Synthesis of Polycatenanes Through Molecular Design

Home , Descriptor (Chemistry)

SYNTHESIS OF POLYCATENANES

THROUGH MOLECULAR DESIGN

QIONG WU

Submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Department of Macromolecular Science and Engineering

CASE WESTERN RESERVE UNIVERSITY

January, 2017 CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Qiong Wu

Candidate for the degree of Doctor of Philosophy.

Committee Chair:

Dr. Stuart Rowan

Committee Member:

Dr. Emily Pentzer

Committee Member:

Dr. Jon Pokorski

Committee Member:

Dr. Michael Hore

Date of Defense: 08/24/2016

Table of Contents Chapter 1. Introduction ...... 1

1. The Topological Bond ...... 1

2. The Synthetic Approach of Topologically Bonded Molecules ...... 3

2.1 Statistical Approach ...... 3

2.2 Templated Synthesis ...... 4

3. Macromolecules Containing Topological Bonds ...... 7

4. Main-chain Polyrotaxanes and Slide-ring Materials ...... 9

4.1 Synthesis of Poly[n]rotaxanes ...... 9

4.2 Applications of poly[n]rotaxanes ...... 11

4.3 Slide-ring gels (SLGs) and Slide-ring materials (SLMs) ...... 16

5. Topological main-chain polyrotaxanes ...... 22

6. Main-chain polycatenanes ...... 23

6.1 Poly[n]catenane ...... 23

6.2 Main-chain poly[2]catenane ...... 27

6.3 Radial polycatenane ...... 29

7. Side-chain polyrotaxanes and polycatenanes ...... 30

8. Interlocked polymeric materials ...... 30

9. Polycatenane network ...... 32

10. Conclusion ...... 33

Chapter 2. Optimizing the Formation of [3]Catenane through Molecular Design† ...... 41

1. Introduction ...... 41

2. The Effect of the Linker Group on the Thread ...... 45

3. The Effect of Side Chain Bulkiness of the Macrocycle ...... 48

4. Isolation and characterization of [3]catenane 8...... 52

5. Conclusion ...... 56

6. Experimental ...... 57

6.1 Materials and Methods ...... 57

6.2 Synthesis of Precursors and Components ...... 58

Chapter 3. Templated Synthesis toward Bip-containing Polymers and Catenanes ...... 70

1. Introduction ...... 70

2. Results and Discussion ...... 72

2.1 The metallo-supramolecular polymer ...... 72

2.2 The ring-closing metathesis (RCM) reaction ...... 77

2.3 Isolation of Bip-containing polymer and catenanes...... 83

2.4 The formation of polymer 17 and the improvement toward polycatenane ...... 85

2.5 Comparison with the Biphenyl-linked Monomer 16 ...... 88

2.6 Synthesis templated by Fe-MSP ...... 91

2.7 Hydrogenation of the double-bond ...... 92

3. Conclusion ...... 94

4. Experimental ...... 95

4.1 Materials and Methods ...... 95

4.2 Synthesis of Components ...... 97

Chapter 4. The Synthesis of Poly[n]catenanes ...... 106

1. Introduction ...... 106 ii

2. The Synthesis and Characterization of Polycatenane ...... 111

2.1 Component Design ...... 111

2.2 Polycatenane Synthesis and Purification ...... 113

2.3 Proof of Polycatenane Structure ...... 121

2.4 Polycatenane Architecture Determination ...... 136

2.5 Metallo Response of Polycatenane ...... 153

2.6 Side-reaction and Improvements ...... 154

3. Conclusion ...... 156

4. Supporting Information ...... 157

4.1 Experimental Details ...... 157

4.2 Supplementary Figures and Tables...... 165

Chapter 5. Synthesis of Poly[n]catenane using Fe2+ as Templating Metal ...... 173

1. Introduction ...... 173

2. Synthesis of Polycatenane Templated with Fe2+ ...... 174

3. Architecture Study of Fe2+ Templated Polycatenanes ...... 176

3.1 Molecular weight ...... 176

4. Attempts to Improve the Synthesis ...... 184

5. Conclusion ...... 186

6. Supporting Information ...... 187

6.1 The assembly of Fe-alt-MSP ...... 187

6.2 DOSY study of Fe-alt-MSP ...... 189

6.3 The Synthesis of 22Fe (and 22Fe-Oligo) via RCM Reaction ...... 190

6.4 Purification and Fractionation of Polycatenane ...... 190

iii

6.5 GPC-MALLS and 1H-NMR Study ...... 191

Chapter 6. Toward High-yield Synthesis of Linear Polycatenane ...... 192

1. Introduction ...... 192

2. Macrocycle with a Hexaethylene Glycol Linker Moiety ...... 196

3. Designing of Macrocycle with Rigid Bisphenol Z Linker Moiety ...... 199

4. Toward the Synthesis of Polycatenanes Containing the 5th Generation Macrocycle ...... 210

5. Application Perspective of Polycatenanes ...... 211

5.1 Mechanical property of a single polycatenane molecule ...... 211

5.2 Mechanical properties of polycatenane materials and composites ...... 212

6. Conclusion ...... 213

7. Supporting Information ...... 214

7.1 UV-Vis titration of G2MC ...... 214

7.2 DOSY study of Fe-G5-MSP ...... 215

7.3 Experimental ...... 216

List of Figures

Figure 1.1 The entanglement of two polymer chains. 1

Figure 1.2. Topological bonded molecules. 2

Figure 1.3. Movement of the macrocycle along the polymer chain of a polymeric [2]rotaxane. 3

Figure 1.4 Synthesis of topologically bonded molecules via statistical approach. 4

Figure 1.5. The templated approach. 5

Figure 1.6. Templated synthesis toward [2]rotaxane (a) or [2]catenane (b). 6

Figure 1.7. Interlocked molecules with more complicated topological structures. 7

Figure 1.8. Classification of topologically bonded macromolecules. 8

Figure 1.9. Efficient synthesis of poly[n]rotaxane from a linear polymer. 10

Figure 1.10. The molecular abacus. 11

Figure 1.11. The molecular shuttle machine. 13

Figure 1.12 Molecular wire, antenna and tubes. 14

Figure 1.13. (a) Illustration of biomimetic poly[n]rotaxane templated synthesis of sequence-

controlled polymer and (b) the chemical structure of the rotaxane template 15

Figure 1.14. The slide-ring gels. 16

Figure 1.15. A slide-ring hydrogel containing ionic functional groups. 18

Figure 1.16. Anti-scratch coating based on slide-ring materials. 19

Figure 1.17. Slide-ring elasticity. 20

Figure 1.18. Solvent-dependent mechanical property of slide-ring gel. 21

Figure 1.19. Topological polyrotaxanes. 22

Figure 1.20. A poly[2]rotaxane actuator. 23

Figure 1.21. Topological main-chain poly[n]catenane (a) and poly[2]catenane (b). 24

Figure 1.22. The poly[n]catenane structure is useful for everyday life. 24

Figure 1.23. Retrosynthetic analysis of poly[n]catenane. 25

Figure 1.24. Synthesis attempts of poly[n]catenanes. 26

Figure 1.25. Synthesis of main-chain poly[2]catenane by the reaction of X and Y. 27

Figure 1.26 The Hunter-Vogtle [2]catenane with bulky cyclohexane groups. 28

Figure 1.27 Reduction of amide improves the topological mobility. 28

Figure 1.28 Sauvage’s flexible poly[2]catenane. 29

Figure 1.29. Synthesis of radial polycatenane from pseudopolyrotaxane. 29

Figure 1.30. The switching process of pendent poly[2]catenane. 30

Figure 1.31. Polymeric [2]rotaxane (a), [2]catenane (b), and trefoil knot (c). 31

Figure 1.32. The polymeric [2]catenane. 32

Figure 1.33 Polymerization of 1,2-dithane (a) forms a polycatenane network (b). 33

Figure 2.1. Synthesis strategies of [3]rotaxane. 42

Figure 2.2 . Binding of Bip and transition metal. 44

Figure 2.3 Synthesis of [3]catenane via component design. 46

Figure 2.4 Comparison of linker groups 47

Figure 2.5. The effect of macrocycle side-chain bulkiness. 48

Figure 2.6 The synthesis of macrocycle 4. 49

Figure 2.7. The 1H-DOSY (aromatic region) of crude [3]catenates with n-hexyl or ethyl side

chains on macrocycle. 51

Figure 2.8. 1H-NMR comparison (aromatic region) of [3]catenane 8, with ring-closed thread 9

and macrocycle 4. 53 vi

Figure 2.9. 1H-NOESY of 8 showing selected cross-component NOE correlations. 54

Figure 2.10 MALDI-TOF of 2-10. Inset shows the isotope pattern. 55

Figure 2.11. TOF/TOF tandem mass spectrometry of 2-10 (m/z = 4117). 56

Figure 3.1 (a) Templated synthesis of DNA from another DNA as template and (b) templated

synthesis of protein from DNA and RNA templates. 70

Figure 3.2 Template synthesis of hollow organic nanotube from supramolecular assembled

structures. 71

Figure 3.3 Assembly of the metallosupramolecular polymer (MSP) and retro synthetic analysis

of polycatenane 15. 72

Figure 3.4 Monomer 14 with xanthene linker moiety (a), and 16 with biphenyl linker moiety (b).

1 2+ Figure 3.5 Full H-NMR showing the titration of Zn-MSP 14n+2·Zn 2n. 74

1 2+ Figure 3.6 H-NMR of monomer 14 and Fe-MSP 14n+2·Fe 2n. 74

Figure 3.7 DOSY of Monomer 14 (a) and its corresponding Zn-MSP (b) and Fe-MSP (c). 76

Figure 3.8 Illustration of the metathesis reaction of Zn-MSP (a) and the monomer 14 under high

concentration ADMET (b) or low concentration RCM (c). 78

Figure 3.9. 1H NMR comparison of the product from Zn-MSP templated synthesis reaction at 2.5

mM of 14, the ADMET and RCM of 14 without metal templation (synthesized at

concentration of 50 and 2.5 mM of 14, respectively), as well as starting material 14 and

purified 18. 79

Figure 3.10 GPC-RI of the product synthesized by Zn-MSP templated metathesis (a), ADMET at

high concentration without template (b), the reactant Monomer 14 (c) and RCM of 14 at

low concentration (d). 82 vii

Figure 3.11 1H-NMR (partial) and GPC (RI) of the crude and purified product. 84

Figure 3.12 Illustration of the bimodal distribution of polymer 17 with, and the formation of

catenane 15 and cyclized monomer 18 during Zn-MSP templated synthesis. 86

Figure 3.13 The monomer 14 shows (a) excellent catenane conversion toward a [3]catenate, but

(b) very low catenane conversion toward polycatenate. An improved strategy is proposed

Figure 3.14 Illustration of the metathesis reaction of monomer 16 with biphenyl liner moiety. 88

Figure 3.15 1H-NMR (partial) comparison of biphenyl-linked monomer (a) and its corresponding

metathesis products synthesized by low concentration “RCM” condition (b), high-

concentration ADMET condition (c), and MSP templated approach (d). 89

Figure 3.16 Partial 1H-NMR comparison of the of products synthesis from the monomers 14 and

16. 90

Figure 3.17 1H-NMR (partial) comparison of the crude polymer product synthesized by Zn-MSP

template (a) and Fe-MSP template. 92

Figure 3.18 Potential functionalization of the hydrogenated polycatenane. 93

Figure 3.19 Schematic illustration of the hydrogenation of 17 (a), and the comparison of GPC-RI

(b) and 1H-NMR (partial) of 17 and 20. 94

Figure 4.1 Topological flexibility of the polycatenane chain. 107

Figure 4.2. Reported synthetic approaches toward main-chain polycatenane and [5]catenanes. 108

Figure 4.3. Efficient one-bond-formation ring-closing synthesis of [3]catenane (a), the failure of

synthesizing polycatenane from Zn-MSP (b), and our modified approach toward the

synthesis of polycatenane (c). 110

Figure 4.4. Structure of (a) the monomer 14 and (b) the cyclized monomer 18. 111 viii

Figure 4.5 The molecular design of DiMC (a) and the structure of the first candidate 23 (b). The

ring-structure forbids the self-assembly between two DiMCs (c), but we still need to prevent

the self-compromised binding (d) during the molecular design. 113

2+ 2+ Figure 4.6. Comparison of (a) the UV-Vis spectra of 23, 24, 23·Fe and 242·Fe , and (b) the

plot of absorption of the Fe2+ titration with 23 and 24 (at λ = 382 nm) versus Fe2+/Bip ratio.

114

Figure 4.7 Self-assembly of the Zn-alt-MSP by NMR titration. 115

Figure 4.8. Synthesis of poly[n]catenane 22. 116

Figure 4.9. 1H-NMR of the polycatenane reaction product along with starting materials and

possible byproducts. 118

Figure 4.10. Summary of an efficient chromatography-free purification of polycatenane 22. 119

Figure 4.11. NMR study on the polycatenane 4-1 purification process. 120

Figure 4.12. Gradient-selected 1H-1H COSY of polycatenane 22. 123

Figure 4.13. Gradient-selected 1H-13C HMQC of polycatenane 22. 124

Figure 4.14. Gradient-selected 1H-13C HMBC of polycatenane 22. 125

Figure 4.15. NMR study on the ratio of red/blue rings in polycatenane 22. 126

Figure 4.16. 1H-1H NOESY of polycatenane 22. 128

Figure 4.17. 1H-1H NOESY of the 1:1 mixture of 23 and 17. 129

Figure 4.18. 1H-1H NOESY of the 1:1 mixture of 23 and 18. 130

Figure 4.19 GPC spectra of (a) the purified polycatenane 22 and the peak deconvolution fitting

and (b) shown with its isolated fractions 22a through 22d. 133

Figure 4.20 MALDI-TOF of polycatenane 22b. 135

Figure 4.21. (a) The fragments of polycatenanes are observed, which results in non-Poisson

distribution; and (b) the fragments of non-interlocked polymers are hardly observed, and the

polymer quasimolecular ion follows Poisson distribution. 136

Figure 4.22 [12]Catenanes with linear (a), branched (b) or cyclic (c) architectures. 137

Figure 4.23. Synthesis of oligomeric catenanes by controlling the stoichiometry of the starting

materials. 139

1 Figure 4.24. The H NMR comparison of HB peaks in polycatenane and oligomeric catenanes.

141

Figure 4.25. The (a) inversion-recovery spectra and (b) spin-echo CPMG spectra of the

polycatenane 22b with different τ. 143

1 Oligo Figure 4.26. H NMR comparison of deconvoluted HB peak in catenane 22 compared with

other oligomeric catenanes and polycatenane. 145

Figure 4.27 Preferred formation of red chain-ends. 146

Figure 4.28. The isolation of 22Cyclic-oligo and its NMR comparison with 22Oligo and the residue

catenane after purification. 147

Figure 4.29 1H-NMR (partial) comparison of polycatenanes and cyclic oligomeric catenane. 148

Figure 4.30 Polycatenanes synthesized from Zn-alt-MSP under different metathesis

concentration. 152

Figure 4.31. Metallo responsive conformation change of a polycatenane. 153

Figure 4.32. The formation of branched, cyclic polycatenane and byproduct. 156

Figure 5.1 Illustration of the formation of branched/linear/cyclic polycatenane mixture with Zn2+

as templating metal and the proposal of using Fe2+ as templating metal to optimizing the

synthesis of linear polycatenane. 174 x

Figure 5.2 The molecular structure of 14 and 23 and the assembly of Fe-alt-MSP monitored by

1H-NMR showing the aromatic region. 175

Figure 5.3 Synthesis approach of polycatenane using Fe2+ as templating metal ion. 176

Figure 5.4 GPC spectra of. (a) 22Fe compared with 22Zn , and (b) peak deconvolution of 22Fe and

1 Fe Figure 5.5. The H-NMR of the HB peak of (a) of polycatenane 22 and its four fractions; and

(b) the comparison of 22Fec-d and 22Znc-d. 179

Figure 5.6. (a) The backbiting reaction in metathesis polymerization. (b) Zn-alt-MSP is short and

rigid, not likely for the backbiting. (c) Fe-alt-MSP is rigid but long, so backbiting may

happen. (d) The formation of cyclic Fe-alt-MSP increases the local concentration of double-

bonds which is susceptible to backbiting reactions. 182

Figure 5.7 Proposed local-concenrated mechanism behind the high conversion toward branched

polycatenanes (a) and the proposed mechanism of the formation of cyclic polycatenane (b)

using Fe2+ as templating metal ion. 183

Figure 5.8 Modified synthesis of polycatenane. 184

Figure 5.9 GPC and 1H-NMR (partial) of polycatenane 22Fe-Oligo and its fractions 22Fe-Oligoa-c. 185

Figure 6.1 Summary of (a) the Zn-alt-MSP and Fe- alt-MSP templated synthesis of

polycatenane, (b) the side-reaction and (c) the shortening effect of the side reaction on the

molecular weight of polycatenane. 194

Figure 6.2 Suppression of the side reaction by redesigning a larger macrocycle (a) and the design

of new generation macrocycle (b). 196

Figure 6.3. (a) The structure and (b) the synthesis of G2MC with hexaethylene glycol linker

moiety. 197 xi

Figure 6.4 (a)Illustration of the self-compromise binding of the G2MC 28; and (b) the UV-Vis

2+ 2+ 2+ spectra comparison of the iron complex of G2MC (28·Fe ), 23·Fe and 242·Fe as well as

23, 24 and 28. The inset shows the structure of 24. 198

Figure 6.5 Stucture of the BPA (a), BPZ (b), the linker synhtesized from BPZ (c) and the

molecular structure of the 3rd generation macrocycle with BPZ linker moiety. 200

Figure 6.6 The UV-Vis spectrum of the iron complex of G3MC (30·Fe2+) compared with

2+ 2+ 2+ 28·Fe , 23·Fe and 242·Fe . 201

Figure 6.7 Functionalization possiblities of G3MC for improving solubility. 203

Figure 6.8 The molecular structure of the 4th generation macrocycle and its retro synthetic

analysis. 204

Figure 6.9 Full 1H NMR of G4MC 31 (purity: ca.85%). 205

Figure 6.10 The molecular structure (a) and synthesis (b) of the 5th generation macrocycle and its

retro synthetic analysis. 206

Figure 6.11 Full 1H-NMR of 32 (purity: ca. 95%) with peak assignment. 207

Figure 6.12 (a) The UV-Vis titration of 32 with Fe(NTf2)2, (b) comparison of 32, 23, 24 and their

2Bips/Fe2+ complex, and (c) the normalized peak intensity (392 nm, normalized by Bip

Moiety concentration) vs Fe2+/Bip ratio. 209

Figure 6.13 Assembly of Fe-G5-MSP by titration monitored by 1H NMR. 211

Figure 6.14. Prospective mechanical property study of polycatenane by single-molecular force

spectroscopy on an AFM cantilever. 212

Figure 6.15. Composite materials of polycatenane and traditional non-interlocked polymers. 213

xii

List of Tables

Table 2.1 Diffusion coefficient (D) and hydrodynamic radius (Rh) study...... 47

Table 2.2 Diffusion coefficients and hydrodynamic radii of the [3]pseudorotaxanes and

[3]catenates with n-hexyl or ethyl side-chains on macrocycles measured by 1H-DOSY. .... 52

Table 3.1 Diffusion Coefficient (D) and Hydrodynamic radius (Rh) of monomer 14 and

corresponding MSPs...... 75

Table 3.2 GPC-MALLS Molecular Weight Measurements ...... 82

Table 3.3 Molecular weight of purified polymer 17 synthesized by Zn-MSP template, ADMET

and RCM...... 85

Table 4.1 Absolute Molecular Weight of 4-1a through 4-1d...... 134

Table 4.2 Ratio of chain-end rings to total rings for polycatenanes shown in Figure 4.22...... 138

Table 4.3 T1 and T2 values of the HB peaks of polycatenane 22b at different region at −25 °C.

...... 143

Table 4.4 NC calculation of polycatenanes compared with [3]catenane and oligocatenane...... 149

Table 4.5 Calculation of NC excluding the contribution of cyclic polycatenane ...... 150

Table 4.6 Proposed relationship of 1H NMR and polycatenane architecture...... 152

Table 4.7 Diffusion coefficient and hydrodynamic radii of mostly linear polycatene 22c and

2+ * corresponding polycatenate 22c·Zn n ...... 153

Table 4.8 End-to-end DP of branched polycatenane and DP of mostly linear polycatenanes. .. 155

Table 5.1 NC calculation and architecture study of polycatenane 22Fe and its fractions...... 179

Fe-Oligo Table 5.2 NC calculation and architecture study of polycatenane 22 and its fractions...... 185

xiii

List of Supporting Figures and Tables

Figure S 4.1. The UV-Vis Titration of DiMC 23 (a) and Bip 24 (b) with Fe2+. 166

Figure S 4.2. DOSY of the Zn-alt-MSP (a) and Monomer 14. 167

Figure S 4.3. 1H-1H COSY (partial) of Zn-alt-MSP before and after the metathesis reaction 168

Figure S 4.4. Inversion-recovery and spin-echo CPMG of polycatenane measure at 25°C. 169

Cyclic-oligo Figure S 4.5. NMR integration of the HB peak for 22 170

Figure S 4.6. DOSY of mostly linear polycatenane 22c (a) and corresponding polycatenate

2+ 22c·Zn 2n+1 (b). 171

Figure S 5.1. 1H NMR comparison (aromatic region) of Fe-alt-MSP and Zn-alt-MSP. 188

Figure S 5.2. DOSY study of (a) Fe-alt-MSP, (b) Zn-alt-MSP and (c) Monomer 14 under 2.5

mM concentration (w.r.t. 14 moiety). 189

Figure S 6.1 UV-Vis titration of G2MC. The four spectra after reaching 2:1 Bip/Fe2+ ratio are

overlapped. 214

Figure S 6.2. 1H DOSY of Fe-G5-MSP measured in 1,1,2,2-tetrachloromethane-d2 at 25 °C, 500

MHz. 215

Table S 4.1. T1 and T2 of polycatenane measure at room temperature...... 169

xiv

Acknowledgements

First of all, thank Dr. Stuart Rowan for training me to a Ph.D. with all gracious advice, education and support during the past five years. Thank my committee members: Dr. Pentzer, Dr. Pokorski and Dr, Hore for the help on my graduation. Thank Dr. Rudy Wojtecki for teaching me a lot during my first 2 years of study. Special thank to all my collaborators: Xiaolong Lang (GPC-

MALLS), Dr. Prashansa Agrawal and Dr. Dale Ray (NMR) and Dr. Jim Faulk (MALDI-TOF).

Finally, and most importantly, thank you to my wife, Guolin Shang, for taking care of me during my PhD study. Thank my parents and parents-in-law for their support.

Synthesis of Polycatenanes Through Molecular Design

Abstract

QIONG WU

Main-chain poly[n]catenanes are topological macromolecules constructed by mechanically interlocking many small cyclic monomer rings into a long chain. The mechanical bonds provide a large degree of freedom to the polymer backbone and poly[n]catenanes are predicted to have exceptional molecular flexibility and robustness. However, the synthesis of main-chain poly[n]catenanes has not been achieved. This dissertation will discuss the first successful synthetic approach toward main-chain poly[n]catenanes, utilizing the ring-closing reaction of a pre- assembled metallosupramolecular polymer (MSP) template from rationally designed components with an isolation yield of 75~80%. Nuclear magnetic resonance (NMR) and gel permeation chromatography-multiangle laser light scattering (GPC-MALLS) confirms the mechanical interlocking structure and high molecular weight of the polycatenanes. Further study shows the product is a mixture of linear, branched and cyclic polycatenanes, which can be separated by size exclusion chromatography. Progress toward increasing the yield of polycatenane synthesis and improving the formation of linear polycatenane will also be described.

xvi

Chapter 1. Introduction

1. The Topological Bond

Entanglements of polymer chains play a very important role in viscoelasticity and rubber elasticity.1 For two entangled polymer chains as shown in Figure 1.1, the physical entanglement cannot be classified as either a covalent bond nor a supramolecular interaction. In order to detach the entanglement, without breaking any covalent bond, the polymer chains need to either move apart in different directions or one of the chains needs to reptate past the other chain until a chain end is found (which obviously depends the molecular weight of the polymer). As such, at short time scales, the entanglement can be considered as a class of a topological or mechanical bond.

Figure 1.1 The entanglement of two polymer chains.

In a molecule, the topological bond refers to the mechanically interlocking that brings two or more non-covalently attached components into one molecule. A prototypical example are the two interlocked rings in a [2]catenane shown in Figure 1.2a. Although there are no covalent bonds between the rings (or macrocycles), it is impossible to separate them without breaking any covalent bonds. In our daily life, the interlocked ring structure is also widely used for macroscopic chains.

Therefore, these molecules are named as catenanes which is derived from “catena”—the word

“chain” in Latin.2 The number “[2]” indicates there are two rings mechanically bonded together into one catenane.

Besides catenanes, rotaxanes, knots and interlocking networks also contains topological bonds. In a rotaxane molecule (figure 1.2b), two bulky stoppers that are covalently attached at both ends of the dumbbell topologically confine the macrocycle within the molecule. The number [2] also refers to the number of mechanically bonded components. If the stoppers are not present this structure is no longer a molecule but a supramolecular species (held together by supramolecular interactions between the macrocycle and thread) called a pseudorotaxane and it is one of the most important synthetic precursors toward both rotaxanes or catenanes. For a molecular knot, it contains one component that is topologically bonded with itself. Figure 1.2c shows the simplest molecular knot—a trefoil knot.

(a) (b) (c)

Figure 1.2. Topological bonded molecules. Figure illustrates a [2]catenane (a), a [2]rotaxane (b), and a trefoil knot (c).

A topological bond behaves distinctly from any covalent bonds or supramolecular interactions.3 For example, in a polymeric [2]rotaxane (Figure 1.3), the macrocycle can travel as far as the polymer chain reaches. Therefore, a topological bond has a theoretically unlimited degree of freedom on motions. On the contrary, covalent bonds may only show limited rotational mobility and very little elongation, compression or bending. Although supramolecular interaction may show good chain dynamics as a consequence of their ability to rapidly exchange,4,5 the robustness of supramolecular interactions is usually much lower than the mechanical bond which will not fail 2

before covalent bond failure.2 In a topologically bonded molecule, the mobility of topological bond is also tunable by molecular design, providing useful building blocks for smart materials and molecular machines.6,7

Figure 1.3. Movement of the macrocycle along the polymer chain of a polymeric [2]rotaxane.

2. The Synthetic Approach of Topologically Bonded Molecules

2.1 Statistical Approach

Statistical approach is a primitive strategy to access rotaxanes, (pseudo)polyrotaxanes and catenanes.6,8 Here there is no enthalpic driving force (such as specific supramolecular interactions) to enhance the formation of the interlocked species. For example, by using a macrocycle (e.g. crown ether) as a solvent for a reaction enhances the chance that the solvent macrocycle maybe threaded by reactant molecule as shown in Figure 1.4. Based on molecular modeling and

6 experimental studies, a minimum ring size of ca. C24 is required for the threading to happen.

Nevertheless, the product formation is usually uncontrollable and the yield is low—an early attempt toward a [2]catenane by Wasserman in 1960s only showed ca. 0.0001% conversion.9 This approach is also not capable for the synthesis of interlocked molecules with more complicated structures.

Figure 1.4 Synthesis of topologically bonded molecules via statistical approach.

2.2 Templated Synthesis

Ensuring that the two components are oriented correctly in order to form the targeted interlocked species, via the use of a template, is the key toward high yielding and more complicated interlocking structures.8,10 The template can be a covalent bond or a designed supramolecular interaction including metal-ligand binding,11 π-π stacking,12 hydrogen bonding,13 and hydrophobic effects,14 etc. Compared with the use of covalent templates, the supramolecular assemblies are much more popular for their convenience and versatility. Their performances are directly related to their conformational selectivity, binding strength and the structural design of precursor molecules.

Some widely used examples of supramolecular motifs used to access interlocked molecules are shown in Figure 1.5: (a) cyclodextrin (CD) is well known for its hydrophilic outside and hydrophobic cavity, in which a hydrophobic molecule can thread in a aqueous environment, driven by hydrophobic effect;15,16 (b) Sauvage’s metal templated synthesis based on the strong coordinating between 2,9-diphenyl-1,10-phenanthroline (dpp) and copper(I) ion;11 and (c) Stoddart’s π-π stacking approach with electron-rich and electron-poor aromatic moieties.12 4

Figure 1.5. The templated approach. The synthesis of interlocked molecules based on hydrophobic effect (a), metal-ligand interaction (b) or π- π stacking (c).

Mechanical bonds can be accessed via various approaches that employ supramolecular interactions. Figure 1.6 summarizes the templating approaches toward efficient synthesis of

[2]rotaxanes and [2]catenanes. The supramolecular interactions are playing the role of either holding a precursor with preferred conformation, or driving one or more components to the desired position, or both a combination of both.17 Depending on the strength of the supramolecular interaction, these techniques can promote the high yielding (even near quantitative) synthesis of interlocked molecules,18,19 and therefore open the door toward molecules with more complicate structures, as shown in Figure 1.7. Moreover, the synthesis of main-chain type polymeric rotaxanes and catenanes also becomes possible, the classification and synthesis of which will be discussed in the next section.

Figure 1.6. Templated synthesis toward [2]rotaxane (a) or [2]catenane (b). Black dotted line illustrates the templating interactions that drive to the formation of interlocked molecules.

Figure 1.7. Interlocked molecules with more complicated topological structures. Image shows the Solomon link20 (a), trefoil knot21 (b), pentafoil knot22 (c) and Borromean link23 (d) synthesized by templated approaches. Images are courtesy with permission from http://www.beautifulchemistry.net/ Copyright © 2016 USTC & TUP.

3. Macromolecules Containing Topological Bonds

Macromolecules with topological bonds, also called interlocked macromolecules, can be classified into four groups shown in Figure 1.8: (1) topological main-chain type (a,b); (2) non- topological main-chain type (c,d); (3) side-chain type (e,f); (4) polymeric interlocked molecules

(g); and (5) interlocked networks (h).2,17 In topological main-chain type polymers— poly[3]rotaxane (a) and poly[n]catenane (b), the topological bonds are an intrinsic part of the polymer backbone and contribute to the degree of polymerization (DP). For the non-topological main-chain type poly[n]rotaxane (c) and radial poly[n]catenane (d), although the mechanical bond is on the polymer main chain it is not an intrinsic part of the backbone. The side-chain type polyrotaxanes or polycatenanes (e,f) do not have the topological bond as part of the polymer backbone. Figure 1.8g shows an example of polymeric interlocked molecules—the polymeric 7

[2]catenane, in which two cyclic polymers are catenated together. Figure 1.8h is an example of polycatenated network. Details of these types of interlocked macromolecules will be discuss in the following sections.

Figure 1.8. Classification of topologically bonded macromolecules. Image illustrates the topological main-chain poly[3]rotaxane (a) and poly[n]catenane (b); non-topological main-chain poly[n]rotaxane and radial poly[n]catenane (d); side-chain poly[2]rotaxane (e) and poly[2]catenane (f); polymeric [2]catenane (g); and polycatenaned network (f).

4. Main-chain Polyrotaxanes and Slide-ring Materials

4.1 Synthesis of Poly[n]rotaxanes

Poly[n]rotaxanes (Figure 1.8c) have attracted a significant amount of research interest because of their versatility and ease of synthesis. The first synthesis of pseudopolyrotaxane dates back to 1976 for the synthesis of an “inclusion polyamide” by Ogata et al. as a polyamide train passing though many β-CD tunnels.14 Pseudopolyrotaxanes can also be prepared by polymerization in the presence of macrocycle as discussed in the statistical approach.

Pseudopolyrotaxanes usually behave differently from its individual components exhibiting different thermal and electrical properties, solubility and crystallinity, etc. due to the unique polyrotaxane structure. However, it is under debate that some examples reported as pseudopolyrotaxanes (prepared by polymerization of the monomer in the presence of macrocycles) may actually be radial polycatenanes (see Figure 1.8d), since some researchers believe the macrocycle will dethread from linear polymer in the absence of stopper groups, while the cyclization during polymerization a way to interlocked these macrocycles.17 The rate of dethreading observed in pseudopolyrotaxanes will be related to the strength of the interaction between the polymer backbone and the ring as well as the molecular weight of the polymer.

In 1990, Harada et al. reported the formation of pseudopolyrotaxane with cyclodextrin (CD) and poly(ethylene glycol) (PEG) and poly(propylene glycol) (PPG) by simply mixing. For PEG

200 or larger, the hydrophobic effect drives the PEG chain into the hydrophobic cavity of α-CD.16

Therefore, the molecular recognition mechanism provides an easy approach toward poly[n]rotaxane: mixing and end-capping, as shown in Figure 1.9. Other macrocycles e.g. crown

ethers,24 cucurbiturils,25 cyclophane26 and macrocycles containing ligands27 are also widely studied.

Figure 1.9. Efficient synthesis of poly[n]rotaxane from a linear polymer.

To convert a pseudopolyrotaxane into a polyrotaxane requires the addition of the large stopper group (Figure 1.9). A wide variety of chemistries have been employed here with the development of more efficient coupling reactions and click chemistries greatly improving the yield of the stopper attachment.

Besides the approach that starts with the macrocycle threads the polymer (either premade or during the polymerization) through it, poly[n]rotaxanes can also be prepared by ring-clipping28 or ring-opening polymerization of a [2]catenane.29 In addition, using larger macrocycles, doubly- stranded (pseudo)rotaxanes and (pseudo)polyrotaxanes have also been successfully prepared.30–32

4.2 Applications of poly[n]rotaxanes

4.2.1 Molecular abacus and shuttles

Topologically bonded molecules open the door toward various series of molecular machines. The movement of macrocycle along the polymer chain (previously discussed, as shown in Figure 1.3) inspires the molecular abacuses and shuttles. Shigekawa and coworkers demonstrated a controlled movement of a-CD ring(s) reversibly along PEG chain pushed mechanically by the tip of scanning tunneling microscope (STM, see Figure 1.10a), which is analogous to the operating of an abacus at the molecular scale.33 The molecular abacus can be precisely controlled for single ring movements (Figure 1.10b) and multiple ring movements

(Figure 1.10c). The major significance of this study is providing a good answer to a fundamental question about molecular machines: how can we precisely operate the machines to do an actual job.17

Figure 1.10. The molecular abacus. 11

Figure shows the structure illustration (left column) and STM image (right column) of the poly[n]rotaxane molecular abacus (a), simple shuttling movements (b) and pair shuttling movements (c). Courtesy with permission from Ref. [33]. Copyright © 2000, American Chemical Society.

The molecular abacus can also be considered as a STM tip powered molecular shuttle.

Through the careful design of the polyrotaxane molecule, the macrocycle is able to shuttle between two or more specific destinations upon certain stimulus, such as light, heat, or chemicals (solvent, pH, redox agents, metal ion), etc. Figure 1.11a shows a thermally responsive poly[n]rotaxane in which β-CDs shuttles between the end and center of the polymer chain. At low temperature the β-

CDs is observed to be interacting with the fluorescein moieties at both terminals. With increasing temperature, however, the interaction between β-CDs and PPG becomes significant, driving the

β-CDs toward the central PPG. 34

More precisely controlled molecular shuttle can be designed by engineering particular stations on the rotaxane and using easy controllable stimuli responsive transitions. As shown in

Figure 1.11b, when the porphyrin “P” harvests the visible-light and induces the electron transfer

+ to fullerene acceptor “A”, an electron shifts from tetrathiafulvalene “D1” oxidize it to “D1 ” which

4+ + repels the macrocycle “R ” toward “D2”. In the absence of light, “D1 ” is reduced back to “D1”, and its stronger interaction with “R4+” pulls the macrocycle backwards.35,36

Figure 1.11. The molecular shuttle machine. Figure shows a thermal responsive polyrotaxane (a)34 and a light powered molecular shuttle (B).36 Courtesy with permission from Ref. [34] Copyright © 1999, American Chemical Society (a) and Ref. [35] Copyright © 2009, Royal Society of Chemistry (b), respectively.

4.2.2 Insulated molecular wire, molecular antenna and nanotubes

Abacus and shuttles are based on the movement of the topological bond, while some other applications are based on the unique core-shell structure of poly[n]rotaxanes. Threading a conductive polymer chain through insulating macrocycles, can result in an insulated molecular wire. This concept was first reported by Anderson and coworkers, as shown in Figure1.12a.26

Encapsulated by water-soluble cyclophane macrocycles, the molecular wire has good water solubility without any modification on the conjugated backbone.

Figure 1.12 Molecular wire, antenna and tubes.

The structure of a water-soluble insulated molecular wire (a),26 molecular antenna (b)37 and the synthesis of α-CD molecular nanotube.38 Courtesy from Ref. [17] with permission. Copyright © 2004, Springer- Verlag.

Ueno and co-workers reported a molecular antenna based on PEG/α-CD polyrotaxanes

(Figure1.12b).37 By functionalizing a portion of α-CDs with naphthalene and terminating the pseudopolyrotaxane with anthracene, photons captured by naphthalene will transfer along the polyrotaxane to the terminal anthracene, resulting in fluorescence.

The polyrotaxane structure has also been used, by Harada and coworkers, as a template to allow access to a molecular tube. The synthetic procedure invovles assembling a high molecular weight PEG/α-CD poly[n]rotaxane, crosslinking the adjacent CD rings followed by removing the stopper and the PEG core. shown in Figure1.12c.38 The molecular weight of the CD tube is up to ca. 20,000, indicating the density of CD rings on the polyrotaxane is very high. The molecular tube

39 can be used for capturing targeted molecules such as C60.

4.2.3 Sequencing translation machine

The macrocycle movement and core-shell structure, discussed in previous two sections respectively, have been employed in the design of sequencing translation machine. Analogous to mRNA that translates the DNA information into a protein structure, conceptually the poly[n]rotaxane can translate its own sequence into a sequence-controlled polymer, as shown in

Figure 1.13.40 After the first monomer “A” attaches to the macrocycle, “A” no longer acts as a stopper and the macrocycle moves forward along the rotaxane. The second monomer “B” inserts between macrocycle and “A”, forming a growing chain. By repeating the two steps, the macrocycle will eventually dethread from the semi-dumbbell with sequenced polymer attaching on it.

Figure 1.13. (a) Illustration of biomimetic poly[n]rotaxane templated synthesis of sequence- controlled polymer and (b) the chemical structure of the rotaxane template Reprinted with permission from Ref. 40. Copyright © 2013, American Association for the Advancement of Science.

4.2.4 Building block for slide-ring materials

Another important application of poly[n]rotaxanes are as the precursors to various slide- ring gels and materials. The synthesis, properties and applications of these materials will be discussed in the next section.

4.3 Slide-ring gels (SLGs) and Slide-ring materials (SLMs)

In 2001, Ito and coworkers obtained a gel by crosslinking the macrocycle of the PEG/α-

CD poly[n]rotaxanes in solvent (Figure 1.14a).3,41 After gelation the material becomes extremely soft but super stretchy, and shows extraordinary ability on swelling. Due to the free movement between polymer chain and the macrocycle, the crosslinking is a topological bond that works like a pulley (Figure 1.14b).3,42 The pulley effect allows the network to rearrange and release the stress until fully extend, while traditional chemical gels tend to structurally degrade as stress is accumulated on the short chains (Figure 1.14c).42 Solid-state SLMs have also be prepared by crosslinking the polyrotaxane with a polymer in the absence of solvent. 3,43,44

Figure 1.14. The slide-ring gels. Synthesis of SLG by crosslinking the polyrotaxane (a), the pulley effect of topological crosslinks (b), and the stress accumulation of chemical gel versus the stress release of the SLG. Courtesy with permission from (a) Ref. [3] Copyright © 2014, Wiley Periodicals, Inc., (b,c) from Ref. [42] Copyright © 2007, Nature Publishing Group. 16

The slide-ring material is a great example showing the unique properties of topologically bonded polymers. Imran and coworkers prepared a polyelectrolyte slide-ring hydrogel with the polyrotaxane as a crosslinker. As expected, the hydrogel shows superior swelling ability up to

62,000% (Figure 1.15a) as the pulley effect allows polymer chains to fully extend beyond the crosslinking points.45 Unlike most supramolecular interactions, the robustness of the topological bond allows the slide-ring materials to access unusual and interesting mechanical properties. As we see in Figure 1.14b, the slide-ring hydrogel is highly stretchable, and even resistant to knife cutting.

The cutting resistance represents the behavior of topological bond in macromolecules. For conventional coating, the stress generated by sharps can easily break the chemical crosslinking network, resulting in a permanent scratch (Figure 1.16a). On the contrary, the topological crosslinking of a slide-ring material (SRM) easily relaxes the stress to reversible deformation and finally heals the stretch by elasticity (Figure 1.16b).3 The whole process is contributed by the collaboration of topological flexibility and the entropic elasticity.46,47 SRM based materials have been commercialized as anti-scratch coatings for automotive and electronics applications.48,49

Figure 1.15. A slide-ring hydrogel containing ionic functional groups. (a) The synthesis of polyelectrolyte SLG, and (b) the SLG shows 62,000% swelling in water, and (c) extraordinary stretchability and cutting resistance. Courtesy with permission from Ref. [45], Copyright © 2014, Nature Publishing Group.

Figure 1.16. Anti-scratch coating based on slide-ring materials. Image demonstrates the mechanism (a) and performance (b) of anti-scratch coating based on slide-ring materials. Courtesy with permission from Ref. [3] Copyright © 2014, Wiley Periodicals Inc.

Dynamic mechanical studies on SRMs show the presence of multiple elastic plateaus: the conventional rubbery elasticity and what has been proposed as the sliding elasticity (Figure

1.17a).46 For the PEG/CD SLG, the sliding of CD along the PEG chain is much slower than the micro-Brownian motion of PEG chain segment. Therefore, the temperature or frequency region between the glass transition of PEG and the movement onset of sliding is the rubbery elasticity stage, where the topological bond is frozen and the network behaves like a covalently crosslinked elastomer. At increased temperature or lowered frequency that allows the sliding movement of the

CD, SRM will be able to transit to a sliding relaxation, reaching the sliding elasticity plateau. The sliding transition comes with further decrease in modulus so the SRM will be extremely soft. The elasticity at sliding stage is different from that on the rubbery stage. As shown in Figure 1.17b, the entropy decrease of the polymer chain results in the entropic elasticity at the rubbery stage. After

sliding, the deformation of polymer chain relaxes but the uncrosslinked CD rings are squeezed together simultaneously.47 The heterogeneous distribution of CD rings generates an entropy loss which works against the chains rubbery elasticity, and the two entropic effects contribute to the sliding elasticity of the sliding stage. The stress-sensitive property has also been studied as pressure-responsive membranes.50

Figure 1.17. Slide-ring elasticity. Figure shows two elastic stages of slide-ring material (a) and the entropic elasticity after sliding (b). Courtesy with permission from: (a) Ref. [46], Copyright © 2011, Nature Publishing Group; (b) Ref. [47] Copyright © 2013, American Chemical Society.

The modulus of a SLG also strongly depends on the solvent it is swollen.46 Figure 1.18a shows a ca. 30-fold increase of Young’s modulus by adding water to the PEG/CD gel swollen in dimethyl sulfoxide (DMSO). Upon the addition of water as a poorer solvent, CD rings aggregate like nodules at the crosslinking points (Figure 1.18b), and the pulley effect is considerably suppressed. Therefore, it will behave more like a conventional chemical gel with much higher modulus (the rubbery stage).

Figure 1.18. Solvent-dependent mechanical property of slide-ring gel. Young’s modulus of SLG swollen in different solvent (a); the aggregation of CDs in poor solvent suppressed the pulley effect (b). Courtesy with permission from Ref. [46] Copyright © 2011, Nature Publishing Group

The application potential for SLGs as an electrolyte has also been examined in Lithium ion batteries, which are currently the most promising candidates for clean energy storage devices.51

For high storage density applications in electric vehicles or other devices, liquid electrolyte based batteries are highly vulnerable to physical damage (e.g. car accident) that may cause fire after the leakage of flammable electrolyte. A safe alternative would be a solid polymer electrolyte but this suffers from insufficient ionic conductivity; therefore, an organogel electrolyte with good mechanical properties a potential alternative candidate. The superior swelling ability allows SLGs 21

to swell up to 97 vol% electrolyte solution, retaining >95 % ionic conductivity of the liquid electrolyte. The stress-resistance property also prevents the electrolyte gel from fracturing under up to 85% compression.51

5. Topological main-chain polyrotaxanes

Topological polyrotaxanes have topologically bonded polymer main-chain. The structures showing in Figure 1.19ab are also known as “daisy-chain” polyrotaxane, which is a self- complementary array built up by one monomer through intermolecular recognition.7,52 For non- daisy-chain type polyrotaxanes, a [3]rotaxane is minimum for the repeat unit. Figure 1.19cd represents two types of poly[3]rotaxanes: singly-threaded (or single-stranded, c) and doubly- threaded (or double-stranded, d).

Figure 1.19. Topological polyrotaxanes. Topological poly[an]daisy-chain type poly[2]rotaxane (a); the poly[c2]daisy-chain type poly[2]rotaxane (b); and poly[3]rotaxanes with singly threading (c) or doubly threading (d) structures. Note: the descriptor of the daisy-chain, [c2] or [an], refers to cyclic/acyclic structure and the number of monomer units that make up the daisy chain structure. For details please refer to Reference 52.

The synthesis of topological polyrotaxanes has proved very difficult. The preparation of poly[an]daisy-chain,53 poly[c2]daisy-chain54,55 and the single-stranded poly[3]rotaxane56–58 have been reported only a few times, while the synthesis of doubly-threaded poly[3]rotaxanes remains an unmet challenge.

Topological polyrotaxanes are good candidate of actuators for artificial muscles. Based on the topological movement of the polymer chain, engineered stimuli-responsive will trigger the shuttling movement which will expand/contract the polymer chain. As shown in Figure 1.20, this concept was proven by an acid-based actuation of the poly[c2]daisy-chain, and an reversible molecular activation was observed.54

6. Main-chain polycatenanes

6.1 Poly[n]catenane

Poly[n]catenane, a molecular structure first proposed in 1953 by Frisch et al.8 as shown in

Figure 1.21 (a), is one of the most challenging molecules in synthetic chemistry. However, the

chain structure is very common in daily life (Figure 1.22). The beautiful polycatenane structure is the most popular one for necklace and bracelets designers. Its topological flexibility also brings great convenience for handling, storage and transportation. This structure is also popular in heavy- duty towing facilities for their superior structural mechanical property.

Figure 1.21. Topological main-chain poly[n]catenane (a) and poly[2]catenane (b).

Figure 1.22. The poly[n]catenane structure is useful for everyday life. A necklace with polycatenane structure (a) can be easily folded into a small bag or box (b) due to the flexibility of the chain; and a polycatenane chain for dockside crane (c) represents the robustness of this structure.

Although the successful synthesis of a poly[n]catenane has reported to date, researchers have been trying for more than 20 years using a variety of different approaches. A plausible strategy is the ring-closing of two entangled chains (Figure 1.23). From the polymer chemistry perspective, however, the entangled monomer contains four functionalities, which will be more likely to result in crosslinking rather than catenane formation. As such it is not a feasible approach 24

to polymerize a tetrafunctional monomer into a linear polymer since hyper-branching or even crosslinking are difficult to avoidable.2,6,17

In 1991, in a paper about the synthesis of polyrotaxanes, Shaffer and Tsay also proposed that cyclization of a linear entangled polymer might yield polycatenane, as shown in Figure

1.23b.59 However, to date this approach has not been successfully used to polycatenanes and some researchers believe that the low yield of cyclization reactions inhibits the likely hood of this strategy.6,8 Although the macrocyclic synthetic methodology advancements during the past two decades has greatly improved the yields of this class of molecules.

Figure 1.23. Retrosynthetic analysis of poly[n]catenane. Synthesis from two entangled semicircles and its actual product (a); and from entangled linear polymer proposed by Shaffer et al. in 1991 (b).

Up to date, the most successful approach toward poly[n]catenane is Stoddart’s stepwise synthesis,60 as shown in Figure 1.24a. After synthesizing the [3]catenane by templated ring- clipping reaction with two pre-formed macrocycles, further ring-clipping reaction yields linear

[5]catenane which is the longest reported linear poly[n]catenane isolated so far. Branched

[7]catenane as well as linear [4]catenane or other byproducts were also formed during the reaction.

With limited yields on each step, however, this approach is not suitable for the synthesis of much larger linear poly[n]catenanes.

In 2004, Takata and coworkers reported the synthesis of laddered poly[2]catenanes via

Diels-Alder polymerization (Figure 1.24b), and proposing further scissoring the double-bonds and

X-Y bonds yields the linear poly[n]catenane.61 This strategy is circumventing the cyclization reaction which is the major obstacle of poly[n]catenane formation. They successfully synthesized the bridged poly[2]catenane61 and also demonstrated the successful double bond scissoring on a small molecule derivative62. However, designing a system with easy cleavable X-Y bonds without interfering the Diels-Alder polymerization remains a big challenge.

Compared to other topological macromolecules, poly[n]catenane is very attractive target as its polymer chain is constructed solely by mechanical interlocking of cyclic monomers, and therefore possess the maximum topological flexibility. In order to get more knowledge about poly[n]catenanes, developing an effective synthetic strategy is currently the first priority.

Figure 1.24. Synthesis attempts of poly[n]catenanes. Stoddart’s Olympiadane: the synthesis of linear [5]catenane (a) and Takata’s protocol of converting from laddered poly[2]catenane by scissoring both double-bonds and X-Y bonds.

6.2 Main-chain poly[2]catenane

In comparison to the poly[n]catenane, the synthesis of main-chain poly[2]catenane (Figure

1.21b) is much more feasible owing to the high-efficient synthesis of [2]catenanes with variable structures that have been developed over the years. After the first synthesis of oligo[2]catenane by

Geerts and cowokers in 1995,63 various poly[2]catenanes have been reported64–68 in the following few years via similar synthetic methods—polymerization of a [2]catenane or copolymerization with other monomers (Figure 1.25).

Figure 1.25. Synthesis of main-chain poly[2]catenane by the reaction of X and Y.

Compared with fully topologically bonded poly[n]catenane, poly[2]catenanes much less topological bonds in their main-chain. Nevertheless, these mechanical bonds should also provide good topological flexibility to the polymer chain. However, poly[2]catenanes synthesized from

Hunter-Vogtle type [2]catenane13 (Figure 1.26) didn’t show the free rotation of the catenane ring even at very high temperature. The bulky cyclohexane groups on both macrocycles of the

[2]catenane prevent the rotations of the two rings and so inhibit the flexibility. When R1~R4 are not the same, conformational isomers can also be isolated.6 Even without the cyclohexane groups, the topological movement of amide-type catenanes is also severely hindered by hydrogen bonding between the rings. To solve this problem, Takata and coworkers reported that the reduction of amide into amine (Figure 1.27) greatly improves the topological mobility.69

O O N O N O H N H N H H R R R R 1 3 H 2 H 4 H N H N N O N O O O

Figure 1.26 The Hunter-Vogtle [2]catenane with bulky cyclohexane groups. The R groups can be used for functionalizing and polymerization into poly[2]catenanes.

Figure 1.27 Reduction of amide improves the topological mobility.

The first poly[2]catenane with good topological flexibility was reported by Sauvage and coworkers (Figure 1.28). The Kuhn segment length of this poly[2]catenane was measured to be

2.7 nm, which is a typical value of a flexible polymer in good solvent, although the poly[2]catenane contains ca. 50 wt% of highly rigid 2,9-di([1,1'-biphenyl]-4-yl)-1,10-phenanthroline moiety. It is an excellent example of accessing highly flexible materials with very rigid components (e.g. the bisbiphenyl phenanthroline moiety in poly[2]catenane) which is only available with topological bonded structure. A number of useful polymers (e.g. conjugated polymers and many stimuli- 28

responsive polymers) necessarily contain rigid components. The mechanical bond opens a door to utilize these building blocks for highly flexible materials.

O O O O O R O O R R' O N O N O O N N O O n R O R' R O O O O O O

Figure 1.28 Sauvage’s flexible poly[2]catenane.

6.3 Radial polycatenane

As a non-topological main-chain polycatenane, the structure of radial poly[n]catenane is analogous to the non-topological poly[n]rotaxane discussed in previous sections. The synthesis of radial polycatenane is shown in Figure 1.29: it can be prepared by cyclic polymerization of a pseudo[2]rotaxane70 (or ring-opening cyclic polymerization of [2]catenane), or from cyclization of a pseudopoly[n]rotaxane or poly[n]rotaxane.71 However, the synthetic difficulty has resulted in not much attention by researchers.

Figure 1.29. Synthesis of radial polycatenane from pseudopolyrotaxane. 29

7. Side-chain polyrotaxanes and polycatenanes

The molecular motion of topologically bonded small-molecular rotaxanes and catenanes attracts researcher’s attention for their applications in molecular machine and electronic devices.

Grafting them on polymers or surfaces can bring the molecular devices into applicable materials.

For example, Grzybowski and coworkers prepared a pendent poly[2]catenane from Stoddart’s bistable switchable [2]catenane, as shown in Figure 1.30. The poly[2]catenane can assemble into various polymer superstructures and maintain the switchable property of [2]catenane moieties.

8. Interlocked polymeric materials

Interlocked polymeric materials refer to topologically bonded polymers, including polymeric rotaxanes, polymeric catenanes and polymeric knots. As previously discussed, a polymeric

[2]rotaxane (Figure 1.31a) is a main-chain poly[n]rotaxane with fewer macrocycles and therefore

not considered as a new topological architecture molecule. The polymeric [2]catenane (Figure

1.31b) with one small-molecular ring and one polymeric ring is also similar to the radial polycatenane.

Figure 1.31. Polymeric [2]rotaxane (a), [2]catenane (b), and trefoil knot (c).

Polymeric [2]catenane with two interlocking cyclic polymer, however, is different from any types of polycatenanes discussed above. Due to the difficulty of ring-closing a telechelic polymer, synthesis of a polymeric [2]catenane by cyclization of a polymer chain is much more challenging than small molecular [2]catenanes. Successful preparation was reported by templated72,73 and statistical74 approaches. Takano and coworkers reported a targeted cyclization reaction method.75

After reacting with metallic potassium, the cyclic polystyrene-b-polyvinyl naphthalene acts as the reactant for the cyclization of telechelic polyisoprene, improving the chance of interlocking after cyclization. Similar to covalently linked block copolymers, catenated polystyrene/polyisoprene

(PS-c-PI) also shows lamella nanophase separation (Figure 1.32a). The domain spacing of PS-c-

PI is between their linear block copolymer and cyclic block copolymer with same molecular weight. Recently, Advincula and coworkers reports a high-yield approach by ring-expansion polymerization from small molecular [2]catenane, as shown in Figure 1.32b.76 This approach fundamentally circumvents the cyclization of a polymer chain, opening a door of large-scale synthesis of polymeric [2]catenane. Moreover, the ring-expansion approach also provides a

feasible method for the synthesis of polymeric trefoil knot77 (Figure 1.31c) which is difficult to synthesize by other methods.6

Figure 1.32. The polymeric [2]catenane. (a) The transmission electron micrograph (TEM) of a polymeric [2]catenane PS-c-PI, and (b) a schematic the ring-expansion synthesis of polymeric [2]catenane from small molecule [2]catenane. Courtesy with permission from: (a) Ref. [75] Copyright © 2008, American Chemical Society; (b) Ref. [76] Copyright © 2012, Royal Society of Chemistry.

9. Polycatenane network

The slide-ring gels/materials are a polyrotaxane network. Network structure are also able to be built with catenanes. Endo and coworkers report many unusual behaviors of cyclic polymerized

1,2-dithiane in the absence of linear thiol (Figure 1.33a). After polymerization, the cyclic poly(1,2-

Dithiane) shows rubber-like behavior including rubbery plateau on stress-strain curve and dynamic viscoelasticity measurement, gel-like swelling in good solvents and decreasing of Tg with increasing molecular weight. All these behaviors are not observed in linear poly(1,2-DIthiane) or conventional cyclic polymers and are explained by the proposal that a polycatenane network

(Figure 1.33b) is formed during the polymerization.

Figure 1.33 Polymerization of 1,2-dithane (a) forms a polycatenane network (b).

10. Conclusion

Different from either covalent bond or supramolecular interactions, the topological bond is a mechanical interlocking interaction that brings two or more non-covalently linked components into one molecule with a topological structure. Topologically bonded molecules including rotaxanes, catenanes and knots can be efficiently synthesized by templating methods. In these molecules, the topological bonds are able to provide extra free or tunable movements based on the molecular design.

Incorporating topologically bonded molecules into polymers provides the polymer materials with structural flexibility and/or molecular-level stimuli-responsive properties. These materials can be classified into main-chain topological polymers, main-chain non-topological polymers, side-chain type polymers, polymeric interlocked molecules and networks. In the family of topologically interlocked macromolecules, some family members can be easily synthesized by current techniques, including main-chain poly[n]rotaxanes, slide-ring materials based on polyrotaxanes, poly[2]catenanes and side-chain polyrotaxanes/polycatenanes. The unique structures and properties have been studied and have shown application potential for ultraflexible elastomers, molecular machines, nanoelectronics as well as models for fundamental polymer

science studies or templates for more complicated structures. For other family members, synthetic difficulties still limit further studies on their unique/unusual properties. In particular, the main- chain poly[n]catenane is predicted to be the most flexible molecule in the family, and may show a number of unique properties. However, the synthesis of poly[n]catenane remains a huge challenge.

Developing an effective and efficient approach to prepare the poly[n]catenane is the top priority and also the main focus of this thesis.

References:

1 Y. H. Lin, Polymer Viscoelasticity: Basics, Molecular Theories, Experiments, and

Simulations, World Scientific, 2011.

2 Z. Niu and H. W. Gibson, Chem. Rev., 2009, 109, 6024–6046.

3 Y. Noda, Y. Hayashi and K. Ito, J. Appl. Polym. Sci., 2014, 131, 40509.

4 Y. Jia and X. Zhu, Chem. Mater., 2015, 27, 387–393.

5 S. Hackelbusch, T. Rossow, P. van. Assenbergh and S. Seiffert, Macromolecules, 2013,

46, 6273–6286.

6 J. P. Sauvage and C. Dietrich-Buchecker, Molecular Catenanes, Rotaxanes and Knots,

Wiley-VCH Verlag GmbH, Weinheim, Germany, 1999.

7 L. Fang, M. A. Olson, D. Benítez, E. Tkatchouk, W. A. Goddard III and J. F. Stoddart,

Chem. Soc. Rev., 2010, 39, 17–29.

8 H. W. Gibson, M. C. Bheda and P. T. Engen, Prog. Polym. Sci., 1994, 19, 843–945.

9 E. Wasserman, J. Am. Chem. Soc., 1960, 82, 4433–4434.

10 G. Gil-Ramírez, D. A. Leigh and A. J. Stephens, Angew. Chemie Int. Ed., 2015, 54, 6110–

6150.

11 J. P. Sauvage and J. Weiss, J. Am. Chem. Soc., 1985, 107, 6108–6110.

12 P. R. Ashton, T. T. Goodnow, A. E. Kaifer, M. V. Reddington, A. M. Z. Slawin, N.

Spencer, J. F. Stoddart, C. Vicent and D. J. Williams, Angew. Chemie Int. Ed., 1989, 28,

1396–1399.

13 C. a Hunter, J. Am. Chem. Soc., 1992, 114, 5303–5311.

14 N. Ogata, K. Sanui and J. Wada, J. Polym. Sci. Polym. Lett. Ed., 1976, 14, 459–462.

15 J. P. Behr and J. M. Lehn, J. Am. Chem. Soc., 1976, 98, 1743–1747.

16 A. Harada and M. Kamachi, Macromolecules, 1990, 23, 2821–2823.

17 T. Takata, N. Kihara and Y. Furusho, in POLYMER SYNTHESIS, SPRINGER-VERLAG

BERLIN, Berlin, Heidelberg, 2004, vol. 171, pp. 1–75.

18 M. Weck, B. Mohr, J.-P. Sauvage and R. H. Grubbs, J. Org. Chem., 1999, 64, 5463–5471.

19 R. J. Wojtecki, Q. Wu, J. C. Johnson, D. G. Ray, L. T. J. Korley and S. J. Rowan, Chem.

Sci., 2013, 4, 4440.

20 C. D. Pentecost, K. S. Chichak, A. J. Peters, G. W. V. Cave, S. J. Cantrill and J. F.

Stoddart, Angew. Chemie Int. Ed., 2007, 46, 218–222.

21 N. Ponnuswamy, F. B. L. Cougnon, J. M. Clough, G. D. Pantos and J. K. M. Sanders,

Science, 2012, 338, 783–785.

22 J.-F. Ayme, J. E. Beves, D. A. Leigh, R. T. McBurney, K. Rissanen and D. Schultz, Nat.

Chem., 2011, 4, 15–20.

23 K. S. Chichak, S. J. Cantrill, A. R. Pease, S.-H. Chiu, G. W. V. Cave, J. L. Atwood and J.

F. Stoddart, Science, 2004, 304, 1308–1312.

24 S. J. Cantrill, M. C. T. Fyfe, A. M. Heiss, J. Fraser Stoddart, A. J. P. White and D. J.

Williams, Chem. Commun., 1999, 1251–1252.

25 C. Meschke, H.-J. Buschmann and E. Schollmeyer, Macromol. Rapid Commun., 1998, 19,

59–63.

26 S. Anderson and H. L. Anderson, Angew. Chemie Int. Ed., 1996, 35, 1956–1959.

27 P. L. Vidal, M. Billon, B. Divisia-Blohorn, G. Bidan, J. M. Kern and J. P. Sauvage, Chem.

Commun., 1998, 629–630.

28 J. Wu, K. C.-F. Leung and J. F. Stoddart, Proc. Natl. Acad. Sci., 2007, 104, 17266–17271. 36

29 S. Kang, B. M. Berkshire, Z. Xue, M. Gupta, C. Layode, P. A. May and M. F. Mayer, J.

Am. Chem. Soc., 2008, 130, 15246–15247.

30 A. Harada, J. Li and M. Kamachi, Nature, 1994, 370, 126–128.

31 P. R. Ashton, P. T. Glink, C. Schiavo, J. F. Stoddart, E. J. T. Chrystal, S. Menzer, D. J.

Williams and P. a Tasker, Angew. Chemie Int. Ed., 1995, 34, 1869–1871.

32 X. Tong, P. Gao, X. Zhang, L. Ye, A. Zhang and Z. Feng, Polym. Int., 2010, 59, 917–922.

33 H. Shigekawa, K. Miyake, J. Sumaoka, A. Harada and M. Komiyama, J. Am. Chem. Soc.,

2000, 122, 5411–5412.

34 H. Fujita, T. Ooya and N. Yui, Macromolecules, 1999, 32, 2534–2541.

35 V. Balzani, A. Credi and M. Venturi, Chem. Soc. Rev., 2009, 38, 1542.

36 S. Saha, A. H. Flood, J. F. Stoddart, S. Impellizzeri, S. Silvi, M. Venturi and A. Credi, J.

Am. Chem. Soc., 2007, 129, 12159–12171.

37 M. Tamura, D. Gao and A. Ueno, Chemistry, 2001, 7, 1390–1397.

38 A. Harada, J. Li and M. Kamachi, Nature, 1993, 364, 516–518.

39 Y. Liu, Z.-X. Yang, Y. Chen, Y. Song and N. Shao, ACS Nano, 2008, 2, 554–560.

40 B. Lewandowski, G. De Bo, J. W. Ward, M. Papmeyer, S. Kuschel, M. J. Aldegunde, P.

M. E. Gramlich, D. Heckmann, S. M. Goldup, D. M. D’Souza, A. E. Fernandes and D. a.

Leigh, Science, 2013, 339, 189–193.

41 Y. Okumura and K. Ito, Adv. Mater., 2001, 13, 485–487.

42 K. Ito, Polym. J., 2007, 39, 489–499.

43 J. Araki and K. Ito, Soft Matter, 2007, 3, 1456.

44 K. Ito, Curr. Opin. Solid State Mater. Sci., 2010, 14, 28–34.

45 A. Bin Imran, K. Esaki, H. Gotoh, T. Seki, K. Ito, Y. Sakai and Y. Takeoka, Nat. 37

Commun., 2014, 5, 5124.

46 K. Ito, Polym. J., 2012, 44, 38–41.

47 K. Kato, T. Yasuda and K. Ito, Macromolecules, 2013, 46, 310–316.

48 L. Nissan Motor Co., Nissan Licensing Bussiness, http://www.nissan-

global.com/EN/LICENSE/PDF/technology01.pdf.

49 L. Nissan Motor Co., Yet2.com Tech of the Week, 2010, September, tow0059972.pdf.

50 C. Katsuno, A. Konda, K. Urayama, T. Takigawa, M. Kidowaki and K. Ito, Adv. Mater.,

2013, 25, 4636–4640.

51 N. Sugihara, Y. Tominaga, T. Shimomura and K. Ito, Electrochim. Acta, 2015, 169, 433–

439.

52 S. J. Cantrill, G. J. Youn, J. F. Stoddart and D. J. Williams, J. Org. Chem., 2001, 66,

6857–6872.

53 M. Zhang, S. Li, S. Dong, J. Chen, B. Zheng and F. Huang, Macromolecules, 2011, 44,

9629–9634.

54 L. Fang, M. Hmadeh, J. Wu, M. A. Olson, J. M. Spruell, A. Trabolsi, Y. Yang, M.

Elhabiri, A.-M. Albrecht-Gary and J. F. Stoddart, J. Am. Chem. Soc., 2009, 131, 7126–

7134.

55 E. N. Guidry, J. Li, J. F. Stoddart and R. H. Grubbs, J. Am. Chem. Soc., 2007, 129, 8944–

8945.

56 T. Oku, Y. Furusho and T. Takata, J. Polym. Sci. Part A Polym. Chem., 2003, 41, 119–

123.

57 P. Wei, X. Yan and F. Huang, Chem. Commun., 2014, 50, 14105–14108.

58 T. Sato and T. Takata, Polym. J., 2009, 41, 470–476. 38

59 T. D. Shaffer and L.-M. Tsay, J. Polym. Sci. Part A Polym. Chem., 1991, 29, 1213–1215.

60 D. B. Amabilino, P. R. Ashton, A. S. Reder, N. Spencer and J. F. Stoddart, Angew.

Chemie Int. Ed., 1994, 33, 1286–1290.

61 N. Watanabe, Y. Ikari, N. Kihara and T. Takata, Macromolecules, 2004, 37, 6663–6666.

62 N. Watanabe, N. Kihara and T. Takata, Org. Lett., 2001, 3, 3519–3522.

63 Y. Geerts, D. Muscat and K. Müllen, Macromolecular Chemistry and Physics, 1995, 196,

3425–3435.

64 J. Weidmann, J.-M. Kern, J. Sauvage, Y. Geerts, D. Muscat and K. Müllen, Chem.

Commun., 1996, 1243–1244.

65 J.-L. Weidmann, J. Kern, J. Sauvage, D. Muscat, S. Mullins, W. Köhler, C. Rosenauer, H.

J. Räder, K. Martin and Y. Geerts, Chem. Eur. J., 1999, 5, 1841–1851.

66 D. Muscat, A. Witte, W. Köhler, K. Müllen and Y. Geerts, Macromol. Rapid Commun.,

1997, 18, 233–241.

67 D. Muscat, W. Köhler, H. J. Räder, K. Martin, S. Mullins, B. Müller, K. Müllen and Y.

Geerts, Macromolecules, 1999, 32, 1737–1745.

68 C. Hamers, O. Kocian, F. M. Raymo and J. F. Stoddart, Adv. Mater., 1998, 10, 1366–

1369.

69 T. Takata, J. Shoji and Y. Furusho, Chemistry Letters, 1997, 881–882.

70 F. Bitsch, C. O. Dietrich-Buchecker, A. K. Khemiss, J. P. Sauvage and A. Van Dorsselaer,

J. Am. Chem. Soc., 1991, 113, 4023–4025.

71 M. Okada and A. Harada, Macromolecules, 2003, 36, 9701–9703.

72 Ö. Ünsal and A. Godt, Chem. Eur. J., 1999, 5, 1728–1733.

73 A. Bunha, P.-F. Cao, J. Mangadlao, F.-M. Shi, E. Foster, K. Pangilinan and R. Advincula, 39

Chem. Commun., 2015, 51, 7528–7531.

74 Y. Gan, D. Dong and T. E. Hogen-Esch, Macromolecules, 2002, 35, 6799–6803.

75 Y. Ohta, Y. Kushida, D. Kawaguchi, Y. Matsushita and A. Takano, Macromolecules,

2008, 41, 3957–3961.

76 P.-F. Cao, A. Bunha, J. Mangadlao, M. J. Felipe, K. I. Mongcopa and R. Advincula,

Chem. Commun., 2012, 48, 12094.

77 P.-F. Cao, J. Mangadlao and R. Advincula, Angew. Chemie Int. Ed., 2015, 54, 5127–5131.

Chapter 2. Optimizing the Formation of [3]Catenane

through Molecular Design†

1. Introduction

There is an intense and growing interest in the design, assembly and utilization of mechanically interlocked molecules, such as catenanes and rotaxanes,1 that show significant promise for the advancement of a number of nanoscience and nanotechnology arenas. Intense efforts have led to a rapid maturation of the field, bringing dramatic improvements in synthetic methodologies which can now produce a variety of increasingly complex interlocked small molecule and polymeric architectures with remarkable efficiency.2–6 With efficient synthetic methodologies developed, these unnatural products have reached an exciting stage as they are now being investigated for applications as diverse as molecular machines/switches,7 molecular logic gates and information ratchets,8,9 memory devices,10 drug delivery,11 actuator,12 Slide-ring gels13 and switchable surfaces,14 to name a few. Interlocking architectures are also anticipated to impart unusual properties to bulk polymer materials. For example, polymeric catenanes are predicted to exhibit large loss moduli, low activation energies for viscous flow and rapid stress relaxation.15,16

Experimentally, however, the properties mechanically interlocked architectures impart to bulk materials are not completely understood, primarily due to difficulties in their synthesis. Motivated by finding ways to access polymeric catenanes we report herein studies aimed at accessing model

[3]-catenanes in high “crude” yield by optimizing the structure of the starting components.

† This chapter is adapted from: R. J. Wojtecki, Q. Wu, J. C. Johnson, D. G. Ray, L. T. J. Korley and S. J.

Rowan, Chem. Sci., 2013, 4, 4440.

Figure 2.1. Synthesis strategies of [3]rotaxane. Synthesis requires 2-bond formation from (a) [2]pseudorotaxane or (b) doubly-threaded [3]pseudorotaxane; and (c) 1-bond formation from [3]pseudorotaxanes.

[3]Catenanes have been synthesized by a variety of synthetic strategies and methodologies.

For instance, several donor-acceptor templated [3]catenanes have been reported that are synthesized from two components and require two covalent bond forming reactions to yield the product in ca. 50% yield.17,18 One successful route to mechanically interlocked structures is the use of metal-ligand coordination, which can be a highly effective way to produce mechanically interlocked molecules if the ligands are carefully selected to coordinate a metal ion in the appropriate orientation.19,20 In terms of metal-templated approaches, there are reports of the dimerization of metallo[2]pseudorotaxanes that require two covalent bond forming reactions to access [3]catenanes (Fig. 2.1a) with 58%21 and 70%22 yields. Metal-templated [3]catenane has also been produced via a double ring-closing approach from a doubly-threaded [3]pseudorotaxane, also requiring two covalent bond forming reactions with a 71% yield (Fig. 2.1b).23 Interested in seeing if the formation of a [3]catenate could be maximized, the present research focuses on the ring closing of metallo[3]pseudorotaxanes which only requires the formation of one bond to access the

[3]catenane (Fig. 2.1c). Interestingly, examples of [3]catenanes synthesized via a single ring closure step are rare. One elegant example was reported by the Leigh group, which utilized

hydrogen bond templating and a single ring closing reaction to access a [3]catenane in 21% yield; the [2]catenane was also a product of this reaction.24 To date we are not aware of a system that utilizes metal templating to produce [3]catenanes via 1-bond forming ring closing reaction.

A variety of metal-ligand combinations have been used to access topologically interlocked molecules. For example, Sauvage, who pioneered this approach, utilized bidentate (bipyridine, phenanthroline) and tridentate (terpyridine) ligands combined with transition metal ions that form tetrahedral (e.g. copper(I)) or octahedral (e.g. ruthenium(II)) complexes to access rotaxanes,25,26 catenanes,27,28 and molecular knots.29 These ligands work well and have been employed by a number of other groups, including Stoddart,30 Swager,31,32 Mayer,23 Advincula33,34 and

Nakazono,35 to access a wide range of complex interlocked species. Leigh and coworkers have not only utilized a range of different metal-binding ligands (including pyridine36 and bipyridine37 as well as imine containing tridentate ligands38) to access metal-templated catenanes and rotaxanes, but also pioneered the use of templating metal ions that function concurrently as a catalyst to promote the formation of a mechanical bond, a process termed active-metal templating.19,37

We have previously reported synthetic routes which allow facile access to multigram quantities of the tridentate ligand, 2,6-bis(N-alkyl-benzimidazolyl)pyridine (Bip) (Figure 2.2a) and its derivatives.39 Bip is a V-shaped ligand that binds with transition metal ions, e.g. Zn2+ or

Fe2+, forming an octahedral complex with two the Bips bound perpendicular to one another (as shown in the crystal structure, Figur 2.2b).40 This ligand and its metal:ligand induced self- assembly processes have subsequently been utilized to produce a range of structurally dynamic polymers41 that display a range of properties such as photo-healing,42 shape memory,43 sensing,44 facile processability and enhanced mechanical stability,45 interfacial assembly46 and chemo- 43

responsive liquid crystalline polymers.47 One advantage of the synthetic routes developed to access the Bip derivatives is the ability to easily control the size of their R1 substituents (Figure

2.2a). This not only imparts the potential to tailor the solubility of the resulting compounds but also offers the intriguing possibility of being able to influence translational movements within an interlocked compound by simply altering the size of the substituent.48 To date there have not been any reports that have utilized the Bip ligands in the metal-templating approach to catenanes.

However, it is worth noting that Piguet and coworkers utilized the Bip ligand to form a supramolecular interlocked catenane, where metal-Bip coordination was utilized as the ring closing step.49 Thus one goal of this present work is to investigate the potential of Bip derivatives, in combination with metal templating, to yield [3]catenanes, and as such expand the repertoire of ligands that can be utilized to access mechanically interlocked compounds.

Figure 2.2 . Binding of Bip and transition metal. (a) The V-shaped Bip ligands bind transition metal ions (e.g. Zn2+ or Fe2+) at 2:1 stoichiometry, and (b) the octahedral crystal structure of Bip/Zn2+ complex (hydrogen, counter-ion and solvent are not shown). The two Bips are colored differently to clearly observe their relationship.40

2. The Effect of the Linker Group on the Thread

In the one-bond formation strategy the threading molecule is the key component in the synthesis of [3]catenane as it determines the nature of the ring-closing reaction. Wojtecki50 pioneered the designing of the thread molecule for the assembly and synthesis of [3]catenate, which can be quantitatively converted to [3]catenane by demetalation (Figure 2.3a). Ring-closing olefin metathesis reaction was selected, which is proven to be an effective approach to access catenanes by a variety of templating strategies from metallosupramolecular templates.27,51 As shown in Figure 2.3b, the thread molecule 1 or 2 contains (i) two Bips for the assembly of metallo[3]pseudorotaxane precursor, (ii) two double-bonds for the ring-closing and (iii) a linker group that controls the conformation of the molecule. In comparison, the macrocycle 3 or 4 (Figure

2.3c) contains only one Bip. Upon adding two equivalents of transition metal salt, e.g. Zn(ClO4)2 or Fe(ClO4)2, into the 1:2 mixture of thread/macrocycle, the component self-assemble into the desired [3]pseudorotaxane structure. In the absence of coordinating solvent or counter-ion, the metal-ligand interaction drives all Bips and metal ions to assemble into supramolecular structures with full 2-Bip/1-metal binding (in order to maximize the enthalpy gain) and smallest assembly size (to minimize the entropic penalty). Under precise titration conditions, the [3]pseudorotaxane structure shown in Scheme 2.2a is shown to be the quantitative product as predicted.50 The cyclic structure of the macrocycle prevents its Bip from forming any 2:1 Bip/metal complexes with itself; therefore the Bip of macrocycle has can only form the 2:1 metal complex with a Bip from the thread and a metal ion.

Figure 2.3 Synthesis of [3]catenane via component design. (a) Synthesis of Bip-containing [3]catenates via one-bond formation, which after demetalation yields a [3]catenane; (b) the thread molecules which consists of two Bips, two reactive double-bonds and a linker group of either biphenyl or xanthene; and (c) the macrocycle containing one Bip with either n-hexyl or ethyl side-chain.

After quantitatively assembly of the components into the [3]pseudorotaxane, Wojtecki studied two different linker groups, biphenyl and xanthene (Figure 2.2), in order to understand their impact on the conversion toward [3]catenate. DOSY studies on their corresponding

2+ 2+ [3]pseudorotaxanes (1·(3·Zn )2 and 2·(3·Zn )2) show a notable difference in the diffusion coefficients (and hydrodynamic sizes):5.68×10−6 cm2/s (0.93 nm) and 7.31×10−6 cm2/s (0.72 nm), respectively (Table 2.1). The para-substituted biphenyl linker with free rotation about its central

2+ single bond allows the [3]pseudorotaxane 1·(3·Zn )2 to adopt an extended “S-like” conformation 46

(as demonstrated by molecular modeling) with a larger hydrodynamic volume and therefore lower diffusion coefficient. On the other hand, the “kinked” xanthene moiety means the

2+ [3]pseudorotaxane 2·(3·Zn )2 prefers to form a more “C-like” compact conformation with a smaller hydrodynamic volume and therefore higher diffusion coefficient. It is proposed that this

2+ difference in conformation results a low yield of 5·Zn 2 (ca. 63% based on DOSY, not isolated,

2+ caused by the more open S-like conformation), while the C-like conformation of 2·(3·Zn )2

2+ facilitates the ring-closing reaction resulting in a higher yield of 6·Zn 2 (ca. 83% based on DOSY and 38% isolation yield).

Figure 2.4 Comparison of linker groups (a)The thread with the biphenyl linker adopts a “S-like” conformation while (b) the one with the xanthene 2+ linker prefers to adopt a “C-like” conformation. The assembled [3]pseudorotaxanes (1·(3·Zn )2 or 2+ 2·(3·Zn )2) thus have different conformations which impacts the yield of the appropriate [3]catenate 2+ 2+ (5·Zn 2or 6·Zn 2, respectively).

Table 2.1 Diffusion coefficient (D) and hydrodynamic radius (Rh) study.

6 2 Sample D × 10 (cm /s) Rh (nm)

2+ 1·(3·Zn )2 5.68 0.93

2+ 2·(3·Zn )2 7.31 0.72

3. The Effect of Side Chain Bulkiness of the Macrocycle

The molecular design of the ‘C’-thread molecule was a big step toward the efficient synthesis of [3]catenane, but the goal of quantitative conversation from the [3]pseudorotaxane to a interlocked [3]catenate via one-bond formation had not yet been achieved. One hypothesis for this lack of quantitative conversion was the bulky nature of the hexyl group on the macrocycle 3.

It is possible that the repulsion of bulky hexyl groups expands/destabilizes the [3]pseudorotaxane

2+ 2·(3·Zn )2 reducing the effectiveness of the subsequence ring-closing reaction. Thus my focus in this research project was to further improve the yield of [3]catenate by optimizing the size of the side-chain bulkiness, as summarized in Figure 2.5. Thus a study was undertaken to investigate the synthesis of the [3]catenates using the same thread molecule 2, with either n-hexyl macrocycle 3 or the less bulky ethyl macrocycle 4. Both these threads will adopt the more “C-like” conformation, and will allow us to examine if the bulkiness of the N-alkyl group effects the subsequent ring- closing reaction.

Figure 2.5. The effect of macrocycle side-chain bulkiness. 2+ (a) Bulky n-hexyl side-chains inhibits the ring-closing reaction of 2·(3·Zn )2, which may disassemble into [2]pseudorotaxane (which can yield the [2]catenate) and the metallated macrocycle; (b) the 2+ [3]pseudorotaxane 2·(4·Zn )2with its less-bulky ethyl chains should have a reduced steric effect on the 2+ conversion to 8·Zn 2. 48

Thus the ethyl macrocycle 4 was synthesized to replace the hexyl 3 for all the studies.

2+ The synthesis of 4 is shown in Figure 2.6. The [3]pseudorotaxane 2·(4·Zn )2 was prepared by mixing 50 mg (28.28 mmol) of 2 (dissolved in 1 mL CHCl3), 68.6 mg (56.56 mmol, 2 eq.) of 4

(dissolved in CHCl3) and 21.1 mg (56.56mmol, 2 eq.) Zn(ClO4)2·6H2O (dissolved in MeCN) followed by removing the sovend under vacuum.

Figure 2.6 The synthesis of macrocycle 4.

To study the conversion of the [3]pseudorotaxane into the [3]catenate 1H-DOSY is a valid tool as long as the relative NMR relaxation weighting of the components is similar (or corrected).52 Figure 2.7 shows that the “C-like” thread 2 converted mainly (ca. 83%) to

2+ [3]catenate 6·Zn 2, however, a significant amount of metallated macrocycle and the byproduct

2+ [2]catenate 7·Zn are also observed from the DOSY with roughly 1:1 amount indicated by similar peak areas. Their diffusion coefficient follows Dmacrocycle > D[2]catenate > D[3]catenate which is reversely correlated to their hydrodynamic volume. Shifting to the non-bulky ethyl side-chain in the macrocycle 4 results in what appears to be almost exclusive formation (ca. 97%) of the

2+ [3]catenane 8·Zn 2 as determined by diffusion NMR.

To try and understand a reason for the dramtic improvement in yield Table 2.2 compares the diffusion coefficients and hydrodynamic radii (Rh), as measured by diffusion NMR, of the

2+ 2+ 2+ [3]pseudorotaxanes 2·(3·Zn )2 and 2·(4·Zn )2 and their corresponding [3]catenates 6·Zn 2 and

2+ 8·Zn 2, respectively. It is interesting to note that macrocycle side-chain appears to influence the

2+ 2+ size of the assemblies (0.72 nm versus 0.65 nm, for 2·(3·Zn )2 and 2·(4·Zn )2, respectively), but

2+ 2+ not on the [3]catenates (0.70 nm versus 0.71 nm, for 6·Zn 2 and 8·Zn 2, respectively). The Rh

2+ 2+ of ethyl [3]pseudorotaxane 2·(4·Zn )2 is slightly smaller than ethyl [3]catenate 8·Zn 2, indicating the formation of a very compact assembly that presumably favors the ring-closing reaction toward catenanes (Figure 2.3a). The hexyl side-chain derivative [3]pseudorotaxane

2+ 2·(3·Zn )2 forms a larger assembly, consistent with steric repulsion of its larger hexyl side chains.

2+ In order to ring close to the [3]catenate 6·Zn 2 it has to contract slightly in size, which will further enhance the steric hindrance of the side chain. In addition, the supramolecular structure of

2+ 2·(3·Zn )2 is under equilibrium with [2]pseudorotaxane and free macrocycles (see Figure 2.5a), although the binding constant between Bip and Zn2+ is high (ca. 106 M−2).53 The bulky hexyl group may also push the balance toward the disassembled structures that results in the formation of

[2]catenate byproducts. It is important to note the diffusion NMR data of the crude reaction does

2+ suggest that the [3]catenate 2·(3·Zn )2 is still the major product so the hexyl side chains do not inhibit formation of the [3]catenate but they do hinder it enough to allow, energetically the 50

formation of side products. As such, this highlights the need for careful molecular design in order to access high yields of these structures.

Figure 2.7. The 1H-DOSY (aromatic region) of crude [3]catenates with n-hexyl or ethyl side chains on macrocycle. 2+ (a) Ring closure of the hexyl [3]pseudorotaxane 2·(3·Zn )2 shows three peaks on the Y-axis corresponding to the metallated macrocycle 3·Zn2+, the [2]catenate byproduct 7·Zn2+ and the target [3]catenate 6·Zn2+; 2+ 2+ while (b) the ethyl [3]pseudorotaxane 2·(4·Zn )2 only shows one peak corresponding to [3]catenate 8·Zn , suggesting a near-quantitative conversion toward the product.

Table 2.2 Diffusion coefficients and hydrodynamic radii of the [3]pseudorotaxanes and [3]catenates with n-hexyl or ethyl side-chains on macrocycles measured by 1H-DOSY.

6 2 Sample Side-chain D ×10 (cm /s) Rh (nm)

2+ n-Hexyl, 2·(3·Zn )2 7.31 0.72 [3]Pseudorotaxane 2+ Ethyl 2·(4·Zn )2 8.09 0.65

n-Hexyl, 6·Zn2+ 7.59 0.70 [3]Catenate Ethyl, 8·Zn2+ 7.42 0.71

4. Isolation and characterization of [3]catenane 8.

The [3]catenane 8 was isolated by demetallation of [3]catenate 8·Zn2+ and chromatography

(with silica gel) to remove the Grubbs catalyst and any minor amount of byproducts. In the end the isolated yield of pure [3]catenane 8 was only 55%; the major loss of product yield comes from the silica gel absorption. The [3]catenane was fully characterized by various 1D and 2D NMRs and MALDI-TOF/TOF mass spectrometry techniques. The [3]catenane structure is confirmed by these characterizations which also provides valuable information for the future characterization of

Bip-containing poly[n]catenanes.

The 1H-NMR of the [3]catenane 8 was compared with the macrocycle 4 and 8 the ring- closed thread 9 (the aromatic region is shown in Figure 2.8). All aromatic peaks show dramatic chemical shift after incorporated into interlocking catenane structure. The peaks highlighted in yellow (e.g. protons K and k in the [3]catenane, between 8.1 and 8.2 ppm) are very distinctive, and can be used as an indicator for the formation of Bip-containing catenanes. To confirm the interlocking structure, 1H-NOESY of [3]catenane 8 was also carried out at reduced temperature to slow down motion of the rings. NOESY cross-peaks between two rings are observed, and are

highlighted in Figure 2.9. Note that the NOE signal intensity is proportional to 1/r6 where r is the distance, the cross peak in NOESY is hardly observable when the distance of two protons are larger than 5 angstroms. Therefore, simply mixing 4 and 9 in solution does not show any cross peaks.

Figure 2.8. 1H-NMR comparison (aromatic region) of [3]catenane 8, with ring-closed thread 9 and macrocycle 4.

Some unique peaks of [3]catenane 8 are highlighted in yellow. Measured in CD2Cl2 ,600 MHz. 53

Figure 2.9. 1H-NOESY of 8 showing selected cross-component NOE correlations.

Measured in CD2Cl2 at −17 °C with 0.05 sec mixing time.

The MALDI-TOF of [3]catenane (Figure 2.10) indicates the molecular ion peak at 4117 for [M]+Na+, the mass of which as well as the isotope distribution pattern perfectly matches the calculated value. The tandem mass spectrometry (MALDI-TOF/TOF, Figure 2.11) is also a solid evidence for the interlocked structure. A clean fragmentation pattern of a [3]catenane is observed, which shows solely the corresponding [2]catenane and both rings. The pattern corresponds to systematic loss of the peripheral macrocycles from the parent catenane without the formation of any species of intermediary mass.

Figure 2.10 MALDI-TOF of 2-10. Inset shows the isotope pattern.

Figure 2.11. TOF/TOF tandem mass spectrometry of 2-10 (m/z = 4117). Inset shows the mechanism of using tandem mass spectrometry technique for the analysis of catenanes. The selected parent molecule will undergo random bond-breakage fragmentations. For a [3]catenane parent molecule, the bond-breakage may happens on the central ring (a) or one of the two outer rings (b). Therefore the fragments will be [2]catenane and individual components.

5. Conclusion

We have described the preparation and optimization of a series of Bip-containing

[3]catenates that require a single ring closure reaction to yield the final interlocked molecule.

DOSY NMR was used to follow the self-assembly of the [3]metallopseudorotaxanes and their subsequent ring closing reaction to yield the [3]catenates. While previous work had shown that ring closing reaction is very sensitive to the linking moiety between the two Bip units in the thread, this work also showed that the N-alkyl substituents on macrocycle also play a role in this step.

Through judicious selection of the two components the crude yield of the [3]catenate could be optimized, resulting in predominantly one product. Such efficient formation of a [3]catenate opens

the door to larger interlocked architectures, an area that is pursued in future chapters of this thesis.

Furthermore, it was also demonstrated that the metal ion template could be removed to yield the

[3]catenanes and these compounds were characterized by a variety of NMR and mass spectroscopy techniques (although further work is needed to improve the efficiency of the demetallation/purification procedure). The data from these initial [3]catenanes does suggests that the N-alkyl moiety may be used to impact the conformation of these interlocked species. Given that the synthetic procedures to access Bip derivatives can be easily modified to allow access to a variety of N-substituted we are interested in understanding further how the different N-substituents can be used to impact the rotational and translational motions of these interlocked compounds.

This strategy may also open a door toward the synthesis of main-chain poly[n]catenanes.

6. Experimental

6.1 Materials and Methods

All solvents, potassium carbonate, sodium bicarbonate, and sodium hydroxide were purchased from Fisher Scientific. Deuterated solvents were purchased from Norell, Inc. for NMR studies. Hoveyda-Grubbs Generation II catalyst and all other chemicals and reagents were purchased from Sigma Aldrich. Solvent was distilled from suitable drying agents under argon.

All NMR spectra were recorded on a Varian Inova 600 MHz NMR spectrometer (150.8

MHz for 13C-NMR) with Highland Techmology Model L700 pulse gradient driver. Mass spectrometry was measured on Bruker Autoflex III MALDI-TOF/TOF mass spectrometer using dithranol as matrix with sodium trifluoroacetate as doping salt.

All DOSY experiments were performed on dilute solution in 5mm NMR tubes at constant temperature of 25 °C with Bipolar pulse-pair simulated-echo (Dbppste) pulse sequence with 50 microseconds diffusion delay, 10 seconds of relaxation delay and 3.5 seconds of acquisition time.

The gradient strength was calibrated using the internal TMS as standard (1% V/V). The NMR spectrum was processed in MestReNova using Bayesian DOSY transformation algorithm. The

The MALDI-TOF/TOF was performed without collision-induced dissociation.

6.2 Synthesis of Precursors and Components

6.2.1 Synthesis of Ethyl Macrocycle 2-4

Step 1 of 3:

Hexaethylene glycol mono-p-toluenesulfonate was first prepared according to a literature procedure.54 0.091 mol (39.72g, 3eq.) of the monotosylate together with 0.030mol (5.04g) of 3,5- dihydroxybenzoate and 0.087mol (12.00g, 3 eq.) of K2CO3 are added to a 250 mL round bottom flask equipped with a stir-bar, a condenser, and was then flushed with argon. 100 mL dry DMF is then added by cannula and the mixture is heated in an oil bath to 75 °C and allowed to react for 24 hrs. After reaction the DMF is removed under reduced pressure and resulting mixture triturated with CHCl3 and then filtered. The filtrate was then collected and the solvent removed under reduced pressure to yield a yellow-brown oil. The oil was purified by column chromatography

(silica gel, ethyl acetate, 3% methanol) to yield the pure diol 10 in a 70% yield. 1H NMR (600

MHz, CDCl3) δ 7.17 (d, J =2.3 Hz, 2H), 6.68 (t, J = 2.3 Hz, 1H), 4.12 (t, J = 4.8 Hz, 4H), 3.87 (s,

3H), 3.84 (t, J = 4.8 Hz, 4H), 3.73 – 3.69 (m, 8H), 3.68 – 3.62 (m, 24H), 3.61 – 3.57 (m, 8H). 13C

NMR (151 MHz, CDCl3) δ 166.91, 159.94, 132.06, 108.21, 107.08, 72.65, 70.95, 70.79, 70.73,

70.67, 70.64, 70.41, 70.31, 70.04, 69.73, 67.92, 61.77, 61.54, 52.39, 29.85. MALDI-MS: m/z

719.0 ([M]+Na+).

Step 2 of 3:

4.1 mmol (2.86 g) of 10, 10.25 mmol (1.95 g, 2.5 eq.) of p-toluenesulfonyl chloride (TsCl) and a catalytic amount of 4-Dimethylaminopyridine (DMAP) (0.205 mmol, 0.025g, 0.05 mol. %) was added to a 50mL round bottom flask (equipped with a stir bar) followed by 20 mL of dichloromethane (DCM) and 4.5 mL of triethylamine (TEA). The mixture was flushed with argon, submerged in an ice bath and stirred for 18hrs. The reaction was then allowed to warm to room temperature at which point the solvent was removed under reduced pressure. The resulting material was triturated with ethyl acetate and filtered. The filtrate was collected and the solvent removed under reduced pressure yielding yellow-brown oil. The resulting material was purified by column chromatography (silica gel, ethyl acetate, 3% methanol) to yield the 11 in a 50% yield. 1H-NMR

(600 MHz, CDCl3) δ 7.78 (d, J = 8.3 Hz, 4H), 7.32 (d, J = 8.0 Hz, 4H), 7.17 (d, J = 2.4 Hz, 2H),

6.67 (t, J = 2.4 Hz, 1H), 4.16 – 4.10 (m, 8H), 3.88 (s, 3H), 3.84 (t, J= 4.8 Hz, 4H), 3.72 – 3.69 (m,

4H), 3.69 – 3.58 (m, 24H), 3.56 (s, 8H), 2.43 (s, 6H). 13C NMR (151 MHz, CDCl3) δ 166.91, 59

159.96, 144.97, 133.22, 132.09, 130.02, 128.15, 108.21, 107.06, 71.02, 70.92, 70.80, 70.75, 70.70,

69.77, 69.45, 68.86, 67.95, 52.41, 21.83. MALDI-MS: m/z 1027 ([M]+Na+).

Step 3 of 3:

A 100mL dropping funnel containing 11 (1.000g, 0.995 mmol) and Ethyl Bip (0.549 g,

0.995 mmol, prepared by literature procedure39) in 65 mL of DMF was fitted to a 250 mL round bottom flask containing Cs2CO3 (3.000g, 9.2 mmol) and stir bar. The reaction vessel was flushed with argon before anhydrous DMF (100 mL) was added by cannula. The reaction was submerged in an oil bath and heated to 70°C while rapidly stirring. The dissolved mixture of components in the dropping funnel was added dropwise to this suspension over the course of 24 hrs. The reaction was stirred at 70°C for totally 72 hrs, after which the solvent was removed and the crude was triturated with CHCl3 and filtered. The filtrate was collect and solvent was removed under vacuum.

The resulting oil was purified by column chromatography (silica gel, chloroform/methanol

1 gradient) to yield 2-4 as light yellow oil in 24% yields. H-NMR (600 MHz, CDCl3) δ 8.29 (d, J =

7.8 Hz, 2H, k), 8.03 (t, J = 7.8 Hz, 1H, l), 8.00 (s, 2H, h), 7.59 (d, J = 8.7 Hz, 4H, g), 7.56 (dd, J =

8.4, 1.5 Hz, 2H, i), 7.46 (d, J = 8.4 Hz, 2H, j), 7.08 (m, 6H, b, f), 6.48 (t, J = 2.3 Hz, 1H, c), 4.72

(q, J = 7.2 Hz, 4H, s), 4.28 (m, 4H, e), 3.94- 3.47 (m, 50H, d, a), 1.37 (t, J = 7.2 Hz, 6H, t). 13C

NMR (151 MHz, CDCl3) δ 166.79, 159.80, 158.43, 150.56, 150.12, 143.67, 138.18, 136.25,

135.21, 134.59, 131.96, 128.40, 125.50, 123.26, 118.19, 115.85, 110.51, 107.90, 106.89, 71.19,

70.93, 70.80, 70.70, 70.63, 70.58, 70.54, 70.47, 70.19, 69.57, 68.19, 67.56, 52.33, 40.14, 15.67.

MALDI-TOF: m/z 1235 ([M]+Na+).

6.2.2 Synthesis of thread 2-2.

Step 1 of 3: synthesis of 3,6-bis(bromomethyl)-9,9-dimethyl-9H-xanthene (2-14)

3,6,9,9-tetramethyl-9H-xanthene (4.16 g, 0.0175 mol, prepared according to the literature procedure55) and N-bromosuccinimide (NBS) (6.20 g, 0.0348 mol, 2 eq.) were added to a round bottom flask along with a catalytic amount (20 mg, 0.122 mmol, 0.6 eq.) of azobisisobutyronitrile

(AIBN). The round bottom was then equipped with a condenser and flushed with argon for a period of three minutes. 100 mL of anhydrous benzene was then cannulated into the round-bottom bringing the concentration of the 3,6,9,9-tetramethyl-9H-xanthene to 175 mM. The round-bottom flask was immersed in an oil bath and heated to 75 °C for 18 hrs, after which a precipitate was noted. Upon cooling to room temperature an increased amount of precipitate was observed and the reaction filtered. The filtrate was collected and the solvent removed under reduced pressure to yield an oil. Addition of hexanes to this oil leads to the further precipitation of a white solid. This was filtered again and filtrate collected and the solvent again removed under reduced pressure to yield an off white powder. This powder was recrystallized from ethanol to yield the 12 in a 38%

1 yield. Melting Point: 138.8-139.8°C. H NMR (CDCl3) δ 7.37 (d, J = 8.0 Hz, 2H), 7.12 (dd, J = 61

13 8.0, 1.8 Hz, 2H), 7.09 (d, J = 1.8 Hz, 2H), 4.48 (s, 4H), 1.61 (s, 6H). C NMR (CDCl3) δ 150.42,

137.41, 130.31, 126.87, 124.05, 117.08, 34.18, 32.98, 32.53. MALDI-MS: m/z 396.0 ([M]+H+).

Step 2 of 3.:

Hexyl Bip (1.547g, 2.33 mmol, prepared by literature procedure39), 4-hexen-1-tosylate

(0.592 g, 2.33 mmol), and K2CO3 (1.3 g, 9.11mmol) were added to a 25mL round bottom flask.

Anhydrous N,N’-Dimethylformamide (10 mL) was then added and the reaction flushed with argon.

The round bottom flask was then immersed in an oil bath at 75°C and the reaction stirred for 18hrs.

DMF was then removed under vacuum and the resulting dry material was triturated with chloroform. The resulting heterogeneous mixture was filtered and the organic layer collected and the solvent removed. The remaining solid was purified using column chromatography (silica gel, chloroform, 2% methanol) to yield 15 in 49% as an off white solid. Melting Point: 91.2-94.1°C.

1 H NMR (600 MHz, CDCl3): δH 0.64 (m, 6H), 1.05 (m, 8H), 1.10 (m, 4H), 1.66 (d, 3H, J= 4.8

Hz), 1.74 (m, 2H, J= 7.2 Hz), 1.87 (m, 2H, 6.6 Hz, J= 7.2 Hz), 2.19 (q, 2H, 6.6 Hz, J= 7.2 Hz),

4.01 (t, 2H, 6.6 Hz), 4.71 (t, 4H, J=7.2 Hz), 5.48 (m, 2H), 6.97 (d, 2H, AB system, J= 8.4 Hz),

7.00 (d, 2H, AB system, J= 9.0 Hz), 7.58-7.41 (m, 8H), 7.66 (bs, 1H), 7.97 (s, 1H), 8.01 (s, 1H),

8.06 (t, 1H, J= 7.8 Hz), 8.32 (dd, 2H, J= 7.8 Hz, J= 1.8 H). δ C (151 MHz, CDCl3) 158.67, 156.84,

150.73, 150.58, 149.86, 149.78, 143.16, 142.98, 138.62, 137.03, 136.66, 135.42, 135.24, 134.16,

133.30, 130.46, 128.70, 128.54, 125.90, 125.82, 123.53, 123.50, 118.02, 117.86, 116.43, 115.10,

110.70, 110.63, 67.59, 45.27, 31.34, 30.23, 29.30, 29.12, 26.52, 22.54, 18.12, 13.93. MS (MALDI, positive): m/z 746 ([M]+H+), 768 ([M]+Na+).

Step 3 of 3:

2-15 (0.700 g, 0.939 mmol), 2-14 (0.185 g, 0.467 mmol) and K2CO3 (0.500 g, 3.62 mmol) were added to a 8mL conical vial. Anhydrous DMF (3 mL) was then added and the reaction flushed with argon. The vial was then immersed in an oil bath at 70°C and stirred for 18hrs. DMF was then removed with vacuum and the solids were stirred with chloroform and filtered. The filtrate was collected and the solvent removed under reduced pressure. The resulting material was purified using column chromatography (triethylamine pretreated silica, eluted with chloroform/hexane gradient) to yield the product 2 in 50% yield. Melting Point: 193.1-195.2°C. 1H NMR (600 MHz,

CDCl3): δ 0.662 (m, 12H, X), 1.079 (m, 24H, U, V, W), 1.662 (m, 12H, A, R), 1.752 (m, 8H, T),

1.880 (m, 4H, N), 2.188 (m, 4H, O), 4.025 (t, 4H, J= 6.6 Hz, M), 4.730 (t, 8H, J= 7.2 Hz, S), 5.192

(s, 4H, E), 5.491 (m, 4H, P, Q), 7.014 (d, 4H, J= 8.7 Hz, F’), 7.129 (d, 4H, J= 8.7 Hz, F), 7.592

(m, 24H, G, G’, I, I’, J, J’), 8.034 (s, 4H, H, H’), 8.073 (t, 2H, J= 7.9 Hz, L), 8.341 (d, 4H, J= 7.8

Hz, K). 13C NMR (150.8 MHz, CDCl3): δ 143.7, 140.7, 138.4, 136.44, 136.41, 136.3, 135.7,

135.68, 134.8, 134.3, 130.544, 128.7, 128.6, 128.3, 127.6, 126.0, 125.7, 128.3, 127.6, 126.0, 123.7,

118.32, 118.27, 115.5, 115.1, 110.7, 110.71, 70.1, 68.8 67.6, 67.5, 45.3, 31.4, 30.3, 29.4, 29.2,

29.18, 29.1, 26.6, 22.7, 18.3, 14.1. MALDI: m/z 1727 ([M]+H+), 1749 ([M]+Na+).

6.2.3 Synthesis of ring-closed thread 2-11

9 mL of DCM was added to the thread 2-

2 (0.103 g, 0.0599 mmol) in a 20mL round bottom flask attached with condensor. Argon was bubbled through this solution for 30 minutes and the reaction which was kept under an argon atmosphere. 30 mol% Hoveyda-Grubbs Generation II metathesis catalyst (0.011g, 0.0180 mmol) dissolved in 1 mL of the de-oxygenated DCM was injected, bringing the final concentration to 6 mM. The reaction was refluxed for 24 hours followed by deactivation of catalyst with ethyl vinyl ether. The solvent was removed and the product purified by triethylamine pretreated silica gel

1 column (hexane/CHCl3 gradient) to obtain the product in 35% yield. H NMR (600 MHz, CD2Cl2):

δ 8.29 (m, 4H, K), 7.98 (m, 4H, L), 7.92 (s, 2H, H), 7.88 (s, 2H, H’), 7.56-7.40 (m, 18H, I, I’, J,

J’, B, G, G’), 7.33 (m, 2H, B), 7.14 (d, J= 7.8 Hz, 2H, D), 7.08 (s, 2H, C), 6.98 (d, J = 7.0 Hz, 4H,

F), 6.92 (d, 3H, J = 7.0Hz, F’), 6.89 (d, J=7.0Hz, 1H, F’), 5.53 (m, 2H, P’), 5.26 (m, 4H, E), 4.68

(t, J = 7.1Hz, 4H, S), 4.62 (m, 4H, S), 3.98 (t, J=6.3 Hz, 3H, M), 3.86 (t, J=6.1 Hz, 1H, M), 2.31

(m, 8H, U), 1.88 (m, 3H, O), 1.82 (m, 2H, O), 1.62 (m, 6H, A), 1.27 (m, 4H, N), 0.97 (m, 24H, W,

13 V, U), 0.58 (m, 12H, X). C NMR (150.8 MHz, CD2Cl2): δ (151 MHz, cd2cl2) 159.00, 157.89,

151.20, 151.08, 151.04, 150.68, 144.03, 138.48, 138.13, 136.28, 136.19, 136.12, 135.00, 134.43,

131.06, 130.43, 129.91, 128.77, 128.72, 127.18, 125.87, 123.34, 121.83, 118.26, 118.15, 116.19,

115.39, 115.28, 115.01, 111.12, 111.08, 111.01, 78.12, 77.34, 69.47, 67.30, 62.61, 56.24, 45.53, 64

45.48, 34.50, 34.40, 32.74, 32.51, 32.26, 31.73, 31.71, 30.57, 30.53, 30.27, 30.09, 29.94, 29.78,

29.68, 29.54, 29.20, 29.04, 28.47, 28.41, 27.18, 26.85, 26.80, 26.70, 25.39, 24.01, 23.27, 23.16,

22.93, 22.91, 15.62, 14.46, 14.35, 14.07. (MALDI, positive): m/z 1693 ([M]+Na+).

Reference:

1 J. F. Stoddart, Chem. Soc. Rev., 2009, 38, 1802.

2 M. Beyler, V. Heitz and J.-P. Sauvage, Chem. Commun., 2008, 5396.

3 T. J. Hubin and D. H. Busch, Coordination Chemistry Reviews, 2000, 200-202, 5–52.

4 F. Aricó, J. D. Badjic, S. J. Cantrill, A. H. Flood, K. C.-F. Leung, Y. Liu and J. F.

Stoddart, in Templates in Chemistry II, eds. C. A. Schalley, F. Vögtle and K. H. Dötz,

Springer Berlin Heidelberg, Berlin, Heidelberg, 2005, vol. 249, pp. 203–259.

5 K. Kim, Chem. Soc. Rev., 2002, 31, 96–107.

6 I. Aprahamian, O. Š. Miljanic, W. R. Dichtel, K. Isoda, T. Yasuda, T. Kato and J. F.

Stoddart, Bulletin of the Chemical Society of Japan, 2007, 80, 1856–1869.

7 A. R. Pease, J. O. Jeppesen, J. F. Stoddart, Y. Luo, C. P. Collier and J. R. Heath, Accounts

of Chemical Research, 2001, 34, 433–444.

8 C. P. Collier, E. W. Wong, M. Belohradský, F. M. Raymo, J. F. Stoddart, P. J. Kuekes, R.

S. Williams and J. R. Heath, Science, 1999, 285, 391–394.

9 V. Serreli, C.-F. Lee, E. R. Kay and D. A. Leigh, Nature, 2007, 445, 523–527.

10 J. E. Green, J. Wook Choi, A. Boukai, Y. Bunimovich, E. Johnston-Halperin, E. DeIonno,

Y. Luo, B. a Sheriff, K. Xu, Y. Shik Shin, H.-R. Tseng, J. F. Stoddart and J. R. Heath,

Nature, 2007, 445, 414–417.

11 K. K. Cotí, M. E. Belowich, M. Liong, M. W. Ambrogio, Y. a Lau, H. a Khatib, J. I. Zink,

N. M. Khashab and J. F. Stoddart, Nanoscale, 2009, 1, 16.

12 B. K. Juluri, A. S. Kumar, Y. Liu, T. Ye, Y. Yang, A. H. Flood, L. Fang, J. F. Stoddart, P.

S. Weiss and T. J. Huang, ACS Nano, 2009, 3, 291–300.

13 K. Ito, Polym. J., 2007, 39, 489–499.

14 J. Berná, D. a Leigh, M. Lubomska, S. M. Mendoza, E. M. Pérez, P. Rudolf, G. Teobaldi

and F. Zerbetto, Nature Materials, 2005, 4, 704–710.

15 T. Pakula and K. Jeszka, Macromolecules, 1999, 32, 6821–6830.

16 J.-L. Weidmann, J. Kern, J. Sauvage, D. Muscat, S. Mullins, W. Köhler, C. Rosenauer, H.

J. Räder, K. Martin and Y. Geerts, Chem. Eur. J., 1999, 5, 1841–1851.

17 F. B. L. Cougnon, N. A. Jenkins, G. D. Pantoş and J. K. M. Sanders, Angew. Chemie Int.

Ed., 2012, 51, 1443–1447.

18 J. M. Spruell, A. Coskun, D. C. Friedman, R. S. Forgan, A. A. Sarjeant, A. Trabolsi, A. C.

Fahrenbach, G. Barin, W. F. Paxton, S. K. Dey, M. A. Olson, D. Benítez, E. Tkatchouk,

M. T. Colvin, R. Carmielli, S. T. Caldwell, G. M. Rosair, S. G. Hewage, F. Duclairoir, J.

L. Seymour, A. M. Z. Slawin, W. A. Goddard, M. R. Wasielewski, G. Cooke and J. F.

Stoddart, Nat. Chem., 2010, 2, 870–879.

19 J. D. Crowley, S. M. Goldup, A.-L. Lee, D. a Leigh and R. T. McBurney, Chem. Soc.

Rev., 2009, 38, 1530.

20 J. E. Beves, B. A. Blight, C. J. Campbell, D. A. Leigh and R. T. McBurney, Angew.

Chemie Int. Ed., 2011, 50, 9260–9327.

21 C. O. Dietrich-Buchecker, A. Khemiss and J. P. Sauvage, Journal of the Chemical

Society, Chemical Communications, 1986, 1376. 66

22 J. D. Megiatto and D. I. Schuster, Chem. Eur. J., 2009, 15, 5444–5448.

23 M. Gupta, S. Kang and M. F. Mayer, Tetrahedron Letters, 2008, 49, 2946–2950.

24 D. a Leigh, J. K. Y. Wong, F. Dehez and F. Zerbetto, Nature, 2003, 424, 174–179.

25 J.-C. Chambron, J.-P. Collin, V. Heitz, D. Jouvenot, J.-M. Kern, P. Mobian, D. Pomeranc

and J.-P. Sauvage, European Journal of Organic Chemistry, 2004, 2004, 1627–1638.

26 J. A. Faiz, V. Heitz and J.-P. Sauvage, Chem. Soc. Rev., 2009, 38, 422–442.

27 M. Weck, B. Mohr, J.-P. Sauvage and R. H. Grubbs, J. Org. Chem., 1999, 64, 5463–5471.

28 C. Hamann, J. Kern and J. Sauvage, Inorganic Chemistry, 2003, 42, 1877–1883.

29 C. O. Dietrich-Buchecker and J.-P. Sauvage, Angew. Chemie Int. Ed., 1989, 28, 189–192.

30 K. S. Chichak, S. J. Cantrill, A. R. Pease, S.-H. Chiu, G. W. V. Cave, J. L. Atwood and J.

F. Stoddart, Science, 2004, 304, 1308–1312.

31 M. J. MacLachlan, A. Rose and T. M. Swager, J. Am. Chem. Soc., 2001, 123, 9180–9181.

32 P. H. Kwan, M. J. MacLachlan and T. M. Swager, J. Am. Chem. Soc., 2004, 126, 8638–

8639.

33 P.-F. Cao, A. Bunha, J. Mangadlao, M. J. Felipe, K. I. Mongcopa and R. Advincula,

Chem. Commun., 2012, 48, 12094.

34 P.-F. Cao, J. Mangadlao and R. Advincula, Angew. Chemie Int. Ed., 2015, 54, 5127–5131.

35 S. Saito, E. Takahashi and K. Nakazono, Org. Lett., 2006, 8, 5133–5136.

36 V. Aucagne, K. D. Hänni, D. A. Leigh, P. J. Lusby and D. B. Walker, J. Am. Chem. Soc.,

2006, 128, 2186–2187.

37 V. Aucagne, J. Berná, J. D. Crowley, S. M. Goldup, K. D. Hänni, D. A. Leigh, P. J.

Lusby, V. E. Ronaldson, A. M. Z. Slawin, A. Viterisi and D. B. Walker, J. Am. Chem.

Soc., 2007, 129, 11950–11963. 67

38 D. A. Leigh, P. J. Lusby, S. J. Teat, A. J. Wilson and J. K. Y. Wong, Angew. Chemie Int.

Ed., 2001, 40, 1538–1543.

39 B. M. McKenzie, A. K. Miller, R. J. Wojtecki, J. C. Johnson, K. A. Burke, K. A. Tzeng,

P. T. Mather and S. J. Rowan, Tetrahedron, 2008, 64, 8488–8495.

40 X. Fan, J. Yuan, Y. Bai, J. Kong and H. Wu, Acta Crystallographica Section E Structure

Reports Online, 2012, 68, m1072–m1072.

41 R. J. Wojtecki, M. a Meador and S. J. Rowan, Nature Materials, 2011, 10, 14–27.

42 M. Burnworth, L. Tang, J. R. Kumpfer, A. J. Duncan, F. L. Beyer, G. L. Fiore, S. J.

Rowan and C. Weder, Nature, 2011, 472, 334–337.

43 J. R. Kumpfer and S. J. Rowan, J. Am. Chem. Soc., 2011, 133, 12866–12874.

44 M. Burnworth, S. J. Rowan and C. Weder, Chem. Eur. J., 2007, 13, 7828–7836.

45 M. Burnworth, D. Knapton, S. J. Rowan and C. Weder, Journal of Inorganic and

Organometallic Polymers and Materials, 2007, 17, 91–103.

46 J. R. Kumpfer and S. J. Rowan, ACS Macro Letters, 2012, 1, 882–887.

47 B. M. McKenzie, R. J. Wojtecki, K. A. Burke, C. Zhang, A. Jákli, P. T. Mather and S. J.

Rowan, Chem. Mater., 2011, 23, 3525–3533.

48 D. A. Leigh, A. Murphy, J. P. Smart, M. S. Deleuze and F. Zerbetto, J. Am. Chem. Soc.,

1998, 120, 6458–6467.

49 C. Piguet, G. Bernardinelli, A. F. Williams and B. Bocquet, Angew. Chemie Int. Ed.,

1995, 34, 582–584.

50 R. J. Wojtecki, Q. Wu, J. C. Johnson, D. G. Ray, L. T. J. Korley and S. J. Rowan, Chem.

Sci., 2013, 4, 4440.

51 D. A. Leigh, R. G. Pritchard and A. J. Stephens, Nat. Chem., 2014, 6, 978–982. 68

52 T. Zhao, H. W. Beckham and H. W. Gibson, Macromolecules, 2003, 36, 4833–4837.

53 S. J. Rowan and J. B. Beck, Faraday Discuss., 2005, 128, 43–53.

54 O. Kocian, K. W. Chiu, R. Demeure, B. Gallez, C. J. Jones and J. R. Thornback, Journal

of the Chemical Society, Perkin Transactions 1, 1994, 527.

55 A. J. Caruso and J. L. Lee, J. Org. Chem., 1997, 62, 1058–1063.

Chapter 3. Templated Synthesis toward Bip-containing

Polymers and Catenanes

1. Introduction

From crown ethers to various molecular rotaxanes, catenanes and knots, templated synthesis is known as a milestone of synthesizing molecules with unique or complex structures in high yield. In addition to its success in small molecule chemistry, templated synthesis is also a fast-growing field in polymer chemistry. One great example that polymer chemists are trying to mimic from the mother nature is the templated synthesis of DNA, RNA and proteins. As shown in

Figure 3.1, the synthesis of DNA, RNA and proteins are all based on using a template, which is a

DNA or a RNA molecule. The template synthesis provides accurate control on the monomer sequence and molecular weight of these biomacromolecules, which is fundamental for the growth, living and reproduction of all living creatures.

Figure 3.1 (a) Templated synthesis of DNA from another DNA as template and (b) templated synthesis of protein from DNA and RNA templates. Reproduced with permission from http://learn.genetics.utah.edu, © Genetic Science Learning Center, University of Utah.

Inspired by the protein synthesis, Lutz and coworkers reported the synthesis of sequence- controlled polymer from a polyrotaxane template, which has been discussed in Chapter 1 (Figure

1.13). Polyrotaxane template can also be used for the synthesis of a cyclodextrin molecular tube

(Figure 1.12c, Chapter 1).

Besides using a polymer as template to synthesize another polymer, the supramolecular assembly can also be used as template for the synthesis of polymers. Zimmerman and coworkers reported the synthesis of organic nanotube as shown in Figure 3.2.1 The porphyrin-cored dendrimer was first assembled in to a supramolecular cylinder structure, which acts as the template for the metathesis crosslinking of the terminal olefin shell. Finally, the core was removed, yielding the organic nanotubes.

In this thesis, we are trying to use a polymeric supramolecular structure to template the synthesis of poly[n]catenanes. For example, the assembly of monomer 14 upon the addition of

2+ 2+ 2+ Zn (or Fe ) yields the metallosupramolecular polymer (MSP) 14n+2·Zn 2n, which conceptually

can yield the polycatenane 15 after ring-closing, as shown in Figure 3.3. The advantage of this process is disengaging the polymerization process (the supramolecular assembly) and the covalent macrocyclization reaction, as the former requires a high concentration while the latter needs very low concentration. The advantage of using Bip-metal interaction is its strong binding constant that limits dissociation under the dilute cyclization reaction conditions, and also allows for the easy removal of metal ion.

Figure 3.3 Assembly of the metallosupramolecular polymer (MSP) and retro synthetic analysis of polycatenane 15.

2. Results and Discussion

2.1 The metallo-supramolecular polymer

2.1.1 The design of monomer

In Chapter 2, we concluded that the xanthene moiety is a better candidate than the biphenyl moiety for the linker group as it provides a more “C-like” conformation favorable for the ring- closing reaction. Therefore, the xanthene-based linker was chosen for this study with the biphenyl being used as a control. Chapter 2 also shows that for the two interlocked macrocycles, N-ethyl side-chains are better than n-hexyl as a consequence of reduced bulkiness; however, the monomer with ethyl side-chains show very limited solubility. Therefore, we chose n-butyl substituents as a

compromise between solubility and steric bulk. The targeted structures of monomer 14 and control monomer 15 are shown in Figure 3.4.

Figure 3.4 Monomer 14 with xanthene linker moiety (a), and 16 with biphenyl linker moiety (b).

2.1.2 The assembly of metallo-supramolecular polymer (MSP)

The assembly of the metallosupramolecular polymer (MSP) is performed by the titration of a

36.0 mM acetonitrile solution of Zn(NTf2)2 into a 12 mM chloroform solution of monomer 14, which is monitored by 1H-NMR (Figure 3.5). After the addition of metal ions, most of the ligand peaks are shifted, with the peaks corresponding to HA, HB and HC being effected the most upon coordination with the metal ion. The inset shows the zoomed in highlighted area, and reveals some several very small peaks next to the main peak, which could be related to the chain-end and/or the presence of macrocycles. However, to date no definitive assignment has been made.

After addition of 1 equivalent of Zn2+ ions (relative to 14) the solution becomes highly viscous and slightly yellowish in color, consistent with the formation of the Zn-MSP. Fe2+ has a higher

10 −2 2 2+ 6 −2 3 overall binding constant to Bip (>10 M ) than Zn ions (ca. 10 M ) Addition with 1 equivalent of Fe(NTf2)2 yields a viscous deep purple solution indicative of metal ion coordination. The NMR of this Fe-MSP (Figure 3.6) show even bigger peaks shifts upon Fe2+ 73

coordination relative to Zn2+ coordination. The NMR of the Fe2+ containing MSP was successfully measured as Fe2+ is in its low-spin state, which is consistent with the 2:1 Bip/Fe2+ complex reported in literature.4

1 2+ Figure 3.5 Full H-NMR showing the titration of Zn-MSP 14n+2·Zn 2n.

Inset shows the magnified region. Monomer 14 was dissolved in CDCl3 (12 mM) and Zn(NTf2)2 was dissolved in MeCN-d3 (36.0 mM). After titration, the solution is ca. 3:1 v/v CDCl3/MeCN-d3 with a Bip concentration of 18 mM.

1 2+ Figure 3.6 H-NMR of monomer 14 and Fe-MSP 14n+2·Fe 2n.

Monomer 14 was dissolved in CDCl3 (12 mM) and Fe(NTf2)2 was dissolved in MeCN-d3 (36 mM). After titration, the solution is ca. 3:1 v/v CDCl3/MeCN-d3 with a Bip concentration of 18 mM.

The accurate molecular weight determination of any supramolecular polymer is difficult to do. So in our initial studies we used diffusion ordered NMR spectroscopy (DOSY) to investigate the hydrodynamic volume of these MSPs. Figure 3.7a-c compares the DOSY of monomer 14, the

Zn-MSP and the Fe-MSP, respectively. The measured diffusion coefficients and their respective hydrodynamic radii (calculated using the Stokes-Einstein Equation) are shown in Table 3.1. At the same monomer 14 concentration (2.5 mM), the diffusion coefficient of both MSPs are much smaller that of the monomer, indicating the 14 is successfully assembled into larger supramolecular structures upon the addition of either Zn2+ or Fe2+. The Fe-MSP shows a smaller diffusion coefficient and therefore has a larger hydrodynamic radius than Zn-MSP, which is consistent with the stronger binding between Bip and Fe2+ than Zn2+ (at 2:1 Bip/metal ion molar ratio).

Both MSPs are soluble in DCM, which is the most widely used solvent for metathesis reactions. Therefore, after thoroughly drying and redissolving the MSP in DCM, the solution can be used directly for the ring-closing metathesis reaction at the desired reaction concentration.

Table 3.1 Diffusion Coefficient (D) and Hydrodynamic radius (Rh) of monomer 14 and corresponding MSPs. 7 2 * Sample D x 10 cm /s Rh (nm)

Monomer 14 12.1 1.13

Zn-MSP 5.03 2.71

Fe-MSP 3.40 4.01

* : Measured by DOSY in C2D2Cl4 at 25 °C. Concentration is 2.5 mM (based on 14) for all samples.

Figure 3.7 DOSY of Monomer 14 (a) and its corresponding Zn-MSP (b) and Fe-MSP (c).

DOSYs are measured in C2D2Cl4 (containing 1% v/v TMS) at 25 °C. The concentration of 14 is 2.5 mM for all samples.

2.2 The ring-closing metathesis (RCM) reaction

The RCM was performed by the addition of the Hoveyda-Grubbs catalyst (2nd generation) into the refluxing DCM solution of the Zn-MSP (Figure 3.8a), at a concentration of 2.5 mM (for monomer 14, unless otherwise mentioned) in order to favor the ring closing reaction. Refluxing

DCM not only increases the reaction kinetics, but also helps in removing the 2-butene condensate.

After reaction, the solution becomes slightly cloudy due to the decreased solubility of (some of) the products in DCM. The precipitation will dissolve upon the addition of a good solvent, e.g.

DMF, indicating they are not cross-linked. The metal ions can be removed by either adding a stronger binder, e.g. Zn2+ can be removed by the addition of ethylene diamine, or react with aqueous solution of NaOH to convert Zn2+ to sodium zincate. Fe2+ can be converted to iron (III) hydroxide by the addition of tetrabutylammonium hydroxide and then removed by filtration). As a control, monomer 14 can also be polymerized by acyclic diene metathesis polymerization

(ADMET) at higher concentrations (50 mM) into a non-interlocked linear polymer 17 (Figure

3.8b). Furthermore, monomer 14 was also subjected to the RCM conditions at a concentration of

2.5 mM, which should predominately yield the cyclized products (e.g. 18 as well as possibly cyclicdimers, cyclictrimers etc.) and less of the linear polymerized species (Figure 3.8c).

Figure 3.8 Illustration of the metathesis reaction of Zn-MSP (a) and the monomer 14 under high concentration ADMET (b) or low concentration RCM (c).

The crude products from the MSP reaction as well as the starting monomer and non-MSP control reactions were then studied by 1H-NMR (Figure 3.9). The spectrum of the product from the MSP reaction looks similar to that of the non-interlocked polymer 17 and monomer 14, indicating the major product is not the polycatenane or an interlocked species. However, if we take a careful look, the yellow-highlighted area shows a few new peaks that do not appear in monomer or any control. According to Chapter 2, chemical shift at this area is consistent with the formation of Bip-containing catenanes. Based on the NMR integration, the conversion of the monomer to interlocked species is ~30%. It is also worthwhile noting that the small peak at 8.30 ppm, which contributes ~5% of the integration (green-highlighted), lines up with the ring-closed monomer that

is also observed in RCM. The metathesis reaction of 14 at 2.5 mM yields mainly 18 (50% calculated from by NMR, 22% isolated).

Figure 3.9. 1H NMR comparison of the product from Zn-MSP templated synthesis reaction at 2.5 mM of 14, the ADMET and RCM of 14 without metal templation (synthesized at concentration of 50 and 2.5 mM of 14, respectively), as well as starting material 14 and purified 18.

The spectra of monomer 14, polymer 17 and the product are very similar, but the double- bond region (5.4 ~ 5.6 ppm) shows distinct differences. Using the staring material with >97% trans double-bond, the monomer has predominantly the trans-olefin geometry. However, in polymer 17 the double bond predominantly the cis isomer. The normalized integration of the double-bond peaks also decreased to ~50% of the original value, indicating the metathesis reaction is complete.

According to the literature, the cis/trans ratio of ADMET reaction depends on the structure of the monomer and catalyst, but there is no well-accepted mechanism behind this phenomenon.

Generally, ADMET reaction usually yields product with predominantly trans conformation, however, our ADMET product shows a cis/trans ratio of 78:22. A computational study by Cavallo and coworkers suggest that the formation of cis conformation is under kinetic control and is generated at the beginning of the reaction.5 Usually as the reaction proceeds, the cis product is gradually converted to trans product, which is thermodynamically more stable. However, this assumes that the catalyst is (a) active enough to open the 1,2 substituted double bonds, (b) steric hindrance of the cis double bond is not an issue and (3) the cis-product is soluble in the reaction mixture. As such one possible explanation is to our cis ADMET product is our reaction undergoes a kinetic controlled process. We do observed decreasing solubility during the RCM reaction of the

MSP and precipitation out of the solution starts to occur, which would make the product less accessible to the catalyst.

The Zn-MSP templated synthesis product shows a similar cis/trans ratio of 74:26. During the reaction, the precipitation of the product was also observed. Therefore, the kinetic controlled mechanism may also work for the Zn-MSP templated synthesis as the precipitated cis-product are not accessible to the catalyst for converting to its trans- isomer. It is interesting to note that for the

Zn-MSP templated synthesis, the metathesis reaction could been seen as undergoing a

(supramolecular) RCM rather than ADMET. For an RCM reaction, ring strain is another important consideration that can impact corresponding to the formation of cis- (or Z-) over trans- (or E-) products. However, as we see no difference in the cis:trans between the MSP-RCM and ADMET processes it appears that ring strain does not impact this. 80

The molecular weight of these samples was studied by GPC-MALLS, as shown in Figure

3.10, and the calculated absolute molecular weight is listed in Table 3.2. The product from the Zn-

MSP reaction shows a bimodal distribution with two broad peaks at 7.6 and 8.8 minutes, corresponding to Mn of 32.8 kDa and 10.39 kDa, respectively. The ADMET product 17 shows an elution peak at 8.2 min with Mn of 14.5 kDa, which is in between of the two peaks the Zn-MSP templated product. The monomer 14 shows a very narrow peak at 9.2 min corresponding to a Mn of 1.68 kDa (calculated by the MALLS detector), which is very close to its actual value (1.65 kDa).

The RCM reaction product shows generally three peaks: the first one is between 7.7 and 8.9 min and corresponds oligomers with average DP of 5 (Mn = 6.37 kDa). The second peak is at 9.1 min with Mn = 2.67 kDa, which is likely a mixture of the dimer and unreacted monomer. The third peak at 9.7 min is very strong and narrow with a Mn of 1.77 kDa, which corresponds to the cyclized monomer 18. Although the elution time of cyclized monomer 18 (9.7 min) is obviously longer than that of the monomer 14 (9.2 min), their actual molecular weights calculated by MALLS are very similar. This is because 18 has much smaller hydrodynamic volume than 14 due to the formation of cyclic structure; however, their actually molecular weights are similar and accurately detected by MALLS regardless of the column calibration. This result also supports the accuracy of using MALLS to calculate the absolute molecular weight of a macromolecule with non-linear architecture. Moreover, the 9.6 min peak also presents in the Zn-MSP templated product and the

ADMET product, indicating the formation of 18 as a byproduct in these reactions, which is in agreement previous NMR studies (highlighted green in Figure 3.9).

Figure 3.10 GPC-RI of the product synthesized by Zn-MSP templated metathesis (a), ADMET at high concentration without template (b), the reactant Monomer 14 (c) and RCM of 14 at low concentration (d).

Table 3.2 GPC-MALLS Molecular Weight Measurements Sample Peak Elution Time Range (min) Mn (kDa) Ð

Overall (6.8~9.4) 15.58 1.35 Zn-MSP templated 6.8~8.2 32.8 1.26 Crude Product 8.2~9.4 10.4 1.18

17 synthesized by ADMET 7.2~9.2 14.5 1.47

Monomer 14 9.0~9.5 1.68 1.00

Overall (7.7~9.8) 2.2 1.57

7.7~8.9 6.37 1.22 Crude RCM product 8.9~9.3 2.67 1.10

9.3~9.8 1.77 1.04

2.3 Isolation of Bip-containing polymer and catenanes.

Based on the above study, the NMR suggests that the Zn-MSP product contains ~65% Bip- containing non-interlocked polymer 17, ~5% cyclized monomer 18, and the rest of ~30% is probably polymeric/oligomeric catenane 15. As such attempts were made to isolate the catenated products from this reaction mixture.

Although the chemical composition of polymer 17 and the oligocatenane 15 are very similar, 15 should have higher mobility of freedom as a consequence of the mechanical bond. Therefore, it may be expected that the oligocatenanes 15 should have better solubility than 17 on account of a larger entropy gain during dissolving. By adding a small amount of DMF into dried (demetalated) crude product, it was observed that some of the crude product dissolved. By repeating this process, we separated the product into a DMF-soluble fraction and DMF-insoluble fraction. The DMF solution was then precipitated into acetonitrile to precipitate out the DMF soluble material.

Both fractions are studied by NMR and GPC-MALLS, respectively, as shown in Figure

3.11. As predicted the DMF-insoluble fraction is predominantly polymer 17. GPC-MALLS shows a bimodal distribution with two peaks at 7.7 min and 8.7 min, corresponding to the molecular weight of 31.4 kDa and 12.1 kDa, respectively. The overall polydispersity index (Ð) is 1.54, which is lower than the most probably distribution value (2.0) presumably due to the removal of small molecular weight fractions. According to 1H NMR DMF successfully dissolves the oligocatenane

15 as expected (yellow highlighted area in Figure 3.11a) as well as some linear polymer 17 and cyclized monomer 18. GPC-MALLS shows only one elution peak at 8.9 min, indicating the DMF- dissolved mixture has similar hydrodynamic volume. The number average molecular weight (Mn)

of the DMF-soluble fraction is 11 kDa with a Ð of 1.15, corresponding to an average degree of polymerization (DP) of 7.

Figure 3.11 1H-NMR (partial) and GPC (RI) of the crude and purified product.

The isolation yield of polymer 17 is ~55%, and the retrieved DMF-soluble material containing the mixture of 15, 17 and 18 (in the ratio of 10:2:3, based on NMR integration) is ~25% of the original mass. Because of the low yield and low molecular weight of the oligocatenane 15, we will focus on improving the yield and molecular weight for the synthesis of polycatenane in the following Chapter 4~6, rather than focusing on further purifying and study of the oligocatenane

15.

What is interesting to see is the effect that templating has on the resulting molecular weight of the polymer 17. Table 3.3 compares the Mn of 17 obtained from the MSP-templating approach

(at 2.5 mM) with the ADMET/RCM (i.e. with no metal ion) reactions run at 50 mM and 2.5 mM, respectively. It can be clearly seen that the using the metal ion templating allows access to much higher molecular weights than can be obtained by the metathesis reaction of 14 processed at the 84

same concentration (2.5mM). Furthermore, it even allows access to higher molecular weight polymer than the ADMET carried out at much higher concentrations such as 50mM: The first peaks of the GPC of the MSP templated product show larger molecular weight than are obtained with ADMET approach at 20 times the concentration. As such the MSP successfully acts as the template that helps to access products with higher molecular weight that can be achieved without the presence of the metal ion.

Table 3.3 Molecular weight of purified polymer 17 synthesized by Zn-MSP template, ADMET and RCM. Method Average cis/trans Yield Mn (kDa) Double-bond ratio Total 19.8 3.0 55% MSP templated Peak1 31.4 -- 35%* (2.5 mM, purified) Peak2 12.1 -- 20%*

ADMET (50 mM) 14.5 3.3 80%

RCM (2.5 mM) 4.7 -- 48%**

*: Calculated from their relative mass fraction GPC (Figure 3.11b) out of 55% isolation yield. **: Calculated from GPC.

2.4 The formation of polymer 17 and the improvement toward polycatenane

A possible explanation for the formation of high molecular weight polymer 17 from the

MSP is proposed in Figure 3.12a. For the MSP, the intra-metathesis reaction between ① and ② is competing with that between ② and ③. The former one results in the formation of the catenane

15 or macrocycle 18, while the latter one yields polymer 17 after demetalation, as illustrated in

Figure 3.12b. The MSP may also under go intermolecular reaction, forming a branched

supramolecular structure that will dramatically increase the resulting molecular weight of 17

(obtained after demetalation), which would be consistent with two molecular weight distributions.

Figure 3.12 Illustration of the bimodal distribution of polymer 17 with, and the formation of catenane 15 and cyclized monomer 18 during Zn-MSP templated synthesis.

Our study in chapter 2 shows monomer 14 has the ability to ring-closing to access the

[3]catenane in >97% conversion (i.e. the conversion of metallo complex toward the [3]catenane,

Figure 3.13a), which is the reason it was chosen for the polycatenane synthesis. However, our study for Zn-MSP shows that is not optimized for the synthesis of polycatenane from the metallosupramolecular polymer, which is believed due to the inefficient ring-closing between ① and ② discussed above (Figure 3.13b). To improve the synthesis of a polycatenane, we proposed

a strategy by designing a ditopic macrocycle (red ring, Figure 3.13c) and co-assemble into a next generation alternating MSP, which will be the main focus for Chapter 4.

Figure 3.13 The monomer 14 shows (a) excellent catenane conversion toward a [3]catenate, but (b) very low catenane conversion toward polycatenate. An improved strategy is proposed (c) by assembly with ditopic macrocycle, which will be focused on Chapter 4.

To further investigate the effect of the MSP-templating on the metathesis polymerization reaction, the xanthene linker moiety in monomer 14 was replaced with a biphenyl moiety, 16 (as discussed in Figure 3.4b). Based on the study in Chapter 2, monomer 16 which contains the biphenyl moiety is less favorable toward the formation of a catenane structure. Therefore, we

2+ expect that the metathesis reaction of MSP 16n+2·Zn 2n should yield little-to-no catenane (Figure

3.14a) and the product should be predominantly the non-interlocked linear polymer 19, which, of course, can also be synthesis by the metathesis of 16 under high concentration reaction conditions

(Figure 3.14b). The metathesis of 16 at low concentration may also not yield much cyclic monomer

20 either since the cyclization toward 20 should be hindered on account of unfavorable conformational effects.

Figure 3.14 Illustration of the metathesis reaction of monomer 16 with biphenyl liner moiety. Due to the unfavored “s-like” conformation, biphenyl-linked monomer is unlikely to form any catenane structure during MSP templated synthesis (a). Metathesis reaction under low-concentration (i.e. the RCM condition) are also less likely toward cyclic product 20 (b). The major product should always be the linear polymerized product 19 (polymer or oligomer).

2.5 Comparison with the Biphenyl-linked Monomer 16

To prove this assumption, we synthesized and characterized the product of these three reactions by 1H-NMR (Figure 3.15). Clearly, the dominant product of all the reactions is polymer

19, the NMR of which is very similar to the starting material 16. Based on NMR, the conversion toward ring-closed species 20 (and possibly also its cyclic dimer, trimer, etc.) is indeed minor

(green-highlighted). It can be more clearly observed when comparing to the xanthene-linked monomer 14 as shown in Figure 3.16, in which the two bottom spectra in red are from the xanthene-monomer 14 while the other two in cyan on top are from biphenyl-monomer 16. Clearly, the xanthene-monomer shows some conversion to oligocatenanes using the Zn-MSP templated

synthesis while biphenyl-monomer shows little-to-no catenane formation (Figure 3.16a). The ring- closing metathesis of each monomer at similar dilute concentrations (without templating) also shows that xanthene-monomer results in much higher conversion toward its corresponding ring- closed monomer.

Figure 3.15 1H-NMR (partial) comparison of biphenyl-linked monomer (a) and its corresponding metathesis products synthesized by low concentration “RCM” condition (b), high-concentration ADMET condition (c), and MSP templated approach (d).

Figure 3.16 Partial 1H-NMR comparison of the of products synthesis from the monomers 14 and 16. Product synthesized by Zn-MSP templated approach (a) or low-concentration metathesis (b) from either xanthene-linked monomer 14 (bottom, red) or biphenyl-linked monomer 16 (top, cyan). Yellow highlighted area shows the peak of catenanes and green highlighted area shows that of ring-closed monomer.

The biphenyl-linked polymer 19 or ring-closed monomer 20 and other cyclic species are insoluble in THF or DMF (chloroform is the only solvent for them we found this far) and as such, due to instrumental limitations, we are not able to carry out any GPC characterization on these products at this time. However, based on the conversion of double-bonds we estimate that the average DP of the products are 5.6 (MSP templated), 4.76 (high-concentration ADMET) and 4.55

(low-concentration metathesis), respectively (For detail, see the experimental section). The DP of all these polymers are very low and similar presumably on account of the solubility of the polymer

(in DCM) decreasing significantly with increasing molecular weight, precipitating out of the solution hindering further growth.

2.6 Synthesis templated by Fe-MSP

As previously discussed in this chapter (section 2.3 and Figure 3.12), the unwanted formation of polymer 17 (if polycatneanes are being targeted) occurs via the metathesis side- reaction between ② and ③ as well as the intermolecular reaction between two Zn-MSP chains.

Besides of Zn2+, other metal ions, such as Fe2+, can also be used to template the formation of the

2+ supramolecular polymer, resulting in the Fe-MSP 14n+2·Fe 2n (as previously discussed in Figure

3.6). In comparison, at a 2:1 Bip/metal ion ratio, the overall binding constant for Fe2+ (>1010 M−2)2 is much larger than that of Zn2+ (ca. 106 M−2)3. Therefore, we are able to perform the ring-closing metathesis reaction of Fe-MSP at a much lower concentration (e.g. 0.25 mM, w.r.t. 14 for all metathesis reactions unless mentioned elsewhere) than that of Zn-MSP (2.5 mM), which should significantly hinder the intermolecular side reactions.

The 1H-NMR of crude product synthesized from Fe-MSP (RCM reaction at 0.25 mM) was compared to that synthesized from Zn-MSP (RCM reaction at 2.5 mM), as shown in Figure 3.17.

For Fe-MSP templated synthesis, the formation of catenane product is improved to ca. 55% (based on NMR integration) compared to ca. 30% for Zn-MSP templated synthesis. The formation of 18 by Fe-MSP templated synthesis at 0.25 mM is ca. 10% (NMR integration), which is higher than

5% synthesized by Zn-MSP at 2.5 mM. This result indicates that to further increase the yield toward the synthesis of polycatenanes, the intramolecular side reactions (between ② and ③) must be avoided. The improved approaches toward this goal will be studied in Chapter 4~6 of this thesis as will a more detail discussion of the NMR characterization of the catenated species.

Figure 3.17 1H-NMR (partial) comparison of the crude polymer product synthesized by Zn-MSP template (a) and Fe-MSP template. Green highlighted area shows the formation of cyclized monomer 18 and yellow highlighted area shows the formation of catenane 15.

2.7 Hydrogenation of the double-bond

The double-bonds on the polymer 17 main-chain and the macrocycle of polycatenane 15 are vulnerable to oxidation, which will decrease the molecular weight of polymer. To solve the problem, we can selectively hydrogenation of the double-bond without hydrogenating the aromatic rings. Besides, after hydrogenation 15 becomes more stable, allowing us to do many types of functionalization in future, as shown in Figure 3.18.

Figure 3.18 Potential functionalization of the hydrogenated polycatenane.

As a proof-of-concept study, the hydrogenation reaction on the ADMET produced 17 was carried out using p-toluenesulfonyl hydrazide as reductant, as shown in Figure 3.18. The NMR spectra before and after hydrogenation are very similar except for the totally disappearance of the double-bond peaks. GPC shows slightly longer elution time after hydrogenation, but the actually average molecular weight is larger. The larger molecular weight is probably due to the removal of some smaller molecular-weight during purification; the Ð is decreased from 1.51 to 1.41. The slightly longer elution time indicates the hydrodynamic volume becomes smaller after hydrogenation possibly due to decreased rigidity by hydrogenation of the double bond. These results prove the successful hydrogenation of Bip-containing polymer without any observable degradation of the polymer, which is very important for future application in the hydrogenation of polycatenanes.

Figure 3.19 Schematic illustration of the hydrogenation of 17 (a), and the comparison of GPC-RI (b) and 1H-NMR (partial) of 17 and 20. Blue highlighted area indicates the total disappearance of the double-bond peaks in 1H-NMR.

3. Conclusion

In this chapter, monomers with either xanthene or biphenyl liner moiety were successfully assembled into metallosupramolecular polymers upon the addition of Zn2+ (or Fe2+). For xanthene- lined monomer 14, metathesis reaction of zinc templated metallosupramolecular polymers (Zn-

MSP) yields mainly the linear polymerized monomer 17 and very small amount of catenanes and cyclized monomer 18. Compared to the same polymer 17 synthesized from ADMET, Zn-MSP templated synthesis is able to obtain much higher molecular weight (up to 32.8 kDa, versus 14.5 kDa for ADMET). However, the Zn-MSP templated product shows a bimodal molecular weight 94

distribution, possibly due to the occurrence of some intermolecular metathesis reactions. The biphenyl-linked monomer 16, on the other hand, yields almost no catenane or cyclized monomer due to its more preferred “S-like” conformation that does not favor ring-closing. By switching from Zn2+ to Fe2+, we are able to perform the metathesis reaction at much lower concentration. For xanthene-linked monomer, the conversion toward a catenaned/interlocked product is improved although the major product is still the non-interlocked polymer. Therefore, we must suppress the intramolecular side reaction in order to synthesize the target polycatenane in good yield, which will be discussed in Chapter 4-6. Finally, we also successfully hydrogenated the double-bond after the metathesis reaction without any unwanted side reactions. Hydrogenation opens the door to many ways of potential future functionalization of the polycatenanes.

4. Experimental

4.1 Materials and Methods

4.1.1 Chemical and Solvents

Dichloromethane (DCM) for the metathesis reaction was purchased from ACROS

Organics (extra dry, contains no stabilizer) and distilled over CaH2 under argon atmosphere before using. Dimethylformamide (DMF) was purchased from ACROS (extra dry with molecular sieves in an AcroSeal bottle) was used as received. Deuterated solvents were purchased from ACROS.

All other solvents and anhydrous potassium carbonate were purchased from Fisher Scientific and used without purification. Zinc bistriflimide was purchased from Strem Chemicals and stored in vacuum desiccator. Irom (II) bistriflimide was prepared according to literature.6 All other chemicals were purchased from Sigma-Aldrich and used without further purification.

4.1.2 Instrumentations

All NMR spectra (except DOSY) was recorded on a Varian Inova 600 MHz spectrometer

(150.8MHz for 13C) at 25 °C (unless otherwise mentioned). All DOSYs were measured on a Bruker

Ascend Advance III 500 MHz spectrometer equipped with CryoProbe using the bipolar pulse pair stimulated echo (Dbppste) pulse sequence at 25 °C. All chemical shifts were calibrated to TMS for all measurements. NMR spectra were processed by MestReNova software, and DOSY spectra was processed with Bayesian DOSY transform. GPC-MALLS was measured on Agilent 1260 infinity GPC system with PLgel MIXED-C column, Wyatt DAWN HELEOS MALLS detector and Wyatt Optilab T-rEX RI detector, using THF was used as the eluent.

4.1.3 General Procedure of Metathesis Reactions

Olefin metathesis reactions, including RCM and ADMET, were used for the synthesis of compounds 15, 17, 18, 19 and 20. The reactions were performed by dissolving the starting material in distilled DCM at the desired concentration and the solution was heated to reflux and bubbled with argon for 30 minutes to remove the dissolved oxygen. Then, the first of two portions of

Hoveya-Grubbs 2nd generation catalyst dissolved in 0.5 mL distilled DCM was added, and the solution was bubbled with argon for another 30 minutes. The amount of each portion of catalyst was calculated by the following equation: m = V × 0.2 mL/mg, where m is the amount of catalyst

(in mg) and V is the volume of the reaction mixture (in mL). The reaction was refluxed for 24 hours. Then the second portion of catalyst (same amount as first portion) was added after 24 hours followed by further bubbling with argon for 30 minutes. The reaction solution was kept at reflux for another 24 hours. At which point the solution was cooled to room temperature followed by the addition of excess ethyl vinyl ether to deactivate the catalyst. 96

4.1.4 Purification of Polymers 17 and 19

For the non-templated synthesis of 17 and 19 the product was dissolved in chloroform and precipitated into MeCN to remove the deactivated catalyst.

For the syntheses of 17 prepared via the MSP templated method, a very small amount of

DMF (10 µL per 1mg crude product) was added into the dried product followed by shaking on a lab shaker overnight to dissolve the cyclic and catenane products into DMF. The DMF solution was then carefully decanted. The remaining non-soluble products were dried under vacuum. The

DMF solution that dissolves the catenane products (as well as cyclized 18) was also collected, and the solvent was removed under vacuum to collect the catenane product for studies.

4.2 Synthesis of Components

4.2.1 Monomer 14 or 18

Monomer 14 or 18 was synthesized in two steps from Butyl-Bip 21:

Step 1: 5g (8.23 mmol) of 21 (synthesized via a literature procedure)7, 2.09 g (8.23 mmol,

8 1 eq.) of tosylated 4-hexen-1-ol (literature procedure) and 3.4 g (24.69 mmol, 3 eq.) of K2CO3 were added into a 100 mL round bottom flask equipped with stir bar. The flask was flushed with argon and then 40 mL anhydrous DMF was added by cannula. The reaction mixture was stirred at

70 °C for 24 hours. DMF was then removed under vacuum and the solid was stirred with chloroform and filtered. The filtrate was collected and the solvent removed under reduced pressure.

The resulting material was purified using column chromatography (silica gel, chloroform/methanol gradient as eluent) and recrystallization (chloroform/methanol mixture).

1 Yield: 46% (theoretical yield is 50%), as yellow crystals. H NMR (600 MHz, CDCl3) δ 8.30 (d,

J = 7.8 Hz, 2H), 8.07 (t, J = 7.8 Hz, 1H), 8.03 (d, J = 1.6 Hz, 1H), 8.01 (d, J = 1.6 Hz, 1H), 7.62–

7.58 (m, 2H), 7.58 – 7.55 (m, 4H), 7.50 (d, J = 8.4 Hz, 2H), 7.03 – 7.00 (m, 2H), 6.98 – 6.94 (m,

2H), 5.54 – 5.44 (m, 2H), 4.76 (t, J = 7.4 Hz, 4H), 4.02 (t, J = 6.5 Hz, 2H), 2.19 (q, J = 6.8 Hz,

2H), 1.88 (p, J = 6.7 Hz, 2H), 1.75 (p, J = 7.5 Hz, 4H), 1.68 – 1.66 (overlapped, 3H), 1.15 (h, J =

13 7.4 Hz, 4H), 0.73 (t, J = 7.4 Hz, 6H). C NMR (126 MHz, CDCl3) δ 158.48, 156.05, 150.60,

150.53, 149.83, 149.69, 143.15, 142.95, 138.40, 136.63, 136.40, 135.34, 135.21, 134.02, 133.60,

130.29, 128.58, 128.38, 125.75, 125.67, 123.26, 117.94, 117.86, 116.12, 114.91, 110.52, 67.41,

44.82, 32.17, 29.13, 28.96, 19.90, 17.97, 13.53. MALDI-TOF MS: 690.49 ([M]+H+).

Step 2 for 14: In a 50 mL round bottom flask equipped with stir bar, 2.1g (3.0 mM) of the product from step 1, 0.6 g (1.5 mM, 0.5 eq.) of 3,6-bis(bromomethyl)-9,9-dimethyl-9H-xanthene

(synthesized by literature procedure)8 and 1.97 g (6.0 mM, 2 eq.) of anhydrous cesium carbonate was added, followed by flushing with argon. Then 15 mL anhydrous DMF was injected into the reaction. The mixture was stirred at 70 °C for 24 hours. DMF was then removed under vacuum and the solid was stirred with chloroform and filtered. The filtrate was collected and the solvent removed under reduced pressure. The resulting material was purified using column chromatography (triethylamine pretreated silica, hexane/chloroform/methanol gradient from

1 75/25/0 to 0/99/1 (v/v/v) as the eluent). Yield of 14: 52%, white solid. H NMR (600 MHz, CDCl3)

δ 8.35 (d, J = 7.8 Hz, 4H, B), 8.07 (t, J = 7.9 Hz, 2H, A), 8.03 (d, J = 5.1 Hz, 4H, C), 7.63 (d, J =

8.2 Hz, 4H, F), 7.61 (d, J = 8.2 Hz, 4H, F’), 7.58 (dd, J = 8.8, 3.2 Hz, 4H, E), 7.50 (dd, J = 8.5, 3.3

Hz, 4H, D), 7.46 (d, J = 8.1 Hz, 2H, H), 7.20 (overlapped, 4H, I, J), 7.11 (d, J = 8.2 Hz, 4H, G),

7.01 (d, J = 8.2 Hz, 4H, G’), 5.49 (quint, J = 5.6 Hz, 4H, P, Q), 5.12 (s, 4H, L), 4.76 (t, J = 7.4 Hz,

8H, U), 4.02 (t, J = 6.4 Hz, 4H, M), 2.19 (q, J = 6.7 Hz, 4H, O), 1.88 (quint, J = 6.9 Hz, 4H, N),

1.75 (quint, 8H, V), 1.66 (overlapped, 12H, K, R) 1.15 (h, J = 7.5 Hz, 8H, W), 0.73 (t, J = 7.4 Hz,

13 12H, X). C NMR (126 MHz, CDCl3) δ 158.48, 158.11, 150.74, 150.71, 150.46, 150.04, 150.02,

143.46, 138.21, 136.68, 136.25, 136.12, 135.55, 135.49, 134.62, 134.12, 130.33, 129.63, 128.47, 99

128.37, 126.59, 125.76, 125.57, 123.16, 122.16, 118.12, 118.07, 115.41, 115.25, 114.94, 110.54,

110.51, 69.54, 67.42, 44.81, 33.94, 32.61, 32.22, 29.16, 28.99, 19.92, 18.01, 13.58. MALDI-TOF

MS: 1635.05 ([M]+Na+).

Step 2 for 18: In a 50 mL round bottom flask equipped with stir bar, 2.1g (3.0 mM) of the product from step 1, 0.38 g (1.5 mM, 0.5 eq.) of and 1.97 g (6.0 mM, 2 eq.) of anhydrous cesium carbonate was added, followed by flushing with argon. Then 15 mL anhydrous DMF was injected into the reaction mixture. The mixture was stirred at 70 °C for 24 hours. DMF was then removed under vacuum and the solid was stirred with chloroform and filtered. The filtrate was collected and the solvent removed under reduced pressure. The resulting material was purified using column chromatography (triethylamine pretreated silica, hexane/chloroform/methanol gradient from

1 75/25/0 to 0/99/1 (v/v/v) as the eluent). Yield of 18: 61%, white solid. H NMR (600 MHz, CDCl3)

δ 8.35 (d, J = 7.8 Hz, 4H), 8.09 – 8.03 (overlapped, 6H), 7.68 – 7.54 (overlapped, 20H), 7.50 (dd,

J = 8.5, 3.9 Hz, 4H), 7.12 (d, J = 8.1 Hz, 4H), 7.01 (d, J = 8.2 Hz, 4H), 5.49 (m, 4H), 5.18 (s, 4H),

4.76 (t, J = 7.4 Hz, 8H), 4.02 (t, J = 6.4 Hz, 4H), 2.19 (q, J = 6.8 Hz, 4H), 1.88 (quint, J = 6.8 Hz,

4H), 1.75 (quint, J = 7.6 Hz, 9H), 1.15 (q, J = 7.5 Hz, 8H), 0.73 (t, J = 7.4 Hz, 12H). 13C NMR

(126 MHz, CDCl3) δ 158.46, 158.13, 150.74, 150.01, 143.42, 140.58, 140.41, 138.20, 136.21,

136.10, 135.53, 135.46, 134.65, 134.12, 130.30, 128.47, 128.37, 128.05, 127.41, 125.76, 125.55,

123.14, 118.09, 115.25, 114.91, 110.48, 69.90, 67.41, 44.79, 32.20, 29.14, 28.96, 19.91, 17.97,

13.55. MALDI-TOF MS: 1558.13 ([M]+H+)

100

4.2.2 Polymerized Xanthene-Monomer 17

17 was synthesized via the ADMET of 14 as following:

In a 5 mL conical vial equipped with stir bar and condenser, 100 mg (0.06 mmol) of 14 was dissolved in 1.2 mL of DCM (final concentration 50 mM). Then the ADMET reaction was performed following the olefin metathesis procedure described above (Chapter 3 Section 4.1.3).

According to the procedure, two portions of Hoveyda-Grubbs 2nd generation catalyst (0.24 mg per portion) was added at the beginning and after 24 hours, respectively. It is worthy to note that when bubbling argon to remove oxygen under the reflux condition, the DCM will slowly escape from the system even with a very efficient condenser. However, during the addition of catalyst

(dissolved in 0.5 mL of DCM), the 0.5 mL DCM offsets the evaporated DCM, maintaining a

1.2mL volume after oxygen removal. After that, the reaction was protected by argon filled balloon and the evaporation of DCM is neglectable with a good condenser. After reaction, the product was purified by the procedure described above in Chapter 3 Section 4.1.4. Briefly, the 1.2 mL of reaction mixture was precipitated into 20 mL of MeCN. The reaction vial was rinsed twice with

0.4 mL of DCM, which were also precipitated in to MeCN. The suspension was sit in fridge overnight to allow complete precipitation. The product was collected by filtration and washed three times with cold MeCN. Yield is quantitative and the product is light brown solid. The color is due

101

1 to the trace amount residue deactivated catalyst. H NMR (600 MHz, CDCl3) δ 8.35 (d, 4H, B),

8.07 (t, 2H, A), 8.03 (s, 4H, C), 7.65 – 7.55 (m, 12H, F, F’, E), 7.50 (dd, 4H, D), 7.46 (d, 2H, P),

7.20 (overlapped, 4H, Q, R), 7.11 (d, 4H, G), 7.01 (d, 4H, G’), 5.56 – 5.48 (m, 2H, K cis/trans),

5.12 (s, 4H, S), 4.76 (t, 8H, U), 4.03 (t, 4H, H), 2.32-2.21 (m, 4H, J cis/trans), 1.89 (quint, 4H, I),

1.74 (quint, 8H, V), 1.66 (s, 6H, T), 1.14 (quint, 8H, W), 0.73 (t, 12H, X). 13C NMR (126 MHz,

CDCl3) δ 158.42, 158.07, 150.69, 150.43, 149.98, 143.39, 138.17, 136.63, 136.20, 136.09, 135.49,

135.44, 134.60, 134.12, 130.13, 129.68, 129.61, 128.44, 128.35, 126.56, 125.52, 123.12, 122.12,

118.09, 118.03, 115.38, 115.19, 114.88, 110.47, 69.52, 67.32, 44.76, 33.90, 32.56, 32.17, 31.93,

29.70, 29.11, 28.97, 22.70, 19.88, 19.79, 14.13, 13.52.

4.2.3 Ring-closed Monomer 18

17 was synthesized by RCM of 14 as following:

In a 50 mL round bottom flask, 50 mg (0.03 mmol) of 14 was dissolved in 12 mL of DCM following the procedure described above (Chapter 3, Section 4.1.3). After reaction and catalyst deactivation, the solvent was removed under vacuum and the product purified by column chromatography with triethylamine neutralized silica gel and a hexane/chloroform/methanol 102

gradient from 75/25/0 to 0/99/1 (v/v/v) as the mobile phase. Yield: 22%. The product is colorless

1 waxy solid which is a mixture of cis and trans isomers. H NMR (600 MHz, CDCl3, cis/trans isomers are highly overlapped) δ 8.32 – 8.26 (overlapped, 4H, B), 8.03 – 7.91 (overlapped, 6H, A,

C), 7.54 – 7.29 (overlapped, 16H, F, F’, E, H), 7.14 – 7.07 (overlapped, 4H, I, J) 6.99 – 6.89

(overlapped, 8H, G, G’) , 5.52 – 5.46 (overlapped, 2H, P), 5.25 (S, 4H, L), 4.70 – 4.59 (overlapped,

8H, U), 4.03 – 3.84 (overlapped, 4H, M), 2.32 – 2.20 (overlapped, 4H, O), 1.93 – 1.82 (overlapped,

4H, N), 1.70 – 1.59 (overlapped, 14H, K, V), 1.08 – 0.99 (overlapped, 8H, W), 0.66 – 0.58

13 (overlapped, 12H, X). C NMR (126 MHz, CDCl3) δ 158.33, 157.43, 150.54, 150.49, 149.97,

143.35, 143.30, 138.07, 137.43, 136.12, 135.38, 134.39, 133.98, 130.43, 129.89, 129.22, 128.35,

128.32, 128.27, 126.61, 125.44, 123.17, 121.09, 117.94, 115.62, 114.90, 114.76, 114.49, 110.43,

69.12, 66.69, 44.66, 33.82, 32.49, 32.10, 28.61, 28.49, 19.80, 13.47. MALDI-TOF MS: 1558.03

([M]+H+).

4.2.4 Hydrogenated Xanthene-Monomer 20

100 mg of the ADMET polymerized polymer 17 (containing 0.064 mmol of double bond),

26.7 mg (0.32 mmol) of p-toluenesulfonyl hydrazide and 32.5 mg (0.32 mmol) of diisopropylamine was dissolved in 0.64 mL of o-dichlorobenzene. The solution was purged with

103

argon before being reacted at 110 °C for 24 hours. After cooling down to room temperature, the product was collected by precipitation of the reaction mixture into cold MeCN followed by filtration to yield light brown solid in 95% (color is due to residue deactivated catalyst from the

1 reactant.). H NMR (600 MHz, CDCl3) δ 8.34 (d, J = 7.8 Hz, 4H, B), 8.07 (t, J = 7.9 Hz, 2H, A),

8.03 (t, J = 2.1 Hz, 4H, C), 7.66 – 7.56 (overlapped, 12H, E, F, F’), 7.50 (dd, J = 8.6, 2.7 Hz, 4H,

D), 7.46 (d, J = 8.1 Hz, 2H, P), 7.20 (overlapped, 4H, Q, R), 7.11 (d, J = 8.3 Hz, 4H, G), 7.02 (d,

J = 8.3 Hz, 4H, G’), 5.12 (s, 4H, S), 4.76 (t, J = 7.3 Hz, 8H, U), 4.04 (t, J = 6.5 Hz, 4H, H), 1.84

(quint, J = 7.8 Hz, 4H, I), 1.75 (quint, J = 7.4 Hz, 8H, V), 1.66 (s, 6H, T), 1.53 (br, 4H, J), 1.44

13 (br, 4H, K), 1.14 (q, J = 7.5 Hz, 8H, Vi), 0.73 (t, J = 7.4 Hz, 12H, X). C NMR (126 MHz, CDCl3)

δ 158.47, 158.07, 150.70, 150.67, 150.43, 149.98, 143.39, 138.18, 136.63, 136.23, 136.09, 135.50,

135.44, 134.61, 134.07, 129.61, 128.44, 128.36, 126.56, 125.53, 123.13, 122.12, 118.09, 118.03,

115.38, 115.20, 114.87, 110.47, 69.53, 68.07, 44.77, 33.91, 32.57, 32.18, 29.70, 29.36, 29.32,

26.05, 19.88, 13.53.

104

Reference:

1 Y. Kim, M. F. Mayer and S. C. Zimmerman, Angewandte Chemie, 2003, 115, 1153–1158.

2 M. Enamullah and W. Linert, Journal of Coordination Chemistry, 1996, 40, 193–201.

3 S. J. Rowan and J. B. Beck, Faraday Discuss., 2005, 128, 43–53.

4 A. W. Addison, S. Burman, C. G. Wahlgren, O. A. Rajan, T. M. Rowe and E. Sinn,

Journal of the Chemical Society, Dalton Transactions, 1987, 2621.

5 N. Bahri-Laleh, R. Credendino and L. Cavallo, Beilstein Journal of Organic Chemistry,

2011, 7, 40–45.

6 M. P. Sibi and G. Petrovic, Tetrahedron: Asymmetry, 2003, 14, 2879–2882.

7 B. M. McKenzie, A. K. Miller, R. J. Wojtecki, J. C. Johnson, K. A. Burke, K. A. Tzeng,

P. T. Mather and S. J. Rowan, Tetrahedron, 2008, 64, 8488–8495.

8 R. J. Wojtecki, Q. Wu, J. C. Johnson, D. G. Ray, L. T. J. Korley and S. J. Rowan, Chem.

Sci., 2013, 4, 4440.

105

Chapter 4. The Synthesis of Poly[n]catenanes

1. Introduction

Among all mechanically interlocked polymers discussed in Chapter 1, the main-chain polycatenane is perhaps the more intriguing as its entire polymer main-chain consists of macrocycles held together by the topological bond. Therefore, assuming free rotation of the rings, these polymers possess the maximum degree if structural flexibility throughout the entire molecule no matter whether the macrocycles are rigid or flexible. As shown in Figure 4.1a, a flexible polycatenane chain is conceptually able to switch reversibly between its fully extended and highly coiled conformations, with no bond angle or torsion angle strain during the switching process. The flexibility is also potentially controllable through the non-covalent interaction between the macrocycles. As such by altering the strength of the non-covalent interactions it is possible to control the topological mobility of the macrocycles and therefore switch the polycatenane chain between highly flexible and rigid rod structures. As mentioned in Chapter 1, materials based on polycatenanes will likely possess large loss modulus and low activation energy for flow, and have the potential to act as outstanding energy damping materials, elastomers with excellent toughness and stimuli-responsive mechanical properties.

106

Figure 4.1 Topological flexibility of the polycatenane chain. (a) The flexible polycatenane chain can switch between fully extended and highly coiled chain conformations; (b) by managing the mobility between macrocycles, polycatenanes can also reversibly switch between flexible chains and rigid rods.

The synthesis of main-chain polycatenanes, however, is one of the most challenging topics in organic and polymer chemistry. During the past few decades, supramolecular chemists, such as

Stoddart, Sauvage, Harada, Gibson and Sanders, pioneered this field and successfully developed high-yielding templated synthesis of rotaxanes, catenanes, knots, polyrotaxanes, poly[2]catenanes and side-chain polycatenanes. The synthesis of a poly[n]catenane, however, has still not been achieved.

107

Figure 4.2. Reported synthetic approaches toward main-chain polycatenane and [5]catenanes.

As previously discussed in Chapter 1, since 1990s many attempts toward the synthesis of main-chain polycatenanes have been reported. Briefly, one may propose a direct synthesis from two entangled semicircles or pseudo[3]rotaxane (Scheme 4.1a) by reacting A with B. However, these tetrafunctional monomers can also act as good crosslinkers resulting in polymer gels or nanogels. The problem here is that the cyclization reaction gives the highest yield under low concentration reaction conditions while the polymerization reaction requires high concentrations to access high molecular weights. Shaffer and coworkers proposed a modified approach that disengaged the polymerization and cyclization steps by preforming a linear supramolecular polymer followed by ring-closing (Figure 4.2b), although they never tried this reaction in the paper as their polymer was not soluble enough. In fact, to date, there have been no report of any

108

successful preparation of polycatenane via this approach. Kihara proposed an alternative approach that circumvents the difficult cyclization reaction by accessing a ladder-like bridged poly[2]catenane by using a Diels-Alder reaction in the polymerization step. Conceptually the desired poly[n]catenane could then be accessed by cleavage of the internal double bonds. However to date no successful preparation of a polycatenane has been reported using this approach. To date,

Stoddart and coworkers have reported the longest linear main-chain [n]catenane which was synthesized using a stepwise approach (Scheme 4.1d). By sequentially clipping on the macrocycles a linear [5]catenane, also termed Olympiadane, could be synthesized in 0.3% overall yield over 2 steps. More recently, Iwamoto and coworkers modified the step-wise approach by dimerizing two

[3]pseudorotacatenanes into a linear [5]catenane in a 12% yield.

Although the stepwise approach allows access to linear oligocatenanes, the multiple repetitive steps required are tedious and the yield for each step is relatively low meaning that such an approach is not a practical way for the synthesis of polycatenanes. We propose that disengaging the polymerization and cyclization reactions is key for the successful synthesis of polycatenanes.

Our approach is to assemble a metallo-supramolecular polymer (MSP) followed by an efficient ring-closing step to yield a poly[n]catenate, which can then be converted in a poly[n]catenane by removal of the metal ion. In Chapter 2 we demonstrated a successful candidate—the threading monomer with xanthene linker—that quantitatively converts the [3]pseudorotaxane to [3]catenate via intramolecular ring closing reaction (Figure 4.3a). Assembling this monomer into a homo- supramolecular polymer Zn-MSP, however, does not result in the efficient ring-closing reaction to yield the macrocycle, as discussed in Chapter 3. Instead, the majority product turns to be the linear polymerized monomer while trace amount of catenane species can be detected. As shown

109

in Scheme 4.2b, the results indicate that the reaction between ② and ③ is preferred to that between

① and ②. Therefore, rationally redesigning the current system is required.

Figure 4.3. Efficient one-bond-formation ring-closing synthesis of [3]catenane (a), the failure of synthesizing polycatenane from Zn-MSP (b), and our modified approach toward the synthesis of polycatenane (c).

To promote the conversion toward polycatenanes, one strategy is co-assembling the monomer with a pre-formed cyclic molecule, i.e. a ditopic macrocycle (DiMC) that contains two

Bip moieties, to access a supramolecular alternating copolymer Zn-alt-MSP, as shown in the retrosynthetic analysis (Figure 5.3c). DiMC helps isolate ② and ③ hindering the reaction between them and as such facilitates the desired reaction between ① and ② required to access poly[n]catenanes 22.

110

2. The Synthesis and Characterization of Polycatenane

2.1 Component Design

To assemble the supramolecular polymer Zn-alt-MSP the monomer 14, which contains a xanthene linker and n-butyl side-chains, was designed as the threading component (Figure 4.4a).

The xanthene linker helps the molecule adopt a “C-like” structural conformation in order to favor the ring-closing reaction (as discussed in Chapter 2), and the n-butyl side-chains provides solubility while at the same time should have a lower steric hindrance compared to n-hexyl side- chains (as discussed in Chapter 3). The two double-bonds can react via an olefin metathesis reaction, forming the ring-closed monomer 18 with a ring-size of 63 atoms—which is enough to thread two Bip-containing macrocycles (as demonstrated in Chapter 2).

Figure 4.4. Structure of (a) the monomer 14 and (b) the cyclized monomer 18.

An appropriate DiMC can be designed by simply connecting two Bips with two linker moieties (Figure 4.5a). The nature of the linker group is key as it controls the size and flexibility of the macrocycle while the side-chain will affect both the solubility and the ring closing reaction

(as discussed in Chapter 2). The initial target candidate was 23 (Figure 4.5b). Tetraethylene glycol linkers gave 23 a ring-size of 68-atoms, which is larger than 18 (Figure 4.4b) to ensure the desired

111

doubly threading of two Bip moieties through one DiMC cavity. The n-butyl side-chain provides a balance between adequate solubility of the metallosupramolecular polymer (MSP) Zn-alt-MSP, while trying to minimize steric bulk. In order to maximize the enthalpy gain during the MSP self- assembly, all Bip units need to bind with a metal ion (e.g. Zn2+ or Fe2+) while all metal ions need to coordinate two Bip units. In order to access high molecular weight assemblies it is critical that the DiMC must not be able to form a 1:1 complex with the metal ion (where both its Bip units are binding to the same metal ion). The assembly of two DiMC (Figure 4.5c) is already forbidden due to the covalently bonded cyclic structure of DiMC and the geometry restriction of Bip/metal coordination (for details, see Figure 2.2 in Chapter 2). However, it is important to rule out the possibility of self-binding of the DiMC, shown in Figure 4.5d, to ensure that the Bip moiety of the

DiMC can only bind with a metal ion together with a Bip moiety from monomer 14 to form the

Bip2/metal complex. If the DiMC can only participate in this Bip2/metal complex the self-assembly of all the components into the desired metallo-supramolecular alternating copolymer Zn-alt-MSP will occur, assuming 14, 18 and Zn2+ (or Fe2+) are at the exact molar ratio of 1:1:2 and allowed to fully equilibrate.

112

Figure 4.5 The molecular design of DiMC (a) and the structure of the first candidate 23 (b). The ring-structure forbids the self-assembly between two DiMCs (c), but we still need to prevent the self-compromised binding (d) during the molecular design.

2.2 Polycatenane Synthesis and Purification

2.2.1 Self-assembly of the Supramolecular Polymer Precursor

The self-binding behavior of 4-5 was investigated via the titration with Fe2+ions, which was monitored by Ultraviolet–visible spectroscopy (UV-Vis, see Figure S4.1a, supporting information). Figure 4.6a shows the full UV-Vis spectra of the complexes formed with Fe2+ with either Bip 24, DiMC 23, and their complexes formed with Fe2+, respectively, respectively, at a 2:1

Bip/metal stoichiometry and the same Bip concentration. As can be seen two distinct spectra were observed. Both show a new peak at 382 nm consistent with the formation of a metal complex that can be clearly observed by titration (Figure S4.1, supporting information). However, of particular interest is the broad peak between 550~600 nm that corresponds to the metal-ligand charge transfer

2+ 2+ 1 (MLCT) between Fe and Bip, and is a characteristic peak for the Bip2/Fe complex. This peak is observed for the Bip complexed with Fe2+, but it is either non-existent or very weak at best for

113

2+ 2+ the DiMC Fe complex, indicating the formation of Bip2/Fe is not preferred for 23. The main- absorption peak at 382 nm for both components were plotted against the metal/Bip ratio, as shown

2+ in Figure 4.6b. Once the metal/Bip ratio reaches 0.5, the Bip forms stable Bip2/Fe complex and the intensity remains constant upon further addition of metal ion. Contrary to this is the UV

2+ titration of DiMC 23 which shows that it does not form a stable Bip2/Fe complex and in fact will

2+ form a 23:Fe 2 complex. This result clearly shows that macrocycle 23 does not prefer the formation of the unwanted 1:1 self-binding complex, and thus will favor the assembly of the

2+ desired Zn-alt-MSP 14n+1·23n+1·Fe 2n+1.

2+ 2+ Figure 4.6. Comparison of (a) the UV-Vis spectra of 23, 24, 23·Fe and 242·Fe , and (b) the plot of absorption of the Fe2+ titration with 23 and 24 (at λ = 382 nm) versus Fe2+/Bip ratio.

The self-assembly of the Zn-alt-MSP was carried out through careful titration of the metal ions in to a 1:1 mixture of 14 and 23 followed by NMR (Figure 4.7). Compared with monomer 14, the chemical shift of HA, HB and HC of DiMC 23 are shifted slightly up-field, possibly due to the conformational strain of the macrocyclic structure. Therefore, it is easily to confirm the 1:1 molar ratio of 14 and 23 by monitoring the integration of HA. After that, the solution of Zn(NTf2)2 was

114

added into the mixture. As we see in Figure 4.7c~e, almost all the aromatic proton peaks show a dramatic change in chemical shift, where the shift of HA, HB and HC are especially obvious. After the addition of 2 eq. of Zn2+, a spectrum of Zn-alt-MSP is obtained. All peaks are broadened on account of the increase in molecular weight. The formation of Zn-alt-MSP is also confirmed by diffusion-ordered NMR spectroscopy (DOSY, see Figure S4.2, supporting information). The diffusion coefficient (D) of Zn-alt-MSP is 3.54×10−7 cm2/s which is much smaller than that of monomer 14 (D=1.21×10−6 cm2/s), and the hydrodynamic radius of Zn-alt-MSP is as 3.4 times as that of the monomer 14.

Figure 4.7 Self-assembly of the Zn-alt-MSP by NMR titration. 1 Aromatic region of H-NMR (600 MHz) of (a) DiMC 23 (in CDCl3); (b) mixture of 23 and monomer 14 at 1:1 molar ratio (in CDCl3); and then, adding into the mixture with (c) 1 eq. or (d) 1.5 eq. of Zn(NTf2)2; and finally the self-assembled Zn-alt-MSP after the addition of 2 eq. of Zn(NTf2)2. The metal salt was dissolved in CD3CN.

115

2.2.2 Synthesis of Polycatenane

Polycatenate 22 was synthesized via the ring-closing metathesis reaction of Zn-alt-MSP using the Hoveyda-Grubbs 2nd generation catalyst in DCM (Figure 4.8) heated to reflux. The reflux of DCM helps the removal of the condensate 2-butene and increases the reaction kinetics. The reaction progress could be monitored by the disappearance of the terminal methyl group of 14 (see

Figure S4.3, supporting information). All the resulting metal ion complexes in the reaction mixture can be readily demetalated either by the addition of a stronger competitive ligand (e.g. ethylenediamine) or by washing the DCM solution of the metal ion complexes with an aqueous solution of NaOH to convert zinc ions into sodium zincate. This process will also include

2+ converting the target polycatenate 22·Zn 2n+1 to the polycatenane 22 (Figure 4.8).

Figure 4.8. Synthesis of poly[n]catenane 22.

1 The region of the H-NMR corresponding to the HA, HB and HC protons on the Bip moiety is shown in Figure 4.9. It compares the possible non-interlocked byproducts—the linear polymerized monomer 17 and cyclized monomer 18, a mixture of both starting materials, monomer 14 and DiMC 23, and the resulting crude product. Comparing the NMRs of the three samples shows that there are significant changes in the chemical shifts of these aromatic protons.

The region highlighted in yellow is of particular interest and significance as this chemical shift region has previously been shown to correspond to the HA, HB and HC in Bip-containing small

116

molecule catenanes (see Chapter 2, Figure 2.8) and as such is a strong indicator that interlocked species are obtained from this reaction. It is important to note, however, that in the crude product peaks are also observed at ca. 8.30 and 8.35 ppm, which correspond to the presence of a small amount of DiMC 23 and linear polymer 17, indicated by red and blue dashed line, respectively.

Dialysis of the crude product against chloroform was carried out in order to remove any low molecular weight species. Both the polycatenane peaks (δ: 8.11∼8.27 and 7.83∼7.925) and the linear polymer 17 (δ = 8.35, 8.07 and 8.02, blue dashed line) remain, consistent with these peaks being related to relatively large molecular-weight polymers. However, all peaks that correspond to DiMC (red dashed line) disappeared, confirming that they were indeed related to the free DiMC small molecules, which are not mechanically bonded with the polymeric species. The chemical shift change of HB is especially diagnostic here and does not overlap with any other proton peaks, and as such can be a good indicator of catenane formation. Thus, although the components have the same chemical structure of their Bip moieties, the chemical shift of the HB peaks in the polycatenanes are shifted upfield compared to non-catenated analogues (14, 17, 18 or 23), presumably due to the fact that the interlocked aromatic Bip moieties are shielding each other.

Based on combining all our data together (including the work in Chapter 2) we can define the non- catenane chemical shift region as 8.270~8.370 ppm and catenane chemical shift region as

8.090~8.270 ppm (yellow region in Figure 4.9).

117

Figure 4.9. 1H-NMR of the polycatenane reaction product along with starting materials and possible byproducts. 1 H-NMR (partial, 600 MHz, CDCl3) comparison of the linear polymerized monomer 17, the cyclized monomer 18, the starting 1:1 mixture of monomer/DiMC, the crude polycatenane product 22 and dialyzed polycatenane 22.

2.2.3 High-efficiency Purification of Polycatenane

Although DiMC can be readily removed by dialysis, the removal of the polymeric byproduct 17 from the polycatenane is difficult on account of their similar polarity and solubility.

Traditional polymer purification techniques, such as precipitation, dialysis and chromatography, show poor separation of byproduct 17 and polycatenane 22. However, their actual polymer

118

architectures are distinct (linear vs. interlocked) and so investigations were undertaken to see if this architectural difference could be exploited to develop a high-efficiency chromatography-free purification of polycatenane to remove the non-polycatenane byproducts.

The interlocked nature of a catenane is known to have a more stable ligand coordination with a metal ion.2 Therefore, upon gradual addition of Zn2+ ion to the mixture, the Bip moieties in the polycatenane will be bound prior to any of the Bip moieties in the none interlocked byproducts, as suggested by Figure 4.10. Therefore, if we careful control the metal equivalency (e.g. 60~70%) the polycatenanes will be at least partially metallated while byproducts (and possibly some oligomers) will remain unmetallated. Metalation dramatically changes the solubility of these compounds to the extent that simple trituration of the mixture with chloroform/hexane selectively precipitates the metalated poly[n]catenanes and removes the unmetallated byproducts/oligomers.

2+ The obtained purified polycatenate 22·Zn 2n+1 can then be readily converted to a purified polycatenane 22 via standard demetallation conditions.

Figure 4.10. Summary of an efficient chromatography-free purification of polycatenane 22.

119

The purification process was evaluated via NMR. The crude product contains ~15% byproducts based on the integration of HB peak (Figure 4.11a). After the addition of Zn(NTf2)2 (to reach 60% metallation), all of the byproducts remains unmetallated as the integration value of byproduct over that of all HB peaks (sum of metallated and unmetallated) does not change (Figure

4.11b). As we see in Figure 4.11c, both byproducts are successfully removed with no observable leftover, and the filtrate (Figure 4.11d) contains both byproducts and a small amount of catenanes, which are presumably predominately oligomers. The amount of byproduct is usually ca. 10% (for each of 17 and 23), and the overall yield of purified polycatenane is as high as ca. 75%. The experimental detail of this purification protocol is available in the supporting information.

Figure 4.11. NMR study on the polycatenane 4-1 purification process. 1 Partial H-NMR (600 MHz) of: (a) crude polycatenane containing by products (CDCl3); (b) after 60% metalation by Zn(NTf2)2, the byproduct remains non-metalated (CDCl3/MeCN-d3, 3:1 v/v); (c) purified polycatenane contains no byproduct (CDCl3); (d) the filtrate contains both byproducts and a small amount of catenanes (CDCl3). 120

2.3 Proof of Polycatenane Structure

2.3.1 Proof of Mechanical Interlocking Structure

While the chemical shifts of the aromatic protons in the 1H NMR are in line with 22 consisting of catenated structures, more definitive evidence for existence of such interlocked species in the structure was sought. There are two key methods that has been used previously to demonstrate the interlocked nature of small molecule catenanes:3

(1) via the presence of DiMC 23 moiety (the red ring) in the product. The only way for 23 to be incorporated in the product is if it is interlocked with the blue component derived from 18. In addition, if the polymer architecture is the targeted alternating catenane structure 22 then it should contain ca. 50% red rings which have to be interlocked by 50% blue ring (cyclized monomer 18 moiety).

(2) via the nuclear Overhauser effect spectroscopy (NOESY). Rings that are interlocked will be in close proximity to each other and as such inter-ring nuclear Overhauser effect (NOE) cross peaks will be observed in the catenane that are not observed in a simple mixture of the individual components.

Of course NMR can be used to for both methods however in order to do this it is important to be able to assign the peaks in the 1H NMR. To do this a range of 2D NMR techniques, including homonuclear correlation spectroscopy (COSY), heteronuclear multiple quantum coherence spectroscopy (HMQC) and heteronuclear multiple bond coherence spectroscopy (HMBC), were used to assign all the peaks of the polycatenane, as shown in Figures 12, 13 and 14, respectively.

1 1 3 H- H COSY allows us to correlate two protons with JH-H coupling, i.e. the two hydrogen atoms connected to two adjacent carbons. Starting from some easy-finding peaks with unique chemical shift (e.g. A, P and U, Figure 4.12), we are able to assign all the peaks on the pyridine ring (A, B), 121

4 the reactive chain-end moiety (M~P) and the butyl side chain (U~X). The long-distance JH-H coupling is usually very weak and not observable by COSY. However, in some scenarios such as

4 coupling across π-systems, JH-H may be observed albeit as a weaker peak, but can provide important information. For example, the carbon CL is between an oxygen and a tertiary carbon so

3 there is no JH-H coupling for HL. However, a weak cross peak was observed corresponding to its

4 JH-H between HL and HI/HJ, which help the assign of proton HI and HJ, and then HH.

Due to the complexity of the spectrum and large quantity of tertiary carbons COSY is not sufficient for the assignment of all the peaks and as a result HMQC and HMBC were also carried

1 out. HMQC measures the JH-C coupling which correlates the proton with the carbon it is directly

2 connected to, while the HMBC measures JH-C that correlates the proton with carbons adjacent to the carbon the proton is bonded to. It is important to note that all tertiary carbon peaks do not show

HMQC signals, which provides important additional information on HMBC. Combining HMQC and HMBC we are able to assigning all the protons and carbons. For example, as shown in Figure

4.14, starting from HU we are able to find tertiary Cβ and Cδ They can be differentiated as Cβ will correlate to HB but Cδ does not. Cδ (red and blue are overlapped) can help us locate proton HD Hd,

HE and He (overlapped). Starting from HL, we can locate Cγ (blue) and then HG and HF. From HM and Hh we can locate carbon Cγ’(blue) and Cγ (red) respectively, and then HG’ and Hg. HF , HF’ and

Hf can also be assigned with the help of COSY. The overlapped HI and HJ show two cross peaks from HMQC (Figure 4.13) corresponding to carbon CI and CJ, but they show four cross peaks on

HMBC with the two new ones being Cφ and Cη, which can be differentiated by the fact that Cφ will correlated to HK at the upfield (1~2ppm) but Cη won’t. Based on this strategy, other peaks can also be assigned. 122

Figure 4.12. Gradient-selected 1H-1H COSY of polycatenane 22.

Measured in CDCl3 at 25°C, 600 MHz.

123

Figure 4.13. Gradient-selected 1H-13C HMQC of polycatenane 22.

Measured in CDCl3 at 25°C, 600 MHz.

124

Figure 4.14. Gradient-selected 1H-13C HMBC of polycatenane 22.

Measured in CDCl3 at 25°C, 600 MHz.

125

After the peak assignment, the non-overlapped peaks i, j k from the red ring and P, L from the blue ring was selected (Figure 4.15) to study the ratio of blue/red rings in the polycatenane by the equation:

()/ �/ = = 0.91, where Ix is the NMR peak area integration of peak x. ()/

The ratio indicates that the polycatenane 22 contains 47.6% red rings and 52.4% blue rings.

The only way to have so much red ring in the structure is for it to be interlocked with the blue component. The fact that the ratio is close to 50:50 of the two components is also encouraging for the formation of an alternating poly[n]catenane. It is worth noting that it is possible to have blue- blue interlocking rings but not red-red and this would explain to the presence of slightly more blue than red rings.

Figure 4.15. NMR study on the ratio of red/blue rings in polycatenane 22. 126

We have previously shown that inter-ring NOE cross peaks can be observed in the Bip- derivative [3]catenanes at low temperatures.3 Therefore, the 1H-1H NOESY experiment of 22 was carried out at -17 °C in CDCl3 at a concentration of 1% w/v (10 mg/mL). While multiple positive

NOE cross-peaks are observed in the spectrum, the key inter-ring cross-peaks are labeled in Figure

4.16. Based on the dipole-dipole spin-lattice relaxation theory, the NOE intensity is proportional to r−6, where r is the distance between the nuclei.4 Therefore, NOE is highly sensitive to the distance and an NOE is generally not observable for two atoms farther than 0.5 nm. Thus the only way for 4-1 in dilute (1% w/v) CDCl3 solution to exhibit inter-ring NOEs is for the rings to be interlocked. To further back this conclusion up the NOESY of the 1:1 mixture (based on the molar ratio of Bip moiety) of 17 and 23 and also the 1:1 (molar ratio) mixture of 17 and 18 at the same

Bip concentration was measured (Figure 4.17 and Figure 4.18). Critically, as expected, no inter- ring cross-peaks between the two components is observed in these two control mixtures.

127

Figure 4.16. 1H-1H NOESY of polycatenane 22. Selected NOE cross-peaks between monomer rings and DiMC rings are labeled. Cross-peaks in red color indicates a positive phase NOE signal. Measurement condition: 600 MHz, −17 °C, CDCl3.

128

Figure 4.17. 1H-1H NOESY of the 1:1 mixture of 23 and 17.

No cross-peaks between two components are observed. Measurement condition: 600 MHz, −17 °C, CDCl3.

129

Figure 4.18. 1H-1H NOESY of the 1:1 mixture of 23 and 18.

No cross-peaks between two components are observed. Measurement condition: 600 MHz, −17 °C, CDCl3.

130

2.3.2 Determination of large molecular weight

While these NMR studies have proved we have an interlocked architecture, the size of these molecules have not been determined. Gel permeation chromatography (GPC) is the most widely used method to analyze the molecular weight of polymers. As a consequence of the unique structure of polycatenanes, traditional GPC based on column calibration and working curves

(including the “universal calibration”) will not provide an accurate measurement of the interlocked polymer. However, when coupling GPC with multi-angle laser light scattering (MALLS) technique and refractive index (RI) detector, the GPC-MALLS system can measure the accurate absolute molecular weight of any polymer when the dn/dc value of the polymer is available. Both weight average and numeric average molecular weight (Mw and Mn, respectively) can be measured directly from MALLS regardless of column calibration. Therefore, all molecular weight measurement for polycatenane is based on GPC-RI-MALLS system with 18 scattering angles. The dn/dc value for the polycatenane 22 was determined to be 0.185 mL/g under the same condition of GPC-MALLS measurement.

Figure 4.19a and Table 4.1 shows the GPC results of polycatenane 22. The average Mn is

21.4 kDa, indicating an average degree of polymerization (DP) of 13.9. The polydispersity index

(Ð) is 1.437, which is smaller than the theoretically expected value (2.0) and is presumably due to the removal of oligomeric catenanes during the purification process. The RI plot shows a main peak at 8.75 min that including a shoulder 8.2 min, and a small peak at 9.4 min. Peak deconvolution fitting shows that the GPC spectrum follows a trimodel distribution that contains three Gaussian peaks. Based on the fact that the polycatenane sample is purified and contains no other polymers, all three peaks belong to polycatenanes while each Gaussian peak corresponds to an individual polydispersity distribution. Therefore, the trimodel distribution result implies that the polycatenane 131

sample contains probably three (and maybe more) different architectures (e.g. linear, branched, etc.). The fitted peak areas of ①, ② and ③ are 29%, 61% and 10%, respectively, which might imply the relative amount of each architecture.

2.3.3 Fractionation of polycatenane

For better analysis, the polycatenane was fractionated to separate these different architectures and also access samples with a smaller Ð. Polycatenane 22 was passed through a hand-loaded size-exclusion chromatography (SEC) column using 2:1 v/v chloroform/ethanol as mobile phase. During SEC separation, polycatenane molecules were fractionated by their hydrodynamic size with the larger sized molecules eluted earlier. Polycatenane 22 was fractionated into four fractions, 22a (largest), 22b (large), 22c (small), and 22d (smallest). Each fraction was further studied by GPC and NMR for molecular weight and architecture studies. As shown in

Figure 4.19b and Table 4.1, the GPC results show successful size-based fractionations with very narrow Ð. The average Mn of 22a is 85.6 kDa (Ð ≈ 1.113), which is consistent an average of degree of polymerization (DP) of the poly[n]catenane of 55. The smallest molecular weight fraction 22d has an average Mn of 12.16 kDa (Ð ≈ 1.115), which represents an average of an [8]catenane. 2*

Therefore, we can reasonably assume that 22a and 22b predominately contiain the same architecture A, 22c consists mostly to architecture B, while 22d contains mainly architecture B and C.

132

Figure 4.19 GPC spectra of (a) the purified polycatenane 22 and the peak deconvolution fitting and (b) shown with its isolated fractions 22a through 22d.

133

Table 4.1 Absolute Molecular Weight of 4-1a through 4-1d.

Sample Relative Mass Mw (kDa) Mn (kDa) Ð

22 (unfractionated) 100% 30.8 21.4 1.437

22a 2% 95.3 85.6 1.113

22b 22% 45.5 38.5 1.182

22c 50% 19.73 17.35 1.138

22d 26% 13.56 12.16 1.115

Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry is a powerful tool for polymer analysis. However, this technique is usually limited to polymers with narrow polydispersity (e.g. Ð <1.2) due to the mass discrimination effect of polydispersed samples.

Therefore, we performed MALDI-TOF fractionated polycatenanes with narrow polydispersity

(1.1 <Ð < 1.2), as shown in Figure 2.20 for polycatenane 22b. Polycatenanes with molecular weight up to 39 kDa shows clearly defined peaks, corresponding to [2]~[25]catenanes, confirming the successful synthesis of polycatenane with large molecular weight. The distance between two adjacent peaks falls between 1526 and 1575, which matches the mass of both rings (1530.8 and

1556.8 for red and blue, respectively) within reasonable experimental error. However, different from a Poisson mass distribution of traditional polydispersed polymers, the intensity of polycatenane decreases with increasing molecular weight. We propose the following reason for this phenomenon, as shown in Figure 4.21. The larger a polycatenanes is, the higher chance it fragments. Therefore, a polycatenane molecule will break into two smaller polycatenanes, which may fragment again and again until everything becomes oligomers, especially [2] and [3]catenanes.

These oligomeric catenane fragments built up high intensity peaks at the low molecular weight

134

range. Therefore, the observed MALDI-TOF peaks of polycatenane is the combination of polycatenane quasimolecular ions and fragments, which is expected to be a decay in intensity with increasing molecular weight (Figure 4.21a). On the contrary, the fragmentation of a non- interlocked polymer will result in shorter chains with random length and large polydispersity. Due to the polydispersity, these fragments could hardly build up a high-intensity peak and thus not easily observable. Therefore, the peaks we observed from non-interlocked polymers are expected to be only their polymer quasimolecular ions with Poisson distribution (Figure 4.21b).

Figure 4.20 MALDI-TOF of polycatenane 22b. Matrix: ditranol, no salt added, measured in linear positive mode.

135

Figure 4.21. (a) The fragments of polycatenanes are observed, which results in non-Poisson distribution; and (b) the fragments of non-interlocked polymers are hardly observed, and the polymer quasimolecular ion follows Poisson distribution.

2.4 Polycatenane Architecture Determination

2.4.1 The architecture of polycatenane

There are three broad classes of main-chain polycatenanes architectures that can be imagined based on the synthesis, namely linear, branched or cyclic. Figure 4.22 shows three

[12]catenanes with each architecture. To quantitatively characterize the architecture, we can define the number of chain-ends per molecule (NC): A linear polycatenane will have two chain- ends in each polymer chain and thus NC(linear)=2, while a cyclic poly[n]catenane will have 136

NC(cyclic)=0 and a branched poly[n]catenane will have NC(branched)>2. It is worthwhile to note that in the Zn-alt-MSP templated synthesis, the branched polycatenane comes from the formation of larger blue rings (>cyclic monomer) and as such NC(branched) can only be 4, 6, 8, 10… and cannot be less than 4. For the mixture of different architectures, the average NC will be a number in between these ideals, e.g. a mixture of linear and cyclic species will have 0

Figure 4.22 [12]Catenanes with linear (a), branched (b) or cyclic (c) architectures.

It is difficult to directly count the exact number of chain-ends of each molecule. However, if we already know the molecular weight (and thus the DP) of the polycatenane, we can calculate

NC from the ratio of the chain-end rings to the total rings (RC/T). Take the three [12]catenanes in

Figure 4.22 as example, their calculated RC/T is shown in Table 4.2. Here the NC will be equal to the degree of polymerization times this ratio (NC = DP × RC/T). The RC/T can be directly measured by NMR peak area integration as long as the chain-end peaks do not overlapped with other peaks and can be distinguished. As previous discussed in Figure 4.9, the doublet peak corresponding to the proton HB of the Bip moiety is not overlapped with any other peak and appears very sensitive to its environment. Its chemical shift can vary from 8.10 to 8.37 depending on whether the Bip moiety is located in an acyclic, a cyclic or a catenated molecule. Therefore, it is a potential good candidate for chain-end analysis. If we can measure the peak area for a HB corresponding to the 137

chain-end (IChain-end) and the total peak area of that proton (ITotal), we will have RC/T = 2×IChain- end/ITotal, where the 2 relates to the fact that each ring has two Bip moieties but only one of them is located on the chain-end. Therefore, we can calculate NC based on the following equation:

2×��×� � = �

Table 4.2 Ratio of chain-end rings to total rings for polycatenanes shown in Figure 4.22.

Architecture NC DP RC/T

(a) Linear 2 12 1:6

(b) 4-Armed 4 12 1:3

2.4.2 Determination of chain-end peak from NMR

Unlike conventional polymers where the chain ends are usually chemically distinct all the macrocycles in a polycatenane chain, either the blue or the red ones, are chemically identical. As such, it is difficult to functionalize the chain-end by a chemical reaction or a chemical-shift agent.

Therefore, to confirm the chemical shift of the chain-end peak for the HB proton, we synthesized some oligomeric catenanes for comparison, including the [3]catenane 25, along with the oligomeric catenane 22Oligo and 15Oligo (shown in Figure 4.23), which should have stronger intensity at the chain-end peak. To obtain these oligomers, the key was to vary the stoichiometry of the three components during the assembly of the MSP. For example, to access the [3]catenane

25 the three components 14, 23 and metal ion were used in a ratio of 1:2:2 (Zn2+), while to access the oligomeric catenane 22Oligo the components were used in a ratio of 2:2:3 (Fe2+). For the synthesis of 15Oligo the ratio of 14:Fe2+ was 5:4. It is worthwhile noting that for the oligomer we used Fe2+ ions to template the assembly of the MSP as when Zn2+ ions were used the yield of the 138

oligomer was very low. Like Zn2+ the Fe2+ ions also form a 2:1 Bip:metal ion complex but it exhibits a much larger binding constant to the ligand than Zn2+. On account of this the reaction was also carried out at a much lower concentration 0.25mM vs 2.5 mM used for the Zn-templated reaction, which we hoped would limit any inter-chain reactions. When using this stronger binding template metal ion followed by the catenane purification protocol (discussed in section 2.2.3), the isolated yield of the 22Oligo was much higher (55%). For more details on the synthetic procedures used see supporting information.

Figure 4.23. Synthesis of oligomeric catenanes by controlling the stoichiometry of the starting materials. Note: while Zn2+ can also be used for the synthesis of 22oligomer and 15oligomer the yield is very low and using Fe2+ as the templating metal ion and a lower concentration dramatically improves the yield.

1 The H-NMR HB peaks of these oligomers and polycatenanes are compared in Figure 4.24.

The first oligomer candidate is the [3]catenane 25 which shows a strong doublet at δ 8.25 ppm, another doublet at 8.21 ppm and a multiplet at 8.17 ppm. Corresponding to the three groups of

139

peaks of 25, three catenane peak regions can be defined as (I) (II) and (III) for the chemical shift of 8.235~8.270, 8.195~8.235, and 8.150~8.195 ppm, respectively. In addition, for oligomeric 15 and 22, there are some catenane peaks that extend to the chemical shift of 8.150~8.090, and we define this region as (IV). From this data, we can reasonably assume region (I) corresponds to the chain-end as it is the only region that meets all the requirements: present in all samples, and shows stronger intensity in oligomers while weaker intensity in polycatenanes. Moreover, region (I) has a chemical shift that is closest to the non-catenane region. This makes sense the chemical shift of the HB peak in the catenane structure is due to the topological structure, and it can be expected that the effect will be much less prominent for the chain-ends compared to those in the center of the polymer. As such, the chemical shift of HB of chain-end should be shifted less upfield than the non-chain-end HB peaks.

140

1 Figure 4.24. The H NMR comparison of HB peaks in polycatenane and oligomeric catenanes.

To further support this assumption, the NMR relaxation studies on the polycatenane samples was also used to identify the chain-ends.5–7 For a synthetic polymer in solution, the spin- lattice relaxation time (T1) generally increases with increasing mobility; therefore, the chain-end peaks generally have a larger T1 than the interior backbone peaks for the same nuclei due to their 141

5,6 increased mobility. In addition, the spin-spin relaxation time (T2) is also related to molecular motion, which means that polymers usually show a much smaller T2 than small molecules, and

7 chains ends will show a larger T2 than the internal polymer chain on account of faster motion.

Thus the measurement of T1 and T2 of the different HB peaks in these samples can be used to help us determining which one(s) correspond to the chain-end peak(s). The NMR relaxation study at room temperature (25 °C) shows that the T1 and T2 of peaks in of region (I) are both larger than that of region (II) and (III) (see Figure S4.4 and Table S4.1, supporting information) However, a note of caution is required here; the polycatenane chain is different from traditional polymer chains as there will be topological mobility of each macrocycle rings and thus it can be expected that even internal macrocycles will have more mobility than is usually observed for repeat units in a traditional all covalent polymer backbone. To limit this effect as much as we can T1 and T2 studies were also carried out at very low temperatures (−25 °C) to reduce the topological mobility.3

The T1 of the polycatenane fraction 22b was measured by the inversion-recovery method, shown in Figure 4.25a. The peaks in region (II) and (III) show a null signal after a waiting time

τ = 1.03s and 1.00s, respectively. Simultaneously, region (I) shows a strong negative peak at 1.03s which indicates it has not fully relaxed at this time point. As calculated in Table 4.3, the T1 of the peak in region (I) is much longer than that in other regions. In the T2 measurement (using the spin- echo Car-Purcell-Meiboom-Gill (CPMG) method8), the doublet in region (I) shows a much slower decay of the echo signal intensity than those in regions (II) and (III). At τ = 0.550s (Figure 4.25b) a total disappearance of the peaks in regions (II) and (III) occurs but not in region (I). This indicates that the peak in region (I) has a larger T2 value. As summarized in Table 4.3, the doublet in region

(I) shows both obviously longer T1 and T2 and that further supports the chain-end assignment for region (I). 142

Figure 4.25. The (a) inversion-recovery spectra and (b) spin-echo CPMG spectra of the polycatenane 22b with different τ. 1 H-NMR showing the HB peak. Measured at −25 °C in CDCl3, 600MHz. The 90° pulse width for all HB protons in three regions are measured to be the same value, which is used for both T1 and T2 measurements.

Table 4.3 T1 and T2 values of the HB peaks of polycatenane 22b at different region at −25 °C. * ** ** Region δ Range/ppm T1/s T2/ms

(I) 8.190 ~ 8.230 1.66 80.3

(II) 8.145 ~ 8.190 1.49 56.9

(III) 8.100 ~ 8.145 1.44 68.6

*: Peaks shift slightly upfield at reduced temperature and the regions are adjusted accordingly. **: Measured in CDCl3 (containing 0.3 v/v% TMS), 600 MHz.

In Figure 4.24, the chain-end peak of 22Oligo and 15 Oligo do not appear as a simple doublet

(as they do in the other samples), and its chemical shift does goes just beyond what we have labelled region (I). This might due to the difference between a blue chain-end and a red chain-end.

Therefore, the chain-end peak of 22Oligo was deconvoluted and compared with other catenane

143

samples (Figure 4.26). Deconvolution result shows two doublet peaks ( ① / ③ and ② / ④ , respectively) of the 22Oligo chain-end peaks. The coupling constants (J) of both doublet ①/③ and

3 ②/④ are calculated to be ca. 7.8 Hz, a normal value for JH-H coupling in the pyridine moiety. The chemical shifts of the doublet splits ②/④ line up with that of the blue-blue oligomeric catenane, shown in blue dotted line in Figure 4.13, while ①/③ matches perfectly that of the [3]catenane 25 and the polycatenane 22b. Interestingly, in Figure 4.17, all polycatenane samples shows only one chain-end doublets that correspond to the red chain-end and little-to-no blue chain-ends are observed. A possible explanation for this is shown in Figure 4.27. Under an ideal reaction scenario, the polycatenane should contain 1:1 ratio of red/blue chain-ends. However, as we have seen above a major side-reaction converts ~10% of the monomer 14 into byproduct 17 either by reacting with an adjacent blue monomer within the Zn-alt-MSP or by the uncomplexed monomer reacting with itself. Either process will result in only red ring chain ends, after removal of the metal ions. If incomplete ring closing of 14 does occur then red ring chain-ends will also be selectively formed.

It is also very possible to have blue-blue interlocking structure (highlighted light blue in Figure

4.27). The rearrangement of one monomer moiety per MSP may results in no blue chain-ends.

Therefore, even if blue chain-ends will have as much as 25% of their peak area out of the region assigned as (I), the very small amount of observed blue chain-end in the polycatenanes means it is reasonable to neglect that factor and use the peak area in region (I) as a good approximation of the total chain-end integration (IChain-end) for the NC calculation.

144

1 Oligo Figure 4.26. H NMR comparison of deconvoluted HB peak in catenane 22 compared with other oligomeric catenanes and polycatenane.

Showing the HB peak. Red or blue dashed line shows the red-ring or blue-ring chain-end peals, respectively. Inset spectrum shows the zoomed in peak fitting of the chain-end in oligomeric catenane with both red and blue chain-ends (black line: NMR spectrum; purple line: fitting curve, blue line: fitting peaks; and red line: the fitting error). All spectra are measured at 25 °C in CDCl3 (600MHz) and calibrated by TMS.

145

Figure 4.27 Preferred formation of red chain-ends.

In addition to the chain-end region (I), region (IV) is also very interesting as not all samples show peaks in this area. In fact, the two samples (22Oligo and 15Oligo) that show the large peaks in this area are the ones that use the stronger binding metal ion (Fe2+) as the templating metal ion and much more dilute reaction conditions. During the purification process of the Catenane 22Oligo, after the first purification cycle we isolated 55% 22Oligo which has a ca. 90% purity (contains ca. 10% non-interlocked species). When we did the 2nd purification cycle, we surprisingly isolated a fraction of oligocatenane (ca. 65% by weight) as the major product of the 22Oligo synthesis reaction

(with 36% overall yield), which shows multiple peaks in region (IV) but has barely no peak intensity in the chain-end region (I), as shown in Figure 4.28. The GPC-MALLS study on this fraction shows very low molecular weight (10 kDa, DP=6.47) which excludes the possibility of very high molecular weight being the reason for the low chain end peak intensity. As such this data is consistent with this oligocatenane fraction containing predominantly a cyclic polycatenane architecture. This would make sense as carrying the reaction out under dilute conditions (0.25 mM) with the strong binding Fe2+ metal ion both favor cyclic over polymeric supramolecular species.9,10

146

As such it is reasonable to assume that region (IV) is diagnostic for cyclic polycatenanes or oligocatenanes. This fraction of the 22Oligo reaction product is therefore labelled as 22Cylic-oligo for clarification.

Figure 4.28. The isolation of 22Cyclic-oligo and its NMR comparison with 22Oligo and the residue catenane after purification.

It is worth noting that in addition to 22Cyclic-oligo, polycatenane 22c and 22d also show some peaks in region (IV) (Figure 4.29, highlighted in green) which indicate that those two fractions presumably also contain cyclic polycatenanes. If they are mixture of linear and cyclic architecture

(with little or no branched), we would expect NC <2 for these two fractions.

147

Figure 4.29 1H-NMR (partial) comparison of polycatenanes and cyclic oligomeric catenane.

Showing HB peak only. Peaks for polycatenane HB in region 4 are highlighted. All spectra are measured at 25 °C in CDCl3 (600MHz) and calibrated by TMS.

2.4.3 Calculation of NC

After assigning the chain-end peak in the NMR, we are able to calculate the NC based on the molecular weight measured from GPC-MALLS and the NMR peak area integration based on the following equation, and the results was listed in Table 4.4.

2×��×� � = �

148

Table 4.4 NC calculation of polycatenanes compared with [3]catenane and oligocatenane.

Sample Mn /kDa (MALLS) DP (MALLS) IChain-end ITotal NC

Polycatenane 22 21.4 13.85 1 11.59 2.39 (Unfractionated)

Polycatenane 22a 85.6 55.40 1 12.75 8.69

Polycatenane 22b 38.5 24.92 1 10.89 4.58

Polycatenane 22c 17.35 11.23 1 13.37 1.68

Polycatenane 22d 12.16 7.87 1 12.95 1.22

[3]catenane 25 N/A 3* 1 3.13 1.92

Catenane 22Cyclic oligo 10 6.47 1 66.96 0.19

*: Calculated based on theory, not from MALLS.

There are a number of key points that come out of this analysis. For example, 25 which was targeted to be a low molecular weight [3]catenane on average has an NC close to 2 as to be expected. In addition, the sample 22Cyclic oligo that had peaks in region IV has an Nc close to 0 which is consistent with it containing mainly cyclic polycatenanes as suggested above. The Nc values of the polycatenane fraction of 22a and 22b are > 4 and are consistent with the presence of branched structures. These samples also show the largest molecular weight, which makes sense if the branched as this should occur as a result of inter-chain reactions of linear polycatenanes.

Both polycatenane 22c and 22d have NC values between 1 and 2 and are likely the mixture of linear and cyclic architectures, although we cannot exclude the possibility of containing branched ones, unless we deconvolute the contribution of linear and cyclic species to the NC calculation. Assuming that 22Cyclic oligo is indeed predominantly cyclic catenanes and that the peak 149

region (IV) are unique for the HB peaks in such cyclic catenanes, then the relative integration of

Cyclic oligo the HB peaks in region (IV) to the total integration area of all the HB peaks in the 22 shows that ca. 50% of Hb peaks falls into region (IV) (NMR see Figure S4.5, supporting information).

Therefore, if we assume all cyclic polycatenanes have 50% peak area in region (IV), for a polycatenane sample containing cyclic ones the overall peak area contributed by cyclic ones will be two times of the peak area in region (IV). Therefore, we can modify the above equation to:

2×��×� �, = � − 2×�()

The result of this calculation is listed in Table 4.5. After excluding the contribution of cyclic polycatenane from the Itotal, the NC of both 22c and 22d are around 2, which would suggest that both these samples are mixtures of linear and cyclic architectures with little-to-no branched polycatenane, (otherwise NC,excluding cyclic would be greater than 2) . We are also able to calculated the ratio of cyclic architectures, which is 13% and 29% in 22c and 22d, respectively.

Table 4.5 Calculation of NC excluding the contribution of cyclic polycatenane

NC Mn /kDa DP Sample IChain-end ITotal IRegion(IV) Cyclic % (excluding (MALLS) (MALLS) cyclic)

Polycatenane 17.35 11.23 1 13.37 1.10 13% 1.98 22c

Polycatenane 12.16 7.87 1 12.95 2.33 29% 1.90 22d

150

In summary, the branched largest polycatenane fraction (22a) has an average Mn of up to

85.6k, corresponding to a branched [55]catenane. The largest linear polycatenane fraction (22c) contains 87% linear architecture (with 13% cyclic) can be isolated with Mn up to 17.4 kDa, corresponding to a mostly linear [11]catenane.

2.4.4 The Effect of reaction concentration on the architecture of the polycatenanes

The NC calculation result shows that 22a and 22b are branched, 22c and 22d are the mixture of linear and cyclic. Looking at Figure 4.29 we can we can see for branched 22a and 22b, the peak intensity in region (II) is much stronger than that in region (III); for the mostly linear 22c, however, the intensity in (II) and (III) are similar. For the mixture of linear and cyclic 22d, region (III) is stronger than (II) and multiple peaks will be clearly observed in Region (IV). In summary, region

(I) is chain-end, branched architecture results in higher intensity in (II) and cyclic architecture shows peaks in (IV).

To explore the relationship between the NMR and the polycatenane architecture a concentration study was undertaken. Thus the metathesis reaction of Zn-alt-MSP was performed at different reaction concentration (0.25 mM, 2.5 mM and 10 mM, w.r.t. the 14 moiety) using Zn2+

1 as the templating metal ion. Figure 4.30 show the partial H NMR of the HB region that corresponds to interlocked polymers. It can be expected that at the higher reaction concentration the reaction between two MSP chains will be much easier and therefore branching and even crosslinking will be enhanced, while at the lower concentration cyclic polycatenanes will be preferred (along with non-interlocked side products) and branched species should be inhibited. As such, based on our prior assignments the polycatenane synthesized at 0.25 mM should have high intensity in region

(IV) (cyclic) and low intensity in region (II) (no branching), while high intensity in region (II) but 151

no intensity in region (IV) for the polycatenane synthesized at 10 mM. As can be seen the results in Figure 4.30 perfectly meets these predictions and strengthened our polycatenane architecture

NMR analysis. Table 4.6 gives a summary of the rule of thumb assignment of the polymer architectures based on the 1H NMR.

Figure 4.30 Polycatenanes synthesized from Zn-alt-MSP under different metathesis concentration.

Showing interlocked HB peak only. All spectra are measured at 25 °C in CDCl3 (600MHz) and calibrated by TMS.

Table 4.6 Proposed relationship of 1H NMR and polycatenane architecture. Summary of peak assignments

Region I: (8.235~8.275) Chain end

Region II: (8.195~8.235) Branched mainly (and linear)

Region III: (8.150~8.195) Linear (branched and cyclic)

Region IV: (8.060~8.150) Cyclic

152

2.5 Metallo Response of Polycatenane

Bip-containing polycatenanes should able to switch reversibly between the non-

2+ metallated polycatenane 22 and metallated polycatenate 22·Zn n.(Figure 4.31). Although both red and blue rings are very rigid, the polycatenane processes highly degree of topological

2+ flexibility and should exist as a random coil in solution. However, in the polycatenate 22·Zn n the topological mobility will be turned off by the metal-ligand coordination and polymer should become a rigid rod. This structural change should result in a large difference in hydrodynamic radius, which can be observed by DOSY (Figure S4.6, supporting information) and the result is listed in Table 4.7. A dramatic 38% increase of the hydrodynamic radius (RH) was observed that further supports its polycatenane structure.

Figure 4.31. Metallo responsive conformation change of a polycatenane.

Table 4.7 Diffusion coefficient and hydrodynamic radii of mostly linear polycatene 22c and 2+ * corresponding polycatenate 22c·Zn n 2 Sample Diffusion Coefficient (cm /s) RH (nm)

Polycatenane 22c 4.30 × 10−7 2.29

2+ −7 Polycatenate 22c·Zn n 5.95 × 10 3.17

*: Measured in 1,1,2,2-tetrachloroethane-d2 at 25°C at a concentration of 2.5mM (for monomer 14 moiety).

153

2.6 Side-reaction and Improvements

The polycatenane 22c and 22d contains 87% and 71% linear catenane, respectively, and they are also the major product of the reaction that contributes 50% and 26% mass of the product, respectively and as such the overall yield of linear polycatenane with the new the Zn-alt-MSP templated approach synthesized at 2.5 mM is as high as 40% (with branched (ca. 18%), cylic (ca.

8%), and non-interlocked (ca. 20%) being the rest of the product). However, there are two major issues that still need to be addressed in order to access high molecular weight linear polycatenanes.

(1) The isolation of linear architecture becomes very difficult and pure linear architecture were not able to be isolated.

(2) Although the removal of the non-interlocked byproduct can be achieved efficiently, this side reaction greatly reduces the molecular weight of the product. Given DP and NC, we can calculate the average end-to-end DP (i.e. the DP 2 × the length of the average arm length) of a branched polycatenane with the equation DPend-to-end = 2 ×DPMALLS/NC (For linear polycatenane,

DPend-to-end is equal to DP). Interestingly, when this is done the calculated DPend-to-end for 22a and

22b compares very favorably with the DP of the mostly linear polycatenane 22c and 22d (where the effect of the cyclic polycatenane has been removed), as show in Table 4.8. The DPend-to-end for all samples are around 10, corresponding to the ~10% chance of side reaction for the ring-closing of monomer 14. This number is also consistent with the yield of the non-interlocked side products.

154

Table 4.8 End-to-end DP of branched polycatenane and DP of mostly linear polycatenanes.

Sample Architecture DP NC DPend-to-end

22a Branched 55.4 8.69 12.75

22b Branched 24.92 4.58 10.88

22c Linear* 11.23 1.98 11.23**

22d Linear* 7.87 1.9 7.87**

*: Effect of cyclic polycatenane is excluded.

**: For linear polycatenane, DPend-to-end is equal to DP and is not calculated by equation.

As summarized in Figure 4.32, the product synthesized from Zn-alt-MSP contains multiple architectures: high concentration results in branching, low concentration increases the formation of cyclic polycatenane, while intermediate concentration produces linear along with both branched and cyclic. Unfortunately, the isolation of the linear polycatenane out of the mixed architectures was not achievable at this moment. The metathesis reaction of Zn-alt-MSP also forms the unwanted non-interlocked side products, the polymerized monomer 17 as well as free DiMC 23 that is not mechanically interlocked with polycatenane. The ca 10% chance of side reaction severely reduces the DP~10 and prevents the formation of larger linear polycatenane. Efforts on the improvement of this system will be the main focus of Chapter 5 and 6.

155

Figure 4.32. The formation of branched, cyclic polycatenane and byproduct.

3. Conclusion

In this chapter, the main-chain poly[n]catenane was successfully synthesized and characterized. The results show that the product is the mixture of linear, branched and cyclic polycatenane. The branched polycatenane shows the largest molecular size and a different distribution than the linear or cyclic ones, and are successfully isolated and characterized, with an numeric average molecular weight (Mn) up to 85.6 kDa, corresponding to a poly[55]catenane with low PDI of 1.11. A fraction of polycatenane with majorly linear architecture was also isolated, that contains 87% linear polycatenane and 13% cyclic polycatenane with average Mn of t 17.4 kDa, indicating an average of linear poly[11]catenane. The theoretical yield of linear polycatenane is ca.

156

40%; however, it is very difficult to isolate pure linear polycatenane out of the mixture that contains branched and/or cyclic architectures.

4. Supporting Information

4.1 Experimental Details

4.1.1 Materials and Instrumentations

Dichloromethane (DCM) for the metathesis reaction was purchased from ACROS

All other solvents and anhydrous potassium carbonate were purchased from Fisher Scientific and used without purification. Zinc bistriflimide was perchased from Strem Chemicals and stored in vacuum desiccator. Irom (II) bistriflimide was prepared according to literature.11 All other chemicals were purchased from Sigma-Aldrich and used without further purification.

All NMR spectra (except DOSY) was recorded on a Varian Inova 600 MHz spectrometer

(150.8MHz for 13C) at 25 °C (unless otherwise mentioned). All DOSYs were measured on a Bruker

157

4.1.2 Procedure of NMR studies

T1 relaxation was studied by inversion-recovery method. The T1 is calculated by the equation: T1=τnull/ln(2), where τnull is the waiting time when the measured peaks show a null signal.

8 T2 relaxation was studied by CPMG method. The data was fitted by mono-exponential decay

–Fτ equation A=B×e , where A is the peak area of interest and τ is the waiting time. The T2 calculated by the equation: T2 =1/F from the fitted value. DOSY was measured in a coaxial NMR tube and the sample was added into the inner tube. For NOESY studies, the sample was filtered by a 0.22

µm Teflon filter to remove any particle prior to measurement.

4.1.3 General procedure for the ring-closing metathesis reaction

The starting material was dissolved in distilled DCM to reach the desired concentration and transferred to a round-bottom flask equipped with stir bar and condenser. The solution was heated to reflux and bubbled with argon for 30 minutes to remove the dissolved oxygen. Then one

(of two) portion of of Hoveya-Grubbs 2nd generation catalyst (dissolved in 1 mL of distilled DCM) was added, and the solution was bubbled with argon for another 30 minutes. For all RCM reactions,

The amount of catalyst per portion was calculated by the equation: mcatalyst=0.2V, where mcatalyst is the amount of catalyst per aliquad (in the unit of mg), V is the amount of DCM used to dissolve the starting material to reach the desired concentration (in the unit of mL). The reaction solution was kept at reflux for 24 hours. Then other portion of catalyst was added. The solution was bubbled with argon for another 30 minutes and kept at reflux for another 24 hours. It is very important to allow cold water running through the condenser all the time to ensure efficient condensing of DCM to maintain a constant concentration as the product architecture is sensitive to the reaction concentration. The reaction progress can be monitored by NMR (discussed in Figure S4.3) and a 158

total reaction time of 48 hrs is sufficient for completion. After reaction finished, the solution was cooled to room temperature followed by the addition of excess ethyl vinyl ether to deactivate the catalyst.

4.1.4 Procedures for demetalation and catalyst removal

The polycatenate was dissolved in MeCN . For the Zn2+-polycatenate, the Zn2+ can be removed by adding an excess amount of ethylenediamine to the polycatenate solution. The polycatenane will precipitated out immediately upon the addition of ethylenediamine. The suspension was settled in fridge overnight to allow complete precipitation and the polycatenane was collected by filtration followed by washing with cold MeCN. For Fe2+-polycatenate, the Fe2+ can be removed by adding an excess amount of tetrabutylammonium hydroxide (1M solution in

MeOH) to convert iron into iron oxide. The suspension was also settled in fridge overnight followed by filtration and washing, during which the polycatenane and iron oxide were both collected, and the mixture was triturated with chloroform and filtrated to remove the iron oxide.

The deactivated catalyst is soluble in MeCN, which will be removed simultaneously during the demetalation process.

4.1.5 Protocol for polycatenane and oligocatenane purification

50 mg of demetalated poly(or oligo)catenane crude product was dissolved in 1 mL CDCl3 in an NMR tube. 25 mg of Zn(NTf2)2 (dissolved in 0.5 mL MeCN-d3) was gradually added into the product solution followed by extensive hand shaking and mixing. Monitored by NMR, the metal solution was added until ca. 60% of the non-metallated catenane became metallated. Then the solution was transferred into a via and the solvent was removed under vacuum, yielding a 159

yellow solid. 15 mL of chloroform/hexane mixture (2:1 v/v) was added in to the vial followed by shaking on a vortexer for 1 minute. The mixture was filtrated and the solid was collected into a vial. This process was repeated a number of times until the filtrate is no longer fluorescent. The solid was collected and the metal ion was removed by the demetallation procedure to yield the poly/oligocatenane.

4.1.6 Synthesis of DiMC 23.

DiMC 23 was synthesized by a similar approach reported in literature in three steps:

Step 1: In a 100 mL round-bottom flask equipped with stir bar, 5g (8.23 mmol, 1 eq.) of

21, 11.5 g (4 eq., 32.9 mmol) of tetraethylene glycol mono-p-toluenesulfonate (prepared according to a literature procedure)3 and 6.8g (6 eq., 49.4 mmol) of potassium carbonate were added. The flask was then flushed with argon and 40 mL dry DMF is then added by cannula. The mixture is heated in an oil bath to 70 °C and allowed to react for 24 hrs. After the reaction the DMF is removed under reduced pressure and resulting mixture triturated with CHCl3 and then filtered. The filtrate was collected and the solvent removed under reduced pressure to yield a yellow-brown waxy solid.

The product was purified by column chromatography (silica gel, chloroform/methanol gradient) to yield the pure product 26 in a 89% yield as light yellow waxy solid. 1H NMR (600 MHz,

Chloroform-d) δ 8.39 (d, J = 7.9 Hz, 2H), 8.10 (t, J = 7.9 Hz, 1H), 8.05 (s, 2H), 7.62 – 7.58

160

(overlapped, 6H), 7.51 (d, J = 8.4 Hz, 2H), 7.04 (d, J = 8.4 Hz, 4H), 4.76 (t, J = 7.4 Hz, 4H), 4.21

(t, J = 4.8 Hz, 4H), 3.90 (t, J = 4.8 Hz, 4H), 3.79 – 3.59 (m, 20H), 2.66 (br, 2H), 1.75 (p, J = 7.4

13 Hz, 4H), 1.15 (h, J = 7.4 Hz, 4H), 0.73 (t, J = 7.4 Hz, 6H). C NMR (126 MHz, CDCl3) δ 158.07,

150.66, 149.92, 143.33, 138.24, 136.11, 135.45, 134.43, 128.34, 125.56, 123.12, 118.00, 115.02,

110.52, 72.90, 72.61, 70.80, 70.65, 70.59, 70.51, 70.33, 69.98, 69.77, 67.53, 61.65, 61.51, 44.78,

32.17, 19.87, 13.53. MALDI-TOF MS: 960.69 ([M]+H+).

Step 2: 5g (5.21 mmol, 1 eq.) of 26 and a catalytic amount of 4-dimethylaminopyridine

(DMAP) (0.41 mmol, 0.05g, 0.08 mol. %) was added to a 100mL round bottom flask (equipped with a stir bar) followed by 20 mL of dichloromethane (DCM) and 4.5 mL of triethylamine (TEA).

The mixture was flushed with argon and then submerged in an ice bath. 2.48g (13 mmol, 2.5 eq.) of TsCl (dissolved in 10 mL of DCM) was then added dropwise by syringe. The reaction was allowed to warm to room temperature and stirred for overnight. After reaction completed, the mixture was filtered to remove the salt and the solvent was removed under vacuum yielding yellow-brown oil. The resulting material was purified by column chromatography (silica gel, chloroform/methanol gradient) to yield colorless oil 27 in a 76% yield. 1H NMR (600 MHz,

Chloroform-d) δ 8.35 (d, J = 7.8 Hz, 2H), 8.07 (t, J = 7.9 Hz, 1H), 8.02 (d, J = 1.6 Hz, 2H), 7.80

(d, J = 8.3 Hz, 4H), 7.62 – 7.59 (m, 4H), 7.58 (dd, J = 8.4, 1.7 Hz, 2H), 7.50 (d, J = 8.4 Hz, 2H),

7.33 (d, J = 8.0 Hz, 4H), 7.03 (d, J = 8.7 Hz, 4H), 4.77 (t, J = 7.4 Hz, 4H), 4.22 – 4.18 (m, 4H),

4.18 – 4.14 (m, 4H), 3.91 – 3.88 (m, 4H), 3.76 – 3.73 (m, 4H), 3.71 – 3.68 (m, 4H), 3.68 – 3.65

(m, 4H), 3.63 – 3.59 (overlapped, 8H), 2.43 (s, 6H), 1.75 (p, J = 7.5 Hz, 4H), 1.15 (h, J = 7.4 Hz,

13 4H), 0.73 (t, J = 7.4 Hz, 6H). C NMR (151 MHz, CDCl3) δ 158.07, 150.67, 149.96, 144.72,

143.37, 138.13, 136.06, 135.46, 134.43, 132.99, 129.76, 128.31, 127.94, 125.49, 123.07, 118.03,

161

114.97, 110.44, 70.80, 70.74, 70.66, 70.61, 70.55, 69.75, 69.20, 68.66, 67.52, 53.38, 44.74, 32.15,

21.59, 19.85, 13.48. MALDI-TOF MS: 1268.75 ([M]+H+).

Step 3. A 125mL dropping funnel containing 21 (1.39g, 2.29 mmol, 1eq.) and 27 (2.9g,

2.29 mmol, 1eq.) in 100 mL of anhydrous DMF was fitted to a 1 L two-necked round bottom flask containing Cs2CO3 (2.98g, 9.14 mmol, 4eq.) and a stir bar. The reaction vessel was flushed with argon before anhydrous DMF (450 mL) was added by cannula. The reaction was submerged in an oil bath and heated to 70°C while rapidly stirring. The dissolved mixture of components in the dropping funnel was then added dropwise to this suspension over the course of 2 days. The reaction was stirred at 70°C for a total of 4 days, after which the solvent was removed and the crude product was triturated with CHCl3 and filtered. The filtrate was collect and solvent was removed under vacuum. The resulting solid was purified by column chromatography (TEA pretreated silica gel, chloroform/methanol gradient) and recrystallization

1 (CHCl3/MeCN) to yield 23 as white solid crystals in 53% yield. H NMR (600 MHz,

Chloroform-d) δ 8.30 (d, J = 7.8 Hz, 4H), 8.02 – 7.96 (overlapped, 6H), 7.52 – 7.48 (m, 8H),

7.39 (dd, J = 8.4, 1.7 Hz, 4H), 7.27 (d, J = 8.4 Hz, 4H), 7.00 – 6.94 (m, 8H), 4.56 (t, J = 7.3 Hz,

8H), 4.14 – 4.11 (m, 8H), 3.93 – 3.90 (m, 8H), 3.79 – 3.76 (m, 8H), 3.74 (dt, J = 6.0, 2.0 Hz,

8H), 1.60 – 1.52 (m, 8H), 1.01 – 0.91 (m, 8H), 0.59 (t, J = 7.4 Hz, 12H). 13C NMR (126 MHz,

CDCl3) δ 158.16, 150.49, 150.01, 143.35, 138.05, 135.97, 135.40, 134.18, 128.26, 125.50,

123.11, 117.86, 115.02, 110.45, 70.91, 69.79, 67.65, 44.58, 32.07, 19.74, 13.54, 13.45. MALDI-

TOF MS: 1553.13 ([M]+Na+).

162

4.1.7 Synthesis and fractionation of polycatenane 22.

Polycatenane 22 is synthesized from Zn-alt-MSP with RCM at a concentration of 2.5 mM, 10 mM or 0.25 mM (concentration is based on 14, same below). Product is purified using the catenane purification procedure described in Figure 4.10. Yield: (mixture of all architectures):

75% (for 2.5 mM reaction), yellow waxy solid. The color is due to the residue deactivated catalyst, which can be removed during fractionation. Fractionation was performed by passing the polycatenane through hand-loaded size-exclusion column with 2:1 chloroform/ethanol as eluent and Tosoh Toyopearl HW50 as media (solvent exchanged into eluent prior to use). After fractionation, all fractions are colorless waxy solid, and 95% mass was the recovered. NMR: 1H

NMR (600 MHz, Chloroform-d) δ 8.28 – 8.13 (m, 8H), 7.98 – 7.82 (overlapped, 12H), 7.46 –

7.19 (overlapped, 34H), 7.03 – 6.93 (overlapped, 4H), 6.90 – 6.76 (overlapped, 8H), 6.71 – 6.62

(m, 8H), 5.39 – 5.26 (m, 2H), 5.09 – 4.98 (m, 4H), 4.64 – 4.42 (m, 16H), 3.94 – 3.76 (m, 12H),

3.70 – 3.49 (m, 24H), 2.16 – 1.97 (m, 4H), 1.78 – 1.37 (overlapped, 26H), 1.10 – 0.86 (m, 16H),

13 0.65 – 0.47 (m, 24H). C NMR (151 MHz, CDCl3) δ 158.31, 157.94, 157.63, 150.70, 150.61,

150.51, 150.44, 150.11, 143.46, 138.05, 137.48, 136.09, 135.46, 134.38, 134.29, 134.17, 133.95,

130.20, 129.25, 128.35, 128.19, 127.97, 126.58, 125.40, 123.22, 121.34, 117.89, 115.63, 115.09,

115.01, 114.89, 114.77, 110.55, 110.48, 70.93, 70.87, 70.81, 69.57, 69.09, 67.53, 67.29, 67.06,

46.26, 44.75, 33.84, 32.49, 32.21, 30.20, 29.86, 28.80, 28.71, 27.47, 19.90, 19.70, 13.60.

4.1.8 Synthesis of [3]catenane 25

[3]catenane 25 was synthesized by RCM at 2.5 mM concentration from

2+ [3]pseudorotaxane (23 Zn )2·14 which is prepared by titrating 2 eq. of 23, 1 eq. of 14 and 2 eq. of Zn(NTf2)2 monitored by NMR. The product is purified by column chromatography with 163

triethylamine pretreated silica gel with chloroform/methanol gradient as eluent. Yield: 10%,

1 colorless waxy solid. H NMR (600 MHz, CDCl3) δ 8.27 – 8.14 (m, 12H), 7.98 – 7.82 (m, 18H),

7.45 – 7.17 (m, 50H), 7.01 – 6.93 (m, 8H), 6.88 – 6.77 (m, 16H), 6.69 – 6.64 (m, 4H). 5.43 –

5.36 (m, 2H), 5.09 – 5.00 (m, 4H), 4.65 – 4.35 (m, 24H), 4.07 – 3.82 (m, 20H), 3.82 – 3.54 (m,

48H), 2.23 – 2.00 (m, 4H), 1.90 – 1.66 (m, 24H), 1.66 – 1.40 (m, 34H), 0.67 – 0.42 (m, 36H).

13 C NMR (126 MHz, CDCl3) δ 158.16, 157.70, 157.46, 150.47, 150.30, 149.86, 143.21, 137.97,

137.30, 135.91, 135.25, 134.15, 133.95, 130.08, 129.11, 128.19, 127.98, 125.27, 123.05, 117.67,

115.44, 114.91, 114.83, 114.68, 110.39, 87.58, 85.50, 83.42, 71.02, 70.67, 69.46, 68.93, 67.09,

66.85, 44.57, 32.35, 32.03, 29.70, 28.63, 19.73, 13.45, 13.44. MALDI-TOF MS: 4642.64

([M]+H+).

4.1.9 Synthesis of catenane 22Oligo and 22Cyclic-oligo.

Oligo Oligomeric catenane 22 was synthesized by RCM at 0.25 mM from precursor 232

2+ 142·Fe 3 which is prepared by titrating 2 eq. of 23, 2 eq. of 14 and 3 eq. of Fe(NTf2)2 monitored by NMR. The product is purified by the catenane purification procedure mentioned above to remove the majority of the non-catenane byproducts. Yield: 55% (purity: ca. 90%, containing ca.

10% non-interlocked species indicated by 1H-NMR), yellow waxy solid (color is due to trace amount of residue deactivated catalyst). To isolate the 22Cyclic-oligo, the purified 22Cyclic was put through the catenane purification procedure again. At this time, the 22Cyclic-oligo preferentially binds with metal and was isolated by filtration and demetalation. 164

4.1.10 Synthesis of catenane 15Oligo.

Oligo 2+ Catenane 15 was synthesized by RCM at 0.25 mM from precursor 145·Fe 4 which is prepared by titrating 5 eq. of 14 and 4 eq. of Fe(NTf2)2 monitored by NMR. The product is purified by the catenane purification procedure to remove the non-catenane byproducts. Yield:

60% (purity: ca. 96% in catenane species), yellow waxy solid (color is due to trace amount of residue deactivated catalyst).

4.2 Supplementary Figures and Tables.

4.2.1 The UV-Vis titration of DiMC 23 and Bip 24 with Fe2+

The UV-Vis titration was performed by the following procedure. The solvent used for all

UV-Vis titration is 9:1 v/v chloroform/acetonitrile. After baseline measurement with pure solvent, the UV cell was emptied and dried, followed by the addition of 2 mL of the 15 µM DiMC 23 solution (30 30 uM Bip moiety). The UV-Vis spectra of 23 was measured. Thenthe titration was started by adding 15 uL of the Fe(NTf2)2 solution (containing 400 uM metal ion and 15 uM 23) followed by UV-Vis measuremt. The titration process was continued until ca. 3.6 eq. of metal ion was added. The titration of Bip 24 was done at 30 uM to maintain the same Bip moiety concentration for comparison.

165

Figure S 4.1. The UV-Vis Titration of DiMC 23 (a) and Bip 24 (b) with Fe2+.

166

4.2.2 DOSY of Zn-alt-MSP compared with Monomer 14

Figure S 4.2. DOSY of the Zn-alt-MSP (a) and Monomer 14. Measured in 1,1,2,2-tetrachloroethane-d2 at 25°C. Concentration is 2.5 mM (based on 14) Solvent contains 1 v/v% TMS as internal standard to monitor the consistency of the gradient strength.

167

4.2.3 The conversion of double-bond during metathesis ring-closing reaction

Before reaction, the Zn-alt-MSP shows two crosspeaks corresponding (O,P) and (Q,R).

After reaction, Q and R are removed as the form of butane and only one crosspeak (O,P) was observed.

Figure S 4.3. 1H-1H COSY (partial) of Zn-alt-MSP before and after the metathesis reaction

168

4.2.4 T1 and T2 measurement of polycatenane 22 at room temperature

Figure S 4.4. Inversion-recovery and spin-echo CPMG of polycatenane measure at 25°C.

Table S 4.1. T1 and T2 of polycatenane measure at room temperature.

Region T1/sec T2/sec

(I) 2.29 0.526

(II) 2.21 0.254

(III) 2.18 0.285

169

4.2.5 NMR integration of the HB peak for cyclic oligomeric catenane

Cyclic-oligo Figure S 4.5. NMR integration of the HB peak for 22 Neglecting the very minor chain-end peak that does not belongs to the cyclic catenane, NMR integration of region IV shows 53.5% of total peak area of HB.

170

4.2.6 DOSY spectra of mostly linear polycatenane 22c and corresponding polycatenate.

Figure S 4.6. DOSY of mostly linear polycatenane 22c (a) and corresponding polycatenate 2+ 22c·Zn 2n+1 (b).

171

References:

1 M. Enamullah and W. Linert, Journal of Coordination Chemistry, 1996, 40, 193–201.

2 A. M. Albrecht-Gary, Z. Saad, C. O. Dietrich-Buchecker and J. P. Sauvage, J. Am. Chem.

Soc., 1985, 107, 3205–3209.

3 R. J. Wojtecki, Q. Wu, J. C. Johnson, D. G. Ray, L. T. J. Korley and S. J. Rowan, Chem.

Sci., 2013, 4, 4440.

4 Atta-ur-Rahman and M. I. Choudhary, in Solving Problems with NMR Spectroscopy,

Elsevier, 1996, pp. 187–211.

5 F. Heatley, in NMR Spectroscopy of Polymers, Springer Netherlands, Dordrecht, 1993, pp. 1–49.

6 L. Li, E. B. Twum, X. Li, E. F. McCord, P. A. Fox, D. F. Lyons and P. L. Rinaldi,

Macromolecules, 2013, 46, 7146–7157.

7 Y. Kim, Bio Design, 2015, 3, 14–25.

8 T. Yuwen and N. R. Skrynnikov, Journal of Magnetic Resonance, 2014, 241, 155–169.

9 E. C. Constable, N. Hostettler, C. E. Housecroft, N. S. Murray, J. Schönle, U. Soydaner,

R. M. Walliser and J. A. Zampese, Dalton Transactions, 2013, 42, 4970.

10 H. B. Tanh Jeazet, K. Gloe, T. Doert, O. N. Kataeva, A. Jäger, G. Geipel, G. Bernhard, B.

Büchner and K. Gloe, Chem. Commun., 2010, 46, 2373.

11 M. P. Sibi and G. Petrovic, Tetrahedron: Asymmetry, 2003, 14, 2879–2882.

172

Chapter 5. Synthesis of Poly[n]catenane using Fe2+ as

Templating Metal

1. Introduction

In chapter 4, we demonstrated the successful synthesis of poly[n]catenanes, which consisted of a mixture of linear, branched and cyclic polycatenanes. Among the three architectures, linear poly[n]catenane is of particular interest. The proposed mechanisms of the formation of branched and cyclic architecture is showing in Figure 5.1. When performing the ring-closing metathesis (RCM) at relatively high concentrations (e.g. using Zn2+ions as the template, ca. 10 mM of the monomer), the intermolecular reaction between two metallo-supramolecular polymer

(MSP) chains can readily occur to yield branched polycatenanes. At low concentrations (e.g. 0.25 mM), the formation of the branched polycatenane is suppressed but the amount of cyclic polycatenanes and non-catenane byproducts is increased. At intermediate concentration (e.g. 2.5 mM), the formation of linear polycatenane is improved to 40~50%, but it is very difficult to isolate the linear polycatenane from the mixture that also branched or cyclic polycatenanes. In order to improve the yield of linear poly[n]catenane, a modified synthetic approach is required. In this chapter we focus on investigating the use of a different templating ion, namely Fe2+.

Bip is known to form a 2:1 complexes with Fe2+ ions which will allow the assembly of the supramolecular polymer (Figure 5.1).1 Relative to the Zn2+ions, Fe2+ ions exhibit a much stronger overall binding constant (1010 M−2 vs 106 M−2 for Zn2+)1,2 that should allow the formation of supramolecular polymers at the low concentrations (e.g. 0.25 mM or lower) and which in turn should suppress the branching reaction during the RCM step.

173

Figure 5.1 Illustration of the formation of branched/linear/cyclic polycatenane mixture with Zn2+ as templating metal and the proposal of using Fe2+ as templating metal to optimizing the synthesis of linear polycatenane.

2. Synthesis of Polycatenane Templated with Fe2+

2+ 2+ The assembly of Fe supramolecular polymer (Fe-alt-MSP) 14n+1·23n+1·Fe 2n+1 was monitored by 1H-NMR (Figure 5.2). The presence of Fe2+ did not interfere the NMR measurement as the Fe2+is in the low-spin state upon binding with Bip at 2 Bips/1 Fe2+ stoichiometry.3 After fully metalation, all the peaks in the aromatic region significantly shift (especially the HA, HB and

2+ HC), in a similar manner to what is seen upon metalation with Zn (see Figure S5.1 for NMR comparison). After removing the solvent under vacuum, the Fe-alt-MSP could form a deep purple film. DOSY studies also confirms the formation of high molecular species. The diffusion coefficient of Fe-alt-MSP (1.95 × 10 −7 cm2/s) is much smaller than that of Zn-alt-MSP (3.54 × 10 174

−7 cm2/s) measured at same condition (see Figure S5.2, supporting information), indicating the Fe- alt-MSP has much larger hydrodynamic radius, which is in agreement with the larger binding constant between Fe2+ and Bip.

Figure 5.2 The molecular structure of 14 and 23 and the assembly of Fe-alt-MSP monitored by 1H- NMR showing the aromatic region. Inset shows the Fe-alt-MSP film obtained after sovent removal. NMR was measured in CDCl3 (before metalation) or CDCl3 containing ≤ 25% MeCN-d3 (during and after metalation) at 25 ºC, 600 MHz. The molecular structures of 14 and 23 are also shown.

175

2+ After obtained the Fe-alt-MSP 14n+1·23n+1·Fe 2n+1, the ring-closing metathesis (RCM) reaction was performed under very dilute condition (0.25 mM according to 14), as illustrated in

Figure 5.3. The corresponding polycatenane product (labeled as 22Fe). The product synthesized by

Zn-alt-MSP at 2.5 mM reaction concentration (Chapter 4) was labeled as 22Zn for clarification.

Similar to 22Zn, 22Fe was also purified (yield: ca. 80%) and fractionated into four fractions: 22Fea through 22Fed, in the order of decreasing hydrodynamic volume, for the more detailed studies.

Figure 5.3 Synthesis approach of polycatenane using Fe2+ as templating metal ion.

3. Architecture Study of Fe2+ Templated Polycatenanes

3.1 Molecular weight

The molecular weight of polycatenane 22Fe was determined by GPC-MALLS as shown in

Figure 5.4a. Compared to 22Zn, 22Fe shows one broad peak and the majority of 22Fe elutes faster than 22Zn, indicating a large average hydrodynamic volume. The unfractionated polycatenane 22Fe has an average molecular weight ca. 23.2 kDa with a polydispersity index (Ð) of 1.32, which is

Zn Zn larger than 22 (Mn=21.4 kDa, Ð =1.437). The small peak of 22 at 9.4 min corresponding probably to the cyclic polycatenane was not observed in 22Fe. However, peak deconvolution

(Figure 5.4b) shows that it is contributed by three Gaussian peaks that may corresponds to different architectures.

176

Figure 5.4 GPC spectra of. (a) 22Fe compared with 22Zn , and (b) peak deconvolution of 22Fe and

177

Figure 5.4c shows the fractionated 22Fea-d. The isolation of the polycatenane into four different molecular weight fractions was successful and the polydispersity indexes (Ð) are all low

(< 1.3). Their relative mass, molecular weight and Ð are compared in Table 5.1. Fraction 22Fea mainly corresponds to Gaussian peak ①, fraction 22Feb corresponds to peaks ① and ②, 22c is predominantly peak ② but may have small contributions to peak ① and ③, while 22d corresponds to peaks ② and ③.

As discussed in Chapter 4, region (I) and (IV) are corresponds to chain-end and cyclic polycatenane. The intensity ratio of (II) and (III) implies the branching—higher intensity in region

(II) indicates more branching. Figure 5.5a shows the NMR spectra of 22Fe and 22Fea-d. The chain- end peak (Region I) increases along with the decreasing of molecular size from 22Fea to fraction-

22Fed. However, the unfractionated polycatenane 22Fe as well as all fractions have relatively strong peaks in region (II), which is indicative of branched polycatenanes. In addition, there are peaks in region IV for the 22Fed fraction (and also very slightly in 22Fec), which are proposed to correspond to cyclic polycatenane architectures. If this assignment is correct the presence of cyclic, as well as non-cyclic species, would explain the bimodal distribution of 22Fed observed in the GPC spectrum.

To quantitatively study the branching, we calculated the average number of chain-ends per molecule (NC). As discussed in chapter 4, for linear, cyclic or branched polycatenane, NC will be

2, 0 or >2, respectively. NC was calculated (Table 5.1) using the GPC-MALLS molecular weight and NMR integration, in a similar manner to what is reported in Chapter 4.

178

1 Fe Figure 5.5. The H-NMR of the HB peak of (a) of polycatenane 22 and its four fractions; and (b) the comparison of 22Fec-d and 22Znc-d.

Measured in CDCl3 at 25 °C, 600 MHz.

Table 5.1 NC calculation and architecture study of polycatenane 22Fe and its fractions.

Isolated NMR * Sample Mn /kDa Ð NC DPend-to-end Mass % Itotal/Ichain-end

22Fe 100% 28.2 1.32 6.44 5.67 6.4

22Fea 11% 57.4 1.08 8.50 8.74 8.5

22Feb 36% 32.6 1.15 6.20 6.81 6.2

22Fec 35% 20.9 1.19 5.65 4.79 5.6

22Fed 18% 16.0 1.27 7.12 2.90 -- 22Fed (cyclic N/A 16.0*** N/A 4.56 4.54 4.6 subtracted)** * : DPend-to-end calculation excludes the cyclic polycatenane. For details, see Chapter 4, Section 2.6 **: For details about cyclic subtraction, see Chapter 4, Figure 4.19. *** : Due to peak overlapping, we are unable to measure the Mn of only the linear proportion. We neglect the change of Mn during the subtraction of cyclic proportion. 179

The calculation results show that the NC of all fractions are greater than 2. Actually, if we

Fe subtract the contribution of cyclic component in 22 d, NC are all larger or similar to 4. For the

Fe polycatenane 22 or its fractions, the NC of 4 or greater, i.e. there are on average 4 or more chain- ends per molecule, indicating that on average each chain has at least one branch.

It is reasonable to expect the two larger fractions 22Fea-b to be predominately branched, as the branching comes from the inter-chain reactions of linear polycatenates. However, for the two

Fe smaller fractions 22 c-d, the NC of them (4.79 and 4.54, respectively, excluding cyclic polycatenane) are much larger than 2, indicates they are also majorly branched architectures.

To further study these two fractions, figure 5.5b compares them with the two smaller fractions of

Zn2+ templated product 22Znc-d (Chapter 4). The major differences are 22Fec-d shows much higher

Fe relative intensity in region (I) and (II) but lower intensity in (III). The Mn weight of 22 c (20.9 kDa) and 22Fed (16.0 kDa) are larger than 22Znc and 22Znd, respectively. However, strong intensity of chain-ends in 22Fec-d show they have large number of chain-ends and the length of each branch are probably very short.

To quantitatively analysis the length of branches, the DPend-to-end (i.e. the DP 2 × the length of the average arm length, see Chapter 4, Section 2.6 for details) were calculated and shown in

Fe Fe Table 5.1. The DPend-to-end of 22 c and 22 d are especially small (5.6 and 4.6, respectively, compared to Ca. 10 for 22Zn), indicating there is a branching point ca. every 3 interlocked rings

(i.e. 1/4 of the rings are branching points). Since the branching results from the dimerization of the blue ring while the red ring cannot induce branching, simple math shows that ca. 50% of the blue rings forms dimers (or trimers, tetramers, etc, although the possibility of them should be much lower).

180

The number of branching is unusually high. To find out a possible explanation, we can compare their precursors for metathesis reaction, the Fe-alt-MSP and Zn-alt-MSP in DCM solution. Firstly, the Fe-alt-MSP assemblies are much larger than Zn-alt-MSP, which means that under the same concentration, the total quantity of Fe-alt-MSP assemblies are less than that of Zn- alt-MSP in a certain voume of solution. Secondly, the metathesis reaction concentration for Fe- alt-MSP (0.25 mM) is much more dilute than that of Zn-alt-MSP (2.5 mM). Therefore, the chance of two Fe-alt-MSP assemblies to meet with each other is much lower than that of Zn-alt-MSP.

The reason for branching should not come from the inter-molecular reaction from linear polycatenates.

One possible reason is the branching reaction happens within the same Fe-alt-MSP assembly, which is analogous to the “backbiting” effect in polymer chemistry. The backbiting is very common in metathesis polymerization due to the dynamic exchange of double-bonds (Figure

5.6a).4 In our system, both Zn- and Fe-alt-MSPs are very rigid due to the high aromatic content.

The short rigid rod-like structure of Zn-alt-MSP limits the chance of backbiting reaction (Figure

5.6b). For Fe-alt-MSP, however, its very long length may overcome its rigidity making the backbiting possible (Figure 5.6c). Moreover, due to the strong binding constant between Fe2+ and

Bip, the large enthalpy gain can overcome the entropy penalty, forming cyclic Fe-alt-MSP assemblies (Figure 5.6d). The cyclic structure greatly increases the local concentration of reactive chain-ends, which becomes more susceptible for back-biting reaction.

181

Figure 5.6. (a) The backbiting reaction in metathesis polymerization. (b) Zn-alt-MSP is short and rigid, not likely for the backbiting. (c) Fe-alt-MSP is rigid but long, so backbiting may happen. (d) The formation of cyclic Fe-alt-MSP increases the local concentration of double-bonds which is susceptible to backbiting reactions.

182

It is worthy to note that once the backbiting reaction happens, the local concentration of the reactive chain-ends further increases as due to the newly formed covalent bond that acts as a covalent template (Figure 5.7, yellow highlighted area). Therefore, multiple branching reactions are likely occurred while the side-reaction (green highlighted) decreases the molecular weight, yielding very highly branched polycatenanes, which may be a good reason for the ca. 50% of the ring-closed monomers dimerized into branching points. Therefore, the formation of linear polycatenane may occur but in a very minor way, and as such no linear-rich fraction was isolated from 22Fe.

Figure 5.7 Proposed local-concenrated mechanism behind the high conversion toward branched polycatenanes (a) and the proposed mechanism of the formation of cyclic polycatenane (b) using Fe2+ as templating metal ion.

183

4. Attempts to Improve the Synthesis

In order to solve this problem, we proposed a modification on the synthesis condition. As discussed in chapter 4, varying the stoichiometry for Monomer 14 and DiMC 23 from 1:1 (which yields a long Fe-alt-MSP) to 10:11 should reduce the length of the assembly, results in the formation of a metallo-supramolecular oligomer (MSO) named as Fe-alt-MSO with a theoretical

DP of ~21, as shown in Figure 5.8. Fe-alt-MSO should have a shorter length and the backbiting should be prevented. To avoid inter-chain reaction, we also further reduced the RCM concentration from 0.25 mM to 0.1 mM.

Figure 5.8 Modified synthesis of polycatenane.

After reaction, we purified the product (named 22Fe-Oligo) and fractionated by SEC column

Fe-Oligo into three fractions, named as 22 a-c, respectively, followed by GPC, NMR studies and NC calculation (Table 5.2). GPC spectra (Figure 5.9a) shows good fractionation of 22Fe-Oligo into three fractions with different molecular weight and very low Ð. As expected the molecular weight becomes much smaller than 22Fe indicating successful molecular weight controlling. The top 4% largest fraction 22Fe-Oligo is only 28.7 kDa, which is only a half as large as 22Fea (57.4kDa) and the other two smaller fractions 22Fe-Oligob-c are also much smaller. NMR spectra indicates there are some cyclic peaks in region (IV) for the smallest fraction 22Fe-Oligoc (Figure 5.9b), which is also observed as a shoulder peak in GPC (Figure 5.9a, red). 184

Figure 5.9 GPC and 1H-NMR (partial) of polycatenane 22Fe-Oligo and its fractions 22Fe-Oligoa-c. GPC was measured with THF as eluent, and the intensity of unfractionated 22Fe-Oligo was multiplied by 3. 1 The H-NMR was measured in CDCl3 at 25 ºC, 600 MHz.

Fe-Oligo Table 5.2 NC calculation and architecture study of polycatenane 22 and its fractions.

Isolated NMR Sample Mn /kDa Ð NC DPend-to-end Mass % Itotal/Ichain-end

22Fe-Oligo 100% 13.1 1.15 4.11 4.13 4.1

22Fe-Oligoa 4% 28.7 1.06 4.14 8.97 9.0

22Fe-Oligob 47% 15.2 1.11 4.06 4.85 4.8

22Fe-Oligoc 49% 9.87 1.08 3.72 3.43 --

Fe-Oligo 22 c * N/A 9.87 N/A 2.58 4.95 5.0 (cyclic subtracted) * : Due to peak overlapping, we are unable to measure the Mn of only the linear proportion. We neglect the change of Mn during the subtraction of cyclic proportion.

185

Based on the NC and DPend-to-end calculation in Table 5.2, all fractions show NC >4 which

Fe-Oligo are mainly branched. DPend-to-end of the two smaller fractions 22 b-c are 4.8 and 5.0, respectively, which is similar to 22Fec-d (5.6 and 4.6, respectively). The formation of branching is not reduced.

In order to figure out the reason, we studied the diffusion coefficient of the RCM reactant, Fe-alt-

MSO, by DOSY. The diffusion coefficient of Fe-alt-MSO is very low (below the detection limit of DOSY). The fact the diffusion coefficient could not be calculated this tell us the size of Fe-alt-

MSO should be very large. To estimate the size, the DCM solution of Fe-alt-MSO was injected through a 0.22 µm Teflon syringe filter. No particle was blocked by the filter, suggesting that their diameter is smaller than 0.22 µm. Therefore, we proposed that the reason for the large size of Fe- alt-MSO in the diffusion experiment is due to aggregation, which may due to using an excess amount of DiMC 23 (1.1 eq w.r.t. 14) that has poor solubility in DCM. If aggregation of Fe-alt-

MSO does occur this would result in severe branching reactions, yielding branched polycatenanes.

Therefore, although using excess DiMC 23 successfully reduces the molecular weight of 22Fe-Oligo, no improvement on the prevention of yielding highly branched polycatenane was observe. The proposed reason is the aggregation of Fe-alt-MSO due to insufficient solubility. In order to improve the solubility which is limited by DiMC 23, the macrocycle need to be redesigned, which will be the main focus of Chapter 6.

5. Conclusion

In conclusion, we successfully synthesized polycatenanes using Fe2+ as the templating metal ion. The stronger binding constant between Bip and Fe2+allows us to do the RCM reaction at much lower concentration without the dissociation of the MSP. However, the products are predominately

186

branched polycatenanes. One possible reason is large or cyclic Fe-alt-MSPs may result in backbiting, i.e. the branching reaction with in the same molecule. We tried to control the size of the MSP by using excess DiMC 23 to assemble the oligomeric Fe-alt-MSO. However, data suggests that the Fe-alt-MSO are aggregated in DCM solution possibly due to its poor solubility.

The DCM-insoluble DiMC 23 is likely the main reason for insufficient solubility. In order to solve this problem, we need to redesign a better macrocycle which will be discussed in Chapter 6.

6. Supporting Information

6.1 The assembly of Fe-alt-MSP

Monomer 14 and Macrocycle 23 was dissolved in CDCl3 with a concentration of ca. 150 mg/mL and ca. 30 mg/mL, respectively. Fe(NTf2)2 was dissolved in MeCN-d3 in ca. 75 mg/mL.

Firstly, 14 and 23 was mixed in a NMR tube into exactly 1 to 1 stoichiometry (for oligomeric

1 MSP, 1 to 1.1) monitored by H-NMR. Then, the Fe(NTf2)2 solution was added in to the NMR tube until no free Bip peak is observed. The solution was equilibrated at 45 ºC under argon atmosphere for 2 days. Finally, the solvent is removed under vacuum. Fe-alt-MSP yields a deep purple film while Fe-alt-MSO does not. The 1H NMR comparison of Fe-alt-MSP and Zn-alt-MSP is shown in Figure S5.1.

187

Figure S 5.1. 1H NMR comparison (aromatic region) of Fe-alt-MSP and Zn-alt-MSP.

Condition: 2.5 mM w.r.t. 14, 25 °C, C2D2Cl2, 500 MHz.

188

6.2 DOSY study of Fe-alt-MSP

Figure S 5.2. DOSY study of (a) Fe-alt-MSP, (b) Zn-alt-MSP and (c) Monomer 14 under 2.5 mM concentration (w.r.t. 14 moiety). Measured in 1,1,2,2-tetrachloroethane-d2 (containing 1% v/v TMS) at 25 °C, 500 MHz. TMS was used as internal standard to monitor the gradient strength.

189

6.3 The Synthesis of 22Fe (and 22Fe-Oligo) via RCM Reaction

For 22Fe: in a 250 round bottom flask, 100 mg of the Fe-alt-MSP film is redissolved in 90 mL DCM to reach 0.25 mM concentration of monomer 14 moiety. The solution was heated to reflux (45 ºC) under argon atmosphere and bubbled with argon for 30 min. Then the first (of three)

18 mg of Hoveyda-Grubbs 2nd generation catalyst (dissolved in 2 mL of DCM) was injected, following by bubbling argon for another 30 minutes. The reaction was kept at reflux for 24 hours.

Then the second (of three) 18 mg of Hoveyda-Grubbs 2nd generation catalyst (dissolved in 2 mL of DCM) was injected, following by bubbling argon for 30 minutes. The reaction was refluxed for another 24 hours. Finally, the third (of three) 18 mg of Hoveyda-Grubbs 2nd generation catalyst

(dissolved in 2 mL of DCM) was injected following by bubbling argon for 30 minutes and refluxing for 24 hours. (For 22Fe-Oligo, 25 mg of Fe-alt-MSO dissolved in 125 mL of DCM for 0.1 mM 14 moiety. 25 mg of catalyst was added each time). After cooled down, 5 mL of ethyl vinyl ether was added and stirred for 5 hours to fully deactivate the catalyst. Then the Fe2+ was converted to iron hydroxide by adding 2 mL of tetrabutylammonium hydroxide solution (1.0 M in methanol) and stirred for 2 hours. DCM suspension was triply washed with water and filtered to remove iron hydroxide, followed by solvent removal under vacuum. The catalyst was removed by redissolving the crude product in 2 mL chloroform and precipitate into 20 mL cold acetonitrile.

6.4 Purification and Fractionation of Polycatenane

The crude product was purified by the catenane purification protocol described in Chapter

4 section 2.2.3. Polycatenane was fractionated by hand-loaded size-exclusion column with Tosoh

Toyopearl HW-50 as media and 2:1 v/v chloroform/ethanol as eluent.

190

6.5 GPC-MALLS and 1H-NMR Study

GPC-MALLS study was performed by Agilent 1260 GPC system with Agilent PLgel column, Wyatt Optilab T-rEX RI detector and Wyatt DAWN Heleos II MALLS detector. The eluent is THF with a flow rate of 1mg/mL.

1 H-NMR of polycatenanes was studied by Varian 600 MHz NMR in CDCl3 contain 0.03 vol%

TMS (unless mentioned elsewhere) at 25 ºC, and all chemical shift was calibrated by TMS for accurate comparison.

References:

1 M. Enamullah and W. Linert, Journal of Coordination Chemistry, 1996, 40, 193–201.

2 S. J. Rowan and J. B. Beck, Faraday Discuss., 2005, 128, 43–53.

3 A. W. Addison, S. Burman, C. G. Wahlgren, O. A. Rajan, T. M. Rowe and E. Sinn,

Journal of the Chemical Society, Dalton Transactions, 1987, 2621.

4 C. W. Bielawski and R. H. Grubbs, Prog. Polym. Sci., 2007, 32, 1–29.

191

Chapter 6. Toward High-yield Synthesis of Linear

Polycatenane

1. Introduction

In chapter 4, we successfully synthesized the mixture of linear, branched and cyclic polycatenanes in ~75% yield, which contains ~55% linear polycatenane (ca. 40% yield). However, it is very difficult to isolate the linear polycatenane from its branched and cyclic isomers. Therefore, in chapter 5 we tried to improve the yield of linear polycatenane by using a stronger ligand binding metal ion, namely Fe2+ as the templating metal ion; however, this modification failed to yield any significant linear polycatenanes, and instead resulting in the formation of mostly branched and cyclic polycatenanes (Figure 6.1a). While the exact reason for this interesting result is not known one possible rationale for this difference may be related to reduced solubility of the higher molecular weight Fe2+-based metallosupramolecular polymer (MSP) which causes MSP aggregation in solution and results in branched polycatenanes.

Another significant issue of the current system is for the Zn2+ templated synthesis the DP of linear polycatenane appears to be limited to 8~11, and the DPend-to-end for branched polycatenanes

(the average length from one chain-end to the another) is also limited to 11~13. For Fe2+ templated

2+ synthesis, the DPend-to-end is even lower (4.6~8.5). Conceptually, for the Zn templated synthesis, such a limitation must be the result of either (1) incomplete ring closing reaction and/or (2) the occurrence of the intramolecular side reaction shown in Figure 6.1b. For Fe2+ templated synthesis, backbiting and or other intra-molecular branching also results in short-chain branched

1 polycatenanes with small DPend-to-end. Our results (by monitoring the vinyl protons by H NMR) suggest that we have >97% of the double-bonds of MSP are converted to the 1,2,disubstituted 192

olefin metathesis product. This suggests that the biggest reason for the reduction in the polycatenane DP is related to the side reaction. The desired polycatenane-forming reaction occurs between the end functional groups on the same monomer (① and ②) to form the blue macrocycle.

However, a side reaction where the reaction occurs between the functional groups of two adjacent monomers (② and ③) will result the formation of the linear polymer 17. As discussed in Chapters

3 and 4, the introduction of the red macrocycle 23 helps to significantly reduce the adjacent monomer side reaction from ~60% to ~10%. However, if ~10% of the reaction yields the linear byproduct this will still severely decrease the molecular weight of polycatenane. For example, according to the Carothers equation, DP = 1/(1-p) where p in this case is the percent of monomer converted to the polymer. For the polycatenane reaction, if we assume that that the length on the

MSP is infinite, then p can be thought of as the percentage of monomer ring closed between ① and ②. Based on ~97% metathesis conversion and ~10-15% side reaction, we can estimate that p

≈ 82-87%. It is also worthy to note that in our system, due to the presence of pre-formed cyclic

DiMC 23, each successful ring-closing contributes 2 increment of DP. Therefore, we modify the equation to fit our system as DP = 2/(1-p) ≈ 11-15. While this number is slightly larger than the

Zn DP or DPend-to-end of 22 (7.9~12.8) it is a good first level approximation of the DP in our system.

Figure 6.1c shows schematically how the different side reactions, even if we start from a long MSP precursor, reduce the molecular weight of polycatenane toward the DP of ~10. The difference between the two presumably is related to the fact the MSP is not infinite and there are presumably cyclic MSP species.

193

Figure 6.1 Summary of (a) the Zn-alt-MSP and Fe- alt-MSP templated synthesis of polycatenane, (b) the side-reaction and (c) the shortening effect of the side reaction on the molecular weight of polycatenane.

It is also worth noting that branching adds an addition complicating factor when it comes to this reaction. The branched polycatenane can be presumably obtained by the inter-chain

(side-)reaction of two linear MSPs, where the blue monomer component may form a cyclic dimer

194

or larger cyclic species. For each such inter-chain reaction, the overall DP would on average double during this branching process, but the NC (number of total chain-ends per molecule) will also be doubled. Therefore, to calculate the DP of average linear component of the branched polycatenanes we can use the DPend-to-end (which is the average DP between any two chain ends in a branched polycatenane, calculated by DPend-to-end = 2×DP/NC) as a close estimation. In comparison, the

Zn DPend-to-end of branched 22 fractions is 10.9~12.8, which is slightly larger than the DP of linear

22Zn fractions (7.9~11.2). As has been discussed the in Chapters 4 and 5 this inter chain reaction may be aided by poor solubility (or aggregation) of the MSPs in a bad DCM solvent. Thus the goal for this chapter is to develop a new macrocycle component that (1) eliminates (or at least greatly minimizes) the adjacent monomer side reaction (between ② and ③) and (2) enhances the solubility of the Met(Zn or Fe)-alt-MSP to reduce aggregation in solution.

1.1 Macrocycle Design

In order to suppress the adjacent monomers side-reaction between ② and ③, it is proposed to design and synthesize larger macrocycles (Figure 6.2) that will result in a larger spacing of between the functional groups ② and ③. Moreover, if this larger macrocycle is also more soluble it will be possible to reduce the size of the N-alkyl side chains from butyl back to ethyl. Based on the results in chapter 2, using N-ethyl substituted Bip units on the red macrocycle (instead of the more sterically bulky butyl) will reduce steric hindrance and increase the efficiency of the desired ring closing of monomer 14. Therefore, Figure 6.2b show the conceptual targeted next generation macrocycle with longer linkers between the N-ethyl-Bip moieties, with the proviso that the new

195

macrocycle also needs exceptional solubility to prevent the solubility-related problems found in chapter 5.

Figure 6.2 Suppression of the side reaction by redesigning a larger macrocycle (a) and the design of new generation macrocycle (b).

2. Macrocycle with a Hexaethylene Glycol Linker Moiety

The 2nd generation macrocycle 28 (G2MC) was designed by extending the tetraethylene glycol linker to a hexaethylene glycol linker, while also replacing the butyl side chains to ethyl, as shown in Figure 6.3. The ring-size increased to 80 atoms (up from 68 atoms for the original macrocycle 23), and the long hexaethylene glycol moiety also provides good solubility that

196

overcomes the decreased solubility contribution of the ethyl chain. Due to similar structure, 28 was synthesized using the same chemistry as for the DiMC 23.

Figure 6.3. (a) The structure and (b) the synthesis of G2MC with hexaethylene glycol linker moiety.

With the longer flexible linker moiety, the first concern is whether it can self-bind with the templating metal ion (M2+, either Zn2+ or Fe2+). As shown in Figure 6.4a, if G2MC 28 can bind with metal by itself into a stable complex, then the G2MC-MSP will not be the entropic preferred

2 2+ product, and a significant percentage of 28·Fe and MSP 14n+1·M 2n will be formed. It has already

2+ been shown in chapter 3 that MSP 14n+1·M 2n will be convert mainly to the byproduct linear polymer 17. The UV-Vis spectra of Bip 24, DiMC 23 and G2MC 28 are almost overlapped (Figure

197

6.4b). However, after adding 0.5 equivalent (versus 1 eq. of Bip moiety) of Fe2+ to each compound

2+ 2+ 2+ respectively, 242·Fe and 28·Fe are still overlapped while that of 23·Fe is not. This result indicates that 28 has the ability to form a stable 2:1 Bip: Fe2+ complex by itself and the hexaethylene glycol glycol linker moiety is too long and flexible to be a useful component in the synthesis of polycatenanes by our methodology.

Figure 6.4 (a)Illustration of the self-compromise binding of the G2MC 28; and (b) the UV-Vis 2+ 2+ 2+ spectra comparison of the iron complex of G2MC (28·Fe ), 23·Fe and 242·Fe as well as 23, 24 and 28. The inset shows the structure of 24.

198

3. Designing of Macrocycle with Rigid Bisphenol Z Linker Moiety

The stable self-binding metal ion complex of 28 shows that we need to redesign the macrocycle with a more rigid liner moiety that can prevent the macrocycle from collapsing to from the self-binding 2:1 Bip complex with the metal ion. A well-known easy-obtainable rigid component would be a bisphenol moiety. For example, bisphenol A (Figure 6.5a) has been widely used in polycarbonate materials and epoxy resins to improve their molecular rigidity. Among all the members of bisphenol family, bisphenol Z (BPZ) was selected as a possible base rigid unit of the linker. Compared with bisphenol A, BPZ keeps the same rigid structure but replacing the gem dimethyl unit with a cyclohexane moiety that should help to increase solubility. Two ethylene glycol moieties were connected on both ends of Bisphenol Z to form the semi-rigid linker with an appropriate length. The resulting 3rd generation macrocycle 30 (G3MC, Figure 6.5b), which has a ring size of 76 atoms, between 28 (80 atoms) and 23 (68 atoms). The synthesis of 30 is shown in

Figure 6.5c.

199

Figure 6.5 Stucture of the BPA (a), BPZ (b), the linker synhtesized from BPZ (c) and the molecular structure of the 3rd generation macrocycle with BPZ linker moiety. TEA: triethylamine; DMAP: 4-Dimethylaminopyridine.

During the purification of 30, however, we found that it has limited solubility in a range of solvents, including chloroform, DCM, methanol, ethanol, THF, acetonitrile or any mixtures of them. Luckily, the saturated solution of G3MC has a concentration of ~10 µM in 9:1 v/v

Chloroform/MeCN which is sufficient for the UV-Vis study that is usually performed at 15 µM

(or 30µM for Bip moiety), but 10 µM (20 µM for Bip moiety) will show descent intensity. Before the addition of the metal ion, its UV-Vis absorption spectrum is also similar to that of Bip 24,

200

DiMC 23 and G2MC 28. After adding 0.5 eq (w.r.t. the Bip moiety) of Fe2+ G3MC 29·Fe2+ has a

2+ 2+ different spectrum from either Bip 242·Fe and G2MC 28·Fe , suggesting it has a reduced tendency to for a stable self-binding 2:1 Bip:metal complex. It should be noted that its UV spectrum is also different from 23·Fe2+ the exact reason for this is not know as of yet but may be related to differences in the preferential bind of the system and the fact that due to limited solubility this UV for 30 was carried out under more dilute conditions. Given the limited solubility of 30 we did not pursue this macrocycle more.

Figure 6.6 The UV-Vis spectrum of the iron complex of G3MC (30·Fe2+) compared with 28·Fe2+, 2+ 2+ 23·Fe and 242·Fe . 2+ 2+ 2+ 2+ Concentration: 15 µM for 28·Fe , 23·Fe and 242·Fe and ca. 10 µM for 30·Fe due to saturation. For comparison the intensity of 30·Fe2+ was normalized by concentration for comparison.

201

The data suggests that the BPZ-based linker may be a good candidate, but to confirm this it was important to improve its solubility. To avoid increasing any steric hindrance that may hurt the RCM reaction, it was decided to modify the linker moiety rather than increasing the length of the N-alkyl groups on the Bip. Based on the structure of 29, there are three possible sites for attaching a long alkyl chain (Figure 6.7). Synthetically, accessing BPZ-derivatives with modifications on (a) and (b) are relatively easy while (c) is more of a synthetic challenge. For site

(a) the retro synthetic analysis (Figure 6.8a) shows that it can be easily synthesized in three steps using a variety of commercially available glycidyl ethers, as shown in Figure 6.8b. We chose the

2-ethylhexyl glycidyl ether as an example and successfully synthesized the 4th generation macrocycle (G4MC, 30). However, G4MC has two main drawbacks: (1) due to the presence four chiral carbons on the macrocycle backbone, the product is actually a mixture of multiple chiral isomers that may induces complexity of the corresponding polycatenanes during future study; (2) these chiral centers makes the NMR spectrum of 31 very complicated. Figure 6.9 show the NMR of roughly purified 31 (after one column chromatography and one recrystallization in

CDCl3/MeCN, purity ca. 85%). Due to the presence of eight chiral centers, although the aromatic region (6.7~8.4 ppm) is not impacted, multiple complex peaks show up in the 3.3~4.9 ppm region, corresponding to the glycerin ester moiety (highlighted in yellow). One of the glycerin ether peaks is partially overlapped with the peak on N-ethyl side chain of the Bip (highlighted in blue) which is important for further structural studies. The aliphatic region is also messed up by the chiral carbon (highlighted in green.). Therefore, we think this macrocycle is not a suitable candidate for further studies and further purification was not performed.

202

Figure 6.7 Functionalization possiblities of G3MC for improving solubility.

203

Figure 6.8 The molecular structure of the 4th generation macrocycle and its retro synthetic analysis.

204

Figure 6.9 Full 1H NMR of G4MC 31 (purity: ca.85%).

To avoid these potential problems, the alternative route would be to access the BPZ derivative with the solubilizing chains on site b (Figure 6.7), e.g. the 5th generation macrocycle

(G5MC, 32) as shown in Figure 6.10a. The G5MC can be synthesized from commercially available

2-dodecylphenol in four steps shown in Figure 6.10b, and the full 1H-NMR of 32 is shown in

Figure 6.11 with a purity of ca. 95% after chromatography and recrystallization. By attaching the

205

long alkyl chain on the bisphenol, we totally avoid any chiral centers and this newest generation macrocycle shows exceptional solubility with no observable bad effect on the future studies.

Figure 6.10 The molecular structure (a) and synthesis (b) of the 5th generation macrocycle and its retro synthetic analysis. 3-MPA: 3-mercaptopropionic acid; CTAC: cetyltrimethylammonium chloride.

206

Figure 6.11 Full 1H-NMR of 32 (purity: ca. 95%) with peak assignment.

Some impurity peaks were circled. Measured at 25 °C, CDCl3, 600 MHz.

207

The UV-Vis titration and self-binding studies of 32 is shown in Figure 6.12. UV-Vis titration of 32 with Fe(NTf2)2 (Figure 6.12a) shows an linear increasing of new peak at 392 nm and the decreasing of the unmetalated Bip absorption peak (340 nm) until the ratio of 32:Fe2+ reaches

1:2 (1Bip/Fe2+). Figure 6.12b compares the UV-Vis spectrum of 32, 23, 24 and their 2Bips/Fe2+ complexes. The three unmetalated compounds show very similar UV absorption, but their

2Bips/Fe2+ complexes show very different absorption spectra. The metal-ligand charge transfer

2+ 2+ band (500-650 nm) that indicates 2Bips/Fe binding is clearly shown in 242·Fe but not in any of the two macrocycles. In comparison at 2:1 Bip/Fe2+ ratio, 32·Fe2+ shows higher intensity of unmetallated Bip absorption (340 nm) than 23·Fe2. It can be more clearly observed in Figure 6.12c.

When titrating Fe2+ to 0.5 eq. of metal (vs Bip moiety, light blue dashed line in Figure 6.12c) , the metal/Bip complex peak (392 nm) of 24 stops increasing, indicating the formation of stable

2Bips/Fe2+ complex. DiMC 23 reaches ca. 90% of the final intensity, indicating when one Bip binds with Fe2+ the other Bip is influenced, although not fully bound into 2Bips/Fe2+ complex. In comparison, 32 reaches ca. 50% of the final intensity with approximately linear growth. Therefore, when one Bip moiety of 32 is bound with metal, the other one is almost not impacted, indicating that the di-dodecyl BPZ linker moiety of 32 is more effective in preventing the self-binding than the tetraethylene glycol linker moiety of 23. Therefore, the G5MC 32 with good solubility, large ring size and no self-binding behavior is expected to be a very good macrocycle for designing the next generation system for polycatenane synthesis.

208

Figure 6.12 (a) The UV-Vis titration of 32 with Fe(NTf2)2, (b) comparison of 32, 23, 24 and their 2Bips/Fe2+ complex, and (c) the normalized peak intensity (392 nm, normalized by Bip Moiety concentration) vs Fe2+/Bip ratio. 209

4. Toward the Synthesis of Polycatenanes Containing the 5th Generation

Macrocycle

The assembly of G5MC 32 with monomer 14 and Fe(NTf2)2 into metallo supramolecular polymer (Fe-G5-MSP) was achieved by titration monitored by NMR, as shown in Figure 6.13.

Previously, we were using the peak HB and Hb to adjust the ratio of DiMC 23 and monomer 14.

However, for 32 and 14 their HB and Hb peaks are fully overlapped. It makes sense that we assume the shift of Hb in 23 is due to ring strain applied to the Bip moiety, but the larger ring-size of 32 does not show much strain to the Bip moiety of 32. To achieve the 1:1 stoichiometry of 32 and 14, we can use the peak HI and HJ (on the xanthene moiety of 14) and Hj on the BPZ moiety of 32 to achieve the same goal shown in Figure 6.13. Then the Fe(NTf2)2 solution was titrated, which is monitored by the shift of protons HA, HB and HC (as well as Ha, Hb and Hc), which is analogous to the formation of Zn-alt-MSP. A purple viscous solution was obtained after titration, indicating the formation of Fe-G5-MSP. The diffusion coefficient of the Fe-G5-MSP is measured to be 3.19 ×

10−7 cm2/s (Figure S6.2, supporting information) which between Zn-alt-MSP and Fe-alt-MSP

(3.54 and 1.95 × 10−7 cm2/s, respectively), confirming the polymeric assembly of Fe-G5-MSP.

The reason for smaller Fe-G5-MSP than Fe-alt-MSP may be due to the small amount of impurity

(ca. 5%) in 32 acting as chain-terminator. Further purification of 32 may increase the size of Fe-

G5-MSP, which will be an interesting candidate for future study.

210

Figure 6.13 Assembly of Fe-G5-MSP by titration monitored by 1H NMR.

For titration, 32 and 14 were dissolved in CDCl3 at 20 mg/ml, respectively. Fe(NTf2)2 were dissolved in MeCN-d3 at 60 mg/ml. Measured at 25 °C, 600 MHz.

5. Application Perspective of Polycatenanes

5.1 Mechanical property of a single polycatenane molecule

The mechanical property of a single molecule can be measured by the single-molecule force spectroscopy.1 One example is using an AFM cantilever to apply stress on the molecule and measure the strain, as shown in Figure 6.14).1 The AFM pulling force results in the topological motion of the polycatenane rings. Due to the mechanical interlocking structure, a linear poly[n]catenane molecule is expected to switch between coiled and extended conformations with

211

no bond angle or torsion angle strain. Therefore, the stretching force is expected to be low and fluctuating within a narrow range until the whole chain is fully stretched, while each fluctuation is expected to be correlated to the rotation of a ring. If the ring-rotation movement is turned off by metallating the Bip-containing polycatenane with Zn2+ or Fe2+, the force spectroscopy of the resulting polycatenate is expected be similar to traditional rigid polymers.

Figure 6.14. Prospective mechanical property study of polycatenane by single-molecular force spectroscopy on an AFM cantilever.

5.2 Mechanical properties of polycatenane materials and composites

Traditional mechanical property measurements, such as tensile test or dynamic mechanical analysis (DMA), usually requires at least a few grams of materials to prepare several high-quality films, which is hardly achievable for poly[n]catenane synthesis. With limited amount of materials

(e.g. < 100 mg), however, the mechanical property of bulk polycatenane material can still be measured by advanced techniques such as micro- or nanoindentation2 on a polycatenane-coated holey substrate.

212

Another approach is using polycatenane as a building block for the “composite materials” of polycatenane and traditional non-interlocked polymers. Polycatenanes can be used as one of the blocks of the block-copolymers (Figure 6.15a) or as the crosslinker (Figure 6.15b) to study the effect of mechanical property by introducing the polycatenane into polymer materials. The topological movement can also be tuned by metallation/demetallation of the polycatenane.

Figure 6.15. Composite materials of polycatenane and traditional non-interlocked polymers. Polycatenane can be used as (a) a block of block-copolymers or (b) the crosslinker.

6. Conclusion

In this chapter, we successfully designed and synthesized a series of ditopic macrocycles toward the goal of larger ring size, good solubility and no self-binding. Eventually, the 5th generation macrocycle G5MC 32 meets all the requirements. 32 has a much ring size of 76

(compared to 68 for 1st generation DiMC), four dodecyl chain with exceptional solubility, ethyl side-chain on Bip moiety with no steric hindrance and no chiral center in the whole molecule. UV-

Vis titration study shows 32 has lower tendency to self-bind than DiMC 23. NMR study also confirms the successful assembly of metallosupramolecular polymer Fe-G5-MSP with monomer

14 and Fe2+. Therefore, 32 is an interesting candidate for our future study on improving the yield, degree of polymerization and linear architecture content. 213

7. Supporting Information

7.1 UV-Vis titration of G2MC

Figure S 6.1 UV-Vis titration of G2MC. The four spectra after reaching 2:1 Bip/Fe2+ ratio are overlapped.

214

7.2 DOSY study of Fe-G5-MSP

Figure S 6.2. 1H DOSY of Fe-G5-MSP measured in 1,1,2,2-tetrachloromethane-d2 at 25 °C, 500 MHz.

215

7.3 Experimental

7.3.1 Synthesis of G2MC

G2MC 28 was synthesized in three steps:

Step 1: Both Ethyl Bip 29 and mono-tosylated hexethylene glycol were synthesized according to literature.3 In a 50 mL round bottom flask equipped with stir bar, 3.0 g (5.44 mmol) of 29, 7.1 g (3 eq.) of mono-tosylated hexethylene glycol, 4.5 g (6 eq.) of anhydrous K2CO3 and

27 mL of anhydrous DMF was added. The reaction was stirred under argon atmosphere at 70 °C for 24 hours. After cooling down to room temperature, the solvent was removed under vacuum.

The crude product was triturated with chloroform and filtered to remove salt. The product was purified by column chromatography with chloroform/methanol gradient as the mobile phase. Yield:

216

1 55%, colorless oil. H NMR (600 MHz, CDCl3) δ 8.37 (d, J = 7.8 Hz, 2H), 8.08 (t, J = 7.8 Hz, 1H),

8.03 (d, J = 1.6 Hz, 2H), 7.63 – 7.57 (overlapped, 6H), 7.52 (d, J = 8.4 Hz, 2H), 7.06 – 7.02 (m,

4H), 4.83 (q, J = 7.2 Hz, 4H), 4.21 (t, J = 4.9 Hz, 4H), 3.90 (t, J = 4.9 Hz, 4H), 3.78 – 3.59

(overlapped, 40H), 1.40 (t, J = 7.2 Hz, 6H).

Step2: In a 25 mL flask equipped with stir bar, 1.75 g (1.62 mmol) of the step 1 product

1.48 g (9 eq.) of triethylamine and catalytic amount of N,N-dimethylaminopyridine (DMAP) was dissolved in 6 mL of anhydrous DCM and kept stirring under argon atmosphere in an ice water bath. Then 0.93 g (3 eq.) of tosyl chloride (dissolved in 4 mL of DCM) was added dropwise. The ice was allowed to melt and the reaction solution was kept stirring overnight under argon atmosphere. After removing the solvent, the crude product was purified by column chromatography with chloroform/methanol gradient as mobile phase. Yield: 91%, colorless oil.

1 H NMR (600 MHz, CDCl3) δ 8.38 (d, J = 7.8 Hz, 2H), 8.08 (t, J = 7.9 Hz, 1H), 8.03 (d, J = 1.6

Hz, 2H), 7.81 – 7.77 (m, 4H), 7.62 – 7.57 (overlapped, 6H), 7.52 (d, J = 8.5 Hz, 2H), 7.33 (d, J =

8.0 Hz, 4H), 7.06 – 7.01 (m, 4H), 4.83 (q, J = 7.2 Hz, 4H), 4.20 (t, J = 4.9 Hz, 4H), 4.18 – 4.13 (m,

4H), 3.90 (dd, J = 5.6, 4.1 Hz, 4H), 3.75 (dd, J = 5.9, 3.6 Hz, 4H), 3.73 – 3.56 (overlapped, 32H),

2.44 (s, 6H), 1.41 (t, J = 7.2 Hz, 6H). MALDI-TOF MS: 1388.11 ([M]+H+).

Step3: In a 500 mL two-neck flask equipped with stir bar and a 100 mL additional funnel,

1.10 g (0.79 mmol) of the step 2 product and 0.44 g (1 eq.) of 29 was dissolved in 100mL of anhydrous DMF and added into the addition funnel. 1.03 g (4 eq.) of Cs2CO3 and 100 mL of anhydrous DMF was added directly into the flask. The reaction mixture was stirred at 70 °C under argon atmosphere, then the stopcock of additional funnel was turned on to allow very slow addition over 3 days. The reaction was kept for totally 4 days. After cooling down to room temperature, the solvent was removed under vacuum. The crude product was triturated with chloroform and filtered 217

to remove salt. The product 28 was purified by column chromatography with triethylamine pretreated silica gel and chloroform/methanol gradient as the mobile phase. Yield: 38%, colorless

1 oil. H NMR (600 MHz, CDCl3) δ 8.35 (d, J = 7.8 Hz, 4H), 8.00 (t, J = 7.8 Hz, 2H), 7.98 (d, J =

1.6 Hz, 4H), 7.55 – 7.50 (m, 8H), 7.45 (dd, J = 8.4, 1.6 Hz, 4H), 7.32 (d, J = 8.4 Hz, 4H), 6.99 –

6.94 (m, 8H), 4.65 (q, J = 7.2 Hz, 7H), 4.13 (t, J = 4.9 Hz, 8H), 3.87 (dd, J = 5.6, 4.1 Hz, 8H), 3.74

(dd, J = 6.2, 3.5 Hz, 8H), 3.72 – 3.66 (overlapped, 24H), 1.24 (t, J = 7.2 Hz, 12H). MALDI-TOF

MS: 1594.94 ([M]+H+).

7.3.2 Synthesis of G3MC

G3MC 30 was synthesized in three steps:

Step 1: 10.0 g (37.3 mmol) of BPZ, 18.6 g (4 eq., 149 mmol) of 2-bromoethanol, 30.9 g (6 eq., 224 mmol) of K2CO3 and 75 mL of anhydrous DMF was added into a 125 mL pressure flask equipped with stir bar. The pressure flask was sealed and the mixture was stirred at 100 °C for 24

218

hours. Caution: appropriate safety shield must be used for reaction in glassware under pressure.

After cooling down, the mixture was poured into 750 mL of deionized water. The product was extracted with diethyl ether 3 times, followed by removing solvent under vacuum and column chromatography with hexane/DCM gradient as eluent. Yield: 76%, colorless oil. 1H NMR (600

MHz, CDCl3) δ 7.17 (d, J = 8.5 Hz, 4H), 6.82 (d, J = 8.6 Hz, 4H), 4.06 – 4.02 (m, 4H), 3.95 – 3.90

(m, 4H), 2.24 – 2.18 (m, 4H), 1.54 (q, J = 5.8 Hz, 4H), 1.51 – 1.45 (m, 2H). MALDI-TOF MS:

378.88 ([M]+Na+).

Step 2: In a round bottom flask equipped with stir bar and addition funnel, 5 g (14 mmol) of the step 1 product, 8.5 g of triethylamine (6 eq.) and catalytic amount of DMAP (16 mg, 0.01 eq, 14 mmol) was dissolved in 50 mL of anhydrous DCM in the flask. 5.9 g of TsCl (2.2 eq., 30.8 mmol, dissolved in 40 mL of DCM) was added into the funnel. The flask was stirred in ice water bath under argon atmosphere and the TsCl solution was added dropwise. The ice was allowed to melt and the reaction was stirred overnight. After remove the solvent under vacuum, the product was purified by chromatography with DCM/hexane gradient as eluent. Yield: 95%, colorless oil.

1 H NMR (600 MHz, CDCl3) δ 7.79 (d, J = 7.8 Hz, 4H), 7.31 (d, J = 7.8 Hz, 4H), 7.11 (d, J = 8.4

Hz, 4H), 6.68 (d, J = 8.3 Hz, 4H), 4.33 (t, J = 4.7 Hz, 4H), 4.09 (t, J = 4.8 Hz, 4H), 2.42 (s, 6H),

13 2.18 (t, J = 5.6 Hz, 4H), 1.54 – 1.49 (m, 4H), 1.49 – 1.43 (m, 2H). C NMR (151 MHz, CDCl3) δ

155.72, 145.02, 141.85, 133.02, 129.95, 128.21, 128.12, 114.30, 68.31, 65.48, 45.18, 37.45, 26.48,

22.99, 21.77.

Step 3. In a 500 mL two-neck flask equipped with a 150 mL addition funnel and a stir bar,

1 g (1.5 mmol) of the step 2 product dissolved in 150 mL of DMF was added into the funnel. 0.83 g (1 eq.) of 29, 1.96 g of Cs2CO3 and 150 mL of DMF was added directly into the flask. The reaction mixture was stirred at 70 °C under argon atmosphere, then the stopcock of additional 219

funnel was turned on to allow very slow addition over 3 days. The reaction was kept stirring at

70 °C for another 1 day (4 days total). After cooling down to room temperature, the solvent was removed under vacuum. The crude product was triturated with chloroform and filtered to remove salt. The product 30 was purified by column chromatography with triethylamine pretreated silica gel and chloroform/methanol gradient as the mobile phase. Yield: 11%, white solid. 1H NMR (600

MHz, CDCl3) δ 8.39 (d, J = 7.9 Hz, 4H), 8.08 (t, J = 7.9 Hz, 2H), 8.04 (d, J =1.7 Hz, 4H), 7.64 –

7.61 (m, 8H), 7.59 (dd, J = 8.5, 1.7 Hz, 4H), 7.52 (d, J = 8.4 Hz, 4H), 7.07 – 7.04 (m, 8H), 6.90 –

6.87 (m, 8H), 6.84 – 6.80 (m, 8H), 4.83 (q, J = 7.2 Hz, 8H), 4.36 (dd, J = 6.0, 3.8 Hz, 8H), 4.34 –

4.30 (m, 8H), 2.23 (m, J = 5.8 Hz, 8H), 1.55 (q, J = 5.8 Hz, 8H), 1.52 – 1.46 (m, 4H), 1.41 (t, J =

7.2 Hz, 12H). MALDI-TOF MS: 1742.92 ([M]+H+).

220

7.3.3 Synthesis of G4MC

G4MC 31 was synthesized in three steps:

Step 1: 10.0 g (37.3 mmol) of BPZ, 27.8 g (4 eq., 149 mmol) of 2-ethylhexyl glycidyl ether,

15.5 g (3 eq., 112 mmol) of K2CO3, 12.1g (1 eq., 37 mmol) of Cs2CO3 and 75 mL of anhydrous

DMF was added into a 250 mL round bottom flask equipped with stir bar. The system was flushed with argon and stirred at 80 °C for 24 hours. After cooling down, the mixture was poured into 750 mL of deionized water. The product was extracted with diethyl ether 3 times, followed by removing solvent under vacuum and column chromatography with hexane/ethyl acetate gradient

221

as eluent. Yield: 83%, white solid. 1H NMR (600 MHz, Chloroform-d) δ 7.17 – 7.13 (m, 4H), 6.83

– 6.79 (m, 4H), 4.12 (p, J = 5.4 Hz, 2H), 4.01 – 3.94 (m, 4H), 3.59 – 3.50 (m, 4H), 3.41 – 3.31 (m,

4H), 1.56 – 1.21 (m, 28H), 0.91 – 0.83 (m, 12H). MALDI-TOF MS: 663.58 ([M]+Na+).

Step 2: In a round bottom flask equipped with stir bar and addition funnel, 10 g (15.6 mmol) of the step 1 product, 9.5 g (6 eq. 93.6mmol) of triethylamine and catalytic amount of DMAP (17 mg, 0.01 eq., 0.156 mmol) was dissolved in 60 mL of anhydrous DCM in the flask. 6.5 g of TsCl

(2.2 eq., 34 mmol, dissolved in 40 mL of DCM) was added into the funnel. The flask was stirred in ice water bath under argon atmosphere and the TsCl solution was added dropwise. The ice was allowed to melt and the reaction was stirred overnight. After remove the solvent under vacuum, the product was purified by chromatography with DCM/hexane gradient as eluent. Yield: 88%, colorless oil. 1H NMR (600 MHz, Chloroform-d) δ 7.82 – 7.76 (m, 4H), 7.29 – 7.23 (m, 4H), 7.12

– 7.06 (m, 4H), 6.63 – 6.57 (m, 4H), 4.80 (p, J = 5.0 Hz, 2H), 4.13 – 4.03 (m, 4H), 3.69 – 3.61 (m,

4H), 3.30 – 3.23 (m, 4H), 2.41 (s, 6H), 1.54 – 1.14 (m, 28H), 0.88 – 0.77 (m, 12H).

Step 3. In a 500 mL two-neck flask equipped with a 150 mL addition funnel and a stir bar,

1 g (1.05 mmol) of the step 2 product dissolved in 130 mL of DMF was added into the funnel. 0.58 g (1 eq., 1.05 mmol) of 29, 1.37 g of Cs2CO3 (4eq. 4.2 mmol) and 130 mL of DMF was added directly into the flask. The reaction mixture was stirred at 70 °C under argon atmosphere, then the stopcock of additional funnel was turned on to allow very slow addition over 3 days. The reaction was kept stirring at 70 °C for another 1 day (4 days total). After cooling down to room temperature, the solvent was removed under vacuum. The crude product was triturated with chloroform and filtered to remove salt. The product 31 was purified by column chromatography (with triethylamine pretreated silica gel and chloroform/methanol gradient as the mobile phase) and recrystallized with chloroform/MeCN. Pure product was not isolated. However, the 1H-NMR of 222

purified product (Figure 6.9) and the MALDI-TOF MS (2310.78, [M]+H+) indicates the formation of 31 with a purity of ca. 85% after purification, as white solid.

7.3.4 Synthesis of G5MC 32

G5MC 32 was synthesized in four steps:

Step 1: The 2,2’-dodecyl Bisphenol Z (DBPZ) was synthesized via a literature procedure.4

Briefly, in a 25 mL two-neck round bottom flask equipped with stir bar and condenser, 10 g (38

223

mmol) of 2-dodecyl (purchased from TCI), 202 mg (0.05 eq., 1.9 mmol) of 3-mercaptopropionic acid (3-MPA) and 122 mg (0.01 eq, 0.38 mmol) of cetyltrimethylammonium chloride (CTAC) was added. The flask was heated to 55 °C under argon atmosphere to melt the 2-dodecylphenol, and a homogenous melt was obtained after stirring. Dry HCl gas (prepared by adding concentrated

HCl solution dropwise into solid anhydrous calcium chloride) was bubbled into the melt by a cannula for 1 hour. Then 1.87g (0.5 eq., 19 mmol) of cyclohexanone was added into the melt dropwise. The reaction mixture was stirred at 55 °C under argon atmosphere for 24 hours. After cooling down to room temperature, the crude product was dissolved in diethyl ether, washed with

1M potassium carbonate solution (three times) and water (three times) followed by drying over anhydrous sodium sulfate and filtration. The filtrate was collected and the solvent was removed under vacuum. The crude product was purified by chromatography (silica gel, hexane/dichloromethane gradient as eluent). Yield: 81%, white waxy solid. 1H NMR (600 MHz,

Chloroform-d) δ 6.99 (d, J = 2.4 Hz, 2H), 6.91 (dd, J = 8.4, 2.4 Hz, 2H), 6.64 (d, J = 8.3 Hz, 2H),

2.54 (t, J = 7.7 Hz, 4H), 2.19 (t, J = 5.3 Hz, 4H), 1.58 – 1.49 (m, 10H), 1.34 – 1.20 (m, 36H), 0.88

13 (t, J = 6.9 Hz, 6H). C NMR (126 MHz, CDCl3) δ 150.86, 141.09, 128.92, 127.77, 125.64, 114.81,

45.05, 41.98, 37.46, 31.94, 30.32, 29.86, 29.71, 29.70, 29.68, 29.67, 29.62, 29.55, 29.51, 29.38,

27.02, 26.46, 24.98, 22.96, 22.70, 14.13. MALDI-TOF MS: 626.11 ([M]+Na+).

Step 2: 5 g (8.26 mmol) of DBPZ, 4.13 g (4 eq., 33 mmol) of 2-bromoethanol, 6.9 g (6 eq.,

50 mmol) of K2CO3 and 40 mL of anhydrous DMF was added into a 50 mL pressure flask equipped with stir bar. The pressure flask was sealed and the mixture was stirred at 100 °C for 24 hours.

Caution: appropriate safety shield must be used for reaction in glassware under pressure. After cooling down, the mixture was poured into 500 mL of deionized water. The product was extracted with diethyl ether 3 times, followed by removing solvent under vacuum and column 224

chromatography with hexane/DCM gradient as eluent. Yield: 72%, white waxy solid. 1H NMR

(600 MHz, Chloroform-d) δ 7.04 (d, J = 2.4 Hz, 2H), 6.98 (dd, J = 8.5, 2.4 Hz, 2H), 6.72 (d, J =

8.5 Hz, 2H), 4.04 (t, J = 4.5 Hz, 4H), 3.96 – 3.92 (m, 4H), 2.56 (t, J = 7.7 Hz, 4H), 2.21 (t, J = 5.6

Hz, 4H), 1.56 – 1.45 (m, 10H), 1.34 – 1.20 (m, 36H), 0.88 (t, J = 6.9 Hz, 6H). 13C NMR (126 MHz,

CDCl3) δ 153.84, 141.12, 130.65, 128.89, 125.34, 110.98, 69.10, 61.76, 45.05, 37.38, 31.94, 30.53,

30.07, 29.71, 29.68, 29.58, 29.55, 29.38, 22.96, 22.70, 14.14. MALDI-TOF MS: 692.72 ([M]+H+).

Step 3: In a 50 mL round bottom flask equipped with stir bar and addition funnel, 3 g (4.33 mmol) of the step 2 product, 2.63 g of triethylamine (6 eq., 26 mmol) and catalytic amount of

DMAP (5 mg, 0.01 eq., 0.043 mmol) was dissolved in 15 mL of anhydrous DCM in the flask. 1.82 g of TsCl (2.2 eq., 9.53 mmol, dissolved in 5 mL of DCM) was added into the funnel. The flask was stirred in ice water bath under argon atmosphere and the TsCl solution was added dropwise.

The ice was allowed to melt and the reaction was stirred overnight. After remove the solvent under vacuum, the product was purified by chromatography with DCM/hexane gradient as eluent. Yield:

94%, colorless oil. 1H NMR (600 MHz, Chloroform-d) δ 7.79 (d, J = 8.0 Hz, 4H), 7.31 (d, J = 8.0

Hz, 4H), 7.00 (d, J = 2.4 Hz, 2H), 6.92 (dd, J = 8.6, 2.4 Hz, 2H), 6.59 (d, J = 8.5 Hz, 2H), 4.33 (t,

J = 4.8 Hz, 4H), 4.10 (t, J = 4.8 Hz, 4H), 2.47 – 2.40 (overlapped, 10H), 2.18 (t, J = 5.4 Hz, 4H),

13 1.53 – 1.41 (m, 10H), 1.33 – 1.18 (m, 36H), 0.88 (t, J = 6.9 Hz, 6H). C NMR (126 MHz, CDCl3)

δ 153.41, 144.82, 132.94, 130.92, 129.85, 128.88, 127.93, 125.22, 110.83, 68.28, 65.42, 45.02,

37.35, 31.96, 30.35, 29.93, 29.76, 29.71, 29.60, 29.51, 29.40, 26.43, 22.94, 22.72, 21.66, 14.15.

MALDI-TOF MS: 1000.79 ([M]+H+).

Step 4: In a 1 L two-neck flask equipped with a 250 mL addition funnel and a stir bar, 2.36 g (2.36 mmol) of the step 3 product dissolved in 250 mL of DMF was added into the funnel. 1.15 g (1 eq., 1.5 mmol) of 29, 3.07 g of Cs2CO3 (4eq. 9.4 mmol) and 200 mL of DMF was added 225

directly into the flask. The reaction mixture was stirred at 70 °C under argon atmosphere, then the stopcock of additional funnel was turned on to allow very slow addition over 3 days. The reaction was kept stirring at 70 °C for another 1 day (4 days total). After cooling down to room temperature, the solvent was removed under vacuum. The crude product was triturated with chloroform and filtered to remove salt. The product 32 was purified by column chromatography (with triethylamine pretreated silica gel and chloroform/methanol gradient as the mobile phase) and recrystallized with chloroform/MeCN. Yield: 9%, white solid, ca. 95% purity estimated by 1H

NMR, with full 1H NMR shown in Figure 6.11. Pure product was not isolated. MALDI-TOF MS

2416.45 ([M]+H+) confirms the formation of the product.

Reference:

1 K. C. Neuman and A. Nagy, Nature Methods, 2008, 5, 491–505.

2 C. Lee, X. Wei, J. W. Kysar and J. Hone, Science, 2008, 321, 385–388.

3 R. J. Wojtecki, Q. Wu, J. C. Johnson, D. G. Ray, L. T. J. Korley and S. J. Rowan, Chem.

Sci., 2013, 4, 4440.

3 L. S. Patil, V. S. Suryawanshi, O. B. Pawar and N. D. Shinde, E-Journal of Chemistry,

2011, 8, 2016–2019.

226

Bibliography

Y. H. Lin, Polymer Viscoelasticity: Basics, Molecular Theories, Experiments, and Simulations,

World Scientific, 2011.

Z. Niu and H. W. Gibson, Chem. Rev., 2009, 109, 6024–6046.

Y. Noda, Y. Hayashi and K. Ito, J. Appl. Polym. Sci., 2014, 131, 40509.

Y. Jia and X. Zhu, Chem. Mater., 2015, 27, 387–393.

S. Hackelbusch, T. Rossow, P. van. Assenbergh and S. Seiffert, Macromolecules, 2013, 46,

6273–6286.

J. P. Sauvage and C. Dietrich-Buchecker, Molecular Catenanes, Rotaxanes and Knots, Wiley-

VCH Verlag GmbH, Weinheim, Germany, 1999.

L. Fang, M. A. Olson, D. Benítez, E. Tkatchouk, W. A. Goddard III and J. F. Stoddart, Chem.

Soc. Rev., 2010, 39, 17–29.

H. W. Gibson, M. C. Bheda and P. T. Engen, Prog. Polym. Sci., 1994, 19, 843–945.

E. Wasserman, J. Am. Chem. Soc., 1960, 82, 4433–4434.

G. Gil-Ramírez, D. A. Leigh and A. J. Stephens, Angew. Chemie Int. Ed., 2015, 54, 6110–6150.

J. P. Sauvage and J. Weiss, J. Am. Chem. Soc., 1985, 107, 6108–6110.

P. R. Ashton, T. T. Goodnow, A. E. Kaifer, M. V. Reddington, A. M. Z. Slawin, N. Spencer, J.

F. Stoddart, C. Vicent and D. J. Williams, Angew. Chemie Int. Ed., 1989, 28, 1396–1399.

C. a Hunter, J. Am. Chem. Soc., 1992, 114, 5303–5311.

N. Ogata, K. Sanui and J. Wada, J. Polym. Sci. Polym. Lett. Ed., 1976, 14, 459–462.

J. P. Behr and J. M. Lehn, J. Am. Chem. Soc., 1976, 98, 1743–1747.

A. Harada and M. Kamachi, Macromolecules, 1990, 23, 2821–2823.

227

T. Takata, N. Kihara and Y. Furusho, in POLYMER SYNTHESIS, SPRINGER-VERLAG

BERLIN, Berlin, Heidelberg, 2004, vol. 171, pp. 1–75.

M. Weck, B. Mohr, J.-P. Sauvage and R. H. Grubbs, J. Org. Chem., 1999, 64, 5463–5471.

R. J. Wojtecki, Q. Wu, J. C. Johnson, D. G. Ray, L. T. J. Korley and S. J. Rowan, Chem. Sci.,

2013, 4, 4440.

C. D. Pentecost, K. S. Chichak, A. J. Peters, G. W. V. Cave, S. J. Cantrill and J. F. Stoddart,

Angew. Chemie Int. Ed., 2007, 46, 218–222.

N. Ponnuswamy, F. B. L. Cougnon, J. M. Clough, G. D. Pantos and J. K. M. Sanders, Science,

2012, 338, 783–785.

J.-F. Ayme, J. E. Beves, D. A. Leigh, R. T. McBurney, K. Rissanen and D. Schultz, Nat. Chem.,

2011, 4, 15–20.

K. S. Chichak, S. J. Cantrill, A. R. Pease, S.-H. Chiu, G. W. V. Cave, J. L. Atwood and J. F.

Stoddart, Science, 2004, 304, 1308–1312.

S. J. Cantrill, M. C. T. Fyfe, A. M. Heiss, J. Fraser Stoddart, A. J. P. White and D. J. Williams,

Chem. Commun., 1999, 1251–1252.

C. Meschke, H.-J. Buschmann and E. Schollmeyer, Macromol. Rapid Commun., 1998, 19, 59–

63.

S. Anderson and H. L. Anderson, Angew. Chemie Int. Ed., 1996, 35, 1956–1959.

P. L. Vidal, M. Billon, B. Divisia-Blohorn, G. Bidan, J. M. Kern and J. P. Sauvage, Chem.

Commun., 1998, 629–630.

J. Wu, K. C.-F. Leung and J. F. Stoddart, Proc. Natl. Acad. Sci., 2007, 104, 17266–17271.

S. Kang, B. M. Berkshire, Z. Xue, M. Gupta, C. Layode, P. A. May and M. F. Mayer, J. Am.

Chem. Soc., 2008, 130, 15246–15247. 228

A. Harada, J. Li and M. Kamachi, Nature, 1994, 370, 126–128.

P. R. Ashton, P. T. Glink, C. Schiavo, J. F. Stoddart, E. J. T. Chrystal, S. Menzer, D. J. Williams

and P. a Tasker, Angew. Chemie Int. Ed., 1995, 34, 1869–1871.

X. Tong, P. Gao, X. Zhang, L. Ye, A. Zhang and Z. Feng, Polym. Int., 2010, 59, 917–922.

H. Shigekawa, K. Miyake, J. Sumaoka, A. Harada and M. Komiyama, J. Am. Chem. Soc., 2000,

122, 5411–5412.

H. Fujita, T. Ooya and N. Yui, Macromolecules, 1999, 32, 2534–2541.

V. Balzani, A. Credi and M. Venturi, Chem. Soc. Rev., 2009, 38, 1542.

S. Saha, A. H. Flood, J. F. Stoddart, S. Impellizzeri, S. Silvi, M. Venturi and A. Credi, J. Am.

Chem. Soc., 2007, 129, 12159–12171.

M. Tamura, D. Gao and A. Ueno, Chemistry, 2001, 7, 1390–1397.

A. Harada, J. Li and M. Kamachi, Nature, 1993, 364, 516–518.

Y. Liu, Z.-X. Yang, Y. Chen, Y. Song and N. Shao, ACS Nano, 2008, 2, 554–560.

B. Lewandowski, G. De Bo, J. W. Ward, M. Papmeyer, S. Kuschel, M. J. Aldegunde, P. M. E.

Gramlich, D. Heckmann, S. M. Goldup, D. M. D’Souza, A. E. Fernandes and D. a. Leigh,

Science, 2013, 339, 189–193.

Y. Okumura and K. Ito, Adv. Mater., 2001, 13, 485–487.

K. Ito, Polym. J., 2007, 39, 489–499.

J. Araki and K. Ito, Soft Matter, 2007, 3, 1456.

K. Ito, Curr. Opin. Solid State Mater. Sci., 2010, 14, 28–34.

A. Bin Imran, K. Esaki, H. Gotoh, T. Seki, K. Ito, Y. Sakai and Y. Takeoka, Nat. Commun.,

2014, 5, 5124.

K. Ito, Polym. J., 2012, 44, 38–41. 229

K. Kato, T. Yasuda and K. Ito, Macromolecules, 2013, 46, 310–316.

L. Nissan Motor Co., Nissan Licensing Bussiness, http://www.nissan-

global.com/EN/LICENSE/PDF/technology01.pdf.

L. Nissan Motor Co., Yet2.com Tech of the Week, 2010, September, tow0059972.pdf.

C. Katsuno, A. Konda, K. Urayama, T. Takigawa, M. Kidowaki and K. Ito, Adv. Mater., 2013,

25, 4636–4640.

N. Sugihara, Y. Tominaga, T. Shimomura and K. Ito, Electrochim. Acta, 2015, 169, 433–439.

S. J. Cantrill, G. J. Youn, J. F. Stoddart and D. J. Williams, J. Org. Chem., 2001, 66, 6857–6872.

M. Zhang, S. Li, S. Dong, J. Chen, B. Zheng and F. Huang, Macromolecules, 2011, 44, 9629–

9634.

L. Fang, M. Hmadeh, J. Wu, M. A. Olson, J. M. Spruell, A. Trabolsi, Y. Yang, M. Elhabiri, A.-

M. Albrecht-Gary and J. F. Stoddart, J. Am. Chem. Soc., 2009, 131, 7126–7134.

E. N. Guidry, J. Li, J. F. Stoddart and R. H. Grubbs, J. Am. Chem. Soc., 2007, 129, 8944–8945.

T. Oku, Y. Furusho and T. Takata, J. Polym. Sci. Part A Polym. Chem., 2003, 41, 119–123.

P. Wei, X. Yan and F. Huang, Chem. Commun., 2014, 50, 14105–14108.

T. Sato and T. Takata, Polym. J., 2009, 41, 470–476.

T. D. Shaffer and L.-M. Tsay, J. Polym. Sci. Part A Polym. Chem., 1991, 29, 1213–1215.

D. B. Amabilino, P. R. Ashton, A. S. Reder, N. Spencer and J. F. Stoddart, Angew. Chemie Int.

Ed., 1994, 33, 1286–1290.

N. Watanabe, Y. Ikari, N. Kihara and T. Takata, Macromolecules, 2004, 37, 6663–6666.

N. Watanabe, N. Kihara and T. Takata, Org. Lett., 2001, 3, 3519–3522.

Y. Geerts, D. Muscat and K. Müllen, Macromolecular Chemistry and Physics, 1995, 196, 3425–

3435. 230

J. Weidmann, J.-M. Kern, J. Sauvage, Y. Geerts, D. Muscat and K. Müllen, Chem. Commun.,

1996, 1243–1244.

J.-L. Weidmann, J. Kern, J. Sauvage, D. Muscat, S. Mullins, W. Köhler, C. Rosenauer, H. J.

Räder, K. Martin and Y. Geerts, Chem. Eur. J., 1999, 5, 1841–1851.

D. Muscat, A. Witte, W. Köhler, K. Müllen and Y. Geerts, Macromol. Rapid Commun., 1997,

18, 233–241.

D. Muscat, W. Köhler, H. J. Räder, K. Martin, S. Mullins, B. Müller, K. Müllen and Y. Geerts,

Macromolecules, 1999, 32, 1737–1745.

C. Hamers, O. Kocian, F. M. Raymo and J. F. Stoddart, Adv. Mater., 1998, 10, 1366–1369.

T. Takata, J. Shoji and Y. Furusho, Chemistry Letters, 1997, 881–882.

F. Bitsch, C. O. Dietrich-Buchecker, A. K. Khemiss, J. P. Sauvage and A. Van Dorsselaer, J.

Am. Chem. Soc., 1991, 113, 4023–4025.

M. Okada and A. Harada, Macromolecules, 2003, 36, 9701–9703.

Ö. Ünsal and A. Godt, Chem. Eur. J., 1999, 5, 1728–1733.

A. Bunha, P.-F. Cao, J. Mangadlao, F.-M. Shi, E. Foster, K. Pangilinan and R. Advincula, Chem.

Commun., 2015, 51, 7528–7531.

Y. Gan, D. Dong and T. E. Hogen-Esch, Macromolecules, 2002, 35, 6799–6803.

Y. Ohta, Y. Kushida, D. Kawaguchi, Y. Matsushita and A. Takano, Macromolecules, 2008, 41,

3957–3961.

P.-F. Cao, A. Bunha, J. Mangadlao, M. J. Felipe, K. I. Mongcopa and R. Advincula, Chem.

Commun., 2012, 48, 12094.

P.-F. Cao, J. Mangadlao and R. Advincula, Angew. Chemie Int. Ed., 2015, 54, 5127–5131.

J. F. Stoddart, Chem. Soc. Rev., 2009, 38, 1802. 231

M. Beyler, V. Heitz and J.-P. Sauvage, Chem. Commun., 2008, 5396.

T. J. Hubin and D. H. Busch, Coordination Chemistry Reviews, 2000, 200-202, 5–52.

F. Aricó, J. D. Badjic, S. J. Cantrill, A. H. Flood, K. C.-F. Leung, Y. Liu and J. F. Stoddart, in

Templates in Chemistry II, eds. C. A. Schalley, F. Vögtle and K. H. Dötz, Springer Berlin

Heidelberg, Berlin, Heidelberg, 2005, vol. 249, pp. 203–259.

K. Kim, Chem. Soc. Rev., 2002, 31, 96–107.

I. Aprahamian, O. Š. Miljanic, W. R. Dichtel, K. Isoda, T. Yasuda, T. Kato and J. F. Stoddart,

Bulletin of the Chemical Society of Japan, 2007, 80, 1856–1869.

A. R. Pease, J. O. Jeppesen, J. F. Stoddart, Y. Luo, C. P. Collier and J. R. Heath, Accounts of

Chemical Research, 2001, 34, 433–444.

C. P. Collier, E. W. Wong, M. Belohradský, F. M. Raymo, J. F. Stoddart, P. J. Kuekes, R. S.

Williams and J. R. Heath, Science, 1999, 285, 391–394.

V. Serreli, C.-F. Lee, E. R. Kay and D. A. Leigh, Nature, 2007, 445, 523–527.

J. E. Green, J. Wook Choi, A. Boukai, Y. Bunimovich, E. Johnston-Halperin, E. DeIonno, Y.

Luo, B. a Sheriff, K. Xu, Y. Shik Shin, H.-R. Tseng, J. F. Stoddart and J. R. Heath,

Nature, 2007, 445, 414–417.

K. K. Cotí, M. E. Belowich, M. Liong, M. W. Ambrogio, Y. a Lau, H. a Khatib, J. I. Zink, N. M.

Khashab and J. F. Stoddart, Nanoscale, 2009, 1, 16.

B. K. Juluri, A. S. Kumar, Y. Liu, T. Ye, Y. Yang, A. H. Flood, L. Fang, J. F. Stoddart, P. S.

Weiss and T. J. Huang, ACS Nano, 2009, 3, 291–300.

J. Berná, D. a Leigh, M. Lubomska, S. M. Mendoza, E. M. Pérez, P. Rudolf, G. Teobaldi and F.

Zerbetto, Nature Materials, 2005, 4, 704–710.

T. Pakula and K. Jeszka, Macromolecules, 1999, 32, 6821–6830. 232

F. B. L. Cougnon, N. A. Jenkins, G. D. Pantoş and J. K. M. Sanders, Angew. Chemie Int. Ed.,

2012, 51, 1443–1447.

J. M. Spruell, A. Coskun, D. C. Friedman, R. S. Forgan, A. A. Sarjeant, A. Trabolsi, A. C.

Fahrenbach, G. Barin, W. F. Paxton, S. K. Dey, M. A. Olson, D. Benítez, E. Tkatchouk,

M. T. Colvin, R. Carmielli, S. T. Caldwell, G. M. Rosair, S. G. Hewage, F. Duclairoir, J.

L. Seymour, A. M. Z. Slawin, W. A. Goddard, M. R. Wasielewski, G. Cooke and J. F.

Stoddart, Nat. Chem., 2010, 2, 870–879.

J. D. Crowley, S. M. Goldup, A.-L. Lee, D. a Leigh and R. T. McBurney, Chem. Soc. Rev., 2009,

38, 1530.

J. E. Beves, B. A. Blight, C. J. Campbell, D. A. Leigh and R. T. McBurney, Angew. Chemie Int.

Ed., 2011, 50, 9260–9327.

C. O. Dietrich-Buchecker, A. Khemiss and J. P. Sauvage, Journal of the Chemical Society,

Chemical Communications, 1986, 1376.

J. D. Megiatto and D. I. Schuster, Chem. Eur. J., 2009, 15, 5444–5448.

M. Gupta, S. Kang and M. F. Mayer, Tetrahedron Letters, 2008, 49, 2946–2950.

D. a Leigh, J. K. Y. Wong, F. Dehez and F. Zerbetto, Nature, 2003, 424, 174–179.

J.-C. Chambron, J.-P. Collin, V. Heitz, D. Jouvenot, J.-M. Kern, P. Mobian, D. Pomeranc and J.-

P. Sauvage, European Journal of Organic Chemistry, 2004, 2004, 1627–1638.

J. A. Faiz, V. Heitz and J.-P. Sauvage, Chem. Soc. Rev., 2009, 38, 422–442.

C. Hamann, J. Kern and J. Sauvage, Inorganic Chemistry, 2003, 42, 1877–1883.

C. O. Dietrich-Buchecker and J.-P. Sauvage, Angew. Chemie Int. Ed., 1989, 28, 189–192.

M. J. MacLachlan, A. Rose and T. M. Swager, J. Am. Chem. Soc., 2001, 123, 9180–9181.

P. H. Kwan, M. J. MacLachlan and T. M. Swager, J. Am. Chem. Soc., 2004, 126, 8638–8639. 233

S. Saito, E. Takahashi and K. Nakazono, Org. Lett., 2006, 8, 5133–5136.

V. Aucagne, K. D. Hänni, D. A. Leigh, P. J. Lusby and D. B. Walker, J. Am. Chem. Soc., 2006,

128, 2186–2187.

V. Aucagne, J. Berná, J. D. Crowley, S. M. Goldup, K. D. Hänni, D. A. Leigh, P. J. Lusby, V. E.

Ronaldson, A. M. Z. Slawin, A. Viterisi and D. B. Walker, J. Am. Chem. Soc., 2007, 129,

11950–11963.

D. A. Leigh, P. J. Lusby, S. J. Teat, A. J. Wilson and J. K. Y. Wong, Angew. Chemie Int. Ed.,

2001, 40, 1538–1543.

B. M. McKenzie, A. K. Miller, R. J. Wojtecki, J. C. Johnson, K. A. Burke, K. A. Tzeng, P. T.

Mather and S. J. Rowan, Tetrahedron, 2008, 64, 8488–8495.

X. Fan, J. Yuan, Y. Bai, J. Kong and H. Wu, Acta Crystallographica Section E Structure Reports

Online, 2012, 68, m1072–m1072.

R. J. Wojtecki, M. a Meador and S. J. Rowan, Nature Materials, 2011, 10, 14–27.

M. Burnworth, L. Tang, J. R. Kumpfer, A. J. Duncan, F. L. Beyer, G. L. Fiore, S. J. Rowan and

C. Weder, Nature, 2011, 472, 334–337.

J. R. Kumpfer and S. J. Rowan, J. Am. Chem. Soc., 2011, 133, 12866–12874.

M. Burnworth, S. J. Rowan and C. Weder, Chem. Eur. J., 2007, 13, 7828–7836.

M. Burnworth, D. Knapton, S. J. Rowan and C. Weder, Journal of Inorganic and

Organometallic Polymers and Materials, 2007, 17, 91–103.

J. R. Kumpfer and S. J. Rowan, ACS Macro Letters, 2012, 1, 882–887.

B. M. McKenzie, R. J. Wojtecki, K. A. Burke, C. Zhang, A. Jákli, P. T. Mather and S. J. Rowan,

Chem. Mater., 2011, 23, 3525–3533.

D. A. Leigh, A. Murphy, J. P. Smart, M. S. Deleuze and F. Zerbetto, J. Am. Chem. Soc., 1998, 234

120, 6458–6467.

C. Piguet, G. Bernardinelli, A. F. Williams and B. Bocquet, Angew. Chemie Int. Ed., 1995, 34,

582–584.

D. A. Leigh, R. G. Pritchard and A. J. Stephens, Nat. Chem., 2014, 6, 978–982.

T. Zhao, H. W. Beckham and H. W. Gibson, Macromolecules, 2003, 36, 4833–4837.

S. J. Rowan and J. B. Beck, Faraday Discuss., 2005, 128, 43–53.

O. Kocian, K. W. Chiu, R. Demeure, B. Gallez, C. J. Jones and J. R. Thornback, Journal of the

Chemical Society, Perkin Transactions 1, 1994, 527.

A. J. Caruso and J. L. Lee, J. Org. Chem., 1997, 62, 1058–1063.

Y. Kim, M. F. Mayer and S. C. Zimmerman, Angewandte Chemie, 2003, 115, 1153–1158.

M. Enamullah and W. Linert, Journal of Coordination Chemistry, 1996, 40, 193–201.

A. W. Addison, S. Burman, C. G. Wahlgren, O. A. Rajan, T. M. Rowe and E. Sinn, Journal of

the Chemical Society, Dalton Transactions, 1987, 2621.

N. Bahri-Laleh, R. Credendino and L. Cavallo, Beilstein Journal of Organic Chemistry, 2011, 7,

40–45.

M. P. Sibi and G. Petrovic, Tetrahedron: Asymmetry, 2003, 14, 2879–2882.

A. M. Albrecht-Gary, Z. Saad, C. O. Dietrich-Buchecker and J. P. Sauvage, J. Am. Chem. Soc.,

1985, 107, 3205–3209.

Atta-ur-Rahman and M. I. Choudhary, in Solving Problems with NMR Spectroscopy, Elsevier,

1996, pp. 187–211.

F. Heatley, in NMR Spectroscopy of Polymers, Springer Netherlands, Dordrecht, 1993, pp. 1–49.

L. Li, E. B. Twum, X. Li, E. F. McCord, P. A. Fox, D. F. Lyons and P. L. Rinaldi,

Macromolecules, 2013, 46, 7146–7157. 235

Y. Kim, Bio Design, 2015, 3, 14–25.

T. Yuwen and N. R. Skrynnikov, Journal of Magnetic Resonance, 2014, 241, 155–169.

E. C. Constable, N. Hostettler, C. E. Housecroft, N. S. Murray, J. Schönle, U. Soydaner, R. M.

Walliser and J. A. Zampese, Dalton Transactions, 2013, 42, 4970.

H. B. Tanh Jeazet, K. Gloe, T. Doert, O. N. Kataeva, A. Jäger, G. Geipel, G. Bernhard, B.

Büchner and K. Gloe, Chem. Commun., 2010, 46, 2373.

C. W. Bielawski and R. H. Grubbs, Prog. Polym. Sci., 2007, 32, 1–29.

K. C. Neuman and A. Nagy, Nature Methods, 2008, 5, 491–505.

C. Lee, X. Wei, J. W. Kysar and J. Hone, Science, 2008, 321, 385–388.

236