCN conditions

Conditions Time (h) FT-IR ratio H2O swelling (%) Pure HEMA - 1.60 77.9 Protected Gel - 0.03 none

0.1% TFA 14 0.61 0.1% TFA 18 0.98 41.2 0.1% TFA 24 1.32

1.0% TFA 18 1.62 FT-IR ratio calculated as the integration of the –OH peak at 3390 cm-1 divided by the integration -1 of the C=O peak at 1725 cm . H2O swelling equal to 100*([swollen mass]-[dry mass])/[dry mass] 94

4.2.2 Reduction-Induced Degradation of Organogel

The design of degradable organogels was based on a disulfide, bisacrylate molecule, bis(2-methacryloxyethyl) disulfide (DSDMA) which has been reported in the literature for degradable aqueous micelles and hydrogels.13,16 These aqueous systems have drawn interest for drug delivery and tissue regeneration.17,18 When this crosslinker was polymerized in a 50:50 (by weight) mixture of butyl acrylate (BA) and HEMA, along with minimal amounts of DMF, the radical initiator AIBN and, optionally, a polymer substrate, a gel was formed when heated at 75°C for 30 min under Ar atmosphere. The gel was transparent and homogenous. Importantly, it swelled in all organic solvents tested: methanol, acetone, THF and DCM.

Figure 4.2 HEMA/BA gel made with DSDMA crosslinker. The gel on the left contains no substrate; the gel on the right contains 20 wt. % polyphospholane substrate.

Reductive degradation of the gel was carried out using dithiothreitol (DTT).

Model studies with small molecule and polymeric disulfides showed that in DCM, unlike aqueous systems, DTT reduction of the disulfide required catalytic amount of a base.

Triethylamine (TEA) was chosen as a non-nucleophilic base that would not degrade polymer substrates. Its low boiling point also allowed easy removal following the reaction. With these reagents, a 0.1 M DTT solution was capable of cleaving all

95 disulfides in a material with as low as 0.1% (v/v) TEA in 24 h. Notably, no degradation of the disulfide gels was observed in control studies without the added TEA, meaning that the degradation process could be considered responsive to a TEA trigger.

Although the overall hydrophilicity could be easily controlled by adjusting

HEMA:BA ratio of the gel, it was critical that the crosslinker have a favorable energy of mixing (χ) with potential polymer subjects to maximize threading (see Chapter 5). For example, it was found that poly(valerolactone), a semi-crystalline polymer, did not form homogenous gels with the above system.

Scheme 4.2: Synthesis of α,ω-bis(acrylate) poly(valerolactone) with internal disulfide

Therefore, a disulfide crosslinker which contained poly(valerolactone) blocks was synthesized by initiating a valerolactone polymerization from 2-hydroxyethyl disulfide.

The chain ends were then functionalized with acrylates to create a polymer disulfide crosslinker with a favorable energy of mixing with valerolactone. Gels made in this manner were much less rigid than the previous generation of gels. However, they still remained intact when submerged in DCM for 24 h (Figure 4.3). Dissolution under the aforementioned reduction conditions was achieved in less than 1 h.

Figure 4.3 Gel made from HEMA and PVL-derived disulfide crosslinker with PVL substrate swollen in DCM.

4.2.3 Organogels Synthesized from 1,2-Dithiolanes

The dynamic disulfide bond allowed for the construction of dynamic, healable hydrogels.14 However, the lower molecular weight PEG triblocks (Scheme 4.3) were insoluble in DCM, methanol and acetone at 25 wt. %. By lengthening the PEG segment to Mn=8000, methanol solubility was imparted. A dithiol crosslinker, 2,2′-

(ethylenedioxy)diethanethiol, was added. A cloudy gel formed in several minutes, as indicated by inversion of the vial. However, the gels readily dissolved upon the addition of DCM.

Scheme 4.3 Triblock polymers tested for gelation

Dynamic hydrogels formed from these reagents are thought to operate via a constant exchange of the disulfide bond via attack by free thiol. In organic solvents, it appears the ring opening equilibrium of the cyclic disulfide is lower, such that a free thiol will exchange with a neighboring disulfide bond to reform the cyclic disulfide. This eliminates crosslinks and leads to gel dissolution. In order to prevent this mechanism, we sought to oxidize all thiols to preclude exchange and maintain crosslinks in the presence of excess solution.

To oxidize the organogels, the triblock polymer and dithiol crosslinker were combined, as above. Then a solution of iodine was applied to the surface of the gel and allowed to diffuse inward (Figure 4.4). The brown iodine solution allows this process to 98 be monitored visually. After the gel is fully colored, it can be submerged in DCM without dissolution for up to a day. Evidently, despite the oxidizing conditions, slow disulfide exchange can still occur.

Figure 4.4 Oxidation of disulfide network. After formation of the organogel in a minimum amount of methanol (left), a methanolic solution of iodide is applied (middle) and allowed to diffuse through the gel (middle). After treatment, the gel is metastable in DCM (right).

The gel can be dissolved by reduction of the disulfide crosslinks. For instance, in an aqueous solution of sodium borohydride the gel is completely dissolved in one minute

(Figure 4.5).

Figure 4.5 Reduction of the “permanent” gel. The oxidized gel (left) is submerged in an

aqeous solution of NaBH4 (middle). After two minutes the gel is completely dissolved.

4.3 Experimental Procedures

300, 400 and 500 MHz Varian instruments. FT-IR spectra were obtained from gels dried at 60°C overnight in a vacuum oven. The Vertex 70 spectrophotometer was fitted with a diamond ATR unit. Gel samples were prepared using a razor blade to cut cross sections and spectra were acquired in triplicate.

Synthesis of 2-tetrahydropyranyloxyethyl methacrylate monomer (HEMA-THP):

Adapted from reference.19 3.69 g of 2-hydroxyethyl methacrylate (Aldrich, FW=130.14) was dissolved in 15 mL diethyl ether. 0.480 g of p-toluenesulfonic acid (Aldrich,

FW=172.20) was added and the solution was cooled to 0°C. 3.69 mL of dihydropyran

(Aldrich, FW=84.12, d=0.922) was added dropwise and the reaction was stirred for 3.5 hrs. Solvent was removed by vacuum and the product was purified by column chromatography (4:1 hexanes:ethyl acetate). 3.78 g (62% yield) isolated as a clear

1 liquid. H NMR (CDCl3): δ 6.12 (s, 1H), 5.56 (s, 1H), 4.66 (t, 1H), 4.32 (t, 1H), 3.94 (m,

2H), 3.94 (m, 1H), 3.68 (m, 1H), 3.50 (m, 1H), 1.95 (s, 3H), 1.81 (m, 1H), 1.70 (m, 1H),

1.59 (m, 2H), 1.52 (m, 2H).

100

Synthesis of diacrylamide polyethylene glycol crosslinker: 163.7 mg of diaminopolyethylene glycol, Mn=3680, was suspended in 30 mL of tetrahydrofuran. 37.2

µL of triethylamine (Aldrich, FW=101.19, d=0.725) and 21.6 µL of acryloyl chloride

(FW=90.51, d=1.12) were added and the reaction was allowed to stir for 3 days. The tetrahydrofuran was then removed under vacuum and the residual polymer dissolved and dialyzed in methanol (MWCO=3.5 kDa). The purified methanol solution was passed through celite filter and washed with warm methanol. The filtrate was then concentrated to yield 149 mg of product. NMR showed 90% endgroup functionalization and the primary peak series in MALDI was the functionalized polymer.

Example procedure for formation of protected gel: 10 mg of PEG-diacrylamide were dissolved in 200 µL acetonitrile and added to 490 mg of HEMA-THP. The solution was placed on a vortex mixer for 30 minutes and then 7.6 mg of AIBN was added and dissolved. The mixture was transferred to a plastic vial with rubber septum and purged with nitrogen for ten minutes. The reaction was started by placing the sealed vial in a

75°C bath for thirty minutes. The resulting gel was removed from the vial and soaked in methanol for several hours to removed unreacted components, then dried overnight in a

65°C vacuum oven.

Example procedure for gel deprotection: The THP-protected network was placed in a vial with stirbar and 4.5mL of 0.1% (v/v) trifluoroacetic acid in acetonitrile was added followed by 0.5 mL of deionized water. The sealed vial was allowed to stir gently for 24 hours. Gel was then removed and soaked in methanol for several hours before being dried overnight in a 65°C vacuum oven. 101

Synthesis of bis(2-methacryloxyethyl) disulfide (DSDMA): Preparation was adapted from reference.20 5.0 g (1 eq.) 2-Hydroxyethyl disulfide (Aldrich, FW= 154.25) was dissolved in 50 mL of DCM and 50 mL of chloroform. 13.5 mL (3 eq.) of trimethylamine was added and the solution was cooled to 0°C. 10.6 mL (3 eq.) of acryloyl chloride were added dropwise. Solution was warmed to RT. After 20 h, solution was washed with 2x

DI water, 2x 0.1 M NaHCO3, 1x sat. NaCl. Solution dried over MgSO4 and solvent was removed under vacuum to give 5.13 g of a yellow liquid. Purified with column chromatography (EtOAc:Hexanes) to give 2.26g (28% yield) of a yellow liquid. NMR

1 showed clean product. H NMR (CDCl3): δ 6.41 (m, 1H), 6.10 (m, 1H), 5.84 m, 1H), 4.40

(m, 2H), 2.95 (m, 2H).

Synthesis of poly(valerolactone) disulfide α,ω-diacrylate crosslinker (XL-2): 36 mg of bis-2-hydroxyethyl disulfide (Aldrich, FW=154.25) and 1.17 g of δ-valerolactone

(distilled from CaH2, FW=100.12) were dissolved in 11.6 mL of DCM. 16.2 mg of TBD was added and stirred for 75 minutes, at which point it was removed from the glovebox and quenched with benzoic acid. ([monomer]:[initiator]:[catalyst]=100:2:1) The product was purified via dialysis (MWCO=1 kDa) in methanol for 18 h, 700 mg isolated (59%

1 yield). GPC (PS calibration): Mn=7110, PDI=1.18. H NMR: Mn=3630 g/mol. The product was then dissolved in 10 mL THF and allowed to stir over molecular sieves, then filtered and diluted to 30 mL THF. Pyridine (anhydrous, FW=79.1, d=0.982) and acryloyl chloride (FW=90.51, d=1.12) were added at 0°C under N2. This was allowed to react for

24 h, during which a brown-yellow precipitate formed. The reaction was concentrated under vacuum and dialyzed (MWCO= 1 kDa) against methanol for 24 h. The solvent

102 was removed under vacuum to yield 672 mg of a white polymer product. GPC (PS calib.): Mn=7250 g/mol, PDI=1.15.

Synthesis of disulfide gels: 70 mg of 2-hydroxyethyl methacrylate (distilled,

FW=130.14, d=1.07), 70 mg of butyl acrylate (Aldrich, FW=128.17, d=0.894) were combined with a disulfide crosslinker (either 20 mg of DSDMA or 80 mg of PVL-based disulfide xlinker), 2.0 mg azobisisobutyronitrile (recrystallized, FW=164.21), 400 µL DMF and (optionally) 40 mg of polymer substrate. This was vortexed for 30 min and then sparged with Ar gas for 2 min. This was sealed and heated to 75°C for 30 min. The resulting clear gel was swollen in DCM for 24 hours without loss of structural integrity.

Reduction and dissolution of disulfide gels: Gel was submerged in 5 mL DCM, 0.1

M DTT, 0.1% (v/v) triethylamine (Aldrich). Gel appeared to be completely dissolved in 8 h, after 24 h the solution was filtered through 0.45 um syringe filter without difficulty.

Synthesis of organogels based on asparagusic acid triblock polymer: 30.7 mg of triblock polymer (x=10, y=182, see scheme 4.1) were dissolved in 92 mg of methanol with vortex mixer. 2.4 mg of 2,2′-(Ethylenedioxy)diethanethiol (5 eq., FW=182.30) were added in 17 mg of methanol. This was mixed and allowed to sit for 5 h. 3.3 mg of I2 (5 eq., FW=253.8) were dissolved in 50 mg of methanol and the solution was applied to the surface of the gel. After 5 h, the entire gel had turned brown and was stable against dissolution in DCM for 6 h or indefinitely in methanol.

Reduction and dissolution of oxidized organogels: The gel, treated with iodine, was placed in a vial and bubbling solution of NaBH4 in deionized water was added. The gel quickly turned colorless and dissolved.

103

4.4 References

1. Vintiloiu, A.; Leroux, J.-C. J. Controlled Release 2008, 125, 179.

2. van Esch, J. H.; Feringa, B. L. Angew. Chem., Int. Ed. 2000, 39, 2263.

3. Zou, J.; Zhang, F.; Chen, Y.; Raymond, J. E.; Zhang, S.; Fan, J.; Zhu, J.; Li, A.;

Seetho, K.; He, X.; Pochan, D. J.; Wooley, K. L. Soft Matter 2013, 9, 5951.

4. Fan, J.; Zou, J.; He, X.; Zhang, F.; Zhang, S.; Raymond, J. E.; Wooley, K. L.

Chemical Science 2014, 5, 141.

5. Yang, R.; Peng, S.; Hughes, T. C. Soft Matter 2014, 10, 2188.

6. Wang, C.; Chen, Q.; Sun, F.; Zhang, D.; Zhang, G.; Huang, Y.; Zhao, R.; Zhu, D. J.

Am. Chem. Soc. 2010, 132, 3092.

7. Vidyasagar, A.; Handore, K.; Sureshan, K. M. Angewandte Chemie International

Edition 2011, 50, 8021.

8. Jung, Y. P.; Kim, J.-H.; Lee, D. S.; Kim, Y. H. J. Appl. Polym. Sci. 2007, 104, 2484.

9. Refojo, M. F.; Yasuda, H. J. Appl. Polym. Sci. 1965, 9, 2425.

10. Inaba, Y.; Kobayashi, T.; Yamamoto, H.; Urano, C.; Google Patents: 2008.

11. Kar, T.; Mandal, S. K.; Das, P. K. Chemistry – A European Journal 2011, 17, 14952.

12. Oh, J. K.; Tang, C.; Gao, H.; Tsarevsky, N. V.; Matyjaszewski, K. J. Am. Chem. Soc.

2006, 128, 5578.

13. Ejaz, M.; Yu, H.; Yan, Y.; Blake, D. A.; Ayyala, R. S.; Grayson, S. M. Polymer 2011,

52, 5262.

14. Barcan, G. A.; Zhang, X.; Waymouth, R. M. Journal of the American Chemical

Society 2015, 137, 5650.

15. Wuts, P. G. M.; Greene, T. W. In Greene's Protective Groups in Organic Synthesis;

John Wiley & Sons, Inc.: 2006, p 16.

104

16. Chen, J.; Qiu, X.; Ouyang, J.; Kong, J.; Zhong, W.; Xing, M. M. Q.

Biomacromolecules 2011, 12, 3601.

17. Kamada, J.; Koynov, K.; Corten, C.; Juhari, A.; Yoon, J. A.; Urban, M. W.; Balazs, A.

C.; Matyjaszewski, K. Macromolecules 2010, 43, 4133.

18. Choh, S.-Y.; Cross, D.; Wang, C. Biomacromolecules 2011, 12, 1126.

19. Klaikherd, A.; Nagamani, C.; Thayumanavan, S. J. Am. Chem. Soc. 2009, 131,

4830.

20. Li, Y.; Armes, S. P. Macromolecules 2005, 38, 8155.

105

106

CHAPTER 5

TOPOLOGICAL TRAPPING OF CYCLIC POLYMERS

Portions reprinted (adapted) with permission from

Stukenbroeker, T. S.; Solis-Ibarra D.; and Waymouth, R. M. Macromolecules, 2014, 47 (23), pp 8224–8230

107

*Sections concerning gel trapping with poly(HEMA) gels were adapted from the above publication.

108

5.1 Introduction

Efforts to generate linear poly(iPP)s with molecular weights ≥ 80 kDa were unsuccessful (Chapter 4), therefore we could not utilize the relative intrinsic viscosities1,2 to provide evidence for high molecular weight cyclic structures. We thus sought an alternative method to probe the topology of these materials and reasoned that a

"topological entrapment" method might provide evidence for the presence of cyclic structures. This approach entails the generation of a crosslinked gel in the presence of both linear and cyclic poly(iPP); if a significant fraction of the cyclic structures were to be threaded by the chains of the network, the cyclic structures would be entrapped by the gel whereas the linear chains would be free to diffuse out of the gel. Semlyen3,4 had previously investigated the entrapment of cyclic polyesters and cyclic PDMS (≤ 10 kDa) in crosslinked polymers and had measured the entrapment efficiencies with solvent extraction.

Figure 5.1 A conceptual rendering of the entrapped cyclic polymers (red) in a polyacrylate (blue) network with PEG (green) crosslinks.

109

5.2 Results and Discussion

5.2.1 Poly(iPP) trapping

Hydroxyethyl methacrylate polymers crosslinked with methacrylate functionalized polyethylene glycol served as a convenient crosslinked network that was both chemically compatible with the poly(iPP) and would swell suitably in organic solvent to release the embedded macromolecules. The generation of hydrogels was carried out by radical polymerization (AIBN) of 2-hydroxyethyl methacrylate (HEMA) and a telechelic PEG- diacrylate (1.2 mol%, PEG Mn=400) in the presence of either linear or cyclic poly(iPP). A small amount of DMF was added to help solvate both components throughout polymerization and minimize polymerization-induced phase separation.5,6

Figure 5.2: Image of poly(iPP) infused gels synthesized with (left) and without (right) DMF

Methanol was chosen as an extraction solvent as it swells the gel and is an excellent solvent for poly(iPP). Methanol was added to the hydrogels containing cyclic p(iPP) (Mn = 202 kDa), linear p(iPP) (Mn = 16 kDa), or linear poly(N,N- dimethylacrylamide) (PDMAA, Mn = 117 kDa) and were left to soak at 25 °C with gentle stirring for varying amounts of time. The methanol was then removed by syringe and the methanol washings were evaporated, suspended in DCM, then filtered and dried

110 under high vacuum. NMR analysis of the residue extracted from the gels showed it was greater than 95% poly(iPP).

Shown in Figure 5.3 is a graph of the weight percentage of entrapped polymers that were extracted from the methanol-swollen HEMA gels as a function of extraction time. As seen in this Figure, 98% of the linear poly(iPP) (Mn = 16 kDa) was extracted from the gel after 11 days, whereas only 64% of the poly(iPP) generated with an NHC

(Mn = 202 kDa) was extracted under similar conditions. These data indicate that ≥ 36% of poly(iPP) generated with an NHC remains entrapped in the gel under conditions where a linear poly(iPP) is completely extracted from the gel. These data are consistent with a cyclic topology for the entrapped poly(iPP)3,4 as cyclic chains that are threaded into the network would be unable to diffuse out of the gel.

Figure 5.3: Weight % of cyclic poly(iPP) (circles, Mn = 202 kDa), linear poly(iPP) (squares, Mn = 16 kDa) or linear PDMAA (triangles, Mn = 117 kDa) recovered from HEMA hydrogels as a function of extraction time (% relative to original mass of entrapped polymer). Lack of error bars indicates the standard deviation is smaller than the size of the marker.

111

The observation that high molecular weight linear poly(N,N-dimethylacrylamide)

(PDMAA, Mn = 117 kDa) also extracted with ≥90% efficiency after 11 days implies that the lower extraction efficiency (or greater entrapment percentage) of the cyclic poly(iPP) is not simply due to differences in the molecular weights. Furthermore, competitive extraction of equal portions linear and cyclic poly(iPP) entrapped in the same gel showed that the relative rates of extraction remained constant, suggesting the different molecular weights are not the root cause of the selectivity. The observation that the linear poly(iPP) is efficiently extracted implies that the trapping of the cyclic poly(iPP) is not due to the formation of covalent bonds between the HEMA gel and the p(iPP) created during the radical polymerization. This is supported by a separate control experiment which demonstrated that when methyl methacrylate is polymerized in the presence of poly(iPP) under the same conditions that the gel is synthesized, no shift or broadening of the poly(iPP) peak was observed (Figure 5.4).

PMMA alone

PMMA polym. in presence poly(IPP)

poly(IPP) alone

Index Differential Refractive Refractive Differential

0.00E+00 18 23 28 Retention Volume

Figure 5.4: Methyl methacrylate was radically polymerized in the presence of poly(iPP) with conditions similar to gel synthesis. The resulting polymer contains unchanged poly(iPP) peak in the GPC chromatogram.

112

The HEMA gels containing entrapped poly(iPP) were analyzed by FT-IR spectroscopy to provide evidence for the entrapped poly(iPP). FT-IR spectra of the gels was obtained by slicing the gels and using a diamond ATR to analyze the center portion of the gel (Figure 5.5-5.6). FTIR spectra of gels generated in the presence of either linear or cyclic poly(iPP) revealed the presence of both HEMA-derived carbonyl stretches and phosphate-derived stretches. After extraction with methanol, FTIR spectra of the HEMA gels containing entrapped cyclic poly(iPP) showed clear absorbances associated with the entrapped poly(iPP). In contrast, HEMA gels from which the linear poly(iPP) were extracted did not evidence any signal for residual poly(iPP).

Figure 5.5. FT-IR of hydrogels HEMA gel (black), HEMA hydrogel generated in the presence of cyclic poly(iPP) (blue)

113

Figure 5.6. Top: FT-IR of hydrogels subtracted spectra from Fig. 5.5 (blue), pure poly(iPP) (green), Bottom: FT-IR of hydrogels subtracted cyclic (blue) and linear (red) infused gels after 11 day methanol extraction, pure poly(iPP) (green)

114

The FT-IR data and extraction data provide evidence that the zwitterionic ring- opening polymerization of iPP with an NHC generates poly(iPP) that remains entrapped in HEMA crosslinked networks after methanol extraction for 11 days. On the basis of

Semlyen's work,4 these data strongly imply that the poly(iPP) generated under these conditions have a cyclic topology. The relatively modest trapping efficiency of 36% (11 day extraction) for poly(iPP) could be due to several factors: (a) incomplete threading of the cyclic poly(iPP) into the cross-linked network, (b) methanolic degradation of the poly(iPP) after extraction with methanol, or (c) the poly(iPP) generated from the zwitterionic ring-opening of iPP contains a mixture of linear and cyclic chains. Our current data do not allow us to completely discriminate among these possibilities, but molecular weight analysis of the extracted poly(iPP) provide evidence that methanolytic degradation of the poly(iPP) is in part responsible for the low trapping efficiencies.

Shown in Figure 5.7 are GPC traces of the poly(iPP) extracted from HEMA gels after 8 hours (5.7b) and 11 days (5.7c). Whereas the GPC traces of poly(iPP) generated from an NHC are monomodal (5.7a), the GPC trace of poly(iPP) extracted after 11 days shows both a low molecular weight peak and a high molecular weight shoulder. The presence of a low molecular fraction in the extracted poly(iPP) that is not present in the original sample suggests that under the extraction conditions, some of the poly(iPP) is degraded under the extraction conditions. The origin of the high molecular weight shoulder could be due to the methanolytic ring-breaking of the cyclic poly(iPP) to linear topology, which would be expected to elute at a shorter retention time.7

115

Figure 5.7: GPC traces of extracted cyclic iPP. (a) poly(iPP) generated from 2 (Table 1, entry 1), (b) poly(iPP) extracted from gel after 8 h, (c) poly(iPP) extracted from gel after 11 d.

116

5.2.1a Selective Polymer Retention Not a Molecular Weight Effect

A gel containing equal masses of linear and cyclic poly(iPP) substrate was synthesized and extracted with methanol. The extraction solvent was analyzed after 8 h and 24 h. GPC analysis on this extracted material revealed how much came from the cyclic and linear substrate, respectively.

Table 5.1 Quantities of poly(iPP) extracted in gels containing a 50:50 by mass mixture of linear and cyclic poly(iPP). Note that the fractions of linear and cyclic are of the extracted material. For example, in the 0-8 h wash, 19.4% of the initial poly(iPP) mass was extracted. Of this, 72% was linear, meaning that (.194)(.721)/(.5)=28.0% of the original linear sample was extracted during this period, where 0.5 represents the linear mass fraction of the original substrate.

Total mass fraction of Linear, mass Cyclic, mass

poly(iPP) extracted fraction of extracted fraction of extracted 0-8 h 19.4% 72.1% 27.9%

8-24 h 8.7% 81.8% 18.2%

It can be shown by this experiment that the selective retention of the cyclic sample is not due to the variable rate differences of extraction due to the molecular weight difference.

An informal proof is presented here.

Let C be the initial amount of extractable polymer (non-trapped) and W be the molecular weight of that polymer. Let F(C, W, t) be function describing the total mass of polymer extracted at time t since washing started. The expression for F is unknown but we can describe several characteristics of the function for t>0:

1. At any t the experiment could be “reset” such that C’= C0-F(t), thus both F and its

derivatives are continuous functions.

117

푑퐹 2. The amount of recovered polymer never decreases, thus > 0. 푑푡

3. F never exceeds the initial amount of recoverable polymer, C, which combined

푑2퐹 with 1 and 2 implies F is concave, < 0 (a stronger statement can actually be 푑푡2

made about the derivatives: none of them can ever change sign for t>0).

A representative graph of the function can be drawn.

Consider the case where we have two gels with the same C but different W, analogous to the null hypothesis wherein neither the high nor low molecular weight gels are at all trapped. It has been observed that the initial rate of polymer extracted trends inversely with molecular weight, thus in the figure W1

118

푑퐹 A representative graph of the rate of extraction, , is constructed. 푑푡

The experimental data was collected over two sequential time ranges allowing average rates of extraction to be calculated. In the above graph the ratio rate(W1) is necessarily rate(W2) decreasing with time. This is in contrast to the data, which, over the first 24 hours, shows a nearly constant rate of low MW extraction relative to high MW extraction (Table

5.2, where linear=W1 and cyclic=W2). The only way to reconcile the data with the model is if C1>C2. In other words there is less extractable polymer in the high molecular weight cyclic sample. Since both initially contain the same amount of polymer, the experiment containing high molecular weight sample is shows selective topological trapping independent of the rate of extraction.

Table 5.2 Cumulative extracted fraction of poly(iPP) (%) from poly(HEMA) gel with 16 mol % PEG crosslinker (Mn=3400). LMW=low molecular weight, HMW=high molecular weight.

LMW Linear, mass HMW Cyclic, mass Time (h) Ratio (linear/cyclic) fraction extracted fraction extracted 8 43.3 26.5 1.63

24 64.7 12.5 1.66

119

5.2.2 Catch-and-Release Gels

The first generation of gels demonstrated that topological trapping was possible for poly(iPP). However, for the strategy to progress from a method of interrogating polymer topology to a preparatory procedure for purification of cyclic macromolecules, the entrapped polymers must be recovered. To achieve this, a disulfide bond was incorporated into the gel crosslinker. Thus, in the presence of a reducing agent, the gel could be degraded to linear chains, allowing the entrapped species to be dethreaded and isolated. The development of a reducible disulfide gel is outlined in Chapter 4. It is also worth noting that these second generation gels also utilized a hydrophobic copolymer, butyl acrylate (BA), to impart swelling in DCM. This allowed the use of non- protic solvents and eliminated the possibility of methanolysis during extraction and recovery. The new gels were less sticky and easier to handle owing to the reduced amount of hydrogen bond donors from HEMA.

Figure 5.8 “Catch and Release” gel strategy. The terms extract and recover are used consistently throughout this chapter to refer to the steps indicated here.

For the proposed strategy to be useful, the gel formation and dissolution conditions must be entirely orthogonal to the poly(iPP) substrate. Control experiments

(Figure 5.4, above) indicated that radical polymerizations did not form any covalent

120 linkages with poly(iPP). Additionally a control study of the polymer under the gel degradation conditions (0.1 M DTT, 0.1% v/v TEA in DCM) for 24 h did not result in any change in the polymer molecular weight (Figure 5.9).

Figure 5.9 GPC (RI detector) chromatograms of linear poly(iPP) before (blue) and after (red) treatment with the gel dissolution conditions (24 h 0.1 M DTT in DCM, 0.1 v/v TEA)

Initial studies with the second generation gels displayed similar trapping efficiencies as the first generation gels. For instance the gel was cut into eight pieces

(Figure 5.10) and soaked in DCM. After 24 h, 69% of cyclic poly(iPP) had been extracted, indicating that 31% remained trapped within the gel. This trapped fraction was slightly lower than that trapped in 24 h by the first generation gels (39%), however this may be due to more efficient extracting by cutting up the gel which limits the distance chains must diffuse.

121

Figure 5.10 Catch and release gels containing poly(iPP) (20 wt. %). Clockwise from upper left, the formed gel (A), the gel cut up in vial for extraction (B), the gel following 24 h DCM extraction (C), and finally complete dissolution in 0.1 M DTT in DCM with 0.1% (v/v) trimethylamine (D)

Following extraction, the gel was soaked in a 0.1 M solution of dithiothreitol

(DTT) in DCM with catalytic TEA (0.1 % v/v) for 24 h. After this time, the solution could be filtered easily through a 0.45 µm syringe filter, indicating complete degradation. The degraded solution was dialyzed (3.5 kDa MWCO) against methanol to remove excess

DTT. Following dialysis, the solvent was removed under vacuum to yield a gelatinous solid. Once dried, dissolution was difficult in any solvent, and filtration was impossible.

Very dilute samples could be dissolved for NMR analysis which revealed that the material was 9.6% poly(iPP) by weight. This constituted 18% of the original iPP sample

(or 61% of the poly(iPP) not recovered in the initial extraction). Attempts to further purify the material via sequestration of the free thiol groups or silica chromatography were not successful.

122

Table 5.3 Fraction of cyclic poly(iPP) substrate recovered from generation II gels by DCM extraction (24 h) and following gel degradation (n=3).

% extracted % recovered* % unaccounted

69.3±1.2 18.7±1.4 12.0±1.7

*note that the recovered fraction is the amount of poly(iPP) present in the degraded gel according to NMR, not an isolated yield

To determine if incomplete or reversible disulfide reduction was the source of the insoluble network material, a control experiment was conducted in which no crosslinker was present, only HEMA/BA, cyclic poly(iPP), initiator and solvent. Though no gel formed, upon removal of the solvent under vacuum, a similar insoluble network material was formed. A similar experiment was conducted with linear poly(iPP) instead of cyclic material. In this case, the polymeric material formed was entirely soluble in THF and chloroform, allowing for easy NMR and GPC analysis.

Figure 5.11: Solubility of network material formed by in-situ polymerization of HEMA/BA with poly(iPP). The polymerization solvent (DMF) was removed under vacuum and THF was added. The cyclic poly(iPP) substrate is on the left; linear poly(iPP) substrate is on the right.

123

The presence of the unmodified linear poly(iPP) peak in the GPC chromatogram of the linear experiment (Figure 5.12) further confirms that that no covalent bonding between the acrylate polymers and poly(iPP) substrate is occurring. The broad polymer peak representing the linear free radical polymerization of HEMA/BA has Mn=30.0 kDa and PDI=1.44. It is hypothesized that this polymer, though linear and unbonded to the poly(iPP), threads the cyclic samples and entangles in such a way as to form a network material that is crosslinked and insoluble. The linear sample was diluted in THF and the polyacrylate was precipitated by addition of water to obtain pure linear poly(iPP).

Repeating this procedure for the cyclic samples did not separate the poly(iPP) from the polyacrylate, indicating that the equilibrium for the linear-cyclic threading/complexation is large.

Figure 5.12: GPC trace (RI detector) of material formed by in-situ polymerization of HEMA/BA with linear poly(iPP). The left peak is polyacrylate (Mn=30.0 kDa); the right peak is linear poly(iPP).

This finding was rather surprising. Previous experiments by Tezuka8 found that the polymerization of acrylates in the presence of cyclic poly(THF) did not induce 124 gelation. However, the high molecular weight of the cyclic poly(iPP)—nearly 200 kDa-- likely induces both threading and entanglements that serve to increase the threading equilibrium. The gel point will be reached once Kthreading, presumably a function of the molecular weight, is large enough that there is, on average, multiple threading events per cyclic chain. Future experiments will seek to isolate cyclic poly(iPP) from the degraded gel by limiting the molecular weight of the post-degradation polyacrylate chains.

125

5.2.3 Physical Properties of Topological Gels

The cyclic-infused gels constitute an intriguing class of double-network gels,9,10 whose mechanical, thermal, and rheological properties had not previously been explored. The DSC thermogram of the dried material showed no features (Figure

5.13). The Tg of poly(iPP), -45°C (see chapter 3), was not observed in scans to -75°C.

This suggests the absence of purely poly(iPP) domains, and further that either (1) the poly(iPP) does not reptate in the trapped state (no Tg), or (2) the Tg due to motion within the gel is lower than -75°C. Given the elastic nature of the material (see rheology, below), 1 is unlikely.

Figure 5.13 DSC scan of HEMA/BA gel with PEG crosslinker and cyclic poly(iPP) substrate (13.5 wt. %) No melting or crystallization features are present over the scan -75°C to 25°C.

126

The toughness of the HEMA/BA network made it unlikely that rheological differences between samples containing poly(iPP) would be observed in bulk material.

To lower the moduli of the materials and attempt to observe small differences based on the topology of the trapped species the networks were swollen in diglyme. This is a high boiling, non-toxic solvent in which the gel and poly(iPP) are highly soluble. Furthermore, its relatively high viscosity compared with DCM or methanol limits the rate of extraction of linear chains. Therefore, a sample of entrapped cyclic was prepared by forming a

HEMA/BA (50:50 by wt.) with PEG(Mn=400) crosslinker (1.35 mol %) in the presence of a poly(iPP) sample (20 wt %) prepared via ZROP. This was then washed in DCM for 24 h. Analysis of the extracted material showed 15 mg of poly(iPP) polymer, indicating that

25 mg of poly(iPP) remained entrapped in the material (13.5 wt. % of the material). A comparison sample containing 25 mg of linear poly(iPP) was synthesized in the same manner, and not washed. Finally, a sample containing no poly(iPP) was synthesized and washed.

Rheological measurements were taken with a rotational rheometer. The oscillatory stress strain curves showed a linear response of all materials (Figure 5.15).

Differences in the storage moduli were observed; however, these could be attributed to the different degrees of sample swelling, rather than differences in the macromolecular structure of the material. The loss modulus was negligible, as expected for a permanently crosslinked material. Compression tests showed similar responses.

127

Table 5.4 Samples of HEMA/BA gels with poly(iPP) substrates used for rheological studies.

type of Swelling % storage amount of poly(iPP) 풔풘풐풍풍풆풏 − 풅풓풚 modulus poly(iPP) present 풅풓풚 (oscillatory) none - 965% 4300 Pa

linear 25 mg 409% 1200 Pa

cyclic 25 mg 626% 5200 Pa

Figure 5.14 Pictures of a gel used for rheological study. First the gel is prepared (left), then swollen in diglyme (middle) and finally sliced with razor blade to 1 mm thick segment for oscillatory measurements using 8 mm cross-hatched geometry head.

128

Figure 5.15: Oscillatory stress strain curves of gels (see table 5.4) measured at 1.0 rad/s strain rate. The non-linear behavior at high strain is likely due to gel slippage on the instrument.

To observe the effects of poly(iPP) strain and relaxation within the material, the material was measured across different a range of oscillatory frequencies (Figure 5.16).

However, there was no significant variation in storage modulus. A step strain experiment, in which the stress relaxation was observed following a single oscillatory strain event, showed no significant differences between linear and cyclic infused samples (Figure 5.17).

129

Figure 5.16: Frequency sweep experiment shows no change in storage modulus across the range of oscillatory frequencies tested (0.31 to 31 rad/s).

Figure 5.17: Stress relaxation experiment with gels (Table 5.4). A 1% strain was instantaneously applied and the material stress was recorded.

130

5.2.4 Poly-valerolactone trapping

To extend the scope of the topological trapping method, poly(valerolactone)

(PVL) was investigated as a substrate. This polymer has been extensively studied in our laboratory as a substrate for ZROP,11,12 and the polymer products are known to be cyclic via MALDI-TOF and intrinsic viscosity measurements. This polyester is significantly

13 more hydrophobic than the poly(iPP) and is also semicrystalline (Tm=56°C).

Initial experiments sought to utilize the existing gel platforms that had been developed for trapping poly(iPP). The HEMA gels, which were swellable in methanol but not DCM, were not suitable because the PVL was insoluble in the extraction solvent.

Therefore the PVL was included in a generation II gel (HEMA/BA, disulfide crosslinker).

However, despite multiple attempts and adjustments, the PVL was extracted from the gel with nearly quantitative yield, thus no material had been trapped in the gel. In order to promote threading during gel formation, several solvents were screened that might increase the size of the PVL random coil to increase threading probability. However,

DMF, benzene and methyl ethyl ketone all failed to generate any trapping. The recovered fractions were nearly pure PVL and the molecular weights were unchanged, meaning that extraction was not due to cleavage of the cyclic chain (Figure 5.18).

131

Figure 5.18: GPC (RI detector) trace of cyclic PVL substrate and the extracted fraction from a HEMA:BA crosslinked gel.

Table 5.5 Extracted fraction of PVL from HEMA/BA gels with DSDMA crosslinker.

HEMA:BA Solvent % extracted

50:50 DMF 90%

25:75 DMF 75%

25:75 MEK ~100%

50:50 DMF 87%

gel degraded upon 0:100 DMF extraction gel degraded upon 50:50 benzene extraction

132

Another strategy to increase threading was to use a PVL-based crosslinker. In order to impart chemical compatibility, a α,ω-bis(acrylate) poly(valerolactone) with an internal disulfide was synthesized (chapter 4). This crosslinker was then mixed with the

PVL substrate and then the gel was prepared normally. Despite this, the PVL substrate was again completely extracted. It is unknown why the trapping of PVL is disfavored.

Compared with the poly(iPP) it is a smaller molecular weight, however, the PVL is much larger than the polyesters trapped successfully by Semylen. It is likely an issue of aggregation and forming microdomains within the material due to unfavorable interaction energy between the polyacrylate and the PVL. Though the polymer forms a homogenous concentrated solution with the acrylate monomers, as the molecular weight of the acrylate increases the interaction becomes more disfavored, a phenomenon known as Polymerization-Induced Phase Separation (PIPS).5,6 In this case it is thought that the PIPS forms microdomains that lead to slightly cloudy gels. When poly(methyl- caprolactone), a chemically similar polyester that is amorphous, was tested as a substrate 76% was recovered in 24 h from a cloudy gel.

Table 5.6 Extracted fraction of PVL from HEMA/BA gels with α,ω-bis(acrylate) poly(valerolactone) crosslinker.

HEMA:BA:PVL-xlinker % extracted

50:0:50 90-100% (3 expts.)

25:25:50 90%

133

Figure 5.19: Picture of PVL gel dry (left) and after extraction (right).

The formation of PVL-rich domains within the gel was evident from DSC studies.

The DSC scan of the PVL substrate (Figure 5.20, top) shows a Tm=58°C and Tc=34°C, in agreement with literature values. The DSC of the dried gel material (not extracted) shows both a melting and a crystallization event (Figure 5.20, bottom). Since the crosslinked network does not have any of these features, it means that the PVL is forming crystalline regions which are likely not threaded by polyacrylate. The Tm has been reduced to 55°C and the Tc is reduced to 23°C, indicating that these domains are impacted by the presence of the gel. It is possible that the reduced mobility of the domains within the network leads to a higher barrier for crystallization.

134

Figure 5.20 DSC Scans of the PVL substrate (top) and a PVL-embedded gel (bottom).

135

5.3 Experimental Procedures

Wyatt Heleos II 8-angle light scattering detector and Wyatt T-rex refractive index detector. NMR data was collected on 300, 400 and 500 MHz Varian instruments. DSC samples were prepared by placing gels in a vacuum oven at 100°C for 3 h.

Rheological Studies: Rheological measurements were taken on TA Instruments ARES-

G2 Rheometer. Samples were prepared as described in the text. Directly following 24 h of DCM extraction, the cyclic-containing sample was submerged in diglyme with gentle stirring for 8 h with multiple changes of solvent to ensure the swelling solvent was completely diglyme. A similar protocol was followed for the sample containing no polymer substrate. For the linear-containing sample, the gel was synthesized and dried under vacuum overnight. It was then swelled for 8 h, however the swelling was significantly less than the other samples. It is possible this is because the dried network material did not swell as rapidly as the swollen gels exchanged solvent.

136

Two razor blades were affixed parallel to each other with a 1 mm gap. This was used to cut down the center of the gels and create the rheological samples. All rheological measurements were made with 0.1±0.01 N. Loss of solvent was negligible during the measurement.

FT-IR characterization of the gels: Gels were dried at 60°C overnight in a vacuum oven. The Vertex 70 spectrophotometer was fitted with a diamond ATR unit. Gel samples were prepared using a razor blade to cut cross sections and spectra were acquired.

Synthesis of poly(iPP) embedded gels (generation I): 152 mg 2-hydroxyethyl methacrylate (Aldrich, distilled, FW=130.14) was combined with 8 mg of poly(ethylene glycol) dimethacrylate (Polymer Sciences, PEG Mn = 400; 5 wt. % = 1.2 mol % = 1 crosslink/40 HEMA units) and 0.4 mL of dimethylformamide in a plastic vial. This mixture was vortexed for ten minutes and then added to 40 mg of a poly(iPP) sample.

To this mixture was added 1.6 mg of AIBN (Aldrich, recrystallized, FW=164.21); the resulting solution was purged with argon and placed in a 75°C bath for 30 minutes. The gel was removed and dried at 60°C overnight in a vacuum oven to give a transparent, sticky gel.

Synthesis of poly(N,N-dimethylacrylamide) (PDMAA) embedded gels: These gels were prepared analogously, but with 40 mg of PDMAA (Polymer Source, stated Mn=117 kDa, Mw/Mn = 1.5).

Gel Extractions with methanol: The polymer-embedded gels were extracted by adding

5 mL methanol to a vial containing the gel and a stirbar. With gentle stirring, the sample was allowed to soak for a given amount of time and then the methanol was removed via

137 syringe. The methanol washings were dried and then 3 mL of DCM was added and briefly sonicated to extract soluble products. The washings were syringe filtered and dried under high vacuum. The PDMAA washings were less soluble than poly(iPP) in

DCM, so instead they were analyzed by NMR in d4-methanol to determine oligomeric

HEMA impurities and reported masses were adjusted accordingly.

PMMA polymerization in presence of poly(iPP): 10 mg of linear poly(iPP) was dissolved in 0.1 mL of DMF. 0.7 mg of AIBN and 25 mg of methyl methacrylate were added and vortex mixed for ten minutes. The vial was purged with argon and put in

80°C oil bath for 30 minutes. The resulting product was dried under vacuum for several hours to remove the unreacted monomer and DMF, and then a GPC sample was taken.

The remaining polymer was precipitated into several mL of methanol and filtered to remove PMMA. The methanol supernatant was evaporated to give pure poly(iPP).

Yield = 7.5 mg (75%).

Synthesis of catch and release gels (generation II): 70 mg 2-hydroxyethyl methacrylate (Aldrich, distilled, FW=130.14), 70 mg of butyl acrylate (Aldrich, distilled,

FW=128.17), 20 mg of DSDMA (FW=262.35) and approximately 40 mg of polymer substrate were dissolved in 400 µL of DMF containing 2 mg of AIBN (Aldrich, recrystallized, FW=164.21). This solution was stirred with a vortex mixer for 30 min.

The vial was then purged with Ar for 60 s and placed in a 75°C bath for 30 minutes.

Following this crosslinking reaction, the gel could be removed from the vial as a soft, elastic gel.

Extraction conditions: Gels were cut into approximately 25-50 mg pieces with a razor blade and placed in vial with 5-10 mL of DCM and gently stirred for 24 hours. The DCM was subsequently decanted and evaporated to give the extracted fraction. 138

Recovery conditions: Gels were placed in a 0.1 M DTT solution in DCM with 0.1 v/v % triethylamine. The vials were capped and stirred for 24 h. Following this, the solutions could be dialyzed in methanol to remove the excess DTT.

Quantification of polymer substrate in extracted & recovered samples: For extracted samples, a 1H proton spectrum was acquired. If the recovered material contained more than the substrate, an NMR standard was used to quantify the amount.

1 31 Mesitylene was used for PVL ( H) and NBu4PF6 was used for poly(iPP) ( P).

Figure 5.21 Example of poly(iPP) quantification with NBu4PF6 internal standard using 31P NMR.

139

1 Figure 5.22 H NMR (CDCl3) of crude extracted material containing poly(iPP), unreacted HEMA and DMF.

Additionally, for the PVL gels containing a PVL xlinker, a GPC spectrum was taken to quantify the amount of substrate PVL and xlinker PVL present in the material.

Figure 5.23 GPC chromatogram of extracted fraction from gel with PVL substrate and PVL xlinker, overlayed with GPC traces of each.

140

For recovery of the poly(iPP) from the degraded gel, the material was dissolved in THF.

For cyclic polymers, the amount of THF was as high as 1 mL/mg of substrate in order to achieve reasonable dissolution (still some cloudiness present). Then, deionized water was added to bring the final composition of the mixture 2:1 H2O:THF, and the mixture was centrifuged for 30 min. The polyacrylate material was deposited at the bottom of the tubes. The supernatant liquid was decanted, concentrated, and dialyzed overnight against methanol. The resulting material was pure poly(iPP) by NMR, with no degradation noted by NMR or GPC.

PVL trapping (PVL crosslinker): 80 mg 2-hydroxyethyl methacrylate (Aldrich, distilled,

FW=130.14), 80 mg of PVL crosslinker (Mn=3800), and approximately 40 mg of PVL substrate were dissolved in 400 uL of DMF containing 2 mg of AIBN (Aldrich, recrystallized, FW=164.21). This solution was stirred with a vortex mixer for 90 min (until

PVL was dissolved). The vial was then purged with Ar for 60 s and placed in a 75°C bath for 90 minutes.

141

5.4 References

1. Topological Polymer Chemistry: Progress of cyclic polymers in synthesis, properties and functions; Tezuka, Y., Ed.; World Scientific Publishing: Singapore, 2013.

2. Cyclic Polymers; 2nd ed.; Semlyen, J. A., Ed.; Kluwer: Dordrecht, 2000.

3. Fyvie, T. J.; Frisch, H. L.; Semlyen, J. A.; Clarson, S. J.; Mark, J. E. Journal of

Polymer Science Part A: Polymer Chemistry 1987, 25, 2503.

4. Wood, B. R.; Joyce, S. J.; Scrivens, G.; Semlyen, J. A.; Hodge, P.; O'Dell, R. Polymer

1993, 34, 3059.

5. Kim, J. Y.; Cho, C. H.; Palffy-Muhoray, P.; Kyu, T. Physical Review Letters 1993, 71,

2232.

6. Boots, H. M. J.; Kloosterboer, J. G.; Serbutoviez, C.; Touwslager, F. J.

Macromolecules 1996, 29, 7683.

7. Lee, H. C.; Lee, H.; Lee, W.; Chang, T. H.; Roovers, J. Macromolecules 2000, 33,

8119.

8. Oike, H.; Mouri, T.; Tezuka, Y. Macromolecules 2001, 34, 6229.

9. Gong, J. P.; Katsuyama, Y.; Kurokawa, T.; Osada, Y. Advanced Materials 2003, 15,

1155.

10. Waters, D. J.; Engberg, K.; Parke-Houben, R.; Ta, C. N.; Jackson, A. J.; Toney, M.

F.; Frank, C. W. Macromolecules 2011, 44, 5776.

11. Chang, Y. A.; Waymouth, R. M. Polymer Chemistry 2015, 6, 5212.

12. Shin, E. J.; Brown, H. A.; Gonzalez, S.; Jeong, W.; Hedrick, J. L.; Waymouth, R. M.

Angewandte Chemie International Edition 2011, 50, 6388.

13. Storey, R. F.; Herring, K. R.; Hoffman, D. C. Journal of Polymer Science Part A:

Polymer Chemistry 1991, 29, 1759.

142