N-Heterocyclic Carbene Mediated Zwitterionic Polymerization for Cyclic Polymers

N-HETEROCYCLIC CARBENE MEDIATED ZWITTERIONIC POLYMERIZATION FOR CYCLIC POLYMERS

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

Eun Ji Shin

July 2011

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/tj619sw3147

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.

Wray Huestis

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.

Barry Trost

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

Zwitterionic polymerization involves a propagating species with both positively and negatively charged groups. Previous investigations on zwitterionic polymerization concern alternating copolymerization of nucleophilic and electrophilic monomers and polymerization of isolated stable zwitterionic monomers. More recently, the ring-opening polymerization of cyclic monomers using nucleophilic initiators have been studied. Specifically, the zwitterionic polymerization of cyclic esters using N-heterocyclic carbenes (NHCs) is the focus of this thesis. The N- heterocyclic carbene mediated zwitterionic polymerization of cyclic monomers provides an expedient route to polymers of various architectures, such as cyclic polymers, cyclic gradient copolymers and linear telechelic polymers. The ring-opening polymerization of lactide initiated by NHCs generates cyclic poly(lactide)s of defined molecular weight and molecular weight distribution. Kinetic studies implicate a mechanism that involves a slow initiation step and a propagation step that is much faster than depropagation and chain termination by cyclization. Stochastic simulations and chain extension experiments showed that only a fraction of the NHC forms the active zwitterion in solution, leading to both chain extension of the zwitterions and re-initiation of the NHC upon addition of the second batch of monomer. These results prompted investigation of a more efficient way to prepare cyclic block copolymers. The difference in reactivity of NHCs towards different monomers was exploited to synthesize cyclic block copolymers of -valerolactone (VL) and -caprolactone (CL). The faster ring-opening of VL relative to CL resulted in a gradient cyclic copolymer comprised of VL-rich sequences that transition to CL- rich sequences in a cyclic macromolecule, instead of a cyclic diblock copolymer. This work not only provides a simple batch copolymerization protocol to produce cyclic gradient copolymers, but also demonstrates the marked difference in reactivity of the NHCs compared to metal catalysts, which produce random copolymers.

Stereocomplexation behavior has been observed in blends of linear poly(L- lactide) and linear poly(D-lactide). The influence of topology on the formation of

v stereocomplex was investigated using blends of linear and cyclic poly(lactide)s prepared by NHC mediated zwitterionic polymerization. The linear/cyclic and cyclic/cyclic blends all form stereocomplexes when annealed. Analyses of data from various characterization techniques indicate that the cyclic topology does not impede the formation of stereocomplexes. The purity of the cyclic polymers is always a concern in the synthesis and physical property studies. Attempts to identify and quantify the linear contamination in cyclic poly(-caprolactone) samples are described. Esterification reactions targeting the hydroxyl endgroups of linear contaminants were not successful, but the macroinitiator approach where the linear contaminant in a cyclic polymer sample is used as the macroinitiator to grow polymers to identify and remove the linear contamination shows promise. A cyclic polymer more robust to post-polymerization chemistry may be needed for more thorough purity studies.

vi PREFACE

Chapter 1 is a review of zwitterionic polymerization with a particular emphasis on the synthesis of cyclic polymers. It is intended to give the reader a background on the brief history of zwitterionic polymerization and systems which use N-heterocyclic carbenes to give interesting architectures, such as cyclic polymers. Chapter 2 describes polymerization of lactide with the N-heterocyclic carbene, 1,3- dimesitylimidazol-2-ylidene (IMes) by sequential addition of monomer. The products obtained are analyzed in the light of the mechanism and kinetics of this system. Also, a short section on the effect of different solvents on the polymerization of lactide with IMes is included. The introduction part and the stochastic simulations and experiments relating to them have been published in the Journal of the American Chemical Society 2009, 131, 4884-4891. The experiments in section 2.3 were performed together with Dr. Wonhee Jeong. Chapter 3 shows the stereocomplexation in various blends of linear and cyclic poly(lactide)s. The effect of the cyclic topology on the crystallization and stereocomplexation was explored. The linear poly(lactide)s were synthesized by a CPIMA SURE program undergraduate student Alexandra Jones. The wide-angle X-ray scattering experiments were performed by Dr. Jihoon Kang at Seoul National University, Korea. This work will be submitted to Macromolecules for publication. Chapter 4 outlines the synthesis and characterization of cyclic gradient copolymers using N-heterocyclic carbenes. This work was published in Angewandte Chemie, International Edition 2011, 50, 6388-6391. The reactivity ratios of - valerolactone and -caprolactone were determined by Silvia Gonzalez and Dr. Wonhee Jeong and 1,3-diethyl-4,5-dimethylimidazol-2-ylidene was synthesized by Hayley Brown. Chapter 5 details the synthesis of cyclic poly(-caprolactone) and the efforts to identify, quantify and remove linear contamination in these cyclic polymers. Parts of section 5.2 has been published in Macromolecules 2011, 44, 2773-2779. Synthesis of 1,3,4,5-tetramethylimidazol-2-ylidene via the deprotonation of the imidazolium salt was performed by Hayley Brown. All of the work in this thesis is done by me, except where noted.

vii viii ACKNOWLEDGEMENTS

I would like to express my deepest gratitude to my advisor Professor Waymouth. He is the ideal research advisor and has become a role model for me. Without his ideas and support, my Ph.D would have never been possible. Thank you so much. Dr. Hedrick has always been a great co-advisor and I thank him for always supportive and enthusiastic about my work. I would like to thank my committee members, Professor Trost and Professor Huestis for giving me a lot of feedback and advice on my research and research proposals. Also, I thank Professor Frank for being a substitute member on one of my research proposals. Professors Kanan and Sattely were so kind to find the time to be the non-readers for my thesis defense. I am so glad that I had the privilege to discuss research with such great minds. I thank Professor Do Yoon for the continuing support he gave me even after I finished my degree with him. Wonhee Jeong was the person who led me into the field of polymer synthesis. Ever since I met him he has been so kind and patient in teaching me everything I know and giving advice on not only research but also ways of life. I am grateful for everything. Another person that made my Ph.D possible was Hayley Brown. She is so smart and such a great person to work with. I owe her a lot and cannot thank her enough. The Waymouth group has been such a fun place to work in. I have met so many good friends. I thank Kyung-sun Son for always being there. She is the nicest person I’ve ever met. I could talk to her about everything and I hope this relationship lasts. I loved and miss the jokes of Matt Kiesewetter. His presence made my trips to lab 1 so eventful and enjoyable. Not only that, he gave so much advice when I got stuck with my research. I really appreciate his help. Liz Kiesewetter was the model ‘lab mother’ and I thank her for always being so nice to me. I thank David Pearson for all his help he gave me whenever I had a synthetic problem. Justin Edward makes me proud that I was his mentor, if only for a short time. I wish him all the best in the

ix future. Dr. Sangjin Jeon has been and still is a great person to go for advice on anything. I thank him for being such a great senior. I would like to thank all other past and present members for making my time in the Waymouth lab so productive and pleasant. I should also thank all the collaborators who helped and contributed to my work; Dr. John Pople, Dr. Jihoon Kang, Allie Jones. Without them my work would not have been complete. Yeonju Kwak and Jean Chung joined the Ph.D program with me in 2006. Despite their young age, they have been really good friends and I am very lucky to have met them. I am very sorry that I won’t be able to be there for their thesis defense but I wish them all the best. I wouldn’t have been able to get my degree without the support from my friends. They gave me strength whenever I felt lonely being alone here. Thank you, Kyungsuk, Jongeun, Sojin, Jooyeon, Bokyung, Jungwha, Hwayeon and the Ebert family. I would like to thank Dr. Hyunwook Ro for being such a great ‘brother’ and taking care of me. My family, I cannot thank enough. I would like to thank my parents for always being supportive of what I do; I am who I am thanks to them. Thank you Heewon, little sister, for all those ‘no content’ emails you sent to cheer me up. I am grateful to my big sister, Misook and brother-in-law Patrick. It was nice to have family close by. Thank you, Jinjoo, for being such a sweet sister-in-law. My parents- in-law are really the best. There would not be any other who would allow their daughter-in-law to go off by herself to study abroad. Thank you so much for being so understanding and patient. Lastly, I would like to thank my husband, Jin Hyun, who has suffered the most during the past 5 years. I wouldn’t have been able to get my degree and do the research I like without his patience and support. I hope I can give him such support someday in the future. Thank you so much.

x TABLE OF CONTENTS

ABSTRACT v PREFACE vii ACKNOWLEDGEMENTS ix LIST OF TABLES xiii LIST OF FIGURES xiv LIST OF SCHEMES xviii SYMBOLS AND ABBREVIATIONS xix

CHAPTER 1: Zwitterionic Polymerization for Cyclic Polymers 1

1.1 Introduction 2 1.2 Zwitterionic Polymerization for Cyclic Polymers 5 1.3 Zwitterionic Polymerization for Different Architectures 11 1.4 Future Outlook 15 1.5 References 17

CHAPTER 2: Chain Extension Experiments and Solvent Effects on N-Heterocyclic Carbene Mediated Zwitterionic Polymerization of Lactide 21 2.1 Introduction 22 2.2 Chain Extension 25 2.3 Nature of Zwitterions: Solvent Effect 33 2.4 Summary 34 2.5 Experimental Section 35 2.6 References 48

CHAPTER 3: Stereocomplexation in Cyclic and Linear Poly(lactide) Blends 51 3.1 Introduction 52 3.2 Results and Discussion 52 3.3 Summary 58 3.4 Experimental Section 59 3.4.1 Experimental Procedues 59 3.4.2 Characterization Data 62 3.5 References 68

CHAPTER 4: Zwitterionic Copolymerization: Synthesis of Cyclic Gradient Copolymers 71 4.1 Introduction 72 4.2 Results and Discussion 73 4.3 Conclusion 78 4.4 Experimental Section 79 4.4.1 Experimental Procedures 79 4.4.2 Determination of Reactivity Ratios 84 4.4.3 Synthesis and Characterization Data of Linear Copolymers 86 4.4.4 Synthesis of Crystalline-Amorphous Copolymers 88 4.5 References 90

CHAPTER 5: Synthesis of Cyclic Poly(-caprolactone) and Attempts to Identify and Remove Linear Contamination 93 5.1 Introduction 94 5.2 Synthesis of Cyclic Poly(-caprolactone) 95 5.3 Attempts to Quantify Linear Contaminants 101 5.4 Removal of Linear Contaminants 109 5.5 Summary 111 5.6 Experimental Section 113 5.7 References 116

xii LIST OF TABLES

Table 2.1 Selected polymerization data 26 Table 2.2 Characterization data for samples in Table 2.1 26 Table 2.3 Monomer conversion during the second polymerization 39 Table 3.1 Characterization of polylactides used in this study 54 Table 3.2 Blends of polylactides used for this study 54

Table 3.3 Lamellar thickness (Lc) and long period (Lp) 57 Table 3.4 Absolute molecular weights of the samples, determined by light scattering 62 Table 3.5 Fraction of iii tetrad 65 Table 3.6 Composition analysis calculated from homonuclear decoupled 1H NMR spectra 66 Table 4.1 Polymerization and characterization data for cyclic gradient

copolymers generated by using Me2IEt 75 Table 4.2 Selected samples for microstructure analysis 76 Table 4.3 Characterization of various linear samples 88 Table 4.4 Crystalline – amorphous copolymers 89 Table 5.1 Selected data for zwitterionic ring-opening polymerization of CL with NHCs 97 Table 5.2 Selected data of polymerization performed using thione derived carbene 3 (entries 1 – 4) and imidazolium salt derived carbene 3 (entries 5 – 7) 99 Table 5.3 Calculation of % linear for samples in Figure 5.6 104

xiii LIST OF FIGURES

Figure 1.1 (a) First-order kinetic plots of various initial conditions. The data for

[M]0 = 0.6 M, [I]0 = 0.006 M were fitted to determine the rate constants (solid line) and these rate constants were used to predict for different initial conditions (the dotted lines). (b) Molecular weight and polydispersity index versus conversion plot for two different initial conditions. 5 Figure 1.2 Electron density correlation function K(z) for (a) linear PCL and (b)

cyclic PCL versus time during isothermal crystallization at 45 ºC. Lc is

lamellar thickness and Lp is long period 9 Figure 1.3 (a) Crystal growth rates from SAXS and (b) half-times of

crystallization as a function of Mw from DSC for cyclic and linear PCL isothermally crystallized (45 ºC) 10 Figure 1.4 (a) Representative molecular weight and polydispersity index versus conversion plot. (b) SEC chromatograms of poly(N-Bu-glycine) before (M/I = 50, dashed curve) and after (M/I = 25, solid curve) the chain- extension experiment. The dot-dashed curve shows the data for an independently synthesized polymer sample of M/I = 75. (Figures reprinted with permission from J. Am. Chem. Soc. 2009, 131, 18072- 18074. Copyright 2009, American Chemical Society.) 12 Figure 1.5 Potential monomers to broaden the area of zwitterionic polymerization 16

Figure 2.1 (a) First-order kinetic plot of the polymerization data at [M]0 = 0.6 M

and [I]0 = 0.006 M. Each point is the average of three runs. The line is

a linear fitting. (b) Plots of Mn and PDI versus monomer conversion. The line is the theoretically predicted molecular weight for an ideal living polymerization 23 Figure 2.2 Gel permeation chromatograms of PLLA homopolymer (entry 1, dashed) and P(LLA-b-DLA) copolymer (entry 7, solid) 27

xiv Figure 2.3 Differential scanning calorimetry traces of P(LLA-b-DLA) copolymer (entry 7) 28

Figure 2.4 Stochastic simulations of polymerization kinetics at (a) [M]0 = 0.6 M

and [I]0 = 0.006 M, (b) [M]0 = 0.9 M and [I]0 = 0.006 M. Ni,0 is the

initial number of initiators, and Ni, Nz, and Nc are the number of initiators, zwitterions, and cyclized chains, respectively 29 Figure 2.5 GPC traces of crude products from the polymerizations (a) (1) (dashed) and (2) (solid) and (b) (3) (dashed) and (4) (solid) in Scheme 2.2 31 Figure 2.6 (a) First-order kinetic plots of the polymerization data in THF and in

1,4-dioxane at [M]0 = 0.6 M and [I]0 = 0.006 M. (b) Plots of Mn and PDI versus monomer conversion for the polymerization data in THF

and in 1,4-dioxane at [M]0 = 0.6 M and [I]0 = 0.006 M. The lines are

linear fittings for Mns 33

Figure 2.7 Mark-Houwink plots of linear and cyclic (a) poly(L-lactide)

homopolymers; []cyclic/[]linear = 0.74 and (b) poly(L-lactide-b-D-

lactide) copolymers; []cyclic/[]linear = 0.75 37 Figure 3.1 Differential scanning calorimetry scans of (a) cyclic PLLA (b) linear PLLA + linear PDLA (c) cyclic PLLA + linear PDLA and (d) cyclic PLLA + cyclicPDLA 55 Figure 3.2 Wide-angle X-ray scattering patterns after (a) annealing at 150 oC (90 oC for c-PLLA) for 24 hours and (b) cooling from the melt to room temperature 56 Figure 3.3 Mark-Houwink plots of (a) linear and cyclic PLLA and (b) linear and

cyclic PDLA. The ratios ([]cyclic/[]linear) are 0.74 62 Figure 3.4 Homonuclear decoupled 1H NMR spectra of (a) linear PLLA and (b) cyclic PLLA. Peak assignments are done according to reported literature 63 Figure 3.5 SAXS profiles (left) and one-dimensional electron density autocorrelation functions (right) of (a) linear PLLA, (b) cyclic PLLA,

xv (c) linear PLLA + linear PDLA, (d) cyclic PLLA + linear PDLA and (e) cyclic PLLA + cyclic PDLA 67 Figure 4.1 Schematic diagrams of linear and cyclic gradient copolymers 72 Figure 4.2 Zwitterionic ring-opening copolymerization of VL and CL 74 Figure 4.3 (a) Conversion of each monomer with time and (b) molecular weight

with conversion for MVL:MCL = 1:1, [Mtot]0 = 1 M, Mtot/I = 100 76 Figure 4.4 Melting points of various samples compared with literature values (ref 27). Two melting peaks from one sample are drawn as two symbols connected with a vertical line 77 Figure 4.5 Mark-Houwink plots of all samples in Table 4.1 and their respective linear analogues 81 Figure 4.6 13C NMR spectra of the four samples in Table 4.2. (CL-VL means the methylene signal of the -valerolactone unit connected to a -

caprolactone, i.e. CL-O-CH2-CH2-CH2-CH2-CO- ) 82

Figure 4.7 Composition analysis for copolymerization at MVL:MCL = 1:1, [Mtot]0 =

1 M, Mtot/ I = 100. (a) Cumulative fraction of VL in copolymer and (b) instantaneous fraction of VL with conversion 82 Figure 4.8 Differential scanning calorimetry scans of samples in Table 4.2 83 Figure 4.9 Wide-angle X-ray scattering patterns of selected samples 83 Figure 4.10 Determination of reactivity ratios using Fineman-Ross method 85 Figure 4.11 Graph relating the amount of CL in starting solution to the amount of CL which will be incorporated into the polymer 85 Figure 4.12 DSC scans of the crystalline – amorphous copolymers 89 Figure 5.1 NHCs used for the zwitterionic ring-opening polymerization of lactones 96 Figure 5.2 First-order kinetic plots for polymerization of CL with carbenes 1(), 2() and 3(). The polymerizations were carried out under the

conditions, [M]0 = 1.0 M and [I]0 = 0.01 M 98

xvi Figure 5.3 Plots of molecular weight (filled symbols) and polydispersity index (open symbols) with conversion for polymerization done with carbene

1 at [M]0 = 1.0 M 98 Figure 5.4 Mark-Houwink plots of the linear () and cyclic () PCL sample (entry 5, Table 5.1) 100 Figure 5.5 Esterification reactions used in this study 103 Figure 5.6 Representative GPC traces for the reaction of (a) linear PCL and (b) cyclic PCL with pyrene butanoyl chloride and TEA in THF in the presence of a linear standard 104 Figure 5.7 NMR spectra of (a) linear PCL and (b) cyclic PCL treated with TMS- Cl (in the presence of TEA) 105 Figure 5.8 NMR spectrum of a linear PCL reacted with Triaz carbene for 2 days (85 % conversion). The peaks labeled as ‘before’ and ‘after’ are the

signals from the methylene (–CH2–OH) protons before and after reacting with Triaz, respectively 106 Figure 5.9 GPC trace (RI detector signal) of linear PCL treated with MTC-pyrene in the presence of TU/DBU 108 Figure 5.10 Removal of linear contamination using trimethylaluminum treated silica 109 Figure 5.11 Mark-Houwink plots of a cyclic PCL sample before (□) and after (■) filtration through linear scavenger 110 Figure 5.12 GPC traces (RI detector signals) before (dashed) and after (solid) filtration for (a) a successful case and (b) a failed case 110 Figure 5.13 Removal of linear contamination using diethylzinc treated polystyrene resins 111

xvii LIST OF SCHEMES

Scheme 1.1 General mechanism of zwitterionic polymerization 2 Scheme 1.2 Charge cancellation coupling polymerization of isolable zwitterions 3 Scheme 1.3 Ring-opening polymerization of pivaloactone with pyridine initiators 4 Scheme 1.4 IMes mediated zwitterionic polymerization of lactide 6 Scheme 1.5 Zwitterionic ring-expansion polymerization of -lactones with SIMes 7 Scheme 1.6 Living zwitterionic ring-opening polymerization of N-alkyl-N- carboxyanhydrides to give cyclic homo- (top path) and cyclic diblock copoly(-peptoid)s (bottom path) 11 Scheme 1.7 Zwitterionic ring-opening copolymerization of -valerolactone and - caprolactone 14 Scheme 1.8 Zwitterionic ring-opening polymerization of ethylene oxide 15 Scheme 2.1 Proposed zwitterionic polymerization of lactide to generate cyclic poly(lactide) 22 Scheme 2.2 Sequential polymerization of lactide 31 Scheme 3.1 Polymerization of lactide with IMes 53 Scheme 5.1 Proposed mechanism for the zwitterionic ring-opening polymerization of CL 96 Scheme 5.2 Strategy for detecting and quantifying linear contamination in cyclic PCL samples 101 Scheme 5.3 Reaction of linear PCL with Triaz 105 Scheme 5.4 Unsuccessful step in the synthesis of 1,3-diphenyl-4-pyrene-4,5- dihydro-1H-1,2,4-triazol-5-ylidene alcohol adduct 107 Scheme 5.5 Polymerization of MTC-pyrene in the presence of a cyclic PCL sample 108

xviii SYMBOLS AND ABBREVIATIONS

[M]0 Initial monomer concentration [I]0 Initial initiator concentration ki Rate constant for initiation kp Rate constant for propagation kc Rate constant for cyclization kd Rate constant for depropagation Tm Melting temperature Ta Annealing temperature Hm Heat of melting Lc Lamellar thickness Lp Long period [] Intrinsic viscosity AlMe3 Trimethylaluminum CL -caprolactone CuAAC Cu(I)-catalyzed azide-alkyne cycloaddition DBU 1,8-diazabicyclo[5.4.0]undec-7-ene DCM dichloromethane DLA D-lactide DLS Dynamic light scattering DP Degree of polymerization (repeat units/initiator) DSC Differential scanning calorimetry EO Ethylene oxide ESI-MS Electron spray ionization-mass spectroscopy Et2Al(OMe) Diethylaluminum methoxide GPC Gel permeation chromatography IMe4 1,3,4,5-tetramethylimidazol-2-ylidene IMes 1,3-bis(-2,4,6-trimethylphenyl)imidazol-2-ylidene, 1,3- dimesitylimidazol-2-ylidene IPr 1,3- diisopropylimidazol-2-ylidene LA Lactide LCCC Liquid chromatography at the critical condition LLA L-lactide M/I Monomer to initiator ratio MALDI-TOF Matrix-assisted laser desorption/ionization – time of flight (mass spectrometry) Me2IEt 1,3-diethyl-4,5-dimethylimidazol-2-ylidene Me2IPr 1,3-diisopropyl-4,5-dimethylimidazol-2-ylidene Mn Number average molecular weight MTC(-OH); (-OR) 5-methyl-2-oxo-1,3-dioxane-5-carboxylic acid; -R ester Mw Weight average molecular weight NCA N-carboxyanhydride NHC N-heterocyclic carbene NMR Nuclear Magnetic Resonance

xix PCL Poly(-caprolactone) PCOE Poly(cyclooctene) PDI Polydispersity index PDLA Poly(D-lactide) PEO Poly(ethylene oxide) PLA Poly(lactide) PLLA Poly(L-lactide) PS Polystyrene REMP Ring-expansion metathesis polymerization REP Ring-expansion polymerization RI Refractive index ROP Ring-opening polymerization SAXS Small-angle X-ray scattering SEC Size exclusion chromatography SIMes 1,3-dimesitylimidazolin-2-ylidene Sn(Oct)2 Tin(II) ethylhexanoate TBD 1,5,7-triazabicyclo[4.4.0]dec-1-ene TEA Triethylamine THF Tetrahydrofuran TMS-Cl Trimethylsilylchloride Triaz Triazol-5-ylidene TU 1-(3,5-bis-trifluoromethylphenyl)-3-cyclohexylthiourea UV Ultraviolet VL -valerolactone WAXS Wide-angle X-ray scattering ZROP Zwitterionic ring-opening polymerization

xx CHAPTER 1

Zwitterionic Polymerization for Cyclic Polymers

1 1.1 Introduction In zwitterionic polymerization the counter ion is covalently bound to the growing chain. So, the propagating chain has both positively and negatively charged end groups.1 In 1960, Szwarc proposed polymerization of zwitterions and analogous dipolar species from observing the copolymerization of isobutene and cyanomethacrylate.2 While neither of the monomers polymerized on its own, the copolymerization yielded a 1:1 copolymer. Szwarc also predicted that cyclic tetramers could be formed in this reaction. Zwitterionic polymerization has been investigated for acrylonitrile, acrylic acid, malonate monomers with tertiary amine or phosphine initiators. The general characteristics of these systems have been discussed in a review by Johnston.3 The general mechanism for zwitterionic polymerization is shown in Scheme 1.1. The relative amount of each species will depend on the initiator–monomer combination, solvent and temperature of the reaction.

R3NCH2 C CH2 C R NCH C CH C CH C etc. R3N R3NCH2 C 3 2 2 2

C C R3N C CH2 R N 3 R3N

Scheme 1.1. General mechanism of zwitterionic polymerization.

Initiation, i.e. the formation of the zwitterion by reaction of the initiator and monomer is slower than propagation and reversible. When inert salts are added, the cyclic ion pairs dissociate and pair with the ions of the salt so that monomer addition is no longer opposed by loss of chain entropy or enthalpy required to separate the chain ends. So the addition of inert salt increases the rate of polymerization. The polarity (solvating power and dielectric constant) of the solvent also influences the polymerization. In solvents of low dielectric constant, the equilibrium concentration of the cyclic ion pairs will increase, so macrocycles are more likely to be formed.

2 Charge cancellation coupling of small bipolar species to form macrozwitterions, proposed by Szwarc, was demonstrated by aryl substituted sulfonium zwitterions (tetrahydrothiophenium hydroxide) (Scheme 1.2(a)).4 These zwitterions were heated to enable polymerization and a combination of cyclic and linear oligomers were observed. The ratio of linears to cycles increased with increasing the solvent dielectric. This work was extended by Odian and co-workers in studies where isolable zwitterions, 1-[4-[(4-hydroxy-1-naphthyl) thio]butyl]quinuclidinium hydroxide inner salts, were prepared by reaction of tetrahydro-1-(4-hydroxy-1-naphthyl)thiophenium hydrochloride salt and quinuclidine.5 Also, 2-mercaptoalkyl-2-oxazolines were polymerized via a zwitterion intermediate formed by intramolecular proton transfer followed by charge cancellation coupling (Scheme 1.2(b)).6

Scheme 1.2. Charge cancellation coupling polymerization of isolable zwitterions.

The ring-opening polymerization of cyclic monomers such as lactones and N- carboxyanhydrides also proceeds through zwitterionic intermediates. To elucidate the mechanism of ring-opening polymerization of lactones by tertiary amines, Kricheldorf used MALDI-TOF mass spectroscopy to study the ring-opening polymerization of pivalolactone by pyridine initiators.7 Linear chains having one pyridinium ion and one - CO2 ion as end groups were observed as products and no cyclic polymers were observed. These observations suggested that pyridine functions as a nucleophilic

3 initiator by ring opening the lactone to generate a zwitterionic carboxylate, which propagates by an anionic mechanism. (Scheme 1.3)

Scheme 1.3. Ring-opening polymerization of pivaloactone with pyridine initiators.

N-heterocyclic carbenes (NHCs) are a class of potent bases and nucleophiles.8- 11 It was shown that NHCs are good organocatalysts for transesterification reactions,12,13 which led to their investigation as catalysts for ring-opening polymerization of lactones.14,15 Polymerization of lactide using the N-heterocyclic carbene 1,3-dimesitylimidazol-2-ylidene (IMes) in the presence of an alcohol initiator efficiently produces linear polylactide. The polymerization is extremely rapid, well controlled and exhibits some features of living polymerization, such as linear relationship between molecular weight and conversion and narrow polydispersity. It has been proposed that it is more likely that the nucleophilic NHC activates the monomer toward attack from the initiating / propagating alcohol to form a zwitterion rather than protonation of the carbene with the initiating alcohol followed by the nucleophilic addition of the alkoxide (H-bonding alcohol activation mechanism). However, the H-bonding alcohol activation mechanism cannot be totally ruled out since theoretical calculations predicted that the H-bond alcohol activation mechanism has a lower barrier than the nucleophilic mechanism in the gas phase or in polar aprotic solvents.16 This mechanistic competition between nucleophilic and general- base mechanisms is commonly found in many nucleophilic/basic organic catalysts.17 Mechanistic studies to test for the viability of the nucleophilic mechanism

4 demonstrated that in the absence of alcohol initiators, IMes could mediate the zwitterionic ring-opening polymerization of lactide to generate cyclic poly(lactide)s. In this chapter, zwitterionic polymerization to prepare cyclic polymers will be discussed. Polymerization mechanism, the effect of the nature of the NHC and monomer and reactivity of different monomer towards different NHCs will be discussed. Zwitterionic polymerization for preparing various architectures will also be described.

1.2 Zwitterionic Polymerization for Cyclic Polymers Lactide polymerization with 1,3-dimesityimidazol-2-ylidene (IMes) The zwitterionic ring-opening polymerization of lactide (LA) with IMes yields cyclic poly(lactide)s of molecular weights between 5,000 – 30,000g/mol with polydispersities (polydispersity index, PDI = Mw/Mn) below 1.5 for monomer conversions less than 90 %.18 The molecular weights increase with increasing monomer conversion and the molecular weight distributions (PDI) are narrow, being less than 1.3 up to 80 % conversion.

(a) (b)

Figure 1.1. (a) First-order kinetic plots of various initial conditions. The data for [M]0

= 0.6 M, [I]0 = 0.006 M were fitted to determine the rate constants (solid line) and these rate constants were used to predict for different initial conditions (the dotted lines). (b) Molecular weight and polydispersity index versus conversion plot for two different initial conditions.

5 The molecular weight versus conversion plots exhibit a non-zero intercept near 0 % monomer conversion, and are relatively insensitive to the initial monomer-to-initiator ratio (Figure 1.1(b)). These observations suggest a slower rate of initiation relative to propagation, so that only a small fraction of the carbenes are converted to active zwitterions which propagate rapidly and liberate the carbene to generate cyclic poly(lactide)s. Scheme 1.4 shows the proposed mechanism for this reaction. In the proposed mechanism, initiation involves the reversible formation of the initial zwitterion followed by addition of second monomer to form Z2. Depropagation and termination by cyclization steps are also included.

Scheme 1.4. IMes mediated zwitterionic polymerization of lactide.

Kinetic and mechanistic investigations19 have been carried out, which -2 -1 implicate a mechanism that involves a slow initiation step (ki = 0.274 M s , second- -1 -1 order in [M]), and a propagation step (kp = 48.7 M s , first-order in [M]) that is much -1 -1 faster than initiation, cyclization (kc = 0.0575 s ) and depropagation (kd = 0.208 s ). In the proposed model, the narrow PDIs are explained by the fast propagation relative to the macrocyclization and the slow initiation at high monomer conversion due to a second-order dependence on [M] for initiation. The cyclization, although slow relative to the propagation, becomes significant at high conversions placing limits on the molecular weights that can be achieved. Simulations of the polymerization system based on the rate constants provided insight into the evolution of the concentration of

6 various species (active zwitterions, free carbene and cyclized chains) with conversion.

The results showed that the concentration of active species ([I]0-[I]) is relatively constant between conversions of 25 % to 75 %, which explains the linear increase in molecular weight with increasing conversion. These results strongly support that nucleophilic activation of the monomer by NHCs is viable. The selectivity for the formation of high molecular weight macrolactones, even at relatively high monomer concentrations, is likely a consequence of the enforced proximity of the zwitterionic chain-ends. Also, the kinetic model rationalizes some of the unique features of the zwitterionic ring-opening polymerization of lactide with IMes and provides a useful mechanistic framework to optimize these polymerizations as a strategy to generate well-defined cyclic polyesters. Consequently, the proposed model guides further investigations on the development of new NHC initiators that enable polymerization with faster initiation and slower cyclization as compared to IMes.

-Lactone polymerization with carbene1,3-dimesitylimidazolin-2-ylidene (SIMes) The ring-opening polymerization of 4-membered ring -lactones using the saturated carbene1,3-dimesitylimidazolin-2-ylidene (SIMes) yields cyclic poly(-lactone)s.20 This reaction proceeds via a novel mechanism that involves the reversible collapse of the zwitterionic intermediate to a neutral imidazolidine spirocycle.

Scheme 1.5. Zwitterionic ring-expansion polymerization of -lactones with SIMes.

7 Isolation of spirocycle S provides compelling evidence for the nucleophilic mechanism. These spirocyclics also serve as competent initiators for the zwitterionic ring-expansion polymerization of β-lactones to generate cyclic polyesters of well- defined molecular weight. Analysis of polymers obtained at different conversions shows that the SIMes-mediated zwitterionic polymerization of β-propiolactone displays characteristics of a living polymerization; molecular weights linearly track monomer conversion and the molecular weight distributions remain lower than 1.3 to high conversions. The remarkable degree of control of this polymerization is presumably due to the generation of a small amount of reactive zwitterionic intermediates by the reversible formation of spiro macrocycles. The low concentration of reactive alkoxides during polymerization suppresses side reactions, such as termination and elimination reactions analogous to modern controlled polymerizations.21,22

-Caprolactone polymerization using 1,3-dialkyl-4,5-dimethylimidazol-2-ylidenes The rates and selectivities of NHCs depend sensitively on both the nature of the carbene and the lactone monomer. The ring strain of the monomer, thermodynamics of the monomer ring-opening reaction, and the steric and electronic character of the NHC are a few of the factors that influence the choice of monomer and carbene combination. While the aryl-substituted carbene IMes is very active for lactide, it is much less active for -caprolactone (CL) and -valerolactone (VL). For these lactone monomers, more basic and less sterically hindered carbenes, 1,3-dialkyl-4,5- dimethylimidazol-2-ylidenes (where alkyl are methyl, ethyl and isopropyl) are more effective than IMes. Due to different relative rates of initiation and propagation between each system, different monomer and carbene combinations allow the synthesis of cyclic polymers in different molecular weight regimes. For the lactide/IMes system the molecular weights were limited to less than Mn = 30,000 g/mol. However, zwitterionic polymerization of CL with 1,3-dialkyl-4,5- dimethylimidazol-2-ylidenes yield cyclic poly(-caprolactone)s with molecular 23 weights up to Mn = 114,000 g/mol. The polydispersities of these polymers are

8 higher (Mw/Mn = 1.4 – 2.2) than in the LA/IMes system, especially at high conversion. This may be due to due to more pronounced transesterification by carbene and/or shuffling of growing chains.18 These high molecular weight cyclic poly(-caprolactone)s (PCL) allow us to study physical properties of cyclic polymers in the melt. To this end, crystallization of the cyclic PCL was investigated by synchrotron small-angle X-ray scattering (SAXS) experiments and differential scanning calorimetry (DSC).23

(a) (b)

Figure 1.2. Electron density correlation function K(z) for (a) linear PCL and (b) cyclic

PCL versus time during isothermal crystallization at 45 ºC. Lc is lamellar thickness and Lp is long period.

The cyclic PCLs prepared by zwitterionic polymerization were well characterized and highly entangled, the molecular weight being over 10 times the 24,25 entanglement molecular weight of linear PCL (Me ~ 3,000 g/mol). High molecular weight cyclic PCL crystallizes with a similar lamellar thickness and long period spacing as linear PCL (Figure 1.2), but the crystallization of cyclic PCL is faster than that of linear PCL for molecular weights greater than 75,000 g/mol (Figure 1.3). These results imply that the polymer topology does not have a significant influence on the crystal structure or morphology, but can have a significant influence on the rate of crystallization from the melt.

(a) (b)

Figure 1.3. (a) Crystal growth rates from SAXS and (b) half-times of crystallization as a function of Mw from DSC for cyclic and linear PCL isothermally crystallized (45 ºC).

Almost at the same time, two other reports on the crystallization of cyclic PCL compared to linear PCL were published. In the first work, Saalwächter et al.26 synthesized cyclic PCL of molecular weights between 50 and 80 kg/mol using the ring expansion polymerization (REP) of cyclic tin initiators. The linear analogues were prepared by cleavage of the Sn-O bond by ethanedithiol. This technique provided cyclic and linear PCL pairs of same number of monomer units. However, the characterization of the cyclic samples was not sufficiently provided; higher elution volume of cyclic PCL compared to linear gave evidence for cyclic topology but the MALDI-TOF mass spectra showed identical structures for linear and cyclic due to cleavage of Sn-O bond in the cyclic during measurement giving uncertainty in the purity of the cyclic sample. Also, it is not clear how the Sn inside the cyclic polymer would affect the physical properties. In the other study, Müller et al.27 used Cu(I)- catalyzed azide-alkyne cycloaddition (CuAAC) reaction to prepare the cyclic PCLs. The molecular weights ranged from 2 to 7.5 kg/mol, which by viscosity studies was shown to be the intermediate molecular weight range where the chains are not completely free of entanglements.28 This ‘click reaction’ route has the advantage of being able to provide exact linear and cyclic analogues, but is limited to relatively low

10 molecular weights to retain high efficiency of the coupling reaction and purity of the cyclic sample. Also, due to the low molecular weight, the influence of the triazole unit in the ring may not be negligible during crystallization. Despite the various factors that may influence the crystallization kinetics, both studies showed that cyclic PCL crystallized faster than the linear counterpart and the authors attributed this to faster diffusion or less entanglement of the cyclic chains, in agreement with our results.

1.3 Zwitterionic Polymerization for Different Architectures Cyclic block copoly(-peptoid)s Cyclic poly(-peptoid)s can be synthesized with controlled molecular weights and narrow polydispersities via NHC mediated ring-opening polymerization of N- substituted N-carboxyanhydrides.29 ESI-MS showed polymeric species with structure consistent with NHC-poly(-pepetoid) spirocyclic adduct and when treated with dithranol free cyclic poly(-peptoid)s were obtained.

Scheme 1.6. Living zwitterionic ring-opening polymerization of N-alkyl-N- carboxyanhydrides to give cyclic homo- (top path) and cyclic diblock copoly(- peptoid)s (bottom path).

11 The molecular weight increases linearly with conversion while the PDI remains narrow even at high conversions (Figure 1.4(a)). This suggests that the rate of cyclization to yield the cyclic polymer is much slower than propagation. The polyamide backbone, which is stable towards transamidation, contributes to the high stability of the cyclic poly(-peptoid)s towards inter- and intra-chain transfer reactions that broaden the polydispersity and terminate the growing chains. The reaction is proposed to go through a living zwitterionic propagating species with minimal chain transfer.

(a) (b)

Figure 1.4. (a) Representative molecular weight and polydispersity index versus conversion plot. (b) SEC chromatograms of poly(N-Bu-glycine) before (M/I = 50, dashed curve) and after (M/I = 25, solid curve) the chain extension experiment. The dot-dashed curve shows the data for an independently synthesized polymer sample of M/I = 75. (Figures reprinted with permission from J. Am. Chem. Soc. 2009, 131, 18072-18074. Copyright 2009, American Chemical Society.)

The livingness of the zwitterionic species was also tested by a chain extension experiment. Cyclic diblock copoly(-peptoid)s were obtained via sequential monomer addition. Each polymerization step was taken to complete conversion. The size exclusion chromatography (SEC) analyses showed increase in molecular weight upon

12 addition of second monomer and the polydispersity remained narrow. No homopolymers were obtained suggesting 100 % initiation of the NHC in the first step.

Cyclic gradient copolyesters Unlike the cyclic poly(-peptoid) system described in the previous section, NHC- mediated zwitterionic polymerization of cyclic esters, such as lactide and - caprolactone, shows less than 100 % initiation of the carbene. Also, the rate of cyclization is not negligible so that there would be carbene liberated upon cyclization on the time scale of the reaction. When chain extension experiments were conducted for the lactide/IMes system,19 it was shown that chain growth of the propagating zwitterions was accompanied by polymerization by residual carbene. Here, the residual carbene can be the uninitiated carbene initiators (due to slow initiation) and those liberated upon cyclization. Consequently, the sequential addition of different monomers would result in a blend of cyclic block and homopolymers. Stochastic simulations based on the proposed kinetic model of this system predicted the number of various species during the polymerization. The calculations are in agreement with the experimental results in that only a fraction of the carbenes initiate to form zwitterions (33 % at [M]0 = 0.6 M, 50 % at [M]0 = 0.9 M) and at high conversion the amount of free carbene initiator increases, implying the liberation of carbene initiator upon cyclization. It is clear that a pure cyclic block copolymer cannot be prepared using sequential addition of monomers for the cyclic polyester systems. So, a different approach that takes advantage of the different reactivity of monomers towards a N- heterocyclic carbene initiator was employed in attempt to synthesize cyclic block polyesters. It has been shown that different monomers show different reactivities towards organocatalysts when copolymerized.30 This is in contrast to metal catalysts where there is no selectivity observed.31,32 This kind of different reactivity has also been discovered for a NHC catalyst system, where -valerolactone (VL) and -caprolactone (CL) were copolymerized using 1,3,4,5-tetramethylimidazol-2-ylidene as the catalyst

13 and benzyl alcohol as the initiator. Estimates for the reactivity ratios for VL and CL were determined to be rVL = 9.0 and rCL = 0.24 by the Fineman-Ross method. The large difference in reactivity ratios between VL and CL suggested that this system would be competent for a one-pot block copolymerization. The zwitterionic copolymerization of VL and CL was carried out with the N- heterocyclic carbene 1,3-diethyl-4,5-dimethyl-imidiazol-2-ylidene (Me2IEt) in the absence of an alcohol initiator in toluene solution at room temperature.33

Scheme 1.7. Zwitterionic ring-opening copolymerization of -valerolactone and - caprolactone.

The cyclic copolymers obtained were not the expected block copolymers, rather they showed a gradient, the composition of the monomers within the polymer chain gradually changing from VL to CL. The polymerization of each monomer showed characteristics of batch gradient copolymerization,34,35 where the composition of the monomers constantly changed with conversion. The gradient sequences and ability of VL and CL to co-crystallize lead to cyclic copolymers with melting points that are slightly lower than the homopolymers but higher than the random copolymers. This large difference in reactivity between VL and CL with NHC catalysts, coupled with sufficiently long lifetimes of the growing zwitterions provides an expedient synthesis of gradient cyclic copolymers (cyclic P(VL-grad-CL)) comprised of VL-rich sequences that transition to CL-rich sequences within a cyclic macromolecule. Due to the fact that these copolymers are cyclic, a sharp comonomer interface and a gradient interface exist within the same molecule (Scheme 1.7). This

14 synthetic approach provides a strategy for generating unusual topologies and sequences, whose properties need further investigation.

,-Difunctionalized linear poly(ethylene oxide)s It has been shown that ethylene oxide can be polymerized by NHCs such as 1,3- diisopropylimidazol-2-ylidene (IPr).36 In this system, cyclization of the growing zwitterion was not observed. Linear poly(ethylene oxide) was exclusively obtained. By choosing the appropriate terminating agent, a variety of -difunctionalized PEOs, such as -bis(hydroxy)-telechilic, -azido--hydroxy- and -benzyl-- hydroxy-PEOs were prepared. A nucleophilic mechanism to generate a zwitterionic imidazol-2-ylidinium alkoxide was proposed. Linear diblock copolymer, poly(ethylene oxide)-b-poly(-caprolactone) was synthesized by sequential ZROP (zwitterionic ring-opening polymerization) of the corresponding monomers, without isolation of the first PEO block. The -difunctionalization and the ability to prepare diblock copolymers provide indirect evidence for a living zwitterionic intermediate.

Scheme 1.8. Zwitterionic ring-opening polymerization of ethylene oxide.

Considering the proposed mechanism, it is not at all obvious why this system does not cyclize to give cyclic PEO. The rate of cyclization may be much slower compared to that in other systems such as lactide/IMes, but the choice of the appropriate quenching agent could induce cyclization to produce cyclic PEO.

1.4 Future Outlook Most of the cyclic polymers prepared by zwitterionic polymerization presented here are polyesters. For post-polymerization chemistry on the cyclic polymers and more

15 practical applications, the monomer scope needs to be expanded to give more robust polymer backbones. To this end, carbosiloxanes are good monomer candidates. NHCs were also shown to catalyze the ring-opening polymerization of carbosiloxanes in the presence of initiating alcohols to give linear poly(carbosiloxane)s.37 This system has been explored in the absence of alcohol initiators and shows potential for preparation of high molecular weight cyclic poly(carbosiloxane)s. Zwitterionic polymerization could also provide a facile route to synthetic mimics of biological molecules. Ring-opening polymerization of phospholane monomers38 to give cyclic poly(phosphoester)s could provide a synthetic mimic of cyclic DNA. Also, previous studies on the polymerization of N-carboxyhydrides39 and dithiolane-2,4-diones40 using amine initiators, such as pyridine, to give cyclic polypeptides and cyclic poly(thioglycolide)s, respectively, can serve as a starting point for well-controlled zwitterionic polymerization to prepare cyclic biopolymers.

Figure 1.5. Potential monomers to broaden the area of zwitterionic polymerization.

16 1.5 References (1) Odian, G. Principles of Polymerization; 4th ed.; Wiley-Interscience: Hoboken, 2004. (2) Szwarc, M. Makromol. Chem. 1960, 35, 132-158. (3) Johnston, D. S. Adv. Polym. Sci. 1982, 42, 51-106. (4) Schmidt, D. L.; Smith, H. B.; Yoshimin. M; Hatch, M. J. J. Polym. Sci., Part A: Polym. Chem. 1972, 10, 2951-2966. (5) Cangiano, D. L.; Odian, G.; Schmidt, D. L. J. Polym. Sci., Part A: Polym. Chem. 1996, 34, 801-809. (6) Gunatillake, P. A.; Odian, G.; Tomalia, D. A. Macromolecules 1987, 20, 2356- 2362. (7) Kricheldorf, H. R.; Garaleh, M.; Schwarz, G. J. Macromol. Sci., Pure Appl. Chem. 2005, A42, 139-148. (8) Arduengo, A. J. Acc. Chem. Res. 1999, 32, 913-921. (9) Bourissou, D.; Guerret, O.; Gabbai, F. P.; Bertrand, G. Chem. Rev. 2000, 100, 39-91. (10) Enders, D.; Niemeier, O.; Henseler, A. Chem. Rev. 2007, 107, 5606-5655. (11) Herrmann, W. A.; Kocher, C. Angew. Chem. Int. Ed. 1997, 36, 2162-2187. (12) Grasa, G. A.; Kissling, R. M.; Nolan, S. P. Org. Lett. 2002, 4, 3583-3586. (13) Nyce, G. W.; Lamboy, J. A.; Connor, E. F.; Waymouth, R. M.; Hedrick, J. L. Org. Lett. 2002, 4, 3587-3590. (14) Connor, E. F.; Nyce, G. W.; Myers, M.; Mock, A.; Hedrick, J. L. J. Am. Chem. Soc. 2002, 124, 914-915. (15) Nyce, G. W.; Glauser, T.; Connor, E. F.; Mock, A.; Waymouth, R. M.; Hedrick, J. L. J. Am. Chem. Soc. 2003, 125, 3046-3056. (16) Lai, C. L.; Lee, H. M.; Hu, C. H. Tetrahedron Lett. 2005, 46, 6265-6270. (17) Kiesewetter, M. K.; Shin, E. J.; Hedrick, J. L.; Waymouth, R. M. Macromolecules 2010, 43, 2093-2107. (18) Culkin, D. A.; Jeong, W. H.; Csihony, S.; Gomez, E. D.; Balsara, N. R.; Hedrick, J. L.; Waymouth, R. M. Angew. Chem. Int. Ed. 2007, 46, 2627-2630.

17 (19) Jeong, W.; Shin, E. J.; Culkin, D. A.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2009, 131, 4884-4891. (20) Jeong, W.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2007, 129, 8414-8415. (21) Matyjaszewski, K.; Xia, J. H. Chem. Rev. 2001, 101, 2921-2990. (22) Moad, G.; Rizzardo, E.; Thang, S. H. Aust. J. Chem. 2005, 58, 379-410. (23) Shin, E. J.; Jeong, W.; Brown, H. A.; Koo, B. J.; Hedrick, J. L.; Waymouth, R. M. Macromolecules 2011, 44, 2773-2779. (24) Gimenez, J.; Cassagnau, P.; Michel, A. J. Rheol. 2000, 44, 527-547. (25) Ramkumar, D. H. S.; Bhattacharya, M. Polym. Eng. Sci. 1998, 38, 1426-1435. (26) Schaler, K.; Ostas, E.; Schroter, K.; Thurn-Albrecht, T.; Binder, W. H.; Saalwachter, K. Macromolecules 2011, 44, 2743-2754. (27) Cordova, M. E.; Lorenzo, A. T.; Muller, A. J.; Hoskins, J. N.; Grayson, S. M. Macromolecules 2011, 44, 1742-1746. (28) Izuka, A.; Winter, H. H.; Hashimoto, T. Macromolecules 1992, 25, 2422-2428. (29) Guo, L.; Zhang, D. H. J. Am. Chem. Soc. 2009, 131, 18072-18074. (30) 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, 8574-8583. (31) Storey, R. F.; Herring, K. R.; Hoffman, D. C. J. Polym. Sci., Part A: Polym. Chem. 1991, 29, 1759-1777. (32) Storey, R. F.; Hoffman, D. C. Makromol. Chem. Macromol. Symp. 1991, 42-3, 185-193. (33) Shin, E. J.; Brown, H. A.; Gonzalez, S.; Jeong, W.; Hedrick, J. L.; Waymouth, R. M. Angew. Chem. Int. Ed. 2011. (34) Matyjaszewski, K.; Ziegler, M. J.; Arehart, S. V.; Greszta, D.; Pakula, T. J. Phys. Org. Chem. 2000, 13, 775-786. (35) Zaremski, M. Y.; Kalugin, D. I.; Golubev, V. B. Polym. Sci. Ser. A 2009, 51, 103-122.

18 (36) Raynaud, J.; Absalon, C.; Gnanou, Y.; Taton, D. J. Am. Chem. Soc. 2009, 131, 3201-3209. (37) Lohmeijer, B. G. G.; Dubois, G.; Leibfarth, F.; Pratt, R. C.; Nederberg, F.; Nelson, A.; Waymouth, R. M.; Wade, C.; Hedrick, J. L. Org. Lett. 2006, 8, 4683-4686. (38) Penczek, S.; Klosinski, P. In Model of Biopolymer by Ring-Opening Polymerization; Penczek, S., Ed.; CRC Press: Boca Raton, FL, 1990, p 291. (39) Kricheldorf, H. R.; Von Lossow, C.; Schwarz, G. J. Polym. Sci., Part A: Polym. Chem. 2006, 44, 4680-4695. (40) Kricheldorf, H. R.; Lomadze, N.; Schwarz, G. Macromolecules 2007, 40, 4859- 4864.

20 CHAPTER 2

Chain Extension Experiments and Solvent Effects on N-Heterocyclic Carbene Mediated Zwitterionic Polymerization of Lactide

21 2.1 Introduction Cyclic macromolecules have fascinated chemists, biologists, and materials scientists for decades.1-3 The simple topological constraint of connecting the ends of a large linear macromolecule into a ring has a significant influence on the structure, dynamics and properties of the polymer, but our understanding of the origin of these effects is only beginning to take shape.3-5 The synthetic challenges in generating high molecular weight cyclic polymers are formidable as the unfavorable entropy of cyclization presents both thermodynamic and kinetic constraints for large chains.6 Experimental efforts to generate large ring macromolecules typically involve the coupling of two reactive ends of a linear telechelic precursor at high dilutions,2,3,7-10 or the use of polymeric supports.11,12 More recently, ring expansion strategies have been devised utilizing cyclic ruthenium13-16 and cyclic tin alkoxide initiators17 to generate high molecular weight cyclic polymers. We recently reported the zwitterionic ring-opening polymerization of lactones by N-heterocyclic carbenes as a strategy to generate cyclic polyesters (Scheme 2.1).18,19

Scheme 2.1. Proposed zwitterionic polymerization of lactide to generate cyclic poly(lactide).

Zwitterionic polymerizations are a class of ionic polymerizations where both the anion and cation are attached to the same polymer chain.20,21 Szwarc had suggested, as early as 1960, that zwitterionic polymerizations might lead to cyclic structures,22 but only recently was this demonstrated experimentally.18,23-25 In contrast

22 to many other zwitterionic polymerizations,20,21 the ring-opening polymerization of lactones by N-heterocyclic carbenes generates cyclic polylactones with high purity and exceptional control of molecular weight and molecular weight distribution.18,23 The cyclic structure of the products were confirmed by various experimental techniques including 1H NMR spectroscopy, mass spectrometry, dynamic light scattering (DLS), and gel permeation chromatography (GPC).18

(a) (b)

Figure 2.1. (a) First-order kinetic plot of the polymerization data at [M]0 = 0.6 M and

[I]0 = 0.006 M. Each point is the average of three runs. The line is a linear fitting. (b)

Plots of Mn and PDI versus monomer conversion. The line is the theoretically predicted molecular weight for an ideal living polymerization.

The zwitterionic polymerization of lactide exhibits several notable features (Figure 2.1) : (1) the polymerization rates are rapid, reaching complete conversion within minutes, (2) relatively high molecular weight macrocycles are generated even at high initial monomer concentrations [M]0 = 0.6 – 1.0 M (in THF), (3) the molecular weight distributions are narrow (polydispersities, PDI = Mw/Mn < 1.5) and (4) the molecular weights (Mn = 5 – 30 kg/mol) increase with monomer conversion. The proposed mechanism for these reactions (Scheme 2.1) involves the nucleophilic attack of the carbene initiator (I = IMes) to generate an initial zwitterionic intermediate. Addition of monomers to the zwitterion leads to chain growth by generation of higher zwitterions (Zn); macrocyclization of the zwitterions (Zn) generates cyclic

23 poly(lactide)s (Cn) with liberation of the carbene. This proposal provides a plausible explanation for the generation of cyclic polymers; nevertheless there are several features that are seemingly at odds with the observed polymerization behavior. In particular, the evolution of molecular weight with monomer conversion and the narrow molecular weight distributions are indicative of a chain-growth process with minimal chain transfer. If the rate of cyclization (Rc) were competitive with the rate of propagation (Rp), then cyclization would constitute a chain-transfer event since the liberated carbene should initiate the growth of another chain. Under these conditions, the molecular weights would be expected to remain constant with monomer conversion and the molecular weight distribution should correspond to one predicted 26 for chain growth with chain transfer, that is Mw/Mn = 2.0. As this is not observed experimentally, several unusual kinetic criteria need to be met to explain the observed polymerization behavior. In a report on the detailed kinetics and mechanism of this polymerization,19 the time evolution of the monomer concentration, molar mass and PDI were analyzed by means of numerical and stochastic simulations for the initiation, propagation, termination, and chain transfer steps. Numerical fitting of the lactide polymerization data at [M]0 = 0.6 M and [I]0 = 0.006 M (up to more than 90 % conversion) to the differential equations d[I] ‐  k [I][M ]2  k [I] [I] dt i c 0

d[M ] 2 ‐  2ki[I][M ]  k p[M ] kd [I]0 [I] dt provides good agreement with the kinetic data at various concentrations and yields the -2 -1 -1 -1 -1 estimates for the rate constants: ki = 0.274 M s , kp = 48.7 M s , kd = 0.208 s , and -1 kc = 0.0575 s . This kinetic model illuminates a number of unique features of these ring-opening polymerization reactions. First, the polymerization rates are very high; -1 -1 the rate constant for propagation (kp = 48.7 M s ) is on the order of enzymatic rate -1 27 constants for ring-opening polymerization of lactide (kcat = 2-225 s ) and faster than 28 -1 -1 29 -1 -1 that for metal alkoxides (NaOR, kp = 0.7 M s at 80 °C; ZnOR, kp = 2.2 M s ).

24 Second, the narrow PDIs are a consequence of the high rate of propagation relative to cyclization (Rp/Rc = kp[M]/kc = 850[M]) and the inefficiency of re-initiation by the carbene initiator liberated upon cyclization, due to the low rate of initiation that is second order in [M]. This polymerization system was further explored by performing chain extension reactions, which are described in the following section. Experiments in which more monomer was added during polymerization in combination with stochastic simulations gave more insight into the speciation of the carbene initiator. Also, some polymerization data in a different solvent other than tetrahydrofuran is presented to show indirect evidence of the propagating species being the proposed zwitterion.

2.2 Chain Extension The addition of more lactide monomer to the living growing chain during polymerization should result in increase in molecular weight. Also, if different kinds of lactides (rac-lactide, L-lactide or D-lactide) are used, the products would be cyclic diblock copolymers.

Polymerizations using L-lactide and rac-lactide as the two monomers via sequential addition using the N-heterocyclic carbene, 1,3-dimesitylimidazol-2-ylidene, as the initiator, were carried out. It has already been shown that IMes polymerizes L- 18 lactide with retention of stereochemistry, producing isotactic poly(L-lactide). So, these polymerizations are expected to produce cyclic copolymers with isotactic and atactic blocks (entries 2 – 5, Table 2.1). The cyclic topology was checked by comparing the dilute solution viscosities with the linear analogues (see Experimental Section). The cyclic samples showed lower intrinsic viscosities compared to the linear in agreement with previous experiments and theoretical predictions.5 It should be mentioned that the reaction times are so short, it is practically impossible to take an aliquot of the first part of the polymerization and add the second monomer simultaneously. So, when an aliquot sample was needed, a separate polymerization of the first part of the sequential polymerization was performed.

25 Table 2.1. Selected polymerization data.

Table 2.2. Characterization data for samples in Table 2.1.

26 The cyclic poly(L-lactide-b-rac-lactide) copolymers showed lower melting points and optical rotation values compared to those of the homopolymer (PLLA) (entry 1, Table 2.1). The melting points (Table 2.2) showed no significant depression with decreasing L-lactide content, which is indicative of the presence of the isotactic block.

Figure 2.2. Gel permeation chromatograms of PLLA homopolymer (entry 1, dashed) and P(LLA-b-DLA) copolymer (entry 7, solid).

In addition, cyclic copolymers of L-lactide and D-lactide (poly(L-lactide-b-D- lactide), P(LLA-b-DLA)) were synthesized (entries 6 – 7, Table 2.1). A representative gel permeation chromatogram of a cyclic copolymer (entry 7) is compared with that of a polymer from the reaction quenched at the first step ([M]0 = 0.6 M, M/I = 100, t = 30 s). Figure 2.2 shows that the molecular weight of the polymer increased with addition of the second monomer. This means that on this time scale the propagating species is living. The lower molecular weight shoulder (on the right) in the GPC trace of the cyclic copolymer indicates that upon addition of the second batch of monomer, the uninitiated (residual) carbenes are initiated (rate of initiation is second-order in monomer) resulting in the production of small molecular weight cyclic homopolymers of the second monomer. The possibility of this peak coming from “dead” chains

27 formed by cyclization (kc) cannot be ruled out. So, the polymer produced by sequential polymerization is more likely a blend of the block copolymer and homopolymer of the second monomer. Cyclic P(LLA-b-DLA) exhibits a melting point much higher than that of cyclic PLLA or PDLA homopolymers.30 This may be due to the combination of (1) the stereocomplex interaction31 between different or within copolymer chains and (2) the interaction between the PLLA block of the copolymer and the small PDLA homopolymers. It is known that blends of linear PLLA and linear PDLA form stereocomplexes that crystallize into more densely packed crystals showing higher melting temperatures and different physical properties compared to the parent homopolymers.31,32 The small peak at the lower temperature (136 oC) is thought to come from the homopolymer produced by re-initiation of carbene or the cyclized “dead” chains, which is consistent with the GPC results.

Figure 2.3. Differential scanning calorimetry traces of P(LLA-b-DLA) copolymer (entry 7).

Stochastic simulations33-35 were performed to provide more insight into this re- initiation phenomenon. The simulations provide information on the estimation of the 36 amount of free carbenes (I), active zwitterions (Zn), and “dead” chains (Cn) at any

28 monomer conversion (given initial inputs of [M]0, [I]0, and rate constants; see

Experimental Section). For example, kinetic simulations for polymerizations at [M]0

= 0.6 M and [I]0 = 0.006 M reveal that even at 50 % monomer conversion, a significant fraction of initiators still remain (Ni/Ni,0 = [I]/[I]0 = 0.67; Ni,0 is the initial number of initiators and Ni is the number of initiators) due to the slow initiation relative to propagation (the solid line in Figure 2.4). At higher monomer conversion, the fraction of initiators increases due to liberation of the carbene initiator by macrocyclization. Importantly, as initiation at high monomer conversion is suppressed because of low [M] and the second-order dependence of initiation on [M], initiators start accumulating after 50 % monomer conversion. The observation that this system exhibits some characteristics of “living polymerization” but deviates from living behavior can be readily understood by considering the evolution of the amount of active species during the polymerization (the dashed lines, Nz/Ni,0, in Figure 2.4).

The fraction of active zwitterions (Nz/Ni,0) increases initially and reaches a maximum (for example, 0.33 in Figure 2.4(a) and 0.50 in Figure 2.4(b) at 50 % monomer conversion). Subsequently, the fraction of active zwitterions decreases upon cyclization of the macrozwitterions.

(a) (b)

Figure 2.4. Stochastic simulations of polymerization kinetics at (a) [M]0 = 0.6 M and

[I]0 = 0.006 M, (b) [M]0 = 0.9 M and [I]0 = 0.006 M. Ni,0 is the initial number of initiators, and Ni, Nz, and Nc are the number of initiators, zwitterions, and cyclized chains, respectively.

29 The molecular weight is formally given by the ratio of the concentration of converted monomer ([M]0 – [M]) to the concentration of active species ([I]0 – [I]). Since [I]0 – [I] changes with time, the evolution of molecular weight with monomer conversion would not be expected to follow that for a “living” polymerization where [I]0 – [I] is constant and typically equal to [I]0. The observed increase in molecular weight with monomer conversion is a consequence of the slow cyclization (compared to propagation) and the relatively constant [I]0 – [I] between 25 – 75% monomer conversion.

The simulation results for polymerizations at higher [M]0 (Figure 2.4(b)) clearly show an important consequence of the second-order dependence of initiation on [M]: polymerization at a higher initial monomer concentration generates a larger fractional amount of active species. The simulations help define the kinetic conditions under which chain extension of the active zwitterions might be expected. To test these predictions, chain extension experiments were carried out by adding a second charge of monomer during the course of the polymerization (Scheme 2.2). An initial polymerization of rac-lactide (0.70 mmol) was performed in a THF solution (1.16 mL, [M]0 = 0.6 M, and [I]0 = 0.006 M). After t1 = 60 s (94 % monomer conversion), a 1.2 M THF solution of rac-lactide (1 mL) was added (bringing the monomer concentration back to 0.6 M), and then stirring was continued for an additional 30 s (t2) before being quenched with CS2. It was observed that only 30 % of monomer was converted in the second polymerization. Analysis of the crude samples by GPC reveals that the addition of more monomer causes the formation of a small shoulder in the low molecular weight regime, but no discernible change in the main peak (Figure 2.5(a)). The stochastic simulations based on the proposed mechanistic model predict Ni/Ni,0 = 0.95, Nz/Ni,0 = 0.05, and Nc/Ni,0 = 0.55 for the first polymerization after t1 = 60 s (94 % monomer conversion, Figure 2.4(a)). Notably, the simulations suggest that the “dead” chains cyclized during the first polymerization constitute more than 90 % of the main peak at 94 % monomer conversion. Also, the experimentally observed most probable molecular weight (Mp) of the second shoulder

30 peak is 7.8 kg/mol, which is close to that for the polymerization at low monomer conversion. These experimental and simulation results implicate that the addition of additional monomer after 94 % conversion results in re-initiation rather than chain extension and that the shoulder observed at lower molecular weight (Figure 2.5(a)) corresponds to re-initiated polymers.

Scheme 2.2. Sequential polymerization of lactide.

(a) (b)

Figure 2.5. GPC traces of crude products from the polymerizations (a) (1) (dashed) and (2) (solid) and (b) (3) (dashed) and (4) (solid) in Scheme 2.2.

31 In contrast, if the initial polymerization is carried out at higher monomer concentration and the second monomer charge is added at a lower conversion, chain- extension is observed. For example, polymerization of rac-lactide (1.04 mmol) in

THF at higher initial monomer concentration (1.15 mL, [M]0 = 0.9 M, and [I]0 = 0.006 M) for 10 seconds proceeded to 59 % conversion (Figure 2.5(b)). The same polymerization is carried out, but a second monomer charge (1.15 mL of a 0.9 M THF solution) is added after t1 = 10 s (making the monomer concentration 0.6 M), and the reaction was continued for an additional 30 s (t2) before being quenched with CS2. Analysis of the crude samples by 1H NMR spectroscopy reveals that 64 % of monomer was converted during the polymerizations (49 % of monomer was converted in the second polymerization; see Experimental Section). GPC traces of the crude samples show that the addition of more monomer causes an increase in the molecular weight of the main peak and the formation of a small shoulder in the low molecular weight regime (Figure 2.5(b)). These results indicate that under these conditions, chain extension of the propagating zwitterions occurs, but is accompanied by some re-initiation of polymerization by the residual carbenes in solution. These observations are consistent with the simulations (Figure 2.4(b)) that predict a significant fraction of both active zwitterions (Nz/Ni,0 = 0.48) and carbenes (Ni/Ni,0 = 0.52) at 59 % conversion. The shoulder observed at Mp = 13.8 g/mol (Figure 2.5(b)) is likely due to re-initiation from the carbenes in solution, but may also include the “dead” chains cyclized during the first polymerization (Nc/Ni,0 = 0.18). While it is likely that the kinetic simulations do not represent all of the kinetic features of these polymerizations, the agreement between the predictions of the kinetic simulations and the chain extension experiments provide strong support for the kinetic model and illuminate the factors that define the kinetic lifetime of the active zwitterions. The experiments show that under certain conditions the propagating species is living allowing for growth of the chain upon addition of more monomer, but it is always accompanied by re-initiation of residual carbene. So, to achieve a pure cyclic block copolymer, a different polymerization should be used.

32 2.3 Nature of Zwitterions: Solvent Effect Due to the extremely rapid polymerization rates, the characterization of the zwitterionic intermediates proposed in Scheme 2.1 has proven challenging. Indirect evidence for the formation of zwitterions was provided by the isolation of spiro cycles derived from β–lactones and saturated N-heterocyclic carbenes,23 and the reaction of IMes with benzoyl chloride to generate the imidazolium salts.37 Nevertheless, the intermediacy of zwitterions provides a rationale for the efficiency of macrocyclization to generate the cyclic polymers as the ion-pairing of the chain ends can compensate for the unfavorable entropy of maintaining the chain ends in close proximity.6,38

(a) (b)

Figure 2.6. (a) First-order kinetic plots of the polymerization data in THF and in 1,4- dioxane at [M]0 = 0.6 M and [I]0 = 0.006 M. (b) Plots of Mn and PDI versus monomer conversion for the polymerization data in THF and in 1,4-dioxane at [M]0 = 0.6 M and [I]0 = 0.006 M. The lines are linear fittings for Mns.

As the solvation of the zwitterionic pairs should influence the relative rate of propagation and cyclization, the effect of different solvents on the polymerization behavior was investigated. For these studies, polymerizations of rac-lactide in 1,4 dioxane were performed at [M]0 = 0.6 M and [I]0 = 0.006 M and compared with polymerizations in THF under the same conditions. Figure 2.6(a) shows that the polymerization rate in 1,4-dioxane is slower than that in THF. Furthermore,

33 polymerizations in 1,4-dioxane generated PLAs of higher molecular weight at the same conversions relative to polymerizations in THF (Figure 2.6(b)). The faster rates observed in THF might be explained in terms of the better solvating properties of THF (dielectric constant: 7.6, dipole moment: 1.8 D, DN (donor number): 20) relative to 1,4-dioxane (dielectric constant: 2.2, dipole moment: 0.45 D, DN: 15).39 The cationic and anionic chain ends of the zwitterionic propagating species might be expected to dissociate more readily in THF than in 1,4- dioxane, thereby resulting in the faster chain growth.20 Faster polymerization rates in solvents of higher dielectric constant have been observed in other zwitterionic polymerizations.20 The higher molecular weights obtained in 1,4-dioxane might be attributed to slower initiation in 1,4-dioxane relative to THF. However, further studies in a broader range of solvents are warranted and are the subject of ongoing investigations.

2.4 Summary In summary, the zwitterionic ring-opening polymerization of lactide provides a fast and efficient method for the synthesis of cyclic poly(lactide)s of defined molecular weight and molecular weight distribution. Kinetic and mechanistic investigations reveal that the rate of initiation is second-order in [M] and slower than propagation, and the rate of propagation is faster than that of cyclization. Chain extension experiments and simulations both show the presence of a living species that is able to grow upon addition of more monomer. Also, there is a significant amount of uninitiated carbene present, such that when more monomer is added, some of these initiate to result in a mixture of high molecular weight chain extended cyclic copolymers and low molecular weight cyclic polymers. Polymerizations in solvents with different dielectric constants showed different rates and molecular weights, giving indirect evidence for the zwitterionic intermediate.

34 2.5 Experimental Section General Considerations. All reactions and polymerizations were performed in a drybox or with Schlenk techniques under nitrogen. 1H nuclear magnetic resonance (NMR) spectra were recorded at room temperature on either Varian 400 or 500 MHz spectrometer, with shifts reported in parts per million downfield from tetramethylsilane and referenced to the residual solvent peak. Gel permeation chromatography (GPC) was performed in tetrahydrofuran (THF) at a flow rate of 1.0 mL/min on a Waters chromatograph equipped with four 5 μm Waters columns (300 mm×7.7 mm) connected in series. A Viscotek S3580 refractive index detector and Viscotek GPCmax autosampler were employed. The system was calibrated using monodisperse polystyrene standards (Polymer Laboratories). The system with a triple detection system (Viscotek, Houston, TX) including a light scattering detector (right- angle and low-angle light scattering) and viscometer was used to measure the intrinsic viscosity with molecular weight of the polymers. Correction for any angular dissymmetry factor in the RALS data was performed in the TriSEC software using the viscometer signal. The angular dissymmetry correction is negligible because the polymers studied are relatively small compared to the laser wavelength (610 nm).

Materials. Tetrahydrofuran (THF) was distilled from sodium/benzophenone after being degassed three times via freeze-pump-thaw cycles. 1,4-dioxane was dried over sodium and distilled. Then it was distilled from sodium/benzophenone after being degassed via freeze-pump-thaw cycles. Carbon disulfide (CS2) was purchased from

Aldrich and distilled from CaH2. rac-Lactide, L-lactide and D-lactide were supplied from Purac, and purified by drying over CaH2 in a THF solution, and subsequent sublimations. The monomer was stored in a drybox under nitrogen. Deuterated chloroform (CDCl3) was purchased from Acros and used as received. Anhydrous methanol was purchased from Aldrich and used as received. 1,3-dimesitylimidazol-2- ylidene (IMes) was prepared according to the literature procedure.40

35 Representative procedure for polymerization of rac-lactide with IMes. In a drybox, a solution of IMes (1.06 mg, 3.46 x 10-3 mmol) in THF (or 1,4-dioxane) (80 μL) (from a stock solution) was added to a stirred solution of rac-lactide (50.1 mg, 0.348 mmol) and mesitylene (4.07 mg, 0.0339 mmol; an internal standard) in THF (or 1,4-dioxane) (0.5 mL) (from a stock solution). The reaction mixture was stirred for the desired reaction time at room temperature. Carbon disulfide (0.1 mL, 1.7 mmol) was added to terminate the reaction. The solvent was removed and the crude polymer was analyzed by 1H NMR spectroscopy and GPC. The conversion was determined both by using the internal standard and using the monomer and polymer peaks of the methine proton (-CH-) in 1H NMR spectra. The two methods showed good agreement.

Representative procedure for sequential polymerization of lactide with IMes. In a drybox, a solution of IMes (2.12 mg, 6.92 x 10-3 mmol) in THF (0.16 mL) was added to a stirred solution of L-lactide (100 mg, 0.696 mmol) in THF (1.0 mL). The reaction mixture was stirred for the desired reaction time (t1) at room temperature. Then, a solution of rac-lactide (100 mg, 0.696 mmol) in THF (1.0 mL) was added and stirring was continued for t2 at room temperature. CS2 (0.2 mL, 3.4 mmol) was added to terminate the reaction. The solvent was removed and the crude polymer was analyzed by 1H NMR spectroscopy and GPC. The conversion was determined by using the monomer and polymer peaks of the methine proton (-CH-) in 1H NMR spectrum. When needed, polymerizations were performed on a large scale, and the polymers were precipitated with a copious amount of anhydrous methanol and dried under vacuum.

36 (a) (b)

Figure 2.7. Mark-Houwink plots of linear and cyclic (a) poly(L-lactide) homopolymers; []cyclic/[]linear = 0.74 and (b) poly(L-lactide-b-D-lactide) copolymers;

[]cyclic/[]linear = 0.75.

Detailed procedure for polymerization in Scheme 2.2 and Figure 2.5 For (a): (1) A solution of IMes in THF (0.006 M) was added to a solution of LA in THF (0.6

M) and stirred for 60 seconds and then quenched by addition of CS2. The conversion was 93.5 % and the GPC trace of this reaction is the dashed line in Figure 2.5(a). (2) A solution of IMes in THF (0.006 M) was added to a solution of LA in THF (0.6 M) and stirred for 60 seconds. Then a solution of LA in THF with a concentration of 1.2 M was added to the reaction solution (making the total monomer concentration to become approximately 0.6 M again) and stirred for 30 seconds. Then the reaction was quenched by addition of CS2. The total conversion was 51.7 % and the conversion of the second batch of monomer was 29.8 %. The GPC trace of this reaction is the solid line in Figure 2.5(a).

According to the simulations, under the conditions [M]0 = 0.6 M, [I]0 = 0.006 M at 93.5 % conversion (Figure 2.4(a)) there is only about 10 % zwitterions and the rest is uninitiated carbene. So, when the second solution of monomer is added, the uninitiated carbenes initiate and form polymer that have the molecular weight similar to that obtained under the conditions [M]0 = 0.6 M, [I]0 = 0.003 M and 29.8 %

37 conversion, which is the lower molecular weight peak in Figure 2.5(a). Since there is only a small amount of zwitterions, there is no discernable growth in the growing polymer chains upon addition of more monomer.

For (b): (3) A solution of IMes in THF (0.006 M) was added to a solution of LA in THF (0.9

M) and stirred for 10 seconds and then quenched by addition of CS2. The conversion was 58.7 % and the GPC trace of this reaction is the dashed line in Figure 2.5(b). (4) A solution of IMes in THF (0.006 M) was added to a solution of LA in THF (0.9 M) and stirred for 10 seconds. Then a solution of LA in THF with a concentration of 0.9 M was added to the reaction solution (making the total monomer concentration to become approximately 0.6 M, so that we can compare with the data at 0.6 M) and stirred for 30 seconds. Then the reaction was quenched by addition of CS2. The total conversion was 63.6 % and the conversion of the second batch of monomer was 48.5 %. The GPC trace of this reaction is the solid line in Figure 2.5(b).

According to the simulations, under the conditions [M]0 = 0.9 M, [I]0 = 0.006 M at 58.7 % conversion (Figure 2.4(b)) there is about 50 % zwitterions and the rest is uninitiated carbene. So, when the second solution of monomer is added, the uninitiated carbenes initiate and form polymer that have the molecular weight similar to that obtained under the conditions [M]0 = 0.6 M, [I]0 = 0.0015 M (since 50 % is zwitterion) and 48.5 % conversion, which is the lower molecular weight peak in Figure 2.5(b). In addition, since there is a significant amount of zwitterions, the growing chains can add monomer to show higher molecular weights (shift of the polymer peak to the left).

Calculation of monomer conversion in the sequential polymerization Let us define the following variables

M1: the amount of monomer in the beginning (in mol)

M2: the amount of monomer added to the mixture at t1 (in mol)

Mtotal = M1 + M2

38 conv1: conversion at t1 (in %) conv2: conversion between t1 and t2 (in %) convtot: conversion at t2 (in %).

Then, the monomer conversion during the second polymerization, conv2, is given by

M tot  convtot  M 1  conv1 conv2   M 1  conv1  M 2   M 1    100  where conv1 is estimated from independent homopolymerizations (see Table 2.3).

Table 2.3. Monomer conversion during the second polymerization.

Details for stochastic simulations The molecular size and size distribution of chains could be studied by means of stochastic simulations of polymerization reactions.34 First, let us define the rate parameters of initiation, propagation, depropagation, and cyclization for a chain with DP (degree of polymerization; DP  0) as follows:

2 rateI  ki[M ] (2.1) rateP  ki[M ] (2.2) rateD  kd (2.3) rateC  kc (2.4) rateK  rateI when DP  0 (2.5)  rateP  rateD  rateC when DP  0

39 where ki, kp, kd, and kc are the rate constants of initiation, propagation, depropagation, and cyclization, respectively. Then, the average life time of chains is given by

1 t  . (2.6) av rateK

A time increment (t) for each step of the simulation is given by

f t  f t  t (2.7) t av rateK

where ft is an adjustable parameter (0 < ft < 1) for both accuracy and efficiency of modeling. For all chains of interest, a uniform random number (RN1; 0 < RN1 < 1) is generated. The life time of i chain is calculated by using

 ln RN1 t  . (2.8) i rateK

If ti > t, then no transformation for i chain occurs. When ti < t and DPi = 0, initiation happens. When ti < t and DPi > 0, another uniform random number (RN2; 0 < RN2 < 1) is generated, and a type of transformation that occurs is determined according to the following probabilities of propagation, depropagation, and cyclization.

rateP P  (2.9) P rateK rateD P  (2.10) D rateK rateC P  (2.11) C rateK

40 After finishing transformations of all chains, values of t, [M], [I], and DPi are updated and recorded for analysis. Based on this algorithm, a computational program was written in Fortran 77. L'Ecuyer’s method41 was utilized to generate random numbers. Using the program and rate coefficients determined from experiments, stochastic simulations of 100000 chains at various [M]0’s and [I]0’s were performed with a ft of 0.0001.

Stochastic simulation program (Fortran 77) Program Stochastic_Kinetics_Simulator c c Stochastic Kinetics Simulator 2.0.3 c Last Modified On 08/25/08 c

implicit real*8 (a-h,o-z) parameter (nmax = 1000000) real*8 t(nmax), cmt(nmax), cit(nmax), dp(nmax), pdi(nmax), : xz(nmax) integer ni(nmax), nz(nmax), nc(nmax), kchain(nmax) common/chain/ nchain, mchain(nmax*10), monomer, initiator character*30 conversion, distribution

write(*,*) 'Stochastic Kinetics Simulator v2.0.3' c c 1. Initialization c

write(*,*) 'Enter file names for conversion and distribution' read(*,*) conversion, distribution

write(*,*) 'Enter the initial monomer concentration in M' read(*,*) cm0

write(*,*) 'Enter the ratio of monomer to initiator' read(*,*) m_i

write(*,*) 'Enter ki, kp, kd, and kc' read(*,*) rki, rkp, rkd, rkc

write(*,*) 'Enter nchain0 (<1000000)' read(*,*) nchain0

41 write(*,*) 'Enter ft (0

write(*,*) 'Enter idum (<0)' read(*,*) idum

write(*,*) 'Enter maximum conversion in %' read(*,*) conv_max

write(*,*) 'Enter kstep' read(*,*) kstep

ci0 = cm0 / real(m_i) cm = cm0 ci = ci0 rkp_ki = rkp / rki nchain = nchain0 monomer0 = nchain * m_i monomer = monomer0 nchain = nchain0 initiator0 = nchain initiator = nchain time = 0.d0 do i = 1, nchain*10 kchain(i) = 0 mchain(i) = 0 end do it = 0 c c 2. Stochastic Kinetics Run c

nstep = 1000000

do istep = 1, nstep

sdt = 0.d0 idt = 0

do jdt = 1, nchain

if ( mchain(jdt) == 0 ) then sdt = sdt + 1.d0 / ( rki * cm * cm ) idt = idt + 1 end if

if ( mchain(jdt) .gt. 0 ) then

42 sdt = sdt + 1.d0 / ( rkp * cm + rkd + rkc ) idt = idt + 1 end if end do dt = ft * sdt / real(idt) do ichain = 1, nchain

if ( mchain(ichain) .lt. 0 ) cycle

if ( mchain(ichain) == 0 ) then rkcm = rki * cm * cm rK = rkcm else rkcm = rkp * cm rK = rkcm + rkd + rkc end if

rn1 = ranf(idum) dti = dt + dlog(rn1) / rK

if ( dti .gt. 0.d0 ) then

rkcm_rK = rkcm / rK rkcmd_rK = ( rkcm + rkd ) / rK rn2 = ranf(idum)

if ( rkcm_rK .gt. rn2 ) then call propagation(ichain) else if ( rkcmd_rK .gt. rn2 ) then call depropagation(ichain) else call cyclization(ichain) end if

end if end do cm = cm0 * real(monomer) / real(monomer0) ci = ci0 * real(initiator) / real(initiator0) conv = ( cm0 - cm ) / cm0 * 1.d2 time = time + dt if ( conv .gt. conv_max ) exit

43 if ( mod(istep, kstep) == 0 ) then

it = it + 1 t(it) = time cmt(it) = cm cit(it) = ci p0 = real(nchain) - real(initiator) p1 = 0.d0 p2 = 0.d0 do jchain = 1, nchain pchain = real(mchain(jchain)) if ( pchain .lt. 0.d0 ) pchain = -pchain p1 = p1 + pchain p2 = p2 + pchain * pchain end do dp(it) = p1 / p0 pdi(it) = p2 / p1 / dp(it)

ni(it) = initiator nc(it) = nchain - nchain0 nz(it) = nchain0 - initiator xz(it) = real(nz(it)) / p0

write(*,91) 't = ', time, 's', '[M] = ', cm, 'M'

end if

end do c c 3. Print results c

open(unit=11, file=conversion, status='unknown') open(unit=12, file=distribution, status='unknown')

write(11,*) 'Wonhee Stochastic Kinetics Simulator v2.0.3' write(12,*) 'Wonhee Stochastic Kinetics Simulator v2.0.3'

do i = 1, nchain lchain = mchain(i) if ( lchain == 0 ) cycle if ( lchain .lt. 0 ) lchain = - lchain kchain(lchain) = kchain(lchain) + 1 end do

write(11,92) '[M]0 = ', cm0, '(M)' write(12,92) '[M]0 = ', cm0, '(M)'

44 write(11,92) '[I]0 = ', ci0, '(M)' write(12,92) '[I]0 = ', ci0, '(M)' write(11,93) '[M]0/[I]0 = ', m_i write(12,93) '[M]0/[I]0 = ', m_i write(11,92) 'kp = ', rkp, '(M-1s-1)' write(12,92) 'kp = ', rkp, '(M-1s-1)' write(11,92) 'ki = ', rki, '(M-2s-1)' write(12,92) 'ki = ', rki, '(M-2s-1)' write(11,92) 'kd = ', rkd, '(s-1)' write(12,92) 'kd = ', rkd, '(s-1)' write(11,92) 'kc = ', rkc, '(s-1)' write(12,92) 'kc = ', rkc, '(s-1)' write(11,94) 'kp/ki = ', rkp_ki write(12,94) 'kp/ki = ', rkp_ki write(11,93) 'nchain0 = ', nchain0 write(12,93) 'nchain0 = ', nchain0 write(11,93) 'nchain = ', nchain write(12,93) 'nchain = ', nchain write(11,94) 'ft = ', ft write(12,94) 'ft = ', ft write(11,93) 'idum = ', idum write(12,93) 'idum = ', idum write(11,93) 'kstep = ', kstep write(12,93) 'kstep = ', kstep write(11,*) write(12,*) write(11,95) 'time (s)', '[M] (M)', '[I] (M)', 'DP', 'PDI', 'Ni', : 'Nz','Nc', 'Xz' write(12,97) 'DP', 'population'

do j = 1, it write(11,96) t(j), cmt(j), cit(j), dp(j), pdi(j), ni(j), nz(j), : nc(j), xz(j) end do

do k = 1, 5000 write(12,98) k, real(kchain(k)) / real(nchain) end do

close(11) close(12)

91 format(A8,F12.5,A2,A10,F10.5,A2,A9,F10.5) 92 format(A12,F10.5,A9) 93 format(A12,I10) 94 format(A12,F10.5) 95 format(A12,A12,A12,A12,A12,A10,A10,A10,A12) 96 format(F12.3,F12.5,F12.7,F12.2,F12.7,I10,I10,I10,F12.7)

45 97 format(A5,A12) 98 format(I5,F12.5)

end

subroutine propagation(ichain)

implicit real*8(a-h,o-z) parameter(nmax = 1000000) common/chain/ nchain, mchain(nmax*10), monomer, initiator

if ( mchain(ichain) == 0 ) then initiator = initiator - 1 monomer = monomer - 1 mchain(ichain) = mchain(ichain) + 1 end if

monomer = monomer - 1 mchain(ichain) = mchain(ichain) + 1

return end

subroutine depropagation(ichain)

implicit real*8(a-h,o-z) parameter(nmax = 1000000) common/chain/ nchain, mchain(nmax*10), monomer, initiator

monomer = monomer + 1 mchain(ichain) = mchain(ichain) - 1

return end

subroutine cyclization(ichain)

implicit real*8(a-h,o-z) parameter(nmax = 1000000) common/chain/ nchain, mchain(nmax*10), monomer, initiator

mchain(ichain) = - mchain(ichain) initiator = initiator + 1 nchain = nchain + 1

46 return end c c randon number generator of L'Ecuyer with Bays-Durham shuffle and added safeguards c

function ranf(idum) integer idum,IM1,IM2,IMM1,IA1,IA2,IQ1,IQ2,IR1,IR2,NTAB,NDIV real*8 ranf,AM,EPS,RNMX parameter(IM1=2147483563,IM2=2147483399,AM=1./IM1,IMM1=IM1-1, : IA1=40014,IA2=40692,IQ1=53668,IQ2=52774,IR1=12211, : IR2=3791,NTAB=32,NDIV=1+IMM1/NTAB,EPS=1.2e-7,RNMX=1.-EPS) integer idum2,j,k,iv(NTAB),iy save iv,iy,idum2 data idum2/123456789/, iv/NTAB*0/, iy/0/ if ( idum.le.0 ) then idum = max(-idum,1) idum2=idum do j = NTAB+8, 1, -1 k = idum / IQ1 idum = IA1 * ( idum - k * IQ1 ) - k * IR1 if ( idum.lt.0 ) idum = idum + IM1 if ( j.le.NTAB ) iv(j) = idum end do iy = iv(1) end if k = idum / IQ1 idum = IA1 * ( idum - k * IQ1 ) - k * IR1 if ( idum.lt.0 ) idum = idum + IM1 k = idum2 / IQ2 idum2 = IA2 * ( idum2 - k * IQ2 ) - k * IR2 if ( idum2.lt.0 ) idum2 = idum2 + IM2 j = 1 + iy / NDIV iy = iv(j) - idum2 iv(j) = idum if ( iy.lt.1 ) iy = iy + IMM1 ranf = min( AM*iy , RNMX )

return end

47 2.6 References (1) McLeish, T. Science 2002, 297, 2005-2006. (2) Large Ring Molecules; Semlyen, J. A., Ed.; Wiley: New York, 1996. (3) Cyclic Polymers; 2nd ed.; Semlyen, J. A., Ed.; Kluwer Academic Publishers: Dordrecht, 2000. (4) Hur, K.; Winkler, R. G.; Yoon, D. Y. Macromolecules 2006, 39, 3975-3977. (5) Roovers, J. In Cyclic Polymers; Second ed.; Semlyen, J. A., Ed.; Kluwer Academic Publishers: Dordrecht, 2000, p 347-384. (6) Jacobson, H.; Stockmayer, W. H. J. Chem. Phys. 1950, 18, 1600-1606. (7) Geiser, D.; Höcker, H. Macromolecules 1980, 13, 653-656. (8) Laurent, B. A.; Grayson, S. M. J. Am. Chem. Soc. 2006, 128, 4238-4239. (9) McKenna, G. B.; Hostetter, B. J.; Hadjichristidis, N.; Fetters, L. J.; Plazek, D. J. Macromolecules 1989, 22, 1834-1852. (10) Rique-Lurbet, L.; Schappacher, M.; Deffieux, A. Macromolecules 1994, 27, 6318-6324. (11) Chisholm, M. H.; Gallucci, J. C.; Yin, H. Proc. Natl. Acad. Sci. 2006, 103, 15315-15320. (12) Wood, B. R.; Hodge, P.; Semlyen, J. A. Polymer 1993, 34, 3052-3058. (13) Bielawski, C. W.; Benitez, D.; Grubbs, R. H. Science 2002, 297, 2041-2044. (14) Bielawski, C. W.; Benitez, D.; Grubbs, R. H. J. Am. Chem. Soc. 2003, 125, 8424-8425. (15) Boydston, A. J.; Xia, Y.; Kornfield, J. A.; Gorodetskaya, I. A.; Grubbs, R. H. J. Am. Chem. Soc. 2008, 130, 12775-12782. (16) Tezuka, Y.; Ohtsuka, T.; Adachi, K.; Komiya, R.; Ohno, N.; Okui, N. Macromol. Rapid Commun. 2008, 29, 1237-1241. (17) Kricheldorf, H. R. J. Polym. Sci., Part A: Polym. Chem. 2004, 42, 4723-4742. (18) Culkin, D. A.; Jeong, W.; Csihony, S.; Gomez, E. D.; Balsara, N. P.; Hedrick, J. L.; Waymouth, R. M. Angew. Chem. Int. Ed. 2007, 46, 2627-2630. (19) Jeong, W.; Shin, E. J.; Culkin, D. A.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2009, 131, 4884-4891.

48 (20) Johnston, D. S. Adv. Polym. Sci. 1982, 42, 51-106. (21) Odian, G. Principles of Polymerization; 4th ed.; Wiley-Interscience: Hoboken, 2004. (22) Szwarc, M. Makromol. Chem. 1960, 35, 132-158. (23) Jeong, W.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2007, 129, 8414-8415. (24) Kricheldorf, H. R.; Lomadze, N.; Schwarz, G. Macromolecules 2007, 40, 4859-4864. (25) Kricheldorf, H. R.; Von Lossow, C.; Schwarz, G. J. Polym. Sci., Part A: Polym. Chem. 2006, 44, 4680-4695. (26) Flory, P. J. Principles of Polymer Chemistry; Cornell University Press: Ithaca, 1953. (27) van der Mee, L.; Helmich, F.; de Bruijn, R.; Vekemans, J.; Palmans, A. R. A.; Meijer, E. W. Macromolecules 2006, 39, 5021-5027. (28) Baran, J.; Duda, A.; Kowalski, A.; Szymanski, R.; Penczek, S. Macromol. Symp. 1997, 123, 93-101. (29) Williams, C. K.; Breyfogle, L. E.; Choi, S. K.; Nam, W.; Young, V. G.; Hillmyer, M. A.; Tolman, W. B. J. Am. Chem. Soc. 2003, 125, 11350-11359. (30) The cyclic PLLA synthesized using IMes exhibits a lower melting point compared to that reported for linear PLLA in the literature. This is due to the lower isotacticity of the PLLA made by IMes. There seems to be significant amount of epimerization of the monomer during polymerization, the mechanism of which still needs more investigation. (31) Tsuji, H. Macromol. Biosci. 2005, 5, 569-597. (32) Ikada, Y.; Jamshidi, K.; Tsuji, H.; Hyon, S. H. Macromolecules 1987, 20, 904- 906. (33) Matyjaszewski, K.; Szymanski, R.; Teodorescu, M. Macromolecules 1994, 27, 7565-7574. (34) Szymanski, R. Macromol. Theory Simul. 1998, 7, 27-39. (35) Szymanski, R.; Baran, J. Macromol. Theory Simul. 2002, 11, 836-844.

49 (36) It is possible that the carbene can react with cyclized chains to regenerate macro-zwitterionic species. We currently have no evidence either for or against this hypothesis and further studies are underway to test whether cyclized chains can be re-initiated for further chain growth. (37) Csihony, S.; Culkin, D. A.; Sentman, A. C.; Dove, A. P.; Waymouth, R. M.; Hedrick, J. L. J. Am. Chem. Soc. 2005, 127, 9079-9084. (38) Markevic, M. A.; Kochetov, E. V.; Ranogaje.F; Enikolop.Ns J. Macromol. Sci., Chem. 1974, A 8, 265-279. (39) Lowry, T. H.; Richardson, K. S. Mechanism and Theory in Organic Chemistry; 3rd ed.; Harper & Row: New York, 1987. (40) Arduengo, A. J.; Dias, H. V. R.; Harlow, R. L.; Kline, M. J. Am. Chem. Soc. 1992, 114, 5530-5534. (41) L'Ecuyer, P. In The Handbook of Simulation; Banks, J., Ed.; Wiley: New York, 1998, p 93-137.

50 CHAPTER 3

Stereocomplexation in Cyclic and Linear Polylactide Blends

51 3.1 Introduction Cyclic polymers differ from linear polymers by one bond linking the chain ends, yet the simple topological constraint1 of linking the ends of a linear chain influence macromolecular properties in ways that remain poorly understood.2-6 A cyclic topology might be expected to influence crystallization; several studies have recently compared the crystallization of cyclic and linear chains.7-12 We have recently reported the synthesis of cyclic crystalline polylactides13 by zwitterionic ring-opening polymerization of lactones.7,13-17 As polylactides are known to form stereocomplexes,18-21 we were intrigued to investigate whether the combination of topological constraints and geometric constraints associated with the formation of stereocomplexes22-24 might influence the ability of cyclic and linear polylactides to crystallize into stable stereocomplexes.25

Stereocomplexation between enantiomeric blends of linear poly(L-lactide) and 26,27 poly(D-lactide) is proposed to be driven by weak CH3O=C hydrogen bond 22 interactions between enantiomeric chains of L- and D-lactide. Polylactide stereocomplexes exhibit higher melting temperatures and different mechanical and physical properties compared to the enantiomeric homopolymers. The orientation of the PLLA and PDLA chains, either parallel or anti-parallel, in the stereocomplex22,23,28,29 provides a constraint on crystallization; this coupled with the topological constraint of a cyclic architecture motivated to investigate whether cyclic and linear polylactides would form stereocomplexes.25

3.2 Results and Discussion Linear and cyclic polylactides were synthesized by the N-heterocyclic carbene 13,15 mediated ring opening polymerization of either L- or D-lactide. L- or D-lactide were polymerized ([M]0 = 0.6 M) using the N-heterocyclic carbene, 1,3- dimesitylimidazol-2-ylidene (IMes, [I]0 = 0.006 M) in tetrahydrofuran (THF) for 30 seconds to give cyclic poly(L-lactide) (PLLA) and poly(D-lactide), respectively

(Scheme 3.1). Linear polylactides were prepared by polymerizing L- or D-lactide

([M]0 = 0.6 M) using benzyl alcohol as the initiator ([I]0 = 0.003 M) and IMes as the

52 catalyst (I/C = 2) in THF for 30 seconds.30 The characteristics of the four polymer samples are shown in Table 3.1. The cyclic topology of the cyclic PLAs was established by comparing the intrinsic viscosities with the corresponding linear polymers (see Experimental

Section). The ratio of the intrinsic viscosities, []cyclic/[]linear, from the Mark- Houwink plots are approximately 0.74 over the molecular weight range shown, similar to that predicted theoretically2 and observed experimentally for similar samples.13 The melting points and optical rotations of the polylactides are lower than those reported in the literature for highly isotactic polylactides.31 Polylactides prepared under these conditions are not perfectly isotactic, as shown by the homonuclear decoupled 1H NMR spectra (see Experimental Section).32 In addition to resonances associated with the iii tetrad, iiiss hexads are evident, implicating competitive epimerization of stereogenic centers by the carbene, as previously reported.13,33

O O N N O H O OH O O n O rt, 30 s O THF O linear PLA

O O O N N OO O n-1 O O O O rt, 30 s O THF O O O O O cyclic PLA Scheme 3.1. Polymerization of lactide with IMes.

53 Table 3.1. Characterization of polylactides used in this study.

Blends of various combinations of the PLAs in Table 3.1 were prepared (Table 3.2) by solvent casting from dicholoromethane.34 The samples were thermally treated as follows; PLLA/PDLA blends were melted at 220 °C for 15 minutes, cooled to 150 °C and annealed for 24 hours. The cyclic and linear PLLA samples were annealed at 90°C for 24 hours. All samples were then analyzed by differential scanning calorimetry (DSC), wide-angle X-ray scattering (WAXS) and small-angle X-ray scattering (SAXS).

Table 3.2. Blends of polylactides used for this study.

As shown in Figure 3.1 and Table 3.2, both the linear and cyclic PLLA/PDLA blends exhibit melting temperatures approximately 50 °C higher than the PLLA or PDLA homopolymers, indicative of the formation of stereocomplexes. The melting points of the blends (178 – 189 °C) are lower than that reported for highly isotactic PLLA/PDLA stereocomplexes;35,36 this is likely due to the lower tacticities of the

54 PLLA and PDLA samples prepared with the carbenes.13 The heats of melting for blends involving cyclic polymers (entries B2, B3 and B4) are also lower than that of the linear PLLA/linear PDLA blend B1, but this is also likely a consequence of the slightly lower tacticity of the cyclic PLAs relative to the linear PLAs.35,36

Figure 3.1. Differential scanning calorimetry scans of (a) cyclic PLLA (b) linear PLLA + linear PDLA (c) cyclic PLLA + linear PDLA and (d) cyclic PLLA + cyclic PDLA.

Analysis of the linear and cyclic PLLA/PDLA blends by wide-angle X-ray scattering provides further evidence for the formation of PLLA/PDLA stereocomplexes. The WAXS profiles of the blends B1, B3 and B4 exhibit peaks at 2 = 12, 21, and 24 ° characteristic of that of PLLA/PDLA stereocomplexes, whereas that of the cyclic PLLA shows peaks characteristic of those of linear PLLA.26 These data unambiguously show that constraining polylactides into a cyclic topology does not impede their ability to form stereocomplexes.

55 For the scattering patterns in Figure 3.2(a), the samples were cooled from the o o melt (220 C) at 5 deg/min to Ta = 150 C, annealed at Ta for 24 hours to allow stereocomplex formation, and then taken off the heater to cool to room temperature for the measurement. The effect of thermal history22,37,38 is shown in Figure 3.2(b); for samples that were not annealed, but simply cooled from the melt to room temperature, only the linear PLLA/linear PDLA blend showed evidence of stereocomplexation. While it is possible that the topology of the cyclic PLAs influences the rate at which they crystallize into stereocomplexes, it is more likely that this behavior is a consequence of the lower molecular weights and slightly higher tacticities of the linear PLLA and PDLA samples.

(a) (b)

Figure 3.2. Wide-angle X-ray scattering patterns after (a) annealing at 150 oC (90 oC for cyclic PLLA) for 24 hours and (b) cooling from the melt to room temperature.

Small-angle X-ray scattering was carried out to estimate the lamellar thickness 39,40 (Lc) and long period spacing (Lp) of linear and cyclic PLLA and the linear and

56 cyclic stereocomplexed blends. The lamellar thicknesses (Lc) and long periods (Lp) were estimated from the one-dimensional electron density autocorrelation function

(see Experimental Section) and are shown in Table 3.3. The lamellar thickness and long period of the linear PLLA (8 nm and 16 nm, respectively) are smaller than those reported previously (Lc ~ 16 nm and Lp ~ 22 nm) for higher molecular weight highly isotactic linear PLLAs,41,42 but comparable to PLLA copolymers containing between 6 – 12 % meso-lactide.43

Table 3.3. Lamellar thickness (Lc) and long period (Lp)

Notably, the lamellar thickness and long period of cyclic PLLA are approximately 20 % larger than those of linear PLLA. The long period (20 nm) is on the order of the extended chain length of this cyclic PLLA sample (approx. 33 nm 44 assuming a 103 helix), implicating the lack of multiple chain-folding for the cyclic chains. The larger lamellar thickness and long period of the cyclic PLLA may be a consequence of a topological constraint on lamellar folding,41,43,45 but further studies are necessary to test the generality of this observation. In previous studies of high molecular weight cyclic and linear poly(-caprolactone)s, we observed no differences in lamellar or long period spacings between the cyclic and linear polyesters, but in the latter case the chain lengths were considerably longer than the lamellar thickness (~ 7 700 nm extended chain length, Lc ~ 9 nm). The lamellar thickness and long period of the cyclic and linear PLLA/PDLA blends were similar to that of the linear PLLA samples when samples were annealed at

57 similar undercoolings T (T = Tm – Ta, for Tm = measured melting point and Ta = annealing temperature). There have been only a few reports on the SAXS analysis of the PLA stereocomplexes.24,46,47 In Tsuji’s report,46 the estimated long period of the stereocomplex that was precipitated from acetonitrile solution at 80 oC and annealed at 216 oC was 12 nm, smaller than that reported for PLLA homopolymer films crystallized from the melt (22 ~ 35 nm). It was noted that this difference in lamellar thickness and long period could be partly due to the different thermal history.20

3.3 Summary

The zwitterionic polymerization of L- and D-lactide with N-heterocyclic carbenes provides a series of crystalline linear and cyclic polylactides. Wide-angle and small angle X-ray scatterings indicate that both linear and cyclic isotactic polylactides crystallize with similar local structures and lamellar spacings as their linear topological isomers, although cyclic PLLA’s exhibit slightly larger lamellar thickenesses than the corresponding linear PLLA’s. Both linear and cyclic blends of PLLA and PDLA form stereocomplexes from the melt; the geometric constraints required from stereocomplexation22-24 do not appear to be compromised by constraining the polylactides into a cyclic chain.

58 3.4 Experimental Section 3.4.1. Experimental procedures

Materials. L-lactide, and D-lactide were provided from PURAC. The monomers were dried with CaH2 in THF, filtered and then sublimed twice. All monomers were stored in a drybox under nitrogen. Tetrahydrofuran (THF) was distilled from sodium/benzophenone and degassed three times via freeze-pump-thaw cycles. Carbon disulfide was purchased from Aldrich and distilled twice from calcium hydride. Anhydrous 2-propanol for the precipitation of cyclic polymer was purchased from Acros and used as received. Anhydrous methanol for the precipitation of linear polymer was purchased from Aldrich and used as received. 1,3-dimesitylimidazolium chloride and potassium tert-butoxide were purchased from Strem and used without further purification. 1,3-dimesitylimidazol-2-ylidene (IMes) was prepared according to the literature procedure.48

Characterization methods. All reactions and polymerizations were performed in a drybox under nitrogen. 1H nuclear magnetic resonance (NMR) spectra were recorded at room temperature on a Varion 400 or 500 MHz spectrometer, with shifts reported in parts per million downfield from tetramethylsilane and referenced to the residual solvent peak. Gel permeation chromatography (GPC) was performed in tetrahydrofuran (THF) at a flow rate of 1.0 mL/min on a Waters chromatograph equipped with four 5 μm Waters columns (300 mm × 7.7 mm) connected in series. A Viscotek S3580 refractive index detector, Viscotek 270 dual detector (light scattering and visocometer) and Viscotek GPCmax autosampler were employed. The system was calibrated using monodisperse polystyrene standards (Polymer Laboratories). Correction for any angular dissymmetry factor in the right-angle light scattering (RALS) data was performed in the TriSEC software using the viscometer signal. The angular dissymmetry correction is negligible because the polymers studied are relatively small compared to the laser wavelength (610 nm). The polymer solution (ca. 10 mg/mL) was prepared by dissolving the polymer in THF. Differential scanning calorimetry (DSC) was performed using a TA Instruments Q100 differential

59 scanning calorimeter. Melting points for the homopolymers were determined by cooling from melt to 90 °C at 0.3 °C/min, annealing for 24 h, cooling to 30 oC and then scanning at a heating rate of 10 °C/min. Melting points for the blended samples were determined by cooling from melt to 150 °C at 10 °C/min, annealing for 24 h, cooling to 30 oC and then scanning at a heating rate of 10 °C/min. The optical rotation was measured in chloroform at room temperature, using a concentration of 0.9 g/dL. For wide-angle X-ray scattering (WAXS) experiments, the polymer was transferred to a thin-walled glass capillary (0.7 mm diameter, Wolfgang Mueller). In a drybox under argon, the sample was first melted at 220 °C for 15 min, and cooled to the desired annealing temperature (90 °C for homopolymers or 150 °C for blended samples). After annealing at that temperature for 24 h, the sample was cooled to room temperature, then taken out of the drybox. WAXS data were collected on a General Area Detector Diffraction System (D8-GADDS, Bruker AXS) equipped with a four- circle diffractometer, a Hi-STAR multiwire area detector, and a Cu Kα X-ray source (λ = 0.154056 nm) at NICEM, Seoul National University in Korea. The data acquired were analyzed using GADDS 3.1.4 software (Bruker AXS).

Representative procedure for preparation of cyclic poly(lactide). In a drybox, under nitrogen, a solution of IMes (10.5 mg, 0.034 mmol) in THF (0.78 mL) was added to a stirred solution of L-lactide (500 mg, 3.47mmol) in THF (5 mL). The reaction mixture was stirred for 30 s at room temperature. Carbon disulfide (0.5 mL) was added to terminate the reaction and the mixture turned red, indicating formation of the adduct of IMes and carbon disulfide.49 The solvent was removed and the crude polymer was analyzed by 1H NMR spectroscopy for conversion. The crude polymer was dissolved in dichloromethane and precipitated from 2-propanol (3 times) to remove residual catalyst and monomer. The polymer was characterized by 1H NMR, 1 1 GPC, DSC and homonuclear decoupled H NMR. H NMR (CDCl3) δ 1.50-1.62 (-

CH3), 5.10-5.24 (-CH-); GPC (THF) Mn = 30,222, PDI = 1.29; DSC Tm = 127, 135 °C,

Hm = 18 J/g; []D = -110 º (c = 0.9 in CHCl3).

60 Representative procedure for preparation of linear polylactide. In a drybox, under nitrogen, a solution of IMes (2.6 mg, 0.0085 mmol) in THF (0.78 mL) was added to a solution of the L-lactide monomer (500 mg, 3.47 mmol) and benzyl alcohol (1.9 mg, 0.018 mmol) in THF (5 mL). The reaction mixture was stirred for 30 s at room temperature. Carbon disulfide (0.5 mL) was added to terminate the reaction and the mixture turned red, indicating formation of the adduct of IMes and carbon disulfide.49 The solvent was removed and the crude polymer was analyzed by 1H NMR spectroscopy for conversion. The crude polymer was dissolved in dichloromethane and precipitated from methanol (3 times) to remove residual catalyst and monomer. The polymer was characterized by 1H NMR, GPC, DSC and homonuclear decoupled 1 1 H NMR. H NMR (CDCl3) δ 1.52-1.63 (-CH3), 5.11-5.24 (-CH-), 4.36 (q, -CHOH,

1H), 7.28-7.43 (m, -C6H5, 5H); GPC (THF) Mn = 14,980, PDI = 1.14; DSC Tm = 150

°C, Hm = 36 J/g; []D = -131 º (c = 0.9 in CHCl3).

Procedure for preparation of blends. The procedure similar to the solvent casting method was used.50 Equal amounts (w/w) of the respective homopolymers were dissolved separately in dichloromethane at concentrations of 10 mg/mL. The two solutions were mixed together in a vial and stirred vigorously overnight. The solvent was then allowed to evaporate slowly at room temperature and then the blend was dried under vacuum to remove residual solvent.

61 3.4.2. Characterization data

Viscosity measurements

(a) (b)

Figure 3.3. Mark-Houwink plots of (a) linear and cyclic PLLA and (b) linear and cyclic PDLA. The ratios ([]cyclic/[]linear) are 0.74.

Table 3.4. Absolute molecular weights of the samples, determined by light scattering.

Degree of polymerization, Sample M (g/mol) PDI w DP Linear PLLA 10,053 1.03 70

Linear PDLA 12,896 1.04 89

Cyclic PLLA 19,327 1.17 134

Cyclic PDLA 20,898 1.15 145

62 Tacticity

Figure 3.4. Homonuclear decoupled 1H NMR spectra of (a) linear PLLA and (b) cyclic PLLA. Peak assignments are done according to reported literature.51

The fraction of iii tetrad (fiii) was determined by deconvoluting the peaks and taking the fraction of the area of the iii tetrad over the whole. The peak fitting procedure of ACD labs software was used. The blue lines show the deconvoluted peaks.

Table 3.5. Fraction of iii tetrad.

Sample Area of iii tetrad Area of whole Fraction of iii, fiii

Linear PLLA 3409301504 3919317916 0.8699

Linear PLDA 4070468096 4545464218 0.8955

Cyclic PLLA 1544126080 1864787306 0.8280

Cyclic PDLA 2774512640 3437649272 0.8071

The composition of the polymers were analyzed from the total normalized integrated intensity of the iis-cored hexad stereosequences in the homonuclear 1 decoupled H NMR spectra. The L-lactide or D-lactide content is estimated from the total iis-cored hexad intensities by assuming a certain meso-lactide fraction and then iteratively corrected for changes in meso fraction calculated from iiiss and isisi intensities.52

65 Table 3.6. Composition analysis calculated from homonuclear decoupled 1H NMR spectra.

Sample % L-lactide % D-lactide % meso-lactide

Linear PLLA 96.06 1.01 2.93

Linear PLDA 0.11 98.08 1.81

Cyclic PLLA 94.54 1.34 4.12

Cyclic PDLA 1.4 94.21 4.38

Small angle X-ray scattering SAXS experiments were performed at beamline 1-4 of the Stanford Synchrotron Radiation Laboratory (SSRL). The X-ray source has a flux of ca. 1010 photons at a wavelength of λ = 1.488 Å. A CCD area detector (Photonic Science) was used to collect the 2-D diffraction: the detector has an active area of 100 mm diameter, which is mapped via a tapered fiber-optic bundle to the CCD chip with a 1024×1024 array of 25 μm square pixels. The samples were prepared by placing the polymers between Kapton films and annealing at 120 oC under nitrogen for 24 hours (in an oven). The films were taken out of the oven, cooled to room temperature and put in the beamline.

From the raw intensity data, Is(q), the of the absolute intensity, i.e. the differential cross section per unit volume, (q), was obtain using the following equation.

I q q  s I  A  e T t 0 s s

66 where I0 is the incident flux, A is the sample exposure area,  is the solid angle element, e is the detector efficiency, Ts is the sample transmission, and ts is the sample thickness, and q is the scattering vector q  2s  4 sin . Fourier transform of (q) gives the one-dimensional electron density autocorrelation function, K(z).39,40

1 1  K(z)  cos qz  4q2q dq r 2 3 0 e 2

The lamellar thickness (Lc) and long period (Lp) of each sample were estimated from the minimum and maximum peaks, respectively.

Figure 3.5. SAXS profiles (left) and one-dimensional electron density autocorrelation functions (right) of (a) linear PLLA, (b) cyclic PLLA, (c) linear PLLA + linear PDLA, (d) cyclic PLLA + linear PDLA and (e) cyclic PLLA + cyclic PDLA.

67 3.5 References (1) Yamamoto, T.; Tezuka, Y. Polym. Chem. 2011, DOI: 10.1039/C1PY00088H (2) Roovers, J. In Cyclic Polymers; Second ed.; Semlyen, J. A., Ed.; Kluwer Academic Publishers: Dordrecht, 2000, p 347-384. (3) McKenna, G. B.; Hostetter, B. J.; Hadjichristidis, N.; Fetters, L. J.; Plazek, D. J. Macromolecules 1989, 22, 1834-1852. (4) Kapnistos, M.; Lang, M.; Vlassopoulos, D.; Pyckhout-Hintzen, W.; Richter, D.; Cho, D.; Chang, T.; Rubinstein, M. Nature Materials 2008, 7, 997-1002. (5) Laurent, B. A.; Grayson, S. M. Chem. Soc. Rev. 2009, 38, 2202-2213. (6) Kricheldorf, H. R. J. Polym. Sci., Part A: Polym. Chem. 2010, 48, 251-284. (7) Shin, E. J.; Jeong, W.; Brown, H. A.; Koo, B. J.; Hedrick, J. L.; Waymouth, R. M. Macromolecules 2011, 44, 2773-2779. (8) Schaler, K.; Ostas, E.; Schroter, K.; Thurn-Albrecht, T.; Binder, W. H.; Saalwachter, K. Macromolecules 2011, 44, 2743-2754. (9) Cordova, M. E.; Lorenzo, A. T.; Muller, A. J.; Hoskins, J. N.; Grayson, S. M. Macromolecules 2011, 44, 1742-1746. (10) Bielawski, C. W.; Benitez, D.; Grubbs, R. H. Science 2002, 297, 2041-2044. (11) Tezuka, Y.; Ohtsuka, T.; Adachi, K.; Komiya, R.; Ohno, N.; Okui, N. Macromol. Rapid Commun. 2008, 29, 1237-1241. (12) Cooke, J.; Viras, K.; Yu, G. E.; Sun, T.; Yonemitsu, T.; Ryan, A. J.; Price, C.; Booth, C. Macromolecules 1998, 31, 3030-3039. (13) Culkin, D. A.; Jeong, W.; Csihony, S.; Gomez, E. D.; Balsara, N. P.; Hedrick, J. L.; Waymouth, R. M. Angew. Chem. Int. Ed. 2007, 46, 2627-2630. (14) Jeong, W.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2007, 129, 8414-8415. (15) Jeong, W.; Shin, E. J.; Culkin, D. A.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2009, 131, 4884-4891. (16) Shin, E. J.; Brown, H. A.; Gonzalez, S.; Jeong, W.; Hedrick, J. L.; Waymouth, R. M. Angew. Chem. Int. Ed. 2011, 50, 6388-6391. (17) Guo, L.; Zhang, D. J. Am. Chem. Soc. 2009, 131, 18072-18074.

68 (18) Fukushima, K.; Kimura, Y. Polym. Int. 2006, 55, 626-642. (19) Kakuta, M.; Hirata, M.; Kimura, Y. Polymer Reviews 2009, 49, 107-140. (20) Tsuji, H. Macromolecular Bioscience 2005, 5, 569-597. (21) Anderson, K. S.; Hillmyer, M. A. Polymer 2006, 47, 2030-2035. (22) Sarasua, J.-R.; Rodriguez, N. L.; Arraiza, A. L.; Meaurio, E. Macromoelcules 2005, 38, 8362-8371. (23) Cartier, L.; Okihara, T.; Lotz, B. Macromolecules 1997, 30, 6313-6322. (24) Li, L.; Zhong, Z.; de Jeu, W. H.; Dijkstra, P. J.; Feijen, J. Macromolecules 2004, 37, 8641-8646. (25) Jones, A. E.; Shin, E. J.; Waymouth, R. M. Polym. Prepr. 2008, 49, 1149- 1150. (26) Ikada, Y.; Jamshidi, K.; Tsuji, H.; Hyon, S.-H. Macromolecules 1987, 20, 904- 906. (27) Morozov, A. N.; Fraaije, J. G. E. M. Macromoelcules 2001, 34, 1526-1528. (28) Okihara, T.; Tsuji, M.; Kawaguchi, A.; Katayama, K.; Tsuji, H.; Hyon, S. H.; Ikada, Y. J. Macromol. Sci., Phys. 1991, B30, 119-140. (29) Brizzolara, D.; Cantow, H. J.; Diederichs, K.; Keller, E.; Domb, A. J. Macromolecules 1996, 29, 191-197. (30) Nyce, G. W.; Glauser, T.; Connor, E. F.; Mock, A.; Waymouth, R. M.; Hedrick, J. L. J. Am. Chem. Soc. 2003, 125, 3046-3056. (31) Chabot, F.; Vert, M.; Chapelle, S.; Granger, P. Polymer 1983, 24, 53-59. (32) Thakur, K. A. M.; Kean, R. T.; Hall, E. S.; Kolstad, J. J.; Lindgren, T. A.; Doscotch, M. A.; Siepmann, J. I.; Munson, E. J. Macromolecules 1997, 30, 2422-2428. (33) Csihony, S.; Nyce, G. W.; Sentman, A. C.; Waymouth, R. M.; Hedrick, J. L. Polymer Preprints 2004, 45, 319-320. (34) Tsuji, H.; Hyon, S. H.; Ikada, Y. Macromolecules 1991, 24, 5651-5656. (35) Tsuji, H.; Ikada, Y. Macromolecules 1992, 25, 5719-5723. (36) Brochu, S.; Prudhomme, R. E.; Barakat, I.; Jerome, R. Macromolecules 1995, 28, 5230-5239.

69 (37) Wang, X. H.; Prud'homme, R. E. Macromol. Chem. Phys. 2011, 212, 691-698. (38) Zhang, J. M.; Tashiro, K.; Tsuji, H.; Domb, A. J. Macromolecules 2007, 40, 1049-1054. (39) Schmidtke, J.; Strobl, G.; ThurnAlbrecht, T. Macromolecules 1997, 30, 5804- 5821. (40) Strobl, G. The Physics of Polymers: Concepts for Understanding Their Structures and Behavior; 3rd ed.; Springer: Berlin, 2007. (41) Baratian, S.; Hall, E. S.; Lin, J. S.; Xu, R.; Runt, J. Macromolecules 2001, 34, 4857-4864. (42) Cho, T. Y.; Strobl, G. Polymer 2006, 47, 1036-1043. (43) Huang, J.; Lisowski, M. S.; Runt, J.; Hall, E. S.; Kean, R. T.; Buehler, N.; Lin, J. S. Macromolecules 1998, 31, 2593-2599. (44) Sasaki, S.; Asakura, T. Macromolecules 2003, 36, 8385-8390. (45) Matushita, Y.; Iwata, H.; Asari, T.; Uchida, T.; ten Brinke, G.; Takano, A. J. Chem. Phys. 2004, 121, 1129-1132. (46) Tsuji, H.; Horii, F.; Nakagawa, M.; Ikada, Y.; Odani, H.; Kitamaru, R. Macromolecules 1992, 25, 4114-4118. (47) Fujita, M.; Sawayanagi, T.; Abe, H.; Tanaka, T.; Iwata, T.; Ito, K.; Fujisawa, T.; Maeda, M. Macromolecules 2008, 41, 2852-2858. (48) Arduengo, A. J.; Dias, H. V. R.; Harlow, R. L.; Kline, M. J. Am. Chem. Soc. 1992, 114, 5530-5534. (49) Nyce, G. W.; Csihony, S.; Waymouth, R. M.; Hedrick, J. L. Chem. Eur. J. 2004, 10, 4073-4079. (50) Tsuji, H.; Hyon, S. H.; Ikada, Y. Macromolecules 1991, 24, 5651-5656. (51) Zell, M. T.; Padden, B. E.; Paterick, A. J.; Thakur, K. A. M.; Kean, R. T.; Hillmyer, M. A.; Munson, E. J. Macromolecules 2002, 35, 7700-7707. (52) Thakur, K. A. M.; Kean, R. T.; Hall, E. S.; Doscotch, M. A.; Munson, E. J. Anal. Chem. 1997, 69, 4303-4309.

70 CHAPTER 4

Zwitterionic Copolymerization: Synthesis of Cyclic Gradient Copolymers

71 4.1 Introduction Gradient copolymers constitute an intriguing class of macromolecules in which the comonomer composition varies continually from the beginning to the end of the linear chain (Figure 4.1).1,2 Experimental and theoretical studies indicate that linear gradient copolymers exhibit distinct physical properties from random and block copolymers as a consequence of their sequence distribution.1,3-6 Because cyclic polymers have no ends, we were intrigued by the challenge of devising an expedient synthesis of cyclic gradient copolymers as these materials would contain both a sharp comonomer interface and a gradient interface in the same molecule.

Figure 4.1. Schematic diagrams of linear and cyclic gradient copolymers.

The topological differences between cyclic and linear macromolecules influence their behavior in ways that remain poorly understood.7-12 Cyclic polymers are more compact and entangle differently as a consequence of the chemical bond that connects the chain-ends.7,10,11,13,14 Cyclic block copolymers exhibit smaller microdomains and, in some cases, different morphologies than linear diblocks as a consequence of topological constraints on phase separation.9,15-17 These studies highlight the need for synthetic strategies to prepare cyclic polymers18-21 with defined sequences to enable studies on the influence of topology and sequence distribution on the self-organization and properties of macromolecules. We have previously described a synthetic strategy for generating cyclic polyesters by the zwitterionic ring-opening of lactones with nucleophilic N- heterocyclic carbenes.22-24 Kinetic studies revealed that the active zwitterions exhibit lifetimes commensurate with that of the growing chains, but attempts to generate

72 block copolymers were frustrated by re-initiation from unactivated carbenes upon addition of a second batch of monomer.24,25 We thus sought another strategy for generating cyclic block copolymers and report here that a one step gradient batch copolymerization1,2 provides a facile method for generating cyclic gradient block copolymers.

4.2 Results and Discussion The synthesis of gradient copolymers in a batch copolymerization requires that all chains grow under similar conditions and the composition of the monomers in the polymerization medium must change continually as the chains grow.1 The first criterion is generally met by living polymerization methods, but it was our hypothesis that the kinetic lifetime of the growing zwitterions might be sufficient to satisfy this first criterion for a gradient copolymerization. The second criterion requires that the reactivity of the two monomers be sufficiently different such that one monomer is consumed more rapidly than the other.1,2 The reactivity of different lactone monomers in ring-opening polymerization of lactones with metal alkoxides is typically not very different,26,27 which would limit the gradients that could be generated. However, as previous studies had suggested that the copolymerization selectivity for organic catalysts are different than that of metal alkoxides,28,29 we carried out experiments to measure the relative reactivity of - caprolactone (CL) and -valerolactone (VL) with N-heterocyclic carbene organocatalysts. The reactivity ratios for the copolymerization of -caprolactone (CL) and -valerolactone (VL) were determined using benzyl alcohol and 1,3,4,5- tetramethylimidiazol-2-ylidene (IMe4) carbene in THF to generate linear copolymers. Copolymerizations were carried out to 10 % conversion to minimize the influence of depolymerization on the determination of the reactivity ratios. 1H NMR was used for determining conversion and 13C NMR integration of the -CL and -VL polymer resonances were used in determining the polymer composition. Estimates for the reactivity ratios for VL and for CL were determined to be: rVL = 9.0 and rCL = 0.24, by the Fineman-Ross method (see Experimental Section). The large difference in

73 reactivity ratios between VL and CL suggested that this system would be competent for a gradient copolymerization.1,2 In addition, these data illustrate the significant influence of the organocatalysts on the comonomer reactivities, as the copolymerization of CL and VL with the carbene IMe4 at 25 °C exhibits a very different copolymerization behavior than Sn(Oct)2 catalysts at 160 °C, for which rVL = 26,27 0.49 and rCL = 0.25.

As the IMe4 catalyst was fast and difficult to control, we investigated the less active carbene 1,3-diethyl-4,5-dimethyl-imidiazol-2-ylidene (Me2IEt) for the generation of cyclic VL/CL copolymers in toluene solution (Figure 4.2).

Copolymerizations of -caprolactone (CL) and -valerolactone (VL) ([Mtot]0 = 1.0 M,

Mtot/I = 100) were carried out using Me2IEt in toluene at room temperature for reaction times ranging from 7 minutes to 2 hours to produce a range of copolymer compositions (Table 4.1).

Figure 4.2. Zwitterionic ring-opening copolymerization of VL and CL.

The resulting copolymers were characterized by NMR, GPC and DSC. The 30 copolymers obtained exhibited polydispersities (Mw/Mn), ranging from 1.7 to 2.4. Evidence that the copolymers were cyclic were obtained by comparing the intrinsic viscosity22,24 of the copolymers produced at each molecular weight by the zwitterionic polymerization versus that of linear VL/CL copolymers prepared with Me2IEt in the presence of an alcohol initiator. The lower intrinsic viscosities observed for the copolymers obtained in the absence of the alcohol compared to their linear analogues prepared in the presence of alcohol initiators, is indicative of a cyclic structure for the former copolymers (see Experimental Section). The ratio of the intrinsic viscosities

74 are calculated to be around 0.6 – 0.78, which is in good agreement with theoretical predictions and other experimental findings.10

Table 4.1. Polymerization and characterization data for cyclic gradient copolymers generated by using Me2IEt.

The copolymerization behavior was characterized by studying the conversion of each monomer with time and molecular weight with conversion. As seen in Figure 4.3, the molecular weight increases steadily with conversion, indicative of a continuous incorporation of both monomers into the growing polymer chain. As shown in Figure 4.3(a), VL converts rapidly to reach 94 % conversion within 5 minutes, while the conversion of CL is more gradual over the 1 hour period. These results reveal that the composition of the monomers in the polymerization medium changes continually as the chains grow,1 implicating the growth of a block consisting predominantly of VL followed by a block consisting predominantly of CL. The cumulative and instantaneous fraction of VL in the copolymer as a function of conversion (see Experimental Section) are also consistent with the formation of a gradient copolymer.

75 (a) (b)

Figure 4.3 (a) Conversion of each monomer with time and (b) molecular weight with conversion for MVL:MCL = 1:1, [Mtot]0 = 1.0 M, Mtot/I = 100.

Table 4.2. Selected samples for microstructure analysis.

a b c c o d e Entry Type %CL Mw (kg/mol) Mn (kg/mol) PDI Tm ( C) Hm (J/g)

1 cyclic NHC 46 98 56 2.4 38/44 30

2 linear NHC 48 43 41 2.1 32/40 40

3 linear SnOct2 47 32 21 1.7 21 54

4 linear diblock 49 20 19 1.2 51/55 69

a Percentage of caprolactone; determined by 13C NMR spectroscopy b Absolute molecular weight; determined by gel permeation c chromatography (GPC) using light scattering Number average molecular weight and polydispersity index (Mw/Mn); determined by GPC with polystyrene calibration d Melting temperature; determined by differential scanning calorimetry (DSC) e Heat of melting; determined by DSC

The microstructure, melting behavior and solid-state structure of the cyclic copolymers were compared to those of a series of linear VL/CL copolymers prepared by different techniques. The microstructure of the copolymers was investigated by 13C NMR spectroscopy and revealed that both cyclic and linear copolymers generated by the Me2IEt carbene (entries 1 and 2 of Table 4.2) contain a larger fraction of homo- dyad (CL-CL and VL-VL) sequences compared to hetero-dyad sequences (CL-VL and VL-CL) (see Experimental Section). In contrast, the copolymer generated by the tin(II) ethylhexanoate catalyst (entry 3 of Table 4.2) shows an equal ratio of the homo- and hetero-dyad sequences, consistent with a more random copolymer. For

76 comparison, a linear P(CL-b-VL) diblock copolymer was prepared in a step-wise fashion by first making a PCL macroinitiator and then growing a second VL block with a 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD) organic catalyst (entry 4 of Table 4.2).28 This diblock copolymer exhibited only two peaks corresponding to the CL-CL and VL-VL homo-dyad sequences. The melting points of the cyclic and linear VL/CL copolymers obtained from the carbene Me2IEt were compared with those obtained from the copolymerization of

VL and CL with Sn(Oct)2 and that of the linear diblock copolymer. The homopolymers of VL and CL are both semicrystalline thermoplastics with melting 26,27 points of 57 °C (Hm = 52 J/g) and 56 °C (Hm = 55 J/g), respectively. Storey and Yoshida31 have previously shown that random (or slightly alternating) copolymers of VL and CL are semicrystalline across the full composition range but that random copolymers exhibit lower melting points than either of the homopolymers with the lowest melting point (21 °C) being observed for copolymers with approximately 60 mol% CL.

Figure 4.4. Melting points of various samples compared with literature values (ref 27). Two melting peaks from one sample are drawn as two symbols connected with a vertical line.

77 The gradient copolymers exhibit different properties from either the random or block copolymers. For similar compositions (approx. 47 % CL), the gradient copolymers obtained from the carbene catalysts exhibit higher melting points (32 – 44 °C, solid squares and circles in Figure 4.4) than those of the random copolymers 26,27 obtained from Sn(Oct)2, and lower melting points than those of the linear diblock copolymers or blends. The melting exotherms of the gradient copolymers are broader than those of the linear diblock copolymer and have smaller heat of melting compared to the homopolymers (see Experimental Section), as might be expected if the two comonomers co-crystallize.26,27 Wide-angle X-ray scattering experiments also provide evidence for the co- existence of CL and VL phases in the solid-state. The cyclic homopolymers exhibit similar WAXS patterns to those of the linear homopolymers.32,33 The WAXS pattern of the cyclic gradient copolymer shows less intense peaks that overlap those of the PVL and PCL homopolymers. In contrast, only an amorphous halo was observed for the random copolymer of similar CL content (see Experimental Section).

4.3 Conclusion In summary, the zwitterionic copolymerization of -valerolactone and - caprolactone provides an expedient synthesis of gradient cyclic (or linear) VL-grad- CL copolymers in a batch copolymerization. The wide difference in reactivity between VL and CL with NHC catalysts, coupled with sufficiently long lifetimes of the growing zwitterions leads to an unusual polymer structure comprised of VL-rich sequences that transition to CL-rich sequences in a cyclic macromolecule. The gradient sequences and the ability of VL and CL to co-crystallize lead to cyclic copolymers with melting points that are similar to the homopolymers. This synthetic approach provides a strategy for generating unusual topologies and sequences, whose properties are under further investigation.

78 4.4 Experimental Section 4.4.1. Experimental Procedures General Considerations. All reactions and polymerizations were performed in a drybox or using Schlenk techniques under nitrogen. 1H (13C) nuclear magnetic resonance (NMR) spectra were recorded at room temperature on a Varian 400 (100) or 500 (125) MHz spectrometer, with shifts reported in parts per million downfield from tetramethylsilane and referenced to the residual solvent peak. Gel permeation chromatography (GPC) was performed in tetrahydrofuran (THF) at a flow rate of 1.0 mL/min on a Waters chromatograph equipped with four 5 μm Waters columns (300 mm x 7.7 mm) connected in series. A Viscotek S3580 refractive index detector and Viscotek GPCmax autosampler were employed. The system was calibrated using monodisperse polystyrene standards (Polymer Laboratories). The system with a triple detection system (Viscotek, Houston, TX) including a light scattering detector (right- angle and low-angle light scattering) and viscometer was used to determine absolute molecular weights of polymers. Correction for any angular dissymmetry factor in the RALS data was performed in the TriSEC software using the viscometer signal. The angular dissymmetry correction is negligible because the polymers studied are relatively small compared to the laser wavelength (610 nm). Differential scanning calorimetry (DSC) was performed using a TA Instruments Q100 differential scanning calorimeter. Melting points were determined on the second scan (after first rapid scan to remove thermal history) at a heating rate of 5 °C/min. Wide-angle X-ray scattering (WAXS) measurements were carried out using a PANalytical X'Pert PRO X-ray diffraction system at 40 kV and 40 mA using sealed-tube Cu Kα (1.542 Å) radiation.

Materials. Toluene was distilled from sodium/benzophenone after being degassed three times via freeze-pump-thaw cycles. -Caprolactone (CL) and -valerolactone (VL) were purchased from Aldrich and distilled from calcium hydride twice. 4- Nitrophenol was purchased from Fluka and purified by recrystallization from toluene and subsequent sublimation. Anhydrous methanol was purchased from Aldrich and used as received. 1,3,4,5-tetramethylimidazol-2-ylidene (IMe4) and 1,3-diethyl-4,5-

79 dimethylimidazol-2-ylidene (Me2IEt) were prepared according to the literature procedure.34

Representative procedure for copolymerization of VL and CL with Me2IEt. In a drybox, a solution of Me2IEt (7.6 mg, 0.05 mmol) in toluene (1 mL) was added to a stirred solution of VL (250 mg, 2.5 mmol) and CL (285 mg, 2.5 mmol) in toluene (4 mL). The reaction mixture was stirred for the desired reaction time at room temperature. 4-Nitrophenol (18.7 mg, 0.13 mmol) was added as a whole to terminate the reaction. The solvent was removed and the crude polymer was analyzed by 1H NMR spectroscopy for conversion. The crude polymer was then precipitated with a copious amount of anhydrous methanol and dried under vacuum. 13C NMR spectroscopy was used to determine the CL content and microstructure of the purified polymer. The number average molecular weights and polydispersities were determined by GPC, and DSC was carried out to determine the melting point and heat of melting. For selected samples, Mark-Houwink analysis and WAXS were performed.

Figure 4.5. Mark-Houwink plots of all samples in Table 4.1 and their respective linear analogues.

Figure 4.6. 13C NMR spectra of the four samples in Table 4.2. (CL-VL means the methylene signal of the -valerolactone unit connected to a -caprolactone, i.e. CL-O-

CH2-CH2-CH2-CH2-CO- ).

(a) (b)

Figure 4.7. Composition analysis for copolymerization at MVL:MCL = 1:1, [Mtot]0 = 1

M, Mtot/ I = 100. (a) Cumulative fraction of VL in copolymer and (b) instantaneous fraction of VL with conversion.2

Figure 4.8. Differential scanning calorimetry scans of samples in Table 4.2.

Figure 4.9. Wide-angle X-ray scattering patterns of selected samples.

83 4.4.2. Determination of Reactivity Ratios

Reactivity ratios of CL and VL with IMe4 in the presence of benzyl alcohol initiator at low conversion : Fineman-Ross method35 Copolymerizations were carried out to 10 % conversion to minimize the influence of depolymerization on the determination of the reactivity ratios.27 1H NMR was used for determining conversion and 13C NMR integration of the γ-CL and β-VL polymer resonances were used in determining the polymer composition.27 To a stirred solution of monomer(s) and alcohol in THF was added a solution of IMe4 in THF. The resultant solution was stirred at room temperature for the desired time. All polymerizations were carried to 10 % conversion at room temperature. The polymerization was quenched with a few drops of CS2. An aliquot was removed for analytical purposes: conversion was determined by 1H NMR spectroscopy, and molecular weights and polydispersities were determined by GPC. After removal of the volatiles, the crude polymer was dissolved in DCM, and precipitated from diethyl ether (or methanol). The precipitate was dried in vacuo to a constant weight. 13C NMR integration of well defined γ-CL and β-VL polymer peaks were used in determining the polymer composition.

dM M r M  M m VL  VL 1 VL CL  VL for low conversion dM M r M  M m CL CL 2 CL VL CL

where MVL and MCL are monomer compositions and mVL and mCL are polymer compositions.

kVLVL kCLCL MVL mVL r1  , r2  and F  , f  kVLCL kCLVL M CL mCL F F 2  f 1 r  r f 1 f 2

A plot of (F/f )(f-1) as ordinate and (F2/f ) as abscissa is a straight line whose slope is r1 and intercept is –r2.

Figure 4.10. Determination of reactivity ratios using Fineman-Ross method.

Figure 4.11. Graph relating the amount of CL in starting solution to the amount of CL which will be incorporated into the polymer.

85 4.4.3. Synthesis and Characterization Data of Linear Copolymers

Representative procedure for copolymerization of VL and CL with Me2IEt in the presence of an alcohol initiator (entries 1 – 5, Table 4.3).

In a drybox, a solution of Me2IEt (3.0 mg, 0.02 mmol) in toluene (1 mL) was added to a stirred solution of VL (250.2 mg, 2.5 mmol), CL (285 mg, 2.5 mmol) and benzyl alcohol (2.7 mg, 0.025 mmol) in toluene (4 mL). The reaction mixture was stirred for the desired reaction time at room temperature. 4-Nitrophenol (18 mg, 0.13 mmol) was added as a whole to terminate the reaction. Similar characterization as the cyclic copolymers was performed.

Representative procedure for copolymerization of VL and CL with Sn(Oct)2 in the presence of an alcohol initiator (entries 6 – 7, Table 4.3).

To a Schlenk containing VL (500.7 mg, 5 mmol), CL (571 mg, 5 mmol) and benzyl alcohol (5.4 mg, 0.05 mmol) was added Sn(Oct)2 (6.3 mg, 0.016 mmol) via a syringe. The reaction mixture was heated to 160 oC and stirred for 1.5 hours. The reaction was quenched by placing the Schlenk in an ice bath. Similar characterization as the cyclic copolymers was performed.

86 Procedure for preparation of linear diblock copolymer: poly(CL-b-VL) (entries 8 – 9, Table 4.3).

N O O N N OH O H O (1) O H rt, 8 h n toluene CL PCL-macroinitiator

N O O O N N O O O O H O H (2) O H O rt n m n toluene VL P(CL-b-VL)

In a drybox, under nitrogen, a solution of 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD) (3.4 mg, 0.024 mmol) in toluene (0.5 mL) was added to a solution of CL (570 mg, 5 mmol) and benzyl alcohol (5.4 mg, 0.05 mmol) in toluene (2 mL). The reaction mixture was stirred at room temperature for 8 hours. The reaction was terminated by addition of acetic acid (~ 5 drops). The solvent was removed and the crude product was precipitated from methanol (200 mL) and then dried under vacuum. The resulting poly(-caprolactone) macroinitiator was characterized by 1H NMR (DP = 77 by endgroup analysis) and GPC. In a dry box, to a solution of the above described PCL (200 mg, 0.0225 mmol using 1H NMR DP) and VL (234 mg, 2.337 mmol) in toluene (2 mL) was added a solution of TBD (1.6 mg, 0.012 mmol) in toluene (0.337 mL). The reaction mixture was stirred for 30 minutes and then the reaction was terminated by addition of acetic acid. The crude product was precipitated from methanol (200 mL) and then dried under vacuum. The resulting poly(CL-b-VL) was analyzed by 1H, 13C NMR, GPC, DSC and WAXS.

87 Table 4.3. Characterization of various linear samples.

Entry t(h) a b c c o d e VL:CL conv. (%) %CL Mn (kg/mol) PDI Tm ( C) Hm (J/g)

1f 100:0 4 28 0 51 1.3 57 58

2 70:30 4 94 32 39 2.1 39/45 37

3 50:50 4 96 48 41 2.1 32/40 40

4 30:70 4 100 72 29 1.8 42/48 51

5f 0:100 8 37 100 90 1.3 56 56

6 60:40 1.5 94 47 21 1.7 21 53

7 50:50 1.5 84 55 36 1.7 21 47

PCL macro- 8 initiator - - - 13 1.4 - -

9 P(CL-b-VL) 1 87 49 18 1.9 51/55 69

10 P(CL-b-VL) 0.5 52 61 21 1.7 51/56 66

a Overall conversion of monomers; determined by 1H NMR spectroscopy b Percentage of caprolactone; determined by 13C NMR spectroscopy c Number average molecular weight and polydispersity index; determined by GPC with polystyrene calibration d Melting temperature; e f determined by differential scanning calorimetry (DSC) Heat of melting; determined by DSC Synthesized using Et2Al(OMe) catalyst

4.4.4. Synthesis of Crystalline-Amorphous Copolymers Further evidence for a gradient copolymer structure was obtained by the zwitterionic copolymerization of VL and 4-methyl--caprolactone (mCL) under conditions exactly analogous to those of the VL/CL copolymerizations. The resultant VL-mCL cyclic copolymer (VL/mCL = 76/24) was completely soluble in acetone, indicating that VL homopolymers (insoluble in acetone) are not generated under these conditions. Furthermore zwitterionic copolymerization of methylvalerolactone (mVL, a 1:1 mixture of 3-methyl--valerolactone and 4-methyl--valerolactone) and CL under similar conditions afforded a soluble cyclic copolymer (mVL/CL = 30/70), indicating that CL homopolymers are also not generated under these conditions. It should be noted that PVL and PCL homopolymers over the molecular weight Mn = 30 kg/mol are insoluble in acetone. These cyclic copolymers were synthesized using a similar procedure as in page 80 and the linear diblock copolymers (prepared for comparison) were synthesized using a similar procedure as in page 87.

88 Table 4.4. Crystalline – amorphous copolymers.

Figure 4.12. DSC scans of the crystalline – amorphous copolymers.

89 4.5 References (1) Zaremski, M. Y.; Kalugin, D. I.; Golubev, V. B. Polym. Sci. Ser. A 2009, 51, 103-122. (2) Matyjaszewski, K.; Ziegler, M. J.; Arehart, S. V.; Greszta, D.; Pakula, T. J. Phys. Org. Chem. 2000, 13, 775-786. (3) Shull, K. R. Macromolecules 2002, 35, 8631-8639. (4) Lefebvre, M. D.; de la Cruz, M. O.; Shull, K. R. Macromolecules 2004, 37, 1118-1123. (5) Cypryk, M.; Delczyk, B.; Juhari, A.; Koynov, K. J. Polym. Sci., Part A: Polym. Chem. 2009, 47, 1204-1216. (6) Yuan, W.; Mok, M. M.; Kim, J.; Wong, C. L. H.; Dettmer, C. M.; Nguyen, S. T.; Torkelson, J. M.; Shull, K. R. Langmuir, 26, 3261-3267. (7) Kapnistos, M.; Lang, M.; Vlassopoulos, D.; Pyckhout-Hintzen, W.; Richter, D.; Cho, D.; Chang, T.; Rubinstein, M. Nat. Mater. 2008, 7, 997-1002. (8) McKenna, G. B.; Hostetter, B. J.; Hadjichristidis, N.; Fetters, L. J.; Plazek, D. J. Macromolecules 1989, 22, 1834-1852. (9) Lescanec, R. L.; Hajduk, D. A.; Kim, G. Y.; Gan, Y.; Yin, R.; Gruner, S. M.; Hogenesch, T. E.; Thomas, E. L. Macromolecules 1995, 28, 3485-3489. (10) Roovers, J. In Cyclic Polymers; 2nd ed.; Semlyen, J. A., Ed.; Kluwer Academic Publishers: Dordrecht, 2000, p 347-384. (11) Semlyen, J. A. Cyclic Polymers; 2nd ed.; Kluwer Academic Publishers: Dordrecht, 2000. (12) Hur, K.; Winkler, R. G.; Yoon, D. Y. Macromolecules 2006, 39, 3975-3977. (13) Suzuki, J.; Takano, A.; Matsushita, Y. J. Chem. Phys. 2008, 129. (14) Minatti, E.; Viville, P.; Borsali, R.; Schappacher, M.; Deffieux, A.; Lazzaroni, R. Macromolecules 2003, 36, 4125-4133. (15) Matsushita, Y.; Iwata, H.; Asari, T.; Uchida, T.; ten Brinke, G.; Takano, A. J. Chem. Phys. 2004, 121, 1129-1132. (16) Zhu, Y. Q.; Gido, S. P.; Latrou, H.; Hadjichristidis, N.; Mays, J. W. Macromolecules 2003, 36, 148-152.

90 (17) Lecommandoux, S.; Borsali, R.; Schappacher, M.; Deffieux, A.; Narayaman, T.; Rochas, C. Macromolecules 2004, 37, 1843-1848. (18) Laurent, B. A.; Grayson, S. M. Chem. Soc. Rev. 2009, 38, 2202-2213. (19) Kricheldorf, H. R. J. Polym. Sci., Part A: Polym. Chem. 2010, 48, 251-284. (20) Bielawski, C. W.; Benitez, D.; Grubbs, R. H. Science 2002, 297, 2041-2044. (21) Yamamoto, T.; Tezuka, Y. Eur. Polym. J. 2011, 47, 535-541. (22) Culkin, D. A.; Jeong, W. H.; Csihony, S.; Gomez, E. D.; Balsara, N. R.; Hedrick, J. L.; Waymouth, R. M. Angew. Chem. Int. Ed. 2007, 46, 2627-2630. (23) Jeong, W.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2007, 129, 8414-8415. (24) Jeong, W.; Shin, E. J.; Culkin, D. A.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2009, 131, 4884-4891. (25) Guo, L.; Zhang, D. H. J. Am. Chem. Soc. 2009, 131, 18072-18074. (26) Storey, R. F.; Hoffman, D. C. Macromol. Symp. 1991, 42-3, 185-193. (27) Storey, R. F.; Herring, K. R.; Hoffman, D. C. J. Polym. Sci., Part A: Polym. Chem. 1991, 29, 1759-1777. (28) 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, 8574-8583. (29) Kamber, N. E.; Jeong, W.; Gonzalez, S.; Hedrick, J. L.; Waymouth, R. M. Macromolecules 2009, 42, 1634-1639. (30) The high polydispersities observed at high conversion may be due to transesterification by carbene and/or shuffling of growing chains. (31) Imasaka, K.; Nagai, T.; Yoshida, M.; Fukuzaki, H.; Asano, M.; Kumakura, M. Eur. Polym. J. 1990, 26, 831-836. (32) Bittiger, H.; Marchess.Rh; Niegisch, W. D. Acta Crystallogr., Sect. B: Struct. Sci 1970, B 26, 1923-&. (33) Furuhashi, Y.; Sikorski, P.; Atkins, E.; Iwata, T.; Doi, Y. J. Polym. Sci., Part B: Polym. Phys. 2001, 39, 2622-2634. (34) Kuhn, N.; Kratz, T. Synthesis 1993, 561-562.

91 (35) Odian, G. Principles of Polymerization; 4th ed.; Wiley-Interscience: Hoboken, NJ, 2004.

92 CHAPTER 5

Synthesis of Cyclic Poly(-caprolactone) and Attempts to Identify and Remove Linear Contamination

93 5.1 Introduction The influence of macromolecular architecture on polymer properties has stimulated the development of new synthetic methods for the generation of well-defined macromolecules.1-8 Synthetic methods to create branched4 and dendritic macromolecules2,7,9 have proven critical to investigations on the influence of branching on macromolecular behavior.10,11 Cyclic polymers constitute one of the simplest topological isomers of linear chains,8,12 yet the simple topological constraint of connecting the ends of a large macromolecule influences the properties in ways that remain poorly understood.13-16 One of the challenges in the study of cyclic macromolecules are the limited set of synthetic methods to prepare large macrocyclic chains.17-20 The formidable entropic constraints of cyclizing large chains21 have stimulated some creative solutions to generate cyclic polymers. The development of new classes of high-yielding coupling reactions20,22-24 and clever strategies for end-to- end coupling using electrostatic self-assembly,25,26 coupling at interfaces,27 in micelles28 or on solid-supports29,30 are representative strategies that have improved the classical methods for cyclizing linear chains at high dilution.12,16-18 Ring expansion strategies,19,31-33 which avoid the entropic penalty of bringing chain-ends in close proximity, provide another powerful strategy for the synthesis of cyclic macromolecules. Despite these advances, synthetic challenges remain, particularly preparing highly pure samples with no linear contaminants has proven to be a difficult task. There are studies showing that a small amount of linear contamination in the cyclic sample significantly changes the rheological behavior of cyclic polymer in the melt state. The melt viscosity of cyclic polymers increases with even trace amounts of linear contaminants.14,15,34,35 Fractional precipitation and preparative size exclusion chromatography (SEC) are the most commonly used methods to purify cyclic polymers from linear contaminants. However, these methods have not been fully successful. Lee et. al. used liquid chromatography at the critical condition (LCCC) to purify cyclic polystyrene from side products.36 In combined size exclusion and affinity chromatography, the critical condition is defined as the conditions under

94 which the retention of the analyte due to the affinity with the support is counterbalanced by the size exclusion. At the critical condition, the retention time of a linear polymer will be independent of the molecular weight. However, at the critical condition for a linear chain, theory predicts that a cyclic polymer of the same structure will elute at a different retention time, that depends on the size of the cyclic polymer chain relative to the pore size.37 Cyclic polystyrene samples that had been purified by fractional precipitation were subjected to preparative LCCC and successfully purified of linear side products. More recently, melt-state magic-angle spinning NMR spectroscopy has been used to identify linear impurity in cyclic poly(cyclooctene) (cPCOE) synthesized by ring-expansion metathesis polymerization (REMP). The NMR spectrum of cPCOE showed lack of endgroup signals in both the olefinic and alkyl regions. The purity of cPCOE was estimated to be over 90 % considering the sensitivity of the technique.33 Also, purification of cPCOE containing ammonium sites (cyclic polyammonium) was successful by using the selective threading of 24- crown-8 to linear polyammonium salts.38 In this work, a zwitterionic polymerization strategy to generate high molecular weight cyclic poly(-caprolactones) (cPCL) is described. A variety of chemical methods were used in an attempt to amplify, detect, and quantify linear contaminants in high molecular weight cyclic poly(-caprolactones). Also, several new methods to purify the cyclic polymer samples were investigated.

5.2 Synthesis of Cyclic Poly(-caprolactone) The zwitterionic ring-opening polymerization39-41 of lactones mediated by N- heterocyclic carbenes (NHC) provides an expedient strategy for generating cyclic polyesters.42-44 The ring-opening polymerization of lactide occurs within minutes at room temperature in the presence of 1,3-bis(2,4,6-(trimethylphenyl)imidazol-2- ylidene) (IMes) to generate cyclic poly(lactide)s (PLAs) with molecular weights in the 42,44 range of Mw = 5000 – 26,000 g/mol. We sought to extend the zwitterionic polymerization to other monomers to target a different molecular weight regime. We have previously shown that the aryl-substituted carbene IMes is inactive for the ring-

95 opening polymerization of -caprolactone (CL), but alkyl-substituted carbenes such as 1,3-diisopropyl-4,5-dimethylimidazol-2-ylidene (1) and 1,3,4,5-tetramethylimidazol- 2-ylidene (3) yield linear PCLs in the presence of alcohol initiators.45 We therefore investigated the zwitterionic ring-opening polymerization of CL using carbenes 1, 1,3- diethyl-4,5-dimethylimidazol-2-ylidene (2), and 3 (Figure 5.1) in the absence of alcohol initiators.

Figure 5.1. NHCs used for the zwitterionic ring-opening polymerization of lactones.

The zwitterionic polymerizations of -caprolactone (CL) were performed in toluene or tetrahydrofuran (THF). In the absence of alcohol initiators, the carbenes 1 – 3 mediate the ring-opening polymerizations of CL at room temperature to produce high molecular weight cyclic PCLs with molecular weights ranging from Mn = 41 kg/mol to Mn = 114 kg/mol (Scheme 5.1).

Scheme 5.1. Proposed mechanism for the zwitterionic ring-opening polymerization of CL.

96 The molecular weights of the PCLs obtained with carbenes 1 – 3 are considerably higher those of the cyclic PLAs generated by IMes,42,44 and the molecular weight distributions are broader, ranging from Mw/Mn = 1.36 (2, 33 % conversion, entry 3,

Table 5.1) to Mw/Mn = 2.16 (2, 74 % conversion, entry 4, Table 5.1).

Table 5.1. Selected data for zwitterionic ring-opening polymerization of CL with NHCs.

The reaction rate is slower for N-alkyl substituted carbenes with larger alkyl substituents (Figure 5.2), but carbenes 1 – 3 all exhibit kinetics that are first-order in caprolactone monomer (Figure 5.2) (carbene 1 exhibits an induction period, after which first order kinetics are observed). However, the evolution of molecular weight does not follow a linear relationship with conversion, and the molecular weights are not very sensitive to the initial ratio of monomer to initiator (M/I), which is similar to the lactide/IMes system42,44 (Figure 5.3). The non-zero intercept in the plot of molecular weight versus conversion suggests slow initiation compared to propagation. In addition, the initiation efficiency seems to be not so high and the rate of cyclization much slower than propagation since the molecular weights obtained are much higher than those theoretically predicted by the monomer to initiator ratios.

Figure 5.2. First-order kinetic plots for polymerization of CL with carbenes 1(),

2() and 3(). The polymerizations were carried out under the conditions, [M]0 =

1.0 M and [I]0 = 0.01 M.

Figure 5.3. Plots of molecular weight (filled symbols) and polydispersity index (open symbols) with conversion for polymerization done with carbene 1 at [M]0 = 1.0 M.

98 In our initial investigations, the carbenes were prepared by the reduction of the corresponding thione (equation (1)), as reported in the literature.46

While performing polymerizations using different batches of carbenes, it was observed that the conversions and molecular weights obtained each time were not consistent (Table 5.2, entries 1 – 4). It was difficult to predict the molecular weight of the polymer produced under the same conditions (initial concentrations, reaction time) when a different batch of carbene was used. The reason for this inconsistent behavior is still not well understood.

Table 5.2. Selected data of polymerization performed using thione derived carbene 3 (entries 1 – 4) and imidazolium salt derived carbene 3 (entries 5 – 7)

However, when the carbenes were prepared by deprotonation of the corresponding imidazolium salts (equation (2)),47 the polymerization behavior was more consistent and reproducible (Table 5.2, entries 5 – 7).48

99 H Cl cat. KOtBu R R rt R R N N NaH N N (2) THF

Evidence for a cyclic topology for the PCL's generated with carbenes 1 – 3 were provided by the absence of observed endgroups in the 1H NMR spectra and dilute solution viscosity measurements. A linear sample of PCL was prepared by the 49,50 ring-opening polymerization of CL with Et2Al(OMe). Comparison of the dilute solution viscosity of the linear PCL with that generated by the zwitterionic polymerization (entry 5 in Table 5.1) by gel permeation chromatography (GPC, in THF) coupled with a light scattering detector and viscometer showed that the cyclic PCL exhibits a lower intrinsic viscosity than that of the linear PCL of similar molecular weight, consistent with a cyclic architecture (Figure 5.4).51

Figure 5.4. Mark-Houwink plots of the linear () and cyclic () PCL sample (entry 5, Table 5.1).

The ratio, []cyclic/[]linear, is 0.72 (± 0.02) for molecular weights between 63 and 114 kg/mol. This ratio deviates from the theoretically predicted value for

[]cyclic/[]linear of 0.66, which assumes an infinitely freely jointed chain at θ

100 conditions, but as discussed by Goddard the ratio of []cyclic/[]linear can vary from 0.71 to 0.64, depending on the degree the solvation of the chains.52 Thus, the dilute solution viscosity measurements provide compelling evidence for a cyclic architecture, although we cannot rule out the presence of minor amounts of linear impurities.

5.3 Attempts to Quantify Linear Contaminants This section describes various attempts to identify and quantify any linear contamination in the cyclic PCL samples. The common theme was to use the hydroxyl endgroup of the linear polymer and attach a chromophore that can be detected by the UV detector connected to the GPC system, which would allow quantification in addition to simple detection of linear contamination (Scheme 5.2).

Scheme 5.2. Strategy for detecting and quantifying linear contamination in cyclic PCL samples.

For a linear polymer containing a UV-active label, the ratio of the area under the UV signal (A) to the intensity of the RI signal (I) is proportional to the ratio of the concentration of the chromophore (cPy) to the concentration of the repeat unit (cru), as in equation (3) (where, k is a machine constant).

dA kUV   cPy A dc c  Py  k  Py (3) I dn c k   c ru RI dc ru ru

101 For a sample that contains both linear and cyclic chains, if the linear chains can be selectively labeled with a chromophore, then the amount of linear chains in the sample can be determined by the amount of label incorporated, assuming that only the linear chains are functionalized with the UV-active label. If only the linear chains of a mixture are labeled, then A/I ratio of the labeled mixture relative to that of a fully labeled linear standard, should provide the percentage of linear chains in the original sample according to the following equation (4).

A I DP % of linear  100  c  l  c (4) I A DP c l l where, l and c stand for the standard linear sample and cyclic sample being analyzed, respectively, and DP is the degree of polymerization of the polymer. Since the molecular weights of the cyclic sample and linear standard are different, the degree of polymerization should be included. As we assume that any linear chains in a sample of cyclic polymers will contain an alcohol endgroup, we first attempted to functionalize the linear chain ends by esterification with a UV-active activated ester (Scheme 5.2, Figure 5.5). Considering that the polymer is a polyester, it is critical that the labeling method be sufficiently mild so that it will not result in opening and/or labeling of the cyclic chains. Several esterification protocols were attempted, as outlined in Figure 5.5. All reactions were carried out in tetrahydrofuran, so that the reaction mixture could be injected into the GPC directly after the reaction. Under a variety of conditions, esterification with carbonyldiimidazole did not proceed even at elevated temperature.

102

Figure 5.5. Esterification reactions used in this study.

Figure 5.6 shows a representative GPC trace of an esterification reaction using pyrene butanoyl chloride and triethylamine (TEA) in THF. In this experiment, a low molecular weight standard linear PCL and the (cyclic or linear) sample to be analyzed were mixed together with pyrene butanoyl chloride and TEA in THF. The reaction mixture was stirred for 4 hours at 55 oC. It is assumed that the standard linear PCL has reacted 100 % with pyrene butanoyl chloride during the reaction time. The reaction mixture was cooled and directly injected into the GPC. The areas under the peaks from each detector were measured using the TriSEC software and the percentage of linear was calculated using equation (4) described above (Table 5.3). The results show that the cyclic sample is over 400 % linear and even the linear sample to be 200 % linear. It seems that the reaction conditions are not mild enough that the cyclic polymer opens and reacts with the reagents. Also, the esterification reaction may not be as efficient as expected, so that there is error in assuming the standard linear PCL reacted 100 %. In addition, under the other reaction conditions, the Yamaguchi53 and Shiina54 esterification reactions, similar results were obtained.

103 (a) (b)

Figure 5.6. Representative GPC traces for the reaction of (a) linear PCL and (b) cyclic PCL with pyrene butanoyl chloride and TEA in THF in the presence of a linear standard.

Table 5.3. Calculation of % linear for samples in Figure 5.6.

To see whether the cyclic polymer is being opened, an end-capping reaction was carried out. When trimethylsilylchloride (TMS-Cl) was used as the end-capping reagent, a peak appeared around 3.6 ppm in the 1H NMR of the cyclic PCL that corresponds to the endgroup of a linear PCL capped with a TMS group (peak g in

104 Figure 5.7(b)). So, it was concluded that both esterification or silylation reactions were unsuitable for selective end-labeling of linear chains due to competitive cleavage/labeling of the cyclic chains.

Figure 5.7. NMR spectra of (a) linear PCL and (b) cyclic PCL treated with TMS-Cl (in the presence of TEA).

Another method was devised that relied on the reaction between alcohols and N-heterocyclic carbenes. Enders’ triazol-5-ylidene (Triaz) is known to react readily with alcohols to form O-H insertion adducts (alkoxy aminals). These adducts are stable at room temperature and reversibly eliminate the alcohol only at 90 oC.55,56 A test reaction using a linear PCL with Triaz was carried out (Scheme 5.3).

Scheme 5.3. Reaction of linear PCL with Triaz.

105 As seen in the NMR spectrum (Figure 5.8), the reaction was very slow, giving 85 % conversion in 2 days. The reaction did not proceed further upon leaving it for another few days.

Figure 5.8. NMR spectrum of a linear PCL reacted with Triaz carbene for 2 days (85 % conversion). The peaks labeled as ‘before’ and ‘after’ are the signals from the methylene (–CH2–OH) protons before and after reacting with Triaz, respectively.

Other problems of this reaction were the low sensitivity of the UV signal and the absorption wavelength of Triaz being at 250 nm, which is also the wavelength at which the carbonyl groups of the PCL absorb. Since there is only one Triaz per chain, the signal is very small and cannot be distinguished with the signal from the polymer carbonyl groups. To try overcome this problem, the synthesis of a triazol-5-ylidene with a pyrene group instead of a phenyl group was attempted. However, the step going from the diamine to the triazole-alcohol adduct was not successful (Scheme 5.4). The presence of the pyrene moiety may be imposing too much steric for the reaction to proceed. Furthermore, it is not certain that even if the alcohol adduct was successfully made, this compound would have the same reactivity (reversible formation of carbene with temperature) as Triaz.

106

Scheme 5.4. Unsuccessful step in the synthesis of 1,3-diphenyl-4-pyrene-4,5-dihydro- 1H-1,2,4-triazol-5-ylidene alcohol adduct.

A “macroinitiator” approach was also attempted. The terminal hydroxyl group of a linear contaminant can be used as the alcohol initiating group for the ring-opening polymerization of functionalized cyclic carbonate monomers to grow a functionalized polycarbonate chain57 on the end of the linear contaminant (Scheme 5.5). Functionalization of cyclic carbonate monomers derived from 2,2-bis(hydroxymethyl) -propionic acid is convenient, so a chromophore such as pyrene can be easily attached to give MTC-pyrene monomer.57,58 The 1-(3,5-bis-trifluoromethylphenyl)-3- cyclohexylthiourea/1,8-diazabicyclo[5.4.0]undec-7-ene (TU/DBU) catalyst system was chosen for the high selectivity of the TU/amine catalysts for tranesterification of cyclic lactones and carbonates relative to open-chain s-cis esters. While cyclic lactones and cyclic carbonates bind to the thiourea, open chain esters, such as ethyl acetate, exhibit no measurable binding affinity for TU under similar conditions.59 The higher H-bond basicity of s-cis esters of lactones/cyclic carbonates60 relative to that of the acyclic s-trans esters of the polymer chain is the likely origin of this high selectivity for ring-opening of monomer relative to transesterification of the chain.61 A test reaction with a linear PCL as the macroinitiator was performed. The GPC retention time changes from 28.6 minutes for the starting linear PCL to 28.52 minutes after the reaction showing that the MTC-pyrene is polymerized on the hydroxyl-terminated linear PCL. However, as seen in the GPC trace, this reaction is accompanied by homopolymerization of the MTC-pyrene (Figure 5.9). This approach

107 could become useful if the initiation efficiency can be improved and the homopolymerization of MTC-pyrene can be suppressed.62

Scheme 5.5. Polymerization of MTC-pyrene in the presence of a cyclic PCL sample.

Figure 5.9. GPC trace (RI detector signal) of linear PCL treated with MTC-pyrene in the presence of TU/DBU.

108 5.4 Removal of Linear Contaminants The main strategy for the removal of linear contaminants was similar to the identification, i.e. use the terminal hydroxyl group. A ‘linear scavenger’ was prepared by treating silica with trimethylaluminum.63 Hydroxyl terminated linear chains in a cyclic PCL sample would be expected to react with the surface Al-(CH3) groups of the

AlMe3-treated silica gel when a mixture of linear and cyclic polymers were eluted through a column of the linear scavenger (Figure 5.10).

Figure 5.10. Removal of linear contamination using trimethylaluminum treated silica.

When a dichloromethane solution of a cyclic PCL is filtered through the linear scavenger column, the shape and position of the polymer peak in the GPC trace did not change significantly (Figure 5.12(a)). The intrinsic viscosity of a filtered cyclic PCL sample is lower than that before filtration (Figure 5.11). Although dilute solution intrinsic viscosity of cyclic polymers is not very sensitive to linear contamination,12 the fact that the viscosity decreased after filtration gives support to the capture of linear contaminants by the linear scavenger. One problem of this linear scavenger is that the filtrations are not always consistent (Figure 5.12). As seen in Figure 5.12(b), something happens to the polymer while going through the linear scavenger column so that the molecular weight distribution is no longer mono-modal. This may be because of transesterification by some different aluminum species on the silica.

109

Figure 5.11. Mark-Houwink plots of a cyclic PCL sample before (□) and after (■) filtration through linear scavenger.

(a) (b)

Figure 5.12. GPC traces (RI detector signals) before (dashed) and after (solid) filtration for (a) a successful case and (b) a failed case.

So, to avoid this problem, polystyrene (PS) resins were tried out. Hydroxyl- functionalized PS resin was treated with diethylzinc to prepare ethylzinc- functionalized PS resin (Figure 5.13).

110

Figure 5.13. Removal of linear contamination using diethylzinc treated polystyrene resins.

It was expected that stirring the cyclic PCL sample with this resin would catch the linear contaminants. However, there was no way to know whether the resin was actually functionalized by the reaction with diethylzinc and a test using a fully linear sample showed that the functionalized PS resin doesn’t even catch all of the linear sample. There was polymer left over after treatment with the resin. So, a more easily characterizable and robust method should be sought.

5.5 Summary The zwitterionic ring-opening polymerization of -caprolactone provides a facile route to high molecular weight poly(-caprolactone). Alkyl-substituted imidazol-2-ylidene carbenes were used to polymerize CL to give high molecular weight cyclic PCLs. The rate of polymerization decreased with larger alkyl groups on the NHC. These systems showed similar polymerization characteristics (slow initiation compared to propagation, low initiation efficiency and slow cyclization compared to propagation) as the lactide/IMes system. The kinetics and mechanism are now being studied with imidazolium salt derived carbenes, which show more consistent behavior compared to the thione derived carbenes. Attempts to identify and quantify linear contamination in cyclic PCL samples using quantitative esterification were not successful due the nature of the polymer being a polyester. However, the macroinitiator approach showed promising results.

111 Scavenging agents exploiting the hydroxyl group of the linear contaminant were tried out to purify cyclic PCL samples. But these agents were not successful due to the difficulties in their characterization and interaction of the agent with the cyclic polymer.

112 5.6 Experimental Section General Considerations. All reactions and polymerizations were performed in a drybox or with Schlenk techniques under nitrogen. 1H nuclear magnetic resonance (NMR) spectra were recorded at room temperature on either a Varian 400 MHz or 500 MHz spectrometer, with shifts reported in parts per million downfield from tetramethylsilane and referenced to the residual solvent peak. Gel permeation chromatography (GPC) was performed in tetrahydrofuran (THF) at a flow rate of 1.0 mL/min on a Waters chromatograph equipped with four 5 μm Waters columns (300 mm×7.7 mm) connected in series. The Viscotek S3580 refractive index detector and Viscotek GPCmax autosampler were employed. The system with a triple detection system (Viscotek, Houston, TX) including a light scattering detector and viscometer was calibrated using monodisperse polystyrene standards (Polymer Laboratories). The right-angle light scattering (RALS) method was used to determine absolute molecular weights of polymers. Correction for any angular dissymmetry factor in the RALS data was performed in the TriSEC software using the viscometer signal. The angular dissymmetry correction is negligible because the polymers studied are relatively small compared to the laser wavelength (610 nm). The polymer solution (ca. 5 mg/mL) was prepared by dissolving the polymer in THF.

Materials. Tetrahydrofuran (THF) was distilled from sodium/benzophenone after being degassed three times via freeze-pump-thaw cycles. ε-Caprolactone was purchased from Aldrich and distilled from calcium hydride twice. Anhydrous methanol was purchased from Aldrich and used as received. 4-Nitrophenol was purchased from Fluka and purified by recrystallization from toluene followed by sublimation. 1,3-diisopropyl-4,5-dimethylimidazol-2-ylidene, 1,3-diethyl-4,5- dimethylimidazol-2-ylidene, 1,3,4,5-tetramethyl-imidazol-2-ylidene46,47 and 49 diethylaluminum methoxide (Et2Al(OMe)) were prepared according to the literature procedures.

113 Representative procedure for preparation of cyclic poly(ε-caprolactone). To a stirred solution of ε-caprolactone (2.28 g, 20.0 mmol) in THF (15.7 mL) was added a solution of IMe4 (3) (24.8 mg, 0.200 mmol) in THF (2.2 mL) at room temperature. The resulting solution was stirred at room temperature for 20 min. 4-Nitrophenol (55.6 mg, 0.400 mmol) was added and stirring was continued for 1 h. Conversion was determined by 1H NMR spectroscopy (40 % conversion). The polymer was purified by precipitation from methanol and analyzed by 1H NMR spectroscopy and GPC. 1H

NMR (CDCl3, 500 MHz): δ 1.32-1.44 (m, CH2CH2CH2), δ 1.58-1.70 (m,

CH2CH2CH2), δ 2.26-2.34 (t, OC(=O)CH2), δ 4.01-4.09 (t, C(=O)OCH2) ppm; GPC

(THF) Mn= 101 kg/mol (conventional PS calibration), PDI = 2.02 (conventional PS calibration), Mn = 89 kg/mol (universal calibration), PDI = 1.45 (universal calibration), and Mw = 129 kg/mol (light scattering).

Representative procedure for preparation of linear poly(ε-caprolactone). To a stirred solution of ε-caprolactone (1.14 g, 9.99 mmol) in THF (8.7 mL) was added a solution of Et2Al(OMe) (1.14 mg, 0.00982 mmol) in THF (0.19 mL) (from a stock solution) at room temperature. The resulting solution was stirred at room temperature for 8 h and the reaction was quenched by adding four drops of acetic acid. Conversion was determined by 1H NMR spectroscopy (90 % conversion). The polymer was purified by precipitation from methanol and analyzed by 1H NMR spectroscopy and 1 GPC. H NMR (CDCl3, 500 MHz): δ 1.32-1.44 (m, CH2CH2CH2), δ 1.58-1.70 (m,

CH2CH2CH2), δ 2.26-2.34 (t, OC(=O)CH2), δ 3.64 (m, CH2CH2OH), δ 3.66 (s,

C(=O)OCH3), δ 4.01-4.09 (t, C(=O)OCH2) ppm; GPC (THF) Mn = 140 kg/mol

(conventional PS calibration), PDI = 1.54 (conventional PS calibration), Mn = 93 kg/mol (universal calibration), PDI = 1.52 (universal calibration), and Mw = 101 kg/mol (light scattering).

Representative procedure for quantitative esterification. Cyclic PCL (11.5 mg, Mw

= 156 kg/mol by light scattering), a standard linear PCL (3.6 mg, Mw = 2500 g/mol by light scattering), pyrene butanoyl chloride (2 mg, 0.0065 mmol) and triethylamine (6.7

114 mg, 0.066 mmol) were mixed in THF (1 mL). The reaction mixture was stirred for 4 hours at 55 oC. The crude product solution was filtered through a syringe filter and then injected directly into the GPC system.

Procedure for MTC-pyrene polymerization with linear PCL. 1-(3,5- bistrifluoromethyl)phenyl-3-cyclohexylthiourea (5 mg, 0.0135 mmol) and 1,8- diazabicyclo[5.4.0]undec-7-ene (2 mg, 0.0131 mmol) were added to a solution of linear PCL (12.6 mg, Mw = 19700 g/mol by light scattering) and MTC-pyrene (38 mg, 0.09 mmol) in DCM (1 mL). The reaction mixture was stirred for 12 hours at room temperature. The reaction was terminated by addition of acetic acid (1 drop). The solvent was removed and the crude product was analyzed by gel permeation chromatography (GPC) using the RI and UV (320 nm) detectors.

Procedure for preparation of linear scavenger and filtration experiment. Trimethylaluminum (2 g) was added to a stirring THF solution (100 mL) of silica gel (10 g), which had been dried under vacuum at 150 oC for 3 days. The reaction mixture was stirred for 4 hours. The solvent was decanted and the product was dried under vacuum. To perform a filtration, the cyclic polymer sample was dissolved in DCM and the solution loaded on a column (a small pipette) filled with the linear scavenger (~ 50 mg). After the polymer solution was filtered, DCM was run through the column several times to collect residual polymer. The solvent was removed and the polymer was precipitated from pentane and then dried under vacuum.

115 5.7 References (1) Bielawski, C. W.; Grubbs, R. H. Prog. Polym. Sci. 2007, 32, 1-29. (2) Grayson, S. M.; Frechet, J. M. J. Chem. Rev. 2001, 101, 3819-3867. (3) Guan, Z. B. Chem. Asian J. 2010, 5, 1058-1070. (4) Hadjichristidis, N.; Pitsikalis, M.; Pispas, S.; Iatrou, H. Chem. Rev. 2001, 101, 3747-3792. (5) Hawker, C. J.; Wooley, K. L. Science 2005, 309, 1200-1205. (6) Kiesewetter, M. K.; Shin, E. J.; Hedrick, J. L.; Waymouth, R. M. Macromolecules 2010, 43, 2093-2107. (7) Rosen, B. M.; Wilson, C. J.; Wilson, D. A.; Peterca, M.; Imam, M. R.; Percec, V. Chem. Rev. 2009, 109, 6275-6540. (8) Tezuka, Y.; Oike, H. J. Am. Chem. Soc. 2001, 123, 11570-11576. (9) Tomalia, D. A.; Frechet, J. M. J. J. Polym. Sci., Part A: Polym. Chem. 2002, 40, 2719-2728. (10) Kapnistos, M.; Semenov, A. N.; Vlassopoulos, D.; Roovers, J. J. Chem. Phys. 1999, 111, 1753-1759. (11) Kasehagen, L. J.; Macosko, C. W. J. Rheol. 1998, 42, 1303-1327. (12) Cyclic Polymers; 2nd ed.; Semlyen, J. A., Ed.; Kluwer Academic Publishers: Dordrecht, 2000. (13) Hur, K.; Winkler, R. G.; Yoon, D. Y. Macromolecules 2006, 39, 3975-3977. (14) Kapnistos, M.; Lang, M.; Vlassopoulos, D.; Pyckhout-Hintzen, W.; Richter, D.; Cho, D.; Chang, T.; Rubinstein, M. Nat. Mater. 2008, 7, 997-1002. (15) McKenna, G. B.; Hostetter, B. J.; Hadjichristidis, N.; Fetters, L. J.; Plazek, D. J. Macromolecules 1989, 22, 1834-1852. (16) Roovers, J. In Cyclic Polymers; Second ed.; Semlyen, J. A., Ed.; Kluwer Academic Publishers: Dordrecht, 2000, p 347-384. (17) Deffieux, A.; Borsali, R. In Macromolecular Engineering. Precise Synthesis, Materials, Properties, Applications; Matyjaszewski, K., Gnanou, Y., Leibler, L., Eds.; Wiley-VCH: Weinheim, 2007, p 875-908. (18) Endo, K. In New Frontiers in Polymer Synthesis 2008; Vol. 217, p 121-183.

116 (19) Kricheldorf, H. R. J. Polym. Sci., Part A: Polym. Chem. 2010, 48, 251-284. (20) Laurent, B. A.; Grayson, S. M. Chem. Soc. Rev. 2009, 38, 2202-2213. (21) Jacobson, H.; Stockmayer, W. H. J. Chem. Phys. 1950, 18, 1600-1606. (22) Laurent, B. A.; Grayson, S. M. J. Am. Chem. Soc. 2006, 128, 4238-4239. (23) Misaka, H.; Kakuchi, R.; Zhang, C. H.; Sakai, R.; Satoh, T.; Kakuchi, T. Macromolecules 2009, 42, 5091-5096. (24) Stanford, M. J.; Pflughaupt, R. L.; Dove, A. P. Macromolecules 2010, 43, 6538-6541. (25) Tezuka, Y.; Mori, K.; Oike, H. Macromolecules 2002, 35, 5707-5711. (26) Tezuka, Y.; Ohtsuka, T.; Adachi, K.; Komiya, R.; Ohno, N.; Okui, N. Macromol. Rapid Commun. 2008, 29, 1237-1241. (27) Ishizu, K.; Ichimura, A. Polymer 1998, 39, 6555-6558. (28) Ge, Z. S.; Zhou, Y. M.; Xu, J.; Liu, H. W.; Chen, D. Y.; Liu, S. Y. J. Am. Chem. Soc. 2009, 131, 1628-1629. (29) Chisholm, M. H.; Gallucci, J. C.; Yin, H. F. Proc. Natl. Acad. Sci. 2006, 103, 15315-15320. (30) Wood, B. R.; Hodge, P.; Semlyen, J. A. Polymer 1993, 34, 3052-3058. (31) Bielawski, C. W.; Benitez, D.; Grubbs, R. H. Science 2002, 297, 2041-2044. (32) Boydston, A. J.; Xia, Y.; Kornfield, J. A.; Gorodetskaya, I. A.; Grubbs, R. H. J. Am. Chem. Soc. 2008, 130, 12775-12782. (33) Xia, Y.; Boydston, A. J.; Yao, Y. F.; Kornfield, J. A.; Gorodetskaya, I. A.; Spiess, H. W.; Grubbs, R. H. J. Am. Chem. Soc. 2009, 131, 2670-2677. (34) Nam, S.; Leisen, J.; Breedveld, V.; Beckham, H. W. Macromolecules 2009, 42, 3121-3128. (35) Roovers, J. Macromolecules 1988, 21, 1517-1521. (36) Lee, H. C.; Lee, H.; Lee, W.; Chang, T. H.; Roovers, J. Macromolecules 2000, 33, 8119-8121. (37) Gorbunov, A. A.; Skvortsov, A. M. Adv. Colloid Interface Sci. 1995, 62, 31- 108.

117 (38) Boydston, A. J.; Grubbs, R. H. Polym. Prepr. (Am. Chem. Soc., Div. Polym. Chem.) 2009, 50, 211-212. (39) Guo, L.; Zhang, D. J. Am. Chem. Soc. 2009, 131, 18072-18074. (40) Johnston, D. S. Adv. Polym. Sci. 1982, 42, 51-106. (41) Kricheldorf, H. R.; von Lossow, C.; Schwarz, G. Macromolecules 2005, 38, 5513-5518. (42) Culkin, D. A.; Jeong, W.; Csihony, S.; Gomez, E. D.; Balsara, N. P.; Hedrick, J. L.; Waymouth, R. M. Angew. Chem. Int. Ed. 2007, 46, 2627-2630. (43) Jeong, W.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2007, 129, 8414-8415. (44) Jeong, W.; Shin, E. J.; Culkin, D. A.; Hedrick, J. L.; Waymouth, R. M. J. Am. Chem. Soc. 2009, 131, 4884-4891. (45) Kamber, N. E.; Jeong, W.; Gonzalez, S.; Hedrick, J. L.; Waymouth, R. M. Macromolecules 2009, 42, 1634-1639. (46) Kuhn, N.; Kratz, T. Synthesis 1993, 561-562. (47) Arduengo, A. J.; Dias, H. V. R.; Harlow, R. L.; Kline, M. J. Am. Chem. Soc. 1992, 114, 5530-5534. (48) Mechanistic and kinetic studies of -caprolactone polymerization using NHCs are now being performed using imidazolium salt derived carbenes. (49) Duda, A.; Florjanczyk, Z.; Hofman, A.; Slomkowski, S.; Penczek, S. Macromolecules 1990, 23, 1640-1646. (50) Hofman, A.; Slomkowski, S.; Penczek, S. Makromol. Chem. Rapid. Commun. 1987, 8, 387-391. (51) McKenna, G. B.; Hadziioannou, G.; Lutz, P.; Hild, G.; Strazielle, C.; Straupe, C.; Rempp, P.; Kovacs, A. J. Macromolecules 1987, 20, 498-512. (52) Jang, S. S.; Cagin, T.; Goddard, W. A. J. Chem. Phys. 2003, 119, 1843-1854. (53) Inanaga, J.; Hirata, K.; Saeki, H.; Katsuki, T.; Yamaguchi, M. Bull. Chem. Soc. Jpn. 1979, 52, 1989-1993. (54) Shiina, I.; Ibuka, R.; Kubota, M. Chem. Lett. 2002, 286-287.

118 (55) Enders, D.; Breuer, K.; Raabe, G.; Runsink, J.; Teles, J. H.; Melder, J. P.; Ebel, K.; Brode, S. Angew. Chem. Int. Ed. 1995, 34, 1021-1023. (56) Coulembier, O.; Dove, A. P.; Pratt, R. C.; Sentman, A. C.; Culkin, D. A.; Mespouille, L.; Dubois, P.; Waymouth, R. M.; Hedrick, J. L. Angew. Chem. Int. Ed. 2005, 44, 4964-4968. (57) Pratt, R. C.; Nederberg, F.; Waymouth, R. M.; Hedrick, J. L. Chem. Commun. 2008, 114-116. (58) Sanders, D. P.; Fukushima, K.; Coady, D. J.; Nelson, A.; Fujiwara, M.; Yasumoto, M.; Hedrick, J. L. J. Am. Chem. Soc. 2010, 132, 14724-14726. (59) 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, 8574-8583. (60) Laurence, C.; Brameld, K. A.; Graton, J.; Le Questel, J. Y.; Renault, E. J. Med. Chem. 2009, 52, 4073-4086. (61) 1,5,7-diazabicyclo[4.4.0]dec-5-ene (TBD) readily transesterifies the backbone of cyclic PCL, so that it cannot be used for this purpose. (62) It has been shown through an independent experiment that DBU polymerizes cyclic carbonate monomers by itself. The initiation of the hydroxyl-terminated linear PCL by the TU catalyst is slow so that DBU-catalyzed polymerization of the cyclic carbonate monomer is competitive. Changing the amine cocatalyst to (-)-sparteine, which usually slows down the reaction rate but shows better control, did not solve the homopolymerization problem. (63) Li, J. H.; DiVerdi, J. A.; Maciel, G. E. J. Am. Chem. Soc. 2006, 128, 17093- 17101.

119