High-Performance Poly(Butylene Terephthalate) Vitrimers

High-performance poly(butylene terephthalate) vitrimers

Citation for published version (APA): Zhou, Y. (2017). High-performance poly(butylene terephthalate) vitrimers. Technische Universiteit Eindhoven.

Document status and date: Published: 18/12/2017

Document Version: Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement: www.tue.nl/taverne

Take down policy If you believe that this document breaches copyright please contact us at: [email protected] providing details and we will investigate your claim.

Download date: 04. Oct. 2021 High‐performance Poly(Butylene Terephthalate) Vitrimers

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de rector magnificus prof.dr.ir. F.P.T. Baaijens, voor een commissie aangewezen door het College voor Promoties, in het openbaar te verdedigen op maandag 18 december 2017 om 16:00 uur

door

Yanwu Zhou

geboren te Hunan, China

Dit proefschrift van het proefontwerp is goedgekeurd door de promotoren en de samenstelling van de promotiecommissie is als volgt: voorzitter: prof. dr. A.P.H.J. Schenning 1e promotor: prof. dr. R.P. Sijbesma 2e promotor: prof.dr.ir. G.W.M. Peters copromotor(en): dr.ir. J.P.A. Heuts leden: prof. dr. F.E. Du Prez (Universiteit Gent) prof. dr. F. Tournilhac (ESPCI ParisTech) prof.dr.ir. L.E. Govaert adviseur(s): dr.ir. J.G.P. Goossens (SABIC)

Het onderzoek of ontwerp dat in dit proefschrift wordt beschreven is uitgevoerd in overeenstemming met de TU/e Gedragscode Wetenschapsbeoefening.

Dedicated to my family.

Yanwu Zhou

High‐performance Poly(butylene terephthalate) Vitrimers

Eindhoven University of Technology, 2017 This research has received funding from SABIC, Begen op Zoom, the Netherlands.

A catalogue record is available from the Eindhoven University of Technology Library ISBN: 978‐90‐386‐4409‐7 Copyright© 2017 by Yanwu Zhou Cover design by Yanwu Zhou Printed by Gildeprint, The Netherland (www. gildeprint.nl)

Table of contents

Summary ...... 1 Chapter 1. Introduction ...... 5

1.1 Motivation ...... 5

1.2 Melt strength ...... 6

1.3 Vitrimers ...... 8

1.4 Melting and crystallization of PBT ...... 12

1.5 Scope and outline of this thesis...... 14

Chapter 2. Poly(butylene terephthalate)(PBT)/glycerol‐based Vitrimers via solid‐state polymerization ...... 17

2.1 Introduction ...... 18

2.2 Results and discussion ...... 19

2.2.1 Incorporation of glycerol into PBT ...... 20

2.2.2 Thermal properties and morphology during solid‐state (co)polymerization ..... 22

2.2.3 Dynamic thermomechanical properties ...... 24

2.2.4 Rheological behavior ...... 26

2.2.5 Stress relaxation experiments ...... 28

2.2.6 Reprocessing ability of the PBT/glycerol‐based semi‐crystalline vitrimers ...... 30

2.3 Conclusions ...... 31

2.4 Experimental section ...... 31

Chapter 3. Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network ...... 35

3.1 Introduction ...... 36

3.2 Results and discussion ...... 36

3.2.1 Synthesis and characterization of the PBT vitrimers ...... 36

3.2.2 Influence of r on the network dynamics ...... 38

3.2.3 Effect of r on the thermomechanical properties ...... 41

3.2.4 Effect of r on the creep resistance ...... 42

3.3 Conclusions ...... 44

3.4 Experimental section ...... 44

Chapter 4. Influence of the morphology on the physical and mechanical properties of PBT vitrimers ...... 45

4.1 Introduction ...... 46

4.2 Results and discussion ...... 47

4.2.1 Influence of the temperature and residence time on thermal properties ...... 48

4.2.2 Influence of oscillatory stress during cooling from the melt on thermal properties 52

4.2.3 Influence of thermal annealing on the dynamic mechanical properties ...... 53

4.2.4 Influence of annealing on tensile properties and creep resistance ...... 56

4.2.5 Recovering the non‐random distribution of dynamic cross‐links above Tm ...... 57

4.3 Conclusions ...... 59

4.4 Experimental section ...... 60

Chapter 5. Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers ...... 63

5.1 Introduction ...... 64

5.2 Results and discussion ...... 64

5.2.1 Synthesis and thermal properties of PBT/triol‐based vitrimers ...... 65

5.2.2 Dynamic mechanical properties ...... 69

5.2.3 Stress relaxation experiments ...... 70

5.2.4 Model molecules study ...... 72

5.2.5 Welding experiments ...... 76

5.2.6 Creep and mechanical properties ...... 78

5.3 Conclusions ...... 82

5.4 Experiment section...... 83 Chapter 6. In situ network formation in PBT vitrimers via processing‐ induced deprotection chemistry ...... 89

6.1 Introduction ...... 90

6.2 Results and discussion ...... 90

6.2.1 In situ network formation during processing ...... 92

6.2.2 Frequency/temperature‐dependent mechanical properties ...... 94

6.2.3 Stress relaxation experiments ...... 95

6.3 Conclusions ...... 97

6.4 Experimental section ...... 98

Chapter 7. Influence of the network nature on the crystallization/melting behavior of PBT ...... 101

7.1 Introduction ...... 102

7.2 Results and discussion ...... 104

7.2.1 Thermal properties of talc‐nucleated PBT, PBT/PC blend and PBT vitrimers .. 105

7.2.2 Why fast scanning calorimetry ...... 106

7.2.3 Critical cooling rate to suppress crystallization during cooling ...... 107

7.2.4 Isothermal crystallization kinetics ...... 109

7.2.5 Critical heating rate to suppress cold crystallization ...... 113

7.2.6 Critical heating rate to suppress reorganization ...... 115

7.3 Conclusions ...... 117

7.4 Experimental section ...... 117

7.5 Acknowledgements ...... 118

Chapter 8. Epilogue ...... 119

8.1 Conclusions ...... 119

8.2 Technology assessment ...... 122

Bibliography ...... 127

Appendix 1 for Chapter 2 ...... 139

Appendix 2 for Chapter 3 ...... 149

Appendix 3 for Chapter 4 ...... 155

Appendix 4 for Chapter 5 ...... 157

Appendix 5 for Chapter 6 ...... 161

Appendix 6 for Chapter 7 ...... 167

Acknowledgements ...... 169

Curriculum Vitae ...... 173

List of publications ...... 174

Summary

Thermoplastics are light, robust, and easy to process but are less temperature and solvent resistance. Thermosets possess excellent mechanical, thermal integrity and largely elastic response to deformations, however, once the reaction is completed, the polymer cannot be reshaped or reprocessed by heat or with solvent. Vitrimers are a relatively new class of materials which form a "bridge" between traditional thermosets and thermoplastics. These materials are covalently crosslinked, but rather than being permanent, the crosslinks are dynamic and can change its topology without decreasing its connectivity because of the exchange reactions. These dynamic crosslinks lead to a higher dimensional stability and creep resistance at service temperatures and (re)processability of the materials at higher temperatures. Poly(butylene terephthalate) (PBT), belonging to the class of engineering thermoplastics and it has been widely used in injection‐molding applications due to its high crystallization rate and good solvent resistance. However, its high molecular weight between entanglements, makes it inherently have no melt strength, which leads to processing challenges for techniques that involve fast elongational flows, such as blow molding, foaming, blown film extrusion, or sheet extrusion. This thesis focuses on using vitrimer chemistry to create a high‐performance PBT vitrimers via incorporation of polyol (i.e., glycerol, 1,1,1‐Tris(hydroxymethyl)propane, and pentaerythritol) in the solid state in the presence of transesterification catalyst (i.e., zinc(II) acetylacetonate). The goal of the thesis is to understand several aspects of these semi‐crystalline vitrimers: chemistry/catalysis  polymerization/network formation kinetics  vitrimer behavior  crystallization and mechanical properties  scale‐up possibility/potential applications. The development of PBT/glycerol‐based vitrimers via a two‐step solid‐state polymerization strategy is investigated in chapter 2. Reactive extrusion is generally employed to incorporate comonomers, but, the volatility of glycerol at the typically high temperatures makes reactive extrusion (250‐270°C) less suitable and we choose solid‐ state (co)polymerization (SSP), which is usually carried out 20−40 °C below the melng temperature of PBT. The obtained PBT vitrimers showed two plateaus: one is at the temperature between Tg and Tm, which is a crystallinity‐dependent plateau regime, and above Tm, PBT vitrimers showed a rubbery plateau, which is original from the dynamic crosslinked network. Also, these semicrystalline vitrimers can be recycled multiple times by

Summary compression molding without substantial loss of dynamic mechanical and thermal properties. The unique features of these new type of semi‐crystalline vitrimers are that there are two individual triggers to control the vitrimer properties, one is catalytic control of vitrimers’ intrinsic dynamics as reported by Capelot, et al. In chapter 3, we proposed a new design parameter, r, which is the molar ratio between the cross‐linker (glycerol) and Zn(II) catalyst. A tailor‐made semi‐crystalline vitrimer can be built via this design parameter r. We show that the cross‐link density of PBT vitrimers is primarily governed by cross‐linker (glycerol). By contrast, the r, ratio between glycerol and Zn(II) catalyst has a strong influence on both the PBT vitrimer elasticity and stress relaxation time, which enables an independent tuning of the stiffness and rubbery plateau modulus. Creep experiments demonstrated that the PBT has better creep resistance compared to neat PBT at service condition (T < Tm) when the PBT vitrimers possess similar stiffness as PBT. These results indicated tailor‐made PBT vitrimers with tunable network dynamics and viscoelastic properties could be achieved via controlling r, the molar ratio between cross‐linker and [Zn2+]. The other one is to tune the crystalline morphology of PBT vitrimers via physical processes, such as crystallization and thermal annealing, to control the intrinsic mechanical properties. In chapter 4, we show that the dynamic nature of the exchangeable network imparted by Zn2+‐catalyzed transesterification can lead to two main consequences: 1) randomization of the initial blocky chemical microstructure, which deteriorates its thermal, (dynamic) mechanical properties and creeps resistance below melting temperature. Moreover, the randomization is accelerated when an oscillatory stress is applied during the crystallization process. 2) Restoration of the blocky chemical microstructure via thermal annealing or crystallization‐induced dynamic crosslink segregation on cooling from the melt. Therefore, tuning the crystalline/amorphous morphology can serve as a practical way to enhance the thermal, mechanical and creep performance of semi‐crystalline PBT vitrimers. The use of the different transesterification kinetics between glycerol (two primaries and one secondary hydroxyl groups) and 1,1,1‐tris(hydroxymethyl)propane (three primary hydroxyl groups) catalyzed by zinc acetylacetonate to tune the viscoelastic properties of bulk materials through simply altering the cross‐linker are demonstrated in chapter 5. The materials cross‐linked with 1,1,1‐tris(hydroxymethyl)propane showed a fast network build up during SSP, but slow stress relaxation kinetics in the resulting network as compared to the materials cross‐linked with glycerol using the same cross‐linker and Zn(II) catalyst contents. These macroscopic differences in gelation and stress relaxation kinetics were investigated by using small‐molecule kinetic model experiments. The effect of the nature of the cross‐linker on the viscoelastic properties of the PBT bulk material is further

Summary evaluated regarding thermal, thermomechanical, weldability, creep resistance and mechanical properties. Our study demonstrates that the chemical architecture of the cross‐ linker (triol), which in addition to chemical composition, type and content of catalyst, can serve as another design parameter to alter the physical and viscoelastic properties of PBT vitrimers. In chapter 6, we show a new processing‐induced debenzalation process, which shares the same deprotection mechanism as common acid‐promoted debenzalation, with an in situ network formation approach. This process is tunable via playing with the processing temperature and 5,5‐bis(hydroxymethyl)‐2‐phenyl‐1,3‐dioxane (BPO) content. A similar dynamic network, regarding network density, melt elasticity and exchange reaction kinetics, was obtained via both debenzalation strategies for the material containing the same BPO and Zn(II)‐based transesterification catalyst content. This solvent‐free deprotection procedure allows high production rates of PBT vitrimer products via injection molding with the combination of low viscosity during processing and vitrimer characteristics in the final product. The influence of the network properties on the melting and crystallization behavior of poly(butylene terephthalate) was determined for a wide range of cooling/heating rates and isothermal temperatures by fast scanning calorimetry in chapter 7. The critical cooling rate to suppress crystallization is dependent on the network properties, i.e., talc concentration, molecular entanglements, PBT/PC blend, and dynamic vitrimer network density. For isothermal crystallization, the annealing temperature dependence of crystallization peak‐ time for PBT shows a bimodal curve with two minima. The heterogeneous nucleation density can be tuned by the talc concentration, which worked as heterogeneous nuclei. The crystallization via homogeneous nucleation is independent on talc concentration, but the homogeneous nucleation density is influenced by both the physical and vitrimer network. For an amorphous material, only the PBT vitrimer with a rubbery plateau modulus above melting point around 2 MPa can suppress the cold crystallization and subsequent melting process at a heating rate as high as 5000 °Cs‐1. Furthermore, the reorganization during the melting of recrystallized crystals is not influenced by network properties at the heating rate range between 10 or 5000 °Cs‐1. In conclusion, the present work explores both fundamental and applied aspects of poly(butylene terephthalate) vitrimers. We have demonstrated that PBT vitrimers can be synthesized from the relative cheap feedstock, such as glycerol, 1,1,1‐ tris(hydroxymethyl)propane, pentaerythritol, with existing polymerization process with or without the additionally added catalyst. Importantly, PBT vitrimers can be produced with high production rates via injection molding with the combination of low viscosity during processing and vitrimer characteristics in the final product. Products made from PBT

Summary vitrimers can maintain their shape above the melting point of PBT, hence permitting to envision thermal and electrical insulation in an extended temperature range. A property which can be useful to keep the integrity of electrical insulators in case of accidental overheating. High crystallization rate in combination with high melt strength, PBT vitrimers can also be used for reflow soldering process. Furthermore, these insights about PBT vitrimers not only expanded the application area of PBT, but also opened up the possibility for additive manufacturing, and high‐added value application for making dynamic three‐ dimensional (3D) structures capable of reversible shape changes or locomotion without mold, which is appealing for soft robotics, tissue engineering, and deployable devices in aerospace.

Chapter 1

Introduction

1.1 Motivation

Poly(butylene terephthalate) (PBT) is a semi‐crystalline engineering plastic which is mechanically strong and solvent resistant. As a semi‐crystalline polymer with very fast crystallizing characteristics, PBT does not usually require nucleation agents for reducing processing cycle times and, thus, is particularly well suited for injection‐molding applications.1,2 In our daily life, a wide range of applications are known for PBT, such as printed circuit board connectors, automobile parts, devices for the electrical, electronics and medical sector. An overview of the chemical structure of PBT, physical properties and application areas is summarized in Figure 1.1.

PBT

Semi‐crystalline T = 45‐55 °C g T = 220‐230 °C m Crystallinity = 35‐50 % Good solvent resistance High crystallization rate

Figure 1.1. Physical properties and application range of poly(butylene terephthalate) (PBT).

PBT, however, possesses a relatively poor melt strength, and it flows above its melting temperature of approx. 225 °C under its own weight, which will cause melt dripping. This melt dripping behavior significantly limits the use of PBT in high‐temperature applications.3,4 The low melt strength is caused by the high molecular weight between 5 entanglements (Me) of PBT. Commercial PBT, with Mn ranging from 5,000 to 45,000 g/mol, is hardly entangled.5

Introduction

1.2 Melt strength

The lack of melt strength makes PBT difficult to process with certain processing methods that are dominated by extensional flow, e.g., blow molding, foaming, and thermoforming.6 In general, there are several ways to improve the melt strength of thermoplastics.7 For a polymer melt, the relationship between molecular weight (M) and zero‐shear viscosity (0) is well known, as is schematically shown in Figure 1.2.

8 Figure 1.2. The relation between zero shear viscosity (0) and molecular weight (M).

Up to a critical molecular weight (Mc, typically twice the average molecular weight between entanglements (Me)) of a polymer, the viscosity increases linearly with molecular weight (M < Mc; 0  M), but above Mc the viscosity increases exponentially (M > Mc; 0  M3.4).11 Hence, the first method to increase the melt strength is increasing the molecular weight and polydispersity index by a solid‐state (co)polymerization (SSP) method, which is a process conducted below the melting temperature, but above the glass transition temperature of a polymeric material. These reactions utilize the molecular mobility of the chains above the glass transition temperature and polymerization during SSP exclusively takes place in the non‐crystalline part of the material; more specifically, only the mobile amorphous fraction can be modified during the SSP as shown in Figure 1.3.9–11

Chapter 1

Figure 1.3. A representative three‐phase morphology of semi‐crystalline polymers.12–15

For a polycondensate polymer, it is essential to remove the condensate in order to push the reaction forward. Normally, during a SSP process the reactor is continuously flushed with an inert gas (e.g. nitrogen) or kept under reduced pressure for the purpose of removing the formed condensate and the protection of the material against oxidative degradation. SSP is a complex process and various reaction conditions can be tuned in order to get the desired polymeric material, such as temperature,11,16 gas flow rate,11,17,18 gas type,9 particle size and geometry,18,19 degree of crystallinity of the material,20 end‐group content, initial molecular weight,9,11,21 and catalysts.11,22,23 The reactions which take place during SSP can be divided into three classes,10,24 viz. inner‐inner, outer‐inner and outer‐outer reactions. The inner‐inner reactions during SSP are basically ester‐interchange reactions and do not lead to an increase in molecular weight of the sample. The outer‐inner reactions are the alcoholysis or acidolysis of ester‐groups within the backbone by a hydroxyl or carboxylic acid end‐group, respectively. During this process the number average molecular weight of a polymer does not change and no condensate is formed. Both these types of reactions lead to randomization of the copolyester backbone. The only reaction in which a condensate is formed and thus a molecular weight build‐up is observed is the outer‐outer reaction. This reaction proceeds by the (trans)esterification of two end‐groups, where one condensate molecule is formed. The second approach to increase the molecular weight and hence the melt strength is chain extension with bifunctional chain extenders during reactive extrusion or compounding by reacting with its terminal groups (‐OH or ‐COOH), such as diepoxides,25–31 triphenylphosphite,32–35 bis(oxazoline)36–42 and pyromellitic dianhydride.43 For this reactive extrusion approach, the maximum increase in the molecular weight depends on the content of the end groups that react with the chain extender. Taking into account the thermal

Introduction and/or mechanical degradation during the extrusion, increasing the molecular weight of

PBT to above Mc  2Me, is, however, very difficult. A third option is to change the molecular topology, such as introducing branching points. If the branches are few and long enough to entangle, the melt viscosity will be higher than that of a corresponding linear polymer of the same molecular weight due to the strain hardening.44–46 A fourth option is the introduction of multiphasic structures, such as in microphase‐ separated PBT‐based blends1 and elastomers.47 A fifth option is the introduction of cross‐links by, for instance, radiation‐induced cross‐ linking method,48–50 which uses high energy radiation (electron beams from electron accelerators or gamma radiation from Cobalt‐60 sources) to improve the heat resistance of PBT in the solid state via producing reactive species, like cations, anions and free radicals. Conventionally, the permanent network forms a thermoset and they cannot be reshaped, processed or recycled at the end‐of‐life cycle. In this Ph.D. thesis, we focus on using covalent adaptable networks51–53 as a melt strength modification method for PBT, specifically, transesterification‐based vitrimer chemistry.54 In this way, we can create a semi‐crystalline PBT vitrimer that can be used as a thermoset at room temperature but maintaining the processability, weldability and dimensional stability at high temperature (e.g., above Tm). 1.3 Vitrimers

Traditionally, synthetic polymer materials are classified into two major groups: thermosets and thermoplastics. Thermosets have excellent mechanical properties, solvent resistance, abrasion resistance, and load‐bearing capacity, due to their covalently cross‐ linked structure. In contrast, thermoplastics can be molded through injection or extrusion with the application of heat or solution process, while most thermosets must be polymerized in the mold to set the desired shape since they generally lack the ability to be reprocessed after curing. Strong and durable like thermosets, yet moldable and recyclable like thermoplastics, vitrimers form a “bridge” between thermoplastics and thermosets. These materials, pioneered by Leibler and co‐workers,54–56 are a new class of chemically cross‐linked polymers in which thermally stimulated exchange reactions permit the change of the network topology while keeping the number of bonds and cross‐links constant. As shown in Figure 1.4, the vitrimer concept was initially demonstrated via transesterification reactions in a ‐hydroxyl‐ester network synthesized by stoichiometric amounts of epoxy and carboxylic acid groups in the presence of a Zn(II) catalyst. When the temperature is high enough, the exchange reaction pair (hydroxyl‐ester bonds) as indicated by the arrows and circles in the permanent network will be activated. During the exchange

Chapter 1 reactions the reactants and products are the same and the equilibrium constant is a temperature independent unity; only the rate of exchange is temperature dependent rather than the reaction extent.53 This network integrity is also evidenced by the insolubility of the epoxy/fatty acid‐based network in trichlorobenzene at high temperature for a long time, i.e., 180 °C, 16h.

Figure 1.4. Exchange process via transesterification in polyester epoxy resins. Reproduced with permission from reference 52.

The frequency(time)‐dependent mechanical and flow properties of vitrimers are often characterized by rheometry. The characteristic vitrimer behavior is shown in Figure 1.5. As depicted in the illustrative plot of G’ versus frequency, for an ideal thermoset (black solid line), it shows a temperature‐ and ‐independent plateau modulus before degradation, while a vitrimer (blue solid line) exhibits liquid‐like stress relaxation at low frequencies (long time‐scales), and the modulus increases at higher frequencies upon increasing temperature. Moreover, a vitrimer shows a constant plateau modulus because of a constant cross‐link density. This relaxation behavior is also evidenced by stress relaxation experiments, in which an instant strain is applied and the modulus is monitored as a function of time. As can be seen from the illustrative plot of modulus versus time, the ideal thermoset does not relax stress before degradation, in contrast, a vitrimer can totally relax its internal stress via topology rearrangements, and the relaxation is significantly shifted toward shorter time‐scales upon increasing temperature due to the enhanced exchange reaction kinetics.

Introduction

Figure 1.5. G’ versus frequency for an ideal thermoset (permanent network) and vitrimer.53, 54

The volume/viscosity‐temperature characteristics of vitrimers and thermoplastics are shown in Figure 1.6. Upon heating, the vitrimer shows two transition temperatures, a classical glass transition temperature like a thermoplastic (Figure 1.6A) and another topology freezing transition temperature (Tv), where a plastic flow or topology rearrangement is enabled,54 as schematically shown in Figure 1.6B. This topology freezing transition temperature (Tv) is well separated with the classical glass transition temperature 56 (Tg) and can be tuned via catalysis. Vitrimers flow like a vitreous glass above Tv and the viscosity follows a gradual Arrhenius‐type viscosity change with temperature instead of the typical Williams‐Landel‐Ferry (WLF) behavior due to the isodesmic mechanism of transesterification.53 Therefore, the vitrimer can be processed with injection molding at elevated temperature or reshaped by in situ heating without preciously controlling the temperature which is impossible for traditional thermoplastics and thermosets.54

Chapter 1

(A) (B)

Figure 1.6. (A) Volume–T characteristics of a thermoplastic polymer. At T < Tg (Amorphous I), the physical state is conventionally referred to as glass, at T > Tg (Amorphous II) as liquid. (B) Volume–T characteristics of a vitrimer depicting two glass transitions: the classical Tg and Tv, a glass transition that reflects topology freezing upon cooling. Viscosity–T characteristics of (C) a thermoplastic polymer, above Tg, the viscosity of a thermoplastic polymer follows a WLF power law with the temperature and (D) vitrimer, above Tv, the viscosity follows an Arrhenius law in with the temperature. Reprinted with permission from reference 56. Copyright © 2012 American Chemical Society.

Since the discovery of vitrimers by Leibler and coworkers in 2011,54 on one side, research has been directed towards the development of new types of associative exchangeable chemistries, i.e., boronic ester‐based transesterification,57,58 transamination between amines and imines (including vinylogous urethane59,60 and hindered urea61–63), hydroxyl mediated transcarbamoylation,64–66 transalkylation of triazolium67 or sulfonium salts,68 transimination or imine metathesis,69–73 thiol‐disulfide exchange or (aromatic) disulfide metathesis,74–80 siloxane‐silanol exchange reaction,81,82 olefin83,84 and dioxaborolane metathesis.85 In addition, besides the associative exchangeable chemistries, the materials based on dissociative dynamic bonds also can show vitrimer‐like behavior at

Introduction the temperature window where the equilibrium constant is high enough, i.e., Diels‐Alder chemistry,86 hindered urea chemistry87 and trans‐N‐alkylation.88 On the other side, researchers have continuously been looking for practical applications for high‐performance or functional material innovations, i.e., high‐performance commodity (polystyrene (PS), high‐density polyethylene(HDPE) and poly(methyl methacrylate) (PMMA))85 or engineering plastics (polylactide(PLA)89 and poly(butylene terephthalate)(PBT)90,91) with high melt strength and dimensional stability, (3D) dynamic actuators based on liquid‐crystalline elastomers,92,93 shape memory polymers triggered by photo‐94–97 or thermal‐induced plasticity,65,98–101 rubber recycling,102 recyclable, reconfigurable and weldable vitrimer (nano)composites,100,103–109 renewable or bio‐based epoxy vitrimers,110–112 nanoimprint lithograph113 and 3D printing.114 These developments on new chemistry and applications of vitrimer materials were recently reviewed by Du Prez et al.115 and Xie et al.116 1.4 Melting and crystallization of PBT

The crystalline morphology and degree of crystallinity of semi‐crystalline polymers are strongly influenced by the conditions applied during processing, and are of major importance for the final physical and mechanical properties.117 For instance, if a product from a crystalline polymer produced via injection molding is only partially crystallized or forms a thermodynamically unstable mesosphase during molding, it may undergo further crystallization during secondary operations, storage, and shipment or in final use, leading to changes in physical properties. Poor control of crystallization during molding may thus lead to widely varying part‐to‐part properties and dimensional variations. Therefore, it is important to understand how the dynamic cross‐links of the semi‐crystalline PBT vitrimer may influence the melting and crystallization behavior of PBT. In the following part, the literature studies on the melting and crystallization behavior of PBT will be summarized. PBT is one of the semi‐crystalline polymers that shows the well‐known double melting behavior as shown in Figure 1.7.118–123 It is generally accepted that the double melting peaks

(low‐Tm and high‐Tm) observed on the conventional DSC curves are associated to the melting and recrystallization of less perfect lamellar crystals into thicker and more perfect lamellar crystals.15,124–126.

Chapter 1

2.0 T 1.5 c

1.0 Cooling 0.5

0.0

‐0.5 Heating T ‐1.0 m,low Heat flow, exo up [w/g] ‐1.5

Tm,high ‐2.0 0 50 100 150 200 250

Temperature [C] Figure 1.7. Thermograms obtained from the second heating and first cooling run of the PBT.

PBT is the fastest crystallizing polyester, and its crystallization rate is comparable to polyamide‐6 which has a similar melting temperature as PBT.1 The crystal morphology or polymorphism is highly dependent on the cooling rate when the material is cooled from the melt without any disturbance, i.e., in quiescent conditions. PBT has two triclinic crystalline forms, the α‐form and the β‐form (Figure 1.8).127–130 The α‐form crystalline structure is mainly obtained when PBT is crystallized from the molten or the glassy state. The β‐form structure forms when an α‐form crystalline structure is held under stress, and trans‐ forms reversibly to the α‐form on removal of the stress. The c‐axis (fiber axis) length of the α‐ crystalline unit cell, ca. 11.6 Å, is shorter than that of β‐form, ca. 13.0 Å, because the conformations of four methyl groups in α‐and β‐forms are gauche‐trans‐gauche and all‐ trans conformations, respectively. A mesomorphic phase of PBT forms only when an amorphous PBT is stretched at room temperature.131 The mesomorphic phase has been regarded as a smectic structure by the discoverer. The smectic structure transforms into the α‐form upon heating. The length of the smectic periodicity is 11.69 Å, which corresponds to the c‐axis length of α ‐crystalline unit. This formation of mesomorphic phase of PBT is similar to poly(ethylene terephthalate) (PET)8‐10 and poly(ethylene naphthalate) (PEN), but different to poly(butylene naphthalate) (PBN),132 polyamide‐6,6,133 polyamide‐ 11,134 and isotactic polypropylene (iPP)135,136 whose mesomorphic phases are generally formed by rapid quenching from the molten state.

Introduction

Figure 1.8. The polymorphism of PBT. α‐ and ‐form crystalline structure and 2D‐WAXD image of mesophase are reprinted with permission from references 126 and 130, respectively. Copyright©1975 American Chemical Society and Copyright©2010, Rights Managed by Nature Publishing Group. 1.5 Scope and outline of this thesis

The objective of this thesis is to create a new type of high‐performance semi‐crystalline PBT vitrimer with PBT‐like characteristics, i.e., fast crystallization rate, high stiffness, but also with a high melt strength, and thus, a better dimensional stability above the melting temperature. To achieve this aim, a solid‐state (co)polymerization (SSP) method, schematically shown in Figure 1.9, was employed to incorporate multifunctional alcohols (i.e., glycerol, 1,1,1‐tris(hydroxymethyl)propane and pentaerythritol) exclusively into the amorphous region of PBT in the presence of a Zn(II) catalyst. In this way, large crystallizable homo‐PBT blocks are retained. Subsequently, processing‐structure‐property relationships were established on the compression‐molded PBT vitrimers via investigating the influence of network dynamics, chemical microstructure on the thermal, viscoelastic properties, melting and crystallization kinetics.

Chapter 1

Figure 1.9. Schematic representation of a solid‐state (co)polymerization or crosslinking strategy to synthesize PBT/multifunctional alcohol‐based vitrimers with maintained crystallinity.

In Chapter 2, a two‐step solid‐state copolymerization strategy is described to incorporate the volatile glycerol into PBT in the presence of Zn(II) catalyst at constant concentration. The glycerol content is quantified before gelation point by 1H‐NMR spectroscopy. The semi‐crystalline morphology and thermal properties of the obtained copolyester during SSP were followed by (temperature‐modulated) differential scanning calorimetry and the vitrimer behavior of the compression‐molded materials were characterized by dynamic thermal mechanical analysis, stress relaxation experiments and oscillatory frequency sweep. The unique feature of semi‐crystalline PBT vitrimers is that there are two independent handles to control the vitrimer properties. The first one is catalytic control of the intrinsic network exchange dynamics (described in Chapter 3) and the other one is tuning the chemical microstructure in combination with the crystalline/amorphous morphology via crystallization and thermal annealing (described in Chapter 4). In Chapter 3, the efforts are described to tune the PBT vitrimer properties via our proposed network design principle, which is to combine the molar ratio of cross‐linker (glycerol) and Zn(II) in a single parameter, r. For all the PBT vitrimers undergoing the same thermal and mechanical history, the correlation between r and the network density/dynamics, thermo‐mechanical properties and creep resistance is established. Next, the influence of temperature, residence time and oscillatory stress during cooling from the melt on the chemical microstructure are discussed in Chapter 4, physical processes, i.e., thermal annealing and crystallization‐induced segregation of dynamic cross‐links, were employed to restore the blocky chemical microstructure. The influence of the chemical microstructure on the creep and mechanical properties is mapped out. In Chapter 5, the chemical structure effects of different types of cross‐linkers are investigated via comparing 1,1,1‐Tris(hydroxymethyl)propane (TMP) (three primary hydroxyl groups) with glycerol (GLY) (two primary and one secondary hydroxyl groups). A detailed comparison of stress relaxation time, weldability, creep resistance, and mechanical properties between PBT/TMP‐based and PBT/GLY‐based vitrimers with similar cross‐link

Introduction density is carried out. Moreover, two model reaction systems were designed in an attempt to correlate the observed stress relaxation behavior of PBT/TMP‐based and PBT/glycerol‐ based vitrimers to small‐molecule kinetics. In Chapter 6, special attention is paid to the development of a PBT vitrimer with a PBT‐ like production rate by injection molding with the help of protection‐deprotection chemistry. For this purpose, pentaerythritol is first converted to a ketal protected diol (5, 5‐ bis (hydroxymethyl)‐2‐phenyl‐1, 3‐dioxane) (BPO), after incorporation of BPO into PBT backbone in the solid state. An in‐situ network formation in the linear copolyester is achieved via a processing‐induced deprotection process. This new discovery enables one to process the low viscosity (linear or branching) material during injection molding with controllable processing window. The dynamics of the obtained network and possible mechanism are discussed. The in‐situ network formation kinetics are further tuned via playing with the chemical structure of the protection group and the catalyst content. The influence of different protection group and catalyst content on the network dynamics, thermal, thermo‐mechanical and tensile properties are evaluated. In Chapter 7, the effect of supercooling on the crystallization of PBT and PBT vitrimers is investigated by fast‐scanning chip calorimetry (FSC), which permits isothermal measurement of the rate of melt crystallization in a wide range of temperatures from the glass transition temperature (Tg) to the melting temperature (Tm) due to its high cooling capacity and short time constant as compared to conventional differential scanning calorimetry (DSC). Two main research questions are studied in this chapter: i) the influence of the dynamic network on the homogeneous and heterogeneous nucleation of PBT and ii) the influence of the dynamic network on the melting and recrystallization kinetics of PBT. Finally, in Chapter 8, the industrial relevance and technological impact of the results presented in this thesis are described.

Chapter 2

Poly(butylene terephthalate)(PBT)/glycerol‐ based Vitrimers via solid‐state polymerization

ABSTRACT: A new type of semi‐crystalline vitrimer was prepared via the incorporation of glycerol into the amorphous phase of poly(butylene terephthalate) (PBT) by solid‐state (co)polymerization (SSP). A near quantitative incorporation of the glycerol was confirmed by NMR spectroscopy. Wide‐angle X‐ray diffraction, differential scanning calorimetry (DSC) and temperature‐modulated DSC showed that the PBT/glycerol‐based vitrimers maintain the crystallization characteristics of normal PBT. By changing the cross‐link density of the PBT/glycerol‐based vitrimers, a wide range of thermal, rheological and mechanical properties were obtained, e.g., the rubber plateau modulus at 270 °C could be tuned from 0.07 to 3 MPa. The characteristic vitrimer behavior was demonstrated by stress relaxation and oscillatory frequency sweep experiments. In addition, these semi‐crystalline vitrimers can be recycled multiple times by compression molding without a substantial loss in dynamic mechanical and thermal properties.

This chapter was partly reproduced from: Zhou, Y. ; Goossens, J. G. P. ; Sijbesma, R. P.; Heuts, J. P. A. Macromolecules 2017, 50, 6742‐6751.

PBT/glycerol‐based vitrimers via SSP

2.1 Introduction

Vitrimers, pioneered by Leibler and co‐workers,54–56 are a new class of chemically cross‐ linked polymers in which thermally stimulated exchange reactions permit the change of the network topology while keeping the number of bonds and cross‐links constant. The vitrimer chemistry was initially based on transesterification reactions,89,92,94,96,110,137,138 but has been broadened since the vitrimer concept was introduced,59,64,74,75,77,83,84,93,139,140 and was recently reviewed by Du Prez and co‐workers.115 Since their invention, vitrimers have been shown to be useful in a variety of practical applications, such as liquid‐crystalline elastomer actuators,92 multi‐shape memory polymers,96 and rubber recycling.102,141 In the current work, we focus on poly(butylene terephthalate), which is a semi‐ crystalline polyester of great commercial importance and widely used in injection‐molding applications due to its high crystallization rate and good solvent resistance.1 However, 5 commercial PBT, with Mn ranging from 5,000 to 45,000 g/mol, is hardly entangled. Its lacks of entanglements inherently gives the material a low melt strength, leading to processing challenges for techniques that involve elongational flows, such as blow molding, foaming, film or sheet extrusion.6,46 In practice, the melt strength describes the resistance of the polymer to extensional deformation and the melt must have a sufficiently high viscosity to withstand the high strain. In general, there are several ways to improve the melt strength of PBT.7 The first one is increasing the molecular weight and polydispersity index by solid‐ state polymerization or chain extension by reactive extrusion with bifunctional chain extenders, such as diepoxy29 and pyromellitic dianhydride.43 Raising the molecular weight of PBT above its critical molecular weight (Mc) is, however, very difficult. A second option would be to change the molecular topology, such as introducing branching points (if the branches are few and long enough to entangle, the melt viscosity will be higher than that of a corresponding linear polymer of the same molecular weight due to the strain hardening).44 A third option is the introduction of multiphasic structures, such as in (micro)phase‐separated PBT‐based blends1 and thermoplastic elastomers.47 Here, we propose a fourth method via introducing dynamic cross‐links to increase the melt strength, while maintaining processability. In this study, a new type of semi‐crystalline vitrimer is developed by incorporation of glycerol (opened epoxy analogue)1,33 into poly(butylene terephthalate) via a solid‐state (co)polymerization method in the presence of zinc(II) acetylacetonate hydrate as the transesterification catalyst. Often, reactive extrusion is used to incorporate comonomers, as for example done in the reference 33. However, the volatility of glycerol at the typically high temperatures make reactive extrusion less suitable and, therefore, we used solid‐state (co)polymerization (SSP). This has the additional benefit of maintaining a high degree of

Chapter 2 crystallinity when high quantities of cross‐linker are used. The transesterification reaction between the glycerol and the PBT segment in the amorphous phase in the presence of Zn2+ during SSP would initially lead to branching and ultimately to a network. The preparation, morphology and vitrimer behavior of the cross‐linked copolyesters are discussed in this chapter. 2.2 Results and discussion

In general, solid‐state (co)polymerization of PBT is performed with a continuous flow of nitrogen in order to remove the condensate (1,4‐butanediol). However, due to the volatility of glycerol at high temperatures under the continuous nitrogen flow, the evaporation rate of glycerol is much faster than its incorporation rate. In order to optimize the preparation method, we compared the built‐in glycerol concentrations of three different preparation strategies, of which the details are discussed in Appendix 1.

Scheme 2.1. Synthesis of poly(butylene terephthalate)/glycerol‐based vitrimers by solid‐state (co)polymerization (SSP).

The best method is schematically shown in Scheme 2.1 and includes a two‐step (co)polymerization strategy. First, a prepolymerization step was performed at 160 °C in a closed glass vial which was pressurized with inert gas to avoid the evaporation of glycerol.

PBT/glycerol‐based vitrimers via SSP

After the total incorporation of the glycerol (24 h for ca. 13 mol% glycerol and 2 mol%

Zn(acac)2), the mixture was transferred to a SSP reactor, and the reaction was continued at ‐1 180 °C with a N2 flow of 0.5 Lmin . The prepared PBT/GLY‐based copolyesters will be abbreviated as Cx, where x indicates the mol % of glycerol (calculated based on the amount of PBT repeat units). The composition as determined by 1H NMR spectroscopy is used for the abbreviations in this Chapter. For all the compounds studied in this Chapter is catalyzed by 2 mol% Zn Zn(acac)2.

2.2.1 Incorporation of glycerol into PBT

The general observations on the molecular weight evolution as a function of reaction time during the two‐step SSP method are shown in Figure 2.1. During the prepolymerization step at 160 °C, an initial drop in Mn due to chain scission was observed for all the compositions studied in this paper (see Appendix 1 for the other compositions). The chain scission is caused by the alcoholysis of the PBT chains by the free hydroxyl groups from the glycerol. The data in Figure 2.1 show that the molecular weight shifts to lower M and the molecular weight distribution becomes broader after prepolymerization. The Mn drops from 21.2 to 3.4 kg/mol for the material containing 13 mol% glycerol catalyzed by 2 mol% Zn2+, and the Ð is increased from 2.2 (physical mixture) to 2.5 (after prepolymerization at 160 °C for 24 h). No cross‐linked copolyesters were formed at 160 °C after 24 h with 2 mol%

Zn(acac)2.

1.4 0 h 1.2 180C 1 h

1.0 180C 0.5 h 160C 24 h M 0.8

0.6 wlog

0.4

0.2

0.0 2.53.03.54.04.55.05.5 log M

Figure 2.1. Evolution of the molecular weight distribution of C13 as a function of tssp. Molecular weights are reported relative to PMMA standards.

Chapter 2

6.8 kg/mol (tssp = 1 h) and the Ð increased from 2.5 to 3.0. This increase in Mn is a result of the polymer chain recombination by transesterification (polycondensation) that takes place between end groups of the chains with elimination of 1,4‐butanediol. When tssp ≥ 3 h, a cross‐linked copolyester was obtained, which is confirmed by insolubility in 1,1,1,3,3,3‐ hexafluoroisopropanol (HFiP). In general, after tssp = 7 h, all the compositions are cross‐ linked.

Table 2.1. Overview of the PBT/glycerol copolyesters prepared by solid‐state polymerization (SSP).

Glycerol Cross‐linked b c d d e tssp Zn(acac)2 content/mol% Mn Mw copolyesters Entry a /h /mol% tssp  (kg/mol) (kg/mol) Tm/ Tc/ c,heating Feed f f f tgel C C /% PBT 0 0 0 0 21.2 46.6 222 194 40

C2 7 2 2.2 2.2 15.2 48.3 219 192 39

C4 7 2 6.1 4.0 12.7 31.9 216 192 36

C7 3 2 8.9 7.0 13.2 35.6 212 183 36

C13 1 2 14.1 13.0 6.8 17.0 209 173 32

C18 1 2 18.6 18.0 6.4 12.8 205 170 31 a tssp, the sample was taken at the time when tssp  tgel (when the material becomes insoluble in HFiP). b c Zn(acac)2 mole percentage based on the amount of PBT repeat units. The feed compositions and the 1 minimum glycerol content obtained at tssp  tgel were determined using H‐NMR spectroscopy and the ratio between the glycerol/PBT repeat units is expressed in mol% and the uncertainty is estimated to be around 5%. dThe molecular weight of the PBT samples measured with relative to PMMA standards in HFiP. eCross‐linked copolyesters were the powder material directly after SSP for 24 h under 0.5 ‐1 f f Lmin N2 flow. Tm, melting temperature, and Tc, crystallization temperature values are the peak values of the melting endotherms and crystallization exotherms from the second heating and first f cooling run, respectively. χc,heating, degree of crystallinity, was determined by dividing the melting 0 enthalpy (ΔHm) of the heating run (obtained via DSC measurements) by the melting enthalpy (ΔH m) 0 ‐1 147 for 100% crystalline PBT. For ΔH m, values of 142 Jg were used in this study.

In order to determine the glycerol content, NMR chemical shifts of the soluble polymer were compared to a model branched copolyester (poly(glycerol‐co‐terephthalate) (PGT) that was synthesized and characterized via different NMR techniques (see Appendix 1 for further details). Based on the assignments of this model branched PGT and the study by Gross et al.142 integration of the peaks at δ=8.10 ppm (1,4‐phenylene) against the peaks from glycerol at δ=4.72‐4.95 ppm (CH2‐O(C=O)) and δ=6 ppm (CH‐O(C=O) allowed for 21

PBT/glycerol‐based vitrimers via SSP determination of the glycerol composition before the gel point during solid‐state 1 (co)polymerization by H‐NMR in a deuterated solvent mixture (CDCl3:TFA‐d = 4:1). The results are summarized in Table 2.1 and the 1H‐NMR data are presented in Appendix 1. It is evident from Table 2.1 that a high degree of incorporation of the glycerol with this two‐step solid‐state (co)polymerization method is obtained. Since the reaction continues after the gel point, the reported values in Table 2.1 can be considered a lower bound of the real amount of incorporated glycerol.

2.2.2 Thermal properties and morphology during solid‐state (co)polymerization

One of the benefits of the solid‐state (co)polymerization method is that the modification only occurs in the amorphous phase and the crystalline phase is largely retained,143 implying that one can obtain a cross‐linked copolyester with a PBT‐like degree of crystallinity and crystallization behavior. In order to investigate the semi‐crystalline character of the cross‐linked PBT‐based copolyester, DSC and temperature‐modulated DSC (MDSC) were employed to follow the thermal properties (melting temperature (Tm), crystallization temperature (Tc) and degree of crystallinity (χc,heating) and morphology (rigid amorphous fraction (αrigid), mobile amorphous fraction (αmobile) and crystalline fraction (χc,heating)) as a function of tssp, respectively. The uncertainty for the different fractions is about 5%. The DSC and MDSC thermograms are presented in the Appendix 1. The thermograms were analyzed by using the first heating run in order to obtain information on the thermal properties of the copolyesters obtained directly after SSP.

As shown in Figure 2.2A, the onset Tm increases with increasing tssp, which changes from 186 to 209 C for C13 after prepolymerization at 160 C for 24 h and solid‐state (co)polymerization at 180 C for 24 h, respectively. Since the relation between the lamellar 144 thickness and Tm is well known via the Gibbs‐Thomson relation, we can conclude that lamellar thickening occurs simultaneously during the cross‐linking process due to the thermal annealing, and this lamellar thickening was also observed during the chain 145 extension of linear PBT/dianol copolyesters in the solid state by Jansen et al. Tm, Tc, and degree of crystallinity are maintained although the molecular architecture of copolyester changes from a branched (tssp < 3 h) to a cross‐linked (tssp ≥ 3h) structure due to the thermal annealing step. Normally crystal perfectioning does not lead to an increased Tc, but to a 146 sharper melting peak and sometimes a higher melting temperature. Here, the rise of Tc by 15 °C in comparison to the value of Tc at tssp = 0 and 0.5 h is probably more related to the presence of the heterogeneities in the melt in the form of glycerol‐enriched microdomains with a higher cross‐link density that act as nucleation sites. In order to erase the effect of thermal annealing and compare the Tm and χc,heating as a function of glycerol content, the

Chapter 2 data from the second heating run were analyzed. The overview of the influence of glycerol contents on the thermal properties of cross‐linked PBT/glycerol‐based copolyesters are summarized in Table 2.1. The Tm and χc,heating of the obtained cross‐linked copolyester decreases with increasing glycerol contents, and the differences between Tm and χc,heating are about 17 C and 9% for PBT and C18, respectively. However, for Tc, a constant plateau is observed for glycerol content  4 mol%. A sharp decrease of 9 C is observed from C4 to C7, and it reaches another plateau around 170 C when glycerol contents are further increased from C7 to C18.

230 0.6 (A) (B) T 220 m,peak 0.5

c,heating 210 0.4 C]

  200 mobile

[ T c m,onset 0.3 T

, 190 m Tc T Fraction [‐] 0.2 180  0.1 rigid 170 0.0 160 0 5 10 15 20 25 0 5 10 15 20 25 t [hour] t [hour] ssp ssp

Figure 2.2. (A) Development of Tm,Tc as a function of tssp measured by DSC at a heating rate of 10 ‐1 °Cmin from ‐50 to 250 °C. The reported data at tssp= 0 h refer to the material after prepolymerization at 160 °C for 24 h. (B) Morphology development during SSP of C13.

To demonstrate that glycerol is exclusively incorporated into the amorphous phase of

PBT during SSP, the morphology of C13 as a function of tssp is studied as an example. Generally, the morphology of a semi‐crystalline polymer is described with a two‐phase model. This two‐phase model comprises a crystalline and an amorphous phase. As reported by Jansen et al.,143 a small fraction of the amorphous phase is confined by the chain segments in the crystal and is, therefore, not accessible for the transesterification reactions during SSP. Thus, it is more appropriate to describe the morphology of PBT/glycerol‐based copolyester by a three‐phase model,148 in which the amorphous phase is subdivided into a rigid amorphous fraction (rigid) and a mobile amorphous fraction (mobile). The MDSC data in Figure 2.2B show that the rigid is not converted into mobile during prepolymerization at 160 °C, the material still exclusively consists rigid amorphous and crystalline fractions. During the solid‐state (co)polymerization at 180 °C, two processes occur simultaneously due to increased chain mobility. Up to the gel point, one process involves the transformation of rigid into mobile, and the other is the crystal perfectioning by

PBT/glycerol‐based vitrimers via SSP transformation of rigid amorphous chain segments into the crystal phase. After gel point

(tssp > 3 h), the mobile decreases from about 35% to 28% but rigid increases from about 13% to 20%. This divergent phenomenon is due to the limited mobility within the mobile amorphous fraction when a cross‐linked network is formed during the SSP. In order to gain information on the crystal morphology, X‐ray measurements were performed on powder samples directly after SSP and pellet samples directly after compression molding of the powder prepared by SSP. Figure 2.3 shows that α‐crystals are unambiguously identified by the position of the 010, ‐101 and 100 diffraction peaks detected at scattering angles 2θ of 17.19, 20.6 and 23.37°, respectively, on using Cu Kα radiation.149 No co‐crystallization of glycerol with PBT was observed. As expected, the diffraction peaks become broader after compression molding, which is due to a reduced lateral crystal size caused by a lower degree of crystallinity. These results corroborate that all the cross‐linked copolyesters are semi‐crystalline, as observed by DSC.

100 111 010 101,111 011 111 101 104 7 6 5 4

3 2 1 15 20 25 30 35 2 () Figure 2.3. Wide‐angle X‐ray diffraction patterns of obtained cross‐linked copolyesters before and after compression molding. The samples are listed as follows: (1) neat PBT (treated with HFiP); samples obtained directly after SSP: (2) C2, (3) C4, (4) C13; samples obtained directly after compression molding: (5) C2, (6) C4, (7) C13.

2.2.3 Dynamic thermomechanical properties

The thermal stability and insolubility were characterized by TGA and swelling experiments, for details see the Appendix 1. In general, the cross‐linked copolyesters show a similar thermal stability as neat PBT as determined by TGA (Appendix 1). The gel fractions of the compression‐molded samples determined by room‐temperature swelling tests in HFiP increase with increasing glycerol contents (Appendix 1).

Chapter 2

The thermomechanical properties of the compression‐molded samples were characterized by dynamic mechanical thermal analysis (DMTA) and the samples were dried at 120 °C under vacuum for 6 h prior to the test. The DMTA results on the materials with different glycerol contents (see Table 2.1) are presented in Figure 2.4 (only 2, 4 and 13 mol% for clarity reasons) and Appendix 1(7 and 18 mol%). The storage modulus (E) as obtained by DMTA of the neat PBT exhibits the expected behavior of a semi‐crystalline polyester.1

After the glass transition region (Tg), it shows a plateau between Tg and Tm, where the crystalline domains act as physical cross‐links, and its modulus depends on the molecular weight between entanglements (Me) and degree of crystallinity. On continued heating, PBT 1 reaches the terminal region and it flows above Tm. The DMTA curves of the cross‐linked

PBT/glycerol‐based copolyesters are similar to that of PBT below Tm, but above Tm, a second plateau, which originates from the cross‐linked network is observed. From PBT to cross‐ linked semi‐crystalline copolyesters (C18), the plateau moduli (at 270 °C) of the materials increased from 0 to 3 MPa, and the Tg increased from 56 to 82 °C, respectively (Figure 2.4 and Appendix 1). Furthermore, we would like to point out two phenomena: 1) the DMTA curve of 13 mol% glycerol shows a strong temperature‐dependent plateau modulus above melting which is probably caused by the randomization of the chemical microstructure, 2) all cross‐linked semi‐crystalline copolyesters show a drop of modulus in the temperature window between 120 °C and 165 °C. This drop is probably caused by a melting transition, and further investigation is discussed in the Chapter 4. In general, the modulus in the rubber plateau increases and melting transition temperature decreases with increasing glycerol content.

103 Neat PBT 102

101 13 mol% 100

‐1 4 mol%

Storage modulus [MPa] 10 2 mol% 10‐2 0 30 60 90 120 150 180 210 240 270 Temperature [C] Figure 2.4. DMTA curves of neat PBT and cross‐linked copolyester with different amounts of glycerol catalyzed by 2 mol% Zn2+, heating rate = 3 °Cmin‐1 and  = 1 Hz.

PBT/glycerol‐based vitrimers via SSP

2.2.4 Rheological behavior

The dynamic nature of the cross‐linked copolyesters was studied by oscillatory frequency sweep experiments in the molten state. Each sample listed in Table 2.1 was tested at sufficiently low strain amplitudes (1%) to probe only the linear viscoelastic properties. The dynamic storage (G) and loss moduli (G) were measured at various temperatures over an angular frequency range of 500 − 0.01 rad/s. Three representave compositions are presented in Figure 2.5A and the data for C18 and C7 are presented in Appendix 1. It was well documented in the comprehensive review on covalent adaptable networks by Kloxin and Bowman53 that materials formed by permanent covalent bonds display a frequency‐independent storage (G) modulus, while frequency‐dependent G and loss moduli (G) are observed for materials containing covalent adaptable networks. As shown in Figure 2.5A, at 240 °C, for neat PBT, the loss moduli is constantly higher than storage moduli (G > G) in the entire frequency range available and it is right in the terminal regime as shown by the typical G"(ω) ~ ω1 and G'(ω) ~ ω2 variation. When 2 mol% of glycerol is incorporated, the viscous modulus is still greater than the elastic modulus, characteristic of a viscous liquid, showing a much lower resistance to flow consistent with rapid exchange kinetics. These results corroborate the absence of a rubber plateau modulus in DMTA (see

Figure 2.4), where the material above Tm looses its resistance to the small stress around 230 °C; the DMTA experiment (Figure 2.4) is carried out at 1 Hz and C2 exhibits a cross‐over frequency around 12 Hz at 230 °C (Appendix 1). When the glycerol content increases to 4 mol%, the material exhibits a power‐law behavior (G G ωn)150,151 and it is similar to the rheological behavior of reversible Diels‐Alder networks below the gel point.152 A drop‐off in the low‐frequency modulus occurs, and a cross‐over in the elastic and viscous moduli is observed for this sample at 250 °C as will be seen below, demonstrating liquid‐like behavior and relaxation at long timescales (Figure 2.7A). Moreover, the cross‐over frequency of G and G at 240 °C for the sample C4 is beyond the lowest frequency tested (0.0016 Hz), which is several orders of magnitude lower than the ∼25 Hz cross‐over frequency of sample with 2 mol% glycerol. When the glycerol content increases from 4 to 13 mol%, a densely cross‐ linked network is formed and the plateau modulus (G0) increases from 0.006 to 0.2 MPa. The sample exhibits ‐independent G and much smaller G values due to the relative longer relaxation time of the dynamic network at 240 °C than the probed frequency. Thus, in the range of frequencies tested, it behaves like a solid‐like gel (G > G) and shows a plateau modulus (G0), taken at the minimum point of G of around 0.2 MPa. These aforementioned frequency‐dependent mechanical phenomenon are original from the transesterification catalyzed by the rather large amount of catalyst present in the system.

Chapter 2

Above melting temperature, the exchange reaction enables the transfer of a crosslink to a vicinal chain, leading to stress relaxation. The cross‐link does not disappear, but rather 'moves around in the network' until stresses are minimized. A similar frequency‐dependent G and G behavior is also observed for chemical gels containing reversible covalent 70,152–154 bonds. Based on the dynamic nature of the cross‐links above Tm, we called this material a PBT vitrimer.

7 13 mol% (B) 5 10 10 13 mol% 4 mol% 6 10 4 mol% 104 

Pas 10 103 2 mol%  4 1  10 PBT

2 10 3 10 G' (open), G'' (closed) [Pa] PBT 2 mol%

1 2 (A) 2 10 10 ‐2 ‐1 0 1 2 10‐2 10‐1 100 101 102 10 10 10 10 10 rad/s [rad/s] Figure 2.5. (A) Storage (open symbols) and loss (filled symbols) moduli and (B) complex viscosity versus angular frequency for PBT/glycerol‐based vitrimers catalyzed by 2 mol% Zn2+ at 240 °C by 25 mm plate‐plate geometry with 1% strain applied.

One of the aims of this work is to increase the melt strength of PBT, because a higher melt strength is essential for processing techniques such as film blowing, foaming and thermoforming. Normally, the melt strength is characterized by uniaxial extensional viscosity, but there is also a power law relationship between the melt strength and the zero‐ shear viscosity, 0, which is the value taken from the Newtonian plateau from the complex viscosity in the terminal region. As reported by Ghijsels et al.,7,155 the melt strengthis proportional to 0. We can see from Figure 2.5B, although the sample C2 shows a similar rheological behavior as neat PBT (G dominates in the tested frequency region), its complex viscosity at ω = 0.1 rad/s is two orders of magnitude higher than neat PBT at 240 °C. When the glycerol content increases from 4 to 13 mol%, the complex viscosity at ω = 0.01 rad/s increases 3 orders of magnitude at 240 °C, which is expected with increasing cross‐link density. These results are consistent with the DMTA data in Figure 2.4 and indicate that a PBT vitrimer with higher melt strength is obtained at higher glycerol contents in the presence of 2 mol% Zn2+.

PBT/glycerol‐based vitrimers via SSP

2.2.5 Stress relaxation experiments

The time‐ and temperature‐dependent stress relaxation properties of the semi‐ crystalline PBT vitrimers were studied by stress relaxation experiments. The experiments were performed on all of the materials in the linear viscoelastic regime (1% strain) and the 56 stress relaxation times () were determined at Gt/G0 = 0.37 (= 1/e). A typical raw stress relaxation curve (G(t) = (t)/0 ) is shown in Figure 2.6A, for C13 heated at 250 °C. It is clear from this Figure, that although the PBT/glycerol‐based vitrimers are crosslinked, they are able to fully relax stresses (G(t)  0) at longer times, indicating that the network is indeed fully dynamic and that any residual stress is negligible.

1.8 1.0 1.6 G0 1.4 0.8

] 1.2 270 C 230 C 0.6 Pa 1.0 0 5 /G 0.8 t G x 10

[ 0.4 0.6 37% G(t) 0.4 0.2 0.2  (A) (B)  0.0 0.0 10‐2 10‐1 100 101 102 103 100 101 102 103 Time [s] Time [s]

1.0  7 Ea 138 kJ/mol 0.8 (18 mol%)

18 mol% 6 Ea  153 kJ/mol 0 0.6

/G 13 mol%

t (13 mol%) G 7 mol% [s] 5 0.4  37% ln Ea  155 kJ/mol 0.2 4 (7 mol%) (C) 270 C (D) 0.0 3 100 101 102 1.83 1.86 1.89 1.92 1.95 1.98 ‐1 Time [s] 1000/T [K ]

Figure 2.6. (A) Non‐treated stress relaxation curve for C13 catalyzed with 2 mol% Zn(acac)2 catalyst at 250 °C. Normalized stress relaxation curves with respect to (B) temperature with C18 and (C) glycerol content at 270C. (D) Variation of the stress relaxation time versus inversed temperature for Cx catalyzed by 2 mol% Zn2+, x is ca. 7, 13 and 18 mol% glycerol.

Chapter 2

For further analyses the stress relaxation data were normalized by the apparent plateau value G0 as indicated in Figure 2.6A. As is clear from Figure 2.6B the normalized relaxation modulus Gt/G0 decreases as a function of time for C13 at the temperature range from 230 to 270 °C (for complete stress relaxation curves of C7 and C18 as a function of temperature, see Appendix 1). The relaxation is significantly shifted towards shorter time‐scales upon increasing temperature, which enhances the rearrangement reaction kinetics with a decrease in relaxation time (τ) from 814 s at 230 °C to only 57 s at 270 °C. Thus, the semi‐ crystalline vitrimer is able to relax stress at temperature above Tm. Decreasing the cross‐ link density has a similar effect as increasing the temperature (Figure 2.6C); the stress relaxes faster when there are fewer cross‐links. The temperature dependence of the relaxation time can be described by the Arrhenius eq 1.

E  a  exp( ) (1) 0 RT  ‐1 Where τ0 is the pre‐exponential factor (s), Ea is the activation energy (J.mol ), R is the ideal ‐1 ‐1 gas constant (Jmol K ) and T is the temperature (K). The activation energy Ea is determined from the slope of the data in Figure 2.6C. An excellent fit is observed in Figure 2.6C, and comparable activation energies of 155, 153 and 138 kJ/mol were determined from the slopes for the samples containing 7 mol%, 13 mol% and 18 mol% Zn2+, respectively.

104 (A) 6 (B)

C4 Ea  163 kJ/mol 4 103 (4 mol%)

2 [s] 

ln 0 102 250 C E  161 kJ/mol 260 C ‐2 a G' (open), G'' (closed) [Pa]  (2 mol%) c, cross‐over 270 C 101 ‐4 10‐2 10‐1 100 101 102 1.84 1.88 1.92 1.96 2.00 2.04 ‐1 rad/s 1000/T [K ] Figure 2.7. (A) Frequency sweep data for C4, and the cross‐over frequency is indicated by the dotted arrow line. (B) Variation of the average relaxation time versus inverse temperature for samples C2 and C4.

Stress relaxation experiment on the samples with glycerol content below 7 mol%, exhibited remarkably fast relaxation as we can see from the representative normalized stress relaxation curves at 250 °C for C13, C4 and C2 in the Appendix 1. Thus, it is impossible 29

PBT/glycerol‐based vitrimers via SSP to determine the relaxation times by stress relaxation experiments. Here, we use oscillatory frequency sweep tests to characterize the average relaxation time (d) of the samples possessing faster relaxation kinetics. The average relaxation time (d) is determined by the cross‐over frequency (ωc) at G = G , tan  =1 (Figure 2.7A and Appendix 1). The activation energy (Ea) of C2 and C4, as determined from the average relaxation times obtained from frequency sweeps at the cross‐over points (Figure 2.7) is in the same order of magnitude with C13 determined from stress relaxation experiments. However, the correlation between these two techniques needs further investigation.

2.2.6 Reprocessing ability of the PBT/glycerol‐based semi‐crystalline vitrimers

Since the dynamic cross‐links could allow for reprocessing of the materials, the recyclability of the semi‐crystalline vitrimers was tested with the vitrimer sample (C18) containing 18 mol% glycerol and 2 mol% Zn2+ as an example. The sample was cut into small pieces and then compression molded at 250 °C for four times (Figure 2.8A).

(B) 103

102 Neat PBT

101 Original 1xrecycled

100 2xrecycled 3xrecycled Storage modulus [MPa] 4xrecycled 10‐1 0 30 60 90 120 150 180 210 240 270 Temperature [C] Figure 2.8. (A) Recycling of the broken pieces by compression molding and (B) storage modulus of the recycled samples.

Before performing dynamic mechanical thermal analysis (DMTA) measurements, the recycled samples were thermally treated at 120 °C for 6 h under vacuum to remove possible moisture. Afterwards, the moisture‐free recycled samples were subjected to DMTA for comparing the dynamic mechanical properties as a function of cycle number. DMTA

Chapter 2 experiments demonstrate nearly full recovery of the dynamic mechanical properties (see Figure 2.8B). 2.3 Conclusions

Semi‐crystalline vitrimers were obtained by the incorporation of volatile glycerol into PBT via a two‐step solid‐state (co)polymerization protocol. Near quantitative incorporation was confirmed via 1H‐NMR spectroscopy up to the gel point. During the first step of this protocol, i.e., a prepolymerization at 160 °C the glycerol is incorporated into the linear polymer chains after which network formation takes place in the second step, i.e. the solid‐ state polymerization at 180C. DSC revealed that the semi‐crystalline character of PBT was maintained in the PBT/glycerol‐based vitrimers in the solid state and modulated DSC confirmed that the cross‐link points are exclusively located in the amorphous phase. Additionally, WAXD showed that the triclinic α‐crystal of neat PBT was maintained in the PBT/glycerol‐based vitrimers and that no co‐crystallization of glycerol took place. In DMTA, two plateaus were found for this type of semi‐crystalline vitrimers: one between Tg and Tm and which is due to the physical cross‐links from the crystalline part of PBT, and the other one above Tm, i.e. the rubber plateau, which is due to the cross‐links. Tunable rheological behavior, ranging from a typical thermoplastic to a solid‐like gel, and stress relaxation times were achieved by controlling the glycerol content while keeping the Zn2+ concentration constant. Finally, it was found that several recycling steps did not significantly alter the thermomechanical properties of the materials. 2.4 Experimental section

Materials

Special grade poly(butylene terephthalate) pellets (Mn = 21.2 kg/mol, Mw = 46.6 kg/mol) against poly(methyl methacrylate) (PMMA) standards in 1,1,1,3,3,3‐hexafluoroisopropanol (HFiP) were provided by SABIC (Bergen op Zoom, the Netherlands) and used as received.

Glycerol (≥99.5%), zinc(II) acetylacetonate hydrate (Zn(acac)2) (powder), terephthaloyl chloride (≥99%), trimethylamine (≥99.5%), magnesium sulfate (anhydrous, ≥99.5%), potassium nitrate (BioXtra, ≥ 99.0%), sodium nitrate (≥99.0%), and sodium nitrite (≥97.0%) were all obtained from Sigma‐Aldrich. Hydrochloric acid, anhydrous dichloromethane, methanol, 1,1,1,3,3,3‐hexafluoroisopropanol (HFiP, 99%) and MilliQ water (LC‐MS grade) were obtained from Biosolve. Deuterated chloroform (CDCl3, 99.8 atom% D) and deuterated trifluoroacetic acid (TFA‐d, 99 atom% D) were obtained from Cambridge Isotope Laboratories. All chemicals were used as received unless denoted otherwise.

PBT/glycerol‐based vitrimers via SSP

Solution Preparation of a Physical PBT/Glycerol Mixture. Physical mixtures of PBT and glycerol were prepared from solution using a common solvent approach.145,156–163 As a representative example, for the physical mixture containing

14.1 mol% of glycerol and 2 mol% Zn(acac)2, the dried PBT powder (9.14 g, 41.55 mmol), glycerol (0.64 g, 6.92 mmol) and Zn(acac)2 (0.22 g, 0.83 mmol) as the transesterification catalyst were dissolved in 20 mL of HFIP at 55 °C. After complete dissolution of all compounds, the HFIP was distilled off. As soon as the material started to precipitate, a vacuum (p = 10‐2 mbar) was applied for complete removal of HFiP. Finally, the obtained lump residue was dried under vacuum for 24 h at 30 °C, vitrified in liquid nitrogen, and subsequently ground into powder using an IKA A11 Basic Analytical mill. This powder was subsequently dried under vacuum for a period of 24 h at 30 °C. The molecular weight of the polymer in the mixture was then checked to ascertain that no undesirable reactions, such as transesterification or degradation, occurred during the preparation procedure.

Preparation of PBT/glycerol‐based copolyesters. PBT/glycerol‐based copolyesters were synthesized by a two‐step solid‐state (co)polymerization method. Typically, in the first step, 5 g of PBT/GLY powder containing 2 164 mol% Zn(acac)2 was placed in a closed vial, and pressurized with argon (p < 3 bar) to avoid the evaporation of glycerol. After 24 h at 160 °C, the mixture was transferred to the SSP reactor,165 which was a glass tube (inner diameter = 2.4 cm) with a sintered glass plate at the bottom. A heat exchange glass coil (inner diameter = 0.5 mm) surrounded the reactor and entered the inner glass tube at the bottom just below the glass plate. The nitrogen gas was heated by passing through this coil prior to entering the reactor and its flow was controlled by a flow meter. The powder bed was fixed by addition of glass pearls (diameter = 2 mm) on top of the powder, and the reactor was purged with a nitrogen flow of 0.5 Lmin‐ 1 for 30 min. After flushing, the reactor was placed in a heated salt bath (T = 180 °C). When the temperature inside the reactor reached 180 °C, the measurement of the reaction time

(tssp) was initiated, and the composition and molecular weight were followed until the gel point. After the reaction, the product was cooled down to room temperature under a continuous nitrogen flow, discharged from the reactor, and the obtained polymer dried under vacuum at 120 °C for 6 h. The prepared PBT/GLY‐based copolyesters will be abbreviated as Cx, where x indicates the mol% of glycerol (calculated based on the amount of PBT repeat units). The composition as determined by 1H‐NMR spectroscopy is used for the abbreviations in this Chapter.

Characterization methods Proton and Carbon nuclear magnetic resonance spectroscopy. 1H‐ and 13C‐NMR spectroscopy were performed on a 400 MHz Bruker Avance III spectrometer at 25 °C. For

Chapter 2 the 1H‐NMR experiments, the spectral width was 6402 Hz, the delay time was 5 s and the number of scans was 64. For the 13C‐NMR experiments, the spectral width was 24,154 Hz, the delay time was 2 s and the number of scans was between 2000 and 5000. Samples were prepared by dissolving 15‐50 mg of the crude polyester in 0.8 mL of an 80:20 vol % CDCl3: d‐TFA mixture. Chemical shifts are reported in ppm relative to the residual solvent peak of

CDCl3 (δ = 77.0 ppm). Size Exclusion chromatography (SEC). Molecular weight distributions, the number average molecular weight (Mn) and dispersity (Đ) of the copolyesters were measured on a system equipped with a Waters 1515 isocratic HPLC pump, a Waters 2414 refractive index detector (40 °C), a Waters 2707 autosampler, and a PSS PFG guard column followed by a 2PFG‐linear‐ XL (7 μm, 8 × 300 mm) columns in series at 40 °C. HFIP with potassium trifluoroacetate (3 gL−1) was used as eluent at a flow rate of 0.8 mLmin‐1. The molecular weights were determined relative to PMMA standards (Polymer Laboratories, Mp = 580 Da 6 up to Mp = 7.1 × 10 Da). Differential scanning calorimetry (DSC). Thermal properties were measured using a DSC Q1000 from TA Instruments. The measurements were carried out from ‐50 to 250 °C with heating and cooling rates of 10 °Cmin‐1 under a nitrogen flow of 50 mLmin‐1. The mobile amorphous fraction (MAF) was determined via the heat capacity increase at half‐step (cp) in modulated DSC (TMDSC) mode using the same DSC equipment. An oscillating heat flow signal with a period of 60 s and amplitude of 0.5 °C was used with an underlying heating rate of 2 °Cmin‐1. The copolyester samples, prepared by SSP, were measured in the temperature range from 0 to 180 °C. Compression molding. The materials were compression molded at 250 °C and 100 bar for 25 minutes in a Collin Press 300G and subsequently cooled with water. Dynamic mechanical thermal analysis (DMTA). Compression‐molded samples (ca. 10.0 (Length) × 5.0 (Width) × 1.0 (Thickness) mm) were measured on a DMA Q800 (TA Instruments) with a film tension setup. A temperature sweep from ‐50 to 270 °C was performed with a heating rate of 3 °Cmin‐1 at a frequency of 1 Hz. A preload force of 0.01 N, an amplitude of 10 µm and a force track of 125% were used. The storage modulus and loss modulus were recorded as a function of temperature. The glass transition temperature was calculated from the peak maximum in the loss modulus. Rheometry. Dynamic shear measurements were performed on a stress‐controlled AR‐ G2 Rheometer (TA Instruments) by using a 25‐mm parallel plate geometry and disk‐shaped specimens (25 mm diameter; 1 mm thick). Frequency sweeps from 0.01 to 500 rad/s were performed at a temperature range between 230 and 270 °C with a strain of 1%, which is in the linear viscoelastic regime. Stress relaxation experiments were performed at a temperature range between 230 and 270 °C with a strain of 1% and the relaxation modulus 33

PBT/glycerol‐based vitrimers via SSP was monitored as a function of time. A constant normal force of 20 N was applied to ensure a good contact with the plates. Wide‐angle X‐ray diffraction (WAXD). Wide‐angle X‐ray diffraction measurements were performed on a Rigaku Geigerflex Bragg‐Brentano Powder Diffractometer using Cu radiation, wavelength 1.54056 Å, at 40 kV and 30 mA. The scans were performed with 0.02° steps in 2θ and a dwell time of 3‐15 s in the 2θ range from 5° till 40°. The analyses were performed on the powder samples directly after SSP and pellet samples directly after compression molding of the powder prepared by SSP.

Chapter 3

Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network

ABSTRACT: Vitrimers, which form a bridge between thermosets and thermoplastics, are a class of materials with promising opportunities for modern material innovations. Poly(butylene terephthalate) vitrimers that combine the properties of neat PBT and those of adaptable networks are expected to greatly extend the potential applications of this industrially important engineering plastic. The current study aims at building up a tailor‐ made semi‐crystalline vitrimer through understanding the impact of the dynamics and density of the adaptable network on the physical properties of PBT vitrimers. We show that the cross‐link density of PBT vitrimers is almost exclusively governed by the cross‐linker (glycerol) content, whereas the ratio of the glycerol to the Zn(II) catalyst content strongly influences both the elastic and stress relaxation properties. This enables independent tuning of the tensile storage modulus (E‘) and rubbery plateau modulus. The PBT vitrimer exhibits a better creep resistance than neat PBT at service temperature when a similar crystallinity is maintained.

This chapter is partially adapted from: Zhou, Y.; Groote, R.; Goossens, J. G. P.; Sijbesma, R. P.; Heuts, J. P. A. to be submitted.

Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network

3.1 Introduction

Vitrimers54–56 are a new type of covalent adaptable networks based on an associative exchange mechanism.52,53,115,116 These dynamically cross‐linked networks undergo exchange reactions at high temperature permitting a change of the network topology while keeping the number of bonds and cross‐links constant. The network rearrangement and the concomitant stress relaxation for vitrimers are based on the associative exchange mechanism and are dominated by the rate of the chemical exchange reactions.54–56 Although there exist some examples of catalyst‐free vitrimers,59,64,66,85,110 catalysis offers an efficient way to control the vitrimer behavior (e.g., stress relaxation time, weldability and flowability), and a wide range of catalysts, such as organic bases or inorganic salts, have been used in the literature to control transesterification reactions.55,56,60,89,92– 94,100,103,113,138,166–170 Recently, a new type of semicrystalline vitrimer based on poly(butylene terephthalate) (PBT) prepared via solid‐state (co)polymerization91 or reactive extrusion90 was developed. These vitrimer materials exhibited a macroscopic flow behavior that could be changed from a liquid‐like solution to a solid‐like gel when increasing the cross‐link density while maintaining the ratio of catalyst to PBT repeat units constant.91 If we now realize that the network rearrangements, responsible for flow and stress relaxation are mediated by the Zn(II) catalyst, it is logical to assume that the more Zn(II) catalytic centers are available per crosslink, the faster the stress relaxation. We now define the parameter r as the number of Zn(II) catalysts per crosslink. X r  Zn (1) X gly

In this equation, XZn and Xgly are the molar fractions of Zn(II) catalyst and glycerol with respect to the PBT repeat units, respectively. In the current study we intend to correlate (and control) stress relaxation, network density, thermo‐mechanical properties and creep resistance of these semi‐crystalline vitrimers to Xgly and r. 3.2 Results and discussion

3.2.1 Synthesis and characterization of the PBT vitrimers

Two sets of materials were prepared via a two‐step solid‐state (co)polymerization strategy consisting of a prepolymerization step at 160 °C in a closed vial followed by a polymerization step at 180 °C under a nitrogen flow, as reported in our previous work.91 In material set I, Xgly was kept constant (13.2 mol%) while varying XZn from 0 to 2 mol% with r ranging from 0 to 0.15 (= 2/13.2). In material set II, the molar ratio between glycerol and

Zn(II) was kept constant (r  0.015), but XZn and Xgly were simultaneously varied. The 36

Chapter 3 composition of the prepared copolyesters were determined by 1H NMR spectroscopy and the results together with the thermal properties after cross‐linking are summarized in Table 3.1. Detailed information about the molecular weight and experimental conditions is given in the Appendix 2.

Table 3.1. Overview of the PBT/glycerol‐based copolyesters prepared by solid‐state (co)polymerization (SSP) before gelation.

Thermal properties Sample sets Namea rb c d d d Tg (°C) Tc (°C) Tm (°C) c (%)

Neat PBT PBT ‐ 55 194 222 38

2 0.15 C13 80 167 201 32 (2/13.2)

0.2 0.015 I. Varying r: C13 66 181 214 34 (0.2/13.0) Xgly constant, 0.05 0.0038 varying XZn C13 69 182 215 34 (0.05/13.0)

0 0 C13 62 189 217 35 (0/13.0)

0.2 0.015 C13 66 181 214 34 (0.2/13.0)

0.1 0.016 II. Constant r: C6 64 192 221 39 (0.1/6.2) varying Xgly and 0.05 0.016 XZn C3 59 193 222 38 (0.05/3.0)

0.02 0.016 C1 59 194 222 38 (0.02/1.2) a y The material name was abbreviated as Cx , where x and y indicate the mol% of glycerol and Zn(acac)2 with respect to the PBT repeat units, respectively. bThe glycerol contents were determined just before 1 c gelation using H NMR spectroscopy. The glass transition temperature (Tg) was determined from the d peak maximum of the loss modulus. The melting (Tm) and crystallization (Tc) temperature were the peak values of the melting endotherms and crystallization exotherms, respectively. The degree of crystallinity (c) was determined by dividing the melting enthalpy (ΔHmelting) (obtained via DSC 0 0 147 measurements) by the melting enthalpy (ΔH fuse) for 100% crystalline PBT (ΔH fuse = 142 J/g).

Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network

3.2.2 Influence of r on the network dynamics

3.2.2.1 Constant Xgly while varying XZn

The influence of the dynamic cross‐links on the melt rheology of PBT/glycerol‐based vitrimers was studied by oscillatory frequency sweep experiments at sufficiently low strain amplitudes (1%) to probe the linear viscoelastic properties, a 25 mm plate‐plate geometry at various temperatures over an angular frequency range of 100 − 0.01 rad/s was used. 0.2 C13 (XZn  0.2 mol% and Xgly  13 mol%) with r  0.015 from set I was used as the example to demonstrate the time‐temperature dependent dynamic mechanical properties, and the results are shown in Figure 3.1A.

106 1 240 C 0.8 r  0.6

280 C 0.4

0 37% 5

10 /G t 0.2 0.05

G C G' [Pa] 13 C13 0 2 C13 0.2 C13

(A) (B) r = XZn/Xgly 104  10‐2 10‐1 100 101 102 100 101 102 rad/s Time [s] 10 0.05 C13 9 0 C13 8

7 0.2

[s] C13 

ln 6

5 2 C13 4 (C) 3 1.83 1.86 1.89 1.92 1.95 1.98 ‐1 1000/T [K ]

Figure 3.1. Dynamic properties of material set I (constant Xgly, changing r). (A) Frequency dependence 0.2 of the storage modulus G of C13 (r  0.015) at different temperatures. (B) Normalized stress 0.2 relaxation curves at 270 °C. (The insert (top) shows C13 obtained before (left) and after (right) stress relaxation experiments). (C) Arrhenius plot of the stress relaxation times.

Chapter 3

At 240 °C, the storage modulus G is almost independent of the frequency (showing a plateau value of about 0.46 MPa), but upon increasing temperature, a drop in modulus is observed; this drop in modulus is shifted to higher frequencies for increasing temperature. In the high‐frequency region ( 10‐1 rad/s) and independent of the applied temperature, all the samples exhibited a similar storage modulus, which is characteristic of vitrimers because the temperature only influences the exchange kinetics and this behavior is consistent with the dynamically cross‐linked nature of the material.53 The time‐ and temperature‐dependent stress relaxation dynamics of the semi‐crystalline vitrimers were further studied by stress relaxation experiments. Before performing stress relaxation experiments, the compression‐molded samples were subjected to oscillatory time sweeps; subsequently, the stress relaxation experiments were performed when a steady state plateau is reached (for details see Appendix 2). This procedure was deemed to be more reliable than directly measuring on compression‐molded samples, as we learned from our 91 2 previous work that the DMTA curve of the “virgin” compression‐molded specimen of C13 showed a significant increase in storage modulus above the melting temperature. This increase is believed to be partly due to the reorganization of the initial blocky chemical microstructure of the PBT (prepared by SSP) upon thermal treatment and a detailed study of this reorganization is the subject of a separate publication. Representative plots for material set I are shown in Figure 3.1B with the normalized stress relaxation shear modulus versus time at 270 °C and it is immediately clear that the higher the catalyst content (in this case, the higher r), the faster the relaxation. For this set of materials, one dominant relaxation time is observed; the data can be fitted with the Maxwell model with a single characteristic relaxation time, , which is defined as the time it takes to relax to 1/e of the 54 0 initial stress. Interestingly, C13 (the material without added Zn(acac)2) shows the same characteristic stress relaxation behavior as the other materials in material set I. This unexpected dynamic behavior is possibly caused by the presence of residual catalyst, used in the production of PBT. The Arrhenius‐dependent variation of the stress relaxation time as a function of the inverse temperature above the melting temperature of the PBT vitrimer is shown in Figure 3.1C. All of them show a linear fit and an activation energy in the range of 150‐170 kJ/mol(the activation energy for each material is presented in Appendix 2) with 0.05 C13 as the exception with a slightly higher Ea (200 kJ/mol).

3.2.2.2 Simultaneously varying the XZn and Xgly while keeping r constant

The results of analogous tests on material set II, in which the XZn and Xgly were simultaneously varied while keeping r  0.015, and the results are presented in Figure 3.2. The frequency sweep experiments shown in Figure 3.2A illustrate that all the materials show solid‐like gel characteristics at 250 °C (above Tm), with G > G in the experimental

Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network

‐2 2 0.02 angular frequency range of 10 ‐10 rad/s and the data of C1 suggests much faster exchange dynamics than the other materials. No terminal relaxation regime was observed at 250 °C for any of the studied compositions.

106 1 C 0.2 0.2 13 C13

0.1 0.1 0.05 C C C3 6 6 37% 105 0.05 0 C3 Experimental data /G t G 0.02 C1

0.1 4 10 C 0.02 Maxwell model fitting G' (open), G'' (closed) [Pa] 1 Decreasing X (A) gly (B)  10‐2 10‐1 100 101 102 100 101 102 103 rad/s Time [s]

0.1 8 C6

7 0.2 C13 [s]

 6

ln 0.05 C3 5

(C) 4 1.83 1.86 1.89 1.92 1.95 1.98 ‐1 1000/T [K ] Figure 3.2. Dynamic properties of material set II. (A) Frequency dependence of the storage modulus G (open symbols) and loss modulus G (filled symbols) at 250 °C. (B) Normalized stress relaxation curves at 250 °C. (C) Arrhenius plots of the stress relaxation times.

The stress relaxation behavior depicted In Figure 3.2B shows very similar relaxation 0.02 times for this set with similar r except for the material containing low Xgly ( 1 mol%); C1 shows a nearly one order of magnitude lower relaxation time than the other materials from set II at 250 °C. It is important to note that in this case it is not possible to fit the stress relaxation curve with a single characteristic relaxation time according to the Maxwell model, implying that more than one mechanism for stress relaxation is involved and that the relaxation is not only determined by the exchange reaction rates. Although the 0.02 relaxation time of C1 is much lower than the other materials in set II, i.e, the relaxation

Chapter 3

0.02 2 3 time at 250 °C for C1 is 10 s and 10 s for the others, all materials have a larger relaxation time than that observed in neat PBT (< 10‐1 s)(Appedix 2). In Figure 3.2C, the 0.02 Arrhenius plots for all relaxation times except for those of C1 are shown and the activation energies, Ea, for these materials are lie also in the narrow range of 140‐170 kJ/mol, which is in a good agreement with the Ea values of vitrimers from set I, and our 90,91 previous work. An overview of Ea and pre‐exponential factor (A or 1/τ0) of the different systems is given in Appendix 2. To summarize, the storage modulus (G) of PBT/glycerol‐ based vitrimers is controlled by the glycerol content, while the dynamics of the exchangeable networks are controlled by r.

3.2.3 Effect of r on the thermomechanical properties

The thermomechanical properties of the prepared PBT vitrimers were characterized by using dynamic mechanical thermal analysis (DMTA) and the samples were annealed at 200 °C under vacuum for 6 h prior to the test. In our previous paper on PBT vitrimers,91 we showed that the DMTA curves of the PBT vitrimers are similar to that of PBT below Tm, but above Tm, a rubber plateau, which originates from the dynamic cross‐linked network, was observed. In Figure 3.3, the results are shown for the current sample sets and a similar behavior is observed.

set I set II 3 3 10 0.05 10 PBT C13

102 102 0.1 C6 0.2 C 0.2 C13 13 1

[MPa] 1 [MPa]  10  10 E E 2 C13 0 0 0.1 0 10 10 C3 C13 C 0.02 (A) ‐1 (B) PBT 1 10‐1 10 0 30 60 90 120 150 180 210 240 270 0 30 60 90 120 150 180 210 240 270 Temperature [C] Temperature [C] Figure 3.3. DMTA curves of sample set I (A) and sample set II (B). Heating rate = 3 °Cmin‐1 and  = 1 Hz. The dotted line indicates the temperature used for creep experiments (see below).

2 The results for sample set I are shown in Figure 3.3A and only C13 shows a different behavior: two clear melting transitions above Tg are observed, while the other materials exhibit only one melting transition. We believe that this second melting transition is caused by a network reorganization that leads to a different crystal morphology and is the subject of a further study. The moduli in the rubber region for material set I are approximately the same within the narrow range of 3‐5 MPa at 270 °C and are relatively constant except for

Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network

the entropic temperature dependence. Upon proportionally decreasing XZn and Xgly, in material set II (Figure 3.3B), the rubber plateau modulus decreases. Let us now consider the tensile modulus (E‘) of PBT vitrimers in the temperature region between Tg and Tm. From Figure 3.3A, it is clear that E at 180 °C increases with decreasing r (decreasing XZn while keeping Xgly constant). Since it is well known that the E‘ of semi‐ crystalline materials in this temperature range is controlled by the degree of crystallinity 171 (c), we compared the c for these materials (Table 3.1 and Appendix 2). In general, the

c of vitrimers is lower than the c of neat PBT, which is consistent with the lower E' values for the vitrimers. What cannot be explained by simple consideration of c are the differences in E' for the various vitrimers, as all display the same c. The DMTA curves for sample set II are presented in Figure 3.3B. It is clearly seen that the 0.2 materials in set II, except for C13 , have a similar E‘ as PBT and a similar c (Table 3.1). As 0.2 can be seen from Table 3.1, C13 has a much lower c and is similar to all materials from set I (all the materials contain approximately 13 mol% glycerol). These results are all consistent with the dependence of E' on c and that high glycerol concentrations lead to 91 lower c, which is in good agreement with our previous work. The data shown in Figure

3.3A are now also more easily explained. The lower c of the vitrimers, in the first instance, decreases E‘ with respect to PBT, but the introduction of (dynamic) crosslinks leads to an increase in E‘. Since the crosslinks are dynamic and not permanent it is then expected that the more dynamic the network (i.e., the higher r), the lower the resistance to deformation and hence the lower E'. This is exactly what is observed. Hence we can conclude that it is possible to control the stiffnes ( E') between Tg and Tm by varying c, through the cross‐ linker content (Xgly), and the dynamics of the network, through r. To summarize this section, we can conclude that we can produce PBT vitrimers with tunable stiffness and rubber plateau.

3.2.4 Effect of r on the creep resistance

Creep is one of the most important viscoelastic properties of a polymer material.172 A potential advantage of PBT vitrimers is that they may show a better creep resistance at service temperature (e.g., T < Tm), while maintaining a high rate of the exchange reactions at a processing temperature above Tm. In creep experiments on the PBT vitrimers, a rectangular‐shape sample (the same as for the DMTA test) was first heated to 125 °C, kept isothermal for 30 min, and subsequently a 2 MPa step stress was applied for a period of 60 min. After 60 min, the stress was removed, and the sample was allowed to recover for another 80 min.

Chapter 3

2.5 2.5 (A) 2 Maximum strain set I (B) set II C13 2.0 Creep rate 2.0 0.2 C 0.2 C 1.5 13 1.5 13

% 0.1 0.05 C n PBT C 6 i 13 1.0 1.0 0.05 tra C3 Strain % S PBT 0 0.5 C13 0.5 C 0.02 Irrecoverable strain 1 0.0 0.0 0 30 60 90 120 150 0 306090120150 Time [min] Time [min] Figure 3.4. Creep‐recovery experiments at 125 °C on PBT vitrimers: (A) with different r (set I) and (B) with the same r (set II).

The results of these experiments are shown in Figure 3.4, the maximum strain reflects the combined elastic and viscoelastic response (creep deformation) that occurred during application of the step stress, which was removed after 60 minutes. The strain‐descending portion of the curve at > 60 minutes represents the creep recovery provided by polymer chains which extended their conformations via changing the internal bond lengths and angles as a response to the step stress. The irrecoverable strain represents the permanent change in the sample caused by the flow of polymer chains. An overview of the creep properties for the prepared PBT vitrimers is presented in Table 3.2.

Table 3.2. Overview of the creep properties of the samples neat PBT and PBT vitrimers at 125 °C under a 2 MPa step stress.

E at 125 °C Maximum Creep rate Sample sets name [MPa] creep strain [%] [%/hour] Reference PBTa 613 1.1 0.040

2 C13 162 2.2 0.140

I. Varying r: 0.2 C13 198 1.6 0.011 Xgly constant, varying 0.05 C13 427 0.9 0.020 XZn 0 C13 428 0.8 0.018

0.2 C13 198 1.6 0.011

II. Constant r: 0.1 C6 420 1.1 0.021 varying 0.05 C3 490 1.1 0.014 Xgly and XZn 0.02 C1 595 0.6 0.012 aPBT, the creep data are presented in the Appendix 2. 43

Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network

In general, the neat PBT has a higher tensile modulus (E) at 125 °C than the PBT vitrimers 0.02 (the E of C1 is identical to that of neat PBT), and all the PBT vitrimer samples possess a better creep resistance than neat PBT at 125 °C in terms of maximum creep strain (except 2 0.2 2 for samples C13 and C13 ) and creep rate (except sample C13 ). Therefore, exchangeable cross‐links can improve creep resistance even at high temperature. 3.3 Conclusions

The effect of the molar ratio between cross‐linker (glycerol) and catalyst (Zn(II)) on the viscoelastic properties of PBT vitrimers have been characterized. The rubber plateau modulus of PBT vitrimers depends on/increases with Xgly, indicating that the cross‐link density primarily controls the static elasticity of PBT vitrimers, while the stress relaxation time is governed by XZn or XZn/Xgly. For materials with the same thermo‐mechanical history, the modulus below Tg is similar to that in neat PBT, while the modulus between Tg and Tm is controlled by r; a higher stiffness can be obtained in two ways, (1) for a range of materials with constant Xgly by decreasing r; (2) for materials with constant r by lowering the Xgly and XZn simultaneously. The creep experiment demonstrates that the creep resistance at service temperature (Tg < T <

Tm), i.e., both the creep strain and the steady‐state creep rate, can be controlled by the cross‐link density and the dynamics of the exchangeable network. While the PBT vitrimers maintained a similar temperature‐dependent crystallinity behavior as neat PBT, they exhibited equal or superior creep resistance over neat PBT. In summary, In this paper, we have clearly shown that by controlling the exchange dynamics and crosslink density it is possible to tune the (thermo)mechanical and rheological properties of PBT vitrimers. 3.4 Experimental section

The used materials, characterization and polymerization methods are similar to Chapter 2.

Chapter 4

Influence of the morphology on the physical and mechanical properties of PBT vitrimers

ABSTRACT: The dynamic cross‐links are exclusively located in the amorphous phase, whereas the larger homo‐PBT segments are retained, which the unique blocky chemical microstructure of the PBT vitrimers is obtained via crosslinking or copolymerization in the solid state. This study clearly shows that the crystalline/amorphous morphology imparted by Zn(II) catalyzed transesterification gives two avenues to control the properties: 1) randomization of the initial blocky chemical microstructure as well as the initial heterogeneous distribution of cross‐links in space upon heating the sample above the melting temperature. This randomization process will deteriorate thermal, (dynamic)mechanical properties and creep resistance of semi‐crystalline PBT vitrimers. Furthermore, this randomization upon heating is a pure chemical reorganization process, and it is accelerated when an oscillatory stress field is applied during the heating and crystallization process. 2) Rearrangements of the cross‐links in space by crystallization (non‐ random distribution in space irrespective of the chemical microstructure via thermal annealing or crystallization‐induced dynamic cross‐links segregation on cooling from the melt. This peculiar crystallization‐induced dynamic cross‐links segregation phenomenon render to a practical way to enhance the thermal, (dynamic)mechanical and creep performance of the semi‐crystalline PBT vitrimer.

This chapter is partially adapted from: Zhou, Y.; Groote, R.; Goossens, J. G. P.; Lugger, J. A. M.; Sijbesma, R. P.; Peters, G. W. M.; Heuts, J. P. A. In preparation.

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

4.1 Introduction

As shown in chapter 2, a high melt strength semi‐crystalline PBT/glycerol‐based vitrimers were developed via a solid‐state (co)polymerization (SSP) approach.91 As shown by a number of studies from our group,143,145,156–158,165,173 the solid‐state (co)polymerization technique results in a blocky chemical microstructure that contains pure and crystallizable PBT chain segments and comonomer units or the dynamic cross‐links that are located in the amorphous phase. The exclusive cross‐linking of PBT chains in the amorphous phase allows for the introduction of crosslinking points and to retain large amounts of long crystallizable PBT segments existing in the initially formed crystallites to impart the resulting cross‐linked PBT with an excellent recrystallization ability upon cooling. However, vitrimer chemistry based on transesterification reactions involves dynamic equilibrium reactions,52,54– 56,89,103,106,115,166,174 with the exchange rate governed by temperature, catalyst (in our case we used Zn(acac)2), and residence time in the molten state. Therefore, the semicrystalline morphology with a variation in degree of crystallinity, spherulite size and the local organization of the chain segments within the spherulites, as well as the average molecular weight between cross‐links (Mc) is dependent on the thermomechanical history of the material; this results in a non‐random distribution of the dynamic cross‐links within the amorphous layers of the lamellar stacks. Normally, the Tm and degree of crystallinity of SSP products are higher in the first heating run because of the thermal annealing process during

SSP, but, even during the second heating run, the Tc, Tm, and degree of crystallinity of SSP products are still much higher than those prepared by melt polymerization with similar compositions as evidenced by the previous work of Jansen et al.,143,156–158,175 and Lavilla et al.161,162 In general, The main driving force for randomization of copolyesters prepared by

SSP is generally accepted to be the considerable entropy increase above Tm, where the transesterification is activated to randomize the initially blocky chemical microstructure in the melt state.176–181 This randomization process results in decreased PBT‐homo block lengths and has a large influence on the melting temperature, crystallization behavior and degree of crystallinity of the resulting copolymers. In the current work, firstly, the influence of residence time in the melt and the application of oscillatory stress on the chemical microstructure of the vitrimer material (dynamics, etc.) as probed by melting and crystallization behavior will be carried out. Secondly, the morphology (crystal structure, lamellar thickness) of PBT/glycerol‐based vitrimers with different thermo‐mechanical history will be characterized by wide‐angle X‐ray diffraction (WAXD)/small‐angle X‐ray scattering (SAXS). Thirdly, the influence of the chemical microstructure on the (dynamic) mechanical and physical properties will be investigated. Finally, the restoration of the blocky chemical microstructure of PBT/glycerol‐based vitrimers will be discussed.

Chapter 4

4.2 Results and discussion

The main advantage of cross‐linking or copolymerization in the solid state, compared to copolymerization in the melt, is that the incorporation only occurs in the amorphous phase, whereas the polymer chain segments in the lamellar crystals remain unchanged. In this way, the crystallization behavior is more or less preserved due to the presence of large crystallizable homopolymer PBT blocks. However, upon melting, due to the transesterification with the help of a catalyst, this large crystallizable homopolymer PBT blocks will be randomized into short homopolymer PBT blocks when the comonomer or dynamic cross‐links are fully miscible. It is well known for blends based on polycondensation polymers, such as PBT/poly(ethylene terephthalate)(PET), PBT/bisphenol‐A Polycarbonate(PC), and PBT/polyarylate, that the transesterification reactions occurring in the melt convert the blends first into blocky copolymers and, as the reaction proceeds, random copolymers will eventually be obtained.178–186 As the level of transesterification increases, this randomization process results in decreased homopolymer block lengths and may therefore have a large influence on the melting and crystallization behavior as well as a reduction of degree of crystallinity and yield stress of the resulting copolymers. In the PBT vitrimer case, the dynamic crosslinks, initially exclusively located in the amorphous of PBT, will be activated upon melting. These activated dynamic crosslinks will lead to the partitioning of the initial crystalline part of PBT and will end up with a homogeneous distribution of dynamic crosslinks in space depends on the residence time and temperature of the melt.

Figure 4.1. Schematic showing the different steps and thermo‐mechanical histories for sample 0.2 containing 13 mol% glycerol catalyzed by 0.2 mol% Zn(acac)2 (C13 ) prepared by SSP. The number 1, 2, 3 indicated the thermo‐mechanical histories of the sample with identical composition.

Thus, the first question that needs to be addressed is to what extent the blocky chemical microstructure obtained in the solid state will randomize in the melt at different temperatures and residence times? One particular composition (13 mol% glycerol + 0.2 0.2 mol% Zn(acac)2) was studied in this work, and the sample was named C13 . The general thermo‐mechanical histories are presented in Figure 4.1; the cross‐linked powder from the

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

SSP reactor (i.e., sample 1) was compression molded at 250 °C with a total residence time around 20 min and subsequently quenched at an average cooling rate around 40 °Cmin‐1. Afterwards, the compression‐molded sample was dried in the vacuum oven at 120 °C for 4 h before performing rheological or dynamic mechanical analysis.

4.2.1 Influence of the temperature and residence time on thermal properties

The powder obtained from the SSP reactor was heated to a temperature above Tm at a heating rate of 10 °Cmin‐1 after which the sample was isothermally treated at a predetermined temperature (e.g., 250 °C) for 15 min. A subsequent cooling and heating run between ‐10 and 250 °C at a temperature ramp of 10 °Cmin‐1 was followed for the analysis of Tm, Tc and degree of crystallinity.

0.0 2.0 230 C (B) 240 C ‐0.5 1.5 250 C 270 C ‐1.0 260 C 1.0 260 C 270 C 250 C ‐1.5 0.5 Tm,low Heat flow, exo up [w/g] Heat flow, exo up [w/g] 240 C (A) T 230 C ‐2.0 m,peak 0.0 150 160 170 180 190 200 210 220 230 140 150 160 170 180 190 200 210 220 Temperature [C] Temperature [C] 50

210

Tm,peak C]  [ Tm,low 40 c 200 T , m T T 190 c 30 Degree of crystallinity [%] (C) 180 230 240 250 260 270 Isothermal melt temperature [ C]

0.2 Figure 4.2. (A) and (B) DSC thermograms of C13.2 as a function of isothermal melt temperature (Tiso) with 15 min equilibrium time. (C) Tm,peak (), Tm,low (), Tc () and degree of crystallinity() as a function 0.2 of the temperature of the melt with 15 min equilibrium time for C13 .

Chapter 4

The melting behavior (Figure 4.2A) of the samples shows the evolution of a single broad melting peak into two well‐separated melting peaks upon increasing the temperature of the melt. However, only one crystallization event (Figure 4.2B) is observed regardless of the temperature of the melt. It is clearly shown in Figure 4.2C that both the melting (Tm,low and

Tm,peak) and crystallization (Tc) peak temperature decrease monotonously as a function of temperature of the melt (Tmelt), while the degree of crystallinity remains nearly constant. The double melting behavior is a well‐known phenomenon in PBT, and it is generally ascribed to the melting and recrystallization of less perfect lamellar crystals into thicker and more perfect lamellar crystals.15,124–126 However, the chain mobility of PBT vitrimer is restricted by the dynamic cross‐links, and the mobility is highly dependent on the relaxation 0.2 time of this vitrimer network, i.e., C13 , as shown in Figure 3.1, chapter 3, possesses a characteristic stress relaxation time around 44 min at 230 °C,187 which is estimated to be at least 4 order of magnitude higher than neat PBT at 230 °C. Therefore, the reorganization is hindered due to a longer relaxation time ( 44 min) in the melt than the equilibrium time ( 15 min); when the sample is cooled from 230 to ‐10 °C and heated again (10 °Cmin‐1) to 250 °C, only one melting peak is observed due to the remained heterogeneous distribution of the dynamic crosslinks inherited from the SSP process, and largely retained pure PBT segments which also render a highest crystallization temperature,  192 °C. The first well‐separated double melting peak is observed upon increasing the temperature of the melt to 250 °C, because the characteristic relaxation time ( 12 min) at this temperature is shorter than the equilibrium time ( 15 min) in the melt. As expected, the initial crystalline part of PBT vitrimer is fully portioned by the dynamic cross‐links, which rendered a more homogeneous distribution of dynamic cross‐links in space. As a consequence, upon crystallization, a thinner lamellar crystal is obtained due to the restriction of dynamic crosslinks resulting in a low Tc ( 181 °C).

Figure 4.3. DSC temperature protocol for studying the effect of melt residence time with a ‐1 heating/cooling ramp 10 °Cmin . The ti represents the residence time at 250 °C.

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

In order to understand the randomization as a function of residence time in the melt in the presence of the Zn(acac)2 catalyst, the materials obtained by SSP were isothermally treated at 250 °C (as in our compression‐molding process) with different residence times by using DSC. Upon heating, the dynamic cross‐links are expected to reshuffle in the polymer melt to randomize the distribution of the cross‐link points, as the crosslinker is fully miscible with the amorphous phase of PBT, while upon crystallization during cooling, the randomly distributed cross‐links will be again exclusively confined in the amorphous phase with a tendency to form thick lamellae.188 In order to discriminate between the chemical and physical melt restructuring during randomization, two different approaches with approximately identical total melt residence times were applied and the applied DSC time/temperature protocols are schematically shown in Figure 4.3. The first one is a sequential heating/cooling cycling protocol (Protocol I) and the other one is a continuous heating/cooling protocol with a predetermined and varying residence time in the melt (Protocol II). 1.5 1.5 Cycle 11 Cycle 1 (A) Protocol II (B) 1.0 1.0

0.5 0.5

0.0 0.0

‐0.5 ‐0.5

Residence time

Heat flow, exo up [w/g] ‐1.0 ‐1.0 Heat flow, exo up [w/g] Protocol I ‐1.5 ‐1.5 170 180 190 200 210 220 170 180 190 200 210 220 Temperature [C] Temperature [C] Total residence time at 250 C [min] Total residence time at 250 C [min] 0 102030405060 0 102030405060 220 40 Protocol I Tm,peak 210 38 Protocol I C] Protocol II  [ 200 36 c, peak T , 190 34 m, peak Crystallinity [%] T

180 32 Protocol II

Tc,peak (C) (D) 170 30 024681012 024681012 Cycle number Cycle number

0.2 Figure 4.4. DSC cooling and melting thermograms for C13 obtained by applying (A) protocol I and (B) protocol II. (C) Tc,peak and (D) crystallinity as a function of cycle number and total residence time.

Chapter 4

The cooling and melting thermograms are displayed in Figures 4.4A and 4.4B, and it can be seen that the experimental thermograms obtained under consecutive heating and cooling cycling or isothermal heating are quite similar in terms of melting and crystallization behavior. In these two different approaches, all PBT vitrimer samples show a double melting peak and a crystallization peak during the melting and cooling step. Those peaks shift to lower temperatures upon increasing cycle number or residence time. An overview of the peak melting temperature (Tm,peak), peak crystallization temperatures (Tc,peak) and degree of crystallinity are presented in Figures 4.4C and 4.4D.

In general, Tm,peak and Tc,peak show a steady decrease upon consecutive heating and cooling cycle (protocol I) or isothermal heating (protocol II). A maximum shift of 6 and 8 °C in, respectively, Tm,peak and Tc,peak to lower values can be observed within 11 cycles or 45 min of residence time. For the degree of crystallinity, we can conclude that it remains nearly constant in these two protocols. From the work of Marchese and coworkers it is known that for PBT a length of at least 16 repeat units is necessary for crystallization to occur.189 Thus, the crystallization of PBT‐ based copolymers will become more and more difficult as the homo PBT segments become shorter and the copolymer tends toward a random sequence distribution. Crystallization in crosslinked polymers is rarely observed with high amounts of cross‐linker, such as poly(ethylene terephthalate) (PET) containing around 10 mol% pentaerythritol with a melting endotherm only observed in the first heating cycle.190 Moreover, Hartwig et al.167 showed that acid‐catalyzed transesterification reactions can change the morphology of initially cross‐linked semi‐crystalline epoxy/poly(‐caprolactone) copolymers (with a rubbery plateau around 3 MPa) to a completely amorphous material after annealing at 150 °C for about one hour. Interestingly enough, for our semi‐crystalline vitrimers containing approximate 13 mol% glycerol (4 MPa rubbery plateau modulus) catalyzed by 0.2 mol%

Zn(acac)2, it is not possible to sufficiently randomize the blocky chemical microstructure of PBT in the quiescent melt. When we compare the crystallization curves of cycles 1 and 11, although the peak crystallization temperature decreases upon cycling, those two peaks become sharper. The sharping of the crystallization peak points to the segregation of thinner pure PBT crystalline segments because of a slight modification of the crystalline/amorphous morphology. Since the PBT vitrimers possess a high melt strength and are able to release the internally build‐up stress upon deformation via network topology rearrangements, our curiosity was triggered to investigate the effect of oscillatory stress (normally applied during dynamic mechanical or rheological analysis) on the network topology and dynamics.

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

4.2.2 Influence of oscillatory stress during cooling from the melt on thermal properties

A compression‐molded sample (2) (see Figure 4.1) was used for this comparative study. The experimental protocols are shown in Figure 4.5A, the sample preparation protocol is indicated by (i), and an identical temperature ramp of 3 °Cmin‐1 was used during the preparation. The sample with different thermo‐mechanical history is created with two different instruments, DMTA and DSC. The sample prepared during the DMTA run was subjected to oscillatory stress while the reference sample prepared during the DSC run was in a quiescent melt state. The properties were analyzed by the DSC with a temperature ramp of 10 °Cmin‐1 as indicated by (ii). The preparation of sample 3 during DMTA run was shown in Figure 4.5B, for a compression‐molded sample 2, which is annealed at 120 °C for 4 h. during the first heating run, it shows three modulus dropping transitions, which are glass transition ( 66 °C), unknown transition ( 135 °C) (will be discussed in Figure 4.6) and melting transition( 210 °C).

1.0 168C 103 140C 1st_heating Sample 3 0.5 2nd_heating 102 Reference sample 0.0 [MPa]  E 101 1st_cooling ‐0.5 Heat flow, exo up [w/g]  0 187 C 10 (B) (C) ‐1.0 204C 0 30 60 90 120 150 180 210 240 270 0 40 80 120 160 200 240 Temperature [C] Temperature [C] Figure 4.5. Schematic representation of the thermal protocols used for understanding the role of oscillatory stress as indicated with (i) and subsequent DSC analysis step as indicated with (ii). (B) 0.2 Representative DMTA curves of the heating‐cooling‐reheating cycle for C13 and the thermomechanical history 3 (as shown in Figure 4.1) was created after the 1st cooling run. (C) Thermograms of samples 3 (previously cooling from 270 °C under oscillatory stress) and its reference sample (previously cooling from 270 °C without oscillatory stress) with a temperature ramp of 10 °Cmin‐1 from ‐15 to 250 °C. 52

Chapter 4

On continuous heating, the rubbery plateau originating from the dynamic cross‐linked network is observed. When cooling from 270 °C with a cooling rate of 3 °Cmin‐1, an almost constant rubbery plateau between 150 and 270 °C is observed and a crystallization transition appears at about 165 °C. The differences between the two heating runs are striking, as the E’ at 230 °C for the second heating run is 6 times higher than the first heating run, i.e., 4.4 MPa and 0.8 MPa, respectively. Moreover, the onset of the melting temperature is 18 °C lower than the first heating run and the stiffness of the material decreased at 130 °C from 268 to 122 MPa. The DSC was employed to further analyze the thermal properties of sample 3 (with oscillatory stress) and its reference sample to understand the role of oscillatory stress. It is clearly shown in Figure 4.5C that sample 3 displays a broader melting and crystallization peak than the reference sample and that Tm,peak and Tc,peak of sample 3 were lower by 17 and 29 °C, respectively. The degree of crystallinity was 20 and 34% for sample 3 and reference sample, respectively. Therefore, the presence of the oscillatory stress promotes the repartitioning of crystallizable homopolymer PBT blocks towards a more homogeneous distribution of the dynamic cross‐links (randomized microstructure) from an initially heterogeneous state of the crystallizable homopolymer PBT segments and the dynamic cross‐links (blocky microstructure). One of the benefit of SSP is that one can obtain a semi‐crystalline vitrimer with a retained crystalline phase as neat PBT while dynamic crosslinks are exclusively located in the amorphous phase. However, this blocky chemical microstructure will be randomized upon melting with or without stress as we discussed above. Moreover, due to the fast cooling rate during the compression molding process and long relaxation time of PBT vitrimer, we can expect that crystallization for the crystallizable PBT segments will be suppressed by the dynamic crosslinks when cooling from the melt. Therefore, it is crucial to understand how a physical process, like isothermal annealing, influences the evolution of the crystalline/amorphous morphology, the concomitant (dynamic) mechanical properties and creep resistance.

4.2.3 Influence of thermal annealing on the dynamic mechanical properties

0.2 A compression‐molded sample 2 (C13 ) was annealed in a vacuum oven at different temperatures with a duration of 4 h and we will name this annealing temperature as Tiso. The DMTA curves for the PBT vitrimer after isothermal annealing for 4 h at various temperatures are shown in Figure 4.6A, for the sample directly after compression molding, there is no extra transition can be observed. This extra transition was only observed when the samples were thermally treated at 80 and 120 °C (Tiso > Tg) and the onset temperature for the steep decrease of the modulus is shifted from 120 °C to 135 °C, respectively. This

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

91 steep modulus decrease between Tg and Tm was also observed in our previous work. When the annealing temperature is further increased to 180 °C, the extra transition vanished, and the modulus is comparable with the neat PBT below Tm, for instance, the modulus at 180 °C for the neat PBT and a PBT vitrimer annealed at 180 °C are 172.8 and 146.0 MPa, respectively. This phenomenon is usually observed in polymers with flexible aliphatic chain segments, such as polyethylene191 and linear aliphatic polyesters,192 which is related to an increase of the chain mobility within the crystalline lamellae (c relaxation). To the best of our knowledge, this phenomenon is not observed in polymers with aromatic building blocks such as PBT. A comparative experiment was carried out on compression‐molded neat PBT via using the same thermal annealing process as a PBT vitrimer. In general, when neat PBT is annealed, a second endotherm is often observed in a subsequent heating run at a temperature below that of the original melting peak, and this new melting peak appears to increase with annealing at the expense of the original melting peak.146,193,194 as we learned from the DMTA results, we found that the stiffness of neat PBT increases with increasing annealing temperature (Tiso), while the onset of the melting temperature is nearly constant as determined from the storage modulus curve. However, the modulus drops between Tg and Tm is not observed for neat PBT (see the Appendix 3 for DMTA curves).

0.0 (A)  3 25 C 10 ‐0.5 80C ‐1.0 120C 102 PBT  ‐1.5 180 C 25 C ' [MPa] E 1 200C 10 80 C ‐2.0 120 C 180 C Heat flow,exo up [a.u.] ‐2.5 100 200 C (B) ‐3.0 0 50 100 150 200 250 50 100 150 200 250 Temperature [C] Temperature [C] Figure 4.6. (A) DMTA curves and (B) first heating step of the DSC traces of compression‐molded PBT vitrimer samples after isothermal annealing for 4 h at various temperature.

To figure out whether this process is caused by c relaxation or a thermodynamic transition, differential scanning calorimetry (DSC) was employed to analyze sample 2 after different Tiso and the results are presented in Figure 4.6B. A low temperature melting transition was observed during the first heating run for the sample annealed above Tg, and it increases with increasing the annealing temperature (Tiso). For instance, when the sample wasannealed at 120 °C for 4h, the DSC results indicated a low melting transition

Chapter 4 temperature around 145 °C during the first heating run and the degree of crystallinity is around 2.5%. Considering this clear shift of melting temperature observed by DSC, it can be concluded that this steep decrease of the modulus between Tg and Tm for our aromatic PBT vitrimers is due to a melting transition, and this is probably caused by the recrystallization of the PBT segments that are confined by dynamic cross‐links, and are not able to crystallize during the fast cooling ( 40 °Cmin‐1). 25 40 10 (A) neat PBT (B) degree of crystallinity 35 20 30 1 200C 180C 15 25 [nm] c 120C l

, 20 p

0.1  L L 80 C 10 p 25C 15 Intensity [a.u.] PBT 10 0.01 5 l Degree of crystallinity [%] c 5 SAXS WAXD 0 0 1E‐3 R.T.0 50 100 150 200 0.1 1 10 ‐1 T [C] q [nm ] iso Figure 4.7. (A) X‐ray spectra of obtained PBT/glycerol‐based vitrimers undergo different annealing temperature (Tiso). (B) The long period (Lp ), lamellar thickness (lc), from SAXS and degree of crystallinity obtained from DSC.

In order to further elucidate this hypothesis, both wide‐ and small‐angle X‐ray diffraction/scattering (WAXD/SAXS) experiments were performed to gain additional insight into the influence of thermal annealing on the crystal morphology (Figure 4.7A); the morphology parameters are shown in Figure 4.7B. From the peak maximum position qmax of the SAXS profiles long period (Lp) is calculated, and lamellar thickness (lc) can be estimated (implicitly assuming a two‐phase microstructure) according to equations: 2 , lL  (1) Lp  cpc qmax

The sample directly after compression molding has Lp  13 nm, and it decreases to  9 nm when Tiso  120 °C. The long period further increases upon increasing the annealing temperature (180 °C  Tiso  200 °C). The initial decrease of the average long period (Lp) is probably caused by the occurrence of the thinner lamellae within (the lamellar insertion 195 model) the existing primary lamellar stacks. When the Tiso  180 °C, the average long period (Lp) increases due to the reorganization process resulting in melting and recrystallization of initially formed thinner lamellae into thicker lamellae.15,157 The lamellar thickness (lc), calculated from the long period obtained from the Lorentz‐corrected SAXS patterns shows a similar trend. When annealed at temperature  180 °C, the lamellar

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

thickness of the PBT vitrimers is similar to that of neat PBT (lc  5 nm), indicating that thermal annealing works efficiently in crystal perfectioning in PBT vitrimers.

4.2.4 Influence of annealing on tensile properties and creep resistance

The effect of thermal annealing on tensile properties was investigated. The stress‐strain curves of PBT vitrimers after isothermal annealing for 4 h at various temperature are shown in Figure 4.8A. Both PBT and PBT vitrimer do not show strain hardening during uniaxial deformation, which indicate that both are relatively brittle materials. The tensile properties, namely Young’s modulus (E) and yield stress (σy), are summarized in Figure 4.8B.

70 1500 110 Tiso:180°C (A) (B) 60 PBT, T :180°C iso 100 1200 50 Tiso:200 °C 90 40 900 80 30 600

Stress [MPa] 70 20

Tiso:120 °C Yield stress [MPa]

T : 80 °C Young's modulus [MPa] 300 iso 60 10 Tiso: 25 °C

0 0 50 0 5 10 15 20 25 30 35 25 50 75 100 125 150 175 200 Strain [%] Annealing temperature [C] 5 (C) 3 4 3‐annealed@180C‐4h 3‐annealed@120C‐4h 3

2 2 Strain [%]

125C 0 PBT 0 30 60 90 120 150 Time [min] Figure 4.8. (A) Characteristic stress‐strain curves of the compression‐molded PBT vitrimer (2) with different annealing temperatures. Experiments were performed at room temperature ( 20 °C) using a constant cross‐head speed of 50 mmmin‐1. (B) The tensile modulus, yield stress of PBT (closed symbols) and PBT vitrimer (open symbols) after isothermal annealing at various temperatures. (C) Creep‐recovery experiments at 125 °C on PBT vitrimers with different thermo‐mechanical histories.

After compression molding, the neat PBT shows a Young’s modulus of 744 ± 85 MPa and a yield stress of 58 ± 0.2 MPa, while the PBT vitrimer values (Young’s modulus = 1090 ± 87 MPa, yield stress = 67 ± 1.1 MPa) are slightly higher than for the neat PBT. In general, the 56

Chapter 4

Young’s modulus and yield stress for both the PBT and the PBT vitrimer increase with increasing annealing temperature. The creep results are shown in Figure 4.8C. For the PBT vitrimer (sample 2) containing

13 mol% glycerol and 0.2 mol% Zn(acac)2, it shows a maximum strain difference of 0.5% as compared with neat PBT when a 2 MPa step stress was applied with a duration of 60 min. When we compare samples 2 and 3, which represents different degrees of randomization (chemical microstructure) for the same chemical composition as schematically shown in Figure 4.1, sample 3 shows a 2.7 times larger maximum strain and 16 times larger steady creep rate than sample 2. These results are in a good agreement with our DMTA and DSC results discussed in Figure 4.5. Moreover, we demonstrate that annealing can also enhance the creep resistance of the randomized PBT vitrimer sample. When the sample is isothermally annealed at 120 °C for 4h, the maximum creep strain was decreased from 4.1% to 3.2%, whereas the steady creep rate increased 1.6 times. In contrast, upon increasing annealing temperature to 180 °C, the steady creep rate is 5 times slower while the maximum creep strain is 1.2 times larger than the sample annealed at 120°C, respectively. To summarize, the degree of crystallinity and morphology (heterogeneity distribution of the dynamic cross‐links) have a strong influence on the physicochemical properties of semicrystalline vitrimers, such as the mechanical strength, heat resistance and creep resistance. So far, efforts have been devoted to the recovery of the non‐random distribution of dynamic cross‐links after randomization in the melt via crystallization‐induced dynamic cross‐links segregation, which is isothermal annealing below Tm. In the following part, an attempt to recover the non‐random distribution of dynamic cross‐links in space is discussed.

4.2.5 Recovering the non‐random distribution of dynamic cross‐links above Tm

In order to demonstrate that the randomized chemical microstructure (a homogeneous distribution of dynamic crosslinks) can be restored at a temperature above Tm, a sequential heating/cooling cycling protocol as schematic shown in the top inset of Figure 4.9A was 0.2 employed. The PBT vitrimer (C13 ) with various thermal and mechanical histories was studied by this sequential heating/cooling cycling protocol. It is generally accepted that a homo‐ or copolymer melt with clusters of molecular segments with a more ordered conformation than the corresponding equilibrated random coil will crystallize at a faster rate, due to a depression in free energy change (G=H‐TS) for nucleation from entropic considerations.38 When the homo‐ or copolymer undergoes 0 sequential heating‐cooling cycles well above Tm at a critical holding time (long enough to erase the melt memory), the crystallization temperature (Tc) should be constant as a function of cycle number.196–199

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

1.0 200 (A) 190 Neat PBT 0.8 180 1 C 

/ 170 2 0.6 c, peak T 160

Heat flow, exo up [w/g] 0.4 150 3 (B) 140 100 110 120 130 140 150 160 170 180 190 024681012 Temperature [C] Cycle number Figure 4.9. (A) Representative DSC cooling curves for sample 3 as a function of cycling number

(protocol was shown in the top inset graph). (B) Tc, peak as a function of cycle number.

As can been seen from Figure 4.9B, when the neat PBT is cycled at 250 °C with a 5 min equilibrium time, the Tc, peak is nearly constant (the corresponding DSC curves are presented 0 in Appendix 3) and the equilibrium melting temperature (Tm ) of PBT is around 237 °C as determined by the Hoffman‐Weeks plots (see Appendix 3).200 The compression‐molded sample 2 shows a minor increase of Tc, peak within 3 cycles and followed with a minor decreases of Tc, peak until cycle no. 10. Eventually, it shows a plateau Tc, peak value similar to a sample, obtained directly after SSP. However, sample 3, which is prepared with a cooling ‐1 rate of 3 °Cmin with oscillatory stress, shows a quite strong increase of Tc, peak during the multiple cycling as compared to sample 1 and 2. A well‐pronounced increment of Tc, peak of approx. 24 °C is observed and the final Tc, peak value is approx. 170.4 °C, which is comparable to samples 1 and 2. Is this higher crystallization temperature or higher rate of crystallization caused by melt memory? Although melt memory is not observed for neat PBT under the same experimental conditions, considering the dynamic network of the PBT vitrimer, the equilibrium time for the PBT vitrimer should be longer than for neat PBT because of the longer relaxation time in the melt. In our PBT vitrimer case, the melt memory may be caused by the residual order from traces of previous crystallites or regions of polymer chains confined by less dynamic cross‐links that remain at temperatures above melting.

To verify whether this strong increase of Tc, peak is due to a melt memory effect or reorganization of dynamic cross‐links in the melt, protocol II as shown in Figure 4.3 was used to investigate the effect of residence time up to 45 min at 250 °C in the melt. Both thermal protocols gives very similar results, which is that the Tc, peak of sample 3 increases with increasing equilibrium time or cycle number (Appendix 3). Thus, the increase in Tc, peak is not caused by melt memory, but it is probably due to the network topology rearrangements activated by Zn2+‐catalyzed transesterification, which leads to a more

Chapter 4 heterogeneous distribution of the network junctions in the melt at 250 °C. The plateau value observed for Tc, peak might indicate the formation of lamellae with a certain thickness in the quiescent melt.

Scheme 4.1. A schematic view of reorganization of dynamic cross‐links during crystallization to form a non‐randomized “micro”‐blocky microstructure.

The possible explanation is schematically shown in Scheme 1 with the sample with a random distribution of dynamic cross‐links upon melting will reorganize the dynamic cross‐ links via Zn(II)‐catalyzed transesterification to a homogeneous distribution of these dynamic cross‐links, which is described as randomization. On cooling down the PBT vitrimer melt, crystallization will take place. Since the cross‐links containing segments that cannot co‐ crystallize with PBT, these dynamic cross‐links do not enter the crystal lattice and reside preferentially in the amorphous phase surrounding the lamellar crystals. Driven by the presuming of longer crystallizable PBT homopolymer sequences, this dynamic cross‐links segregation promoted by Zn(II)‐catalyzed transesterification during crystallization will lead to the formation of a non‐random distribution of dynamic cross‐links in space. 4.3 Conclusions

Considering the dynamic nature of the exchangeable network of PBT vitrimers, the effect of temperature of the melt and residence time have been mapped out on a sample crosslinked in the solid state. In the quiescent melt, a slight decrease of the Tm,peak, Tc,peak and crystallinity upon prolonged equilibration time at 250 °C or high temperature of the melt (up to 270 °C) was observed. Thus, the blocky chemical microstructure of the semi‐ crystalline vitrimer obtained by solid‐state polymerization cannot be fully randomized by the exchangeable network in the quiescent melt under the studied conditions. When oscillatory stress is applied during the cooling process, a highly randomized chemical microstructure is obtained as indicated by the lower Tm,peak, Tc,peak and degree of crystallinity as well as broad melting and crystallization peak.

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

The steep decrement of the storage modulus between 120 and 150 °C during DMTA experiment is a melting transition as confirmed by DSC and X‐ray experiments, and this melting transition is probably caused by the recrystallization of the PBT segments confined by dynamic cross‐links during thermal annealing above Tg, which are not able to crystallize during the fast cooling (40 °Cmin‐1). Tensile tests and creep experiments show that a PBT vitrimer with a blocky chemical microstructure is beneficial for its mechanical strength and creep resistance. Peculiar crystallization‐induced dynamic cross‐links segregation is found for the highly randomized vitrimers below Tm (thermal annealing) or upon crystallization from the melt, and this physical phenomenon with the help of Zn2+‐catalyzed hydroxy‐ester interchange reactions accelerates a subsequent crystallization as observed by higher crystallization peaks temperature on cooling from the melt, for instance, a rise of 24 °C for Tc,peak is observed within 10 cycles. The dynamic nature of the exchangeable network of PBT vitrimer contributes to both the melting‐induced dynamic crosslinks randomization driven by entropy increase and crystallization‐induced dynamic crosslinks segregation driven by the lower enthalpy state of crystallites or presuming of longer crystallizable sequences. Therefore, the concept developed in this chapter permits a physical way for tuning the properties of semicrystalline vitrimer, such as thermal properties, mechanical strength and creep resistance. 4.4 Experimental section

Materials The material containing about 13 mol% glycerol and 0.2 mol% Zn2+ was prepared by SSP, 91,187 as described in our previous paper. The neat poly(butylene terephthalate) powder (Mn

= 21.2 kg/mol, Mw = 46.6 kg/mol against poly(methyl methacrylate (PMMA) standards in 1,1,1,3,3,3‐Hexafluoro‐2‐propanol (HFiP)) provided by SABIC (Bergen op Zoom, The Netherlands) was used as the reference. Characterization methods Differential scanning calorimetry. DSC experiments were performed by using a DSC Q1000 from TA Instruments. Hermetically sealed aluminum pans were used with a typical sample weight of 5~6 mg. In general, the crystallization and melting behaviors of the polymers were heated from ‐10 to 250 °C at a heating rate of 10 °Cmin‐1. After a 5 min equilibration time at 250 °C, the samples were cooled to ‐10 °C at a rate of 10 °Cmin‐1. The DSC procedures used are dependent on the thermal protocol of individual experiment. The crystallization and melting temperatures were determined from the peak of the cooling exotherm and heating endotherm, respectively. The crystallinity was calculated by using 142 Jg‐1 as the heat of fusion of 100% crystalline PBT.40 60

Chapter 4

Compression molding. The materials were compression molded at 250 °C and 100 bar for 25 minutes in a Collin Press 300G and subsequently cooled with water. Dynamic mechanical thermal analysis. For the consecutive heating‐cooling‐heating cycle experiment used in this work, compression‐molded samples (ca. 10.0 (Length) × 5.0 (Width) × 1.0 (Thickness) mm) were measured on a DMA Q800 (TA Instruments) with a film tension setup. During the experiment, the strain was oscillated at a frequency of 1 Hz with a peak‐to‐peak amplitude of 10 µm while the temperature was increased from ‐15 to 270 °C at a rate of 3 °Cmin‐1. Once the temperature reached 270 °C, it was maintained for 5 minutes and then decreased to 25 °C at the same rate. The dimension of this sample was measured again at 25 °C; afterwards, a standard DMTA test was performed to study the dynamic mechanical properties after the previous cooling run. Tensile tests. The tensile properties of the samples annealed at different temperatures were evaluated with a Zwick 100 tensile tester equipped with a 10 KN load cell at room temperature. The dog‐bone‐shaped tensile bars were die cut from a compression‐molded polymer sheet with a thickness of 1 mm. The samples were tested with a constant cross‐ head speed of 50 mm.min‐1. At least 5 specimens were averaged to collect the tensile properties for each sample. X‐Ray diffraction spectroscopy. XRD profiles were recorded with a Ganesha lab instrument equipped with a Genix‐Cu ultralow divergence source producing X‐ray photons with a wavelength of 1.54 Å and a flux of 1×108 photons s−1. Diffraction patterns were collected by using a Pilatus 300 K silicon pixel detector with 487×619 pixels of 172 μm2 placed at a sample‐to‐detector distance of 91 mm. Azimuthal integration of the diffraction patterns was performed by utilizing the SAXSGUI software.

Influence of the morphology on the physical/mechanical properties of PBT vitrimers

Chapter 5

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

ABSTRACT: Poly(butylene terephthalate) (PBT) vitrimers are characterized by their high degree of crystallinity and melt strength. We herein demonstrate the use of the different transesterification kinetics between glycerol (two primary and one secondary hydroxyl groups) and 1,1,1‐tris(hydroxymethyl)propane (three primary hydroxyl groups) catalyzed by zinc acetylacetonate to tune the viscoelastic properties of bulk materials through simply altering the cross‐linker. The materials cross‐linked with 1,1,1‐tris(hydroxymethyl)propane showed a fast network build up during SSP, but slow stress relaxation kinetics in the resulting network as compared to the materials cross‐linked with glycerol using the same cross‐linker and Zn(II) catalyst contents. These macroscopic differences in gelation and stress relaxation kinetics were investigated by using small‐molecule kinetic model experiments. The effect of the nature of the cross‐linker on the viscoelastic properties of the PBT bulk material is further evaluated in terms of thermal, thermomechanical, weldability, creep resistance and mechanical properties. Our study demonstrates that the chemical architecture of the cross‐linker (triol), which in addition to chemical composition, type and content of catalyst, can serve as another design parameter to alter the physical and viscoelastic properties of PBT vitrimers.

This chapter is partially adapted from: Zhou, Y.; Goossens, J. G. P.; Pan, J.; Sijbesma, R. P.; Heuts, J. P. A. In preparation.

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

5.1 Introduction

A high melt strength semi‐crystalline PBT vitrimer was developed via the incorporation of glycerol, which is the ring‐opened form of an epoxide containing two primary hydroxyl groups and one secondary hydroxyl group, in the solid state.91 Subsequently, we studied two different routes to tune the viscoelastic properties of PBT/glycerol‐based vitrimers: one chemical route controlling the dynamics of the exchangeable network (Chapter 3)187 and one physical route using the crystallization/melting kinetics to control the localization of the cross‐linker (Chapter 4).201 To our knowledge, the effect of the transesterification kinetics difference between the hydroxyl groups and their subsequent formed ester groups on the viscoelastic properties of PBT vitrimers has not been explored. It is well known that the reactivity of a primary OH is higher than that of a secondary OH in the presence of a Lewis acid transesterification catalyst.202,203 Moreover, Leibler and coworkers demonstrated that the presence of hydroxyl groups is essential for the stress relaxation and welding efficiency of epoxy networks via model studies with small molecules,55 although there is one example of polyester networks whose dynamics are based on ester‐ester exchange reactions in the presence of 1,5,7‐triazabicyclo[4.4.0]dec‐5‐ene.99 Given the importance of the free hydroxyl group in the transesterification reaction in the current study, we will study in this chapter the influence of the chemical structure of the cross‐linker on the viscoelastic properties of PBT vitrimers. For this purpose, 1,1,1‐tris(hydroxymethyl)propane (TMP) with only primary hydroxyl groups was chosen as a cross‐linker to compare with glycerol on two aspects: 1) the kinetics of the network build up during the solid‐state (co)polymerization and 2) the dynamics of the exchangeable network of PBT vitrimers and its influence on the viscoelastic properties of the PBT vitrimers. Furthermore, we will correlate the kinetics of the network formation during solid‐state (co)polymerization and the exchange kinetics of PBT/triol‐ based vitrimers with model materials. 5.2 Results and discussion

There are mainly three types of interchange reactions in a polyester network, i.e., alcoholysis, acidolysis, and esterolysis.188 For our polyester‐based vitrimer systems, the incorporation of the triol cross‐linker into the PBT backbone and the subsequent stress relaxation above its topology transition temperature54,56,174 after forming the network is mainly achieved by the transesterification reaction between the ester and hydroxyl groups and this process is schematically shown in Scheme 5.1.

Chapter 5

Scheme 5.1. Schematic representation of the main reactions involved in the PBT vitrimer during the transesterification‐based stress relaxation process.

In the PBT/GLY‐based vitrimer system, there are four main exchange reaction pairs between hydroxyl and ester groups: 1) primary OH and secondary hydroxyl ester, 2) primary OH and primary hydroxyl ester, 3) secondary OH and primary hydroxyl ester, and 4) secondary OH and secondary hydroxyl ester. The source of the primary OH and primary hydroxyl ester group can be from both neat PBT segments and PBT/GLY segments, while the secondary OH and secondary hydroxyl ester is from PBT/GLY segments. In contrast, in the PBT/TMP‐based vitrimer systems, the main exchange reaction pair is between the primary OH and primary ester, and these two exchange reaction groups can be from both PBT and PBT/TMP segments.

5.2.1 Synthesis and thermal properties of PBT/triol‐based vitrimers

In general, a two‐step solid‐state (co)polymerization strategy was employed to incorporate the triol (TMP and GLY) into PBT backbone as reported in previous chapters. The sample name is abbreviated as Cx(cross‐linker), where x indicates the mol% of cross‐

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers linker with respect to the PBT repeat units. The molecular weight evolutions as a function of reaction time (tssp) during the two‐step SSP method are shown in Figure 5.1.

1.4 1.4 (B) Physical mixture (A) 1 h Physical mixture 3 h 1.2 1.2 1 h 1.0 0.5 h 1.0

0.8 0 h 0.8 0 h

0.6 0.6 wlog M wlog M

0.4 0.4 180C 180 C 0.2 0.2 TMP GLY 0.0 0.0 2.53.03.54.04.55.05.5 2.5 3.0 3.5 4.0 4.5 5.0 5.5 log M log M Figure 5.1. Evolution of the molecular weight distribution of (A) C8(TMP) and (B) C8(GLY) as a function of tssp. Molecular weights are reported relative to PMMA standards. Note: tssp = 0 h is the sample after prepolymerization at 160 °C for 48 h.

During the prepolymerization step at 160 °C, an initial drop in Mn due to chain scission was observed for all the compositions studied in this chapter (see Appendix 4 for other compositions). The chain scission is caused by the alcoholysis of the PBT chains by the free hydroxyl groups from the used triol (TMP or GLY). The data in Figure 5.1A show that the molecular weight shifts to lower M and the molecular weight distribution becomes broader after prepolymerization. The Mn drops from 21.2 to 6.7 kg/mol for the material containing 8 mol% TMP, and the Ð is increased from 2.2 (physical mixture) to 2.5 (after prepolymerization at 160 °C for 48 h). For C8(GLY), a similar molecular weight drop is obtained as C8(TMP) while the Ð is more or less constant after prepolymerization at 160 °C for 48 h. No cross‐linked copolyesters are formed at 160 °C after 48 h with 0.2 mol%

Zn(acac)2 for C8(GLY) and C8(TMP).

In the second step, solid‐state (co)polymerization was performed at 180 °C with a N2 ‐1 flow of 0.5 Lmin in order to remove the condensate 1,4‐butanediol. The Mn increases as a function of tssp before the gel point; for C8(TMP), it increased from 6.7 kg/mol (tssp = 0 h at 180 °C) to 9.2 kg/mol (tssp = 1 h) and the Ð increased from 2.5 to 2.7, while for C8(GLY), the Mn increases to 9.8 kg/mol at tssp = 3 h, but no changes of the Ð is observed. This increase in Mn is a result of polycondensation that takes place between end groups of the chains with elimination of 1,4‐butanediol. When tssp ≥ 3 h (for C8(TMP)) or 7 h (for C8(GLY)), cross‐ linked copolyesters are obtained, which is confirmed by insolubility in HFiP. The molar concentration of triol before the gel point during solid‐state (co)polymerization was studied by by 1H‐NMR spectroscopy in a deuterated solvent mixture

Chapter 5

(CDCl3:TFA‐d = 5:1). For PBT/GLY‐based vitrimer, integration of the peaks at δ = 8.10 ppm

(1,4‐phenylene) against the peaks from glycerol at δ = 4.72‐4.95 ppm (CH2‐O(C=O)) and δ = 6 ppm (CH‐O(C=O) allowed for determination of the glycerol composition before gelation.91 Similarly, for C8(TMP), integration of the peaks at δ = 8.10 ppm (1,4‐phenylene) against the

TMP methyl peaks (‐CH3) at δ = 0.95‐1.20 ppm allowed for determination of the TMP content before gelation. The results are summarized in Table 5.1 and the 1H‐NMR data are presented in Appendix 4. It is evident from Table 5.1 that a high degree of incorporation of the triol with this two‐step solid‐state (co)polymerization method is obtained.

Table 5.1. Overview of the characteristics of the prepared PBT copolymers.

Cross‐linker Gel concentrationa ~t c T d T e T f T f χ f Entry gel fraction d g m c c Before (hour) /°C /°C /°C /°C /% Initialb /%c gelationb PBT ‐ ‐ ‐ ‐ 374 56 222 194 40

C1 (GLY) 1.6 1.2 <24 84 376 61 222 196 40

C1 (TMP) 1.6 1.3 7 82 377 60 220 191 35

C8(GLY) 8.2 7.8 7 74 377 59 217 188 37

C8(TMP) 8.5 8.0 3 79 377 61 209 175 35

gC13(GLY) 14.3 13.2 3 99 378 66 213 183 35

C13(TMP) 14.1 13.5 1 82 376 77 194 145 24 aThe sample name was abbreviated as Cx(cross‐linker), where x indicates the mol% of cross‐linker with respect to the PBT repeat units. bThe cross‐linker content was determined before gelation using 1H‐NMR spectroscopy. c ~tgel: the approximate time when the sample is not soluble in 1,1,1,3,3,3‐hexafluoroisopropanol (HFiP) anymore. d Td, initial decomposition temperature is calculated from the value of onset. e Tg, peak maximum from the loss modulus determined by DMTA. f Tc and Tm are the peak values of the crystallization exotherms and melting endotherms, respectively. f Degree of crystallinity (c)(%) was determined by dividing the melting enthalpy (ΔHmelting) (obtained 0 0 via DSC measurements) by the melting enthalpy (ΔH melting) for 100% crystalline PBT. For ΔH melting, a value of 142 J/g was used in this study.204,205 gC13(GLY): data was taken from Chapter 2.91

The overall thermal history during the preparation is shown in Scheme 5.2. The insolubility, thermal stability, thermomechanical and rheological properties are

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers comprehensively characterized by gel fraction experiments, thermogravimetric analysis, dynamical mechanical thermal analysis, and rheometry. All the samples are insoluble in HFiP at room temperature and show a gel fraction of approx. 80% (see Appendix 4). Furthermore, all the studied PBT vitrimers show a slightly better thermal stability than neat PBT, displaying a decomposition temperature approximately 2‐5 °C higher than that of neat PBT. PBT and PBT/GLY‐based vitrimers show only one weight loss step at 395 and 400 °C, respectively. In contrast, PBT/TMP‐based vitrimers that have two weight loss steps, i.e., C8(TMP) has two peaks centered at 399 and 448 °C, respectively. The results of above‐ mentioned studies are summarized in Table 5.1.

Scheme 5.2. The overall polymerization and thermal history during the preparation of PBT vitrimers. tssp stands for the reaction time during solid‐state (co)polymerization (SSP). tgel stands for the gelation time during SSP.

100 (A) 1.8 (B) C1(TMP) C1(GLY) C8(GLY) 1.2 C13(GLY) 80 PBT C8(TMP) C13(TMP) PBT 0.6 60 C1(TMP)

C1(GLY) 0.0 40 Weight [%] C8(TMP) ‐0.6 C8(GLY) 20

C13(TMP) Heat flow, exo up [w/g] ‐1.2 C13(GLY) 0 ‐1.8 100 200 300 400 500 600 120 140 160 180 200 220 Temperature [C] Temperature [C] Figure 5.2. (A) TGA and (B) DSC thermograms curves of C1, C8, and C13 (GLY and TMP) and neat PBT.

All samples have 0.2 mol% Zn(acac)2. The insert in the bottom of Figure 5.2A shows the optical properties of C1(TMP), C8(TMP), and C13(TMP) from the left to the right.

The different thermal properties between the PBT/triol‐based vitrimers with approximately the same amount of catalyst and cross‐linker are presented in Figure 5.2B and an overview of the thermal properties is given in Table 5.1. In general, the higher the cross‐linker content, the lower the Tm, Tc, and degree of crystallinity. For the PBT/triol‐based 68

Chapter 5 vitrimers with approx. the same amount of cross‐linker content, the PBT/GLY‐based vitrimers possess a higher Tm, Tc and degree of crystallinity than the PBT/TMP‐based vitrimers. For instance, the Tm of C8(TMP) and C8(GLY) are 15 and 5 °C lower than in neat

PBT (Tm = 222 °C), respectively. The Tc of C8(TMP) and C8(GLY) is about 175 °C versus 188

°C, while the neat PBT has a Tc around 194 °C. Furthermore, the degree of crystallinity of C8(TMP) and C8(GLY) show a 4 to 6% lower than neat PBT, which is around 41%. The decreasing degree of crystallinity of the PBT/TMP‐based vitrimers also correlate with their optical appearance. As shown in the inset of Figure 5.2B, the transparency of the materials is increasing with an increase of the TMP content. Interestingly, C1(GLY) shows a 2 °C higher

Tc than neat PBT while their degree of crystallinity is nearly the same. This phenomenon is probably related to the presence of the heterogeneities in the melt in the form of glycerol‐ enriched microdomains with a higher cross‐link density that act as nucleation sites.90,91 In summary, for a similar cross‐linker and Zn(II) catalyst content, PBT/GLY‐based vitrimers exhibits better thermal properties than PBT/TMP‐based vitrimers, while a similar thermal stability as compared with PBT/TMP‐based vitrimers.

5.2.2 Dynamic mechanical properties

The thermomechanical properties of the compression‐molded samples were characterized by dynamic mechanical thermal analysis (DMTA) and the samples were dried at 120 °C under vacuum for 4 h prior to the test. The DMTA results on the materials with different triol contents (see Table 5.1) are presented in Figure 5.3A. The storage modulus (E) as obtained by DMTA of the PBT vitrimers exhibit the expected behavior90,91, and they were similar to that of PBT below Tm, but above Tm, a second plateau, which originated from the dynamic cross‐links was observed.

3 Neat PBT 10 C1(GLY) 105 C13(GLY) 2 C8(GLY) 10

C13(TMP) 1 104

' [MPa] 10 E C8(TMP) C8(TMP) 100 C8(GLY) C1(TMP) G'(open), G''(closed) [Pa] (A) 103 (B) 0 30 60 90 120 150 180 210 240 0.01 0.1 1 10 100 Temperature [C] [rad/s] Figure 5.3. (A) DMTA curves of C1, C8, and C13(GLY and TMP) and neat PBT. It has to be noted that all samples have 0.2 mol% Zn(acac)2. (B) Angular frequency dependence of the storage modulus G (open symbols) and loss modulus G (filled symbols) at 250 °C for C8(GLY) and C8(TMP), for clarity reasons, the other compositions are presented in Appendix 4. 69

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

It is clearly shown in the DMTA curve that C1(GLY) exhibits a nearly identical stiffness as neat PBT, since they have the same degree of crystallinity. For approximately the same amount of cross‐linker, the vitrimers possess a similar rubber plateau modulus. The DMTA results are consistent with the DSC results (Figure 5.2B).

The dynamic mechanical properties in the molten state were studied by oscillatory frequency sweep experiments in the linear viscoelastic region. Figure 5.3B show the frequency‐dependent dynamic mechanical properties of the PBT vitrimers at 250 °C. Taken C8(TMP) and C8(GLY) for an example, the materials show solid‐like gel characteristics with G > G. When the cross‐linker content is higher than 8 mol%, the material demonstrates a nearly constant value for the elastic modulus over the studied angular frequency regime of 0.01‐100 rad/s. As the cross‐linker content is decreased to approximately 1 mol%, a drop in the low‐frequency modulus occurs, demonstrating a liquid‐like behavior and relaxation at long timescales of the C1(GLY) and C1(TMP) (Appendix 4). Here, it is clear shown that a similar shear storage modulus is obtained for C8(TMP) and C8(GLY), while they exhibits a clearly different melt dynamics.

5.2.3 Stress relaxation experiments

To further elucidate the dynamic nature of these PBT vitrimers, the stress relaxation behavior was studied in the linear viscoelastic region. Before performing the stress relaxation experiments, the compression‐molded samples were subjected to an oscillatory time sweep at 250 °C with 1 % strain at 1 Hz and the storage modulus was monitored as a function of time until a steady‐state plateau was reached. The stress relaxation curves of C8(TMP) at different temperatures are shown in Figure 5.4A. Based on the Maxwell model ⁄ ( ) for viscoelastic fluids, the relaxation time can be defined as the time 54–56 needed to achieve G(t)/G0 = 0.37(= 1/e). Representative normalized stress relaxation curves at 250 °C for PBT vitrimers are shown in Figure 5.4B. It is clear from this figure that the relaxation processes of the C1(TMP) and C1(GLY) materials cannot be fitted by a single characteristic relaxation time and this deviation was also reported in Chapter 3 for some systems. For the C13(TMP) system at 250 °C, it relaxes only to 40% of its initial stress after 5400 s, while the C13(GLY) system shows full stress relaxation after around 4000 s. Although the stress relaxation behavior of the C13(TMP) system can be fitted by the Maxwell model and the fitted stress relaxation time is about 10600 s, it is very difficult to experimentally determine the characteristic relaxation time. In order to investigate the influence of the cross‐linker structure, C8 was chosen as the example to demonstrate the differences in activation energy, welding efficiency and creep resistance. The stress relaxation curve can be nicely fitted by the Maxwell model using a single relaxation time suggesting that network rearrangement is the main relaxation 70

Chapter 5 mechanism with the relaxation time for C8(TMP)(~2900 s) being four times higher than C8(GLY)(~700 s) at 250 °C. The activation energy difference between C8(GLY) and C8(TMP) was measured by stress relaxation experiments at different temperatures.

1.0 6x105 C13(TMP)

5x105 0.8 240 C C8(GLY) 5 4x10 230 C 0.6

260 C 0 3x105 /G

t C1(TMP)

G 0.4 Modulus [Pa] 5 37% 2x10 Experiment T 1x105 0.2 C8(TMP) 250 C C1(GLY) Fitting C13(GLY) (A) 270 C 0 (B) ‐2 ‐1 0 1 2 3 4 0.0 10 10 10 10 10 10 10 100 101 102 103 104 Time [s] Time [s]

Ea 175 20 kJ/mol 9 C8(TMP)

8 [s]

 7

E  127 7 kJ/mol 6 a C8(GLY) (C) 5 1.83 1.86 1.89 1.92 1.95 1.98 1000/T [K‐1]

Figure 5.4. (A) Plots of stress relaxation experiments of C8(TMP). (B) Normalized stress relaxation curves at 250 °C for PBT vitrimers (solid line: experimental data; dashed line: Maxwell model fitting). Relaxation times were measured for a 63% relaxation, and (C) variation of the stress relaxation time for C8(TMP) and C8(GLY) versus inverse temperature.

It is clearly shown in Figure 5.4C that the PBT/GLY‐based vitrimer exhibits a lower activation energy (Ea) and relaxes much faster than the PBT/TMP‐based one, i.e., the Ea for C8(TMP) and C8(GLY) are 175 and 127 kJ/mol, respectively. Therefore, the stress relaxation experiments clearly demonstrated that the chemical reaction involved during the network topology rearrangement for PBT/GLY‐based and PBT/TMP‐based vitrimers are different as evidenced by a different activation energy. Furthermore, the stress relaxation time is highly dependent on the chemical structure of the cross‐linker, i.e., the primary/secondary hydroxyl group, and we will attempt to correlate this macroscopically flow behavior with small molecules kinetics.

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

5.2.4 Model molecules study

Scheme 5.3. Model reaction between butyl benzoate and 1‐decanol (1°‐OH) or 2‐decanol (2°‐OH) to mimic primary and secondary hydroxyl ester link formation.

In order to assess the kinetics of the network formation during solid‐state (co)polymerization (SSP) and the exchange kinetics of PBT/triol‐based vitrimers, a detailed study on two sets of model compounds was conducted. Model reaction set 1 is shown in Scheme 5.3. Butyl benzoate (BB), 1‐decanol and 2‐decanol were selected to mimic the gelation process of PBT/GLY‐based and PBT/TMP‐based copolyesters during SSP. For a typical model reaction, the butyl benzoate and 1‐decanol or 2‐decanol were mixed in a stoichiometric ratio in the presence of 0.2 mol% zinc(II) acetylacetonate with respect to butyl benzoate. Afterwards, the resulting mixture was heated in the temperature range between 120 and 190 °C, and gas chromatograms of the reaction were recorded at different time intervals. All the involved compounds could be easily followed by the well‐resolved signals (Appendix 4). If we assume that forward and reverse reactions (1) are both second order in their respective reagents, we obtain： dBB  k1 decanol BB  k1 decyl benzoate butanol (1) dt 12 At the beginning:1-decyl benzoate   butanol  0 and [1‐decanol] = [BB], so

dBB  2 kBB1  decanolkBB    (2) dt 11

k1 was then obtained by fitting the concentration data to the integrated second order rate Eq 3: 11 kt1 BB BB 0 (3)

Chapter 5

Finally, the activation energy was determined by plotting lnk1 against 1/T, according to the Arrhenius equation：

Ea 1 (4) lnk1   lnA RT A representative plot for the change in concentrations for model reaction (1) in Scheme 5.3 at 160 °C is shown in Figure 5.5A.

Butyl benzoate (A) 0.12 2.5 (B) or 1‐decanol k = 3.88 10‐3 Lmolmin‐1 ] 1 ‐1

L 1‐decanol  2.0 0 0.09 Fit 1

k = 5.14 10‐3 Lmolmin‐1 1.5 1 0.06 Fit 2

1.0 1‐decyl benzoate 1/[BB]‐1/[BB]

0.03 ' ‐5 ‐1 k = 2.33 10 Lmolmin 0.5 1‐butanol 1 Concentration [mol 2‐decanol 0.00 0.0 0 50 100 150 200 250 300 0 5 10 15 20 25 30 35 Time [min] Time [min] (C) ‐5 Primary OH ‐6 Ea 66 kJ/mol ‐7

k ‐8 ln ‐9 ‐10 Secondary OH

‐11 Ea 103 kJ/mol

‐12 2.12.22.32.42.5 1000/T[K]‐1 Figure 5.5. (A) A representative plot of the concentration versus time for model reaction (1) at 160 °C. (B) Second order linear fit for model reactions (1) and (2) at 180 °C. (C) Arrhenius plot for determination of activation energy of transesterification in model reactions (1) and (2).

The two reactants, butyl benzoate and 1‐decanol, both decrease from 2.5 to 1.4 molL‐ 1, while for the two products, the concentration of 1‐decyl benzoate increases from 0 to 1.0 molL‐1. However, the concentration of 1‐butanol only increases from 0 to 0.1 molL‐1. This is related to the volatility of 1‐butanol. The reaction temperature (160 °C) is much higher than the boiling point of 1‐butanol (Tboil = 118 °C). Therefore, during the reaction 1‐butanol evaporates and condensates on the glass wall of the vial. The representative second order plot is shown in Figure 5.5B, and the data in the first 30 minutes were used, where the reverse reaction is still negligible. It is evident from this plot that the reaction rate for the 73

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers primary OH with butyl benzoate is approx. 200 times higher than for the secondary OH. These observed differences are consistent with the difference gelation times between C8(TMP) and C8(GLY) during SSP at 180 °C (see Chapter 2),91 where the gelation time for

C8(TMP) (tgel 3 h) is roughly 4 hours earlier than C8(GLY)(tgel7 h). The Arrhenius plots are shown Figure 5.5C. For model reactions (1) and (2), the obtained Ea are approx. 66 and 103 kJ/mol, respectively. The model reaction set 2 is shown in Scheme 5.4. 3‐Phenyl‐1‐propanol (1°‐OH), 1‐ phenyl‐2‐propanol (2°‐OH), 1-decyl benzoate (1°‐ester), and 2‐decyl benzoate (2°‐ester) were selected to mimic the stress relaxation behavior of PBT vitrimer in the molten state.

Scheme 5.4. Chemical structure of the model molecules investigated to mimic the possible exchange reaction in the PBT/triol‐based vitrimers between the free hydroxyl and ester group.

Taking model reaction (4) as an example, 2-decyl benzoate (2°‐ester) and 3‐phenyl‐1‐ propanol (1°‐OH) was mixed in a stoichiometric ratio in the presence of 2 mol% zinc(II) acetylacetonate with respect to the ester bond. Afterwards, the resulting mixture was heated to the desired temperature (180, 190 and 200 °C), and gas chromatograms were recorded at different time intervals to follow the kinetics of model molecules. The reaction rate was defined as following:

 d 2 ester 2 2 (2kk'')2 ester  k ' 2  esterestr 2 e  k ' 2 es t er (5) dt 34 4 00 4

K’3 was calculated by fitting the concentration data of first 30 minutes to the integrated second order rate equation:

11' (6) kt3 22 ester ester 0 k’4 was calculated from the equilibrium constant: 2  ester 1  OH k ' K 3 (7)     k ' 1  ester 2  OH 4

Chapter 5

Figure 5.6A shows the combination of experimental results and the theoretical fitting data of the model reaction (4). Since none of these compounds evaporates at the experimental temperature and both the forward and reverse reaction were taken into consideration, the fit for the first 5 hours is perfect. It is worth to point out that the rate for model reaction (5) was extremely low and its conversion was nearly zero after 5 hours at 190 °C. Subsequently, the activation energies of model reaction (3) and (4) were determined by Arrhenius Eq 4, the Ea value is 108.8 kJ/mol (reaction (3)), 136.2 kJ/mol (backward reaction (4)) and 150.6 kJ/mol (forward reaction (5)) (Figure 5.6B). An overview of the reaction results is shown in Table 5.2.

1.1 experimental data E = 108.8 kJ/mol theoretical fitting ‐3 a 1‐OH + 1‐ester 1.0 180 C ‐4

  190C ‐5 2 ‐OH + 1 ‐ester 0.9 k Ea = 136.2kJ/mol ln ‐6 200 C 0.8

Concentration [mol/L] E = 150.6 kJ/mol ‐7 a (A) 1‐OH + 2‐ester 0.7 0 4000 8000 12000 16000 ‐8 Time [sec] 2.05 2.10 2.15 2.20 2.25 1000/T [K‐1]

Figure 5.6. (A) Second order linear fit for model reaction (4). (B) Arrhenius plot for the model reaction set 2 as shown in Scheme 5.4.

Table 5.3. Overview of the reaction results at 190 °C with the model compounds.

Reaction Rate coefficient/ Ea Equilibrium Type Condition Lmol‐1s‐1 /kJmol‐1 time/h

1‐OH+1‐ester 2.44  10‐2 108.8 2

1‐OH+2‐ester 190 °C/ 1.96  10‐3 150.6 24 2 mol% Zn(II) 2‐OH+1‐ester catalyst 5.39  10‐3 136.2 13

2‐OH+2‐ester 0 n.d.a n.d.a an.d., it is not able to determine within the experimental conditions used in this chapter.

It is clearly shown in Table 5.2 that the exchange reaction rates follow the order: 1°‐OH + 1°‐ester > 2°‐OH + 1°‐ester > 1°‐OH + 2°‐ester >> 2°‐OH + 2°‐ester. These model reactions 75

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers suggest that a vitrimer system composed of primary OH and ester groups should possess a shorter stress relaxation time (fast exchange kinetics) and lower activation energy (Ea). However, the PBT/TMP‐based vitrimers show much slower stress relaxation kinetics than PBT/GLY‐based vitrimers. These results are not expected and more experiments or a better design of the model reaction system needs to be done in order to make solid conclusions. Nevertheless, one possible explanation for this discrepancy could be that the PBT/TMP‐ based vitrimers possess no or very few free hydroxyl groups after curing due to the higher reactivity of 1°‐OH over 2°‐OH and the large excess of ester groups from the PBT backbone.

5.2.5 Welding experiments

In this part, we study how these differences in relaxation time and activation energy will influence the welding behavior of PBT/triol‐based vitrimers. Weldability is one of the unique properties of vitrimers, as for a cured thermoset large chain segment diffusion via reputation is not possible for interfacial welding between two polymer surfaces.206 The welding behavior of these materials was studied experimentally55,71,93,94,106,169 and theoretically.207,208 In this study, we will use one particular composition, C8(GLY) and C8(TMP), as an example to demonstrate how this different temperature‐dependent stress relaxation behavior influences the weldability. In our experiments, two rectangular samples (5.4 mm (width) 0.6 mm (thickness)) were stacked with an approximately 18 mm overlap length and held together under pressure for welding times ranging from 5 to 60 min at 250 °C and subsequently cooled to room temperature via fast cooling (40 °Cmin‐1). The experimental set‐up is schematically shown in Appendix 4. As we know from Chapter 4 that PBT and PBT vitrimers are typically brittle materials, we quenched the welded materials to increase their toughness.209 In this way, we are able to ensure that the material is not broken before the welded part is separated. The weldability was then evaluated by carrying out tensile tests at room temperature with a cross‐head speed of 50 mmmin‐1 on the welded samples and comparing the force at break or strain at break. In Figure 5.7A, when the samples were welded at 250 °C under a 0.5 N axial force, after 5 min welding time, the stress‐strain curve of C8(GLY) shows clear a yield point, necking, and a significantly high toughness because of quenching, eventually, the area between clamp and uncontacted part is broken with a force at break of ∼126 N. The welded part cannot be separated anymore which indicated that the interface is well established due to the fast exchange reactions (Figure 5.7B, photo (1)). Under the same conditions, the C8(TMP) was not welded at all and it is separated immediately after cooling down to room temperature. This striking difference in weldability correlates well with the stress relaxation time differences as shown in Figure 5.4.

Chapter 5

140 5 min, 0.5N, (A) 120 C8(GLY)(1)

100 5 min,10N, 80 C8(TMP)(3) 60 min,0.5N, 60 C8(TMP) (2)

Force [N] 40 5min, 1N, 20 C8(TMP) 0 012345 Displacement [mm]

Figure 5.7. (A) Stress−strain curves of samples welded under diﬀerent condions. (B) The pictures of 3 representative welding conditions after tensile testing; the corresponding tensile curve is indicated with numbers: (1) C8(GLY), 5 min, 0.5 N; (2) C8(TMP), 60 min, 0.5 N; (3) C8(TMP), 5 min, 10 N.

In order to reach a similar welding efficiency as the C8(GLY) for the C8(TMP), the influence of time and axial force (pressure) were studied at constant temperature. A clear trend can be observed, i.e., increasing either the contacting time or pressure accelerates the welding process. As shown in Figure 5.7A, with the increase of welding time from 5 to 60 min while keeping the axial force constant (0.5 N), the force at separation increased from 0 to 106 N and the two welded parts were separated completely at the end of the tensile test (Figure 5.7B, photo (2)). When increasing the pressure from 0.5 to 10 N while keeping the welding time constant (5 min), the force to separation changed from 0 (0.5 N axial force) to 73 N (1 N axial force) and 103 N (10 N axial force) (Figure 5.7A). A higher pressure helps to increase the real contact area by squeezing surface rough topography. Although the force at break increased dramatically with an increase in pressure, the C8(TMP) is still not completely welded together as indicated by the red circle in Figure 5.7B, photo (3). To summarize, on one side, when the same pressure is applied, C8(GLY) with faster relaxation time shows a much better weldability than C8(TMP). A faster network rearrangement rate renders a faster establishment of covalent bond formation in the contact interface, thus a 77 Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers better weldability; on the other side, the true contact area is also essential for the welding efficiency since a high pressure can increase the real contact area and subsequent exchange reaction to form a covalent bond in the contact interface.

5.2.6 Creep and mechanical properties

The effect of chemical structure of the cross‐linker on the creep resistance below and above the melting temperature (Tm) of PBT vitrimers was investigated by elongational creep experiments. Figure 5.8 displays the time‐dependent strain at constant stress (creep) and subsequent strain recovery for C8(GLY), C8(TMP) and neat PBT.

8 14 (B) C8(TMP) (A) 7 12

6 10 5 PBT 8 4 6 3 Strain % Strain % 4 2 C8(GLY) 1 2 0 0 0 50 100 150 200 0 50 100 150 200 250 Time [min] Time [min]

Figure 5.8. Creep resistance for C8(GLY) and C8(TMP) at (A) T < Tm, 150 °C (circle), 180 °C (up triangle

), 200 °C (square). (B) T > Tm, 230 °C (○, ●) and 250 °C (▲, Δ). Neat PBT (black line and symbols), C8(GLY)( blue line and symbols) and C8(TMP) (red line and symbols).

In these experiments, a constant nominal stress0 of 2 MPa or 0.03 MPa was applied to the samples below and above Tm, respectively. An overview of the creep resistance properties for the prepared PBT vitrimers is given in Table 5.3. In general, the neat PBT has a higher tensile modulus (E) than the PBT vitrimer below Tm, and the maximum creep strain or the steady compliance increases monotonously with temperature, which is a characteristic trend for most viscoelastic materials. It is seen that the C8(GLY) material possesses a better creep resistance than neat PBT at T 150 °C in terms of maximum creep strain and creep rate (neat PBT has no resistance above Tm at all). The E for neat PBT and C8(GLY) is about 302 MPa and 194 MPa at 200 °C, however, the maximum strains for neat PBT and C8(GLY) are 5 and 3.5% after 4800 s under a constant load of 2 MPa, respectively. The C8(TMP) shows a poorer creep resistance in comparison to both the neat PBT and the C8(GLY). The C8(TMP) exhibits a lower and broader melting temperature as compared with neat PBT and the C8(GLY)(as can be seen from the DSC curve in Figure 5.2 and DMTA curve

Chapter 5 in Figure 5.3). Thus the C8(TMP) was always partially melted at 200 °C, which leads to a E as low as 108 MPa.

Table 5.4. Overview of the creep properties of the samples neat PBT and PBT vitrimers at T < Tm under

2 MPa step stress and T > Tm under 0.03 MPa step stress.

Creep E Maximum creep Creep rate Irrecoverable Sample name temperature [MPa] strain [%] [%/hour] strain[%] [°C]

150 500 0.8 0.007 0.05

PBT 180 370 2.9 0.08 1.1

200 302 4.9 0.2 2.7

150 342 0.9 0.01 0.1

180 247 2.1 0.09 0.7

C8(GLY) 200 194 3.5 0.3 1.4

230 1.5 3.8 1.0 2.5

250 1.6 12.8 1.7 11.8

150 225 3.9 0.3 1.3

180 149 7.2 0.5 3.1

C8(TMP) 200 108 104 18.3 77.8

230 2.3 1.4 0.2 0.7

250 2.6 9.7 0.6 9.0

In summary, the exchangeable cross‐links can improve the creep resistance of semi‐ crystalline PBT vitrimer at service temperature (e.g., at T < 200 °C) when a comparable degree of crystallinity or E’ is maintained as neat PBT. For a similar cross‐linker, catalyst concentration and thermo‐mechanical history, a general trend for the creep resistance

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

below Tm is stated as follows: C8(GLY) > neat PBT > C8(TMP). Above Tm, C8(TMP) shows a better creep resistance than C8(GLY) and this phenomenon can be explained by two possible ways: (1) C8(TMP) possesses a much longer relaxation time than C8(GLY) at the same temperature; (2) the different exchange dynamics of the network topology leads to a slightly different network density. The creep experiments above Tm also confirm that the PBT vitrimers behave like a viscoelastic liquid. When the stress is released, the material recovers only its initial elastic response and a permanent deformation remains. The mechanical behavior of the PBT vitrimers is further examined by tensile tests on the samples after annealing at 120 °C for 4 h. The Young's modulus (elastic modulus) is an important mechanical parameter of materials because it is related to the ability of a material to resist elastic deformation when loaded and is indicative of the rigidity of the material. In the glassy state, segmental motions are largely restricted and the polymer has a high rigidity, leading to a relatively high Young's modulus and low strain at break. As can be seen from Figure 5.7, neat PBT fails in a brittle manner without yielding with an average elongation at break of 5% at room temperature. In comparison, all tested PBT vitrimers were found to have an increased or comparable elongation at break.

70 (A) 70 (B) C8(TMP) C1(GLY) 60 PBT 60 PBT 50 50 C8(GLY) 40 40 C13(TMP) C1(TMP) 30 30 Stress [MPa] Stress [MPa] 20 20 C13(GLY) 10 10

0 0 0 5 10 15 20 25 0 5 10 15 20 25 Strain [%] Strain [%] Figure 5.9. Characteristic stress‐strain curves of the PBT and PBT vitrimers: (A) PBT/GLY‐based vitrimer and (B) PBT/TMP‐based vitrimer. Experiments were performed on dog‐bone specimens under uniaxial tension at room temperature (23 °C) using a constant cross‐head speed of 50 mmmin‐1.

However, there is no clear trend for the elongation at break in the glassy state for those samples. On the one hand, introducing cross‐links will lead to a decrease of the degree of crystallinity and thus more tie molecules between crystals, which can enhance the strain at break of the material. On the other hand, the cross‐linked network limits the deformation of the material since it is not able to undergo plastic deformation below its glass transition temperature. The Young’s modulus of PBT is 800 ± 50 MPa, and most of the samples show a comparable Young’s modulus as PBT. However, the Young’s modulus of the C8(GLY) and

Chapter 5

C13(GLY) are 1.4 times higher, i.e., about 1100 MPa, respectively. All the materials show a comparable tensile strength as neat PBT, which is around 52  7 MPa, with the C13(TMP) system as an exception. An overview of the tensile properties for the prepared PBT vitrimers is presented in Table 5.3. In summary, with the same amount of cross‐linker and catalyst, the mechanical performance of PBT/glycerol‐based vitrimers are better than PBT/TMP‐ based vitrimers in terms of Young’s modulus, tensile strength and elongation at break.

Table 5.5. Tensile properties of neat PBT and PBT vitrimers.

Young’s modulus Tensile strength Elongation at break Sample name [MPa] [MPa] [%]

PBT 800  50 52  7 5  1

C1(GLY) 671  71 71  1 20  2

C1(TMP) 759  135 58  14 15  15

C8(GLY) 1052  42 54  5 5  1

C8(TMP) 922  162 61  6 10  6

C13(GLY) 1111  49 63  2 26  11

C13(TMP) 749  397 34  15 4  2

To explore the morphology of the fracture surface of the PBT and PBT vitrimers, the samples after tensile test were imaged by using scanning electron microscopy (SEM). As we discussed above, PBT is a relatively brittle material and the fracture surface of PBT is smooth, which is attributed to the hindered plastic deformation due to the relatively high rigidity of the polymer matrix,210 as shown in Figure 5.10A. The same sharp, clear, and ligament shaped fracture surface morphology is also observed for the C8(GLY) (Figure 5.10C) which has a similar elongation at break as neat PBT. The micrographs of the fracture surfaces of the C1(GLY) and C13(GLY) are shown in Figures 5.10B and 5.10D. Since they possess a relatively higher strain at break in comparison to neat PBT and the C8(GLY), which are about 19.6 and 26.0%, respectively, the fracture surface morphology changes from a smooth surface to a rough surface with extensive fibrillation of the matrix. These observations suggest an enhanced ductility of the material which is normally observed for 211 PBT during the deformation above Tg when the molecular mobility is improved. In

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers general, PBT/TMP‐based vitrimers are brittle with C1(TMP) as an exception, therefore a similar brittle fracture surface morphology as neat PBT was observed and the images of PBT/TMP‐based vitrimers are shown in Appendix 4.

Figure 5.10. Representative SEM micrographs of the fracture surfaces of (A) PBT, (B) C1(GLY), (C) C8(GLY), and (D) C13(GLY) at 23 °C. 5.3 Conclusions

A series of semi‐crystalline PBT/triol‐based vitrimers were prepared by the incorporation of two different cross‐linkers, i.e. glycerol (GLY) and 1,1,1‐ tris(hydroxymethyl)propane (TMP) into PBT via a two‐step solid‐state (co)polymerization protocol. The gelation time differences between the PBT/TMP and PBT/GLY‐based copolyesters with a similar cross‐linker and Zn(II) catalyst content during SSP at 180 °C are in good agreement with the small model molecules study with identical Zn(II) catalyst concentrations. For the obtained PBT/triol‐based vitrimers containing a similar triol and Zn(II) content, similar gel contents and rubber plateau moduli were found but with a large difference in thermal properties and stress relaxation time, indicating that the network dynamics is highly dependent on the chemical structure of the cross‐linker, i.e., primary/secondary hydroxyl group. Both C8(TMP) and C8(GLY) show a gel content of around 80% and a plateau modulus of about 2 MPa, while the crystallization temperature

Chapter 5 of C8(TMP) is 13 °C lower than C8(GLY) and its relaxation time is 4 times higher than C8(GLY) at 250 °C. The creep resistance of the PBT/GLY vitrimer was better than PBT/TMP‐based vitrimer at Tg < T < Tm, while at T > Tm, the trend is reversed. The different relaxation times between the PBT/GLY‐based and PBT/TMP‐based vitrimers are also evident from the welding experiments. C8(GLY) with a relatively short relaxation time shows a good adhesion after welding at 250 °C, while C8(TMP) with a relatively long relaxation time has nearly no adhesion. In general, the mechanical performance, i.e., Young’s modulus and tensile strength, of PBT vitrimers are better than of neat PBT in the glassy state, although the PBT/TMP‐based vitrimers have the same characteristic brittle behavior as PBT. Interestingly, the PBT/GLY‐ based vitrimers show a ductile behavior when 13 mol% glycerol is incorporated and this ductile behavior is also confirmed by SEM analysis which shows the fracture surface morphology of PBT/GLY‐based vitrimers changed from a smooth fracture surface to a fibrillated surface due to extensive yielding. This work shows the possibility to tune the viscoelastic properties of PBT vitrimers through simple perturbations of the primary and secondary hydroxyl groups in a multifunctional alcohol, which provides another design parameter for PBT/triol‐based vitrimers with a tunable exchange kinetics besides the common catalysis approach. 5.4 Experiment section

Materials. Butyl benzoate, 2‐decanol, triethylamine and hexadecane, zinc acetylacetonate hydrate

(Zn(II)(acac)2 ·xH2O, powder), benzoyl chloride (99%), 1,4‐butanediol (99%), anhydrous pyridine (99.8%), anhydrous magnesium sulfate (MgSO4, 99.5%), sodium bicarbonate

(NaHCO3, 99.7%), glycerol (GLY, 99.5%), 1,1,1‐tris(hydroxymethyl)propane (TMP, 97%), trifluoroacetic acid (TFA, 99%), butyl benzoate (99%), 2‐decanol (98%), benzoyl chloride (99%), trimethylamine (99%), hexadecane (anhydrous, 99%), 3‐phenyl‐1‐propanol (98%) and 4‐phenyl‐2‐butanol (97%) were all obtained from Sigma‐Aldrich. Acetonitrile, methanol, diethyl ether and MilliQ water (LC‐MS grade) were obtained from Biosolve. Tetrahydrofuran (THF, HPLC grade) was obtained from Rathburn. Deuterated chloroform

(CDCl3, 99.8 atom% D) was obtained from Cambridge Isotope Laboratories. 1‐decanol (>98%) were purchased from TCI Europe. All chemicals were used as received, unless denoted otherwise.

Solution preparation of a physical PBT/triol mixture A common solution approach was used to prepare the physical mixtures containing PBT, triol and catalyst as described in previous chapters.91,187 For example, for the preparation of a PBT/TMP based vitrimer containing 8 mol% TMP and 0.2 mol% catalyst, 75.82 g PBT, 83

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

4.00 g TMP, 0.18 g Zn(acac)2 and 100 ml 1,1,1,3,3,3‐hexafluoroisopropanol (HFiP) as solvent were added into a 250‐ml round bottom flask, and heated to 55 °C in an oil bath under continuous stirring. HFiP was distilled off when the mixture was completely dissolved. When the material started to crystallize, a vacuum was applied to enhance the removal of HFiP. Finally, the obtained material was dried in a vacuum oven at 40 °C for 24 h. The physical mixture was cooled in liquid nitrogen and subsequently ground into powder using an analytical laboratory mill. This powder was then dried in a vacuum oven at 40 °C for a period of 48 h. The PBT/glycerol physical mixture will be abbreviated in this report as C8(GLY), and the PBT/TMP physical mixture will be abbreviated in as C8(TMP). 0.2 mol% of

Zn(acac)2 was used for both samples.

Preparation of PBT/triol‐based copolyesters PBT/triol‐based copolyesters were synthesized by a two‐step solid‐state polymerization 91,187 method. Typically, in the first step, 5 g of PBT/triol powder containing 2 mol% Zn(acac)2 was placed in a closed vial,164 and pressurized with argon (p < 3 bar) to avoid evaporation of glycerol. After 24 h at 160 °C, the mixture was transferred to the SSP reactor,165 which was a glass tube (inner diameter = 2.4 cm) with a sintered glass plate at the bottom. A heat exchange glass coil (inner diameter = 0.5 mm) surrounded the reactor and entered the inner glass tube at the bottom just below the glass plate. The nitrogen gas was heated by passing through this coil prior to entering the reactor and its flow was controlled by a flow meter. The powder bed was fixed by addition of glass pearls (diameter = 2 mm) on top of the powder, and the reactor was purged with a nitrogen flow of 0.5 Lmin‐1 for 30 min. After flushing, the reactor was placed in a heated salt bath (T = 180 °C). When the temperature inside the reactor reached 180 °C, the measurement of the reaction time (tssp) was initiated, and the composition and molecular weight were followed until the gel point. After the reaction, the product was cooled down to room temperature under a continuous nitrogen flow, discharged from the reactor, and the obtained polymer dried under vacuum at 120 °C for 6 h. The prepared PBT/GLY‐based copolyesters will be abbreviated as Cx, where x indicates the mol% of glycerol (calculated based on the amount of PBT repeat units). The composition as determined by 1H‐NMR spectroscopy is used for the abbreviations in this chapter.

Synthesis of model molecules 1‐decyl benzoate. All the compounds were synthesized by a similar procedure. A typical method is described as follows. To a 100 ml round bottom flask, benzoyl chloride (2.64 g, 14.8 mmol) was added dropwise to a solution of 1‐decanol (2.35 g, 14.8 mmol), triethylamine (10 mL) and THF (10 mL). After stirring for 24 h at room temperature, the mixture was extracted with CHCl3 to remove triethylamine hydrochloride, repeatedly

Chapter 5

washed with 10 % NaHCO3 and water. Then the product was dried with anhydrous Na2SO4 overnight, evaporated to dryness and finally purified by column chromatography using an eluent of hexane‐ethyl aecetate, 10:1 mixture. The product was analyzed via 1H‐NMR and 13 1 C‐NMR spectroscopy to verify its structure and purity. H‐NMR (400 MHz, CDCl3, δ in ppm):

0.8 (t, 3H, ‐CH3), 1.2‐1.5 (m, 14H, ‐CH2), 1.8 (m, 2H, ‐CH2), 4.3 (t, 2H, ‐COOCH2‐), 7.4, 7.5, 8.0 13 (m, 5H, aromatic H); C‐NMR (400 MHz, CDCl3, δ in ppm): 14.5 (1C, ‐CH3); 22.5, 25.7, 28.9,

29.1, 29.3, 29.6, 30.1, 32.2 (8C, ‐CH2), 65.8 (1C, ‐CH2), 128.2, 129.1,129.3, 130.2, 131.4, 132.6 (aromatic C), 168.8 (1C, ‐COO‐). 1 2‐decyl benzoate. H‐NMR (400 MHz, CDCl3, δ in ppm): 0.8 (t, 6H, ‐CH3), 1.2‐1.5 (m, 12H,

‐CH2), 1.6‐1.8 (m, 2H, ‐CH2), 5.3 (m, 2H, ‐COOCH‐), 7.4, 7.5, 8.0 (m, 5H, aromatic H). 1 3‐phenylpropyl benzoate. H‐NMR (400 MHz, CDCl3, δ in ppm): 2.0 (m, 2H, ‐CH2), 2.4

(t, 2H, benzylic H), 4.3 (t, 2H, ‐COOCH2‐), 7.2‐8.0 (d, 10H, aromatic H).

Model reactions

Typically, Zn(acac)2 (0.04 g, 0.2 mol%) was added into 1‐decanol/2‐decanol (5.86 g, 37 mmol) in a 100‐mL round bottom flask at 50 ºC with stirring until catalyst powder was completely dissolved. The internal standard hexadecane (250 μL) and butyl benzoate (6.59 g, 37 mmol) were then added into the mixture with continuously stirring. After the catalyst was homogeneously distributed in the mixture, the glass vial was equally divided into eight 10 mL glass vials. The glass vials were subsequently heated in an oil bath to the desired temperature. The experimental temperatures of the model reaction between 1‐decanol and butyl benzoate were 120, 140, 160, 180 °C. However, due to the limited reaction rate for the model reaction between 2‐decanol and butyl benzoate, the temperature was adjusted higher to 160, 170, 180, and 190 °C. Samples were taken at predetermined time intervals with a syringe, and analyzed via GC chromatograph with a FID detector. In order to limit the influence of the reverse reaction, samples from the first 30 minutes were analyzed for determination of the rate coefficient of the forward reaction. Butyl benzoate is abbreviated as BB. Characterization methods Nuclear magnetic resonance spectroscopy. 1H‐NMR spectroscopy was performed on a 400 MHz Bruker Avance III spectrometer at 25 °C. The spectral width was 6402 Hz, the delay time was 5 s and the number of scans was 64. Samples were prepared by dissolving ~15 mg of the crude polyester in 0.8 mL of an 80:20 vol % CDCl3: d‐TFA mixture. Chemical shifts are 13 reported in ppm relative to the residual solvent peak of CDCl3 (δ = 77.0 ppm). For the C‐ NMR experiments, the spectral width was 24,154 Hz, the acquisition time was 1.300 s, the delay time was 2 s and the number of scans was between 2000 and 5000. Samples were prepared by dissolving 6 mg of the sample in 1 mL of CDCl3. Chemical shifts are reported in

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

ppm relative to the residual solvent peak of CDCl3 (δ = 77.0 ppm) and trimethylsilyl peak of (δ = 0 ppm). The data analysis was performed using MestReNova software. Size exclusion chromatography (SEC). Molecular weight distributions, the number average molecular weight (Mn) and polydispersity index (Đ) of the copolyesters were measured on a system equipped with a Waters 1515 isocratic HPLC pump, a Waters 2414 refractive index detector (40 °C), a Waters 2707 autosampler, and a PSS PFG guard column followed by a 2PFG‐linear‐ XL (7 μm, 8 × 300 mm) columns in series at 40 °C. HFIP with potassium trifluoroacetate (3 g.L−1) was used as eluent at a flow rate of 0.8 mLmin−1. The molecular weights were determined relative to poly(methyl methacrylate) standards 6 (Polymer Laboratories, Mp = 580 Da up to Mp = 7.1 × 10 Da). Gel fraction. The mass fraction of the cross‐linked part of the PBT/triol‐based vitrimers, which is defined as the gel fraction, was calculated by the following method. A patch of PBT vitrimer (~0.5 g) was weighed and immersed in HFiP in a 10‐mL glass vial for 36 h at room temperature. The samples were subsequently filtered and dried in a vacuum oven at 120 °C for 3 days. The gel fraction was calculated with the equation: gel fraction% ∗ 100% minitial, mass of the sample before extraction, mdry, mass of the sample after extraction dried in a vacuum oven at 120 °C for 72 h. Differential scanning calorimetry (DSC). The thermal properties were measured using a Q1000 DSC from TA Instruments. The measurements were carried out from ‐50 to 250 °C with heating and cooling rates of 10 °Cmin‐1 under a nitrogen flow of 50 mLmin‐1. The mobile amorphous fraction (MAF) was determined via the heat capacity increase at half‐ step (cp) in modulated DSC (TMDSC) mode using the same DSC equipment. An oscillating heat flow signal with a period of 60 s and amplitude of 0.5 °C was used with an underlying heating rate of 2 °Cmin‐1. The copolyester samples, prepared by SSP, were measured in the temperature range from 0 to 180 °C. Compression molding. The materials were compression molded at 250 °C and 100 bar for 25 minutes in a Collin Press 300G and subsequently cooled with water. Dynamic mechanical thermal analysis (DMTA). Compression‐molded samples (ca. 10.0 (Length) × 5.0 (Width) × 1.0 (Thickness) mm) were measured on a Q800 DMA (TA Instruments) with a film tension setup. A temperature sweep from ‐50 to 270 °C was performed with a heating rate of 3 °Cmin‐1 at a frequency of 1 Hz. A preload force of 0.01 N, an amplitude of 10 µm and a force track of 125% were used. The storage modulus and loss modulus were recorded as a function of temperature. The glass transition temperature was calculated from the peak maximum in the loss modulus.

Chapter 5

Rheometry. Dynamic shear measurements were performed on a stress‐controlled AR‐ G2 Rheometer (TA Instruments) by using a 25‐mm parallel plate geometry and disk‐shaped specimens (25 mm diameter; 1 mm thick). Frequency sweeps from 100 to 0.01 rad/s were performed at a temperature range between 240 °C with a strain of 1%, which is in the linear viscoelastic regime. Stress relaxation experiments were performed at a temperature range between 230‐270 °C with a strain of 1% and the relaxation modulus was monitored as a function of time. A constant normal force of 20 N was applied to ensure a good contact with the plates. Tensile properties. Tensile tests were performed at a constant deformation rate of 50 mmmin‐1 using a Zwick Z100 tensile apparatus with 1 kN load cell. Samples were prepared via compression molding and cut into dog‐bone shaped samples of 35 mm x 2 mm x 1 mm. Scanning electron microscopy (SEM). The morphology of the fractured surface of samples with a thin gold layer was analyzed using JSM‐IT100 scanning electron microscope (SEM) (Japan) operating in a high vacuum mode and an accelerating voltage of 5 kV was used. Gas Chromatography (GC). The concentration of each model compounds at predetermined time interval during a model reaction was monitored by a Varian 450 equipped with a flame ionization detector (FID). Samples were prepared by quenching 0.05 mL of the reaction mixture with 1 mL of chloroform. Thermogravimetric analysis (TGA). Thermal stability studies were performed on a TA

Instruments TGA Q500 machine under a N2 rich atmosphere. Samples were heated at 10 °C.min‐1 from 20 to 600 °C. Temperature calibration was performed using the Curie points of high‐purity aluminum, nickel and perkalloy standards.

Chemical structure effect of the triol cross‐linker on the viscoelastic properties of PBT vitrimers

Chapter 6

In situ network formation in PBT vitrimers via processing‐induced deprotection chemistry

ABSTRACT: Although the network dynamics and mechanical properties of PBT vitrimers can to some extent be controlled via chemical and physical approaches, it remains a challenge to be able to process PBT vitrimers with the same injection moulding conditions as neat PBT. Here, we show a new processing‐induced debenzalation process, which shares the same deprotection mechanism as common acid‐promoted debenzalation, leading to in situ network formation. This process is tunable via playing with the processing temperature and 5,5‐bis(hydroxymethyl)‐2‐phenyl‐1,3‐dioxane (BPO) content. A similar dynamic network, in terms of network density, melt elasticity and exchange reaction kinetics, was obtained via this and standard debenzalation strategies. This solvent‐free deprotection strategy allows for high production rates of PBT vitrimer products via injection moulding with the combination of low viscosity during processing and vitrimer characteristics in the final product.

This chapter is partially adapted from: Zhou, Y.; Goossens, J. G. P.; van der Bergen, S.; Sijbesma, R. P.; Heuts, J. P. A. In preparation.

In situ network formation in PBT vitrimers via processing‐induced deprotection chemistry

6.1 Introduction

Previously, we developed poly(butylene terephthalate)(PBT) vitrimers via incorporation of a triol (glycerol91,187,201 and trimethylolpropane212) into PBT in the solid state. Depending on the chemical composition (i.e., the catalyst and cross‐linker content) and polymerization conditions in the solid state, in most of the cases, the materials were cross‐linked during the solid‐state polymerization. In general, the terminal relaxation regime of PBT vitrimers was rarely observed at angular frequencies down to 10‐2 s‐1 in the temperature range from 250 to 270 °C.90,91,187,201,212 Considering that typical shear rates during injection molding are as high as 10,000 s‐1 and the viscosity of PBT vitrimers can be up to 107 Pas at 250 °C, these PBT vitrimers cannot be processed like neat PBT during the short residence times typical for extrusion and injection molding.213 Here, we propose a controllable way to process PBT vitrimers (based on transesterification exchange reactions) via controlling the network formation with the help of protection‐deprotection chemistry. In this way, we can overcome the relative long relaxation times and high viscosities associated with PBT vitrimers and maintain the high conversion rates as neat PBT by injection molding. Accidentally, when compression molded a linear copolyester with benzal protection groups at 250 °C, we discovered that a cross‐linked material was obtained. This observation was in contrast to the earlier study by Collard et al.,190 in which the polymers were found to be thermally stable to repeated thermal cycling up to 300 °C by differential scanning calorimetry after the incorporation of 5, 5‐bis(phenylcarboxymethyl)‐2‐phenyl‐1,3‐dioxane (BPO) into PBT chains. Driven by this intriguing result and its practical relevance, in this chapter, firstly, the in situ network formation kinetics were investigated with oscillatory time sweep experiment using the powder directly after SSP. Secondly, the vitrimer characteristic properties of the materials obtained via our new processing‐induced deprotection method will be studied and compared with the reference samples obtained via common trifluoroacetic acid (TFA)‐promoted deprotection method. 6.2 Results and discussion

The experimental strategy is schematically shown in Scheme 6.1: pentaerythritol is first converted into 5,5‐bis(hydroxymethyl)‐2‐phenyl‐1,3‐dioxane (BPO) (m.p. 135‐137 °C) via benzaldehyde protection chemistry.190 Subsequently, this diol is incorporated into the PBT backbone via solid‐state (co)polymerization (SSP) to form a linear copolyester in line with our earlier work on PBT modification in the solid state.143,145,156–158,173,175 The linear copolyester is then transformed into a network via processing‐induced deprotection and subsequent crosslinking.

Chapter 6

Scheme 6.1. Schematic representation of using protection‐deprotection chemistry to synthesize PBT/pentaerythritol‐based vitrimers via solid‐state (co)polymerization (SSP).

Materials containing 1, 2.4, 3.5 and 9 mol% of 5, 5‐bis(phenylcarboxymethyl)‐2‐phenyl‐ 1,3‐dioxane (BPO) were prepared by using the solid‐state polymerization method as reported in our previous work (see the Appendix 5 for details).91 The composition obtained 1 from H NMR spectroscopy analysis was used for the abbreviations; e.g., pC1 and deC1, where pC stands for the linear copolyester with the benzal protection group, and deC stands for the linear copolyester with pendant hydroxyl groups after TFA‐promoted deprotection. The chemical structures of the linear copolyester with the benzal protection group or pendant hydroxyl groups are shown in Scheme 6.1. C1 indicates the molar percentage (1 mol %) of mono‐benzal pentaerythritol‐ and 1,4‐BD‐based repeat units. The catalyst loading was fixed at 0.2 mol% with respect to the PBT repeat unit in this study. As it is schematically 91

In situ network formation in PBT vitrimers via processing‐induced deprotection chemistry shown in Scheme 6.2, taken the sample containing 1 mol% crosslinker as an example, before performing rheology and dynamic mechanical thermal analysis (DMTA) experiments, pC1 is compression molded to different shape at 250 °C with a total heating time around 25 min, and this material is named compression‐molded pC1. A further curing step is performed to obtain a fully cured sample, and it is named pX1. For the fully cured reference sample, which is prepared by TFA‐promoted deprotection method, is named deX1.

Scheme 6.2. Schematic representation of the abbreviations for the materials studied in this chapter.

6.2.1 In situ network formation during processing

In order to investigate in situ network formation during the processing, oscillatory time sweep experiments with 1 % strain at different temperatures were employed to monitor the molecular structure changes of the material (pC1) directly after SSP or after debenzalation (deC1) and a ring‐shaped melting chamber setup surrounding the lower plate were charged with the materials (Appendix 5). The time dependence of the storage (G) and loss (G) moduli of the material after SSP is shown in Figure 6.1A. It is immediately clear from this Figure that cross‐linking takes place (G crossing G) during both experiments and that this happens first for the unprotected polymer. The initial complex viscosities for pC1 and deC1 were about 2.1 kPa.s and 3.6 kPa.s at 250 °C and clear cross‐over (gel) points (G =

G) were observed around 65 min and 17 min for pC1 and deC1, respectively. After the cross‐ over point (G = G), both samples were insoluble in 1,1,1,3,3,3‐hexafluoroisopropanol at room temperature and a gel fraction of approx. 76% is obtained for both materials (Appendix 5).

Chapter 6

106 106 (A) (B) pC3.5

pC2.4 deC1 105 105

pC1 pC1

104 104 G'(open), G''(closed) [Pa] G'(open), G''(closed) [Pa] Cross‐over point Cross‐over point 103 103 101 102 103 104 101 102 103 104 Time [s Time [s Figure 6.1. Time dependence of the storage modulus G (open symbols) and loss modulus G (filled symbols) at 250 °C for the linear materials: (A) comparison between protected (pC1) and unprotected

(deC1) material; (B) comparison of different protected copolymers with different amounts of 5, 5‐ bis(phenylcarboxymethyl)‐2‐phenyl‐1,3‐dioxane (BPO).

The cross‐over time can be shortened by increasing either the molar content of BPO or the processing temperature. For instance, the cross‐over times was 65 min, 6.5 min and 4.3 min for pC1, pC2.4 and pC3.5, resp. (Figure 6.1B). With the increase of temperature from 250 to 270 °C, the cross‐over time of pC1 was decreased from 65 min to 36 min (Appendix 5). This new processing‐induced debenzalation process allows the material to be processed as a thermoplastic with a low viscosity for a controllable time, which is beneficial for a processing technique such as injection molding. The linear material prepared by SSP possesses similar thermal properties as neat PBT, such as Tm, Tc and degree of crystallinity

(cr) (the linear pC1, Tm= 221.5 °C, Tc = 193.3 °C and cr = 38.3%; Neat PBT, Tm= 221.6 °C, Tc =

193.7 °C and cr = 42%). Differential scanning calorimetry (DSC) was also employed to characterize the in situ network formation. As shown in Figures 6.2A and 6.2B, upon repeated cycling between 25 and 250 °C, both the crystallization and melting peak of pC3.5 and deC3.5 shift to lower temperatures. In Figure 6.2C, clear decreasing trend for the values of degree of crystallinity, crystallization‐peak temperature (Tc) and melting‐peak temperature (Tm) are observed. This deterioration of thermal properties can be attributed to the retardation of the crystallization process of the PBT vitrimer by the formed branching or cross‐linking points. These DSC results are in line with the time‐dependent G evolution (Figure 6.1A).

In situ network formation in PBT vitrimers via processing‐induced deprotection chemistry

1.5 1.5 (A) (B)

] 1.0 1.0 g / w [ 0.5 0.5 0.0 0.0

ow, exo up ‐0.5 fl Scan number ‐0.5 scan number

‐1.0 Heat flow, exo up [w/g] Heat ‐1.0 C3.5 pC3.5 de ‐1.5 120 140 160 180 200 220 240 120 140 160 180 200 220 240 Temperature [C] Temperature [C] 220 184

215 44 180

210 40 176 C  C 

/ 205 172

m 36 / c T T 200 168 32 195 164 Degree of crystallinity / % 28 190 160 012345678910 Scan number

Figure 6.2. DSC thermograms for (A) pC3.5 and (B) deC3.5. The temperature was cycled between 25 and 250 °C for 8 cycles, holding the polymer at 250 °C for 5 min for each heating and cooling cycle. (C)

Plots of degree of crystallinity (circle) (c), crystallization temperature (up triangle) (Tc) and melting temperature (square) (Tm) vs. cycle number for PC3.5 (filled symbol) and deC3.5 (open symbol).

6.2.2 Frequency/temperature‐dependent mechanical properties

Figure 6.3A shows thefrequency dependence of the G and G data at 250 °C measured for the pX1 and deX1. The fully cured pX1 and deX1 exhibited a solid‐like gel behavior over the entire tested frequency range (i.e. G > G) and displayed a plateau modulus (Gp) for cured pX1 and deX1, taken at the minimum points of G, of around 0.035 and 0.14 MPa, respectively. The solid‐like mechanical responses of the cured materials were also demonstrated by the frequency dependence of the complex viscosity (η*, Pas). Linear relationships between log η* and log with a slope of −1 correspond to a perfectly elasc solid, such as an ideal chemically cross‐linked network.8,214

Chapter 6

107 X1 3 de 10 PBT

5 10 106 102 pX1

5 10 101 ' [MPa] E 3.5 mol% 4 10 100 4 Complex viscosity [Pas] G' (open), G'' (close) [Pa] 10 1 mol% (A) (B) 103 10‐1 0.01 0.1 1 10 100 0 30 60 90 120 150 180 210 240 270 [rad/s] Temperature C] Figure 6.3. (A) Storage (open symbols) moduli, loss (filled symbols) moduli and complex viscosity (η*)

(half‐opened symbols) versus frequency at 250 °C for the cured pX1 and deX1. (B) Comparison of DMTA curves for X1 and X3.5 vitrimers obtained by processing‐induced (bold line) deprotection and reference samples (thin line) (heating rate = 3 °Cmin‐1, frequency = 1 Hz). For clarity reasons, the DMTA curves of C2.4 are presented in the Appendix 5.

The thermo‐mechanical properties of the studied vitrimers are shown in Figure 6.3B, and a non‐vanishing rubbery plateau for both pX1 and deX1 was observed above Tm and the materials exhibit moduli of about 0.24 MPa and 0.38 MPa at 250 °C, respectively. deX1 has a glass transition temperature (Tg) similar to neat PBT, i.e., 55 °C determined by the peak maximum from the loss modulus, while pX1 possesses a higher Tg than cured deX1, which is around 63 °C. This increase in Tg is due to the small difference in cross‐link density between pX1 and cured deX1. In general, the higher the cross‐linker content (BPO), the lower the modulus between glass transition and the melting region and the melting temperature of the material. This moderate modification of the thermomechanical properties is due to a slight change of the degree of crystallinity (Appendix 5).

6.2.3 Stress relaxation experiments

The dynamics of the network obtained by processing‐induced deprotection were further compared with reference samples using shear stress relaxation experiments. The stress relaxation curve presented in Figure 6.4A is a stress relaxation modulus (G(t) = σ(t)/γ0) plot versus time at 290 °C for pX3.5, a network achieved via the processing‐induced deprotection method, and it shows complete stress relaxation. This full stress relaxation indicates that the network is truly dynamic. As expected,91,187 the stress relaxation time can be tuned by temperature and the amount of cross‐linker. pX2.4 was used as the example to demonstrate the temperature‐dependent stress relaxation behavior as shown in Figure 6.4B. The stress relaxation time is significantly shifted towards shorter time‐scales on increasing temperature, which enhances the rearrangement reaction kinetics with a

In situ network formation in PBT vitrimers via processing‐induced deprotection chemistry decrease in relaxation time (τ) from 5061 s at 230 °C to only 325 s at 260 °C. The characteristic stress relaxation time is defined as the time it takes for the samples had relaxed to 1/e (37%) of the initial stress relaxation modulus, analogous to other studies on different vitrimers,1‐3,17,18,22‐27,35,36 Figure 6.4C shows the comparison of the stress relaxation behavior of the material obtained via processing‐induced deprotection method and its corresponding reference sample. For the materials with the same BPO content, the results indicate that the material obtained via the processing‐induced debenzalation has a slightly longer stress relaxation time than its corresponding reference sample, which is consistent with a higher crosslink density. For the materials obtained via the same debenzalation process, the higher the BPO content (i.e., the higher the crosslink density), the longer the stress relaxation time. 5 1.8x10 3x105 (A) (B) 1.5x105

5 1.2x10 2x105

9.0x104 260 C 4 5

Modulus [Pa] 6.0x10 1x10

Modulus [Pa] 250 C 240 C 3.0x104 230 C T 0.0 0 ‐2 ‐1 0 1 2 3 10 10 10 10 10 10 10‐1 100 101 102 103 Time [s] Time [s] 1.0 10 (C) (D)

0.8 8 deX3.5 X1 0.6 de 0 /G [s] t

X3.5  6 G p

0.4 ln 37% pX3.5

0.2 X3.5 4 pX1 de X2.4 250 C p 0.0 0 1 2 3 4 1.76 1.80 1.84 1.88 1.92 1.96 2.00 10 10 10 10 10 ‐1 Time [s] 1000/T [K ]

Figure 6.4. (A) Stress relaxation curve for pX3.5 catalyzed with 0.2 mol% Zn(II) catalyst obtained at 290

°C and (B) different temperatures for pX2.4. (C) Normalized stress relaxation curves for different BPO contents. Relaxation times are measured for a 63% relaxation. (D) Variation of the stress relaxation time for X1 and X3.5 versus inverse temperature obtained via different curing processes. For clarity reasons, the complete normalized stress relaxation curves of pX1, deX1, pX3.5 and deX3.5 are presented in Appendix 5.

Chapter 6

It was well documented in our previous work43, 45 that the stress relaxation behavior of a less developed network (here: pX1 and deX1) cannot be described by the Maxwell model with a single characteristic stress relaxation time. Only the stress relaxation behavior of materials (i.e., pX3.5 and cured deX3.5 ) with a well‐developed network (nearly ‐ independent G and small G)153 relax with a single exponential process and for these material, it is possible to derive the activation energy from the obtained stress relaxation time. Since the relaxation times are controlled by associative exchange reactions, the temperature dependence of the relaxation time can be described by the Arrhenius equation (eq 1).

 0 (/RT)Ea (1) ‐1 In this equation, τ0 is the pre‐exponential factor (s), Ea is the activation energy (Jmol ), R is the ideal gas constant (Jmol‐1K‐1) and T is the temperature (K). An excellent fit is observed in Figure 6.4D and the activation energy Ea is determined from the slope. Comparable activation energies were observed for the materials obtained via the TFA‐ promoted and processing‐induced debenzalation approaches. The values of Ea are 230 and

250 kJ/mol for pX3.5 and deX3.5, respectively. Therefore, we can conclude that the PBT vitrimers obtained via the (solvent‐free) processing‐induced debenzalation process have similar networks as those obtained via TFA‐promoted debenzalation approach; this in turn suggests that similar crosslinking chemistries are involved. 6.3 Conclusions

In summary, the new findings of in situ network formation by the processing‐induced deprotection approach in the presence of carboxylic acid end groups from PBT render a similar network with vitrimer characteristics as in the case of a TFA‐deprotected reference. The cross‐linking kinetics of the protected linear copolyester obtained directly after SSP was followed by oscillatory time sweep experiments at 250 °C and it is shown that the duration of pre‐gel period (low‐viscosity processing window) and cross‐link density of the final network can be tuned at will by playing with the processing temperature and BPO content. We proposed a two‐step mechanism for the in situ network formation via the processing‐ induced deprotection method: first the deprotection of the benzylidene acetal group to form free hydroxyl groups takes place and, subsequently, branching and ultimately network is formed via transesterification in the presence of Zn(II) catalyst. Our in situ network formation strategy shares the same aim as the recent work from Leibler and coworkers;85 they reported a rapid exchange reaction based on the metathesis of dioxaborolanes to enable the commodity vitrimers to be processed with fast production rates and current equipment during extrusion or injection molding.

In situ network formation in PBT vitrimers via processing‐induced deprotection chemistry

6.4 Experimental section

Materials.

Special grade poly(butylene terephthalate) pellets (Mn = 21.2 kg/mol, Mw = 46.6 kg/mol) against poly(methyl methacrylate) (PMMA) standards in HFiP were provided by SABIC (Bergen op Zoom, the Netherlands) and used as received. Glycerol (≥99.5%), zinc(II) acetylacetonate hydrate (Zn(acac)2) (powder), terephthaloyl chloride (≥99%), trimethylamine (≥ 99.5%), were all obtained from Sigma‐Aldrich. Hydrochloric acid, anhydrous dichloromethane, methanol, 1,1,1,3,3,3‐hexafluoroisopropanol (HFiP, 99%) and

MilliQ water (LC‐MS grade) were obtained from Biosolve. Deuterated chloroform (CDCl3, 99.8 atom% D) and deuterated trifluoroacetic acid (TFA‐d, 99 atom% D) were obtained from Cambridge Isotope Laboratories. All chemicals were used as received unless denoted otherwise. Synthesis of 5, 5‐Bis (hydroxymethyl)‐2‐phenyl‐1, 3‐dioxane This product was prepared according to a previous literature procedure.7 Pentaerythritol (18.0 g, 132 mmol) was dissolved in water (130 mL) at 60 °C. The solution was cooled to room temperature undisturbed. Into the stirring solution, concentrated HCl (0.7 mL) was added followed by benzaldehyde (3.0 mL, 30 mmol). After the precipitation formed, more benzaldehyde (11.0 mL, 108 mmol) was added dropwise and the reaction mixture was allowed to stir at room temperature for overnight. The solid was heated with boiling slightly alkali water (Na2CO3 solution) for 10 min and filtered quickly through filter paper followed by washing with hot slightly alkaline water, and Et2O. Recrystallization from toluene gave 5,5‐bis‐(hydroxymethyl)‐2‐phenyl‐1,3‐dioxane as a colorless solid (52 %). 1H

NMR (400 MHz, DMSO‐d6): 3.22 (d, HCOH, 2H, axial exocyclicCH2), 3.70 (d, HCOH, 2H, equatorial exocyclic CH2), 3.80 (d, 2H, C4,6 axial H), 3.90 (d, 2H, C4,6 equatorial H), 4.55 (t, HOCH, 1H, axial OH), 4.62 (t, HOCH, 1H, equatorial OH), 5.40 (s, 1H, benzylic H), 7.40‐7.55 13 (m, 5H, aryl H); C NMR (400 MHz, DMSO‐d6):139, 129, 128, 126 (aromatic C); 100 (C2), 69

(C4,6), 61, 59 (exocyclic CH2); Synthesis of 2‐phenyl‐5, 5‐Bis (phenylcarboxymethyl)‐1, 3‐dioxane This product was prepared according to a previous literature procedure.7 1H NMR (400

MHz, CDCl3): 4.1 (d, 2H, ), 3.70 (d, HCOH, 2H, equatorial exocyclic CH2), 3.80 (d, 2H, C4,6 axial H), 3.90 (d, 2H, C4,6 equatorial H), 4.55 (t, HOCH, 1H, axial OH), 4.62 (t, HOCH, 1H, equatorial OH), 5.40 (s, 1H, benzylic H), 7.40‐7.55 (m, 5H, aryl H); 13C NMR (400 MHz,

CDCl3):139, 129, 128, 126 (aromatic C); 100 (C2), 69 (C4,6), 61, 59 (exocyclic CH2). Synthesis of 2, 2‐bis(hydroxymethyl)propane‐1,3‐diyl dibenzoate A solution of 2.8 g of monobenzalpentaerythritol in 10 ml of pyridine was treated with 3.75 g of benzoyl chloride, which caused an exothermic reaction. After twenty‐four hours,

Chapter 6

100 ml of 10% sulfuric acid was added with cooling by ice‐water bath, and the crude product filtered, washed with water and sodium carbonate solution and recrystallized from isopropyl alcohol (25 ml), long needles of m.p. 120.5 °C; yield 4.6 g, (88%). The solution preparation of a physical PBT/BPO mixture For preparing the physical mixture with 1.2 mol% BPO, the dried PBT 195 powder (78.81 g, 357.8 mmole), mono‐benzal pentaerythritol (BPO) (0.80 g, 4.4 mmol), and zinc acetylacetonate (0.057 g, 0.7 mmol ) were dissolved in 250 mL of 1, 1, 1, 3, 3, 3‐ hexafluoroisopropanol (HFIP). The mixture was refluxed until a clear solution was obtained and subsequently the solvent was removed by distillation under atmospheric and reduced pressure (10‐3 mbar). The PBT/BPO mixture was dried at 30 °C under reduced pressure for

24 h, cooled using liquid N2 and subsequently grounded using a mill (IKA type 10A). Finally, this mixture was dried for an additional 24 h at 30 °C under reduced pressure before being used in SSP. Solid‐state (co)polymerization PBT/BPO‐based copolyesters were synthesized by a two‐step solid‐state polymerization method. Typically, in the first step, 5 g of PBT/BPO mixture containing 0.2 mol% Zn(acac)2 was placed in a closed vial, and pressurized with argon (p < 3 bar) to avoid the evaporation of BPO. After 24 h at 160 °C, the mixture was transferred to the SSP reactor, which was a glass tube (inner diameter = 2.4 cm) with a sintered glass plate at the bottom. A heat exchange glass coil (inner diameter = 0.5 mm) surrounded the reactor and entered the inner glass tube at the bottom just below the glass plate. The nitrogen gas was heated by passing through this coil prior to entering the reactor and its flow was controlled by a flow meter. The powder bed was fixed by addition of glass pearls (diameter = 2 mm) on top of the powder, and the reactor was purged with a nitrogen flow of 0.5 Lmin‐1 for 30 min. After flushing, the reactor was placed in a heated salt bath (T = 180 °C). When the temperature inside the reactor reached 180 °C, the measurement of the reaction time (tssp) was initiated, and the composition and molecular weight were followed. After the reaction, the product was cooled down to room temperature under a continuous nitrogen flow, discharged from the reactor, and the obtained polymer dried under vacuum at 120 °C for 6 h. Debenzalation of the protected copolymers The polymers were added to trifluoroacetic acid (6 mL/g of polymer) at room temperature and stirred until all of the solid material was dissolved. The solution was stirred for an additional 2‐3 h. The polymer solution was then precipitated in methanol. The solid was removed by filtration, washed with methanol (300 mL), and vacuum‐dried at room temperature to afford 2,2‐bis(hydroxymethyl)‐propane‐1,3‐diyl containing copolymers. Characterization methods

In situ network formation in PBT vitrimers via processing‐induced deprotection chemistry

Proton and Carbon nuclear magnetic resonance spectroscopy. 1H‐ and 13C‐NMR spectroscopy were performed on a 400 MHz Bruker Avance III spectrometer at 25 °C. For the 1H‐NMR experiments, the spectral width was 6402 Hz, the delay time was 5 s and the number of scans was 64. For the 13C‐NMR experiments, the spectral width was 24,154 Hz, the delay time was 2 s and the number of scans was between 2000 and 5000. Samples were prepared by dissolving 15‐50 mg of the crude polyester in 0.8 mL of an 80:20 vol % CDCl3: d‐TFA mixture. Chemical shifts are reported in ppm relative to the residual solvent peak of

CDCl3 (δ = 77.0 ppm). Size Exclusion chromatography (SEC). Molecular weight distributions, the number average molecular weight (Mn) and polydispersity index (Đ) of the copolyesters were measured on a system equipped with a Waters 1515 isocratic HPLC pump, a Waters 2414 refractive index detector (40 °C), a Waters 2707 autosampler, and a PSS PFG guard column followed by a 2PFG‐linear‐ XL (7 μm, 8 × 300 mm) columns in series at 40 °C. HFIP with potassium trifluoroacetate (3 gL−1) was used as eluent at a flow rate of 0.8 mL min−1. The molecular weights were determined relative to PMMA standards (Polymer Laboratories, Mp 6 = 580 Da up to Mp = 7.1 × 10 Da). Differential scanning calorimetry (DSC). Thermal properties were measured using a DSC Q1000 from TA Instruments. The measurements were carried out from ‐50 to 250 °C with heating and cooling rates of 10 °Cmin‐1 under a nitrogen flow of 50 mlmin‐1. Compression molding. The materials were compression molded at 250 °C and 100 bar for 25 minutes in a Collin Press 300G and subsequently cooled with water. Dynamic mechanical thermal analysis (DMTA). Compression‐molded samples (ca. 10.0 (Length) × 5.0 (Width) × 1.0 (Thickness) mm) were measured on a DMA Q800 (TA Instruments) with a film tension setup. A temperature sweep from ‐50 to 270 °C was performed with a heating rate of 3 °Cmin‐1 at a frequency of 1 Hz. A preload force of 0.01 N, an amplitude of 10 µm and a force track of 125% were used. The storage modulus and loss modulus were recorded as a function of temperature. The glass transition temperature was calculated from the peak maximum in the loss modulus. Rheometry. Dynamic shear measurements were performed on a stress‐controlled AR‐ G2 Rheometer (TA Instruments) by using a 25‐mm parallel plate geometry and disk‐shaped specimens (25 mm diameter; 1 mm thick). Frequency sweeps from 500 to 0.01 rad/s were performed at a temperature range between 230 and 270 °C with a strain of 1%, which is in the linear viscoelastic regime. Stress relaxation experiments were performed at a temperature range between 230 and 270 °C with a strain of 1% and the relaxation modulus was monitored as a function of time. A constant normal force of 10‐20 N was applied to ensure a good contact with the plates.

100

Chapter 7

Influence of the network nature on the crystallization/melting behavior of PBT

ABSTRACT: The influence of the network properties on the melting and crystallization behavior of poly(butylene terephthalate) was determined for a wide range of cooling/heating rates and isothermal temperatures by fast scanning calorimetry. The critical cooling rate to suppress crystallization is dependent on the network properties, i.e., talc concentration, molecular entanglements, PBT/PC blend, and dynamic vitrimer network density. For isothermal crystallization, the annealing‐temperature dependence of crystallization peak‐time for PBT shows a bimodal curve with two minima. The heterogeneous nucleation density can be tuned by the talc concentration, which worked as heterogeneous nuclei. The crystallization via homogeneous nucleation is independent on talc concentration, but the homogeneous nucleation density is influenced by both the physical entanglements and dynamic cross‐links. For an amorphous material, only the PBT vitrimer with a rubbery plateau modulus above melting point around 2 MPa can suppress the cold crystallization and subsequent melting process at a heating rate as high as 5000 °Cs‐1. Furthermore, the reorganization during the melting of recrystallized crystals is not influenced by the network nature at the heating rate ranging between 10 or 5000 °Cs‐1.

Influence of the network nature on the crystallization/melting behavior of PBT

7.1 Introduction

Poly(butylene terephthalate) (PBT) vitrimers are a new class of semi‐crystalline materials, which are chemically cross‐linked thermosets with reprocessability, repairability and weldability above melting temperature.90,91 These semi‐crystalline vitrimers contain dynamic cross‐links in the amorphous phase of PBT and have a similar crystal morphology as PBT.91 The crystalline morphology and degree of crystallinity of semi‐crystalline polymers are strongly influenced by the conditions applied during processing and are of major importance for the final physical and mechanical properties.117 Poor control of crystallization during molding may thus lead to widely varying part‐to‐part properties and dimensional variations. Therefore, it is important to understand how the dynamic cross‐ links of the semi‐crystalline PBT vitrimer may influence the melting and crystallization behavior of PBT. The crystal morphology, melting behavior and isothermal crystallization kinetics of linear PBT have been extensively because of its wider fields of application.15,124,127,130,144,146,147,195,215–220 PBT is very suitable for injection molding application due to its relatively fast crystallization rate.1,2 In real industrial processing conditions during injection molding, cooling rates in the order of 100 °Cs‐1 are common. In recent years, this fast cooling rate can be achieved via the application of fast‐scanning chip calorimetry (FSC).221 FSC permits isothermal measurement of the rate of melt crystallization in a wide range of temperatures from below the glass transition temperature (Tg) to close 0 to the equilibrium melting temperature (Tm ) due to its high cooling capacity and short time constant of the instrument as compared to conventional differential scanning calorimetry (DSC).222–225 The crystallization kinetics of PBT in the whole temperature range between the glass transition and melting temperature has been studied by FSC. The crystallization half‐ time as a function of isothermal crystallization temperature shows a bimodal distribution with two minima around 70 °C and 135 °C,126,205,226–228 which is similar to many other polymers such as polyamide‐6 (PA‐6),229,230 polyamide‐11 (PA‐11)134 and isotactic polypropylene (iPP).231,232 In general, this dual crystallization behavior is related to the formation of different, thermodynamically stable or unstable crystal structures, including the formation of mesophases at high supercoolings. The study from by Androsch et al.227 showed, however, that for PBT this observed dual crystallization behavior was not related to a change in crystal structure, but to a qualitative change of the nucleation mechanism from heterogeneous nucleation at low supercooling with a relatively low nucleation density to homogeneous nucleation at high supercoolings with a very high number of nuclei. Interestingly, heterogeneous nucleating agents, such as multi‐walled carbon nanotubes (MWCNT)226 and talc,228 only influence the crystallization kinetics in the heterogeneous

102

Chapter 7 nucleation‐dominated temperature range, while the crystallization via homogeneous nucleation is independent on the heterogeneous nuclei concentration in the low crystallization temperature range. Molecular dynamics simulations for the early stage of homogeneous nucleation suggest that the chain segments forming the nuclei showed bundle‐like aggregates with chain‐folded conformations.233,234 This suggests that the dynamics of the whole polymer chain will affect the kinetics of forming the homogeneous nuclei. The non‐isothermal crystallization kinetics of PBT is also well studied by FSC experiment. At slow cooling rates, only the high crystallization‐temperature event was found.235,236 When the cooling rate exceeded 40 Cs‐1, a low crystallization‐temperature event was observed at temperatures lower than 80 °C, depending on the cooling rate.226,227,237 Both crystallization events coexisted until the cooling rate exceeds 200 Cs‐1, then an amorphous sample is formed.226,227,237 PBT is one of the semi‐crystalline polymers that shows the well‐known double melting behavior,118–123 and this multiple melting behavior was studied for a wide range of heating rates using FSC by Furushima et al.126 and Jariyavidyanont et al.238 The observed double melting peaks are assigned to the melting of imperfect crystals (low‐temperature peak) and recrystallized and/or reorganized crystals (high‐temperature peak). Furushima et al.126 mentioned that a very high heating rate is needed to suppress the reorganization, whether the crystals are formed via melt crystallization (heterogeneous nucleation for relatively slowly cooled samples or homogeneous nucleation via intermediate cooling rates), and this melting of recrystallized crystals can be avoided at the heating rate of 100,000 K.s‐1.126,226 When a vitrified polymer glass is subjected to a subsequent heating/annealing close to or just above Tg, it will involve in some cases cold crystallization because of the fast formation of homogeneous crystal nuclei at temperatures close to Tg. It was found for PBT that the critical heating rate to suppress cold‐crystallization is much higher than the critical cooling rate to vitrify the PBT (300‐500 °Cs‐1) on cooling,126,226–228,237 being about 10,000 °Cs‐1.126,226 In this chapter, we study how the network properties may influence the melting/crystallization of PBT and PBT vitrimers. The main research questions that will be addressed in this chapter are : 1) Do the network properties influence the critical cooling rate for quenching a PBT to below glass transition without crystallization; 2) Can we tune the homogeneous and heterogeneous nucleation density of PBT via playing with the type of network; 3) How does the structure of the glass as obtained by quenching from the melt and subsequent annealing close to or just above Tg influence the crystallization kinetics upon heating and what is the effect of the network dynamics on the critical heating rate to suppress cold crystallization; 4) Will the network be able to influence the critical heating rate to suppress reorganization.

103

Influence of the network nature on the crystallization/melting behavior of PBT

Here, we will investigate the influence of network properties by using heterogeneous nucleating agents, i.e., talc as an inorganic filler, a (partially) miscible polymer to adapt the physical network density (PBT/polycarbonate (PC) blend) and a dynamic vitrimer network (PBT vitrimers) on the aforementioned research questions. 7.2 Results and discussion

The structure of this part is as following: firstly, the thermal properties of the materials studied in this chapter will be presented, as well as the vitrimer network characteristics of PBT vitrimers. Secondly, non‐isothermal crystallization kinetic experiment will be employed to study the influence of the network properties on the critical cooling rate. The isothermal crystallization kinetics will be followed by a qualitative analysis of the nucleation density as a function of Tc. Finally, the influence of different network properties on the critical heating rate to suppress cold crystallization and melt‐recrystallization process of PBT will be discussed.

Table 7.1. Characterization of the materials used in this study.

a a Talc PC Glycerol Tm Tc Degree of Entry (wt %) (wt %) (mol %) (°C) (°C) crystallinity (%)a PBT 0 0 0 223 195 41 PBT/talc(0.005)b 0.005 223 201 41 PBT/talc(0.01) b 0.01 223 202 42 PBT/talc(0.025) b 0.025 223 202 41 PBT/talc(0.5) 0.5 224 204 44 PBT/PC(5) 5 222 194 38 C1(GLY)c 1.2 222 196 40 C8(GLY)c 7.8 217 188 37 a a Melting‐peak temperature (Tm) and Crystallization‐peak temperature (Tc) were determined by second heating and first cooling run with a temperature ramp 10 °Cmin‐1, respectively. aDegree of crystallinity was determined by dividing the crystallization enthalpy (ΔHc) of the second heating run 0 (obtained via DSC measurements) by the bulk crystallization enthalpy (ΔH c) for 100% crystalline PBT. 0 ‐1 147 b For ΔH c, a value of 142 Jg was used . PBT/Talc(0.005) and PBT/PC(5), stands for 0.005 wt% talc‐ nucleated PBT and PBT/PC blend containing 5 wt% of PC, respectively. cC1(GLY) and C8(GLY) stand for

PBT/glycerol‐based vitrimers containing 1 and 8 mol% glycerol catalyzed with 0.2 mol% Zn(acac)2, respectively. Thermal properties of the PBT vitrimers are taken from Table 5.1, Chapter 5.

The talc‐nucleated PBT with different weight percentage of talc loadings and PBT/PC (95/5) blend with 5 weight percentage of PC were prepared by melt compounding. The PBT 104

Chapter 7 vitrimers (C1(GLY) and C8(GLY))catalyzed by 0.2 mol% Zn2+ are prepared by solid‐state (co)polymerization and the preparation method was previously described in Chapter 5. The thermal properties of the PBT/talc composites, PBT/PC blend and PBT vitrimers are studied by DSC and the thermograms are presented in Figure 7.1. An overview of the typical thermal properties, such as Tc, Tm, and degree of crystallinity are shown in Table 7.1.

7.2.1 Thermal properties of talc‐nucleated PBT, PBT/PC blend and PBT vitrimers

The peak melting temperature (Tm), peak crystallization temperature (Tc) and degree of crystallinity (c) of the used PBT pellets, determined by DSC from the second heating run, are shown in Figure 7.1A.

3 (A) (B) PBT 2 PBT/PC(5) 2 Increasing talc loading 1 C8(GLY) C1(GLY) 1

0 0

‐1

‐1 Heat flow, exo up [w/g]

Heat flow, exo up [w/g] Tm,low T ‐2 m,high ‐2 180 190 200 210 220 230 160 170 180 190 200 210 220 230 Temperature [C] Temperature [C] 0.0 (C)

‐0.1

‐0.2 T PBT/PC(5) g ‐0.3 Heat flow, exo up [w/g] ‐0.4

‐30 0 30 60 90 120 150 Temperature [C]

Figure 7.1. Standard DSC curves recorded on heating and cooling at 10 °Cmin‐1: PBT and (A) talc‐ nucleated PBT, (B) PBT/PC (5) blend and PBT vitrimers. (C)The heating run for PBT/PC (5) blend in the temperature range ‐50 to 160 °C.

Upon heating, the well‐known double melting behavior of PBT is clearly visible with two well‐separated melting peaks, and these two melting peaks shift to higher temperature with increasing talc content. This shift of the melting temperature for the talc‐nucleated PBT is caused by an increase of the lamellar thickness due to the nucleating effect of the talc particles leading to a higher Tc and concomitant lower undercooling, while the double‐ 105

Influence of the network nature on the crystallization/melting behavior of PBT melting behavior is related to the reorganization process that occurs during the DSC heating run, resulting in melting and recrystallization of less perfect crystals into thicker and more perfect crystals.15,124–126

Talc is known to be a heterogeneous nucleating agent. This leads to a higher Tc. Then the undercooling is lower. The growth rate is also influenced. The mobility is higher so that the overall crystallization kinetics are faster (and leading to a higher degree of crystallinity) as both nucleation and growth rates are higher. As is shown in Figure 7.1A, for example, the peak crystallization temperature, Tc, increases by 6 °C with only 0.005 wt% of talc. An additional increase in the concentration of the talc generates only a modest additional increase of the crystallization temperature. The DSC thermograms of PBT with 5 wt% PC, PBT/PC(5), are shown in Figures 7.1B. In the PBT‐PC case, the competition between liquid‐ liquid phase separation, interfacial transesterification reactions, and crystallization of PBT complicate the phase behaviour of this type of blends.239 Delimoy et al.240 proposed a phase diagram of PBT/PC blends, for which they concluded that above 90 wt% PBT, PBT/PC blends are monophasic homogeneous systems in the melt, although it has to mentioned that this was evidenced by TEM at room temperature after cooling down from the melt. As shown in Figure 7.1C, only one Tg is observed for PBT/PC(5) blend in the temperature range between ‐50 and 160 °C, and this particular amount of PC has a limited effect on the melting

(Tm) and crystallization temperature (Tc) for PBT. In the following discussion, a single‐phase (fully miscible) system is assumed. The thermal properties of PBT vitrimers are presented in Figure 7.1B, C1 shows a slightly higher Tc than neat PBT, and the Tc , Tm and degree of crystallinity decrease with increasing glycerol content, i.e., the Tc decreases from 196 to 188 °C when the glycerol content increases from  1 mol% (C1(GLY)) to  8 mol% (C8(GLY)). In general, PBT vitrimers show a slightly broader crystallization peak than the rest of the materials studied in this chapter, which is due to the broad distribution of the lamellar thickness in the PBT vitrimers.241 Furthermore, the double melting behavior is clearly visible for PBT vitrimers at a heating rate 10 °Cmin‐1. The vitrimer characteristics of C1(GLY) and C8(GLY) were well documented in Figures 5.3 and 5.4. The rubber plateau modulus is about 0.7 MPa and 2MPa at 250 °C for C1(GLY) and C8(GLY), respectively.

7.2.2 Why fast scanning calorimetry

As we know, for a conventional DSC, its maximum cooling rate is about 40 °Cmin‐1 (0.67 °Cs‐1); however, the cooling rates during polymer melt processing can be relatively high. For instance, with injection molding, the local cooling rate on the material surface may reach 1000 °Cs‐1,234 and it is not feasible to directly measure such fast crystallization processes using conventional DSC. Therefore, the non‐isothermal and isothermal crystallization kinetics analysis were done by employing a power‐compensation Mettler‐

106

Chapter 7

Toledo Flash DSC 1 (Figure 7.2A) connected to a Huber intracooler TC100. The typical cooling rates of flash DSC 1 are between 0.1 to 4000 °Cs‐1, and typical heating rates are between 0.5 to 40,000 °Cs‐1. Here, in our experiments, the sample was placed directly on the sensor and pre‐melted at 270 °C for 0.1 s to establish a good thermal contact between the sample and the sensor and a typical PBT sample on the calorimeter membrane is shown in Figure 7.2B. In general, the sample mass was estimated by dividing the measured heat capacity at the glass‐transition temperature in units of JC‐1 on cooling the melt faster than the critical cooling rate to ensure the absence of any crystallization with the literature mass‐ specific heat‐capacity of 0.40 JgC‐1 for amorphous PBT.147

(A) (B)

Figure 7.2. (A) A Mettler Toledo Flash DSC 1 and the chip sensor. (B) POM micrographs of a typical PBT sample on the calorimeter membrane.

7.2.3 Critical cooling rate to suppress crystallization during cooling

A crystallization exotherm peak is always visible during a non‐isothermal crystallization experiment for neat PBT measured by conventional DSC. With increasing cooling rate, the crystallization peak became wider and shifted towards lower temperature.242,243 In order to determine the critical cooling rate to suppress crystallization during cooling, samples were cooled from 270 to ‐90 °C at different cooling rates range from 10 to 2000 °Cs‐1 with fast scanning calorimetry, as schematically shown in Figure 7.3. A plot of the FSC curves recorded on cooling PBT and 0.5 wt% talc‐nucleated PBT at different cooling rates is shown in Figure 7.4. Relatively slow cooling at a rate of 10 °Cs‐1 (bottom curve) for neat PBT reveals a broad high‐temperature crystallization peak at a temperature Tc, high (red dotted line) of 150 °C, and when the cooling rate is around 80 °Cs‐1, the low‐temperature crystallization peak (Tc,low) appears, which is indicated by the blue dotted line. These observed two crystallization events confirmed the previous work of Schick and coworkers.227

107

Influence of the network nature on the crystallization/melting behavior of PBT

Figure 7.3. Temperature‐time profile for the non‐isothermal crystallization kinetic experiments.

‐1 Neat PBT Cooling rate in K s‐1 0.5 wt% talc‐ Cooling rate in K s nucleated PBT 2000 2000

1000 1000 500

300 500

100 300

100 Heat flow [mW, exo up] Heat flow [mW, exo up] 50 50

10 10 0 30 60 90 120 150 180 210 0 30 60 90 120 150 180 210

Temperature [C] Temperature [C] (A) (B) Figure 7.4. FSC curves, heat‐flow rate as a function of temperature, recorded on cooling (A) PBT and (B) talc‐nucleated PBT at different rates between 10 (bottom) and 2000 Ks‐1 (top). The curves were shifted vertically for clarity.

When the cooling rate exceeds 200 °Cs‐1, a fully glassy PBT is obtained and it completely vitrifies at a Tg of around 30 °C. For the talc‐nucleated PBT, the low‐temperature crystallization event is undetectable due to the heterogeneous nucleation effect and the used cooling rate of 2000 °Cs‐1 is not sufficient to completely vitrify the 0.5 wt% talc‐ nucleated PBT. For talc contents  0.025 wt%, an amorphous glass is obtained when cooling rate  2000 °Cs‐1. The critical cooling rate for the PBT/PC (95/5) blend and the PBT vitrimers are lower than neat PBT, which vitrifies with a cooling rate of around 300 °Cs‐1. The crystallization curves for those samples did not shown any apparent crystallization event, 108

Chapter 7 thus it is not possible to further analyze the data. The non‐isothermal crystallization FSC curves of the PBT vitrimers and PBT/PC (5) blend are shown in Appendix 6.

210 Flash DSC 180 Conventional DSC 150 C]  [ c

T 120 PBT

90 PBT/talc(0.025) Tc, high PBT/talc(0.5) 60 PBT Tc, low 10‐3 10‐2 10‐1 100 101 102 103 ‐1 Cooling rate [Cs ] Figure 7.5. Crystallization peak temperature of PBT and 0.025 (PBT/talc(0.025)) and 0.5% (PBT/talc(0.5)) talc‐nucleated PBT as a function of cooling rate. Data obtained on cooling rates lower than 10.0 °Cs–1 were collected by conventional DSC (TA Q1000), otherwise by FSC.

Figure 7.5 shows the peak crystallization temperatures of PBT and talc‐nucleated PBT as a function of the cooling rate, with the open and closed symbols referring to the high‐ and low‐peak crystallization temperature. Due to the limited resolution of the peak crystallization temperature for PBT‐based blends (see Appendix 6), only the neat PBT and talc‐nucleated PBT are presented. In general, the Tc decreases with increasing the cooling rate because of the kinetics of both crystal nucleation and growth. The talc‐nucleated PBT shows a higher Tc than neat PBT because of a higher heterogeneous nucleation density.

Furthermore, the Tc increases with increasing the talc content and the low crystallization peak time cannot be detected in talc‐nucleated PBT.

7.2.4 Isothermal crystallization kinetics

The used time‐temperature profile of the performed isothermal crystallization experiments is shown in Figure 7.6. Samples were melted by heating at a rate of 1000 °Cs‐ 1 to 270 °C, held at this temperature for 1 s, and then cooled at a rate of 5000 °Cs‐1 to the crystallization temperature with the chosen cooling rate to ensure that no crystallization occurs before the isothermal crystallization segment was reached. After the isothermal crystallization step, the sample was first cooled to ‐90 °C to ensure a complete crystallization process, then re‐melted, before continuing the analysis of at an isothermal crystallization step at a 5 or 2.5 °C lower temperature.

109

Influence of the network nature on the crystallization/melting behavior of PBT

Figure 7.6. Temperature‐time profile for the isothermal crystallization kinetic experiments.

Figure 7.7. Set of FSC curves, heat‐flow rate as a function of time, obtained during isothermal crystallization of 0.5 wt% talc‐nucleated PBT at temperatures between 180 °C (front curve) and 60 °C (back curve). Exothermic heat flow is directed upward.

Figure 7.7 shows an example of a series of FSC curves, heat‐flow rate as a function of time, obtained during isothermal crystallization of talc‐nucleated PBT at temperatures between 180 (front curve) and 60 °C (back curve). The temperature interval between two neighboring curves in the temperature range between 170 and 130 °C was 5 °C, while it was lowered to 2.5 °C in the temperature range between 130 and 50 °C. The peak in the various curves of Figure 7.7 is due to crystallization‐related exothermic heat flow, corresponding to the peak‐time of crystallization, being a measure of the gross crystallization rate. The shorter the peak‐time of crystallization the faster is the crystallization process. The curves of Figure 7.7 reveal two different minimum peak‐times of about 0.2 and 0.08 s at high and low crystallization temperatures, respectively, associated to two qualitatively different crystallization events as was also shown in Figure 7.5 for the non‐isothermal crystallization experiments.

110

Chapter 7

PBT/PC 1 PBT 10 blend us s eo ou en ne og ge om tero H He 0 wt% 100

Talc 

Crystallization peak‐time [s] ‐1 10 0.5 wt% 60 80 100 120 140 160 180 Temperature [C] Figure 7.8. Peak‐time of isothermal crystallization of PBT, talc‐nucleated PBT, PBT/PC (95/5) blend as a function of the crystallization temperature.

In Figure 7.8 the quantitative data about the temperature‐dependence of the peak‐time of crystallization of PBT, talc‐nucleated PBT and PBT‐based blends as a function of the crystallization temperature are summarized, as determined from the FSC curves in Figure 7.7. It is well known that nucleation can be either homogeneous or heterogeneous. Homogeneous nuclei are formed due to thermodynamic driving forces by the polymer chains themselves, whereas heterogeneous nuclei are often supported on interfaces, usually on intentionally added nucleating agents. The observed bimodal temperature‐ dependence of the crystallization rate for neat PBT is associated to two different nucleation densities as reported by Wurm et al.26 Two different minimum peak‐times of about 0.2 and 0.08 s at high and low crystallization temperature for neat PBT is clearly shown in Figure 7.9, respectively. The overall crystallization speeds up significantly at high temperatures for PBT in the presence of talc. At 170 °C, crystallization peak‐time decreases from 7.4 for the neat PBT to 3.2 s for the 0.005 wt% talc sample. Addition of 0.01, 0.025 and 0.5 wt% talc reduces the heterogeneous crystallization time further to 1.8, 1.2 and 1 s, respectively. The latter means a 7‐fold crystallization rate increase. However, as is apparent from the figure, talc only influences the heterogeneous nucleation density. The crystallization peak‐time at low temperatures for PBT increases from 0.2 to 1 s for PBT/PC(5) blend. A possible explanation is that the physical network formed from the PBT/PC(5) blend restricts the chain mobility of the PBT in the homogeneous nucleation region.

111

Influence of the network nature on the crystallization/melting behavior of PBT

101

C1(GLY) PBT C8(GLY) 100 us eo en og ter He us eo gen Crystallization peak‐time [s] mo Ho 10‐1 60 80 100 120 140 160 180 Temperature [C] Figure 7.9. Peak‐time of isothermal crystallization of PBT and PBT vitrimers as a function of the crystallization temperature.

The influence of vitrimer network on the isothermal crystallization kinetics are shown in Figure 7.9. PBT vitrimers exhibit a similar bimodal distribution as neat PBT, and the transition temperatures where it changes from heterogeneous to homogeneous nucleation (as indicated with the arrows in Figure 7.10) is approximately 10 and 20 °C higher than neat PBT. Both the heterogeneous and homogeneous nucleation densities decrease with the introduction of vitrimer network. The crystallization peak‐time at 75 °C is 2 and 4.5 times higher than neat PBT for C1(GLY) and C8(GLY), respectively. Furthermore, the homogeneous nucleation temperature range is expanded for PBT vitrimers, i.e., the temperature window is below 140 and 110 °C for C8(GLY) and neat PBT, respectively. Therefore, the vitrimer network has a higher impact on the homogeneous nucleation density than on the heterogeneous nucleation density. Interestingly, the PBT, C1(GLY) and C8(GLY) show an overlap in crystallization peak time around 110 °C. In contrast, the crystallization peak time of the PBT/PC(5) blend is constantly higher than neat PBT in the temperature range between 50 and 110 °C. This overlap crystallization peak time for PBT vitrimers is close to neat PBT, which is 1.5 and 1.1 s, respectively, while for the PBT/PC(5) blend is around 5.8 s. Furthermore, for C8(GLY), it shows another dramatic increase of homogeneous nucleation density when the isothermal crystallization temperature is below 80 °C. In summary, the effect of a vitrimer network on the isothermal crystallization kinetics of PBT has a similar behavior on decreasing the homogeneous nucleation density as PBT/PC blend, which possesses a physical network, but the dynamic nature of the vitrimer network may complicate the crystallization kinetics of PBT.

112

Chapter 7

7.2.5 Critical heating rate to suppress cold crystallization

Cooling at 4000 Cs‐1 Heating rate in Cs‐1 Cooling at 4000 Cs‐1 Heating rate in Cs‐1

10 10 20 50 20 100 200 50 500 100 1000 200 500 2000 1000

5000 2000

(A) Neat PBT (B) PBT/PC(5) blend 5000 0 30 60 90 120150180210240270 0 30 60 90 120 150 180 210 240 270

Temperature [C] Temperature [C] ‐1 ‐1 Cooling at 4000 Cs‐1 Heating rate in Cs‐1 Cooling at 4000 Cs Heating rate in Cs 10 10 20 20 50 50 100 100 200 200 500 500 1000

1000 2000

2000

Temperature [C] Temperature [C] Figure 7.10. FSC curves, heat‐flow rate as a function of temperature, recorded on heating samples at different rates between 10 (top) and 5000 °Cs‐1 (bottom). (A) PBT, (B) PBT/PC(5) blend, (C) C1(GLY) and (D) C8(GLY). The samples were previously cooled at 1 °Cs‐1 from the melt. The curves shifted vertically for clarity. 113

Influence of the network nature on the crystallization/melting behavior of PBT

One of questions posed in the introduction of the chapter was how the structure of the glass obtained by quenching the PBT from the melt and subsequent annealing close to or just above Tg involving in some cases cold crystallization influence the crystallization kinetics and how this cold crystallization influences the subsequent melting and recrystallization kinetics. The FSC curves for the amorphous samples produced by quenching from the melt at 4000 °Cs‐1 are presented in Figure 7.10, the glass transition of PBT and PBT/PC blend occurs between 45 and 60 °C followed by an exothermic peak due to cold crystallization, while this glass transition shifts to higher temperatures and becomes broad for PBT vitrimers. This broad glass transition behavior of PBT vitrimers is probably due to the non‐ random distribution of the dynamic cross‐links in the amorphous phase. The cold crystallization is not suppressed for the PBT/PC blend and the C1(GLY), samples at a heating rate as fast as 5000 °Cs‐1, while the cold crystallization and subsequent melting peaks are disappeared at the same heating rate for the C8(GLY) sample.

(A) C8(GLY) 180 (B) 240 C8(GLY) C1(GLY) 165 230 150 C]

C1(GLY) 

[ PBT/PC(5) 220 135

C] 120 blend 

[ 210 m 105 T PBT 200 PBT 90 c, cold crystallization T 75 190 60 PBT/PC(5) blend 180 45 101 102 103 101 102 103 Cooling rate [Cs‐1] Cooling rate [Cs‐1] Figure 7.11. (A) Melting‐peak and (B) cold‐crystallization‐peak temperature of PBT (), PBT/PC blend (), V1 () and V8 () against heating rate after previously cooled at 4000 °Cs‐1 from the melt.

The melting‐peak (Tm) and cold‐crystallization‐peak (Tcc) temperature as a function of heating rate are shown in Figure 7.11. For PBT and PBT/PC blend, at higher heating rates, the cold crystallization peak is shifted to higher temperatures and the melting peak to lower temperatures. The Tm of PBT is constantly higher than for the PBT/PC blend in the heating rates between 10 to 5000 °Cs‐1. For PBT, the cold crystallization events are clearly present at the entire studied heating rate range, while for PBT/PC(5) blend, when the heating rate reaches 5000 °Cs‐1, the cold crystallization peak areas become significantly smaller and invisible, but a subsequent melting peak is still visible. Interestingly, for the PBT vitrimers, the cold crystallization peak is shifted to higher temperatures, which is similar as PBT and PBT/PC(5) blend; however, the melting peak firstly decreases then increases again at a heating rate above 500 °Cs‐1 and 50 °Cs‐1, for C1(GLY) and C8(GLY), respectively.

114

Chapter 7

Furthermore, the Tm of PBT vitrimers is constantly higher than PBT and PBT/PC blend in the ‐1 heating rates between 10 to 5000 °Cs and the Tm of the PBT vitrimer increases with increasing network density. In summary, it is possible to suppress cold crystallization of PBT at a heating rate  5000 °Cs‐1 via playing with the vitrimer network density.

7.2.6 Critical heating rate to suppress reorganization

Many semi‐crystalline polymers exhibit reorganization effects as characterized with two separated melting peaks during the conventional DSC measurement upon heating.118–123 Therefore, the curve does not show only the melting of the crystallites originally present in the sample. What is the critical heating rate to suppress reorganization? It is reported by Furushima et al.126 for neat PBT that a heating rate as high as 80,000 °Cs‐1 is not sufficient to suppress the reorganization as evidenced by the appearance of a melting peak. Here, we would like to understand the role of a vitrimer network during reorganization. If this reorganization process is just chain segment mobility within the crystalline lamellae, it should not be influenced by the vitrimer network as the cross‐linking points are exclusively located in the amorphous phase, but after some thickening/reorganization this chain mobility will be influenced, since that the chain segments are restricted by the network (not entanglements but real covalent cross‐linking points that on the time scale of the reorganization are not transesterifying to change the network topology. It was reported that the shift in melting temperatures of superheated crystals has a power‐law dependence on heating rate with exponent between 0 and 0.5.126,244 As demonstrated in Figure 7.12A, PBT was allowed to crystallize at a slow cooling rate (1 °Cs‐1) to ‐10 °C and the obtained FSC curves at subsequent heating rates between 10 °Cs‐1 and 5000 °Cs‐1 show two peaks. The peak at lower temperatures (solid line) shifts to higher temperatures with increasing heating rate (dash line) and eventually these two melting peaks merge into one melting peak because of superheating. The PBT/PC(5) blend shows a similar behavior as neat PBT (Figure 7.12B). In contrast, as shown in Figures 7.12C and 7.12D, both the PBT C1(GLY) and C8(GLY) vitrimers show a very broad melting peak as compared to PBT and PBT/PC(5) blend. For C1(GLY), the broadness of the melting peaks increased with the increasing of heating rate. With the increase of the network density, two well‐separated melting peaks are observed and the lower melting temperature peak (solid line) do not merge with the higher melting temperature peak (dash line) with increasing heating rate. The lower melting peak temperatures are nearly constant while the higher melting peak temperatures shift to higher temperatures with increasing heating rate. These higher melting peak temperatures of C8(GLY) are not detectable at a heating rate  2000 °Cs‐1. Therefore, the recrystallization process is not influenced by the dynamic cross‐links from the PBT vitrimers, which is caused

115

Influence of the network nature on the crystallization/melting behavior of PBT by the fact that this reorganization process is just chain segment mobility within the crystalline lamellae.

‐1 ‐1 Cooling at 1 Cs Heating rate in Cs Cooling at 1 Cs‐1 Heating rate in Cs‐1

10 10 20 20 50 50 100 100 200 200 500 500 1000 1000 2000 2000 5000 5000

(A) Neat PBT (B) PBT/PC(5) blend 0 30 60 90 120 150 180 210 240 270 0 30 60 90 120 150 180 210 240 270

Temperature [C] Temperature [C] ‐1 ‐1 ‐1 Cooling at 1 C s Heating rate inC s‐1 Cooling at 1 Cs Heating rate in Cs 10 10 200 200 500 500 1000 1000 2000

2000

5000 (C) C1(GLY) (D) C8(GLY) 5000 0 30 60 90 120 150 180 210 240 270 0 30 60 90 120150180210240270

Temperature [C] Temperature [C] Figure 7.12. FSC curves, heat‐flow rate as a function of temperature, recorded on heating samples at different rates between 10 (top) and 5000 °Cs‐1 (bottom). (A)Neat PBT, (B) PBT/PC(5) blend, (C) C1(GLY), and (D) C8(GLY). The samples were previously cooled at 1 °Cs‐1 from the melt. The curves shifted vertically for clarity. 116

Chapter 7

7.3 Conclusions

In this chapter, we studied the influence of network properties on the melting and crystallization kinetics of PBT and PBT vitrimers by fast scanning calorimetry in the entire temperature range between Tg and Tm. The critical cooling rate for quenching a sample to below glass transition without crystallization is highly influenced by the network properties, i.e., for the neat and 0.5 wt% talc‐nucleated PBT melt need to be cooled faster than 200 and 2000 °Cs‐1, respectively, and this critical cooling rate decreases with decreasing talc loadings. When a physical network, PBT/PC blend, or a vitrimer network is introduced to the amorphous phase of PBT, this critical cooling rate is lower than neat PBT. In isothermal crystallization experiments, we demonstrated that the homogeneous and heterogeneous nucleation density of PBT can be tuned via playing with the type of network. PBT shows two crystallization‐rate maxima at 135 and 70 °C, with characteristic minimum crystallization times of about 0.72 and 0.15 s, respectively. This bimodal distribution of crystallization peak time behavior is confirming the earlier studies.25‐29 Addition of talc as a heterogeneous nuclei, the overall high temperature crystallization process speeds up, but the heterogeneous nuclei do not become dominant and cannot influence the crystallization kinetics. Homogeneous nucleation density is influenced by physical (PBT/PC blend) or vitrimer network because of the dynamics of the whole polymer chain is limited by the physical or vitrimer network, which influences the kinetics of forming the homogeneous nuclei. During the reorganization of the amorphous samples, the cold crystallization is not suppressed for the PBT/PC blend and the V1 samples at a heating rate as fast as 5000 °Cs‐ 1, at this heating rate, only C8(GLY) shows a complete suppression of the cold crystallization of PBT. In general, the higher the heating rate, the higher the cold crystallization temperature. It is shown in this study that the reorganization during the melting of recrystallized crystals is not influenced by either physical or vitrimer network, which is probably caused by the fact that this reorganization process is just chain segment movement within the crystalline lamellae. 7.4 Experimental section

Materials

Special grade poly(butylene terephthalate) pellets (Mn = 21.2 kg/mol, Mw = 46.6 kg/mol) against PMMA standards in 1,1,1,3,3,3‐hexafluoroisopropanol (HFiP) were provided by SABIC (Bergen op Zoom, the Netherlands) and used as received. Talc‐nucleated PBT samples

117

Influence of the network nature on the crystallization/melting behavior of PBT with different amount of talc, and PBT/PC (95/5) blend were also provided by SABIC (Bergen op Zoom, the Netherlands) and used as received. The vitrimer samples were synthesized by solid‐state (co)polymerization and the synthetic method is described in Chapter 5. Characterization methods Differential scanning calorimetry (DSC). DSC was performed by using a DSC Q1000 from TA Instruments. The measurements were carried out from ‐50 to 260 °C with heating and cooling rates of 10 °Cmin‐1 under a nitrogen flow of 50 mLmin‐1. Fast scanning chip calorimetry (FSC). FSC analyses were performed using a power‐ compensation Mettler‐Toledo Flash DSC 1 attached to a Huber intracooler TC100. The calorimeters were purged with dry nitrogen gas at a flow rate of 20 mlmin‐1. Prior loading of the sample, first the empty FSC sensor was conditioned and temperature‐corrected according to the specification of the instrument provider. Samples were prepared from the pellets using a microtome, to obtain thin sections with a thickness less than 20 m which then were further reduced in their lateral width to about 50‐100 m using a scalpel and a stereomicroscope. In this study, the sample was placed directly on the sensor and pre‐ melted at 270 °C for 0.01 s to establish a good thermal contact between the sample and the sensor. Polarized Optical Microscopy (POM). The sample on the chip sensor was placed between crossed polarizers using POM (Zeiss Axioplan 2 microscope) equipped with a Zeiss Axiocam camera. POM image was collected in transmission mode. Processing technique. The materials were compression molded at 260 °C and 100 bar for 15 minutes in a Collin Press 300G subsequently cooled with water. Rheological measurements. Dynamic shear measurements were performed on a strain‐ controlled AR‐G2 Rheometer (TA Instruments) by using a 25 mm parallel plate geometry and disk‐shaped specimens (25 mm diameter; 1 mm thick). Frequency sweeps from 0.01 to 500 rad/s were performed at 260 °C with a strain of 1%, which corresponds to the linear viscoelastic regime. 7.5 Acknowledgements

We thank Dr. Jan Lohmeijer (SABIC) for providing the nucleated PBT, PBT/PC blend and helpful discussions. Dr. Martin van Drongelen (University of Twente, the Netherlands) is gratefully acknowledged for his help with Flash DSC experiments. Dr. Enrico Troisi (SABIC) is gratefully acknowledged for his help with data analysis and helpful discussions.

118

Chapter 8

Epilogue

The goal of this thesis has been to create a new type of high‐performance semi‐ crystalline poly(butylene terephthalate) (PBT) vitrimer with PBT‐like characteristics, i.e., fast crystallization rate, high stiffness, but also with a high melt strength, and thus, a better dimensional stability above the melting temperature. Within the framework of this project, we aimed at the incorporation of multifunctional alcohols such as glycerol (GLY), 1,1,1‐ tris(hydroxymethyl)propane (TMP) or pentaerythritol (PEY) as cross‐linkers into PBT via a two‐step solid‐state (co)polymerization approach in the presence of a transesterification catalyst. These materials are potential candidates for industrial applications. Since PBT vitrimers are made of cheap feedstocks and can be reprocessed, repaired and recycled, we can envision these materials in applications at higher temperatures and as matrix material for composites for the automotive, construction and wind energy sectors. In this chapter, we first recap some of the conclusions from the previous chapters to clarify what we learned from our work. A technology assessment will be carried out to evaluate the advantages and disadvantages of our current solid‐state (co)polymerization method with an example for a practical solution for the scale‐up of a PBT/polyol‐based systems.

8.1 Conclusions

Vitrimer chemistry based on transesterification reactions is a versatile method to improve the performance of engineering plastics, like PBT. In this thesis, we showed that PBT vitrimers can be prepared via a two‐step solid‐state (co)polymerization method with glycerol as the cross‐linker and zinc acetyl acetonate (Zn(acac)2) as the catalyst. The prepared PBT vitrimers have semi‐crystalline characteristics inherited from PBT and a macroscopic flow behavior originated from the covalent adaptable network. Besides, PBT vitrimers unambiguously possess a triclinic –crystal structure as neat PBT. These semi‐ crystalline PBT vitrimers can be recycled multiple times by compression molding without a substantial loss in mechanical and thermal properties. The temperature‐dependent dynamic mechanical properties of semi‐crystalline PBT vitrimers comprise three regimes in the temperature range from ‐15 to 270 °C, i.e., glassstate, the plateau between Tg and Tm, and rubber plateau modulus above Tm. Variation of the [catalyst]/[cross‐linker] ratio in the initial formulation of PBT/glycerol‐based vitrimers affords a fine tuning of the plateau modulus between Tg and Tm, and the rubber plateau

119

Epilogue

modulus above Tm, while the modulus in the glassy state is similar to that of neat PBT( 1‐ 3 GPa). More specifically, the [cross‐linker] primarily controls the rubber plateau modulus and the [catalyst]/[cross‐linker] ratio controls the plateau modulus between Tg and Tm. The network dynamics of PBT vitrimers are primarily controlled by the [catalyst]/[cross‐linker] ratio. Above all, PBT vitrimers exhibit a better creep resistance over neat PBT at service temperatures (< Tm) while a similar crystallinity is maintained. The physical and mechanical properties of semi‐crystalline vitrimers are profoundly influenced by their thermomechanical history, i.e., the temperature of the melt, annealing time and temperature. The initially heterogeneous spatial distribution of dynamic cross‐ links formed during the solid‐state (co)polymerization process will be randomized to a homogeneous distribution of dynamic cross‐links upon melting. This process is driven by an entropy increase due to a morphology change from an ordered crystalline to a random amorphous state and this randomization process is accelerated in the presence of oscillatory stress. Interestingly enough, upon cooling, driven by the formation of a long‐ range crystalline phase, these homogeneously distributed dynamic cross‐links were exclusively relocated in the amorphous phase and the heterogeneous distribution of dynamic cross‐links was recovered. For PBT/polyol‐based vitrimers, we also exploited to use the different transesterification kinetics between glycerol (two primary and one secondary hydroxyl groups) and 1,1,1‐ tris(hydroxymethyl)propane (three primary hydroxyl groups) catalyzed by zinc acetylacetonate for tuning the viscoelastic properties, i.e., stress relaxation time, weldability, creep resistance and (dynamic) mechanical properties, of bulk materials through simply altering the cross‐linker. It is particularly challenging to balance fast exchange reactions at temperatures high enough to process vitrimers at high throughput but without degradation, and very slow exchange reactions for creep resistance at service temperature. In this thesis, we developed a new strategy based on in situ network formation by a processing‐induced deprotection approach in the presence of carboxylic acid end groups from PBT rendering a similar network with vitrimer characteristics as in the case of a TFA‐deprotected reference material. In this way, we have the benefit of processing a linear PBT copolymer for injection molding applications, i.e., short relaxation time, low viscosity, high crystallization rate, and vitrimer characteristics for the cured final part, i.e., high melt strength, recyclability, and weldability. Our in situ network formation strategy shares the same aim as the recent work from Leibler and coworkers;85 they reported a rapid exchange reaction based on the metathesis of dioxaborolanes to enable the commodity vitrimers to be processed with fast production rates and current equipment during extrusion or injection molding.

120

Chapter 8

The critical cooling rate to suppress crystallization is dependent on the network properties, i.e., vitrimer network density and dynamics. For isothermal crystallization, the annealing temperature dependence of the crystallization peak‐time for PBT shows a bimodal curve with two minima. The heterogeneous nucleation density can be tuned by the talc concentration while the homogeneous nucleation is independent on talc concentration, but the homogeneous nucleation density is influenced by both the physical and dynamic covalent cross‐links. For an amorphous material, only the PBT vitrimer with a rubbery plateau modulus above Tm around 2 MPa can suppress the cold crystallization and subsequent melting process at a heating rate as high as 5000 °Cs‐1. Furthermore, the reorganization during the melting of recrystallized crystals is not influenced by either physical or dynamic covalent cross‐links at the heating rate range between 10 and 5000 °Cs‐1. So far, we have just "scratched the surface" of PBT vitrimers, leaving several open questions in this thesis: on the polymer chemistry side, for instance, in chapter 4, we found that The initially heterogeneous spatial distribution of dynamic cross‐links (fully misible with the amorphous phase of PBT) formed during the solid‐state (co)polymerization process will be randomized to a homogeneous distribution of dynamic cross‐links upon melting. Therefore, one of the interesting question will be that what is the influence of the imiscibility of cross‐linker within the amorphous phase of PBT on the vitrimer properties? Can we improve the catalytic efficiency via adding a cocatalyst?245 Why do the model reaction kinetics not match the network dynamics of PBT/GLY and PBT/TMP‐based vitrimers? On the polymer physics side, it is crucial to further understand the processing‐ morphology‐properties relationships of PBT vitrimers. It is known that the morphology of semi‐crystalline polymers is strongly influenced by the conditions applied during processing, which is of significant importance for the final mechanical properties.246–249 Thus, it is crucial to understand how cooling rate, pressure and shear flow influence the mechanical properties of PBT vitrimers? Besides shear flow, the extensional flow of polymer melts plays a dominant role in polymer processes such as film blowing, blow molding, fiber spinning and thermoforming.6,44,46 Therefore, knowing the extensional flow properties of polymer melts is very important to control and predict their ability to be shaped or processed into products. Furthermore, we also attempted to compare our PBT/polyol‐based vitrimers91 with the PBT/epoxy‐based vitrimers studied by Leibler and coworkers.90 However, these materials are prepared by two different preparation techniques, reactive extrusion and solid‐state (co)polymerization for PBT/epoxy and PBT/polyol‐based vitrimers, respectively. These two processes render materials with a different crystalline/amorphous morphology. To accurately compare those two vitrimer systems, besides the morphology, there are more

121

Epilogue parameters that need to be taken into consideration:the network build up, the network structure, transesterification catalyst and processing parameters. Therefore, we feel that such comparison is still premature.

8.2 Technology assessment

PBT vitrimers can be used for injection molding applications as well as process techniques involving fast elongation flow, like thermoforming. Therefore, PBT vitrimers can expand the application area for PBT. Also, PBT vitrimers are expected to be a promising candidate to replace the thermoset‐based composites giving their high stiffness, creep resistance, weldability and recyclability. One of the main drawbacks of the current solid‐state (co)polymerization (SSP) is the standard solution physical mixture preparation approach, in which a very expensive and toxic solvent, 1,1,1,3,3,3‐hexafluoroisopropanol(HFIP) (1000 €/L) was used. Therefore, we tried an alternative production route, which is shown in Scheme 8.1.

Scheme 8.1. Schematic representation of premixing PBT, polyol, and catalyst via fast extrusion, subsequently solid‐state (co)polymerization was employed to increase the MW and cross‐link the material.

Here, we use a DSM Xplore 15 ml twin‐screw mini‐extruder under a nitrogen flow to prepare the mixture for SSP. Typical parameters are as follows: temperature = 260 °C, roto speed = 50‐100 rpm, mixing time after feeding  30 seconds, and nitrogen flow. A typical extrusion experiment is shown in Scheme 8.2. It contains several steps: clean the extruder with cleaning agents and PBT, feeding/mixing and extrusion. The extruded yarn was cut into small length, vitrified by liquid nitrogen, and subsequently ground into powder using an IKA A11 Basic Analytical mill. This powder was then dried under vacuum for a period of 24 h at 50 °C.

122

Chapter 8

Scheme 8.2. A general reactive extrusion process, from cleaning, feeding, mixing to extrusion.

The development of molecular weight during the reactive extrusion process is shown in Figure 8.1A. In general, the molecular weight of PBT is decreasing as a function of extrusion time catalyzed by Zn(acac)2, and the Ð is nearly constant (Ð  2). The molecular weight decreases faster at the first 1 min, and then it is more or less constant as a function of extrusion time. The molecular weight decreases from 83 kg/mol to 55 kg/mol and 35 kg/mol after 10 min extrusion time for the system without and with 2.4 mol% pentaerythritol in the presence of 0.2 mol% Zn(acac)2, respectively. Above all, there is no molecular weight build up during reactive extrusion.

100 3.0 110 (A) Withou PEY (B) 90 With PEY 100 Soluble Insoluble 80 2.5 90

70 80 2.0 Đ 60 70 [kg/mol] w [kg/mol]

w 60 M 50 M 1.5 50 40 Zn(acac) 40 2 30 With PEY 1.0 30 0246810 01234567 t [min] t [hour] extrusion ssp Figure 8.1. (A) Development of Mw and dispersity as a function of textrusion in the presence of zinc acetylacetonate (Zn(acac)2) with and without pentaerythritol.(B) Development of Mw as a function of tssp in the presence of zinc acetylacetonate with pentaerythritol.

The molecular weight of the mixture prepared via a fast extrusion process was built up ‐1 during solid‐state (co)polymerization at 180 °C with 0.5 Lmin N2 flow. Transesterification reactions occur between the hydroxyl groups from pentaerythritol and ester groups from

123

Epilogue the PBT chains present in the amorphous phase. As shown in Figure 8.1B, this recombination reactions lead to the increase of molecular weight and the material is not ‐1 soluble in HFiP when tssp > 4.5 h at 180 °C with 0.5 Lmin N2 flow. We further explored the influence of different ligands of zinc salts at identical pentaerythritol content on the properties of the PBT/pentaerythritol‐based vitrimers. The molecular structures of the catalysts are shown in Scheme 8.3.

Scheme 8.3. Molecular structure of the catalysts used in this chapter.

110 6 (A) Zinc stearate (B) 100 Zinc stearate 5 90 Zinc acetate Zn(OAc) 80 2 Zn(acac) 4 2 Zn(acac) 70 Đ 2

[kg/mol] 3 w 60 M 50 ZnO ZnO 2 40 With PEY With PEY 30 1 01234567 01234567 t [hour] t [hour] ssp ssp

Figure 8.2. Development of (A) Mw and (B) dispersity as a function of tssp in the presence of zinc stearate (upper triangle), zinc acetate (circle), zinc acetylacetonate (down triangle) and zinc oxide (square) with pentaerythritol. The catalyst content was constant at 0.2 mol% concerning the PBT repeat units.

During SSP (Figure 8.2), the molecular weight increases and the effects are dependent on the used ligands: Zinc stearate > Zn(OAc)2 > Zn(acac)2. The Ð also increases similarly: Zinc stearate > Zn(OAc)2  Zn(acac)2. The gelation time for zinc acetate and zinc acetylacetonate catalyzed systems during SSP is 4.5 h, while the time decreased to 4 h for zinc stearate. Zinc oxide (ZnO) is not activated in the PBT/pentaerythritol‐based system, and it has no effect on the molecular weight build up and copolymer architecture change in comparison to the rest of the catalysts studied in this chapter.

124

Chapter 8

2.0 (A) (B) 103 1.5 Zinc stearate 1.0 102 0.5

101

' [MPa] 0.0

E Zn(OAc) PBT 2 ‐0.5 0 Zinc stearate

10 Heat flow, exo up [w/g] Zn(acac)2 ‐1.0

Zn(OAc)2 Zn(acac)2 10‐1 ‐1.5 0 30 60 90 120 150 180 210 240 140 160 180 200 220 240 Temperature [C] Temperature [C] Figure 8.3. (A) The storage modulus of PBT and PBT/PEY‐based copolymers catalyzed with 0.2 mol% Zn2+, heating is at 3 °Cmin‐1 and frequency of 1 Hz and (B) DSC curves of PBT/PEY‐based vitrimers after solid‐state (co)polymerization. All the samples containing 2.4 mol% pentaerythritol.

The thermomechanical properties of the prepared materials are shown in Figure 8.3.

The glass transition temperatures (Tg) of the PBT vitrimers containing 2.4 mol% pentaerythritol catalyzed by 0.2 mol% Zn2+ with different ligands are slightly higher than neat PBT. The Tg is increasing in the following manner: Tg = 55.0 °C (PBT), 58.3 °C (Zinc stearate), 60.4 °C (Zinc acetylacetonate), and 62.1 °C (Zinc acetate). Three PBT vitrimers with identical pentaerythritol and Zn(II) catalyst content exhibit similar DMA curves, although those materials catalyzed by Zn(II) catalyst with different ligands. The thermal properties of those materials characterized by differential scanning calorimetry (DSC) show a similar degree of crystallinity 34%, while it is 37% for neat PBT. The crystallization‐peak temperature of those PBT vitrimers is about 187 °C, which is slightly lower than neat PBT (192 °C). Nevertheless, these DSC results are in good agreement with the results obtained via DMTA. The dynamic behavior of the PBT vitrimers was probed by small‐amplitude oscillatory frequency sweep experiments. The macroscopical flow behavior of the PBT/pentaerythritol‐based vitrimers with 2.4 mol% pentaerythritol catalyzed by 0.2 mol% zinc salts with different ligands are shown in Figure 8.4. In the examined frequency range the materials exhibit a ‐dependent G and shows much smaller G values. Thus, in the range of frequencies tested, it behaves like a solid‐like gel (G > G). The plateau modulus 0 (GN ) taken at the minimum loss modulus point shows a trend in the following order: Zinc stearate > Zn(OAc)2  Zn(acac)2. The materials catalyzed by Zn(OAc)2 and Zn(acac)2 also exhibit a similar viscosity, which is about 2107 Pas taken at 0.01 rad/s. However, the viscosity of the material catalyzed by Zinc stearate is about 2 times higher than the ones

125

Epilogue

7 catalyzed by Zn(OAc)2 and Zn(acac)2, which is about 4  10 Pas. Furthermore, all viscosity curves exhibit a slope of ‐1 which indicates a well‐developed network. 108 Zn(OAc)2

7 10 Zn(acac)2 105 Zinc stearate 6

 10 Pas 105  4 Zn(OAc)2 10 4 Zn(acac) 10

G'(open), G''(closed) [Pa] 2 (A) Zinc stearate 3 (B) ‐2 ‐1 0 1 2 10 10 10 10 10 10 10‐2 10‐1 100 101 102 rad/s [rad/s] Figure 8.4. (A) Elastic (open symbols) and viscous (filled symbols) moduli and (B) complex viscosity versus frequency at 250 °C with 1% strain for the PBT/PEY‐based vitrimer catalyzed by zinc salts with different ligands.

To sum up, we showed that the standard solvent approach is an easy problem to be solved. With the help of reactive extrusion or compounding techniques, it is easy to prepare the mixtures at an industrial scale. For scale‐up of SSP, there are several companies specialized in SSP that can provide services at production levels, such as Cumapol250 and Polymetrix.251 Therefore, PBT vitrimer, produced with cheap feedstocks, can be easily scale‐ up without using HFiP as a solvent.

PBT vitrimer combining a high tensile storage modulus below Tm and vitrimer characteristics (e.g., recyclability, repairability, and weldability) holds a promising opportunity to replace the traditional thermosets matrix for polymer composites application. In this thesis, although we did not cover the area of (short) long glass‐fiber reinforced PBT vitrimers composites or PBT vitrimer nanocomposites, in principle, we can expect a further enhancement of the mechanical performance, creep resistance or extra functionalities (e.g., photo‐thermal effect) of raw PBT vitrimers.252,253 Besides the PBT vitrimer (nano)composites area, there are still many aspects could be explored, i.e., shape‐ memory properties,254,255 solid‐state plasticity,65,99,101,116 steam sterilization and wire and cable insulation applications. Therefore, there is a great future for semi‐crystalline PBT vitrimers.

126

Bibliography

(1) Scheirs, J.; Long, T. E. Modern Polyesters, Chemistry and Technology of Polyesters and Copolyesters; John Wiley & Sons: Chichester, 2004. (2) Fakirov, S. Handbook of Thermoplastic Polymers: Homopolymers, Copolymers, Blends, and Composites; Wiley‐ VCH: Weinheim, 2002. (3) Wang, J. S.; Zhao, H. B.; Ge, X. G.; Liu, Y.; Chen, L.; Wang, D. Y.; Wang, Y. Z. Ind. Eng. Chem. Res. 2010, 49, 4190–4196. (4) Zhao, H.‐B.; Wang, X.‐L.; Guan, Y.; Wang, X.‐L.; Chen, L.; Wang, Y.‐Z. Polymer 2015, 70, 68–76. (5) Miller, S. Macrocyclic polymers from cyclic oligomers of poly(butylene terephthalate), University of Massachusetts, Amherst, 1998. (6) Lim, L.‐T.; Auras, R.; Rubino, M. Prog. Polym. Sci. 2008, 33, 820–852. (7) Ghijsels, A. and De Clippeleir, J. Int. Polym. Process. 1994, 9, 252–257. (8) Rubinstein, M.; Colby, R. H. Polymer Physics; OUP: Oxford, U.K., 2003. (9) Gantillon, B.; Spitz, R.; McKenna, T. F. Macromol. Mater. Eng. 2004, 289, 88–105. (10) Jansen, M. A. G. Modification of poly(butylene terephthalate) by incorporation of comonomers in the solid state, Technische Universiteit Eindhoven, 2005. (11) Vouyiouka, S. N.; Karakatsani, E. K.; Papaspyrides, C. D. Prog. Polym. Sci. 2005, 30, 10–37. (12) Cheng, Stephen Z. D.; Cao, M. Y.; Wunderlich, B. Macromolecules 1986, 19, 1868– 1876. (13) Cheng, Stephen Z. D.; Wu, Zong Quan; Wunderlich, B. Macromolecules 1987, 20, 2802–2810. (14) Cheng, Stephen Z. D.; Wunderlich, B. Macromolecules 1987, 20, 1630–1637. (15) Cheng, S. Z. D.; Panc, R.; Wunderlich, B. Makromol. Chem. 1988, 189, 2443–2458. (16) Papaspyrides, C.D.; Vouyiouka, S. N. Solid state polymerization; John Wiley & Sons: Hoboken, New Jersey, 2009. (17) Jabarin, S. A.; Lofgren, E. A. J. Appl. Polym. Sci. 1995, 32, 5315–5335. (18) Huang, B.; Walsh, J. J. Polymer 1998, 39, 6991–6999. (19) Goodner, M. D.; Gross, S. M.; Desimone, J. M.; Roberts, G. W.; Kiserow, D. J. J. Appl. Polym. Sci. 2001, 79, 928–943. (20) B. Duh. J. Appl. Polym. Sci. 2006, 102, 623–632. (21) Duh, B. J. Appl. Polym. Sci. 2002, 83, 1288–1304. (22) Vouyiouka, S. N.; Papaspyrides, C. D.; Weber, J. N.; Marks, D. N. Polymer 2007, 48, 4982–4989. 127

Bibliography

(23) Filgueiras, V.; Vouyiouka, S. N.; Papaspyrides, C. D.; Lima, E. L.; Pinto, J. C. Macromol. Mater. Eng. 2011, 296, 113–121. (24) Gubbels, E. Melt copolymerization and solid‐state modification as methods to prepare novel bio‐based polycondensates, Technische Universiteit Eindhoven, 2014. (25) Ton Loontjens, Wil Belt, Dirk Stanssens, P. W. Macromol. Symp. 1993, 216, 211– 216. (26) Bikiaris, D. N.; Karayannidis, G. P. J. Polym. Sci. Part A Polym. Chem. 1996, 34, 1337– 1342. (27) Haralabakopoulos, A. A.; Tsiourvas, D.; Paleos, C. M. J. Appl. Polym. Sci. 1998, 71, 2121–2127. (28) Bikiaris, D. N.; Karayannidis, G. P. J. Appl. Polym. Sci. 1998, 70, 797–803. (29) Guo, B.; Chan, C. J. Appl. Polym. Sci. 1999, 71, 1827–1834. (30) On, S. J. A. F.; Leterrier, Y. Polym. Eng. Sci. 2000, 40, 1942–1952. (31) Japon, S.; Boogh, L.; Leterrier, Y.; Manson, J.‐A. E. Polymer 2000, 41, 5809–5818. (32) Jacques, B.; Devaux, J.; Legras, R. Polymer 1996, 37, 1189–1200. (33) Jacques, B.; Devaux, J.; Legras, R.; Nield, E. Macromolecules 1996, 29, 3129–3138. (34) Jacques, B.; Devaux, J.; Legras, R.; Nield, E. Polymer 1997, 38, 5367–5377. (35) Dias, M. L.; Silva, A. P. F.; Eloisa, P. Polym. Eng. Sci. 2000, 40, 1777–1782. (36) Inata, H.; Matsumura, S. J. Appl. Plolym. Sci. 1985, 30, 3325–3337. (37) Inata, H.; Matsumura, S.; Development, P. J. Appl. Plolym. Sci. 1986, 32, 5193–5202. (38) Carbassi, F.; Messina, G.; Guido, I.; Enichem, D.; Fauser, V. G. J. Appl. Polym. Sci. 1993, 50, 1501–1509. (39) Loontjens, T.; Pauwels, K.; Derks, F.; Neilen, M.; Sham, C. K.; Serne, M. J. Appl. Plolym. Sci. 1997, 65, 1813–1819. (40) Christof, W.; Philipp, M.; Mülhaupt, R. J. Appl. Plolym. Sci. 1997, 66, 633–642. (41) Karayannidis, G. P.; Psalida, E. A. J. Appl. Plolym. Sci. 2000, 77, 2206–2211. (42) Böhme, F.; Jakisch, L.; Komber, H.; Wursche, R. Polym. Degrad. Stab. 2007, 92, 2270–2277. (43) Incarnato, L.; Scarfato, P.; Maio, L. Di; Acierno, D. Polymer 2000, 41, 6825–6831. (44) Gahleitner, M. Prog. Polym. Sci. 2001, 26, 895–944. (45) Sun, Y.; Wu, L.; Bu, Z.; Li, B.; Li, N.; Dai, J. Ind. Eng. Chem. Res., 2014, 53, 10380– 10386. (46) Mannion, A. M.; Bates, F. S.; Macosko, C. W. Macromolecules 2016, 49, 4587–4598. (47) Schmalz, H.; Abetz, V.; Lange, R.; Soliman, M. Macromolecules 2001, 34, 795–800. (48) Cleland, M. R.; Parks, L. A.; Cheng, S. Nucl. Instruments Methods Phys. Res. Sect. B 2003, 208, 66–73.

128

Bibliography

(49) Rouif, S. Nucl. Instruments Methods Phys. Res. Sect. B 2005, 236, 68–72. (50) Mizera, A.; Manas, M.; Manas, D.; Holik, Z.; Stanek, M.; Navratil, J.; Bednarik, M. Adv. Mater. Res. 2014, 1025–1026, 256–260. (51) Kloxin, C. J.; Scott, T. F.; Adzima, B. J.; Bowman, C. N. Macromolecules 2010, 43, 2643–2653. (52) Bowman, C. N.; Kloxin, C. J. Angew. Chem. Int. Ed. 2012, 51, 4272–4274. (53) Kloxin, C. J.; Bowman, C. N. Chem. Soc. Rev. 2013, 42, 7161–7173. (54) Montarnal, D.; Capelot, M.; Tournilhac, F.; Leibler, L. Science 2011, 334, 965–968. (55) Capelot, M.; Montarnal, D.; Leibler, L. J. Am. Chem. Soc. 2012, 134, 7664–7667. (56) Capelot, M.; Unterlass, M. M.; Tournilhac, F.; Leibler, L. ACS Macro Lett. 2012, 1, 789–792. (57) Cash, J. J.; Kubo, T.; Bapat, A. P.; Sumerlin, B. S. Macromolecules 2015, 48, 2098– 2106. (58) Cromwell, O. R.; Chung, J.; Guan, Z. J. Am. Chem. Soc. 2015, 137, 6492–6495. (59) Denissen, W.; Rivero, G.; Nicolaÿ, R.; Leibler, L.; Winne, J. M.; Du Prez, F. E. Adv. Funct. Mater. 2015, 25, 2451–2452. (60) Denissen, W.; Droesbeke, M.; Nicolaÿ, R.; Leibler, L.; Winne, J. M.; Du Prez, F. E. Nat. Commun. 2017, 8, 14857. (61) Ying, H.; Zhang, Y.; Cheng, J. Nat. Commun. 2014, 5, 3218. (62) Zhang, Y.; Ying, H.; Hart, K. R.; Wu, Y.; Hsu, A. J.; Coppola, A. M.; Kim, T. A.; Yang, K.; Sottos, N. R.; White, S. R.; Cheng, J. Adv. Mater. 2016, 28, 7646–7651. (63) Fang, Z.; Zheng, N.; Zhao, Q.; Xie, T. ACS Appl. Mater. Interfaces 2017, 9, 22077– 22082. (64) Fortman, D. J.; Brutman, J. P.; Cramer, C. J.; Hillmyer, M. A.; Dichtel, W. R. J. Am. Chem. Soc. 2015, 137, 14019–14022. (65) Zheng, N.; Fang, Z.; Zou, W.; Zhao, Q.; Xie, T. Angew. Chemie ‐ Int. Ed. 2016, 55, 11421–11425. (66) Fortman, D. J.; Brutman, J. P.; Hillmyer, M. A.; Dichtel, W. R. J. Appl. Polym. Sci. 2017, 134, 44984. (67) Obadia, M. M.; Mudraboyina, B. P.; Serghei, A.; Montarnal, D.; Drockenmuller, E. J. Am. Chem. Soc. 2015, 137, 6078‐6083. (68) Hendriks, B.; Waelkens, J.; Winne, J. M.; Du Prez, F. E. ACS Macro Lett. 2017, 6, 930– 934. (69) Deng, G.; Tang, C.; Li, F.; Jiang, H.; Chen, Y. Macromolecules 2010, 43, 1191–1194. (70) Deng, G.; Li, F.; Yu, H.; Liu, F.; Liu, C.; Sun, W.; Jiang, H.; Chen, Y. ACS Macro Lett. 2012, 1, 275–279. (71) Taynton, P.; Yu, K.; Shoemaker, R. K.; Jin, Y.; Qi, H. J.; Zhang, W. Adv. Mater. 2014,

129

Bibliography

26, 3938–3942. (72) Deng, G.; Ma, Q.; Yu, H.; Zhang, Y.; Yan, Z.; Liu, F.; Liu, C.; Jiang, H.; Chen, Y. ACS Macro Lett. 2015, 4, 467–471. (73) Zhang, Y.; Barboiu, M. Chem. Rev. 2016, 116, 809–834. (74) Canadell, J.; Goossens, H.; Klumperman, B. Macromolecules 2011, 44, 2536–2541. (75) Pepels, M.; Filot, I.; Klumperman, B.; Goossens, H. Polym. Chem. 2013, 4, 4955– 4965. (76) Lei, Z. Q.; Xiang, H. P.; Yuan, Y. J.; Rong, M. Z.; Zhang, M. Q. Chem. Mater. 2014, 26, 2038–2046. (77) Rekondo, A.; Martin, R.; Ruiz de Luzuriaga, A.; Cabañero, G.; Grande, H. J.; Odriozola, I. Mater. Horizons 2014, 1, 237–240. (78) Martin, R.; Rekondo, A.; Ruiz de Luzuriaga, A.; Cabañero, G.; Grande, H. J.; Odriozola, I. J. Mater. Chem. A 2014, 2, 5710‐5715. (79) Imbernon, L.; Oikonomou, E.; Norvez, S.; Leibler, L. Polym. Chem. 2015, 4271–4278. (80) Gao, W.; Bie, M.; Liu, F.; Chang, P.; Quan, Y. ACS Appl. Mater. Interfaces 2017, 9, 15798–15808. (81) Zheng, P.; McCarthy, T. J. J. Am. Chem. Soc. 2012, 134, 2024–2027. (82) Schmolke, W.; Perner, N.; Seiffert, S. Macromolecules 2015, 48, 8781–8788. (83) Lu, Y. X.; Tournilhac, F.; Leibler, L.; Guan, Z. J. Am. Chem. Soc. 2012, 134, 8424–8427. (84) Lu, Y. X.; Guan, Z. J. Am. Chem. Soc. 2012, 134, 14226–14231. (85) Röttger, M.; Domenech, T.; Weegen, R.; Breuillac, A.; Renaud, N.; Leibler, L. Science 2017, 356, 62–65. (86) Kuang, X.; Liu, G.; Dong, X.; Wang, D. Mater. Chem. Front. 2017, 1, 111–118. (87) Zhang, L.; Rowan, S. J. Macromolecules 2017, 50, 5051–5060. (88) Obadia, M. M.; Jourdain, A.; Cassagnau, P.; Montarnal, D.; Drockenmuller, E. Adv. Funct. Mater. 2017, DOI：10.1002/adfm.201703258. (89) Brutman, J. P.; Delgado, P. A.; Hillmyer, M. A. ACS Macro Lett. 2014, 3, 607–610. (90) Demongeot, A.; Groote, R.; Goossens, J. G. P.; Hoeks, T.; Tournilhac, F.; Leibler, L. Macromolecules 2017, 50, 6117−6127. (91) Zhou, Y.; Goossens, J. G. P.; Sijbesma, R. P.; Heuts, J. P. A. Macromolecules 2017, 50, 6742–6751. (92) Pei, Z.; Yang, Y.; Chen, Q.; Terentjev, E. M.; Wei, Y.; Ji, Y. Nat. Mater. 2014, 13, 36– 41. (93) Yang, Y.; Pei, Z.; Li, Z.; Wei, Y.; Ji, Y. J. Am. Chem. Soc. 2016, 138, 2118–2121. (94) Yang, Y.; Pei, Z.; Zhang, X.; Tao, L.; Wei, Y.; Ji, Y. Chem. Sci. 2014, 5, 3486–3492. (95) Ube, T.; Kawasaki, K.; Ikeda, T. Adv. Mater. 2016, 8212–8217. (96) Pei, Z.; Yang, Y.; Chen, Q.; Wei, Y.; Ji, Y. Adv. Mater. 2016, 28, 156–160.

130

Bibliography

(97) Li, Z.; Yang, Y.; Wang, Z.; Zhang, X.; Chen, Q.; Qian, X.; Liu, N.; Wei, Y.; Ji, Y. J. Mater. Chem. A 2017, 5, 6740–6746. (98) Lu, L.; Fan, J.; Li, G. Polymer 2016, 105, 10–18. (99) Zhao, Q.; Zou, W.; Luo, Y.; Xie, T. Sci. Adv. 2016, 2, e1501297–e1501297. (100) Yang, Z.; Wang, Q.; Wang, T. ACS Appl. Mater. Interfaces 2016, 8, 21691–21699. (101) Yang, L.; Zhang, G.; Zheng, N.; Zhao, Q.; Xie, T. Angew. Chemie 2017, 129, 12773– 12776. (102) Imbernon, L.; Norvez, S. Eur. Polym. J. 2016, 82, 347–376. (103) Legrand, A.; Soulié‐Ziakovic, C. Macromolecules 2016, 49, 5893–5902. (104) Yu, K.; Shi, Q.; Dunn, M. L.; Wang, T.; Qi, H. J. Adv. Funct. Mater. 2016, 26, 6098– 6106. (105) Taynton, P.; Ni, H.; Zhu, C.; Yu, K.; Loob, S.; Jin, Y.; Qi, H. J.; Zhang, W. Adv. Mater. 2016, 28, 2904–2909. (106) Chabert, E.; Vial, J.; Cauchois, J.‐P.; Mihaluta, M.; Tournilhac, F. Soft Matter 2016, 12, 4838. (107) Horiz, M.; Ruiz de Luzuriaga, A.; Martin, R.; Markaide, N.; Rekondo, A.; Cabañ ero, G.; Rodrí guez, J.; Odriozola, I. Mater. Horiz. 2016, 3, 241–247. (108) Zhang, H.; Xu, X. Compos. Part A Appl. Sci. Manuf. 2017, 99, 15–22. (109) Kissounko, D. A.; Taynton, P.; Kaffer, C. Reinf. Plast. 2017, 1–5. (110) Altuna, F. I.; Pettarin, V.; Williams, R. J. J. Green Chem. 2013, 15, 3360–3366. (111) Ma, Z.; Wang, Y.; Zhu, J.; Yu, J.; Hu, Z. J. Polym. Sci. Part A Polym. Chem. 2017, 1790– 1799. (112) Matharu, A. S.; Ding, C.; Shuttleworth, P. S.; Clark, J.; Makin, S. Green Chem. 2015, 17, 4000–4008. (113) Lyon, G. B.; Cox, L. M.; Goodrich, J. T.; Baranek, A. D.; Ding, Y.; Bowman, C. N. Macromolecules 2016, 49, 8905–8913. (114) Shi, Q.; Yu, K.; Kuang, X.; Mu, X.; Dunn, C. K.; Dunn, M. L.; Wang, T.; Jerry Qi, H. Mater. Horiz. 2017, 4, 598–607. (115) Denissen, W.; Winne, J. M.; Du Prez, F. E. Chem. Sci. 2016, 7, 30–38. (116) Zou, W.; Dong, J.; Luo, Y.; Zhao, Q.; Xie, T. Adv. Mater. 2017, 29, 1606100. (117) Schrauwen, B. A. G.; Janssen, R. P. M.; Govaert, L. E.; Meijer, H. E. H. Macromolecules 2004, 37, 6069–6078. (118) Murthy, N. S.; Curran, S. A.; Aharoni, S. M.; Minor, H. Macromolecules 1991, 24, 3215–3220. (119) Alizadeh, A.; Richardson, L.; Xu, J.; McCartney, S.; Marand, H.; Cheung, Y. W.; Chum, S. Macromolecules 1999, 32, 6221–6235. (120) Androsch, R.; Wunderlich, B. Macromolecules 2000, 33, 9076–9089.

131

Bibliography

(121) Sohn, S.; Alizadeh, A.; Marand, H. Polymer 2000, 41, 8879–8886. (122) Marand, H.; Alizadeh, A.; Farmer, R.; Desai, R.; Velikov, V. Macromolecules 2000, 33, 3392–3403. (123) Wunderlich, B. Thermal analysis of Polymeric Materials; Springer: Berlin, 2005. (124) Yen, J. T.; Runt, J. J. Polym. Sci. Part A Polym. Chem. 1989, 27, 1543–1550. (125) Yasuniwa, M.; Tsubakihara, S.; Ohoshita, K.; Tokudome, S. I. J. Polym. Sci. Part B Polym. Phys. 2001, 39, 2005– 2015. (126) Furushima, Y.; Kumazawa, S.; Umetsu, H.; Toda, A.; Zhuravlev, E.; Schick, C. Polymer 2017, 109, 307–314. (127) Yokouchi, M.; Sakakibara, Y.; Chatani, Y.; Tadokoro, H.; Tanaka, T. Macromolecules 1975, 9, 266–273. (128) M. G. Brereton R. Jakeways, T. Smith and I. M. Ward, G. R. D.; Brereton, M.; Davies, G.; Jakeways, R.; Smith, T.; Ward, I. . Polymer 1978, 19, 17–26. (129) Bornschlegl, E.; Bonart, R. Colloid Polym. Sci. 1980, 258, 319–331. (130) Tashiro, K.; Nakai, Y.; Kobayashi, M.; Tadokoro, H. Macromolecules 1980, 13, 137– 145. (131) Konishi, T.; Miyamoto, Y. Polym. J. 2010, 42, 349–353. (132) Konishi, T.; Nishida, K.; Matsuba, G.; Kanaya, T. Macromolecules 2008, 41, 3157– 3161. (133) Rhoades, A. M.; Williams, J. L.; Androsch, R. Thermochim. Acta 2015, 603, 103–109. (134) Mollova, A.; Androsch, R.; Mileva, D.; Schick, C.; Benhamida, A. Macromolecules 2013, 46, 828–835. (135) De Santis, F.; Adamovsky, S.; Titomanlio, G.; Schick, C. Macromolecules 2007, 40, 9026–9031. (136) Silvestre, C.; Cimmino, S.; Duraccio, D.; Schick, C. Macromol. Rapid Commun. 2007, 28, 875–881. (137) Yu, K.; Taynton, P.; Zhang, W.; Dunn, M. L.; Qi, H. J. RSC Adv. 2014, 4, 48682–48690. (138) Yu, K.; Taynton, P.; Zhang, W.; Dunn, M. L.; Qi, H. J. RSC Adv. 2014, 4, 10108–10117. (139) Obadia, M. M.; Mudraboyina, B. P.; Serghei, A.; Montarnal, D.; Drockenmuller, E. J. Am. Chem. Soc. 2015, 137, 6078–6083. (140) Lei, Z. Q.; Xiang, H. P.; Yuan, Y. J.; Rong, M. Z.; Zhang, M. Q. Chem. Mater. 2014, 26, 2038–2046. (141) Imbernon, L.; Norvez, S.; Leibler, L. Macromolecules 2016, 49, 2172–2178. (142) Kulshrestha, A. S.; Gao, W.; Gross, R. A. Macromolecules 2005, 38, 3193–3204. (143) Jansen, M. A. G.; Goossens, J. G. P.; Wit, G. De; Bailly, C.; Schick, C.; Koning, C. E. Macromolecules 2005, 38, 10658–10666. (144) Konishi, T.; Sakatsuji, W.; Fukao, K.; Miyamoto, Y. Macromolecules 2016, 49, 2272–

132

Bibliography

2280. (145) Jansen, M. A. G.; Goossens, J. G. P.; De Wit, G.; Bailly, C.; Koning, C. E. Macromolecules 2005, 38, 2659–2664. (146) Kim, J.; Nichols, M. E.; Robertson, R. E. J. Polym. Sci. Part B Polym. Phys. 1994, 32, 887–899. (147) Pyda, M.; Nowak‐Pyda, E.; Mays, J.; Wunderlich, B. J. Polym. Sci. Part B Polym. Phys. 2004, 42, 4401–4411. (148) Wunderlich, B. Prog. Polym.Sci. 2003, 28, 383–450. (149) Chen, X.Q., Xu, J. J., Lu, H.B., Yang, Y. L. J. Polym. Sci. Polym. Phys. 2006, 44, 2112– 2121. (150) Chambon, F.; Petrovic, Z. S.; Macknight, W. J.; Winter, H. H. Macromolecules 1986, 19, 2146–2149. (151) Winter, H. H. and Chambon, F. J. Rheol. 1986, 30, 367–382. (152) Adzima, B. J.; Aguirre, H. A.; Kloxin, C. J.; Scott, T. F.; Bowman, C. N. Macromolecules 2008, 41, 9112–9117. (153) Liu, F.; Li, F.; Deng, G.; Chen, Y.; Zhang, B.; Zhang, J.; Liu, C. Y. Macromolecules 2012, 45, 1636–1645. (154) Roberts, M. C.; Hanson, M. C.; Massey, A. P.; Karren, E. A.; Kiser, P. F. Adv. Mater. 2007, 19, 2503–2507. (155) Ghijsels, A.; Ente, J. J. S. M. and Raadsen, J. Int. Polym. Process. 1990, 5, 284–286. (156) Jansen, M. A. G.; Wu, L. H.; Goossens, J. G. P.; Wit, G. D. E.; Bailly, C.; Koning, C. E.; Portale, G. J. Polym. Sci., Polym. Chem. 2008, 46, 1203–1217. (157) Jansen, M. A. G.; Wu, L. H.; Goossens, J. G. P.; de Wit, G.; Bailly, C.; Koning, C. E. J. Polym. Sci., Polym. Chem. 2007, 45, 882–899. (158) Jansen, M. A. G.; Goossens, J. G. P.; Wit, G. De; Bailly, C.; Koning, C. E. Anal. Chim. Acta 2006, 557, 19–30. (159) Sablong, R.; Duchateau, R.; Koning, C. E.; De Wit, G.; Van Es, D.; Koelewijn, R.; van Haveren, J. Biomacromolecules 2008, 9, 3090–3097. (160) Sablong, R.; Duchateau, R.; Koning, C. E.; Pospiech, D.; Korwitz, A.; Komber, H.; Starke, S.; Häußler, L.; Jehnichen, D.; Auf Der Landwehr, M. Polym. Degrad. Stab. 2011, 96, 334–341. (161) Lavilla, C.; Gubbels, E.; Martínez de Ilarduya, A.; Noordover, B. A. J.; Koning, C. E. Macromolecules 2013, 46, 4335−4345. (162) Lavilla, C.; Gubbels, E.; Alla, A.; Martínez de Ilarduya, A.; Noordover, B. A. J.; Koning, C. E.; Muñoz‐Guerra, S. Green Chem. 2013, 16, 1789–1798. (163) Gubbels, E.; Jasinska‐walc, L.; Merino, D. H.; Goossens, H.; Koning, C. Macromolecules 2013, 46, 3975–3984.

133

Bibliography

(164) Cakir, S.; Nieuwenhuizen, M.; Janssen, P. G. A.; Rulkens, R.; Koning, C. E. Polymer 2012, 53, 5242–5250. (165) Jeyakumar, A.; Goossens, H.; Noordover, B.; Prusty, M.; Scheibitz, M.; Koning, C. J. Polym. Sci. Part A Polym. Chem. 2013, 51, 5118–5129. (166) Demongeot, A.; Mougnier, S.‐J.; Okada, S.; Soulié‐Ziakovic, C.; Tournilhac, F. Polym. Chem. 2016, 7, 4486–4493. (167) Arnebold, A.; Wellmann, S.; Hartwig, A. Polymer 2016, 91, 14–23. (168) Yang, H.; Yu, K.; Mu, X.; Wei, Y.; Guo, Y.; Qi, H. J.; Qi, H. J.; Wudl, F. RSC Adv. 2016, 6, 22476–22487. (169) Chen, Q.; Yu, X.; Pei, Z.; Yang, Y.; Wei, Y.; Ji, Y. Chem. Sci. 2017, 8, 724–733. (170) Liu, W.; Schmidt, D. F.; Reynaud, E. Ind. Eng. Chem. Res. 2017, 56, 2667–2672. (171) Cowie, J. M. G.; Arrighi, V. Polymers: chemistry and physics of modern materials; 3rd Edition, Blackie academic and professional, and imprint of Chapman and Hallbishopbriggs: Glasgow, 1991. (172) Ferry, J. D. Viscoelastic properties of polymers; John Wiley & Sons: Chichester, 1980. (173) Gubbels, E.; Jasinska‐Walc, L.; Hermida‐Merino, D.; Goossens, H.; Koning, C. Macromolecules 2013, 46, 3975–3984. (174) Pritchard, R. H.; Redmann, A.‐L.; Pei, Z.; Ji, Y.; Terentjev, E. M. Polymer 2016, 95, 45–51. (175) Gubbels, E.; Jasinska‐Walc, L.; Hermida‐Merino, D.; Hansen, M. R.; Noordover, B.; Spoelstra, A.; Goossens, H.; Koning, C. Polymer 2014, 55, 3801–3810. (176) Fakirov, S; Sarkissova, M.; Denchev, Z. Macromol. Chem. Phys. 1996, 197, 2889– 2907. (177) Fakirov, S.; Sarkissova, M.; Denchev, Z. Macromol. Chem. Phys. 1996, 197, 2837– 2867. (178) Denchev, Z.; Sarkissova, M.; Fakirov, S.; Yilmaz, F. Macromol. Chem. Phys. 1996, 197, 2869–2887. (179) Fernandez‐Berridi, M. J.; Iruin, J.; Maiza, I. Polymer 1995, 36, 1357–1361. (180) Montaudo, G.; Montaudo, M. S.; Scamporrino, E.; Vitalini, D. Macromolecules 1992, 25, 5099–5107. (181) Devaux, J.; Godard, P.; Mercier, J. P.; Touillaux, R.; Dereppe, J. M. J. Polym. Sci. Polym. Phys. Ed. 1982, 20, 1881–1894. (182) Fakirov, S.; Sarkissova, M.; Denchev, Z. Macromol. Chem. Phys. 1996, 197, 2837– 2867. (183) Eguiazabal, J. I.; Fernandez‐Berridi, M. I.; Iruin, J. J.; Maiza, I. J. Appl. Polym. Sci. 1996, 59, 329–337. (184) Kim, J. H.; Lyoo, W. S.; Ha, W. S. J. Appl. Polym. Sci. 2001, 82, 159–168.

134

Bibliography

(185) Tanguy, T.; Bonnet, J. P.; Kudryavtsev, Y. V.; Tessier, M.; Fradet, A. Macromolecules 2010, 43, 9318–9327. (186) Ilarduya, A. M. De; Muñoz‐guerra, S. Macromol. Chem. Phys. 2014, 215, 2138–2160. (187) Zhou, Y.; Groote, R.; Goossens, J. G. P.; Sijbesma, R. P.; Heuts, J. P. A. To be submitted 2017. (188) Fakirov, S. Transesterifications in condensation polymers; Wiley‐VCH: Verlag GmbH: Weinheim, 1999. (189) Marchese, P.; Celli, A.; Fiorini, M. J. Polym. Sci. Part B Polym. Phys. 2004, 42, 2821– 2832. (190) Kriegel, R. M.; Collard, D. M.; Liotta, C. L.; Schiraldi, D. A. Macromolecules 1998, 31, 2475–2479. (191) Hu, W.‐G.; Schmidt‐Rohr, K. Acta polym. 1999, 50, 271–285. (192) Pepels, M. P. F.; Govaert, L. E.; Duchateau, R. Macromolecules 2015, 48, 5845–5854. (193) Yeh, J. T.; Runt, J. J. Polym. Sci. Part B Polym. Phys. 1989, 27, 1543–1550. (194) Yasuniwa, M.; Murakami, T.; Ushio, M. J. Polym. Sci. Part B Polym. Phys. 1999, 37, 2420–2429. (195) Hsiao, B. S.; Wang, Z.‐G.; Yeh, G. F.; Gao, Y.; Sheth, C. K. Polymer 1999, 40, 3515– 3523. (196) Lorenzo, A. T.; Arnal, M. L.; Sánchez, J. J.; Müller, A. J. J. Polym. Sci. Part B Polym. Phys. 2006, 44, 1738–1750. (197) Reid, B. O.; Vadlamudi, M.; Mamun, A.; Janani, H.; Gao, H. H.; Hu, W. B.; Alamo, R. G. Macromolecules 2013, 46, 6485–6497. (198) Mamun, A.; Chen, X.; Alamo, G. Macromolecules 2014, 47, 7958–7970. (199) Chen, X.; Mamun, A.; Alamo, R. G. Macromol. Chem. Phys. 2015, 216, 1220–1226. (200) Pompe, C.; Häußler, L.; Winter, W. J. Polym. Sci. Part A Polym. Phys. 1996, 34, 211– 219. (201) Zhou, Y.; Groote, R.; Goossens, Johannes G. P Lugger, J. A. M.; Sijbesma, R. P.; Peters, G. W. M.; Heuts, J. P. A. In preparation. (202) Posner, G. H.; Oda, M. Tetrahedron Lett. 1981, 22, 5003–5006. (203) Otera, J. Chem. Rev. 1993, 93, 1449–1470. (204) Runt, J.; Miley, D. M.; Zhang, X.; Gallagher, K. P.; McFeaters, K.; Fishburn, J. Macromolecules 1992, 25, 1929–1934. (205) Pyda, M.; Nowak‐Pyda, E.; Heeg, J.; Huth, H.; Minakov, A. A.; Lorenzo, M. L.; Schick, C.; Wunderlich, B. J. Polym. Sci. Part B Polym. Phys. 2006, 44, 1364–1377. (206) Frich, D.; Hall, A.; Economy, J. Macromol. Chem. Phys. 1998, 199, 913–921. (207) Yang, H.; Yu, K.; Mu, X.; Wei, Y.; Guo, Y.; Qi, H. J. RSC Adv. 2016, 6, 22476–22487. (208) Yu, K.; Shi, Q.; Li, H.; Jabour, J.; Yang, H.; Dunn, M. L.; Wang, T.; Qi, H. J. J. Mech.

135

Bibliography

Phys. Solids 2016, 94, 1–17. (209) Abt, T.; Sánchez‐Soto, M.; Martínez de Ilarduya, A. Eur. Polym. J. 2012, 48, 163–171. (210) Zhang, Y.; Broekhuis, A. a.; Picchioni, F. Macromolecules 2009, 42, 1906–1912. (211) Zhang, W.; Li, J.; Shang, Y.; Li, H.; Jiang, S.; An, L. Mater. Des. 2017, 129, 143–150. (212) Zhou, Y.; Pan, J.; Goossens, J. G. P.; Sijbesma, R. P.; Heuts, J. P. A. In preparation (213) Franck, A. Understanding Rheology of Thermoplastic Polymers http://www.tainstruments.com/pdf/literature/AAN013_V_1_U_Thermoplast.pdf. (214) Song, G.; Zhao, Z.; Peng, X.; He, C.; Weiss, R. A.; Wang, H. Macromolecules 2016, 49, 8265–8273. (215) Ludwig, H.‐J.; Eyerer, P. Polym. Eng. Sci. 1988, 28, 143–146. (216) Nichols, M. E.; Robertson, R. E. J. Polym. Sci. Part B Polym. Phys. 1992, 30, 755–768. (217) Kim, H. G.; Robertson, R. E. J. Polym. Sci B Polym. Phys. 1998, 36, 1757–1767. (218) di Lorenzo, M. L. L.; Righetti, M. C. Polym. Eng. Sci. 2003, 43, 1889–1894. (219) Righetti, M. C.; Di Lorenzo, M. L.; Angiuli, M.; Tombari, E. Macromolecules 2004, 37, 9027–9033. (220) Konishi, T.; Miyamoto, Y. Macromolecules 2010, 43, 375–383. (221) Mathot, V.; Pyda, M.; Pijpers, T.; Vanden Poel, G.; Van De Kerkhof, E.; Van Herwaarden, S.; Van Herwaarden, F.; Leenaers, A. Thermochim. Acta 2011, 522, 36– 45. (222) Van Herwaarden, S.; Iervolino, E.; Van Herwaarden, F.; Wijffels, T.; Leenaers, A.; Mathot, V. Thermochim. Acta 2011, 522, 46–52. (223) Iervolino, E.; Van Herwaarden, A. W.; Van Herwaarden, F. G.; Van De Kerkhof, E.; Van Grinsven, P. P. W.; Leenaers, A. C. H. I.; Mathot, V. B. F.; Sarro, P. M. Thermochim. Acta 2011, 522, 53–59. (224) Toda, A.; Androsch, R.; Schick, C. Polymer 2016, 91, 239–263. (225) Androsch, R.; Schick, C. Adv. Polym. Sci. 2017, 276, 257–288. (226) Wurm, A.; Herrmann, A.; Cornelius, M.; Zhuravlev, E.; Pospiech, D.; Nicula, R.; Schick, C. Macromol. Mater. Eng. 2015, 300, 637–649. (227) Androsch, R.; Rhoades, A. M.; Stolte, I.; Schick, C. Eur. Polym. J. 2015, 66, 180–189. (228) Furushima, Y.; Kumazawa, S.; Umetsu, H.; Toda, A.; Zhuravlev, E.; Schick, C. J. Appl. Polym. Sci. 2017, 133, 44739–44750. (229) Mileva, D.; Androsch, R.; Zhuravlev, E.; Schick, C. Polymer 2012, 53, 3994–4001. (230) Van Drongelen, M.; Meijer‐Vissers, T.; Cavallo, D.; Portale, G.; Poel, G. Vanden; Androsch, R. Thermochim. Acta 2013, 563, 33–37. (231) Zia, Q.; Androsch, R.; Radusch, H. J.; Piccarolo, S. Polymer 2006, 47, 8163–8172. (232) Zia, Q.; Radusch, H. J.; Androsch, R. Polymer 2007, 48, 3504–3511. (233) Yamamoto, T. Polymer 2009, 50, 1975–1985.

136

Bibliography

(234) Yamamoto, T. J. Chem. Phys. 2010, 129, 184903. (235) Hobbs, S. .; Pratt, C. Polymer 1975, 16, 462–464. (236) Chang, E. P.; Kirsten, R. O.; Slagowski, E. L. Polym. Eng. Sci. 1978, 18, 932–936. (237) Schawe, J. E. K. J. Therm. Anal. Calorim. 2014, 116, 1165–1173. (238) Jariyavidyanont, K.; Androsch, R.; Schick, C. Polymer 2017, 124, 274–283. (239) Santos, J. M. R. C. A.; Guthrie, J. T. J. Mater. Chem. 2006, 16, 237–245. (240) Delimoy, D. Polymer 1995, 36, 3255–3266. (241) Alsleben, M.; Schick, C. Thermochim. Acta 1994, 238, 203–227. (242) Supaphol, P.; Dangseeyun, N.; Srimoaon, P. Polym. Test. 2004, 23, 175–185. (243) Zhang, L.; Hong, Y. R.; Zhang, T. S.; Li, C. Z. J. Macromol. Sci. Part B 2010, 49, 514– 527. (244) Toda, A.; Hikosaka, M.; Yamada, K. Polymer 2002, 43, 1667–1679. (245) Stanton, M. G.; Gagne, M. R. J. Am. Chem. Soc. 1997, 119, 5075–5076. (246) Fakhreddine, Y. A.; Zoller, P. J. Polym. Sci. Part A Polym. Phys. 1991, 29, 1141–1146. (247) Sato, Y.; Inohara, K.; Takishima, S.; Masuoka, H.; Imaizumi, M.; Yamamoto, H.; Takasugi, M. Polym. Eng. Sci. 2000, 40, 2602–2609. (248) Van Der Beek, M. H. E.; Peters, G. W. M.; Meijer, H. E. H. Macromolecules 2006, 39, 1805–1814. (249) Van Der Beek, M. H. E.; Peters, G. W. M.; Meijer, H. E. H. Macromolecules 2006, 39, 9278–9284. (250) https://www.cumapol.nl/ssp/. (251) http://www.polymetrix.com/process‐technology/solid‐state‐polycondensation‐ ssp/. (252) Kumar, S. K.; Benicewicz, B. C.; Vaia, R. A.; Winey, K. I. Macromolecules 2017, 50, 714–731. (253) Paul, D. R.; Robeson, L. M. Polymer 2008, 49, 3187–3204. (254) Li, Q.; Zhou, J.; Vatankhah‐Varnoosfaderani, M.; Nykypanchuk, D.; Gang, O.; Sheiko, S. S. Macromolecules 2016, 49, 1383–1391. (255) Zhou, J.; Sheiko, S. S. J. Polym. Sci., Part B Polym. Phys. 2016, 54, 1365–1380.

137

Bibliography

138

Appendix 1 for Chapter 2

Comparison three different types of synthetic route

In order to optimize the preparation method, we compared the built‐in glycerol concentrations of three different preparation strategies (Figure A1.1). All the preparation routes contain two steps, which differ in terms of used temperatures. The physical mixture containing 14.1 mol% glycerol and 2 mol% Zn2+ was used to compare these three different preparation routes. Identical reaction times (24 h) during step 1(160 °C) were employed. An initial drop in Mn due to chain scission was expected during step 1, caused by the alcoholysis of the PBT chains by the free hydroxyl groups from the glycerol with the help of Zn2+.

Scheme A1.1. Preparation method screening to minimize the evaporation of glycerol during solid‐ state (co)polymerization.

Figure A1.1. The molecular weight distribution of the copolyesters prepared via three different synthetic route.

139

Appendix

‐1 Route 2 vs 1 (with (2) or without (1) 0.5 Lmin N2 flow at 160 °C). Just before the gel point, the Mn for routes 2 and 1 are 27.9 kg/mol and 12.4 kg/mol, and the minimum glycerol contents are 3 mol% and 9 mol%, respectively (the calculation method is discussed in Figure 2).

Route 3 vs 1(with (3) or without (1) pressure at 160 °C). Just before the gel point, the Mn for route 3 and 1 are 6.8 kg/mol and 12.4 kg/mol, and the minimum glycerol contents are 13 mol% and 9 mol%, respectively. Therefore, the best method is route 3, which includes a two‐step (co)polymerization strategy. First, a prepolymerization step was performed at 160 °C in a closed glass vial which was pressurized with inert gas to avoid the evaporation of glycerol. After the total incorporation of the glycerol (24 h for 13 mol% glycerol and 2 mol% Zn2+), the mixture was ‐1 transferred to an SSP reactor, and the reaction was continued at 180 °C with 0.5 Lmin N2 flow.

Table A1.1. Overview of the PBT/glycerol copolyesters prepared by solid‐state (co)polymerizaiton (SSP) before gelation via different synthetic route.

a a Initial glycerol glycerol contents b b tssp < tgel Zn(acac)2 Mn Mw Entry composition before gelation /h /mol% (kg/mol) (kg/mol) (mol%) (mol%)

1C9 1 2 14.1 9 12.4 28.1

2C3 1 2 14.1 3 27.9 66.4

3C13 1 2 14.1 13 6.8 17.0 aThe compositions were determined using 1H‐NMR spectroscopy and ratio between the glycerol/ BD b 1 2 units expressed in mol %. The molecular weight of the PBT samples measured with HFiP‐SEC. C9 , ‐1 2 2 160 °C 24h without N2 flow, then heating up to 180 °C with 0.5 Lmin N2 flow. C3 , 160 °C, 24h with ‐1 ‐1 3 2 0.5 Lmin N2 flow, then heating up to 180 °C with 0.5 Lmin N2 flow. C13 , 160 °C 24h at closed vial, ‐1 SSP was performed at 180 °C with 0.5 Lmin N2 flow.

140

Appendix

1.4 2+ 3.0 2 mol% Zn 24 h, C2 A 20 1.2 24 h, C4 0 h 18 1.0 24 h, C7 16 2.5 )

14 n 0.8 24 h, C18 12 /M 2.0 w

0.6 10 wlog M (kg/mol)

24 h, C13 n

8 PDI (M

0.4 M 1.5 6 0.2 4 (B) 0.0 2 1.0 2.53.03.54.04.55.05.5 0 2 4 6 8 10 12 14 16 18 20 log M Glycerol content (mol%)

2 Figure A1.2. (A) The molecular weight distribution of Cx (x= 2, 4, 7, 13, 18 mol%) after prepolymerization at 160 °C for 24h. (B) Overview of the PBT/glycerol copolyesters after prepolymerization at 160 °C under pressure.

Branched Poly (glycerol terephthalate)

In a 100 mL dried two‐way round‐bottom flask that is connected to a condenser, glycerol (3.118 g, 34 mmol) was dissolved in 60 mL of anhydrous dichloromethane under mild heating condition (35 °C). Triethylamine (7 g, 70 mmol) was added into the mixture.

Terephthaloyl chloride (6.882 g, 34 mmol) in 40 ml anhydrous dichloromethane (CH2Cl2) was added into the flask using a drop funnel within 30 min while the mixture was stirring. After injection, the reaction was left for an additional hour. The oligomerization was terminated by filtering the triethylamine hydrochloride salt and then extracting excess of solvent followed by precipitation with methanol in an ice bath. Precipitates were filtered and then dissolved in 40 mL of dichloromethane and dried by adding anhydrous magnesium sulfate. The product solution was filtered and solvent was removed by rotary evaporator followed by drying under vacuum at room temperature overnight.

141

Appendix

Figure A1.3. The molecular weight distribution, 1H‐ and 13C‐NMR spectrum of the hyperbranched poly(glycerol‐co‐terephthalate) copolyester recorded in CDCl3/TFA‐d (4:1 v/v) mixture at room temperature.

The branched poly(glycerol terephthalate) with Mn = 4.4 kg/mol and Đ = 2.1, was synthesized to resolve the signals from glycerol in the copolyesters, and it is not soluble in

CDCl3 and CDCl3/HFiP‐d2 (v/v  95:5), but soluble in DMSO‐d6 , HFiP and TFA‐d/CDCl3 (v/v  1:4).

142

Appendix

Figure A1.4. Expanded regions (3.75‐6.25 ppm) from 1H NMR spectra of the copolyester before and 2 after prepolymerization recorded in CDCl3/TFA‐d (5:1 v/v) mixture at room temperature. (1) C14.1 ‐ physical mixture, (2) branched poly (glycerol terephthalate), (3) C13 (tssp = 1 h), (4) C4 (tssp = 7 h), (5)

C2 (tssp = 7 h).

Glycerol can be present in four forms. The glycerol protons CH2‐O(C=O), f, and CH‐ O(C=O), h’, appear as broad signals at 4.72‐4.95 ppm and 6 ppm, respectively. These assignments are consistent with the work of Gross, et al.142 The molar content of glycerol is calculated with equation 1: % 100% (1)

The compositions before gelation are 13.2 (tssp = 1), 4 (tssp = 7 h), 2.2 (tssp = 7 h) mol% for the physical mixture with 14.3, 6.1 and 2.2 mol% glycerol loadings, respectively.

143

Appendix

24 h = t0

t1= 0.5 h

t2 = 1 h

t3 = 3 h

t4 = 7 h t at 160 C.

Heat flow, exo up (a.u.) 0 t ~ t at 180 C. B 1 5 t5 = 24 h 120 140 160 180 200 220 240 Temperature (C) C

Heat flow, exo up (w/g) t1

t0 120 140 160 180 200 Temperature (C) Figure A1.5. (A) DSC thermogram of neat PBT treated with HFiP. DSC first heating (B) and cooling (C) runs of C13 copolyesters as a function of tssp obtained at a heating rate of 10 °C/min from ‐50 to 250

°C. (D) First heating run reversing cp traces and morphology development during SSP of the synthesized C13 copolyesters.

The Δcp value to calculate αmobile by eq 1. χheating was determined from ΔHmelting using eq 2 whereas

αrigid was determined by eq 3. The first heating run was used to calculate αmobile and ΔHmelting and the results are shown in Figure S5D. ∆ (1) ∆ ∆ (2) ∆

1 (3)

144

Appendix

Insolubility after compression molding

The materials were compression molded at 250 °C and 100 bar for 25‐30 minutes in a Collin Press 300G subsequently cooled with water. The circular sample was dried at 120 °C under vacuum for 4‐6 h before measurements.

Scheme A1.2. Photograph of compression molding processing, the powder after SSP was processed into circular rheological sample.

For all the materials prepared by SSP, there is no antioxidant added and this coloration is probably caused by the fast transesterification in the presence of Zn(acac)2 during the melt processing of PBT vitrimers. Thermal stability before and after compression molding

Figure A1.6. TGA curves showing the thermal stability of PBT/glycerol‐based vitrimer before and after compression molding. 145

Appendix

Figure A1.7. (A) Photograph of PBT/glycerol‐based vitrimer swelling at HFiP and gel fraction as a function of glycerol feeding ratio (B).

Insolubility experiments were carried out with samples immersed in the 1, 1, 1, 3, 3, 3‐ Hexafluoro‐2‐propanol at room temperature for 24 h. The gel fraction was calculated with the equation: gel fraction% 100%.

Figure A1.8. (A) Storage modulus, (B) loss modulus, (C) Tg for PBT/glycerol‐based vitrimers with different glycerol contents.

146

Appendix

106 7 (B) (A) 10

106 5

10 7 mol% 

5 Pas 10 230 C  4 104 18 mol% 10 240 C 250 C 3

G'(open), G''(closed) [Pa] 10 2+ 2 mol% Zn 240 C 13 mol% glycerol+2 mol%Zn2+ 103 10‐2 10‐1 100 101 102 10‐2 10‐1 100 101 102 rad/s [rad/s] Figure A1.9. (A) Storage modulus versus ang. frequency for PBT/glycerol‐based vitrimers catalyzed by 2 mol% Zn2+ at 240 °C by 25 mm plate‐plate geometry with 1% strain applied.(B) Complex viscosity versus angular frequency for C13 at different temperature by 25 mm plate‐plate geometry with 1% strain applied.

Figure A1.10. Normalized stress relaxation curves at different temperatures: (A) C7, and (B) C18. (C) Normalized stress relaxation curves at 250 °C for the samples: C2, C4 and C13.

147

Appendix

105

4 10

103

2  10 220 C  230 C G' (closed), G'' (open) (Pa) C2 240 C 101 ‐2 ‐1 0 1 2 3 10 10 10 10 10 10 rad/s

Figure A1.11. Storage (open symbols) and loss (filled symbols) moduli versus ang. frequency for C2 at different temperature by 25 mm plate‐plate geometry with 1% strain applied.

148

Appendix

Appendix 2 for Chapter 3

1.4 48 h@180C, 0 mol%, (C 0) 1.4 13 0 h C D 0.05 mol% 0.02 mol% 1.2 0.05 48 h@180C 96 h, 0.05 mol% 1.2 (C ) 3 0.02 0.05 (C ) (C ) 1 1.0 13 1.0 0.1 mol% 48 h, 0.2 mol% (C 0.1) 0.8 0.2 0.8 6 (C13 )

24 h, 2 mol% 0.6 0.6 2

w (log M) ( )

C w (log M) 13 0 h 0.4 0.4 The same 0.2 0.2 r value 0.0 0.0 48 h@160C 2.53.03.54.04.55.05.5 2.5 3.0 3.5 4.0 4.5 5.0 5.5 log M log M Figure A2.1. Photographs of the high pressure (A) and SSP reactor (B). (C) The molecular weight distribution of material set I, 2, 0.5, and 0.05 mol% Zn2+ after prepolymerization at 160 °C and 180 °C 2+ for 0 mol% Zn with different times. (D) Material set II, the same r (r  0.015) via tuning Xgly/XZn.

Here, all the materials are synthesized via a two‐step polymerization strategy as described in the Chapter 2. First, a prepolymerization step was performed at 160 °C in a high pressure reactor (80 gram scale)(Figure A2.1A), which was pressurized with inert gas to avoid the evaporation of glycerol. The polymerization conditions were slightly changed due to the variation of XZn because it has a strong influence on the polymerization kinetics. For instance, in order to obtain the material with the same molecular weight after prepolymerization step, the experiment is performed at 160 °C for 24, 48, 96 h for sample 2 2+ 0.2 2+ 0.05 2+ C13 (2 mol% Zn ), C13 (0.2 mol% Zn ) and C13 (0.05 mol% Zn ). One exception is 0 2+ sample C13 (0 mol% Zn ), which is prepolymerized at 180 °C for 48 h. After the total incorporation of the glycerol, the mixture was transferred to an SSP reactor ( 5‐ 40 gram 149

Appendix

‐1 scale) (Figure A2.1B). The reaction was continued at 180 °C with a N2 flow of 0.5 L min . The molecular weight after prepolymerization and composition before gelation of the prepared copolyesters are summarized in Table A2.1.

Table A2.1. Overview of the molecular weight of the copolyesters after prepolymerization.

bMolecular weight after r = XZn/ Xgly prepolymerization Sample Sample set name aBefore Mn Mw aInitial gelation (kg/mol) (kg/mol)

2 C13 2/14 2/13.2 3.1 9.3

I. Varying r: 0.2 C13 0.2/14 0.2/13.0 3.8 8.1

Xgly constant, varying 0.05 XZn C13 0.05/14 0.05/13.0 3.2 9.6

0 C13 0/14 0/13.0 6.5 15.0

0.2 C13 0.2/14 0.2/13.0 3.8 8.1

0.1 II. Constant r: C6 0.1/6.5 0.1/6.2 6.8 13.6

varying Xgly and XZn 0.2 C3 0.05/3.5 0.05/3.0 10.8 21.7

0.02 C1 0.02/1.6 0.02/1.2 17.0 31.9 aThe glycerol contents were determined using 1H NMR spectroscopy and ratio between the b glycerol/BD units expressed in mol %, CDCl3:d‐TFA=4:1(v:v) was used as a solvent. The molecular weight of the PBT/glycerol‐based copolyester samples measured with HFiP‐SEC.

150

Appendix

106

105

104

0.02 mol% 103 0.05 mol%

G'(open), G''(closed) [Pa] 0.1 mol% (A) 250 C 0.2 mol% 102 101 102 103 104 Time [s]

Figure A2.2. Time dependence of the storage modulus G (open symbols) and loss modulus G (filled symbols) at 250 °C for compression‐molded material set II: the same r (r  0.015) via tuning Xzn/ Xgly.

107

106

105

104

103

102 Relaxation modulus

101 250 C 100  10‐3 10‐2 10‐1 100 101 102 Time [s] Figure A2.3. Non‐treated stress relaxation curve of neat PBT at 250 °C with 1% strain.

151

Appendix

Table A2.2. Linear fit results of activation energy (Ea) and pre‐exponential factor (A) for PBT/glycerol‐ based vitrimers.

r value and Glycerol Activation energy pre‐exponential Entry ‐1 content/mol% /kJ/mol factor/s

0 (13.0) 172  3 9.78 × 1013

0.004 (13.0) 202  8 1.27 × 1017 Different r value 0.015 (13.0) 167  8 6.98 × 1013

0.15 (13.0) 152  2 6.36 × 1012

0.015 (13) 167  8 6.98 × 1013

The same r value 0.015 (6.2) 137  3 4.66 × 1010

0.015 (3.2) 167  8 1.01 × 1014

0 2 (B) C13 4 2 0.2 C ‐1 C13 13 3 0.05 C 0.2 C13 13

‐2 0 0.05 C13 2 C13

0 ‐3 Neat PBT C13 1 T

Heat flow, exo up [a.u.] c T Heat flow, exo up [a.u.] (A) m Neat PBT ‐4 ‐50 0 50 100 150 200 250 0 ‐50 0 50 100 150 200 250 Temperature [C] Temperature [C] 0 0.02 (D) C1 3 0.05 0.02 ‐1 C3 C1 C 0.1 6 2 C 0.05 ‐2 0.2 3 C13 0.1 C6 1 ‐3 Neat PBT 0.2 C13 Heat flow, exo up [a.u.] Heat flow, exo up [a.u.] T Tc (C) m Neat PBT ‐4 0 ‐50 0 50 100 150 200 250 ‐50 0 50 100 150 200 250 Temperature [C] Temperature [C] Figure A2.4. DSC second heating (A) and (C), first cooling (B) and (D) run of two sets of PBT vitrimers obtained at a heating rate of 10 °Cmin‐1 from ‐50 to 250 °C.

152

Appendix

Figure A2.5. (A) Scheme of a creep response of polymers decribed by a four-parameter (Burger’s) model with spring and dashpot. (B) Creep‐recovery experiments of neat PBT at 2MPa stress at different temperatures.

153

Appendix

154

Appendix 3 for Chapter 4

103

102 25 C 80 C ' [MPa] 1 E 10 120 C 180 C 100 200 C

0 50 100 150 200 250 Temperature [C] Figure A3.1. DMTA curves of compression‐molded neat PBT undergo different thermal annealing process.

1.8 Neat PBT 1.6 Cycling at 250C 1.4 1.2 1.0 0.8 0.6

Heat flow, exo up [w/g] 0.4 0.2 150 160 170 180 190 200 210

Temperature [C] Figure A3.2. DSC thermograms for neat PBT as a function of cycling number at 250 °C with 5 min equilibrium time and subsequent heating at 10 °Cmin‐1 for checking the final thermal properties.

155

Appendix

Time (min) 0 5 10 15 20 25 30 35 40 45 50 55 60 175

170 Protocol l

165 Protocol ll C 

/ 160 c, peak

T T  23.9 C 155 c,peak

150 Sample 3 145 024681012

Cycle number Figure A3.3. Comparison between protocol I and II for sample 3.

260 5 PBT 250 4 212 C 240

3 C) 237.7 C 210 C  (

m 230 T 2 208 C 220 1 206 C Heat flow, endo up (w/g) 210 0 205 210 215 220 225 230 235 240 245 250 210 220 230 240 250 260 Temperature (C) T (C) c Figure A3.4. Hoffman‐weeks plot of neat PBT.

156

Appendix 4 for Chapter 5

Figure A4.1. The expanded regions (01.7 ppm) from 1HNMR spectra of C8(TMP) physical mixture and

C8(TMP) at tssp = 1h recorded in CDCl3/TFA‐d (4:1 v/v) mixture at room.

Figure A4.2. Photo of gelation in HFiP for the cured PBT vitrimers.

157

Appendix

Figure A4.3. (A) Time dependence of the storage modulus G’ (open symbols) and viscous modulus G’’ (filled symbols) at 250 °C for C8(GLY)(triangle) and C8(TMP)( circle).Photo of gelation in HFiP for the cured PBT vitrimers. (B) Angular frequency dependence of the storage modulus G’ (open symbols) and viscous modulus G’’ (filled symbols) at 250 °C for PBT/triol‐based vitrimers.

Figure A4.4. The temperature was controlled by the environmental test chamber (ETC) of the AR G2 rheometer). (A) Assembly of the two rectangular samples, compressed together during the welding process.

158

Appendix

Figure A4.5. GC analysis retention time for each compound in model reaction (1) and (2).

Figure A4.6. Calibration curve for butyl benzoate. 159

Appendix

0.18 2.8 (A) Butyl benzoate 120 C 0.15 2.6 160 C 180C 2.4

0 0.12 2.2 140 C 0.09 1/c‐1/c 2.0 180 C 0.06 160C 1.8

 Concentration [mol/L] 0.03 140 C 120C 1.6 (B) 0.00 0 5 10 15 20 25 30 35 40 0 500 1000 1500 2000 Time [min] Time [sec] Figure A4.7. (A) the linear fit of against time for model reaction (1), (B) Second order linear fit for model reaction (1) ((■) experimental); (‐) fit.

160

Appendix 5 for Chapter 6

4 C 40 0 C 4.0 36 0 3.6 3 32 24 h 28 3.2 6 h 24 2

(kg/mol) 2.8 Ð 3 h n 20 M 16 2.4 1 h dwt/d(logM) 1 12 2.0 0.5 h 8 C 1.6 0 0 h 4 p 9 3.03.54.04.55.05.56.0 0 5 10 15 20 25 30 Log M (g/mol) Reaction time (hour) (A) (B) a' a' c' c b' b' 24h

C 7h 

180 2h

0.5h

6h a

C c  b b a 160

0h 5.65.24.84.44.03.63.2 ppm (C) (D)

Figure A5.1. A) The molecular weight distribution of C0 (closed symbol) and pC1 (open symbol) copolymer during the SSP as a function of tssp; b) development of Mn and Đ as a function of tssp; (c) the three forms of the BPO copolymers during the SSP reaction: free BPO monomer (F), BPO mono‐ester (M) and fully incorporated BPO di‐ester (D); (d) 1H NMR spectra of model compound (1) and copolymer pC9 before and after SSP reaction recorded in C2D2Cl4 at room temperature.

For the C0 physical mixture, the molecular weight increases as a function of reaction time; after tssp = 24 h, Mn becomes close to 40.9 kg/mol. During the prepolymerization at 160 °C, the amount of unreacted hydroxyl groups of the BPO is relatively high. Thus, chain scission is the dominated reaction, which leads to a large drop in molecular weight.

However, as shown in Figure A5.1, an increase in the molecular weight (Mn) of the formed copolymer can be observed during SSP at 180 °C, which is changed from 4.8 kg/mol (tssp = 6 h at 160 °C) to 9.2 kg/mol (tssp = 24 h at 180 °C). This increase is a result of polymer chain recombination by esterification and transesterification (polycondensation) reaction that takes place between the co‐reactive chain ends present in the copolymer with elimination 161

Appendix

of the condensation product 1, 4‐butanediol. The Đ increases from 2.2 (tssp) = 0 h to 2.9 (tssp = 24 h at 180 °C).

The chemical microstructure of the pC9 copolymer will have a blocky structure consisting of unmodified homopolymer PBT blocks (only BD‐T repeat units), which are predominantly present in the crystalline domains, and amorphous parts consisting of both BD‐T and BPO‐ T repeat units. The BPO has two hydroxyl end groups which both can react with PBT. 5, 5‐ Bis(phenylcarboxymethyl)‐2‐phenyl‐1,3‐dioxane (1) was synthesized as a model compound for the copolymers and for investigation of the hydrolysis of the benzal protecting group. Due to the acid‐sensitive nature of the protected repeat unit, 1, 1, 2, 2‐tetrachloroethane‐ d2 was used to acquire NMR spectra under neutral conditions. In common with the spectrum of the model diester 1, the exocyclic methylenes of these units appear as singlets (4.33 and 4.80 ppm) and the benzylic proton appears at 5.50 ppm, which is consistent with the successful incorporation of the benzal‐containing repeat units derived from BPO. The amount of BPO incorporated into the polymer by solid‐state (co)polymerization was consistently lower than the monomer feed ratio, owing to loss of volatile monomer under the used reaction conditions.

Scheme A5.1. The molecular structure before and after acid‐promoted debenzalation.

The benzal‐containing copolymers were deprotected by dissolving the polymers in TFA (6 mL.g‐1 of polymer) (Scheme A5.1) and stirring for several hours with the solution exposed to the atmosphere, and subsequent precipitation in methanol. The 1H NMR spectra of the polymers in 1, 1, 2, 2‐tetrachloroethane‐d2 showed that the diol is deprotected and that the benzaldehyde is completely removed from the polymer after washing it with hot methanol for three times (Figure A5.2(b)). Measurements of molecular weight before and after deprotection (Figure A5.2(a)) indicate that there is no significant decrease in molecular weight as a result of ester hydrolysis during this process.

162

Appendix

Mn (Kg/mol) Mw (Kg/mol) PDI protected polymer 41.9 91.0 2.17 deprotected polymer 40.8 84.9 2.08

Benzal pentaerythritol

024 161820222426283032 7.6 7.4 7.2 7.0 6.8 6.6 6.4 6.2 6.0 5.8 5.6 5.4 5.2 5.0 4.8 ppm Elution time / min (A) (B)

Figure A5.2. a) SEC traces, b) 1H NMR spectra of C1 before (black) and after (red) deprotection at TFA (ca. 6 ml mg‐1).

105 Cross‐linked 250C

linear Branching

4 10

 Gel point 270 C

G'(open), G''(closed) [Pa] (Protected) C1 + 0.2mol%Zn2+ 103 101 102 103 104 105 Time [s Scheme A5.2. (A) Melting ring set up for direct rheological analysis of the powder obtained from the SSP reactor. (B) Time dependence of the storage modulus G (open symbols) and loss modulus G (filled symbols) at 250 and 270 °C for pC1.

Figure A5.3. Insolubility test with pX1, deX1, pX2.4 and pX3.5 (from left to right) for 48 h at room temperature in 1, 1, 1, 3, 3, 3‐hexafluoroisopropanol. 163

Appendix

103

102

101

pC2.4 0 10 C2.4

Storage modulus [MPa] de

10‐1 0 30 60 90 120 150 180 210 240 270

Temperature C]

Figure A5.4. DMTA curves of pX2.4 and deX2.4.

1.0 1.0 230 C

0.8 0.8

0.6 230 C 0.6 0 0 290 C 270 C /G /G t t G G 0.4 0.4 37% 37% 240 C 0.2 270 C 0.2 C3.5 deC1, BPO 260 C 250 C p 250 C 0.0 0.0 ‐1 0 1 2 3 4 100 101 102 103 10 10 10 10 10 10 Time [s] Time [s] 1.0 230 C 0.8

0 0.6 270 C /G t G 0.4 37%

0.2 250 C deC3.5 0.0 100 101 102 103 104 Time [s]

Figure A5.5. Normalized stress relaxation at different temperatures for materials obtained via two different deprotection methods.

164

Appendix

a'1 a'2 b b 2 c 1

a' 1,1,2,2-TCE a'2 1 c b 2 b1

b a1 1 c a2 a1 a 2 b c 2 b 2 b1

8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0 ppm Figure A5.6. 1H NMR spectra of 5, 5‐Bis(hydroxymethyl)‐2‐phenyl‐1,3‐dioxane (BPO) and 5, 5‐

Bis(phenylcarboxymethyl)‐2‐phenyl‐1,3‐dioxane (1) recorded at 1,1,2,2‐tetrachloroethane‐d2

CH bending of the benzene ring ‐OH 3504 cm‐1 After ‐1 ‐C=Ost ‐C‐Oring1042 cm debenzalation

‐CO‐O

‐CH2

Before debenzalation

4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 600 ‐1 ‐1 Wavenumber [cm ] Wavenumber [cm ] Figure A5.7. ATR‐FTIR spectra of the model reaction between 5, 5‐Bis(phenylcarboxymethyl)‐2‐ phenyl‐1, 3‐dioxane and benzoic acid at 180 °C for 2 h.

165

Appendix

3 100 C) 

80 2 60

Weight (%) 40 Pure PBT 1 C1.2(BPO) 20 C2.4(BPO)

C3.5(BPO) Deriv. weight change (%/ 0 0 100 200 300 400 500 600 Temperature (C)

Figure A5.8. TGA curves of the linear PBT/BPO‐based copolyesters.

166

Appendix 6 for Chapter 7

Cooling at 4000 Cs‐1 Heating rate in Cs‐1 10 20 50 100 200 500 1000

2000

5000

(A) 0.5 wt% talc 0 30 60 90 120 150 180 210 240

Temperature [C] Figure A6.1. FSC curves of PBT/talc(0.5) , heat‐flow rate as a function of temperature, recorded on heating samples at different rates between 10 (top) and 5000 °Cs‐1 (bottom). The curves were shifted vertically for clarity.

167

Appendix

PBT/PC(5) blend Cooling rate in K s‐1 ‐1 2000 C1(GLY) Cooling rate in K s 2000

1000 1000

500 500 300 300 100 100

50 Heat flow [mW, exo up]

Heat flow [mW, exo up] 50

10 10 0 30 60 90 120 150 180 210 240 0 30 60 90 120 150 180 210 240

Temperature [C] Temperature [C]

C8(GLY) Cooling rate in K s‐1 2000

1000

500 300 100

50 Heat flow [mW, exo up]

10 0 30 60 90 120 150 180 210 240

Temperature [C] Figure A6.2. FSC curves of PBT/PC(5), C1(GLY) and C8(GLY), heat‐flow rate as a function of temperature, recorded on cooling samples at different rates between 10 (top) and 5000 °Cs‐1 (bottom). The curves were shifted vertically for clarity.

168

Acknowledgements

The challenges which are experienced when entering a new field of experimental work, kept me going for the last four years was only possible due to the support and assistance of many people. I would like to sincerely thank all the people that helped me during my Ph.D. study and have contributed to making it a success. First of all, I would like to thank my promoter, prof.dr. R. P. (Rint) Sijbesma, for accepting me to be part of his group during the reorganization of SPM and his scientific contributions to my project and publications. Secondly, I would like to express my gratitude to dr.ir. J.G.P. (Han) Goossens for giving me the opportunity to perform research on this chain‐of‐knowledge project in collaboration with SABIC, Bergen op Zoom. Dear Han, I admire your broad knowledge and brilliant ideas, which gave me numerous opportunities to explore the possibility of my project in both scientific and practical application directions. Although you are busy with your work in SABIC, Bergen op Zoom, you are always there when I need your help. I wish you all the best in SABIC. Thirdly, I am deeply grateful to dr.ir. J.P.A. (Hans) Heuts for his dedicated daily supervision and enthusiasm about covalent‐adaptable networks. Dear Hans, you are always capable of turning my ideal and possibility in a well‐organized and logical way that is easy and achievable for a four‐year project. Your patience, trust, and critical eye have been crucial to the successful completion of this thesis and my personal development as an independent research scientist and critical thinker. There are no words to describe how grateful I am for your endless support and help to my scientific and personal development during the past three years. All the time spent together, from the scientific discussions at your to the fun (and drinks) we had at the conferences in Gent and Lyon together, will be indelible memories for the rest of my life. I wish you a bright and fruitful academic career in TU/e. I would also like to thank my co‐promotor, prof.dr.ir. G.W.M. (Gerrit) Peters. Dear Gerrit, thank you for giving me the opportunity to perform flow‐induced crystallization experiments in your laboratory and participate the FIC meeting; also, your inspiring stories, scientific insights, and supports have been great assets to my project. I am grateful to Prof. Dr. Filip Du Prez, Prof. François Tournilhac, and prof.dr.ir. L.E. (Leon) Govaert for kindly agreeing to be a part of my defense committee and providing detailed criticism on the work in further improving the quality of this thesis. Special thanks to prof.dr. A.P.H.J. (Albert) Schenning for chairing my Ph.D. defense.

169

Acknowledgements

Special thanks go out to Sabic, for their financial support and for giving me the opportunity to connect and collaborate with multiple people from Sabic. Theo, Frank, Chiel, Jan, Vaidya and Rob for all your support for this project. I especially would like to thank dr. Ramon Groote for his help and scientific inputs for this project and our manuscripts. Special thanks to all my colleagues from SPM and supramolecular chemistry group. With all of your companion, my four‐years journey became joyful and colorful. Dear Bart, thanks for letting me join the “polyconversation” and willing to take over my daily supervision at the moment I lost during the reorganization. Although we have been working together for a short period, I am grateful for your support and contribution to my project. I wish you all the best in Allnex. Dear Benny, thanks for introducing me to the group members on the first day I started in SPM and sharing your (theoretical and practical) knowledge on rheology with me. Dear Pim, thanks for taking care the DSC and DSC measurements for our group and your support on rheology measurements in PTG. Dear Erik, my lab buddy! Thanks for teaching me the tricks of solid‐state polymerization technique and helping me build up my style fumehood. Dear Lily, thanks for taking care the group during the difficult period of the group and your passion for research is very positive energy for the rest of the group. I had a great time with you, Karel, and Hans in Lyon during the EPF 2017! Dear Cristina, you are always calm down and willing to help. Special thanks for taking care my NMR trials in Spain when I was stuck with the analysis of the glycerol composition. Dear Karel, thanks for sharing your broad knowledge and experience with me, and you are a great scientist! I also would like to thank you for giving me an opportunity to present my work at Rolduc Polymer Conference and your hospitality in Kerkrade. Dear Peter, thanks for taking care all the polymer processing equipment (extruders, compression molding, etc.) for our group, and your help during my injection molding and compounding experiments. I also like to express my gratitude to many other SPMers and PTG/e members: Pleunie, Anne, Pauline, Martin, Macro, Ton, Mark, Juliën, Geert, Tamara, Olessya, Mohammad, Joice, Raf, Bjorn, Timo, Stefen. It is also my great privilege to be a member of the Sijbesma group for the past two years. I enjoyed the discussions with these nice colleagues during group meeting and collaboration. Dear Jody, you are so energetic and always enthusiastic towards research and life. Thanks very much for help me with the X‐ray experiments in ICMS and hopefully we can have one publication together shortly. Dear Eveline, I enjoyed the time we spent together during the self‐healing materials summer school 2016, Vlieland and APME 2017 in Gent. Thank you and Jody for organizing the team‐building group activity (bowling), and we sincerely enjoyed! You have a brand new DMTA now (exciting!), and I wish you all the good luck with your Ph.D. project. I would also like to thank Martina, Jessica, Xiao, Bao, Berry, Robert, Marcos, Marko, Subham for their support and accompany during the past two years.

170

Acknowledgements

My thanks also go to the students who worked on this project: Eren, Fabienne, Jikai, and Sjarco, I enjoyed working with you over the past years. I wish you the very best of success in your PhD work and future careers. I could not finish the thesis without collaborating with other researchers. I am grateful to Francois and Adrien (ESPCI, ParisTech), for their kindness to show me the synthesis of vitrimer in their lab and the discussions we had during the project update meeting. I also would like to thank the people from polymer technology group, mechanical engineering department who helped me during the past four years. Especially, I would like to express my gratitude to several colleagues. Dear Enrico (SABIC), I admire your broad knowledge, your passion for research and your willingness to help. I am very grateful for all your help and inspiring discussions for my beamline proposal application/beam time in Grenoble, extensional rheology experiments and flash DSC data analysis with Mathematica. I truly enjoyed our time together in Grenoble, Lunteren, and Kerkrade and I wish you a bright future in SABIC. Dear Martin (TUD), thanks for giving me the introductions for the PVT and Flash DSC instruments. Your support during my experiments in Gemini is very much appreciated. I wish you a fruitful academic career in Twente. Dear Harm, thanks very much for your help with the compression set experiments as well as your willingness to help me out in Grenoble. I wish you all the success in SABIC. 终于毕业了，四年的荷兰生活中，我非常幸运有机会结识一群志同道合的同事和朋友们。向你们说一声珍重，道一声祝福，愿我们珍藏彼此真挚的友谊和一路走来的美好回忆，衷心地祝愿你们前程似锦！首先，要感谢在埃因霍温火车站久久等候的李春亮同学，你年轻有为，乐于助人。在过去的五年里，不但在科研上硕果累累，爱情也是甜甜蜜蜜。衷心祝愿你在 DSM 一切顺利，爱情事业双丰收！娄长，感谢你在 GC‐MS 上的帮助和支持。更要感谢你在 auditorium 给大家创造的畅聊科研之外的生活的机会。沈教授（礼华），非常敬佩你对科研的热情，感谢在 tensile test 和 SEM 的技术支持。更要感谢沈教授组织的烧烤和德州聚会，极大地丰富了我们大家的业余生活。祝你和你女朋友在美国的生活之旅一切顺利。一流和春晖，很荣幸能在 E 村认识两位学霸，衷心祝愿你们在今后的科研生涯里一帆风顺，硕果累累，家庭幸福美满！德磊，你的积极向上的心态和对待生活乐观的态度深深感染我们。希望你在 Eindhoven 一切顺利，事业蒸蒸日上。爽，很荣幸能和你一起成为 SPM 两个“关门” 博士，希望你在 PPG 一切顺利，爱情事业双丰收！宇晶，从微信上得知你已经成为一名父亲，很替你开心。祝愿小宝宝健康快乐地成长，也希望你在 Allnex 一切顺利，家庭美满幸福！蒙潇，你永远充满了正能量，非常感谢你组织的 Chinese new year party, 给你 12 个赞。希望你在陶氏化学一切顺利，爱情甜甜蜜蜜！宝，你总是很犀利，而且厨艺俱佳。祝你在今后的学习中一切顺利，并早日找到一份理想的工作。浩，非常抱歉没能参加你的答辩，祝你在长沙一切顺利，家庭幸福美满！huiyi 和刘 171

Acknowledgements

杰, 祝你们在今后的科研生活里一路披荆斩棘，创造辉煌。Kangbo 和立国，感谢师兄和师姐真诚的帮助和支持，你们坦然和豁达的心态深深地影响着我，让我在遇到困难的时候变得不那么急躁。衷心祝愿你们家庭幸福美满，事业更上一层楼。Miao, 感谢你在我申请 ASML 职位的时候给予的帮助和支持，衷心祝愿你在 ASML 一切顺利，步步高升，同时也祝你和你的家人幸福安康。感谢爸爸妈妈多年的养育和无私的奉献，同时也要感谢岳母千里迢迢来荷兰帮忙照顾我们的生活。希望你们在国内一切顺利，身体健康。感谢舅舅，小姨，姨妈，冬哥，老姨以及其家人对我父母的照顾和支持，在这里也向你们表达我最诚挚的谢意。我最亲爱的老婆，丹童，能娶到你，是我这辈子最大的福气。感谢一路走来你对我的信任和理解；有你相伴，我在人生的奋斗路上才不孤单。时光在流，我们在走，愿我们携手走到时光的尽头。蓓荷，感谢上苍赐给我们如此聪慧漂亮的宝贝女儿，爸爸希望你在荷兰健康快乐地成长。希望你将来成为爸爸妈妈心目中最耀眼的明星。

, Eindhoven, November 9th, 2017

172

Curriculum Vitae

Yanwu Zhou was born on the 5th of October 1985 in Leiyang, Hunan province, China. He obtained his master degree from the Department of Macromolecular Science, Fudan University, Shanghai, China in July 2012. During this period, his research topic was about structure‐property relationships of thiocarbonylthio compounds which are widely used as mediating agents in controlled radical polymerization. This systematic work was published in MACROMOLECULES, and he was honored as the outstanding master graduate of Fudan University. After one year industrial experience with China Banknote Ink co., Ltd, in Shanghai, China, since October 2013, he started his Ph.D. project under the supervision of dr. ir. J.G.P.(Han) Goossens (SABIC) on the high‐performance poly(butylene terephthalate) vitrimers at the laboratory of Polymer Materials (SPM) and subsequently joined the lab of supramolecular chemistry group at the supervision of dr.ir. J.P.A. (Hans) Heuts and prof.dr. R.P. (Rint) Sijbesma. This project is financially supported by SABIC and focuses on understanding the structure‐property relationships of poly(butylene terephthalate) vitrimers. The main results of the studies obtained within the project are presented in this dissertation. Also, he became a “Registered polymer scientist” after completing four modules of the RPK (Register PolymeerKundige) course organized by the National Dutch Graduate School of Polymer Science and Technology (PTN) in May 2016.

173

List of publications

So far, this thesis has resulted in the following publications: Peer‐reviewed journal publications:  Zhou, Y.; Goossens, J. G. P.; Sijbesma, R. P.; Heuts, J. P. A. Poly(butylene terephthalate)/Glycerol‐based Vitrimers via Solid‐state Polymerization. Macromolecules, 2017, 50, 6742–6751.  Zhou, Y.; Groote, R.; Goossens, J. G. P.; Sijbesma, R. P.; Heuts, J. P. A. Tuning PBT vitrimer properties by controlling the dynamics of the adaptable network. To be submitted.

Patent application  Johannes G. P. Goossens, Yanwu Zhou. METHODS OF FORMING POLYMER COMPOSITIONS. U.S. provisional patent, No.62/447,512, 2017.

Other publications related to the author:  Zhou, Y.; He, J.*; Li, C.; Hong, L.; Yang, Y. "Dependence of thermal stability on molecular structure of RAFT/MADIX agents—A kinetic and mechanistic study" Macromolecules 2011, 44, 8446.  Li, C.; He, J.*; Zhou, Y.; Gu, Y.; Yang, Y. "Radical‐induced oxidation of RAFT agents— A kinetic study" J. Polym. Sci., Part A: Polym. Chem. 2011, 49, 1351.  Fan, D.; Liu, Y.; He, J. *; Zhou, Y.; Yang, Y. Porous Graphene‐Based Materials by Thermolytic Cracking. J. Mater. Chem. 2012, 22, 1396.

174