Methacrylated Poly(Ethylene Glycol)S As Precursors

Home , Polypropylene glycol

METHACRYLATED POLY(ETHYLENE GLYCOL)S AS PRECURSORS

FOR SUPERPLASTICIZERS AND UV-CURABLE ELECTRICAL

CONTACT STABILIZATION MATERIALS

A Dissertation

Presented to

The Graduate Faculty of The University of Akron

In Partial Fulfillment

of the Requirements for the Degree

Doctor of Philosophy

Ali Javadi October, 2017

METHACRYLATED POLY(ETHYLENE GLYCOL)S AS PRECURSORS

FOR SUPERPLASTICIZERS AND UV-CURABLE ELECTRICAL

CONTACT STABILIZATION MATERIALS

Ali Javadi

Dissertation

Approved: Accepted:

Advisor Department Chair Dr. Mark D. Soucek Dr. Sadhan Jana

Committee Member Dean of the College Dr. Sadhan Jana Dr. Eric J. Amis

Committee Member Interim Dean of the Graduate School Dr. Miko Cakmak Dr. Chand Midha

Committee Member Date Dr. Toshikazu Miyoshi

Committee Member Dr. J. Richard Elliott

ABSTRACT

Poly(ethylene glycol)s (PEGs) are an important class of polymeric materials. In addition to standard linear PEGs, polymers synthesized from (meth)acrylated PEGs are specially versatile in modern technological applications. Comb-like copolymers derived from (meth)acrylated PEGs, such as polycarboxylate ethers (PCEs) are widely used as hydration and setting modifiers in cement while the working mechanisms in cement hydration have remained uncertain. The first part of this dissertation uncovers correlations between copolymer architecture and setting properties of cement for a range of synthesized PCEs architecture. Adsorption of PCEs on calcium silicate hydrate surfaces involves migration of Ca2+ ions in the acrylate backbone to the calcium silicate hydrate surface and subsequent ion pairing of the anionic polymer backbone with the positively charged surfaces of calcium-silicate-hydrate (C-S-H) gel.

Two consistent sets of property correlations are identified as a function of copolymer design. The adsorption strength of PCEs onto cement pastes, the conductivity, and retardation of cement hydration correlate in the same order. The water-to-cement ratio necessary for processing, zeta potentials, and the fluidity of the cement pastes correlate in a different order. The adsorption set directly correlates with the density of carboxylate groups, leading to strong, flexible ionic packing of multimolecular layers for low density and short length of side chains. The fluidity increases with density and length of PEG side chains, leading to less flexible, flat-on conformation, and lower adsorbed

iii mass. Best dispersion of cement particles and greatest water reduction require a compromise reached by low density and intermediate length of PEG side chains. The mechanisms support the design of cement materials and related particle dispersions.

A new series of ultraviolet (UV) curable electrical contact stabilization materials, which contain polypropylene glycol (PPG)-block-polyethylene glycol (PEG)-block- polypropylene glycol (PPG) capped with methacrylate functional groups on both ends as the reactive polymer and methacrylated PEGs as reactive diluents were developed in the second part of this dissertation. The photo-crosslinking behavior of these formulations was studied in detail. The effects of reactive diluents, including functionalities, molecular weight, and content were investigated via a combination of dynamic rheological measurements and real time Fourier transform infrared (FT-IR) spectroscopy.

In general, it was found that both crosslinking kinetics and dynamic modulus were affected by the selection of reactive diluents. In the case of mono-functional reactive diluents, the conversion rate and ultimate modulus were not significantly influenced by the molecular weight of reactive diluents because they could only form dangling ends in the crosslinked networks. Moreover, approximately 6% decrease in the contact resistance of electrical contact surfaces using these formulations was observed compared to the co-operating surfaces without applying the stabilization materials.

ACKNOWLEDGEMENTS

First and foremost, I would like to thank God for giving me the strength, ability, and opportunity to undertake this research study.

I would like to express my sincere gratitude to my advisor, Dr. Mark D. Soucek, for his continuous support, patience, and immense knowledge. I would like to thank Dr.

Miko Cakmak for his insightful discussions and valuable advice. I would also like to thank Dr. Sadhan Jana, Dr. Toshikazu Miyoshi, Dr. J. Richard Elliott, and Dr. Hendrik

Heinz for their insightful comments and encouragement which incented me to widen my research from various perspectives.

I am thankful for the financial support from the Department of Polymer

Engineering and Imagine Research and Technology Inc. In my daily work, I have been blessed with a cheerful and friendly group of colleagues at the Department of Polymer

Engineering. I would like to thank them for their support.

TABLE OF CONTENTS

LIST OF FIGURES ...... xi

LIST OF TABLES ...... xvii

INTRODUCTION ...... 1

BACKGROUND ...... 6

2.1. Poly(ethylene glycol)s (PEGs) ...... 6

2.2. (Meth)Acrylated PEGs...... 9

2.3. (Meth)Acrylated PEGs as superplasticizers for cement ...... 10

2.3.1. Portland cement ...... 10

2.3.2. Chemical composition of Portland cement ...... 11

2.3.3. Mechanisms of cement hydration ...... 13

2.3.4. Superplasticizers...... 15

2.3.5. Synthesis of PCEs based on (meth)acrylated PEGs ...... 18

2.3.6. Chemical architecture of PCEs ...... 19

2.3.7. Earlier studies on the working mechanism of PCEs ...... 23

2.4. Electrical contact stabilization materials based on PEG/PPG ...... 41

COMB-LIKE SUPERPLASTICIZERS: SYNTHESIS, CHARACTERIZATION, AND

CONDUCTION CALORIMETRY ...... 44 vi

3.1. Introduction ...... 44

3.2. Experimental section ...... 47

3.2.1. Materials ...... 47

3.2.2. Synthesis and characterization of copolymers ...... 48

3.2.3. Conduction calorimetry measurements ...... 49

3.3. Results ...... 50

3.3.1. Polymer synthesis and characterization ...... 50

3.3.2. Conduction calorimetry ...... 52

3.4. Discussion ...... 55

3.4.1. RAFT copolymerization of MMA and MPEGMA in aqueous media ...... 55

3.4.2. Hydration of Portland cement and effects of superplasticizers ...... 57

3.4.2.1. Hydration of alite ...... 57

3.4.2.2. Hydration of Portland cement...... 60

3.4.2.3. Role of molecular architecture of PCEs on hydration of Portland cement .. 60

3.5. Conclusions ...... 62

COMB-LIKE SUPERPLASTICIZERS: FLUIDITY TESTS AND ADSORPTION

MEASUREMENTS ...... 63

4.1. Introduction ...... 63

4.2. Experimental section ...... 71

4.2.1. Slump test ...... 71

vii

4.2.2. Adsorption measurement...... 74

4.3. Results ...... 75

4.3.1. Dispersion properties...... 75

4.3.2. Adsorption of PCEs ...... 79

4.4. Discussion ...... 82

4.4.1. General features of molecular architecture and adsorption ...... 82

4.4.2. Experimental issues in adsorption measurement ...... 83

4.4.3. Fluidity tests and reduction of the water-to-cement ratio ...... 84

4.4.4. Adsorption of PCEs ...... 86

4.5. Conclusions ...... 87

COMB-LIKE SUPERPLASTICIZERS: WORKING MECHANISMS ...... 89

5.1. Introduction ...... 89

5.2. Experimental section ...... 93

5.2.1. Zeta potential measurements ...... 93

5.2.2. Conductivity measurements ...... 94

5.3. Results ...... 95

5.3.1. Zeta potential ...... 95

5.3.2. Conductivity ...... 98

5.4. Discussion ...... 101

5.4.1. Binding mechanism and conformations of the polymers ...... 101

viii

5.4.2. Alternative models for polymer conformations and surface adsorption ...... 106

5.5. Conclusions ...... 113

UV-CURABLE CONTACT STABILIZATION MATERIALS ...... 114

6.1. Introduction ...... 114

6.2. Experimental section ...... 122

6.2.1. Materials and synthesis ...... 122

6.2.2. Sample preparation ...... 124

6.2.3. Film formation...... 125

6.2.4. Rheological measurements before UV curing ...... 126

6.2.5. Real-time rheology during UV curing ...... 126

6.2.6. Real-time FT-IR spectroscopy ...... 127

6.3. Results ...... 128

6.3.1. Synthesis and characterization of PPG-b-PEG-b-PPG dimethacrylate ...... 128

6.3.2. Rheology development during UV curing ...... 130

6.3.3. Effect of reactive diluents ...... 133

6.3.4. Contact resistance of soft gel UV-cured contact stabilization materials ...... 140

6.4. Discussion ...... 142

6.5. Conclusions ...... 147

SUMMARY ...... 149

REFERENCES ...... 153

APPENDIX ...... 177

LIST OF FIGURES

Fig. 2.1. Chemical structure of standard linear PEGs and polymers developed from

(meth)acrylated PEGs...... 9

Fig. 2.2. Transformation of raw meal into clinker ...... 11

Fig. 2.3. C3S crystals in two clinkers: (a) small crystals (smaller than 10 mm) and (b) coarse crystals (coarser than 10 mm). C2S crystals: (c) typical crystal and (d) detail of a

C2S crystal ...... 12

Fig. 2.4. Schematic representation of different stages of heat release during the hydration of an OPC ...... 14

Fig. 2.5. Isothermal calorimetry obtained on mixtures of 80 wt.% alite and 20 wt.% C3A in the presence of variable amounts of gypsum ...... 15

Fig. 2.6. Schematic representation of the comb-like PCEs...... 17

Fig. 2.7. General structure of acrylic-based PCEs...... 21

Fig. 2.8. Possible structure of hyperbranched PCEs ...... 21

Fig. 2.9. Improving the flexibility of backbone in PCEs by introducing styrene as a spacer molecule...... 22

Fig 2.10. Chemical structure of the PCEs with backbone (n) to side chain (m) ratio of n:m

= 2:1, and 6:1; number of EO units (p) in the side chains: p = 23 and 102 ...... 24

Fig. 2.11. Heat flow of cement pastes with w/c 0.35 measured for 72 h. The C3A contents of the cements increase from top to bottom row...... 26

Fig. 2.12. General structure of PCE, PMS, and BNS for investigating the impact of chemical structures on the adsorptions ...... 28

Fig. 2.13. Schematic representation of polymers distributions on the surface of hydrating cements ...... 29

Fig. 2.14. Chemical structure of the copolymers of 2-acrylamido-2-methyl-1- propanesulfonic acid (AMPS) and methoxy polyethylene glycol acrylate (MPEGA) ..... 30

Fig. 2.15. Schematic representation of random PCEs (a) and gradient PCEs (b)...... 31

Fig. 2.16. Effect of sulfate concentration on the random (left) and gradient (right) PCE adsorption ...... 32

Fig. 2.17. TEM images of the developed PNPs ...... 35

Fig. 2.18. TEM images of the PNPs adsorbed onto the cement grains...... 35

Fig. 2.19. Chemical structure of the monomers used in the study of Zhang et al. [74], (a)

AA, (b) SSS and (c) MAPTAC...... 37

Fig. 2.20. Synthesis of the PCEs containing both the carboxyl and sulfonic groups ...... 38

Fig. 2.21. Schematic representation of the adsorption of (a) PAA and (b) PSSS on the surface of cement grains ...... 39

Fig. 3.1. Chemical structure of the synthesized PCEs. The average ratios x:y were 1:1,

2:1, 3:1, 4:1, 5:1, 6:1, and z = 4.54, 9.09, 19.32, 43.18 for the polymers investigated. ... 50

Fig. 3.2. GPC chromatograms of the synthesized PCEs...... 52

xii

Fig. 3.3. The effects of PEG side chain density on the retardation of hydration of cement pastes ...... 53

Fig. 3.4. The effects of PEG side chain length on the retardation of hydration of cement pastes ...... 53

Fig. 3.5. Possible hydrophobic regions formed during the copolymerization of MMA and

MPEGMA in water ...... 57

Fig. 3.6. Different stages in the alite hydration following the Ca2+ concentration evolution as well as the associated heat release ...... 58

Fig. 4.1. Critical and saturation dosages are shown with respect to the dispersing ability of a SP ...... 66

Fig. 4.2. Adsorption of different LS polymers as a function of the added polymer ...... 69

Fig. 4.3. Adsorption of different PCEs as a function of the added polymer ...... 70

Fig. 4.4. Representative pictures for the time dependent slump loss measurements...... 73

Fig. 4.5. The effects of PEG side chain density on the slump loss behavior of the cement pastes (w/c=0.485)...... 76

Fig. 4.6. The effects of PEG side chain length on the slump loss behavior of the cement pastes (w/c=0.485)...... 76

Fig. 4.7. The effects of PEG side chain density on the water reduction capability as a function of dosage...... 77

Fig. 4.8. The effects of PEG side chain length on the water reduction capability as a function of dosage...... 78

xiii

Fig. 4.9. The effects of PEG side chain density on the adsorption isotherms of cement pastes with w/c 0.50 containing PCEs after 15 min of hydration...... 79

Fig. 4.10. The effects of PEG side chain length on the adsorption isotherms of cement pastes with w/c 0.50 containing PCEs after 15 min of hydration...... 80

Fig. 5.1. Schematic representation of the electrostatic repulsion...... 91

Fig. 5.2. Variation in electrostatic potential is shown as a function of the ionic strength of the continuous phase and of the distance from the surface of a colloidal particle ...... 92

Fig. 5.3. A general schematic representation of the effect of an adsorbed chemical admixture on the potential of a charged surface ...... 93

Fig. 5.4. Effects of the grafting density of PCE chains on the zeta potential of cement pastes with w/c ratio of 0.50 and different concentration of added PCEs...... 97

Fig. 5.5. Effects of the length of PEG side chains on the zeta potential of cement pastes with w/c ratio of 0.50 and different concentration of added PCEs...... 98

Fig. 5.6. Effects of the density of PEG side chains on the conductivity of the cement pastes containing PCEs at a w/c ratio of 0.50 as a function of time...... 99

Fig. 5.7. Effects of the length of PEG side chains on the conductivity of the cement pastes containing PCEs at a w/c ratio of 0.50 as a function of time...... 100

xiv

Fig. 5.9. Interactions of polycarboxylate ethers with a model C-S-H surface as a function of the length of PEG side chains according to molecular dynamics simulation in water

((001) facet of tobermorite 14 Å)...... 105

Fig. 5.10. Expected conformations of comb-copolymers in a good solvent according to the model by Gay and Raphaël ...... 107

Fig. 5.11. Adsorbed amount of polycarboxylate ether as a function of the COO‒ content in experiment and calculations by the earlier Flatt-Gay-Raphaël model ...... 111

Fig. 5.12. Schematic of conformations of adsorbed PCEs on hydrated cement particles and relationship to adsorption, hydration, and dispersion...... 112

Scheme 6.1. Chemical structure of MPEGMA and PEGDMA reactive diluents...... 123

Fig. 6.1. FT-IR spectroscopy of PPG-b-PEG-b-PPG (black curve) and PPG-b-PEG-b-

PPG dimethacrylate (red curve)...... 128

Fig. 6.2. 1H NMR spectra of (a) PPG-b-PEG-b-PPG and (b) PPG-b-PEG-b-PPG dimethacrylate...... 129

Fig. 6.3. Development of the elastic (G’) and viscous (G”) modulus as a function of time for formulation M-9-3.8 under a constant frequency of 6 rad/s...... 131

Fig. 6.4. Elastic (G’) and viscous (G”) modulus as a function of frequency for formulation M-9-3.8 (Table 6.1) before (t=0 s) and after UV-radiation exposure at time intervals of t=37 and 1200 s...... 132

Fig. 6.5. FT-IR spectra of formulation M-9-3.8 at four different UV exposure times. .. 133

Fig. 6.6. (a) Development of elastic and viscous modulus and (b) vinyl conversion as a function of UV-exposure time for formulations M-5-3.8 and D-5-3.8...... 135

Fig. 6.7. (a) Development of elastic (G’) and viscous (G”) modulus and (b) vinyl conversion as a function of UV-exposure time for formulations M-5-3.8 and M-9-3.8. 137

Fig. 6.8. (a) Development of elastic (G’) and viscous (G”) modulus and (b) vinyl conversion as a function of UV exposure time for formulations M-9-3.8, M-9-5.7, and M-

9-8.6...... 140

Fig. 6.9. Contact resistance results for the formulations M-9-3.8, M-9-5.7, and M-9-8.6.

The Ref. was measured without any contact stabilization materials...... 142

Fig. 6.10. Schematic representation of the crosslink density in formulations M-5-3.8 and

D-5-3.8. Compared to the mono-functional reactive diluent MPEGMA, the addition of di- functional reactive diluent PEGDMA greatly increases the crosslink density...... 144

Fig. S1. FT-IR spectrum of PCE 3:1-19...... 177

Fig. S2. 1H NMR spectrum of PCE 3:1-19...... 177

xvi

LIST OF TABLES

Table 3.1. Characterization of the synthesized PCEs. The copolymer code "x:y-z” indicates the ratio of carboxylate (x) to ester (y) and the number of PEG monomers in the side chain (z)...... 51

Table 6.1. Summary of UV-curable formulations developed for rheology tests...... 125

Table S1. Theoretical conformations, radius of an adsorbed spherical core Rac, and calculated adsorbed mass at monolayer coverage msat of the synthesized PCEs according to the Flatt-Gay-Raphaël model...... 178

xvii

CHAPTER I

INTRODUCTION

Poly(ethylene glycol)s (PEGs) have good water miscibility and can easily solubilize many poorly water soluble compounds. PEGs are commercially available in different physical states and molecular weights. The versatility of PEG-based materials can be described by their excellent chemical and physical characteristics. In recent years, there is a growing attention in the introduction of different functional groups into the

PEG-based materials. In fact, attachment of different functional groups to the terminal hydroxyl groups of PEGs has expanded their uses. Major advances have been made in the research and development of novel applications for PEG-based systems. In addition to linear PEGs, the polymers produced from (meth)acrylated PEGs are especially important in modern industrial applications. These systems are graft structures having multiple PEG side chains. Due to their particular structures, these polymers are usually biocompatible and water-soluble, showing stimuli-responsive properties.

Polycarboxylate ethers (PCEs), as comb-like copolymers derived from

(meth)acrylated PEGs, are the last-generation superplasticizers (SPs) for cementitious materials. These materials consist of a main chain bearing carboxylic groups, to which

PEG side chains are attached. In fact, Portland cement is a complex product which is generally composed of abundant materials such as limestone and clay. Ordinary Portland

1 cement (OPC) is made from four phases, to which a small amount of calcium sulfate has been added. In order to better understand the effects of chemical admixtures, including

SPs on the properties of cement, it is important to have a good understanding of the chemical and physical consequences of cement hydration. In general, the hydration of

OPC includes dissolution and precipitation processes in a very complex chemical system, hence leading to the setting and hardening of cement. The final products of the hydration of silicate part of cement are a gel of calcium silicate hydrate (C‒S‒H) and calcium hydroxide (CH) as a crystalline phase.

Superplasticizers have found wide applications in cement and concrete industry.

Compared to the normal water-reducing materials, SPs can highly decrease the mixing water needed in a cement mixture. In PCEs, the carboxylic groups dissociated in aqueous media can confer negative charges to the main backbone. In fact, the negatively charged main chains of PCEs are responsible for their adsorptions onto the positively charged cement particles. Moreover, the dispersing ability of PCEs is as a result of the steric hindrance effects of the PEG side chains. While the adsorption properties depend on the number of free carboxylic groups, the steric stabilizations depend on the density and length of the PEG side chains.

Free radical copolymerization of methacrylic acid and a macromonomer bearing

PEG side chain is the most common technique for the synthesis of PCEs. However, this method leads to high polydispersity degree and gradient distribution of PEG side chains along the polymer backbone. These ill-defined structures cannot be used as good models for understanding the exact working mechanisms of PCEs. Recently, reversible addition- fragmentation chain transfer (RAFT) polymerization technique has been used to develop

PCEs having specific compositions. This method could provide an efficient structural control in terms of monomer distributions and molar mass.

The main objective of the first part of this dissertation is to appropriately link the structural factors of well-defined PCEs with their performance in cementitious materials in order to better understand the exact interaction mechanisms. Hence, nine PCEs with constant length of methacrylate main chain (DP = 50) and systematic variation in the grafting density and length of the PEG side chains are synthesized using RAFT polymerization technique. The influence of PCEs on properties of cement pastes is determined using a variety of tests including hydration retardation, slump loss, water reduction, calorimetry, zeta potential, and adsorption isotherms.

In the second part of this dissertation, UV-curable contact stabilization materials based on methacrylated PEGs were designed, synthesized, and tested for electrical contact surfaces. In any electrical contact, the co-operating contacting surfaces can never be perfectly smooth or flat, intrinsically arising from the manufacturing process, even at the atomic level. When observed under sufficiently high magnification, such contacting surfaces may appear fairly rough with number of irregular peaks and pits. Other factors, such as impurities attached to the surface and grain boundaries, may also contribute to the overall surface roughness. Hence, when two co-operating surfaces contact with each other, the surface roughness causes only a small part of the surfaces to be physically connected. Therefore, the transmission of electrical current from one contact surface to the other can only take place at those places where there is real physical contact between two materials. Such co-operating contact surfaces, when used in electrical contact situation, may cause adverse effects to a variety of applications. For example, in high

3 current applications, there may be heat induced deformation or even chemical reaction due to the contact resistance. Also, this discontinuity in alternating signals transmission may add artificial signals to transmission in radio frequency circuitry or exaggerates noise level, causing program crashes, incorrect data transmission, spurious parity, and cyclic redundancy error in the case of computer circuitry.

Commercially available stabilization materials, block copolymers of polyethylene glycol (PEG)-b-polypropylene glycol (PPG), are normally coated onto one or both of a pair of co-operating contact surfaces, e.g. edge card connectors for the purpose of electrical performance improvement. However, after being applied, they remain liquid for the life of application at room temperature. The fact that this liquid can flow or leak significantly restricts its applications. The main aim of this part of the present dissertation is to develop a methacrylated PPG/PEG-based material which can be uniformly coated onto the entire surface of electrical co-operating systems. More importantly, this material is able to be cured under UV light, producing a gel-like coating onto the surfaces. The new material is found to switch from non-conductive to conductive state at very low film thickness.

The outline of this dissertation is as follows. A comprehensive introduction regarding the (meth)acrylated PEGs as superplasticizers for cement as well as the electrical contact stabilization materials is provided in Chapter II. Then, the synthesis, characterization, and conduction calorimetry results of the developed PCEs in this study are discussed in Chapter III. The detailed fluidity tests and adsorption measurements of these well-defined PCEs are described in Chapter IV. The obtained results of zeta potential and conductivity measurements as well as a comprehensive interpretation

4 regarding the working mechanism of PCE adsorption and cement hydration based on both experimental and simulation results are then described in Chapter V. Design, synthesis, and characterization of UV-curable contact stabilization materials based on methacrylated PPGs/PEGs are highlighted in Chapter VI. The dissertation finishes with a summary in Chapter VII. Supporting Figures and Tables can be retrieved from the supporting information in the Appendix section.

CHAPTER II

BACKGROUND

The text of this chapter is a reprint of the material as it appears in: Javadi, A.; Mehr, H.

S.; Soucek, M. D. Polym. Int. DOI: 10.1002/pi.5432

2.1. Poly(ethylene glycol)s (PEGs)

Poly(ethylene glycol)s (PEGs) with the general structure of HO–CH2–[CH2–O–

CH2]n–CH2–OH are an important class of polymeric materials, mostly employed as vehicles in parenteral and oral dosage forms [1-3]. PEGs have good water miscibility and can solubilize various poorly water soluble materials. These characteristics are attributed to the terminal hydroxyl (–OH) groups and ether linkages in the polymer backbones.

PEGs are available in different physical states varying from clear liquids (low molecular weight (Mw)) to semisolids (medium Mw) and solids (higher Mw). PEGs are commercially available in various molecular weights ranging from 200 to 10,000,000 g/mol. In general, PEGs having high molecular weights (∼20,000) are referred to as poly)ethylene oxide(s (PEOs). In comparison with higher molecular weight PEGs, those having lower molecular weights have a large number of –OH groups per weight basis and higher hygroscopic characteristics.

The most versatile method for the synthesis of PEGs with narrow dispersity and precisely controlled chain length is living anionic ring opening polymerization [4]. The

6 versatility of PEG and related materials can be described by their exceptional chemical and physical properties. In fact, PEG is non-toxic, chemically inert, and hence a predestined candidate for several biomedical applications. PEG ensures good solubility in water and common organic solvents, moreover, is capable of preventing undesired interactions with human immune system. Recently, there is a growing interest in the introduction of functional groups into PEG-related materials. Due to this fact that the repeating ethylene oxide (EO) units of PEG have no reactive side groups, PEG can be bound to other compounds via terminal –OH groups. In general, two approaches are used for the functionalization of PEGs [5, 6]. The first one is the alteration of terminal –OH groups through a series of reactions to more active functional groups. The second method is the reaction of difunctional compounds with PEG under controlled conditions so that one of the functional groups can react with PEG and the other one remains active.

However, several steps are mostly conducted to achieve the expected derivatizations.

The ability to attach different functional groups to the terminal ‒OH groups of

PEG polymers has expanded their benefits. PEGylation process, which is considered as the covalent grafting of PEG derivatives onto molecules, can improve biocompatibility and water solubility, especially valuable for drug development [7]. For medical applications, PEGylated materials need extensive characterization using complex analytical techniques to ensure regulative compliance [8, 9]. Homo- and hetero- bifunctional PEG-based systems are specifically suitable as spacers or cross-linkers between chemical entities [10, 11], whereas PEGs with mono-functional structures prevent bridging which can otherwise affect the PEGylation of compounds with bifunctional structures [12, 13]. Bifunctional PEG systems are often used for the

PEGylation of proteins [14, 15], peptides [16], small molecules [17], virus particles and nanoparticles [18-20], and surfaces [21, 22]. The bulky structures of branched PEGs consisting of linear PEG side chains attached to a central core can facilitate single point attachments to different targets via single reactive groups [23]. In addition, multi-arm

PEG derivatives, prepared through ethoxylation of various cores including pentaerythritol, tripentaerythritol, or hexaglycerol, are easily cross-linked into three- dimensional (3D) hydrogel materials [24-26].

Significant advances have been made by the scientific community in the research and development of new applications for PEG-containing systems. For example, these materials have been widely desired because of their potential in drug delivery applications, either through direct PEGylation of therapeutics [14, 27] or via PEG-based vehicles, including nanoparticles [28, 29], dendrimers [30], or micelles [31, 32].

Important factors which influence the bioactivity of PEGylated drugs are the length of

PEG chain, PEGylation site, and conditions for PEGylation reaction. The development of hydrogels is an important use of PEG-based systems [33-35]. These hydrogels can be used for different applications, such as cell culture, adhesives for wound closure, tissue engineering, controlled release matrices, medical devices, and regenerative medicine tools. PEG is the gold standard polymer for cosmetic, pharmaceutical, and medical applications and is utilized for a wide range of products. In recent years, many excellent reviews have been published on topics related to PEGs, such as high molecular weight

PEGs [36], pharmaceutical applications of PEGs [37, 38], introduction of degradation sites into PEGs [39], and advances in polymerization of epoxides, as well as novel polymer architectures available via this route [40].

2.2. (Meth)Acrylated PEGs

In addition to standard linear PEGs, the polymers synthesized from

(meth)acrylated PEGs are specially versatile in modern technological applications. As shown in Fig. 2.1, such polymers are graft structures composed of a carbon-carbon backbone and multiple side chains containing oligo(ethylene glycol) segments. Due to this particular structure, in most cases, these macromolecules are biocompatible and water-soluble. In addition, these polymers show stimuli-responsive properties, which are usually not attainable with linear PEG systems. Recently, it has been demonstrated that these polymers generally exhibit a lower critical solution temperature (LCST) behavior in pure water or physiological media [41-44]. Comprehensive details regarding monomer synthesis and polymerization in these systems can be found in the literature [45, 46].

O O Linear poly(ethylene glycol) O O O

non-linear PEG-analogue

O O O n O O O O O O O O O O

Polymerizable PEG side chain methacrylate n= 2, 3, 4, 8, 20, ... O O O O O n n n n n Fig. 2.1. Chemical structure of standard linear PEGs and polymers developed from (meth)acrylated PEGs.

Although the synthesis and biomedical applications of the polymers based on

(meth)acrylated PEGs have been widely investigated, these systems provide many other crucial applications as well. This chapter is not meant to be an exhaustive review of all

9 the applications of (meth)acrylated PEGs, but focuses instead on the use of these systems in superplasticizers for cement and concrete.

2.3. (Meth)Acrylated PEGs as superplasticizers for cement

2.3.1. Portland cement

Portland cement is considered as a complex product which is made from simple and abundant materials including limestone and clay. In general, precise proportions of these two materials have to be mixed with some additives to produce a raw meal having a precise chemical composition. This process will result in the creation of clinker by a complex pyroprocessing presented in Fig. 2.2. The production cost for Portland cement is thoroughly linked to the cost of the fuel utilized to make a kiln temperature high enough to enable different chemical reactions which transform the raw material into clinker.

Clinker generally comes out of the kiln as grey nodules, darker or paler based on its iron amount. When the iron oxide content is less than 1%, the clinker is whitish, and also the lower content of the iron oxide, the whiter the cement made.

Because no two raw feed materials are totally similar, no two clinkers are identical. In fact, even in the case of a plant which runs two identical kilns fed with same raw meal, the clinkers produced are not absolutely similar due to this fact that no two kilns are identical perfectly. Furthermore, it is not possible to maintain the same pyroprocessing conditions in a kiln all day long in burning and quenching areas, consequently the clinkers created have a certain degree of variability. Before being ground, the clinker is stocked temporarily in a homogenization hall. It seems that this is 10 the best method to make Portland cement with almost constant properties which facilitates the production of concrete having predictable properties [47].

Fig. 2.2. Transformation of raw meal into clinker [48]. Reproduced with permission from KHD Humboldt Wedag.

2.3.2. Chemical composition of Portland cement

Ordinary Portland cement (OPC) is composed of four basic phases, to which a very small amount of calcium sulfate (essentially gypsum) has been added. These main phases include alite (an impure form of tricalcium silicate SiO2.3CaO (C3S)) and belite

(an impure form of dicalcium silicate SiO2.2CaO (C2S)) as well as an interstitial phase composed itself of two main phases, namely tricalcium aluminate Al2O3.3CaO (C3A) and tetracalcium aluminoferrite 4CaO.Al2O3.Fe2O3 (C4AF). The first two minerals create the silicate phase of the Portland cement, and the C3A and C4AF constitute its aluminous

11 phase. In general, C3A and C4AF are crystallized, depending on the composition of raw meal, the maximum temperature in firing zone, and its final quenching after passage in the firing zone. Depending on the final grinding, some gypsum can be dehydrated partially into calcium hemihydrate or totally dehydrated to create a calcium sulfate. The images of C3S and C2S are shown in Fig. 2.3.

Fig. 2.3. C3S crystals in two clinkers: (a) small crystals (smaller than 10 mm) and (b) coarse crystals (coarser than 10 mm). C2S crystals: (c) typical crystal and (d) detail of a C2S crystal [48]. Reproduced with permission from KHD Humboldt Wedag.

In general, the CaO, SiO2, Al2O3, Fe2O3, etc. contents of the Portland cement do not exist separately as oxides. Rather, they are usually combined to create the different minerals which are found in Portland cement. The CaO contents of different Portland cements are always high. For example, these amounts are around 60‒65% for grey cements and around 70% for white cements. In addition, the silica content varies between

20% and 24%. However, the Fe2O3 and Al2O3 contents are more variable. Besides these main oxides, the chemical analysis shows the presence of other oxides with low contents

[47].

2.3.3. Mechanisms of cement hydration

To better understand the effect of superplasticizers on the properties of fresh and hardened cement, it is very important to have a good understanding of the chemical and physical consequences of hydration. In general, the hydration of OPC involves dissolution and precipitation processes in a complex chemical system, resulting in the creation of different hydrates. These processes lead to the setting and hardening of cement. The final products of silicate hydration are a gel of calcium silicate hydrate (C‒

S‒H) as well as a crystalline phase of calcium hydroxide (CH), also called portlandite.

The hydration reactions between calcium sulfate and aluminates lead to the creation of two different phases which are called trisulfoaluminoferrite hydrates (AFt) and monosulfoaluminoferrite hydrates (AFm). The most important phase in AFt is ettringite.

In addition, the AFm is typically composed of positively charged calcium and aluminum platelets on the octahedral coordination with oxygen [49].

As shown in Fig. 2.4, the heat release resulting from the hydration of an OPC generally shows characteristics from the hydration of silicates and aluminates phases and can be divided into five steps. In stage I, the exothermic peak is because of the wetting of cement surface and the fast dissolution of anhydrous phases. Moreover, in the very first minutes, the ettringite precipitates because of the availability of calcium sulfate and the high reactivity of aluminates. This initial sharp peak is followed by a sudden slowdown of the reaction and induction period (stage II). As shown in stage III, the second main peak generally corresponds to the precipitation of main products of the silicates hydration with an increase in the heat release resulting from the simultaneous increase in dissolution of the C3S. Afterwards, the reaction slows down (stage IV) and a small peak

13 also occurs which is probably related to the sulfate depletion point and corresponds to a faster precipitation of the ettringite and a higher dissolution of the C3A. Stage V is a low activity period because of slow diffusion of species in the hardened material. This stage includes another peak corresponding to the creating of AFm resulting from the reaction between C3A and ettringite [49].

Fig. 2.4. Schematic representation of different stages of heat release during the hydration of an OPC [49]. Reproduced with permission from Elsevier.

Studying the major mechanisms with pure phases is the first step to understand the mechanisms involved in cement hydration. However, it is essential to study the interactions between different phases which might lead to other reactions or improve kinetics with respect to those in pure phases. The importance of balance between the silicates, aluminates, and sulfates has been investigated [50] (Fig. 2.5). In the absence of sulfates, it was shown that in a system having 80 wt.% of pure C3S and 20 wt.% of C3A, the aluminate hydration mostly suppresses the silicate hydration. However, the extent of

C3S hydration increases as gypsum is added. Also, this addition delays the sulfate

14 depletion point. At a certain dosage (~4% in the case shown in Fig. 2.5), there is an inversion so that the sulfate depletion point can occur right after the silicate peak.

Although from this dosage on, the addition of more sulfates further delays the sulfate depletion point, it does not affect the silicate peak. It seems this phenomenon is related to availability of aluminum ions in pore solutions.

Fig. 2.5. Isothermal calorimetry obtained on mixtures of 80 wt.% alite and 20 wt.% C3A in the presence of variable amounts of gypsum [50]. Adapted with permission.

2.3.4. Superplasticizers

Superplasticizers (SPs), also called water-reducing admixtures, were developed in the 1970s and have found wide applications in the construction industry. These materials have the ability to decrease the mixing water in a given concrete mixture compared to normal water-reducing materials. They consist of long-chain and high molecular weight anionic surfactants having a large number of polar segments in the hydrocarbon chains.

When adsorbed on cement particles, surfactants can impart strong negative charges,

15 which help to lower surface tension of surrounding water and enhance fluidity of systems considerably. In general, compared to the normal water-reducing admixtures, large amounts of SPs, up to 1% by weight of cement (bwoc), can be introduced into the concrete mixtures without excessive bleeding and set retardation.

It seems that the colloidal size of long-chain particles of admixtures can obstruct the water-bleed flow channels in the concrete, so that the segregation is not encountered in SP-containing concretes. In fact, outstanding dispersion of cement particles in water accelerates the hydration rate. Consequently, the retardation is rarely observed; instead, the acceleration of hardening and setting is a normal occurrence. The first generation of

SPs possessed a bad reputation for rapid loss of slump or consistency. In fact, the currently available SPs frequently contain some retarding materials to offset the rapid loss of consistency achieved after the addition of admixtures.

Compared to 5‒10% reduction in the mixing water which is possible using ordinary plasticizing products, a water reduction in the 20‒30% range can be achieved while maintaining the high consistency in reference concrete. In general, an increase in the mechanical properties is commensurate with a reduction in the water to cement (w/c) ratio. Because of the greater rate of cement hydration in well-dispersed systems, the concrete mixtures containing SPs exhibit higher compressive strengths than the reference concrete with the same w/c ratio. This is of special importance in precast concrete where high early strengths are needed for the rapid turnover of formwork. By means of high cement content and w/c ratios lower than 0.4, it is possible to attain even faster strength development rates.

The last-generation SPs are comb-like copolymers which were introduced in the mid-1980s [51]. This category includes several types of superplasticizers showing the same common comb-shaped structure. In general, comb-like SPs are derived from

(meth)acrylated PEGs and consist of a backbone bearing carboxylic groups, to which non-ionic PEG side chains are attached (Fig. 2.6). These SPs are called polycarboxylate ethers, polycarboxylate esters, or polycarboxylates (PCEs). In fact, the carboxylic substituents, dissociated in water, can confer a negative charge to the main chain. It is believed that the negatively charged main chain is responsible for the SP adsorption onto the positively charged cement particles. In addition, the dispersing characteristics of

PCEs comes from their non-adsorbing side chains, which are responsible for the steric hindrance effects. While the adsorption properties strongly depend on the number of free carboxylic substituents, the steric stabilizations depend on the amount and length of the side chains of the adsorbed polymers [52]. More precisely, the adsorption is affected by all structural factors of the SP, but to different extents [49, 53].

------

Fig. 2.6. Schematic representation of the comb-like PCEs.

2.3.5. Synthesis of PCEs based on (meth)acrylated PEGs

The key to the success of PCEs lies in the fact that they can offer a wide range of possible chemical structures. Since the chemical structure significantly affects the performance of PCEs, tailoring the molecular structures can enable the development of

SPs with different properties which can be used in a broader range of industrial applications. The main parameters determining the performance of PCEs include the length and chemical nature of the backbone, the length, chemical nature, and distribution of the side chains, the anionic charge density, and the type of linkage between backbone functionalities and side chains.

Two main methods are usually used to synthesize the PCEs. One is the free radical copolymerization of a monomer containing carboxylic acid group and a macromonomer bearing side chain. This approach is the most common method as it possesses a simple experimental process and is cost-effective. Also, the radical copolymerization is considered as an ideal technique for the incorporation of different monomers into the main chain. This method leads to the gradient distribution of side chains along the polymer backbone. In addition, it gives rise to the PCEs with high polydispersity degree, in terms of both the size of polymers and the statistical distribution of monomers. Due to this fact that the reactivity of two monomers are different, the PCEs synthesized by free radical copolymerization have a gradient distribution of side chains along the copolymer backbone. For such PCEs, the typical polydispersity index (PDI) is in the range of 2‒3. The other approach for the preparation of PCEs is the esterification or amidation of a preformed backbone containing carboxylic acid groups with mono- functional PEGs. This technique leads to the PCEs with a narrower distribution of

18 molecular weights, because the length of the backbone is fixed and the side chains are uniformly distributed along the backbone.

As mentioned, the composition of monomers along the copolymer chains is hardly controlled using the common synthetic procedures, such as free radical copolymerization or esterification reactions. Hence, the resulting polymers exhibit broad molar mass distributions. In order to understand the working mechanisms of PCEs, the ill-defined structures cannot be utilized as good models. Recently, D’Agosto, Pourchet, and coworkers [54, 55] have used the reversible addition-fragmentation chain transfer

(RAFT) polymerization technique to synthesize PCEs possessing specific compositions.

In fact, the RAFT polymerization provided an effective structural control in terms of molar mass as well as the monomer distributions.

2.3.6. Chemical architecture of PCEs

In general, the side chains utilized in the production of PCEs are made of the PEG with molar masses in the range of 750‒5000 g/mol. In fact, the PEGs having different masses, and hence lengths, can be introduced in the same polymer to balance the charge densities. In addition, the PEG side chains confer high hydrophilicity to the PCE systems.

The hydrophilic properties of such PCEs can be decreased by means of polypropylene oxide (PPO) instead of PEG side chains. Also, they can be copolymerized to develop the mixed side chains of varying water solubility. The PCEs with PPO side chains can provide different performances, such as reduced air entrainment [56].

The PEG chains are typically introduced in both synthetic methods, the esterification approach as well as the synthesis of the macromonomer for radical copolymerization. The general form of these systems is a methoxy-terminated PEG. In fact, the presence of bifunctional PEG during the esterification reaction leads to the crosslinking of polymer chains and the formation of water-insoluble gels. However, the PCEs with OH-terminated PEG side chains can be synthesized by the radical copolymerization of hydroxy-terminated PEG methacrylate and methacrylic acid. Although the market for comb-shaped SPs is dominated by the PCEs with PEGs as side chains, some attempts have been made to utilize other types of side chains. For example, it has been reported that PCEs with PEG and ethoxylated polyamides as side chains can be highly efficient as dispersants, allowing the use of w/c ratio as low as 0.12 [57].

The monomer part of the backbone in the acrylic-based copolymers is acrylic acid. As shown in Fig. 2.7, the bond between these parts and the PEG side chains is usually an ester bond. It seems that the ester bonds of acrylic- or maleic-based SPs are prone to undergo hydrolysis in the aqueous phase of cement as an alkaline medium. The cleavage of some of the PEG side chains can lead to an increase in the number of free ‒

COOH groups. Consequently, PCEs with higher charge density and improved adsorption should be formed over time. In fact, change in adsorption abilities can compensate for flow loss in cement pastes. This phenomenon has been utilized for production of cement dispersants with good slump-retention characteristics [58].

H H2 H CH2 C C C n m HO O O O

O p CH3

Fig. 2.7. General structure of acrylic-based PCEs.

Cross-linked SPs can exhibit similar retention ability. For example, Miao et al.

[59] reported a PCE system with hyperbranched structure (Fig. 2.8). The hydrolysis of ester bonds in this system could provide a supply of SPs for adsorption on cement particles. The resulting PCEs could be more stable towards the hydrolysis of their side chains when methacrylic acid was used instead of acrylic acid. In addition, the hydrolysis of ester bonds can be overcome through substitution with more stable chemical bonds, such as ether, amide, and imide bonds.

Macromonomers based on allyl ether are also copolymerized with different monomers, including acrylic acid, maleic acid, or maleic anhydride to create PCEs. In general, compared to the methacrylic-based PCEs, allyl ether-based SPs exhibit a lower adsorption but improved slump retention [60]. The monomer type in the backbone of

PCEs affects their performance as superplasticizers. For example, although the chemistry of backbone in the acrylic- and maleic-based superplasticizers may look the same, these

PCEs exhibit different adsorption behavior. Mostly, this is due to the additional chelating property of the vicinal COOH groups of maleic segments. The ‒CH3 groups in the α- position with respect to the carbonyl, as for methacrylic- and methallylic-based superplasticizers, decease the mobility of main chain and change the adsorption behavior of the polymer. For instance, the flexibility of the backbone can be improved by introducing spacer molecules, such as styrene monomer (Fig. 2.9). In general, the adjustment of chemistry in superplasticizers allows modifying polymers with different properties.

H H H CH2 C C C n m HO O O O

H2 R' = CH2 C O CH3 n

Fig. 2.9. Improving the flexibility of backbone in PCEs by introducing styrene as a spacer molecule.

2.3.7. Earlier studies on the working mechanism of PCEs

Regardless of the increasing number of research projects, there are still several unsolved problems when using SPs in cementitious materials. The effects such as the early slump loss, poor flow behavior, and strong retardation can be as a result of the incompatibility between the SPs and binders. Many studies on the SP–cement interactions do not investigate the chemical structure of the admixtures used. In order to close this gap, Winnefeld et al. [61] studied a series of admixtures with systematic variations of molecular structures. The objective of this study was to link the structural factors of the PCEs with their performance in cementitious systems. Therefore, different

PCEs were synthesized varying the density and length of the PEG side chains as well as the molecular weight of the copolymers. The influence of these PCEs on the properties of cement pastes was determined by various characterization techniques. Although the decreasing density of the PEG side chains increased the workability, the lengths of the side chains and the molecular weights of the copolymers had a minor effect. The minor influence was probably because of the mushroom-like conformation of the PEG side chains in aqueous solutions. In addition, the adsorption measurements exhibited that the copolymers with higher charge densities could adsorb to a larger extent compared to the copolymers with lower charge densities.

The effect of hydroxy termination of PEG side chains in the performance of PCEs was investigated [62]. Compared to the SPs having conventional methoxy termination, almost similar properties were observed. In fact, because of the similar anionic charge densities and molecular weights for both methoxy and hydroxy terminated polymers, the copolymerization led to similarly structured comb-like copolymers. In comparison with

23 the –OCH3 terminated PCEs, the hydroxy terminated SPs showed slightly better slump retention behavior over time. The measured adsorptions correlated well with the required dosages for achieving high paste flow. In addition, the zeta potential values revealed similar trends for both the ‒OH and –OCH3 terminated copolymers. Moreover, the measurements of isothermal heat flow calorimetry for the SPs in cement did not show significant differences. In general, the results suggested that for the synthesis of PCE admixtures, the –OH terminated PEG-based macromonomers could offer an attractive alternative to the corresponding –OMe terminated macromonomers.

Zingg and coworkers [63] performed a parametric study to link the molecular structure of a series of comb-shaped PCEs with their performances in cement pastes having different tricalcium aluminate (C3A) contents. Hence, three PCEs were synthesized by grafting PEGs on a methacrylic acid copolymer backbone. As shown in

Fig. 2.10, the architectures of copolymers were varied by using different PEG side chains densities and lengths.

n m NaO O O O

O p CH3 Fig 2.10. Chemical structure of the PCEs with backbone (n) to side chain (m) ratio of n:m = 2:1, and 6:1; number of EO units (p) in the side chains: p = 23 and 102 [63].

As shown in Fig. 2.11, the presence of PCEs showed retarding effects on the start of acceleration period of cement pastes. The higher charge density and dosage of the

PCEs resulted in the stronger delay in the settings. Besides, with increasing the contents of C3A in cement pastes, the retardation effects of the PCEs decreased independent of their dosage and architecture. It was found that the workability was influenced by the ionic charge density of the PCEs used. Consequently, the PEG side chain density was the key factor which controlled the adsorption behavior and therefore the strength of steric and electrostatic stabilization. In fact, the steric stabilization could govern the flow behavior of the cement pastes. It was also demonstrated that the role of PEG length was of minor effect if the PCE weight concentrations were compared. However, if the charge densities were taken in account, the PCEs with longer side chain length showed improved dispersion ability, which was due to their stronger steric repulsive forces. Moreover, despite the influence on zeta potential of the cement pastes, the PCEs having high side chain density and length showed minor impact on the paste rheology, which was supported by the adsorption data.

Fig. 2.11. Heat flow of cement pastes with w/c 0.35 measured for 72 h. The C3A contents of the cements increase from top to bottom row. The left column shows the effect of PCE architecture and the right column shows the effect of dosage (PCE 102-6) [63], Copyright 2009. Reproduced with permission from Elsevier.

In another study, Ran et al. [64] investigated the effects of PCEs having variation in PEG side chain length, on the adsorption, zeta potential, dispersion, and rheology of cement suspensions to elucidate the dispersion mechanism of these copolymers. It was found that the adsorption amount was controlled by the COO‒ content in polymer backbones and by the length of PEG side chains as well as the polymer conformation. 26

The adsorption of these copolymers on the cement particles controlled the dispersion and rheological behavior of cement pastes. Although the dispersion effects increased as the adsorbed amounts increased, the dispersing power of these copolymers was dependent on the PEG side chain length. In fact, the long side chains had more dispersion power than the short side chains. The rheological measurements showed that all the suspensions yielded a shear-thinning behavior, except at high shear rate or high dosages where a

Newtonian behavior was evident. The addition of PCEs with long PEG side chains or low ionic contents had little influence on the zeta potentials, while short side chain PCEs having high ionic content induced the largest potentials, because of their different conformation and orientation of the adsorbed copolymers. In addition, it was found that the steric hindrance effects dominated the dispersion behavior of the comb-like PCEs.

However, for the short side chain PCEs with high ionic contents, the electrostatic repulsion and steric hindrance were proposed to be responsible for the dispersion effects.

The structure-property relationships also suggested that a geometrical balance between the main chain and side chains in PCEs was very useful in designing optimum molecular structures of highly efficient dispersants.

The zeta potential of early hydration cementitious products is a key parameter for the superplasticizer adsorption. Plank and Hirsch [65] studied the influence of zeta potential of early cement hydration products, such as ettringite, portlandite, monosulfate, syngenite, and gypsum on the superplasticizer adsorption. Hence, these hydrate phases were prepared and their zeta potential values were measured using an instrument based on electroacoustic technique. As shown in Fig. 2.12, to investigate the impact of chemical structure of SP molecules on the adsorption, some polycondensate admixtures based on

27 melamine (PMS) or naphthalene (BNS) as well as polycarboxylates based on methacrylic acid/methoxy-terminated PEG copolymer (PCE) were synthesized. It was concluded that a positive zeta potential was necessary to achieve high SP adsorption. Among the early cement hydration phases, ettringite and monosulfate showed positive zeta potentials and could adsorb high amounts of SPs. In addition, syngenite, gypsum, and portlandite showed zero or negative zeta potentials and did not adsorb SPs. Therefore, SPs adsorbing on the hydrating cement grains were mostly concentrated on the spots where ettringite was crystallized. The results were mosaic structures for the hydrating cement grains having uneven distribution of polymer molecules on the surfaces. A schematic representation of the creation of adsorbed polymer layers on the cement grains during their early hydration is shown in Fig. 2.13.

SO3Na CH CH3 CH3 2 H2 H2 CH2 C C C C C

O O NaO O CH3 p q

O p CH3 m PCE

H H N N N O H2 C N N

SO Na NaO3S 3 n n

PMS BNS

Fig. 2.12. General structure of PCE, PMS, and BNS for investigating the impact of chemical structures on the adsorptions [65].

The water-soluble copolymers of 2-acrylamido-2-methyl-1-propanesulfonic acid

(AMPS) and methoxy polyethylene glycol acrylate (MPEGA) were synthesized using free radical polymerization method and evaluated dispersant for cement particles [66]. In general, the slump-retaining effects of the developed copolymers were studied in terms of pH, composition, and molecular weight of PEG side chains. These copolymers were characterized by means of Fourier Transform Infrared Spectroscopy (FT-IR) and Nuclear

Magnetic Resonance (NMR) spectroscopy. The dilute solution viscometry measurements were performed to compare the molecular weight effects on the fluidity of copolymers. In addition, the mechanical properties of mortar samples prepared by the synthesized copolymers were investigated to determine the flexural and compressive strengths. It was demonstrated that the reaction pH had a significant effect on the molecular weight of the

PEG-grafted samples, causing an important effect on the fluidity. The copolymers synthesized at a pH of 6 showed the highest fluidity result. Moreover, the copolymers with shorter PEG side chains provided higher viscosity average molecular weight and fluidity values than the copolymers with longer PEG side chains. Although an increase in the molecular weight caused an increase in the fluidity, a decrease in the mechanical

29 properties because of different air contents of these copolymers was observed. Fig. 2.14 shows the general chemical structure of these copolymers.

CH2 CH2 H H 2 OH a O CH3 b N C H O H S n O O O O

APS initiator

CH CH b a HN O O O

CH2 O S O HO n O CH3

Fig. 2.14. Chemical structure of the copolymers of 2-acrylamido-2-methyl-1- propanesulfonic acid (AMPS) and methoxy polyethylene glycol acrylate (MPEGA) [66].

In an interesting study, Pourchet and coworkers [55] used calcite suspensions in equilibrium with respect to calcium hydroxide (Ca(OH)2). This system containing different amounts of sulfate ions was utilized to investigate how the repartition of the anionic MAA segments as well as PEG side chains could affect the polymer behavior in terms of dispersion and adsorption ability. The RAFT copolymerization method allowed developing well defined copolymers differing only in repartition (random or gradient) of the charged segments along the backbone. These copolymers are schematically represented in Fig. 2.15.

This work showed that the repartition of the carboxylate groups along the copolymer backbone had a strong effect on the adsorption in the presence of sulfate ions.

In general, a significant decrease in adsorption was observed as the sulfate concentration increased. This phenomenon was related to the sulfate competitive adsorption on the surfaces covered with calcium ions as well as the subsequent decrease of calcium ions in the solution, providing the modification of calcite interface. As shown in Fig. 2.16, in comparison with the random repartition of the comonomers along the comb-like copolymers, a gradient repartition of the comonomers strongly increased the copolymer adsorption, especially when the concentration of sulfate ion increased. Thus, the adsorption of the equivalent gradient copolymer was significantly less sensitive to the sulfate competitive adsorption. In fact, a gradient copolymer could behave like a random copolymer having a lower esterification ratio, showing the contribution of the less esterified part of the gradient polymer in its adsorption.

Fig. 2.16. Effect of sulfate concentration on the random (left) and gradient (right) PCE adsorption [55], Adapted with permission from Elsevier.

In another study, Alonso and coworkers [67] investigated the compatibility between three different PCEs and four commercially available cements by adsorption, calorimetry, zeta potential, and rheological methods. From the adsorption point of view, the higher the percentage of carboxylate segments in admixture, the more strongly it is adsorbed on cement pastes. According to the rheological results, the optimum ratio of carboxylate to ester groups for the admixtures was found to be in the range of 0.7‒1.2.

Moreover, the fluidizing effect of the superplasticizers on cement pastes was conditioned by the presence of minerals. Despite the low adsorption amounts of the superplasticizers in slag-blended cements, the presence of PCEs produced the steepest declines in the rheological factors. The delay effect of admixtures on cement hydration improved with increasing PCE dosage.

In order to obtain high flow speed, the PCE should be hydrophilic, i.e. hydrophobic segments should be avoided in the polymer structure where possible. In general, the hydrophobic methyl groups are introduced when methacrylic acid is utilized instead of acrylic acid, or when the PEG side chains are terminated with methyl instead

32 of hydroxyl. In addition, the impact of methyl-terminated side chains is more pronounced when the PEG chain is very short. Plank and coworkers [68] reported a model which explained the flow characteristics of concrete and cement, namely a flow speed increasing effect. It was shown that the hydrophilic-lipophilic balance (HLB) value greatly influenced this property. This study demonstrated that the dispersing force, as the

PCE dosage to reach a certain spread value, was independent of the HLB value of the

PCEs. However, the dispersing of a PCE appeared to derive from its adsorbed amount as well as the thickness of adsorbed polymer layer. These findings are important for actual field applications of cements and concretes where the fast flow speeds can allow more rapid placement and complete filling of the congested frameworks. Moreover, this study suggested that the hydrophobic interactions which cause high mortar stickiness could be based on a mechanism which was independent of particle dispersion. In addition, it was found that the differences observed for the flow speed of mortars resulted from the different plastic viscosities, and those were directly linked to the HLB values of PCEs.

The amide-structural PCEs were synthesized by amidation between polyacrylic acid (PAA) and amino-terminated methoxy PEG under different conditions [69]. The effects of synthesis conditions on the reaction rate and flow performance of the cement pastes were studied. FT-IR, NMR, and molecular-weight measurements were used for characterization, and the results confirmed the ideal amide structure and amidation reaction. PCEs with the carboxyl/amino ratio of 4:1 showed the highest adsorption percentage, lowest surface tension, and the good paste fluidity results. Although the application performance in concrete exhibited that this amide-containing PCE had similar slump to that of conventional PCE, it showed better air-entraining ability and bubble

33 retention than those of conventional PCE. Based on these good performances, the amide-

PCE can show broad application prospects.

The effects of polymer latexes on the properties of cement pastes and the interaction between cement and latex particles have been already investigated.

Xiangming et al. [70] utilized semi-batch emulsion polymerization to produce polymer nanoparticle (PNP) dispersions having the particle sizes around 30–50 nm. Polystyrene dispersion was selected as a model to study the possibility of using PNPs as cement dispersant similar to the PCEs. After the copolymerization of styrene with AA and methoxy polyethylene glycol methacrylate (MPEGMA), the polystyrene rich part could form the major part of the NPs, while the PAA and MPEGMA rich polymers stayed at the interface between the NP and the aqueous phase because of their high hydrophilicity.

The influences of the structural factors of the synthesized PNP dispersions on the fluidity of fresh cement pastes were investigated. Moreover, the interaction of PNP dispersion with cement was studied specifically by adsorption, isothermal calorimetry, zeta potential, Transmission Electron Microscopy (TEM), and Mercury Intrusion Porosimetry

(MIP). The adsorption of PNPs on the cement pastes was studied by Total Organic

Carbon (TOC) test on the supernatant which was separated by centrifuging the cement pastes. Fig. 2.17 shows the TEM images of the developed PNPs.

The thickness of adsorption layer of PNPs having the particle size of ~30 nm was remarkably larger than that of PCE, which was reported as ~3–10 nm [71]. Because of the very small size of PCE molecules, it is hardly possible to directly observe the adsorbed layer of PCE on the cement surface, whereas the synthesized PNPs were usually observable by SEM or TEM methods. As shown in Fig. 2.18, the morphology of adsorption layer of PNPs onto cement surface was directly observed by TEM. During

TEM observation, the Energy Dispersive X-ray (EDX) spectroscopy was applied to determine the elemental composition in selected ranges. It was also demonstrated that the adsorption density of PNPs onto the surface of aluminates was higher than that onto the surface of silicates.

Recently, Pourchet and coworkers [72] investigated the fluidity evolution of the inert granular systems as a function of the adsorption of SPs at low adsorption amounts.

In fact, the inert system could be well adapted to control the intrinsic fluidizing performance of comb-like SPs. In general, this inert model paste was composed of the calcite suspensions in a pore solution containing sulfate ions and saturated with regard to calcium hydroxide. This pore solution allowed mimicking the adsorption competition between SPs and sulfate ions. The originality of this work was in the use of a wide variety of superplasticizers. The polymers developed in this study were synthesized with various side chain densities and lengths as well as the modified anionic functions, such as carboxylates, dicarboxylates, and phosphates. It was reported that the substitution for phosphate function showed some original results. With PEG side chains of 1100 g/mol, the phosphate-containing polymer was less effective than the carboxylate-containing polymers. However, the phosphate-containing polymer exhibited the best fluidizing efficiency with PEG side chains of 2000 g/mol.

In another study, Pourchet’s research group studied the effects of the PCE structural parameters on their adsorptions in order to improve the PCE resistance to sulfate competitive adsorptions, hence minimizing their dosages [73]. In general, the PCE adsorptions were investigated on ettringite and calcite which are considered as the surfaces representative of the early hydrating cement. The effects of PEG side chain density and length as well as the PCE anionic function, such as carboxylates, dicarboxylates, or phosphates on the adsorptions were analyzed. In general, modification of the anionic was a good technical way to increase the resistance to sulfate competition.

At an equivalent charge density, the dicarboxylate-containing polymer was less sensitive

36 to sulfate concentrations than the monocarboxylate-containing polymer, while the phosphate-containing PCE was insensitive for the studied concentrations.

Zhang et al. [74] investigated the impacts of charge characteristics of PCE copolymers on their adsorptions and retardation effects on the cement hydration. The copolymers with variation of charge species were developed by copolymerizing a macromonomer with selected cationic and anionic monomers. Adsorptions and impacts on the cement hydration of the monomers including AA, sodium p-styrene sulfonate

(SSS), and [3-(methacryloylamino)propyl]trimethyl ammonium chloride (MAPTAC), as well as their corresponding homopolymers and copolymers were studied in cement pastes by means of TOC and calorimetry tests, respectively. The results showed that in the cement pastes, the retardation and adsorption were not found for the monomers while for their corresponding homopolymers, different extents of adsorption and retardation were observed. The charge characteristics of copolymers strongly determined their adsorption

‒ ‒ + and retardation effects, in the order of ‒COO > ‒SO3 > ≡N ‒. Fig. 2.19 shows the chemical structure of the monomers used in this study. In addition, the structures of copolymers containing both the carboxyl and sulfonic groups are shown in Fig. 2.20.

O Cl OH H C N N 2 O S O H O ONa CH3

(a) (b) (c)

Fig. 2.19. Chemical structure of the monomers used in the study of Zhang et al. [74], (a) AA, (b) SSS and (c) MAPTAC.

O O O O O S OH OH O N n H O

H2 H H2 H2 C C C C C CH a b c HO O O O HN O

O O S O n OH Fig. 2.20. Synthesis of the PCEs containing both the carboxyl and sulfonic groups [74].

In addition to being adsorbed on the surface of positively charged aluminate, PAA was able to be adsorbed on the surface of negatively charged silicate through the bridging calcium ions because of the complexation between these ions and ‒COO‒ groups as illustrated in Fig. 2.21. In contrast, poly(sodium-p-styrene sulfonate) (PSSS) did not have such an effect with calcium ions and the adsorbed amount of PAA was higher than that of

PSSS. Moreover, the adsorption of PAA on the surfaces of negatively charged mineral via bridging calcium ions led to less reduction of the zeta potential values of cement pastes because of the coverage on the anionic surface as well as the extension of shear plane. Besides, the adsorbed PAA molecules could further catch calcium ions in solution to create complexes, which could also weaken the ability of PAA to decline the zeta potential values of the cement pastes to some extent.

It is well known that the calcium ions are the most abundant ions in cementitious systems. In general, they are dissolved in the constituent of interfaces and the aqueous phase of cement pastes [75, 76]. Quite recently, Sowoidnich and coworkers [77] reported that the calcium ions are complexed due to the anionic charges in the functional groups of polymers. Moreover, it was shown that the calcium ion/polymer complexes dispersed in the aqueous solution could grow to larger aggregates as the hydration progressed.

Therefore, this work provided a comprehensive analysis of the polymers in cementitious systems through separating into calcium ions at the interfaces and in aqueous phases. It was reported that the complexation of calcium ions in the solution was mostly dependent on the type and number of functional groups of the polymers. In fact, the interaction of polymers with the calcium containing interfaces was more complicated. At low adsorption levels, the adsorption of superplasticizers on the surface of cement and tricalcium silicate (C3S) depended on the type of anionic charge. Thus, the adsorption of

39 polymers containing sulfonic groups exceeded that of polymers having carboxyl groups.

After the establishment of first adsorption layer, the presence or absence of PEG side chains and the type of grafting of PEG side chains were critical for further adsorption.

This behavior was generally attributed to differences in the conformation of polymers on the surfaces. Similar trends were observed by comparing the adsorption of polymers on cement and C3S. These results indicated that the structure of polymers had strong effect on the adsorption behavior on C3S.

In general, the chemical nature of PEGs can offer desirable solutions in formulating many poorly water soluble compounds as well as potential challenges to chemical and physical stability of the compounds formulated with PEG-based systems.

The nonlinear polymers based on (meth)acrylated PEGs combine a unique set of advantages and are utilized in modern technological applications, such as associative rheology modifiers (thickeners), superplasticizers for cement and concrete, and membranes for lithium ion batteries. Development of polymers based on (meth)acrylated

PEGs for industrial applications is a growing field of research having great promise for the future. With the plethora of PEG-based monomers available today, high extent of polymerization methods, and choice of mechanisms, the ever growing field of nonlinear polymers derived from (meth)acrylated PEGs has broadened the options for industrial applications during the last decade. The increasing requirements for these applications mean an ongoing search for novel polymeric materials with improved properties, which can expand the range of upcoming applications. In addition, with the new reagents and techniques in hand, the exploration of macromolecules based on (meth)acrylated PEGs is at its height, and scientists can be excited regarding the developments still to come.

2.4. Electrical contact stabilization materials based on PEG/PPG

In all electrical contacts, co-operating contact surfaces can never be flat perfectly.

In fact, the contact surfaces appear rough with many irregular peaks when observed under high magnification. Different parameters such as impurities attached to the surfaces and manufacturing processes contribute to these surface roughness [78, 79]. The electric current transmission from one surface to another can only take place at the parts where there are ideal physical contacts between the surfaces [80, 81]. Such surface roughness generally results in the contact resistance, which creates negative effects to a variety of applications [82].

In high current applications, there may be deformation resulting from heat induction [83] or chemical reaction [84, 85] because of the contact resistance. Radio frequency or audio frequency connectors, power scale video cable connectors, etc. are considered as the micro-power current applications. In such systems, the contact resistance can result in a discontinuity in signal flow when the signals are alternating complex waveforms and the voltage potential between contact surfaces changes from positive to negative potential or vice versa. This discontinuity in the alternating signals transmission can also add artificial signals to transmission in the radio frequency circuitry or even amplify the noise level, causing incorrect data transmission, program crashes, and cyclic redundancy error in computer circuitry [86].

A variety of polymeric materials have been used on electrical contact surfaces to minimize the contact resistance and increase the signal transmission [87-89]. Among these polymers, the PEG-b-PPG based systems are commercially used because they can

41 successfully eliminate most discontinuity between co-operating electrical contact surfaces and zero-crossing distortion [86]. It has been reported that these fluid materials can reduce the contact resistance due to this fact that they can easily fill the gaps between co- operating electrical contact surfaces [86]. However, these polymers remain in the liquid state after applied at room temperature. Therefore, they can significantly leak or flow so that their applications are highly restricted. On the other hand, the conduction mechanism of these materials is not well understood yet. Hence, developing a PEG-based material which can be uniformly and permanently applied onto the entire surface of co-operating systems is highly demanding.

Much attention has been recently devoted to the amount of solvent which enters the atmosphere from solution based coatings. Ultraviolet (UV) technology is an instantaneous and effective conversion of a liquid formulation having low viscosity and containing ingredients which are applied to a substrate, into a cross-linked mass by exposure to the radiation source [90]. In UV curing process, the chemical formulations including polymers, reactive diluents, photo-initiators (PIs), and additives can provide targeted properties of the final products. Moreover, a proper selection of the components is needed to enable an efficient curing process. In general, reactive diluents are used to decrease the viscosity of resins to improve processability in various applications. They are utilized in formulating solvent-free coatings to optimize performance properties such as adhesion, flexibility, impact strength, and filler-loading. It has been already reported that reactive diluents have significant effects on the curing behavior and properties of the resulting coatings [91-94]. In fact, the type, functionality, molecular weight, and content of reactive diluents are considered as the most important factors in determining the

42 mechanical [95] and physical properties [94, 96]. Furthermore, the overall response of a system containing reactive diluents to the radiation is greatly affected by the type of reactive diluents used. For example, the reactive diluents with higher degree of unsaturated segments generally have faster responses to the radiation curing compared to the reactive diluents having lower unsaturated functionalities [92, 93]. Hence, the curing speed can be adjusted by the proper ratio of saturated to unsaturated segments in reactive diluents.

Dynamic rheology [97-101] and Fourier transform infrared (FT-IR) spectroscopy

[101-103] are two well established methods that are frequently used to study the kinetics of a UV curing process. The evolution of the UV-crosslinking reaction and the transition from a viscous liquid to viscoelastic gel can be tracked through continuous measurements elastic (G’) and viscous (G”) modulus. [104, 105]. FT-IR spectroscopy has long been used to monitor the specific chemical changes during UV curing by following the absorption bands identified with the reacting functional groups. [106, 107]. In addition, the kinetics of UV curing process can be precisely determined by real time FT-IR spectroscopy technique [101-103, 107].

Following these considerations, a UV-curable electrical contact stabilization material based on PEG/PPG end-capped with methacrylate functional groups can be designed and synthesized. Mono- or di-methacrylated PEG-based reactive diluents can also be utilized for gel-like formulations to be consistent with the synthesized electrical contact stabilization material. A combination of real time rheology and FT-IR spectroscopy can be used for studying the effects of functionalities, molecular weight, and content of these reactive diluents.

CHAPTER III

COMB-LIKE SUPERPLASTICIZERS: SYNTHESIS,

CHARACTERIZATION, AND CONDUCTION CALORIMETRY

3.1. Introduction

The creation of high performance concretes needs the use of organic admixtures.

Among these materials, the dispersing agents are mostly used to improve the rheological properties of the materials before their applications without harming their final properties in terms of mechanical resistance. As already mentioned, the most successful of these admixtures in the market are superplasticizers (SPs) which can act as dispersing agents of cement during the moisturization of concrete. In general, these polymeric compounds are comb-like copolymers incorporating methacrylic acid (MAA) segments and poly(ethylene glycol) (PEG) side chains. Although it is not fully understood yet, the dispersion mechanisms of these systems probably operate via adsorption/defloculation stages [108, 109]. It seems that the carboxylic acid substituents along the polymer backbone provide the adsorption of polymeric materials onto the mineral phases of cement while defloculation step is ensured by the electrosteric or steric repulsion provided by the PEG side chains.

Industrially, these copolymers are produced either through grafting of mono- hydroxylated PEGs onto the preformed poly(methacrylic acid) (PMAA) or through direct

44 copolymerization of MAA and PEG-based macromonomers. In fact, whatever the approach used, the conventional free radical polymerization is utilized either to synthesize the PMAA backbone or to create the graft copolymer. Hence, the final copolymers show broad distributions of molar masses and the compositions in MAA and

PEG segments along the polymer chains are hardly controlled. Although several research studies have been performed [110-112], these ill-defined macromolecular structures cannot serve as good models to understand the accurate mechanisms operating for these superplasticizers. An accurate and effective control of chemical structure in terms of molar mass and distributions of monomers can provide a broad architectural freedom. In addition, this strategy can put the corresponding behaviors as dispersing agents into perspective.

Compared to the classical techniques of living cationic and anionic polymerization reactions, the emergence of living/controlled radical polymerization methods has brought new benefits to the synthesis of polymers [113]. These techniques have also simplified the production of tailor-made polymers via increasing the chemical diversity of beneficial hydrophilic monomers. Among available living/controlled radical polymerization methods, the reversible addition-fragmentation chain transfer (RAFT) techniques is mainly appealing because of this fact that the reaction conditions are similar to the industrial conventional free radical processes [114, 115].

Several RAFT (co)polymerization of low molar mass PEG-based monomers have been reported in the literature [116-120]. However, only few studies [121, 122] had investigated the controlled radical copolymerization of MAA and MPEGMA, until the recent studies by Rinaldi et al. [54]. This was probably due to the difficulty of removing

45 unreacted MPEGMA macromonomers from the synthesized (co)polymer compared with the lower molar mass PEG-based macromonomers. In order to overcome this problem, high MPEGMA conversions should be reached in the polymerizations involving this monomer. To better understand the fluidification mechanisms of the graft copolymers in cement and concrete, Rinaldi et al. [54] performed a research devoted to the synthesis of copolymers of MAA and MPEGMA of relatively high molar mass (1100 g/mol) using the RAFT method. In general, a certain level of control of polymerization was confirmed by a linear increase of the molar mass and low polydispersity index (PDI) observed versus conversion. Hence, copolymers of MAA and MPEGMA with well-defined structures were successfully developed.

The mechanism and kinetics of hydration in different cementitious systems can be investigated using calorimetric and other thermal procedures. In fact, the heat evolution curves reflect the reactions and physical processes occurring in the hydrating systems and leading to transformation of a plastic cement paste into a hardened matrix. It can also give an answer regarding the acceleration or delay because of the additional components introduced into the hydrating system. At early age, the hydration of clinker phases is affected in the presence of superplasticizers. In fact, the heat evolution, hydration of different cements, and formation of hydration products are hampered in the presence of these materials. Superplasticizers can delay the hydration of cement, prolong the initial and final setting time, and slow down the heat evolution process. These admixtures are usually used in large-scale concrete production. They can also give the possibility to avoid the problems when unforeseen delays between mixing and casting of the concrete occur.

The main objective of this chapter was to properly link the structural factors of well-defined PCEs with their performances in cementitious systems in order to better comprehend the interaction mechanisms. Therefore, we designed and synthesized a series of well-defined polycarboxylate-ether based copolymers PCEs with systematic variation in PEG grafting density and length. The interactions between the synthesized PCEs and cement pastes as well as the origin of the diminished attraction forces caused by PCEs were investigated in detail. The calorimetric measurements were carried out to acquire a comprehensive understanding of the heat of hydration evolution for each SP-cement system.

3.2. Experimental section

3.2.1. Materials

An ordinary Portland cement with a Blaine surface area of 3550 cm2 g‒1 was used in this study. Methacrylic acid (Aldrich), methoxy poly(ethylene glycol) methacrylate

(MPEGMA) macromonomers (300 g mol-1, 500 g mol-1, 950 g mol-1, and 2000 g mol-1,

Aldrich), 4-cyanopentanoic acid dithiobenzoate (CPADB) (Aldrich, >97%) as chain transfer agent (CTA), deionized water (DI water), diethyl ether (EMD Millipore, anhydrous), deuterium oxide (D2O) (Cambridge Isotope Laboratories, Inc.), and sodium hydroxide (NaOH) (Aldrich) were used as received. Tetrahydrofuran (THF) (Aldrich,

HPLC, >99.9%) was distilled over Na in the presence of benzophenone. 2,2’-

Azobis(isobutyronitrile) (AIBN) (Aldrich, 98%) was purified by recrystallization from

47 methanol. All other chemicals, unless otherwise discussed, were reagent grade and used as received.

3.2.2. Synthesis and characterization of copolymers

The comb-like PCEs were synthesized by RAFT copolymerization reaction of

MAA with MPEGMAs macromonomers based on the modified reported procedure [54].

A representative procedure for the synthesis of PCE 3:1-19 is described as follows. MMA

(3.67 g, 4.30 x 10-2 mol), MPEGMA 950 g mol-1 (13.60 g, 1.43 x 10-2 mol), CPADB

(0.39 g, 1.395 x 10-3 mol) as a RAFT agent, and deionized water (20.00 g) as a solvent were mixed in a three-neck round bottom flask. Sodium hydroxide (NaOH) (1.00 g,

0.025 mol) was then added to the flask. The mixture was gently stirred by magnetic stir bar under the nitrogen purge in an oil bath at 85 °C. After 30 min, the polymerization was initiated by dropping AIBN (0.0178 g, 1.84 x 10-4 mol) as an initiator solubilized in 4 mL of THF via syringe pump within 3.5 hours and continued for another 20 h. The polymerization was terminated via immediate exposure to air and quenching with an ice bath. After cooling, the mixture was washed with diethyl ether (100 mL) for three times in order to remove unreacted reactants. Finally, the product was dried overnight in a vacuum oven.

Fourier transform infrared (FT-IR) spectroscopy was performed on a Thermo

Scientific Nicolet iS50 FTIR spectrometer using 512 scans at a resolution of 8 cm-1 in transmission mode. The compositions of the synthesized copolymers were determined using 1H NMR (Varian NMRS-500 nuclear magnetic resonance instrument) operating at

500 MHz using deuterium oxide (D2O) as a solvent. The molecular weight and polydispersity index (PDI) of the prepared copolymers were determined by gel permeation chromatography (GPC, Waters) using Bryce-type differential refractometer

(RI) as a detector equipped with TSKgel SuperMultipore PW-H (TOSOH) column, and polyethylene oxide (PEO) and polyethylene glycol (PEG) as standards for the calibration.

The samples were analyzed by using 75 wt. % aqueous solution of these salts; sodium azide (NaN3), potassium phosphate monobasic (KH2PO4), sodium phosphate dibasic dodecahydrate (Na2HPO4·12H2O), and sodium sulfate (Na2SO4). The flow rate was 0.5 mL min-1 and polymer concentration was 1 mg mL-1 solvent.

3.2.3. Conduction calorimetry measurements

The measurements were performed by an isothermal heat flow calorimeter at 20

°C. For each analysis, 6.00 g of cement was weighted into a flask and 2.91 mL of DI water was added (w/c=0.485). Then, the cement paste was mixed with a small stirrer for

3 min. The flask was capped and placed into the calorimeter. It is important to note that the initial heat peak (the first maximum after the addition of DI water) was not recorded because of the external mixing. The heat flow was recorded for 72 h. From the calorimetric curve, the onset of the acceleration period could be derived, which roughly corresponded to the initial setting determined by means of the Vicat needle test.

3.3. Results

3.3.1. Polymer synthesis and characterization

In order to investigate the effect of the PCEs structures (i.e. side chains grafting density and length), nine well-defined PCEs were developed by RAFT polymerization of

MMA and MPEGMA with the same polymerization degree (DP = 50). It was already demonstrated that the reaction conditions induced a random reactivity between the monomers [54]. Fig. 3.1 exhibits the general chemical structure of the superplasticizers synthesized in this study. The length of the side chains was set at values of ~4.54, 9.09,

19.32, and 43.18 and the chosen grafting degrees (y/(x+y) = XX %mol) were 14, 17, 20,

25, 33, and 50%.

H3C H3C

x y

HO O O O

O z CH3

Fig. 3.1. Chemical structure of the synthesized PCEs. The average ratios x:y were 1:1, 2:1, 3:1, 4:1, 5:1, 6:1, and z = 4.54, 9.09, 19.32, 43.18 for the polymers investigated.

All the resulting PCEs were characterized by FT-IR, 1H NMR, and aqueous GPC

(Waters). Fig. S1 (Appendix) shows a typical FT-IR spectrum of PCE 3:1-19. Moreover,

Fig. S2 (Appendix) shows a typical 1H NMR spectrum of PCE 3:1-19, in which all protons are in good agreement with the proposed structure. In general, the signals in the

50 range of ~0.70‒2.25 ppm were assigned to the ‒CH2‒CCH3 and ‒CH2‒CCH3 repeating units of the polymer main chain. The singlet of the –O–CH3 could be observed at ~3.24 ppm. The most intensive signals were observed around ~3.40–3.80 ppm representing the protons of –(CH2CH2O)– in the repeating units of PEG side chains. Moreover, the protons of ethylene oxide unit located between the ester group in the main chain and the ethylene oxide groups in the side chain (–OCO–(CH2CH2O)–) were observed at a slightly higher chemical shift around 3.98–4.25 ppm. The molecular weights of the developed superplasticizers were determined relatively to a calibration using PEO and PEG standards and their characteristics are indicated in Table 3.1. The GPC chromatograms of these well-defined PCEs are also shown in Fig. 3.2.

Table 3.1. Characterization of PCEs. The copolymer code "x:y-z” indicates the ratio of carboxylate (x) to ester (y) and the number of PEG monomers in the side chain (z).

a b c ‒ Copolymer Density of Length of Mn Mw PDI μmol COO / code side chains (x:y) side chains (z) g mol-1 g mol-1 g polymerd PCE 1:1-19 1:1 (50%)e 19.32 19,400 25,300 1.3 988

PCE 2:1-19 2:1 (33%) 19.32 14,400 18,800 1.3 1773

PCE 3:1-19 3:1 (25%) 19.32 12,300 15,000 1.2 2500

PCE 4:1-19 4:1 (20%) 19.32 10,600 12,700 1.2 3149

PCE 5:1-19 5:1 (17%) 19.32 9,400 11,400 1.2 3654

PCE 6:1-19 6:1 (14%) 19.32 8,500 10,300 1.2 4160

PCE 6:1-5 6:1 (14%) 4.54 4,700 5,700 1.2 7518

PCE 6:1-9 6:1 (14%) 9.09 6,100 7,200 1.2 5952

PCE 6:1-43 6:1 (14%) 43.18 13,500 17,800 1.3 2407 a Mn = number-average molecular weight b Mw = mass-average molecular weight c Polydispersity index = Mw/Mn d Calculated values refer to the number of COO‒ per grams of PCEs e Grafting degree% (y/(x+y))

7.00 PCE 1:1-19 6.00 PCE 2:1-19 PCE 3:1-19 5.00 PCE 4:1-19 4.00 PCE 5:1-19 PCE 6:1-19 3.00 PCE 6:1-5 PCE 6:1-9

2.00 PCE 6:1-43 LS, Aux (mV) LS, 1.00

0.00

-1.00 10.0 15.0 20.0 25.0 30.0 35.0 Elution time (min)

Fig. 3.2. GPC chromatograms of the synthesized PCEs.

3.3.2. Conduction calorimetry

The presence of PCEs generally retards the start of acceleration periods in cement pastes. In fact, it is assumed that the PCE molecules can preferably adsorb on C3A and its hydration products. However, they principally retard tricalcium silicate (C3S) hydration and delay the formation of C-S-H. Fig. 3.3 and Fig. 3.4 exhibit the influence of the PEG side chain density and length of the synthesized PCEs on the hydration heat development in the cement pastes. The number of COO‒ groups per grams of the PCEs was increased by decreasing side chain density and length in the polymer samples (Table 3.1).

12 Cement 10 PCE 1:1-19 PCE 2:1-19 8 PCE 3:1-19 PCE 4:1-19 6 PCE 5:1-19 PCE 6:1-19 4

Heat flow (J/(g.h)) 2

0 0 10 20 30 40 50 60 70 Hydration time (h)

Fig. 3.3. The effects of PEG side chain density on the retardation of hydration of cement pastes with 0.2% bwoc PCEs and w/c=0.485 measured by heat flow for 72 h. The settings were significantly delayed by increasing the number of COO‒ groups as a result of decreasing the side chain density in the synthesized PCEs.

12 Cement 10 PCE 6:1-5 PCE 6:1-9 8 PCE 6:1-19 PCE 6:1-43 6

Heat flow (J/(g.h)) 2

0 0 10 20 30 40 50 60 70 Hydration time (h)

Fig. 3.4. The effects of PEG side chain length on the retardation of hydration of cement pastes with 0.2% bwoc PCEs and w/c=0.485 measured by heat flow for 72 h. The settings were significantly delayed by increasing the number of COO‒ groups as a result of decreasing the side chain length in the synthesized PCEs.

As shown in Fig. 3.3 and Fig. 3.4, the settings were strongly delayed by increasing the number of COO‒ groups in the synthesized PCEs. The shoulders observed after the maximum heat evolutions were related to the times when calcium sulfate compounds were consumed and the remaining tricalcium aluminate (C3A) phases reacted to different phases such as monosulfate or calcium aluminate hydrates [61]. In the presence of PCEs, the hydration peaks are originated from the sum of simultaneous hydration processes in different clinker phases and hence, the interpretation of peak shape is difficult. In general, the smallest influence on cement pastes regarding the delay of the onset of acceleration period was observed for PCE 1:1-19. This indicated little interactions between the molecules of PCE 1:1-19 and the mineral phases of the cement paste. It was demonstrated that despite the presence of flexible carbon-carbon and carbon-oxygen bonds in the PEG side chains, the accessibility of COO‒ groups in the

PCEs with high side chain density was highly hindered. In fact, charge densities decreased by increasing the side chain density and length, hence strong interactions were not possible. In addition, PCEs 6:1-19, 6:1-5, and 6:1-9 having higher charge densities led to significant shifts of the main hydration peaks to the right and hence delayed the acceleration periods of the cement pastes. Decreasing the side chain density and length prolonged the setting times.

3.4. Discussion

3.4.1. RAFT copolymerization of MMA and MPEGMA in aqueous media

Some studies have employed methacrylate derivatives containing PEG side chains as co-monomers for the preparation of well-defined macromolecular architectures by controlled radical polymerizations. For example, the statistical RAFT copolymerization of MPEGMA and MAA have been reported by Schubert and coworkers [122], however, little is yet known regarding the behavior of high molar mass PEG macromonomers in a

RAFT polymerization process [121, 123]. In 2009, D’Agosto and coworkers [54] reported the RAFT polymerization of MPEGMA by means of cyanoisopropyl dithiobenzoate (CPDB), AIBN as initiator, and THF as solvent. In this system, the control of MPEGMA homopolymerization was confirmed through the linear increase of the molar masses, the low PDIs observed with conversion, and high chain transfer constant for CPDB. Then, the copolymerization of MPEGMA with MAA was performed in dioxane. High conversion was acquired with a good level of control and without any composition drift. When the copolymerization was performed in the presence of CPADB in water, a strong increase in the polymerization rate was observed with almost quantitative conversion which did not affect the control level of the resulting copolymers.

Since the present study aimed at the synthesis of copolymers of MPEGMA and

MMA in an environmentally friendly solvent for industrial applications, the method developed by D’Agosto and coworkers [54] with a reasonable level of control in water was utilized here. Hence, nine well-defined PCEs were synthesized by RAFT polymerization of MMA and MPEGMA with the same DP of 50 to investigate the exact

55 effects of the side chains density and length on the working mechanism of SPs in cement.

In the literature, several examples exhibited that the RAFT polymerizations of methacrylate-containing derivatives were controlled efficiently by cyanoisopropyl dithiobenzoate (CPDB) as a reversible chain transfer agent [54, 124-126]. In the present work, CPADB as a hydro-soluble CPDB equivalent and the most widely employed CTA in the RAFT synthesis of water-soluble (co)polymers was utilized. In this case, a suitably high pH of the reaction medium was adjusted to favor the solubility of CPADB which could sometimes invite the risk of hydrolysis [127]. A small amount of THF as a co- solvent for AIBN was also used. The copolymerization of MAA and MPEGMA mediated by CPADB was performed in water with AIBN as initiator.

A strong increase in the homopolymerization rate of PEG-based macromonomers was already observed in water instead of benzene as a solvent [128]. Using light- scattering measurements, it was demonstrated that the micellar organization of PEG- containing macromonomers through the polymerizable methacrylate segments, confined in a hydrophobic area, was the origin of this phenomenon. As shown in Fig. 3.5, it seems that the confinements in hydrophobic domains of MAA and MPEGMA can happen in water. These hydrophobic domains can be generated by the complexation of carboxylic acid functions of the MAA with PEG segments or by the clusters of MPEGMA driven via the methacrylate heads or both of them. This phenomenon would lead to better solubilization of CTA and AIBN in these domains, ready to ensure the control of copolymerization. It was already reported that whatever the concentrations of the monomers used in water, the polymerization rates were similar and much higher than those in other solvents, such as 1,4-dioxane [54].

MAA O

O O O n O MPEGMA

Fig. 3.5. Possible hydrophobic regions formed during the copolymerization of MMA and MPEGMA in water [54]. The hydrophobic domains created by the clusters of MPEGMA driven through the methacrylate heads (right) or by the complexation of carboxylic acid functions of the MAA with PEG segments (left). Adapted with permission from John Wiley & Sons, Inc.

3.4.2. Hydration of Portland cement and effects of superplasticizers

3.4.2.1. Hydration of alite

Although C3A is considered as the most reactive phase, the hydration of OPC is dominated by alite (C3S) hydration, because it is the main component of cement (~50‒

70%). The hydration of alite leads to the precipitation of C-S-H which is an amorphous phase with variable stoichiometry. The C-S-H phase is responsible for the development of cement and portlandite (Ca(OH)2 (CH)) as explained by Eq. 3.1:

2 C3S + 6 H2O → C-S-H + 3 Ca(OH)2 (Eq. 3.1)

In general, the hydration of alite can be considered as a six-stage process as shown in Fig. 3.6. These stages can be explained through different rate-limiting

57 mechanisms. Because of this fact that several aspects of these stages are still being debated, understanding the impact of superplasticizers on hydration process of cement is difficult.

Fig. 3.6. Different stages in the alite hydration following the Ca2+ concentration evolution as well as the associated heat release [49]. Reproduced with permission from Elsevier.

The first peak observed during stage 0 is because of the dissolution of alite, which is a highly exothermic process. This stage lasts a few minutes and is then followed by the first deceleration process in stage I. The reasons for this deceleration process are not fully understood yet and many controversial hypotheses have been discussed in the literature.

For example, one of the first hypotheses involves producing a protective membrane of hydrates on the alite surface which may inhibit further dissolution and hence slows down the hydration reactions [129]. This protective membrane can either convert into the more permeable stable hydrate or break because of the osmotic pressure. Stage II (induction period) usually corresponds to the latent time until the critical nucleation or the polymerization of silicates. These events mark the end of low chemical activity and the beginning of the acceleration period. However, some scientists believe that this induction 58 period is only a time in which the process of decreasing rate (stage I) is matched by the process of increasing rate (stage III) [130, 131].

As shown in Fig. 3.6, during the first stages, the concentrations of calcium and silicate ions increase till reaching a critical supersaturation level. In fact, at this critical level the hydrates nucleate and grow. Stage III represents an acceleration period, in which the main peak for heat release corresponds to the massive precipitation of CH and C-S-H responsible for hardening. This is actually coupled with the improved dissolution of C3S, which is responsible for the measured heat release. In this stage, the heterogeneous nucleation and growth of the hydrates can control the rate of hydration, which is dependent on the C-S-H surface area.

Scanning electron microscopy (SEM) images of the early hydrated C3S have already confirmed that C-S-H grows as well-defined needles outside the grains [132]. In fact, these needles form until reaching a complete coverage which corresponds to the maximum of heat release. Instead of producing new needles precipitating on the previously formed ones, the C-S-H precipitates at the interface between the originally formed needles and anhydrous grains, leading to slower growth mechanism (stage IV). At that point, the lower rate of hydration is generally explained via the slower diffusion of ions through the hydrates [49]. Although transition to the diffusion regime due to the denser layer of hydrate has always been suggested for the deceleration stage, it has never met all conditions to clarify the experimental observations. Finally, the limited transport process through the densifying microstructure is responsible for the lower heat release of stage V.

3.4.2.2. Hydration of Portland cement

Portland cement is mainly composed of silicates and aluminates. Hence, its heat release shows characteristics from the hydration of anhydrous phases of these components. The heat release of Portland cement can also be divided to different steps corresponding to the alite hydration. Similar to alite, the exothermic peak in stage I is because of the wetting of cement surface and the fast dissolution of anhydrous phases.

Moreover, the ettringite precipitates in the very first minutes due to the high reactivity of aluminates as well as the availability of calcium sulfate. A sudden slowdown of the reaction and induction period is then observed in stage II. As already discussed, the second peak corresponds to the precipitation of the products of silicates, with an increase in the heat release due to the associated increase in dissolution of C3S hydration (stage

III). Then, the reaction slows down (stage IV). During this deceleration, a small peak occurs which generally represents the sulfate depletion and corresponds to the faster precipitation of ettringite and higher dissolution of C3A. Stage V is considered as a period of low activity because of slow diffusion of species in the hardened cement [49].

3.4.2.3. Role of molecular architecture of PCEs on hydration of Portland cement

The addition of PCEs retards the start of the so-called acceleration period in cement hydration relative to unmodified cement pastes (Fig. 3.3 and Fig. 3.4). The maxima of the heat flow in conduction calorimetry indicates the onset of delayed hydration and the retardation time. The period of retardation depends on the density and length of the PEG side chains, whereby lower density of side chains, corresponding to

60 higher content of COO‒ groups per grams of the PCEs (Table 3.1), scales with longer retardation time. The least delay of the acceleration period was observed for PCE 1:1-19, in which only half the backbone contains carboxylate groups and the PEG side chains are comparatively long (Fig. 3.3). Upon grafting of only one ether side chain per every six methacrylate ions in PCE 6:1-19, the retardation was highest. Adsorption measurements

(Chapter IV) and molecular simulations (Chapter V) show that the interaction of the less ionic backbones such as PCE 1:1-19 with mineral phases of the cement paste was weaker compared to the strongly adsorbing and retarding polymers such as PCEs 4:1-19, 5:1-19, and 6:1-19. A high density of side chains decreases the conformational flexibility of the backbone and sterically hinders access of COO‒ groups to the C-S-H surface.

Similarly, longer side chains shorten the retardation time (Fig. 3.4). The retardation of the acceleration period shortens from PCE 6:1-5 to PCE 6:1-43 as the overall density of COO‒ groups per unit volume of PCE diminishes, the strength of adsorption weakens, and conformations of the acrylate backbone on the surface change from tilted to flat-on (see Chapter V). Longest retardation of all systems is observed for the most ionic PCE 6:1-5, even though all other polymer backbones are already 86% ionic in this PCE 6:1-z series (Table 3.1).

Mechanistically, it is thus proposed that the polymers adsorb onto the particle surface, retard tricalcium silicate (C3S) hydration, and delay the formation of C-S-H.

However, the entire hydration peaks originate from the sum of simultaneous hydration processes of different clinker phases, which contribute to the shape of the peaks.

Adsorption is most relevant for the major phase (C3S, ~55%) as well as for the reactive tricalcium aluminate phase (C3A, ~10%) and its hydration products. The shoulders after

61 the maximum heat evolution are likely related to the consumption of calcium sulfate (few

%) and reaction of remaining tricalcium aluminate (C3A) phases to monosulfate and calcium aluminate hydrates [61]. Dicalcium sulfate phase (C2S, ~25%) and calcium aluminum ferrite (C4AF, ~5%) do not hydrate on these time scales and can be considered inert.

3.5. Conclusions

Retardation by superplasticizers depends on their chemical nature, molecular architecture, dosage, and the time of admixture addition. In this study, comb-like polycarboxylate-ether based superplasticizers (PCEs) were successfully synthesized using controlled RAFT polymerization and studied in the Portland cement suspensions to explain the working mechanisms in cement hydration. In general, the retardation of cement setting depended on the density and length of the PEG side chains and correlated with the amount of the carboxylate groups per unit volume. In fact, a lower density of

PEG side chains as well as a shorter PEG side chains led to a maximum retardation and highest adsorbed mass. Although in the calorimetric measurements, the retardation typically prolongs the induction period, the effects on hydration are not only to delay the onset of main hydration peak. Superplasticizers can also change the maximum heat release and the slope of acceleration period. Such changes can lead to a completely different calorimetric signature in the hydration processes.

CHAPTER IV

COMB-LIKE SUPERPLASTICIZERS: FLUIDITY TESTS AND

ADSORPTION MEASUREMENTS

4.1. Introduction

In recent years, several efforts have been made to investigate which combinations of cements and superplasticizers can be compatible. However, these attempts failed partially due to the large diversity of performance measured with different superplasticizers. Moreover, cementitious materials cannot be completely mastered by constrained standards and the direct case-by-case testing remains the only solution under the suitable operative conditions. It is important to note that some efforts were also made to extend the common standards of superplasticizers, only to realize that they do not always work. On the other hand, some standard organizations already decided to utilize a specific cement as a reference, or a composite made of a mixture of three cements used in a particular place. However, these methods could only inform about a specific cement or mixture being used, and the results were not predictive of the behavior of a broader range of cementitious materials used in practice. Therefore, there is no other approach than direct testing of cement-superplasticizer combinations [133].

Because of this fact that concrete tests are demanding in terms of energy, materials, and time, many researchers have developed simplified testing techniques on

63 mortars which are easy and quick to perform. In fact, developments in rheology have recently established the important links between these simplified tests and the basic rheological properties [134]. In general, the most widely used tests are the flow spread, mini-slump, and Marsh cone tests. In order to evaluate how the rheological properties of cementitious systems change in the timeframe when cements are placed in practice, these test methods are typically performed during the 90‒120 min after mixing. Although the flow spread and mini-slump tests directly relate to the yield stress, the Marsh cone test is controlled mainly by plastic viscosity of the system if the yield stress is low enough. A reasonable correlation between the mini-slump and Marsh cone tests has been already found using poly(naphthalenesulfonate) (PNSs) [135]. The results also suggested that the plastic viscosity depended on the adsorption of superplasticizers, but to a lower extent than the yield stress. Indeed, it could support the idea that Marsh cone test may also be efficient in determining the saturation points of superplasticizers which are the dosages leading to the full surface coverage.

These above mentioned simplified tests are commonly performed on cement pastes rather than concretes. Consequently, it is worth mentioning that the differences between the rheology of these materials usually come from the presence of sands and coarse aggregates in concrete. The pastes between these aggregates are subject to high shear forces. Hence, if prepared alone, they must be mixed intensely to experience the similar shearing conditions as in the concrete. These issues are of even greater importance in the mixtures containing superplasticizers. In addition, the mortar matrix in concrete should retain the aggregates, the suspensions, and in turn, the paste in the mortar should sustain the sand grains [133]. If these conditions are not achieved, the segregation

64 of sand and aggregates occurs. Besides the issues related to the segregation and mixing energy, the composition of aggregates plays a key role. The presence of swelling clays has negative effects on the fluidity of cementitious systems containing PCE superplasticizers [136, 137].

It is frequently observed in cement and concrete that the actual dosages of superplasticizers used in pastes or mortars are overestimated, causing excessive bleeding, retardation, and segregation [133]. Hence, this could discourage operators from using these scaled-down test methods to study the performance of superplasticizers. However, experienced operators are able to overcome these inconveniences easily by adjusting the dosages in cementitious systems depending on the performances, starting from the fast and useful results obtained from those test methods. The use of flow test in the pastes to optimize the superplasticizer dosage can represent a relatively effective technique because operators would target the specific yield stress required to sustain the aggregates of the size to be utilized in concrete [138].

Understanding the interactions between cementitious materials and superplasticizers has been a concern in the field of cement and concrete. Although there have been many studies on this topic, some of them contain unclear points from the viewpoint of appropriateness and reproducibility of the analytical methods [139]. In general, it is well known that superplasticizers work as dispersants mainly after adsorptions on the surfaces of solid phases. This is already demonstrated by the simple observation that, although PEGs do not have dispersing effects, they can be utilized to work as dispersants through giving them the ability to adsorb on surfaces of cement particles via their attachment to the carboxylic groups in PCEs.

The working mechanism of numerous superplasticizers involve the adsorption of water-soluble compounds at solid-liquid interface. In general, the adsorption increases with the polymer dosage until the surfaces of solid particles are completely covered. As shown in Fig. 4.1, the fluidity increases and finally approaches a maximum level which is dictated by the adsorbed conformations of polymers [140]. In fact, the polymer affinity for the surface shows a critical effect on the amount of polymer which should be added before the surface is fully covered. For instance, a polymer having low affinity for the surface must be added in large amounts due to this fact that a large fraction of it remains in solution rather than being adsorbed on the surface. The critical dosage, as another concept with regard to polymer dosage, is defined as the dosage above which the dispersion effect is noticed [141]. However, the results obtained from this dosage are highly dependent on the experimental conditions, including the type of cement, type and chemical structure of the polymer, and water-to-cement (w/c) ratio.

Fig. 4.1. Critical and saturation dosages are shown with respect to the dispersing ability of a SP [140]. Adapted with permission from Elsevier.

It has been demonstrated that the loss of fluidity depends on the amount of the copolymer remaining in the solution [142]. This can be explained by considering the fact that a reserve of copolymer is available in the solution to adsorb on newly produced hydrates [143, 144]. The question of why copolymers which do not adsorb at first would adsorb later has two responses. The first reason is that if there is an equilibrium between the solution and surface of solid particles, the distribution of copolymers depends on the ratio of surface area to solution volume. Therefore, if more surfaces are formed, one part of the non-adsorbed copolymers moves onto the newly formed surfaces, even if the original surfaces are not fully covered. Moreover, the second reason is that during the initial hydration, most of the hydrates created are aluminates for which the copolymers have higher affinity. Thus, it can be considered that the aluminates (not silicates) may be at full surface coverage initially. As long as copolymers are available in the solution, they can widely adsorb on the aluminates, while remaining in the equilibrium with partially covered silicate surfaces. Another consideration is that the workability of superplasticizers is lost much faster in the systems reaching only partial surface coverage than those with full surface coverage [141]. This can be as a result of the ability of adsorbed superplasticizers to modify the chemical kinetics of the cement hydration and the associated formation of new surfaces which have negative effects on workability.

Adsorption is frequently reported in terms of adsorption isotherm through plotting the amount of adsorbed copolymers with respect to the remaining amount in solution.

This technique makes sense if the equilibrium is established between the surface and solution. At the low surface and solution concentrations, several theoretical approaches to the adsorption isotherms have already predicted that the adsorbed amounts should be

67 proportional to the concentrations in solutions [53]. In general, increasing dosages of the compounds having adequate affinity for the surface enriches the surface, while the concentration in solution remains low. In fact, rising concentrations at the surface results in the adsorbed copolymers to enter a semi-dilute regime in which the adsorbates start to show excluded volume interactions. In these situations, the copolymers reorganize themselves on the surface to minimize the free energy. This minimal energy regime corresponds to the adsorption plateau, in which the adsorbed amounts become independent of the concentrations in solution [141].

Different studies on the adsorption of superplasticizers exhibit that the fractions of adsorbed copolymers remain constant as their dosages are increased. These fractions can be different over a large range of dosages. It seems that this linear zone can be interpreted in two different ways. First, it may be related to the polydispersity of the copolymers, with only one fraction of the copolymers adsorbing strongly and the other hardly adsorbing. Second, the free site activity may remain constant because the number of sites occupied in the surface is small. In order to better identify the regions in which the adsorption is proportional to the dosage, it can be helpful to plot adsorption values versus dosages. Fig. 4.2 shows an example for lignosulfonate (LS) superplasticizers [71]. It is obviously shown that the initial part of adsorption is linear with dosage. More detailed analysis shows that the initial slopes are not identical and also lower than unity. In fact, only a fixed fraction of the LS is adsorbed during the initial stage, which likely results from the polydispersity effect. Above this linear zone, some LS polymers are at their plateau and others continue to adsorb slowly with increasing dosage.

Fig. 4.2. Adsorption of different LS polymers as a function of the added polymer [71]. Adapted with permission from Elsevier.

For PCEs, a similar situation is already reported as shown in Fig. 4.3 [71]. An initial linear zone can be observed as a function of dosage. In this case, the slopes are substantially lower compared to those of LS polymers. In addition, PCP-3 does not reach a plateau which is probably as a result of precipitation. At the end of linear regime, the amount of adsorbed polymer is far beyond what is expected if the Langmuir-type isotherm is followed. However, only a limited fraction of the PCEs adsorbs in the linear zone which can be due to the polydispersity effect. In fact, superplasticizers generally show a polydispersed structure, meaning these materials contain several fractions of different structures and/or molecular weights. This can clarify the discrimination which takes place during the adsorption between different fractions [61, 141].

Fig. 4.3. Adsorption of different PCEs as a function of the added polymer [71]. Adapted with permission from Elsevier.

An important problem in measuring the adsorption isotherms of superplasticizers is to identify whether a true equilibrium has been reached. In fact, if the adsorption energy of polymers is high, their rearrangement takes time at high surface coverage. In cementitious materials, the problem is exacerbated by this fact that the surface areas change with time because of chemical reactions, hence limited contact times may only be utilized. Another concern regarding cementitious systems is that the polymers do not uniformly adsorb on different phases in cements. In hydration products and pure cement clinker phases, the superplasticizers preferably adsorb on aluminate phases with positively charged surfaces than on silicates having negatively charged surfaces [65, 145,

146]. Consequently, understanding the interactions of superplasticizers with cement phases is very important for controlling the fluidity of cement pastes and concretes.

Molecular mass of polymers plays a very important role in the adsorption behavior of superplasticizers, regardless of their molecular structure and nature.

Therefore, adjusting the molecular mass of superplasticizers is an approach to modify the

70 performance of these materials. Although for LSs as the derivatives of natural products, this may be performed by ultrafiltration, for the synthetic polymers, it can be performed by means of modifications on synthesis conditions, including temperature, reaction time, and concentrations of monomers and CTAs. As mentioned earlier, superplasticizers are polydispersed in structure and size so that a discrimination can take place during the adsorption process [61]. Depending on the desired performance of PCEs, it is possible to change the chemistry and length of their backbone, side chain density and length, etc.

[58]. It is important to note that all these parameters can affect the adsorption, which in turn affect the rheological properties [61, 111, 147, 148].

4.2. Experimental section

4.2.1. Slump test

The performance of the synthesized copolymers (discussed in Chapter III) in

Portland cement was tested through measuring paste flow, time dependent slump loss, and the maximum achievable water reduction [149, 150]. A mini slump test based on

DIN EN 1015 was used to determine the paste flow. This test was performed as follows: the water-to-cement (w/c) ratio of the cement paste without copolymer was first set to provide a spread of 18±0.5 cm. At this w/c ratio, the dosages of the synthesized copolymers needed to reach a spread of 26±0.5 cm were determined. In general, the copolymer was dissolved in the required amount of water placed in a mixer. It is important to note that when the aqueous copolymer solutions were utilized, the amount of

71 water contained in the copolymer solutions was subtracted from the amount of mixing water.

In a typical experiment, the copolymer was dissolved in the mixing water prior to the cement addition. Then, 300 g of cement was added within 15 s to the water, stirred for

30 s, and then rested for 1 min without stirring and was again stirred for 30 s using the mixer. After the stirring, the cement paste was immediately poured into a Vicat cone

(height 50 mm, diameter 50 mm) placed on a glass plate and the cone was vertically removed. The resulting spread of the cement paste was measured twice, the second measurement was in a 90° angle to the first and averaged to provide the spread value.

For the time dependent slump loss behavior, 300 g of cement was mixed with the required amount of water as described in the method above. After each measurement, the paste was transferred back into the mixer. Before each subsequent measurement, the paste was again stirred for 30 s. The measurements were performed every 30 min and the total period of the measurements was 120 min (Fig. 4.4). For measuring the maximum water reduction achievable, the w/c ratio of the cement paste without the copolymer was set to provide a spread of 26±0.5 cm. Next, the dosages of the synthesized copolymers in the steps of 0.30% by weight of cement (bwoc) were added and the w/c ratio at which the paste showed a spread of 26±0.5 cm was determined. The maximum water reduction achievable with these copolymers was obtained when the w/c ratio could not be lowered any more despite further increases in the copolymer dosages.

1st Measurement

2nd Measurement After 30 min

3rd Measurement After 60 min

4th Measurement After 90 min

5th Measurement After 120 min

Fig. 4.4. Representative pictures for the time dependent slump loss measurements. The measurements were performed every 30 min and the total period of the measurements was 120 min.

4.2.2. Adsorption measurement

The dispersion of SPs in cement pastes is facilitated through the adsorption of SP molecules onto the surface of cement grains. In the present study, the total organic carbon

(TOC) of the solution was determined using the depletion method [55, 149, 151, 152].

The non-adsorbed portions of the copolymers remaining in the solutions at equilibrium conditions were determined by analyzing the TOC of the solution using a TOC analyzer

(Shimadzu, TOC-VCPH, Japan).

In a typical experiment, 16 g of cement and 7.76 g of DI water (w/c=0.485) and the amount of copolymer to be tested were filled into a 50 mL centrifuge tube. The mixture was shaken in an SHZ-C oscillator for 5 min at 300 rpm, and then centrifuged in an HC-3518 high-speed centrifuge for 10 min at 5000 rpm. A clear supernatant solution was collected using a membrane filter with a pore diameter of 0.22 μm. Then, the supernatant solution was diluted with DI water to a suitable concentration for TOC measurement. From the difference between the TOC content of the copolymer reference sample and the TOC content of the resulting supernatant, the adsorbed amount of copolymer on cement of per unit mass could be calculated. In general, the measurements were repeated three times and the average was reported as the final adsorbed amount. To correct the values of adsorption measurements for the organic content of the Portland cements, the TOC of pore solutions of the plain cement pastes were also taken.

Moreover, the TOC of the used DI water was measured. Hence, these background values were taken into account before calculating the consumed amounts of the synthesized

PCEs.

4.3. Results

4.3.1. Dispersion properties

Dispersion and dispersion maintaining abilities are considered as two important factors for evaluating the performance of superplasticizers. The dispersion maintaining ability is usually expressed by slump loss which is defined as the decrease in fluidity with time. In general, previous studies on superplasticizer systems revealed that PCEs with long side chains showed better dispersion and dispersion maintaining abilities. It seems that the dispersibility of PCEs is mostly attributed to the steric hindrance of the side chains [108, 111]. As reported by Roussel and coworkers [153], the mini slump test is an empirical test and a true rheological tool that characterizes an intrinsic property of the material instead of only measuring a density-dependent slump value and test geometry.

The dispersion force of the synthesized PCEs was determined using two different approaches in the present study [149, 150, 154]. In the mini slump test, a cement paste prepared at a w/c ratio of 0.485, which showed a spread of 18±0.5 cm, was utilized as a reference. As exhibited in Fig. 4.5 and Fig. 4.6, the developed PCEs having different

PEG side chains density and length dispersed the cement very well. Increased bleeding of the cement pastes occurred at dosages exceeding 1.2%, as a result of the higher fluctuation of the spread values. As expected, the purified PCE samples showed higher dispersion forces than the samples obtained in the synthesis without any purification. In addition, the slump losses of the cement pastes prepared at the w/c ratio of 0.485 and different PCEs dosages corresponding to the paste flows of 26±0.5 cm were determined

75 over a period of 120 min. The slump of all cement pastes decreased quickly in the first 30 min and slower subsequently.

40.0 PCE 1:1-19 (0.9% bwoc) PCE 2:1-19 (0.9% bwoc) PCE 3:1-19 (0.6% bwoc) 30.0 PCE 4:1-19 (0.3% bwoc) PCE 5:1-19 (0.3% bwoc) PCE 6:1-19 (0.3% bwoc) 20.0

10.0 Slump Slump flow (cm)

0.0 0 30 60 90 120 Time (min)

Fig. 4.5. The effects of PEG side chain density on the slump loss behavior of the cement pastes (w/c=0.485). The slump flow was determined using the same sample at times of 0 min, 30 min, etc. The fluidization ability was significantly decreased by increasing side chain densities.

40.0 PCE 6:1-5 (1.2% bwoc) PCE 6:1-9 (0.9% bwoc) PCE 6:1-19 (0.3% bwoc) 30.0 PCE 6:1-43 (1.2% bwoc)

20.0

10.0 Slump Slump flow (cm)

0.0 0 30 60 90 120 Time (min)

Fig. 4.6. The effects of PEG side chain length on the slump loss behavior of the cement pastes (w/c=0.485). The slump flow was determined using the same sample at times of 0 min, 30 min, etc. PCE 6:1-19 showed better fluidization than PCE 6:1-5 and PCE 6:1-9. In addition, PCE 6:1-43 with very long side chain exhibited significantly low fluidization ability.

For the water reduction achievable with PCEs, a cement paste was prepared at a w/c ratio of 0.630 which gave a spread of 26±0.5 cm and was utilized as a starting system. Subsequently, several pastes containing dosages of 0.30–1.20% bwoc of the synthesized PCEs at incremental steps of 0.30% were prepared. At each dosage, the w/c ratio showing a spread of 26±0.5 cm was determined and the percentage of water reduction obtained was calculated. The results are exhibited in Fig. 4.7 and Fig. 4.8. It was obvious that PCE 6:1-19 possessed a very high water reduction capability. For instance, at a dosage of 0.30% bwoc which is typical for the actual application of many superplasticizers, the water reduction was 21%. The maximum water reduction achieved with PCE 6:1-19 was 50% at 1.20% bwoc dosage. This value compared well with that obtained from already reported PCEs, confirming the excellent dispersion force of this

PCE superplasticizer.

0.7

0.6

0.5

0.4 PCE 1:1-19 0.3 PCE 2:1-19 W/C ratio W/C PCE 3:1-19 0.2 PCE 4:1-19 PCE 5:1-19 0.1 PCE 6:1-19 0.0 0.0 0.3 0.6 0.9 1.2 PCE dosage (% bwoc)

Fig. 4.7. The effects of PEG side chain density on the water reduction capability as a function of dosage. The water reduction capability was significantly decreased by increasing side chain densities.

0.7

0.6

0.5

0.4

0.3

W/C ratio PCE 6:1-5 0.2 PCE 6:1-9 PCE 6:1-19 0.1 PCE 6:1-43 0.0 0.0 0.3 0.6 0.9 1.2 PCE dosage (% bwoc)

Fig. 4.8. The effects of PEG side chain length on the water reduction capability as a function of dosage. PCE 6:1-19 showed better water reduction capability than those of PCE 6:1-5 and PCE 6:1-9. In addition, PCE 6:1-43 having very long side chain exhibited significantly low water reduction capability.

Although long PEG side chains were thought to improve the paste fluidity through steric repulsive forces [110, 111], their combination with high side chain density could prevent the adsorptions of the synthesized PCEs and the steric stabilization of the cement pastes. As shown in Fig. 4.5 and Fig. 4.7, the fluidization ability of the PCEs decreased significantly by increasing side chain densities (25%‒50%). PCEs 4:1-19, 5:1-

19, and 6:1-19 showed very good flow behavior at low concentrations. However, increase in the concentration of PCE 6:1-19 led to bleeding at lower concentrations than those of

PCE 4:1-19 and PCE 5:1-19. This observation was supported by the water reduction experiment, where PCE 6:1-19 showed better water reduction capability than those of

PCE 4:1-19 and PCE 5:1-19. As shown in Fig. 4.6 and Fig. 4.8, with the same side chain density (14%), PCE 6:1-19 having side chain length (z) of ~19 showed the best flow

78 behavior and water reduction capability compared to those of PCEs 6:1-5, 6:1-9, and 6:1-

43.

4.3.2. Adsorption of PCEs

The adsorptions for the synthesized PCEs are presented in Fig. 4.9 and Fig. 4.10.

As shown, for all PCEs developed in this study, the adsorption curves generally had the same trends in which the adsorbed amounts increased rapidly by increasing the PCE dosages to ∼0.4% bwoc. Within this range, the surfaces of cement particles contained more empty sites and were more active in nature. Hence, the adsorption ratios were relatively high. However, at the higher dosages the adsorption amount increased more slowly and gradually approached a constant plateau values which corresponded to the amounts of PCEs required for the monolayer coverage.

5.0

4.0

3.0

2.0

1.0

PCE 1:1-19 PCE 2:1-19 PCE 3:1-19 Adsorbed Amount Amount (mg/g) Adsorbed PCE 4:1-19 PCE 5:1-19 PCE 6:1-19 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 PCE dosage (% bwoc)

Fig. 4.9. The effects of PEG side chain density on the adsorption isotherms of cement pastes with w/c 0.50 containing PCEs after 15 min of hydration. In general, the adsorptions were increased with decreasing PEG side chain density.

5.0

4.0

3.0

2.0

1.0

PCE 6:1-5 PCE 6:1-9 Adsorbed Amount Amount (mg/g) Adsorbed PCE 6:1-19 PCE 6:1-43 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 PCE dosage (% bwoc) Fig. 4.10. The effects of PEG side chain length on the adsorption isotherms of cement pastes with w/c 0.50 containing PCEs after 15 min of hydration. In general, the adsorptions were increased with decreasing PEG side chain length. PCE 6:1-43 with very long PEG side chains significantly showed lower adsorption.

Assuming that PEG side chains are highly flexible, the charge density of backbone should be the significant factor which principally determines the adsorption ability of the PCEs. Previous studies on the cement-water-superplasticizers indicated that the PCE fraction having higher Mw was preferably adsorbed [61, 155]. Yoshioka and coworkers [145] reported that an increase in the C3A content of cement led to higher PCE adsorption and required higher concentrations of PCEs to reach the saturation. This could be an indication of favored PCE adsorption on aluminate phases which was also supported by a study of Zingg and coworkers [63].

In the present study, PCE 1:1-19 having the highest side chain density showed the lowest adsorption among the synthesized PCEs. This result fitted well with the obtained dispersion properties, where PCE 1:1-19 showed the lowest dispersion and dispersion maintaining abilities. In fact, low charge density of PCE 1:1-19 allowed only poor

80 accessibility to the anionic groups and therefore, adsorption was hindered. Molecular simulations also support the importance of surface binding of calcium acrylates units and lower attraction of PEG side chains (Chapter V).

Fig. 4.9 shows that the synthesized PCEs adsorbed increasingly with decreasing side chain density. However, the influence of side chain density was negligible for PCEs

4:1-19, 5:1-19, and 6:1-19 with PEG density lower than 25%. As shown in Fig. 4.10, although the influence of side chain length for PCEs 6:1-5, 6:1-9, and 6:1-19 was almost negligible, PCE 6:1-43 with very long side chain significantly showed lower adsorption and reached saturation at the lower adsorbed carboxyl groups. In fact, the longer side chains of PCE 6:1-43 had larger steric action radii than those of PCEs 6:1-5, 6:1-9, and

6:1-19 and hence, less number of macromolecules were needed to reach the similar fluidization states. Despite slightly lower adsorbed carboxyl groups of PCE 6:1-19 than those of PCE 6:1-5 and PCE 6:1-9, it possessed stronger influence on the workability of the cement paste. This could be attributed to the higher steric repulsive forces resulting from the longer PEG side chains of PCE 6:1-19. Moreover, PCEs having short PEG length usually shows higher tendency for intercalation and formation of organomineral phases which can counteract their dispersion abilities. This is especially the case where a high amount of C3A is largely combined with inadequate supply of sulfate during early hydration of cement.

4.4. Discussion

4.4.1. General features of molecular architecture and adsorption

Due to the entropy terms linked to the conformation changes of the polymer between the surface and solution, the energy balance which controls the polymer adsorption is not trivial [141]. Therefore, the analytical solutions are available in a limited cases for which the conformations can be well defined, usually using scaling laws. In fact, this is considered as the case for linear chains [156]. Beyond these limited cases, the molecular simulations are needed.

Molecular mass has an important role in the superplasticizer adsorption, regardless of the molecular nature. In general, the related reports can be found for various superplasticizers including PCEs [61]. Hence, adjusting the molecular mass of superplasticizers is a method to modify their performances. For synthetic superplasticizers, this approach can be performed through modifying the synthesis conditions, such as chain transfer agents, concentrations of monomers, and reaction time and temperature. Moreover, it should be considered that superplasticizers are polydispersed in structure and size so that a certain discrimination can take place during the adsorption process [61, 141]. Depending on the performance desired, the chemistry and length of the PCE backbone, the ionic groups with respect to the PEG side chains, and the length of the PEG side chains can be changed [58]. All these parameters affect the adsorption, which in turn influences the rheological properties of superplasticizers, as already evidenced in numerous papers [61, 111, 157, 158].

4.4.2. Experimental issues in adsorption measurement

Typically, adsorption measurements are performed by the solution depletion. This method involves preparing a suspension with a known amount of the admixture, separation of the aqueous phase from the suspension, stabilization of the solution against the precipitation, and measuring the admixture content. Multiplying the resulting concentration by the liquid volume provides the total amount of the non-adsorbed admixture [141]. In order to obtain the adsorbed amount, this value can be deduced from the total initial amount of the admixture. Although different choices can be made to report these results, it makes sense to report the results in terms of the surface coverage.

In this case, it is required to know how much admixture is generally adsorbed at the full surface coverage. However, this amount can be higher than the adsorbed one at the plateau in the presence of the competing adsorbing species.

One solution to the above mentioned problem is to use a reference molecule which can reach the true equilibrium plateau. However, rather than the mass ratio to express the surface coverage, it makes more sense to calculate occupied surfaces.

Particularly, this is necessary if the reference plateau is taken with another molecule.

Thus, it is required to have a model for the surface occupied by each molecule. If such a model or a good reference molecule are not available, the next best expression for the surface coverage is to report the mass of adsorbed admixture with respect to the solid surface. Consequently, the obtained surface coverage can be related to the real surface coverage using a numerical constant. As a last option, the adsorption can be simply expressed with respect to the initial mass of the solids. Although this provides the least physical insight, it is the most straightforward option.

4.4.3. Fluidity tests and reduction of the water-to-cement ratio

Two essential criteria for evaluating the performance of superplasticizers are the initial dispersion of cement particles (slump loss) as well as maintenance of fluidity in the following hours to enable processing of concrete. A correlation between lower grafting density of side chains and higher fluidity can be seen (Fig. 4.5). The increase in fluidity for higher carboxylate content in the backbone (from 1:1-19 to 6:1-19) is consistent with stronger adhesion of PCE to the surface, reduction of particle agglomeration, and the observation of longer retardation. However, a mixed correlation is found as a function of the length of side chains (Fig. 4.6). Apparently, PCEs with very long side chains (6:1-43) do not adsorb strongly, do not prevent particle agglomeration well, and result in a larger slump loss. Shorter side chains at the same dosage provide more sustained fluidity, yet the best performance is reached for intermediate chain length (6:1-19). If the side chains are further shortened, such as in 6:1-5, adsorption and hydration retardation can be stronger (Fig. 4.10), however, inter-particle forces are not reduced effectively by highly ionic polymer alone.

Prevention of particle agglomeration, and thus fluidity loss, appears to be most effective in the presence of a strongly adsorbing (ionic) as well as a nonionic polymer component (PEG) or nonpolar functional groups. These relationships are similar to previously proposed mechanisms of agglomeration prevention by organic molecules

(grinding aids) in dry cement [159] and in clay minerals [160, 161]. Eventually, a smaller number of macromolecules with greater radius of gyration containing nonionic side chains (e.g. PCE 6:1-19) reaches better fluidization states than a larger number of smaller, more ionic macromolecules (PCE 6:1-5) per surface area. Previous studies

84 suggested that PCEs with longer PEG side chains show better dispersion and dispersion maintaining abilities attributed to steric hindrance by the side chains [108, 111].

However, very long PEG chains at low degree of esterification (such as PCE 6:1-43) were not included in these studies. The new findings sample the polymer architecture more comprehensively and indicate an optimum length of PEG side chains for best dispersion.

Another significant processing property affected by the PCEs is the water reducing capability, which helps reduce porosity in concrete and is measured by the lowest feasible water-to-cement ratio. The possible water reduction follows the trend in dispersion for all PCEs (Fig. 4.7 and Fig. 4.8). A cement paste with a w/c ratio of 0.630, which gave a spread of 26±0.5 cm, was utilized as a starting system. Then, several pastes containing dosages of 0.30–1.20% bwoc of the synthesized PCEs were prepared in incremental steps of 0.30% bwoc. The w/c ratio showing the same spread of 26±0.5 cm was determined for each dosage. In the best case (PCE 6:1-19), the w/c ratio was reduced from 0.63 to 0.50 at a dosage of 0.30% bwoc that is typical for the practical application of many superplasticizers. The w/c ratio could also be further reduced from 0.63 to 0.32 at

1.20% bwoc. Accordingly, the typical water reduction was 21% and the maximum water reduction up to 50%. The data compare well with previously reported water reduction of

PCEs and confirm the excellent dispersion ability.

The molecular origin for the water reduction capability appears to be the same as for stronger fluidization ability. A high content of COO‒ groups in the PCE backbone is required for strong adsorption, as well as an intermediate length of the PEG side chains is necessary to act as a non-ionic organic spacer between cement particles. If the side chains

85 are too long, weaker PCE adsorption onto the C-S-H surface results. If too short, the ionic nature of the methacrylate backbone favors water inclusion and introduces undesirable agglomeration forces between the organic shells of the cement particles that reduce dispersion, fluidity, and increase the necessary water-to-cement ratio for processing

(while maintaining longest retardation).

4.4.4. Adsorption of PCEs

The trend in retardation of acceleration phase of cement setting correlates with the adsorbed amount of PCEs. Adsorption increases rapidly for dosages up to 0.4% bwoc for all systems as the surfaces of cement particles contain empty sites, below monolayer coverage, within this range and the attraction of available polymers is high. At higher concentration, however, the adsorbed amount increased more slowly and approached a constant plateau which approximately corresponds to monolayer coverage and partial bilayer coverage with PCEs. The orientation of the molecules, however, is shown to differ between flat-on and tilted acrylate backbones (Chapter V), explaining why PCEs of lower side chain volume (PCE 6:1-5) form thicker layers than PCEs of high volume of side chains (PCE 6:1-43 or PCE 1:1-19).

The charge density of the polymers is the principal factor that determines attraction to the calcium silicate hydrate surface. Higher charge density leads to a thicker adsorbed layer that prevents water penetration and delays the hydration reaction. A decrease in the density of nonionic side chains therefore increased the adsorbed amount of PCE (Fig. 4.9). The differences become smaller for low density of side chains, as seen

86 for PCEs 4:1-19, 5:1-19, and 6:1-19. Shorter side chains equally lead to a larger adsorbed amount due to a higher content of carboxylate groups per unit volume of polymer and increased intermolecular attraction of the polymer (Fig. 4.10). The largest gap is seen between PCE 6:1-43 and PCE 6:1-19 due to the significant difference in chain length, and smaller differences between PCEs 6:1-19, 6:1-9, and 6:1-5.

Consistent trends in retardation and adsorption were found in previous studies

[61, 155], and higher molecular weight for a particular PCE x:y-z tends to increase adsorption [52]. An increase in the tricalcium aluminate (C3A) content of cement can also increase PCE adsorption and require higher PCE concentrations for saturation [145]. C3A might thus preferably attract PCEs over C3S and C-S-H [63], possibly related to stronger ionic character [162] and better accessibility of the hydration products to calcium ions and acrylates.

4.5. Conclusions

It was fond that increase in the fluidity for higher carboxylate content in the backbone of PCEs was consistent with the stronger adhesion of the PCEs to the surface and longer retardation. However, a mixed correlation was found as a function of the length of the PEG side chains. In fact, PCEs with very long PEG side chains did not adsorb efficiently and resulted in a large slump loss. Although shorter side chains provided more sustained fluidity at the same dosage, the best performance was reached for the intermediate chain length. A high content of COO‒ groups in the PCEs was necessary for strong adsorption, and an intermediate length of the PEG side chains was

87 required to act as a non-ionic spacer between cement particles. The charge density of the polymers is the main parameter that determines attraction to the C-S-H surface. Higher charge density of PCEs led to a thicker adsorbed layer which could prevent water penetration and delayed the hydration reaction. Consequently, a decrease in the density of

PEG side chains increased the adsorbed amount of PCEs. However, these differences became smaller for low density of PEG side chains. Shorter PEG side chains led to a larger adsorbed amount because of a higher content of COO‒ groups per unit volume of polymer and increased the intermolecular attraction of the polymers. The largest gap was seen between PCE 6:1-43 and PCE 6:1-19 because of the significant difference in their chain length, and smaller differences were seen between PCEs 6:1-19, 6:1-9, and 6:1-5.

CHAPTER V

COMB-LIKE SUPERPLASTICIZERS: WORKING MECHANISMS

5.1. Introduction

The effect of superplasticizers on cementitious materials comes from introducing repulsive interparticle forces which decrease the overall attractive forces between cement particles. Dispersion (van der Waals( forces arise because the local fluctuations in polarization within a particle induce a correlated response in the other one [163]. This interaction is attractive for two particles created by the same isotropic matter. Due to the correlation of the fluctuations of dipoles in condensed media, the range of interparticle dispersion forces is larger than that of individual dipoles [164].

In water, solid surfaces are mostly charged because of this fact that water is a good solvent for ions due to its high dielectric permitivity. In fact, the surface charges arise from the specific adsorptions of ions or ionic macromolecules and/or the dissociation of surface groups. At a charged interface, the distribution of ions is described very well by the Gouy-Chapman model. This double-layer model distinguishes between two different zones, the Stern (inner) and diffuse (outer) layers. The Stern layer is based on the counter-ions immobilized through the surfaces of particles and the diffuse layer consists of mobile ions. The concentration of counterions is generally higher close to the

89 surfaces than that in the bulk solutions. However, the ions in diffuse layer exhibit an opposite behavior [164].

Ions in the double layer give rise to a specific electrical potential which decays exponentially with the distance according to Eq. 5.1 [164]:

‒KX ΨES = Ψ0 e (Eq. 5.1)

where Ψ0 is considered as the surface potential and the decay length, also called as the

Debye length, is given by K‒1 which is defined based on Eq. 5.2:

‒1 휖휖0푘퐵푇 K = √ 2 2 (Eq. 5.2) 푒 ∑푖 푐푖푍푖

where ε is considered as the relative dielectric constant of water, ε0 is the permittivity of vacuum, kB is the Boltzmann constant, T is the absolute temperature, e is the charge of electron, and c is the bulk concentration of the ith electrolyte of the valence Z.

The Debye length, as the diffuse layer thickness, represents distance between the surfaces within which there is a significant repulsive potential. As shown in Fig. 5.1, when two surfaces having identical charges approach, the electrical double layers overlap and develop an excess ion concentration. In fact, this phenomenon results in an osmotic pressure which calls for dilution of ions between surfaces through drawing water from the bulk solution. Hence, the surfaces must move apart with the presence of more water between particles. Therefore, the electrostatic repulsive forces do not need physical contact between particles.

Fig. 5.1. Schematic representation of the electrostatic repulsion. When the charged surfaces approach, the double layers overlap and produce an excess ion concentration, creating an osmotic force which resists the approach of charged surfaces [164]. Reproduced with permission from Elsevier.

As shown in Fig. 5.2, the Debye length generally decreases with increasing the ionic strength. At the high salt concentrations, screening of the surface charges by means of ions in solution is more efficient, and the range of electrostatic repulsion is mostly decreased. This is very important in cement suspensions having high ionic strength in the range of 100‒200 mM [165]. Consequently, the electrostatic repulsion in cement suspensions is only felt at the short distance. The attractive van der Waals forces are predominant in this range. It is important to note that the magnitude evaluation of the electrostatic force in cementitious suspensions is complicated because these systems do not behave ideally and the ion activities cannot be approximated to concentrations [166].

Moreover, the divalent calcium ions are able to induce attraction forces between the negatively charged surfaces of the C-S-H. In fact, this effect can be important with time,

91 due to the increasing number of the C-S-H contact points when the hydration proceeds

[167, 168].

Fig. 5.2. Variation in electrostatic potential is shown as a function of the ionic strength of the continuous phase and of the distance from the surface of a colloidal particle [164]. Reproduced with permission from Elsevier.

It is important to note that the surface and Stern potentials cannot be measured directly. In fact, zeta potential is the experimentally determined potential. Zeta potential is considered as the potential at shear plane and lies in the diffuse layer further from the surface than Stern plane [164]. Consequently, the use of zeta potential can probably underestimate the values of the electrostatic force because of the intrinsic limitations in the zeta potential measurements (Fig. 5.3). Microelectrophoresis is one of the most common methods for measuring the zeta potential values. However, this technique needs the use of highly diluted suspensions which are not the representatives of real cement pastes. Moreover, the particles tend to settle when their sizes are larger than a few micrometres in diameter, hence only very fine particles can be measured. The second

92 method for measuring the zeta potential is electroacoustics which can be used in concentrated suspensions to monitor the interactions between superplasticizers and cement [65, 169].

Fig. 5.3. A general schematic representation of the effect of an adsorbed chemical admixture on the potential of a charged surface [110]. Adapted with permission from Elsevier.

5.2. Experimental section

5.2.1. Zeta potential measurements

Electroacoustic technique enables to determine the zeta potential of highly solids loaded suspensions [70, 170]. Such suspensions can truly represent the conditions existing in a mortar or concrete. In the present study, zeta potentials of the cement pastes were determined using a DT 1201 Electro acoustic Spectrometer (Dispersion

Technology, Inc., USA). This instrument can measure the vibration current induced by an acoustic wave which causes moving the aqueous phase relative to the cement particles. A potential difference, designated as zeta potential, is produced from that and can be measured. The pH-meter and zeta dip probe were calibrated. In order to prevent segregation, all samples were measured in a beaker and stirred (350 rpm). To insure the purity, the titration unit was washed several times prior to use with the titrant. In a preliminary test series, the cement pastes utilized in this study showed a stable zeta potential after 10 min of hydration. To investigate the exact impact of the synthesized

PCEs having different chemical structures on the zeta potential of the cement paste, concentration series with constant titration increments of the diluted PCE solution of

0.05% bwoc were measured. This experimental setup can measure the impact of delayed addition of PCEs.

5.2.2. Conductivity measurements

The raw data obtained for the zeta potential are affected by the fact that pore solutions of the cement pastes are strongly charged with several ionic species and charged polymers [63]. Thus, background measurements were performed in this study using the following method. The pore solution of each cement paste was first extracted by filtration and then diluted until the conductivity of the cement paste after 10 min was reached to ∼15–25 mS/cm. Because the conductivity of cement pastes increases with time, the background measurements of different conductivity levels from 15 to 25 mS/cm with increment of 1 mS/cm were measured. Then, the PCEs were titrated to the 18

94 mS/cm pore solutions. A background measurement was performed after each titration step. Subsequent to each experiment, the raw data collected by the zeta probe was recalculated through applying the analogous conductivity/PCE titration background files.

5.3. Results

5.3.1. Zeta potential

Studying the interparticle forces in the cement-superplasticizer interactions is very important. The surface charges, availability of counter-ions, and structures of the diffuse double layers control the behavior of approaching particles. The zeta potential can represent the potential at slip surface where a particle in motion separates from its surrounding liquid phase. In the present study, the cement pastes after 10 min reached a steady state with zeta potentials between ‒4.6 and ‒6.0 mV. These zeta potentials were in good agreement with values reported in the literature [63, 171]. The negative zeta potentials in these systems could be related to the negative zeta potential of calcium silicates as the major components in the cement. However, positive or both, negative and positive zeta potentials have also been reported for cement pastes [146, 158, 172-174]. In fact, the cement pastes used in those studies differed in the reactivity of their components, the composition of their pore solutions, and their mineralogical compositions. In general, most publications lack detailed information regarding the sample factors and the experimental setup utilized and therefore, direct comparisons of the values are very difficult. In this study, the zeta potential values for the plain cement pastes were found to be stable between 10 and 90 min of hydration.

As shown in Fig. 5.4, the titration of cement pastes of PCEs 3:1-19, 4:1-19, 5:1-

19, and 6:1-19 having higher charge densities led to the zeta potential shifts to the positive charges and reached the saturation concentrations below ∼0.2% bwoc. PCE 6:1-

19 reached its saturation concentration faster than other PCEs. A likely explanation is the preferred adsorption of Ca2+ ions, which neutralize the acrylate anions in the PCEs, to the particle surface. This mechanism is consistent with the observation in molecular dynamics simulation (section 5.4.1) and clearly explains the trend. It was also noted that

PCEs with higher charge densities adsorbed faster. For the effect of side chain length, it was observed that the zeta potential values of PCE 6:1-5 stayed negative, PCE 6:1-9 showed a zeta potential shift close to zero, and PCE 6:1-19 exhibited the most positive zeta potential among all the synthesized PCEs. PCEs with longer side chains can thus change the negative zeta potential values to positive values. Again, the adsorptions of

Ca2+ ions from the acrylate anions in the PCEs are stronger for longer side chains that lead to flat-on adsorption as shown by the molecular simulations (Fig. 5.9). Ca-acrylate thereby would emerge upon initial conversion of Na-acrylate with superficial portlandite

(Ca(OH)2), which is present in excess on the surface, to Ca-acrylate and NaOH.

An outlier is the negative zeta potential for PCE 6:1-43, however, a large part did not adsorb and remained in the pore solution. Although the weak adsorption of PCE 6:1-

43 molecules could be measurable with the zeta dip probe, they were too weak to resist the filtration procedures. In fact, the dominance of PEG side chains in the polymer reduces adsorption due to lower affinity of PEG to the C-S-H surface. Therefore, less

2+ Ca ions and acrylate are available for adsorption per unit mass to neutralize the zeta potential. An alternative explanation is that at the same weight concentrations, more

96 carboxylic acid groups are available with PCE 6:1-5 compared to those of PCE 6:1-19, and the carboxylate groups could cause larger negative zeta potentials for shorter PEG chains. This argument does not hold, however, as the zeta potential without PCEs is the largest negative value, therefore supporting the role of Ca2+ ions in the acrylate as drivers for adsorption and neutralization of the zeta potential.

4.0

2.0

0.0

-2.0 PCE 1:1-19 PCE 2:1-19 -4.0 PCE 3:1-19 PCE 4:1-19 zeta potential (mV) -6.0 PCE 5:1-19 PCE 6:1-19 -8.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 PCE dosage (% bwoc)

Fig. 5.4. Effects of the grafting density of PCE chains on the zeta potential of cement pastes with w/c ratio of 0.50 and different concentration of added PCEs. More ionic polymers lead to positive zeta potentials in the order PCE 1:1-19 to PCE 6:1-19 due to adsorption of more calcium ions from the calcium acrylate groups.

4.0

2.0

0.0

-2.0

-4.0 PCE 6:1-5 PCE 6:1-9

zetapotential (mV) -6.0 PCE 6:1-19 PCE 6:1-43 -8.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 PCE dosage (% bwoc)

Fig. 5.5. Effects of the length of PEG side chains on the zeta potential of cement pastes with w/c ratio of 0.50 and different concentration of added PCEs. The highest zeta potential was observed for intermediate chain length (PCE 6:1-19) and followed the trend in dispersion efficiency. The increasing zeta potential for higher dosage shows that Ca2+ ions in the acrylate are attracted to the particle surface upon PCE adsorption in all cases.

5.3.2. Conductivity

As shown in Fig. 5.6 and Fig. 5.7, in contrast to the stable values of the zeta potential of cement pastes, the conductivity increased with evolving hydration time. This parameter is mostly influenced by dissolution and formation of hydrate phases [63]. In these systems, the cement pastes consistently exhibited lower increase of conductivity with increasing the PCE contents throughout the preliminary titrations. In particular, PCE

6:1-5 and PCE 1:1-19 showed the lowest and highest increase of conductivity values with increasing the copolymer contents, respectively. This fact could be due to a thicker upright adsorption layer for PCE 6:1-5 (Fig. 5.8) versus a loose, less well adsorbed PCE

1:1-19 layer due to the dominance of weakly adsorbed PEG. The trends in Fig. 5.6 clearly show how dominance of acrylate lowers the increase in conductivity and dominance of

PEG increases the conductivity. Strength of adsorption and thickness of the adsorbed layer can act as a diffusion barrier towards and a conductivity increase.

22.0

21.0

20.0 PCE 1:1-19 19.0 PCE 2:1-19 PCE 3:1-19 PCE 4:1-19

18.0 PCE 5:1-19 Conductivity Conductivity (mS/cm) PCE 6:1-19 17.0 10 15 20 25 30 35 40 45 50 55 60 Time (min)

Fig. 5.6. Effects of the density of PEG side chains on the conductivity of the cement pastes containing PCEs at a w/c ratio of 0.50 as a function of time. The initial conductivities are higher for PCEs with larger acrylate content and stronger adsorption, related to availability of more ions per unit volume (PCE 1:1-19 to 6:1-19). The order inverts over time related to inhibition of surface dissolution and hydration processes by adsorption of the PCEs, then showing highest conductivity for the least ionic PCE with high density of side chains.

22.0

21.0

20.0

19.0 PCE 6:1-5 PCE 6:1-9 18.0

Conductivity Conductivity (mS/cm) PCE 6:1-19 PCE 6:1-43 17.0 10 15 20 25 30 35 40 45 50 55 60 Time (min) Fig. 5.7. Effects of the length of PEG side chains on the conductivity of the cement pastes containing PCEs at a w/c ratio of 0.50 as a function of time. The initial conductivity correlates with higher acrylate content and ionic strength (PCE 6:1-43 to 6:1-5). The order inverts over time as the more ionic PCEs (6:1-5) become more effective to inhibit surface dissolution and hydration, then showing lowest conductivity.

100

5.4. Discussion

5.4.1. Binding mechanism and conformations of the polymers

Molecular dynamics simulations were employed by our collaborators in the

University of Colorado-Boulder to probe the adsorption mechanism and binding conformations of PCEs to the calcium silicate hydrate surfaces for the first time according to our knowledge. Upon contact of the PCEs with the tobermorite surface, binding of some Ca2+ ions pertaining to the acrylate groups onto the tobermorite surface occurs as a first step (Fig. 5.8 and Fig. 5.9). The calcium ions coordinate siloxide groups

(≡SiO-) that retain lower calcium coordination numbers upon creation of a (hkl) surface than in the bulk material. Calcium coordination with the negatively charged mineral surface is consistent with neutralization, or reversal, of the negative zeta potential towards positive values (Fig. 5.4 and Fig. 5.5). The simulations also show for all systems that the acrylate backbone subsequently closely approaches the C-S-H surface while the

PEG side chains are rarely adsorbed (Fig. 5.8 and Fig. 5.9). This observation is consistent with higher adsorbed amount of PCEs with lower PEG content such as 6:1-5 and lower adsorbed amounts of PCEs with high PEG content such as 1:1-19 in experiment.

Adsorption energies are computed in a range of -10 to -40 kcal/mol per 16mer

PCE depending on the polymer architecture and on the composition of (hkl) tobermorite

14 Å surfaces (Fig. 5.8 and Fig. 5.9). The corresponding adsorption energy per monomer is from -0.6 to -2.5 kcal per mol acrylate monomer at ~35% surface coverage. The data are in excellent agreement with AFM measurements of adhesion of PCEs onto C-S-H, which translate to a range of -1.7 to -2.6 kcal per mol acrylate monomer (-1.0 to -1.5·10-

101

18 J per chain for a PCE 3.3:1-23 with DP = 83) [52]. The most closely related system in the simulation, PCE 3:1-9, has an adsorption energy of -2.5 kcal per mol acrylate monomer, demonstrating the predictive value of cement simulations using IFF (Fig. 5.8 b) [175].

In detail, adsorption of the calcium polycarboxylate ethers onto the calcium silicate surface occurs via migration of calcium ions to empty silicate sites on all cleavage planes, and is then followed by the formation of ion pairs between the COO‒ groups of the polymer and the Ca2+ ions on the mineral surface (Fig. 5.8). Major contributions to adsorption energies result from COO‒ ··· Ca2+ contacts within 3 Å distance [176-178].

The polyethylene oxide side chains avoid direct contact with the tobermorite surface. It is less favorable for ether oxygen atoms (charge -0.40e) in comparison to water oxygen atoms (charge -0.82e) to form part of the hydration of shell of calcium ions on the C-S-H surface, even though PEG could act as a multidendate ligand in crown ethers.

For high density of side chains such as in PCE 1:1-9, the acrylate backbone has limited conformational flexibility and intermittently adsorbs onto the silicate surface (Fig.

5.8 a). As the density of side chains decreases, the acrylate backbone gains conformational flexibility and spends up to 100% of time in direct contact with the tobermorite surface, for example, in PCEs 3:1-9 and 7:1-9 (Fig. 5.8 b,c). The binding energy, as a facet average, is strong for all side chain densities, up to -40 kcal/mol for a

16mer PCE at a surface coverage of ~35%. The strength of binding is related to the formation of ion pairs as well as hydrogen bonds (Fig. 5.8 d, e). As the ionic character of the polymers increases, the number density of polymers per surface area increases when higher surface coverage is allowed, supported by tilted conformations and stronger ionic

102 cohesion of the polymer film. The result is a larger adsorbed mass per unit area as seen in experiment.

The length of the side chains has a similarly strong effect on conformations (Fig.

5.9). PCEs with shorter side chains such as in PCE 7:1-3 orient the acrylate backbone tilted or upright relative to the surface, driven by favorable ionic interactions between neighbor molecules and ample conformational flexibility (Fig. 5.9 a). The adsorption energy per molecule is then weaker although side-by-side packing of multiple molecules can lead to much higher adsorption energy (Fig. 5.9 e), explaining a larger adsorbed mass per unit area in experiment. For somewhat longer PEG side chains in PCE 7:1-6, the orientation of the acrylate backbone was intermediate between upright and flat-on as the side chains begin to inhibit ionic interactions between neighbor backbones (Fig. 5.9 b).

For longer PEG side chains such as in PCE 7:1-9, the acrylate backbone preferred tilted or flexible flat-on orientations and strongly adsorbed onto the tobermorite surface with the PEG side chains oriented upward and away from the surface (Fig. 5.9 c). PCEs with longer side chains and larger molecular volume show larger negative adsorption energies due to binding of a higher number of COO‒ groups in the backbone (Fig. 5.9 b,c), whereby the backbone adjusts to available Ca2+ ions by characteristic rotations (Fig. 5.9 d). Direct surface contact is reduced and interactions between neighbor acrylate backbones are more difficult to achieve, which still explains favorable adsorption yet lower adsorbed mass in total.

103

a PCE 1:1-9 b PCE 3:1-9 c PCE 7:1-9 d

~35% surface coverage < 3 Å c CSH ··· Ca2+ ion pairs b CSH/Ca2+ ··· COO- ion pairs

ΔEads = -37 10 kcal/mol -41 5 kcal/mol -30 7 kcal/mol tcontact = 60-100% 90-100% 100% e

< 3 Å a SiOH ··· -OOC H bonds b Si O Ca C H 1 nm

Conformational flexibility of the backbone Number density of oligomers per surface area Adsorbed mass per surface

Fig. 5.8. Adsorption mechanism of polycarboxylate ethers onto a model C-S-H surface as a function of the density of side chains according to molecular dynamics simulation in water ((100) surface of tobermorite 14 Å). (a-c) Initial conformations of the oligomers in top view and the composition of one molecular layer of the calcium silicate surface, including Ca2+ ions, SiO- groups, SiOH groups, and water (upper panels). Equilibrium conformations in side view show strong binding for all polymers, yet the contact time of the polymer with the surface (closer than 3 Å) increases from PCE 1:1-9 to PCE 7:1-9 (lower panels). A high density of side chains (PCE 1:1-9) leads to less flexible conformations of the main chain and intermittent surface contact compared to low density of side chains (PCE 7:1-9). The more ionic PCE 7:1-9 assumes tilted, partly detached conformations that can increase the number density of adsorbed molecules at high surface coverage. (d) Major contributions to adsorption by binding of Ca2+ ions from the acrylate backbone to the mineral surface, followed by the formation of ion pairs (<3 Å distance) between the acrylate backbone and surface Ca2+ ions. PEG side chains are not attracted to the surface. (e) Hydrogen bonds between surface silanol groups and oxygen atoms in the carboxylate groups of the polymer backbone add to the binding energy.

104

a PCE 7:1-3 b PCE 7:1-6 c PCE 7:1-9

ΔEads = -11 4 kcal/mol ΔEads = -22 5 kcal/mol ΔEads = -30 7 kcal/mol

c 1 nm a O H Si Ca C

Horizontal orientation of acrylate backbone Less conformational flexibility Lower adsorbed mass per surface area

d e PCE 7:1-3

4 C-C bonds

Neighboring - - COO- groups + +

Fig. 5.9. Interactions of polycarboxylate ethers with a model C-S-H surface as a function of the length of PEG side chains according to molecular dynamics simulation in water ((001) facet of tobermorite 14 Å). (a-c) Equilibrium conformations in side view. All oligomers are attracted with significant adsorption energies and found 100% of time in contact with the surface. PCEs with shorter side chains (PCE 7:1-3) prefer upright orientation and show favorable ionic interactions with nearest neighbors (see panel e). Longer side chains (PCE 7:1-9) prefer tilted and flexible horizontal conformations of the backbone. (d) The conformation of the acrylate backbone adjusts to enable binding of COO‒ groups to locally specific patterns of Ca2+ ions on the surface (highlights by red circles). Esterified acrylate groups often detach two carboxylate groups to accommodate the side chain and support binding of acrylate ions further apart. (e) Association of multiple PCE 7:1-3 oligomers at larger surface coverage in the simulation. Inter-chain electrostatic interactions at high volume density of COO‒ groups in the polymer increase cohesion of the PCE film, which correlates with higher adsorbed mass and longer retardation of hydration.

105

5.4.2. Alternative models for polymer conformations and surface adsorption

The conformations of comb-like copolymers in solution have been previously classified [52, 179] based on molecular structure, viscosity, light scattering, and small angle neutron scattering measurements [112, 180, 181]. The model by Gay and Raphaël

[179] uses the number of repeating units n, each containing N monomers in the backbone and a side chain containing P monomers to predict conformations and solution properties of PCEs (Fig. 5.10). The PCEs in this study assume the conformations of decorated chains (DC) for PCE 6:1-5, flexible backbone worms (FBW) for most other structures, and semi-flexible backbone worms (SBW) for PCE 1:1-19. The expected conformations in molecular simulation are the same, even though the chain length is shorter, and were found to be in agreement with these schematic depictions.

The Gay-Raphaël model, however, cannot describe adsorbed conformations on the surface. The estimation of likely conformations of surface-adsorbed polymers has been attempted by the Flatt-Gay-Raphaël model, which uses further Flory minimization of the energy [52]. The FBW conformation model can be extended to account for the adsorbed conformation of the comb copolymer through a random walk of hemispheres.

The radius of an adsorbed spherical core (Rac) can then be calculated using the Eq. 5.3

[52]:

1/5 푎푝 7/10 −1/10 푅푎푐 = (2√2 (1 − 2휒) ) 푎푝푃 푁 (Eq. 5.3) 푎푁

where χ is the Flory factors of side chains, 푎P is the size of side chain unit, 푎N is the size of backbone unit, and P and N have been previously defined. A constant value of χ

106 measured in pure water in the semi-dilute regime was additionally assumed [182]. The adsorbed amount of copolymer at saturation plateau, msat, is achieved by Eq. 5.4 [52,

183]:

1/10 −3/10 푚푠푎푡 ≈ 푃 푁 (Eq. 5.4)

However, the assumptions in the Flatt-Gay-Raphaël model are too simplistic. The comb-like homopolymers in solution must be non-ionic and the chemistry of the surface is disregarded, among other conditions. Therefore, critical details of the identified mechanisms are excluded and the mismatch between the model and experiment is inevitable (Fig. 5.11 and Table S1).

a 5.0 b DC FBW PCE 1:1-19 SBW

4.0

3.0

PCE 2:1-19 SBS 2.0 PCE 3:1-19

Log(n) / Log(N) / Log(n) PCE 5:1-19 PCE 4:1-19 PCE 6:1-5 1.0 PCE 6:1-43 PCE 6:1-9 PCE 6:1-19 FBS

0.0 0.0 1.0 2.0 3.0 4.0 5.0 Log(P) / Log(N)

Fig. 5.10. Expected conformations of comb-copolymers in a good solvent according to the model by Gay and Raphaël (ref. [179]). The synthesized PCEs are highlighted. (a) Possible conformations include a decorated chain (DC), flexible backbone worm (FBW), stretched backbone worm (SBW), stretched backbone star (SBS), and flexible backbone star (FBS). The dashed lines show trends among PCEs as a function of side chain density and side chain length. (b) Nomenclature of n, N, and P in the Gay-Raphaël model (from ref. [52]). The polymer code can be written as (N-1):1-P with a total number of monomers n·N.

107

In consequence, the combination of experiment and atomistic simulation is a novel way to elucidate a consistent working mechanism of the PCEs and first molecular interpretations. Any number and chemistry of systems can be explored; the major limits are the accessible length and time scale of simulations. Simulations with the Interface force field take into account the specific surface and polymer chemistry. Hydration properties and predicted binding energies are in excellent agreement with experiment (see section 5.4.1). The expected accuracy is better than ±10% as shown for chemically well- defined silica surfaces as a function of pH in contact with specific peptides [178]. The new modeling tools for cement systems in combination with experiment may enable discoveries up to the large nanometer scale.

5.4.3. Working mechanism of PCE adsorption and cement hydration

The various techniques have provided key insights into the working mechanisms of PCEs, which are combined in the following. Adsorption of PCEs onto C-S-H surfaces involves primarily hydrated species of C3S and C3A. The mechanism includes (1) ion pairing of Ca2+ ions from the acrylate backbone with the hydrated mineral surfaces, (2) adsorption of acrylate ions from the PCE backbone onto the mineral surface via Ca2+ ···

COO‒ ion pairs, (3) and much weaker hydrogen bonds between COO‒ ions and silanol groups on the mineral surface. PEG side chains are not attracted to the mineral surface.

The strength of adsorption of PCEs on cement particles is controlled by the COO‒ content in the polymer backbones. The molar amount of COO‒ ions per gram of polymer

(Table 3.1) is proportional to the adsorbed amount per surface (or mass of cement) for

108 different grafting density of side chains (Fig. 5.11 a) and for different length of side chains (Fig. 5.11 b). The adsorbed amount reaches saturation at ≥5 mmol COO‒ per gram of polymer. Long side chains prevent attractive ionic polymer-polymer interactions, thus leading to a lower adsorbed mass, especially for a level of COO‒ below ~2000‒3000

μmol COO‒/g polymer (Fig. 5.11 b).

Consistent with these observations is the tendency toward upright orientation of acrylate backbones and thicker polymer layers for more ionic PCEs due to attractive ionic polymer-polymer interactions on the mineral surface (Fig. 5.12). The 50mer acrylate backbones tend towards a flat-on orientation on the mineral surface for high density of

PEG side chains (Fig. 5.12 a) and toward a more flexible, tilted orientation for low density of PEG side chains, especially for short length (Fig. 5.12 b,c). This trend is supported by atomistic simulations. The adsorption mechanism and adsorbed conformations thus change significantly depending on the PEG content of the polycarboxylate ether.

Although the adsorbed amounts of PCEs 6:1-5 and 6:1-9 are larger than that of

PCE 6:1-19, the fluidity (dispersion efficiency) of PCE 6:1-19 was significantly higher than those of PCEs 6:1-5 and 6:1-9 (Fig. 5.12 c). It is proposed as an explanation that nonionic PEG side chains must be of a certain length to mitigate ionic forces between the polymer films surrounding the cement particles, and therefore support dispersion and higher fluidity. The reversal of zeta potentials follows the same trend as the improvement in fluidity and reduction of water content for all PCEs. The results clearly confirm that

PCE adsorption and dispersion ability are two different parameters that need to be

109 considered separately, while both have several associated properties that follow the same trend.

Retardation of cement hydration is mainly due to the adsorption of PCEs onto the surface of cement particles that modifies the growth kinetics and morphology development of early hydrates. The clear relationship between the adsorbed amount, the inverse equilibrium conductivity, and retardation supports that hydration reactions are delayed proportional to the strength of binding of the PCE. The operating mechanism is steric hindrance of water penetration by the surface-adsorbed polymer film.

The observed trends of the effects of side chain density and length of PCE on the kinetics of cement hydration are consistent with earlier experimental studies, although less well-defined comb copolymers of higher polydispersity were typically used [144,

147, 184, 185]. The lack of affinity of PEG side chains to C-S-H surfaces and reduced electrostatic attraction of particles in the presence of PCEs of high PEG content were observed [110, 186]. Coiled conformations of PEG side chains reported in experiment agree with observations for longer PEG chains in the simulation (N>30) [175]. The PEG layer thickness was suggested to scale as N 0.6 with the length of side chains N by Atomic

Force Microscopy (AFM) on MgO surfaces [180, 187]. However, simple existing concepts of steric hindrance do not explain the complexity of mechanisms and properties.

Quantitative insight into ionic versus nonpolar interactions, binding mechanisms, binding energies, and polymer conformations from the molecular level is essential for rational molecular design.

110

5.0 5.0

4.5 (a)a 4.5

4.0 PCE 5:1-19 4.0

3.5 PCE 6:1-19 3.5 PCE 3:1-19 PCE 4:1-19 3.0 3.0

2.5 2.5

2.0 2.0 PCE 2:1-19 1.5 1.5 PCE 2:1-19 1.0 PCE 6:1-19 1.0 PCE 1:1-19 PCE 4:1-19 PCE 3:1-19 0.5 0.5

PCE 5:1-19 (mg/g) Amount Adsorbed Calculated Experimental Adsorbed Amount (mg/g) Adsorbed Experimental 0.0 0.0 0 1000 2000 3000 4000 5000 μmol COO‒/g polymer

5.0 5.0 (b) 4.5 b PCE 6:1-5 4.5 PCE 6:1-9 4.0 PCE 6:1-19 4.0

3.5 3.5

3.0 3.0 PCE 6:1-43 2.5 2.5

2.0 2.0

1.5 1.5 PCE 6:1-43 1.0 1.0

0.5 0.5

PCE 6:1-19 PCE 6:1-9 (mg/g) Amount Adsorbed Calculated PCE 6:1-5 Experimental Adsorbed Amount (mg/g) Adsorbed Experimental 0.0 0.0 1000 3000 5000 7000 9000 μmol COO‒/g polymer

Fig. 5.11. Adsorbed amount of polycarboxylate ether as a function of the COO‒ content in experiment and calculations by the earlier Flatt-Gay-Raphaël model [52, 183]. (a) The effect of the density of PEG side chains. (b) The effect of the length of the side chains. Higher density of COO‒ groups increases adsorption in both series. The FGR model does not correlate with measurements.

111

b Best particle Loosely grafted dispersion side chains, medium length Lower w/c ratio (e.g. 6:1-19)

H2O

a c

Densely grafted Loosely grafted side chains, side chains, medium length short length (e.g. 1:1-19) (e.g. 6:1-5)

H2O

C-S(/A)-H hydration layer Anionic backbone of PCE Neutral PEG side chains Water Cation (e.g. Ca2+) - 2- Anion (e.g. OH , SO4 ) Colloid boundary

Fig. 5.12. Schematic of conformations of adsorbed PCEs on hydrated cement particles and relationship to adsorption, hydration, and dispersion. (a) A high density of side chains of medium length leads to flat-on, intermittent binding of the backbone to the surface. Water penetration and hydration are relatively faster. (b) A low density of side chains of medium length leads to more consistent contact with the surface, increased conformational flexibility of the backbone, and tilted orientations of the backbone with respect to the surface. Ionic attraction between neighbor backbones, the adsorbed mass per unit area, and barriers to hydration increase. The nonionic portion of side chains leads to best reduction of agglomeration between particles. (c) Low density of side chains and short length of the side chains further increase conformational flexibility of the acrylate backbones, enable upright orientations, and strong ionic attraction to the surface. A thicker protective film is formed via strong inter-chain interactions. The barrier to hydration is then largest, yet dispersion and fluidity are somewhat reduced. Blue arrows symbolize interactions of water with the particle surface and PCEs backbones.

112

5.5. Conclusions

The fluidity, water reduction, and zeta potential values follow a different trend as a function of polymer structure in comparison to the adsorption and retardation of hydration. As already mentioned, best dispersion and greatest water reduction was achieved for the PCEe having low density and intermediate length of the PEG side chains. The zeta potential also showed a maximum for these PCEs because the upright- trending conformations of more highly charged backbones could enable less interfacial charge separation in an electric field. The systematic study of well-defined PCEs demonstrated for the first time the adsorption mechanisms, conformations, and orientations of the adsorbed polymer layers, which were determined by the ionic interactions with C-S-H surface and by the geometric balance between the main chain of the PCEs and the PEG side chains. Macroscopic and nanoscale measurements were explained from the molecular scale, enabling specific interpretations to influence the working mechanisms of cement setting. Nevertheless, the complexity of interactions in cement minerals shall not be underestimated. The chemistry of surface regions and the size of cement particles give rise to many simultaneous interactions that continue to leave open questions from the nanoscale to macroscale.

113

CHAPTER VI

UV-CURABLE CONTACT STABILIZATION MATERIALS

6.1. Introduction

In all electrical contacts, the real contacting surfaces are not continuously planar or perfectly smooth. If magnified sufficiently, these surfaces show the general appearance of a mountain range, having a large number of irregular peaks and pits. However, the actual transmission of electrical current from one surface to the other one only occurs at those parts where there is a real physical contact between the co-operating contact surfaces. These co-operating contact surfaces are broadly categorized. For instance, high current applications consist of plug-in communications as well as standby power applications. Besides, low power applications include low level radio frequency, audio frequency, video, and computer circuitry.

In addition to the above-mentioned issues, there may be contact resistance because of oxidized metal or other external materials in an electrical contact situation. In addition, there may be a higher resistance if there is any varnish material present. Hence, virtual insulation between contacting surfaces can occur under certain circumstances. In high current applications, heating deformation because of the resistance of materials used in contacting surfaces may occur. Moreover, there is a possibility of chemical reactions which take place due to this heat.

114

In all low power current applications, such as low current video cable connectors, audio or radio frequency connectors, or cable connectors, there can be situations when there is not an appropriate signal power available to confirm that there is a consistent maintenance of the signal flow. In terms of alternating complex waveform signals, the current flow can be temporarily disturbed at zero-crossing conditions, in which the voltage potential between contact surfaces changes from a directed negative to positive potential, or vice versa. In such conditions, the current flow can be re-established only after there is sufficient voltage rise to break down the potential gap between co-operating contact surfaces.

The discontinuous behavior in radio frequency circuitry at the zero-crossing, can cause line reflections which may enhance artifacts in signals. For instance, the zero- crossing discontinuity in video signals may show up as imperfect chroma demodulation or video ghosts due to the apparent noise. In computer circuitry, the zero-crossing can appear as amplified noise, and in certain conditions there can be incorrect data transmission, rectification artifacts, or cyclic redundancy error conditions.

Octadecyl alcohol-doped palm oil materials have been already used as contact stabilization materials in electronic industry, mainly in respect of low power audio and radio frequency as well as computer data transmission applications. However, in all vegetable oil-based materials, there is a tendency for the oils to be cross-linked during their use, especially in the presence of metallic materials which can act as catalysts. This cross linking creates a varnish, by which an insulative property occurs. Therefore, using such materials, even though the initial results were encouraging, they proved to be of no value due to the varnishing and resulting contact insulating characteristics.

115

A material which could show both low and high resistance, and be activated electrically so as to be switched from one resistance state to another was already developed [188]. Cuddy et al. [189] developed another material based on a polyether copolymer which could inhibit the production of dross during wave soldering. This copolymer floats on top of hot liquid solder and it should comprise a block co-polymer of a dihydroxyphenol together with one lower alkylene oxide, having at least 20% by weight of oxyethylene segment, with a molecular weight in the range of 500 to 3000. The only purpose of that material was to inhibit dross through floating on the surface of liquid solder throughout the wave soldering procedures.

Angell et al. [190] developed a solid electrolyte which was formed of a cross- linked elastomeric complex material charged with ionizable salts of high ionic conduction. The particular purpose for that material was to permit its use as solid electrolyte for use at high temperatures. Non-ionic difunctional block polymers terminated with primary hydroxyl groups having the molecular weights ranging from

1000 to 15,000, are defined with the trade mark PLURONIC™. These block polymers are polyoxyalkylene derivatives of the propylene glycol and are available in liquid, powder, or cast-solid forms. They are used as defoamers, demulsifying agents, stabilizers, binders, wetting agents, dispersing agents, and chemical intermediates.

In order to be effective at micro-power and higher power levels as well as at frequencies ranging from DC to 500 mHz, a contact stabilization material should have the following characteristics. First, it must display conductive effects when it is utilized between conductive surfaces including metals, mainly in metal to metal electrical contacts. However, this material should not show conductive properties when it is applied

116 to different insulators. These insulators include insulating material between the outer conductive shells and the inner conductive studs or wires used in video connectors and co-axial radio frequency, or the insulating materials between data terminal strips for computer connectors [191].

A contact stabilization material can be sufficiently liquid so that it may be easily applied to the electrical contact surfaces. Besides, the contact stabilization material is supposed to have an appropriately high surface tension that capillary action causes the material to migrate between electrical contact surfaces. Even when the contact stabilization material is spread in a very thin film thickness, it should have a low vapor pressure to remain in place for the service life of connector contact surfaces. Moreover, a good contact stabilization material must exhibit properties such that if the material catalyzes any reaction in the presence of metallic contact surfaces, then the resultant material should also show the same properties.

When a contact stabilization material is utilized, it is better to have sufficient detergent action so that any coated contaminant which may be present on a contact surface, should be lifted or wiped away when the contact stabilization material is applied.

In addition, the contact stabilization material should not degrade the commonly used insulation materials or plastics that are used in electrical applications. Also, the contact stabilization material should not be corrosive to any metals that are utilized. The contact stabilization material should have low toxicity for safety reasons [191]. When a contact stabilization material is applied, the flow characteristics of the electrical signal current between co-operating surfaces will be improved. In fact, this improvement can be as a

117 result of having lower contact resistance between surfaces, having stabilized contact resistance between surfaces, and decreasing the likelihood of contact noise.

A contact stabilization material for use with electrical contact surfaces was already developed under the tradename Stabilant 22™ [191]. This material is based on at least one copolymer or block copolymer of polypropylene glycol (PPG) together with polyethylene glycol (PEG), or a plurality of block copolymers of PPG together with PEG.

These materials, especially when they have low molecular weights, may be fluid and tend to be runny. However, it was found that after cross-linking, the longer polymer chains are still benign and show acceptable desired characteristic as discussed above. The block copolymers of PEG and PPG, having the molecular weights in the range of 1,400‒2,800, have been found to be the most useful materials. Moreover, it is important to note that higher molecular weight block copolymers can be suitable for use with higher temperature surfaces and exhibit the essential capillary action at those temperatures. In this invention, the block copolymers were useful when the PPG and PEG were present in the range of 2‒98 wt.% [191].

It was suggested that the conductive mechanism using Stabilant 22™ may be as a result of tunnel diode transmission effect. In fact, the film thickness effects of this material were such that the normal insulation gaps were not affected adversely. This material may need to be more or less liquid at its application temperatures. In addition, it may be intended for the operation at very low or high temperatures. Hence, higher molecular weight materials could be developed in circumstances where it was known that they should work in higher or lower temperatures. Furthermore, these contact stabilization materials could be diluted with certain diluents in order to have larger

118 surface area coverage. However, if a diluted material was applied to a surface that was incompatible with the diluent, either an undiluted contact stabilization material or another having a different diluent should be used [191].

Despite the fact that Stabilant 22™ is able to improve the performance of electrical contact co-operating surfaces, this material is fluid at room temperature. This fluidity, due to its flowing or leakage, can restrict its applications, especially when the electrical co-operating surfaces are required to disconnect and re-connect for multiple times. In this case, the stabilization materials can be removed as a result of the relative motion of co-operating surfaces during disconnecting and re-connecting, which rapidly exhausts the stabilization effects. Besides, the leakage of fluid stabilization materials might contaminate the devices as well. Thus, a material which can equally or better improve the performance of electrical contact surfaces that can be quickly solidified on the co-operating electrical contact surfaces is in need.

In recent years, much attention has been devoted to the amount of solvent that enters the atmosphere from coatings and related products. In the early 1970s, almost 90% of industrial coatings were low solids in nature and contained 10‒20% by weight solids

(excluding pigments) and the remainder was solvent. Industrial and architectural coatings were characterized as being solventborne. Although latex paints are currently prevalent, oil-based paints still have a remarkable place in the market due to their outstanding characteristics and properties. However, their odor and slow drying nature remain as deficiencies. Cure-on-command technology in coatings, adhesives, and inks is the instantaneous and efficient conversion of a low viscosity liquid formulation containing ingredients that has been applied to a substrate, into a cross-linked and polymerized mass

119 by exposure to a radiation source. In general, the cure-on-command technology is highly effective and 100% solids in nature. The radiation used in the curing process is usually

UV radiation, electron beam (EB) energy, electromagnetic radiation, coherent or laser radiation, and visible radiation or light [90, 192].

The electromagnetic spectrum is derived from electromagnetic waves of radiation which are recognized by fluctuations of electric and magnetic fields. It includes the total range of electromagnetic wavelengths and extends from zero to infinity. On the low end of the range, cosmic rays, gamma rays, and hard X-rays are located and on the high end of the range, microwave, radar, radio, television, and electric waves are placed [193,

194]. The visible portion of the spectrum having the wavelengths in the range 400‒760 nm is known as light. Moreover, the invisible portion of the spectrum beyond the violet end of the visible region is known as UV light. The UV region can be classified as UV-C,

UV-B, and UV-A which contain regions in the range 15‒280, 280‒320, and 320‒400 nm, respectively [192]. The range from 800 to about 30000 nm is known as the infrared (IR) region of the spectra and the microwave region begins above this is the microwave region and so on. The cure-on-command technology typically uses the radiation curing because of numerous advantages such as rapid cure, low capital investment, high solids, high productivity, removal of solvents, possible multiple operations, high cross-linking density, and potential for improved chemical, abrasion, and stain resistance [90].

In the cure-on-command technology, volatile organic compounds (VOCs) are near zero. Although solvent is added to some specialized formulations for viscosity reduction, it is typically easy to fit the equipment with a solvent recovery system or to stay within regulatory limits. Using this technology, heat-sensitive substrates including

120 wood, paper, plastics, and printed circuit boards can be coated and cured. In fact, there is little buildup of heat in the substrate when thin films are involved. However, careful selection of formulation ingredients and line speed should be considered when heat buildup is produced through the preparation of thick films. It can be well understood that radiation-curing lines are able to run much faster than conventional coating lines. In addition, the multiple operations which can be accomplished on a single pass through the line, result in increases in productivity. For these reasons, the cure-on-command is a technology that will be experiencing faster growth in the near future [90, 195].

The UV curable coatings are always in competition with thermally curable systems of the water-based, solvent-type, or powder coatings. The UV curing process is mainly determined by the desired application of the coating. In fact, the end-product determines the substrate to be coated. This substrate may be a protective coat for window frames, an overprint varnish for paper cards, an abrasion resistant clear coat for parquet, or a colored base coat and a clear coat for metal coils. The type, characteristics, and thickness requirements of the coatings are governed by the economics of the coating process and the properties such as abrasion resistance, color effects, high gloss appearance, flexibility, hardness, etc. The chemical formulations including base resins, photoinitiators (PIs), reactive diluents, and additives can provide these targeted properties. In addition, a suitable selection of the components should be performed in order to enable an efficient curing process. The materials costs, the equipment set-up, and the whole coating process design have to be considered in order to calculate the total costs of the coating process [90, 196].

121

The economic factors of the UV curing have been already discussed by Stowe

[197]. In general, the UV curing process offers many advantages over competitive coatings. These advantages are basically related to costs, performance, and environmentally compliance. Because of this fact that the general comparison of process economics is not possible, it should be performed rather in case to case studies.

Therefore, the UV curing significantly relies on the required application properties with the chemistry chosen to achieve the performance requirements and the UV curing equipment applied to provide a fast cure in order to meet the economical aspects of coating technology [90].

Following these considerations, new materials based on methacrylated

PEGs/PPGs are synthesized. These materials should be able to uniformly coat onto the entire surface of electrical co-operating systems. Moreover, these new materials are able to be cured under UV light and produce gel-like structures. After being cured, the soft gels can provide the possibility of re-inserting multiple times without scratching and affecting the electrical properties.

6.2. Experimental section

6.2.1. Materials and synthesis

Poly(propylene glycol)-block-poly(ethylene glycol)-block-poly(propylene glycol)

(PPG-b-PEG-b-PPG) (Mn ~2,700 g/mol) was purchased from Sigma-Aldrich.

Methacryloyl chloride, trimethylamine (TEA), 2-hydroxy-2-methylpropiophenone

(photo-initiator), and dichloromethane were purchased from Sigma-Aldrich and used

122 without further purification. BYK®-333 was kindly provided by BYK Additives &

Instruments and added as a wetting agent. Poly(ethylene glycol) dimethacrylate

(PEGDMA) with the average molecular weights of ~350 g/mol (repeating unit (n) of ~5) was purchased from Polysciences Inc. Methoxy poly(ethylene glycol) monomethacrylate

(MPEGMA) with the average molecular weights of 300 g/mol (n~5) and 500 g/mol (n~9) were purchased from Sigma-Aldrich and used as received. Both PEGDMA and

MPEGMA were used as reactive diluents and their chemical structures are shown in

Scheme 6.1.

Scheme 6.1. Chemical structure of MPEGMA and PEGDMA reactive diluents.

PPG-b-PEG-b-PPG dimethacrylate was synthesized based on the reported procedure [198, 199] and described as follows. PPG-b-PEG-b-PPG (Mn~2,700 g/mol)

(54 g, 0.020 mol) was dissolved in 100 mL anhydrous dichloromethane into a three-neck round bottom flask. TEA (4.86 g, 0.048 mol) was then added at 0 °C in an ice bath.

Methacryloyl chloride (5.00 g, 0.048 mol) was dissolved in 24 mL anhydrous dichloromethane and added dropwise to the PPG-b-PEG-b-PPG solution by syringe pump. Then, the mixture was stirred for 24 h at room temperature under nitrogen to complete the reaction. The resultant mixture was later filtered by vacuum filtration to separate the insoluble triethylamine salts, followed by neutral alumina filtration in order

123 to remove excess triethylamine. Then, rotavap was used to remove excess dichloromethane solvent. Finally, the purified product was dried in a vacuum oven at room temperature for 24 h.

The synthesized PPG-b-PEG-b-PPG dimethacrylate was characterized by Fourier transform infrared (FT-IR) spectroscopy with a Thermo Scientific Nicolet iS50 FT-IR spectrometer using the transmission mode at a resolution of 8 cm-1 and averaging from

512 scans. The compositions were determined using 1H NMR (Varian NMRS-500 nuclear magnetic resonance instrument) operating at 500 MHz with deuterium oxide

(D2O) as a solvent.

6.2.2. Sample preparation

The UV-curable electrical contact stabilization materials developed in this study consisted of the PPG-b-PEG-b-PPG, synthesized PPG-b-PEG-b-PPG dimethacrylate, mono-functional (MPEGMA, n~5 or 9) or di-functional (PEGDMA, n~5) reactive diluents, photo-initiator (2-hydroxy-2-methylpropiophenone), and a wetting agent. For rheology study, each formulation was prepared by first mixing the PPG-b-PEG-b-PPG dimethacrylate and reactive diluent with different molar ratios (1.0:3.8, 1.0:5.7, and

1.0:8.6) in an amber glass vial. The photo-initiator (1 wt.%) and wetting agent (0.5 wt.%) were then added to each mixture. The resulting mixtures were mixed by a vortex for 30 s and the homogeneous solutions were finally obtained at room temperature. All formulations are summarized in Table 6.1. As an example, formulation code M-5-3.8 indicates the formulation containing mono-functional MPEGMA reactive diluent (M)

124 having 5 repeating units and the molar ratio (reactive diluent to PPG-b-PEG-b-PPG dimethacrylate) of 3.8.

Table 6.1. Summary of UV-curable formulations developed for rheology tests.

Formulation codea Reactive diluent Repeating unit (n)b Molar ratioc M-5-3.8 MPEGMAd 5 3.8 D-5-3.8 PEGDMAe 5 3.8 M-9-3.8 MPEGMA 9 3.8 M-9-5.7 MPEGMA 9 5.7 M-9-8.6 MPEGMA 9 8.6 a all formulations are based on PPG-b-PEG-b-PPG dimethacrylate, reactive diluent, photo-initiator, and wetting agent. b Number of repeating units in reactive diluents. c Molar ratio of reactive diluent to PPG-b-PEG-b-PPG dimethacrylate. d Mono-functional MPEGMA reactive diluent. e Di-functional PEGDMA reactive diluent.

6.2.3. Film formation

In order to coat the material onto one or both of a pair of co-operating contact surfaces with uniform thicknesses, flow coating was applied to prepare polymer films with thickness in the submicron regime [200]. The flow coating device comprises a knife blade fixed at a certain distance above a stationary stage. Typical gap heights range from tens of microns to hundreds of microns. After depositing/wicking of a small amount of pre-cured materials between the blade and the substrate, accelerating the blade with the substrate being rigidly fixed to the stage or vice versa can produce a thin layer of material with desired thickness. The flow coating process is considered as a result of the competition between the capillary forces holding the polymer solution between the knife blade and the substrate and the frictional drag exerted on that same solution when the blade (or substrate) is pulled away [201, 202].

125

A usual method of applying the stabilization material onto a connector is proposed as follows. First, the connector was disconnected and then 0.1 mL of the stabilization material was coated to one of the co-operating electrical contact surfaces by flow coating with coated thickness less than 3µm. Subsequently, UV radiation exposure was applied for 90 s, after which the coated material solidified and showed a gel-like structure.

Finally, the connector was re-connected.

6.2.4. Rheological measurements before UV curing

The pre-cured rheological properties of the samples were characterized by a dynamic strain controlled rheometer (TA Instruments ARES-G2). All the experiments were performed on a parallel plate fixture with a diameter of 25 mm at 25 °C. The linear viscoelastic (LVE) range of each sample was determined by a dynamic strain sweep followed by a frequency sweep.

6.2.5. Real-time rheology during UV curing

A Malvern Instruments (Bohlin Gemini 200) rheometer system with a parallel plate fixture of 25 mm diameter was utilized to track the dynamic rheological properties during UV curing. A specially designed transparent UV curing bottom plate was placed in this system to allow the UV-radiation to pass through. The UV-radiation was generated by a S2000 Omnicure Spot UV-lamp at a fixed intensity of 40 mW/cm2 throughout each experiment. The intensity was calibrated by a UV-Power Puck II radiometer prior to each experiment. The sample thickness was kept at ~500 µm and all the experiments were

126 performed at 25 °C. For the time sweep, all the samples were exposed to UV-radiation at t=0 with a frequency value of 1.0, 6.0, or 10.0 rad/s. The frequency sweep was applied immediately after the UV-exposure and was terminated at different time periods to study the gelation behavior of these materials. All the dynamic oscillatory measurements were carried out under controlled strain of 0.02 in order to endure the linear viscoelastic regime of all samples.

6.2.6. Real-time FT-IR spectroscopy

The FT-IR spectra were recorded by a Alpha-T (Bruker Optics, Billerica, MA,

USA) trasmission mode in juction with an external UV-lamp (S2000 Omnicure). The intensity of UV-radiation was maintained at 40 mW/cm2 (same as the rheological measurements) for all samples. The pre-cured liquid sample was sandwiched between two KBr crystals with a diameter of 19 mm and then inserted inside the FT-IR spectrometer sample chamber. Each IR spectrum obtained during curing process was an average of six scans with a resolution of 4 cm-1. The C=C stretching vibrations of the vinyl functional groups (~1636 cm-1) were selected to monitor and calculate the conversion rate which could be obtained from the integration of the C=C absorption bands at the beginning of the experiment and any subsequent time t, using the following

Eq. 6.1:

퐴(0)−퐴(푡) 푋(푡) = (Eq. 6.1) 퐴(0) where 푋(푡) is the conversion at time t, 퐴(0) is the area of the initial band, and 퐴(푡) is the area at time t.

127

6.3. Results

6.3.1. Synthesis and characterization of PPG-b-PEG-b-PPG dimethacrylate

The synthesis pathway for the preparation of PPG-b-PEG-b-PPG dimethacrylate is shown in Scheme 6.2. As shown in Fig. 6.1, both PPG-b-PEG-b-PPG and PPG-b-PEG- b-PPG dimethacrylate samples exhibited similar FT-IR patterns except the presence of a weak signal at 1720 cm-1 for the dimethacrylate polymer because of the carbonyl stretching mode. Since the amount of methacrylate groups compared to the aliphatic polymer chain was low, the appearance of weak absorption band due to the carbonyl groups was expected.

Scheme 6.2. Synthesis of PPG-b-PEG-b-PPG dimethacrylate.

Fig. 6.1. FT-IR spectroscopy of PPG-b-PEG-b-PPG (black curve) and PPG-b-PEG-b- PPG dimethacrylate (red curve).

128

In 1H NMR spectrum, the signal at 1.13 ppm was assigned to the propylene glycol methyl (CH3) protons. In addition, the signals around 3.36–3.66 ppm represented the methylene protons as well as the protons of –(O‒CH(CH3)CH2)‒ in the repeat units of the main chains. A comparison of the 1H NMR spectra between the starting and modified triblock copolymers indicates the presence of two vinyl hydrogen resonances (Ha and Hb) at 5.50 and 6.00 ppm for the modified triblock copolymer, which confirms the presence of the methacrylate end groups.

Fig. 6.2. 1H NMR spectra of (a) PPG-b-PEG-b-PPG and (b) PPG-b-PEG-b-PPG dimethacrylate.

129

6.3.2. Rheology development during UV curing

Several new materials derived from the synthesized PPG-b-PEG-b-PPG dimethacrylate, mono-functional (MPEGMA, n~5 or 9) or di-functional (PEGDMA, n~5) reactive diluents, photo-initiator (2-hydroxy-2-methylpropiophenone), and a wetting agent were prepared and tested. They could be uniformly coated onto the entire surface of electrical co-operating systems, and more importantly, they were able to be cured under

UV light as a suitable source for production line. The formulations were modified several times based on the measured contact resistance values. The modified formulations based on the balanced ratios of PPG-b-PEG-b-PPG dimethacrylate, reactive diluent, and initiator showed decreased contact resistance compared to the non-coated surface.

However, after being UV cured at small thicknesses and inserted into the edge connectors, they were broken due to the friction between the connectors. To overcome this problem, a better understanding about the exact function of reactive diluents in the

UV curing process of these materials was really needed.

A representative graph of the elastic (G’) and viscous (G”) modulus for formulation M-9-3.8 (Table 6.1) as a function of UV curing time is shown in Fig. 6.3.

Prior to the UV-exposure, the sample exhibited a typical viscous liquid behavior with a much higher viscous modulus (0.9910 Pa) than elastic modulus (0.0368 Pa) [203].

Nevertheless, after being exposed to the UV-radiation, both G’ and G” values immediately increased with time and G’ exceeded G” beyond 37 s, which is in the vicinity of the gel point as already discussed by Winter-Chambon Criterion [104, 105].

The gel point in a chemically crosslinking system is defined as the critical extent of reaction where the polymer chains start forming network and transition from a viscous

130 liquid state to a viscoelastic gel [204]. After the gel point, the sample showed elastic behavior for the rest of curing process, having the final plateau G’ value (t=1200 s) greater than the G” value. The G” values leveled off instantly after the gel point and reached plateau whereas the G’ values continued increasing, which implied that the sample started to behave more elastic with the almost negligible viscous characteristics.

Fig. 6.3. Development of the elastic (G’) and viscous (G”) modulus as a function of time for formulation M-9-3.8 under a constant frequency of 6 rad/s.

A frequency sweep was also performed before UV-radiation (t=0) and immediately after removing the UV-radiation at different time intervals (t=37 and 1200 s), and the results are summarized in Fig. 6.4. The G’ and G” curves at t=0 s further verified the viscous characteristics of the sample prior to the UV-exposure, which possessed a much greater G” value than G’ throughout the entire testing frequency range.

Moreover, both G’ and G” were frequency dependent with different slopes, revealing a viscous liquid state. Nevertheless, at t=37 s (around the gel point), the elastic and viscous modulus were almost identical and demonstrated the same dependency on the frequency of oscillation. Note that the graphs in Fig. 6.4 are plotted on a logarithm scale, meaning 131 that both the G’ and G” showed the same power-law behavior (G’, G” ∼ωn) with the same exponent n (relaxation exponent, value is 0.5). With G’=G” ~ ω0.5, the steady shear viscosity and equilibrium shear modulus can be obtained using Eq. 6.2 and Eq. 6.3:

−0.5 휂0 = lim(퐺"/휔) = 퐶 lim(휔 ) = ∞ (Eq. 6.2) 휔→0 휔→0

0.5 퐺∞ = lim(퐺′) = 퐶 lim(휔 ) = 0 (Eq. 6.3) 휔→0 휔→0 where 휂0 and 퐺∞ are the steady shear viscosity and equilibrium shear modulus, respectively [104, 205]. 휂0=∞ and 퐺∞ = 0 are exactly the features of a material at the liquid/solid transition point. When the sample was further exposed to the UV-radiation

(t=1200 s), the elastic modulus became greater than the viscous modulus throughout the frequency range, and more importantly, independent of the frequency, both of which are characteristics of a highly cross-linked material.

Fig. 6.4. Elastic (G’) and viscous (G”) modulus as a function of frequency for formulation M-9-3.8 (Table 6.1) before (t=0 s) and after UV-radiation exposure at time intervals of t=37 and 1200 s.

132

6.3.3. Effect of reactive diluents

The real time FT-IR technique was utilized to monitor the curing kinetics of these systems. The infrared spectra of formulation M-9-3.8 at four different UV-exposure times are shown in Fig. 6.5 as a representative of all formulations. Previous to the UV- exposure, the C=C absorption band of vinyl groups (~1636 cm-1) was quite evident, whereas this band was notably diminished as the curing reaction proceeded. As the reaction approached completion, the C=C band almost dropped to the zero absorption at

780 s, implying that the vast majority of the C=C functional groups were reacted. The changes of the C=C band areas were used to calculate the vinyl conversions and plotted as a function of UV-exposure time for each formulation (Figures 6.6b, 6.7b, and 6.8b).

Fig. 6.5. FT-IR spectra of formulation M-9-3.8 at four different UV exposure times.

Fig. 6.6a shows the modulus development of time sweep for the UV-curable formulations M-5-3.8 and D-5-3.8 based on mono-functional or di-functional reactive diluents in order to study the functionality effects. Both the elastic and viscous modulus of di-functional formulation showed an instant increase after being exposed to the UV-

133 radiation, whereas mono-functional formulation exhibited a gradual increase in the elastic and viscous modulus. Moreover, the gel point shifted from 45 s to 8 s as the functionality of the reactive diluents was increased from one to two.

The evolution of vinyl conversion for mono-functional and di-functional formulations M-5-3.8 and D-5-3.8 was obtained from FT-IR spectra (Fig. 6.6b), further confirming that the reactive diluents having more vinyl functionalities exhibited faster responses to the radiation process. At the initial stage of UV-exposure, the vinyl conversion of di-functional formulation D-5-3.8 immediately increased to a stable value of ~0.95 within a few seconds (~18 s). However, the vinyl conversion curve of mono- functional formulation M-5-3.8 demonstrated a slow increase after being exposed to the

UV-radiation and reached the stable value of ~0.93 at a much longer time (~780 s). The lower ultimate conversion value of formulation M-5-3.8 compared to that of formulation

D-5-3.8 could be ascribed to the chemical structures of the reactive diluents utilized.

134

Fig. 6.6. (a) Development of elastic and viscous modulus and (b) vinyl conversion as a function of UV-exposure time for formulations M-5-3.8 and D-5-3.8.

The ethylene oxide repeating units of the reactive diluents used in this study could also have an effect on the curing kinetics and properties of the resulting UV-curable formulations. In order to investigate the effects of PEG side chain length, two mono- functional reactive diluents MPEGMA having different repeating units of n=5 and 9 were utilized. Before UV-radiation, both formulations M-5-3.8 and M-9-3.8 exhibited a

Newtonian fluid behavior throughout the testing frequency range. Compared to formulation M-5-3.8, formulation M-9-3.8 showed a higher viscosity over the entire frequency range because of its higher molecular weight. The evolution of elastic and viscous modulus vs. UV-exposure time for formulations M-5-3.8 and M-9-3.8 is shown

135 in Fig. 6.7a. As shown, both samples showed a rapid increase in G’ and G” values after being exposed to the UV-radiation [206, 207]. Compared to formulation M-5-3.8, formulation M-9-3.8 containing the reactive diluent having higher molecular weight consistently maintained its larger G’ and G” values till 142 s and 105 s, respectively.

Then, the G’ and G” of formulation M-5-3.8 started to show higher values than those of formulation M-9-3.8, mainly due to the higher reaction rates as a result of the larger mobility of MPEGMA free radicals with shorter PEG chains [93].

As already mentioned, the viscosity of reactive diluent MPEGMA with longer

PEG side chain (n=9) was considerably higher than that of the reactive diluent with shorter PEG side chain (n=5). This difference in viscosity could play a significant role on the rheological properties especially in the initial curing stages. In fact, a larger increase of elastic and viscous modulus could be observed for formulation M-9-3.8 at the initial curing. Additionally, an earlier gel point appeared accordingly as illustrated in Fig. 6.7a, indicating that formulation M-9-3.8 reached transition from a viscous liquid to a viscoelastic gel sooner than formulation M-5-3.8. Although further investigations would be needed to understand this behavior, the stronger hydrogen bonding in formulation M-

9-3.8 may result in the closer association of PEG/PPG chains, resulting in an earlier gelation compared to that of formulation M-5-3.8 [208].

Beyond 142 s, the elastic (G’) modulus curve of formulation M-5-3.8 almost overlapped with that of formulation M-9-3.8. This observation also happened for the viscous (G’’) modulus curve after 105 s. In fact, this phenomenon revealed that the ultimate dynamic modulus was insignificantly influenced, if not completely unaffected, by the molecular weight of the mono-functional reactive diluents used in this study. This

136 could be explained by the fact that the PEG side chains of mono-functional reactive diluents only formed dangling ends in the cross-linked networks and did not contribute to the elastic modulus of these systems [204]. As shown in Fig. 6.7b, the vinyl conversion vs. UV-exposure time for both formulations M-5-3.8 and M-9-3.8 reached ~0.65 conversion within 50 s with almost identical conversion speed, which again proved the comparable mobility of both reactive diluents.

Fig. 6.7. (a) Development of elastic (G’) and viscous (G”) modulus and (b) vinyl conversion as a function of UV-exposure time for formulations M-5-3.8 and M-9-3.8.

137

The UV curing process of all developed formulations in this study was based on the copolymerization between PPG-b-PEG-b-PPG dimethacrylate and the mono- functional or di-functional reactive diluents [209, 210]. By modifying the content of each component, the mechanical properties and curing kinetics of the resulting cross-linked polymer networks could also be greatly varied. To study the quantitative effects of reactive diluents, three different amounts of reactive diluents including 40 wt.%, 50 wt.%, and 60 wt.% were selected to perform a time sweep of dynamic rheological measurement under external UV-radiation. For these systems, the molar ratios of

MPEGMA reactive diluent to PPG-b-PEG-b-PPG dimethacrylate were 3.8 (formulation

M-9-3.8), 5.7 (formulation M-9-5.7), and 8.6 (formulation M-9-8.6), respectively.

The evolution of elastic (G’) and viscous (G”) modulus as a function of UV- exposure time for formulations M-9-3.8, M-9-5.7, and M-9-8.6 are shown in Fig. 6.8a.

These formulations exhibited Newtonian fluid behavior duration the frequency range prior to the UV-exposure. The complex viscosity of these systems decreased in the following order of M-9-8.6

As shown in Fig. 6.8a, both elastic (G’) and viscous (G”) modulus curves demonstrated a three regime development: a gradual increase in the beginning (regime i) followed by a rapid increase (regime ii), and eventually a gradual leveling off (regime iii). In a free radical polymerization, these regimes could approximately ascribed to the

138 photo initiation, propagation, and termination steps, respectively. Furthermore, formulation M-9-8.6 with higher reactive diluent content showed lower modulus values

(both G’ and G”) at a given UV-exposure time.

Fig. 6.8b shows the development of vinyl conversion for formulations M-9-3.8,

M-9-5.7, and M-9-8.6 after being exposed to the UV-radiation. Each sample exhibited an instant increase in conversion and then quickly decreased and reached a stabilization level. As the reactive diluent content increased, the conversion decreased at a given time and the curve leveled off later, both of which are indicative of reduced reaction rate at the higher reactive diluent content. Since a lower degree of unsaturation has lower response to the radiation [211], formulation M-9-8.6 having more mono-functional reactive diluent was less sensitive to the radiation. Consequently, increase of vinyl conversion was slower in formulation M-9-8.6 compared to those of formulations M-9-3.8 and M-9-5.7.

139

Fig. 6.8. (a) Development of elastic (G’) and viscous (G”) modulus and (b) vinyl conversion as a function of UV exposure time for formulations M-9-3.8, M-9-5.7, and M- 9-8.6.

6.3.4. Contact resistance of soft gel UV-cured contact stabilization materials

The method of applying the contact stabilization material on electrical contact surfaces should ensure that at least one of a pair of co-operating contact surfaces have a substantial portion coated by a very thin film of the stabilization material in order to enhance the flow of electrical signal current between two surfaces. Meanwhile, the thickness of coated material on the electrical contact surfaces cannot exceed a few micrometers; otherwise the excessive material after being cured will be squeezed and

140 accumulated between co-operating surfaces because of the relative motion of the surfaces, hence obstructing the metal-to-metal contact. The contact stabilization materials, herein, could mask the whole surfaces and act as insulating layers between electrical contact surfaces.

Based on the rheological studies performed, three formulations M-9-3.8, M-9-5.7, and M-9-8.6 were selected and used for contact resistance tests. It was found that the thickness of coated stabilization material within ~3 µm was appropriate to improve the electrical current flow characteristics. As shown in Fig. 6.9, ~6% decrease in contact resistance of electrical contact surfaces using these formulations was observed compared to the co-operating surfaces without applying the stabilization materials. In these tests, six identical gold plated edge cards were prepared and inserted into six identical connectors in order to fix the contacts in series, which consequently provided six sets of edge card/connector pairs for contact resistance testing. The measurements were performed at the back of the connectors, for both before and after the application of the contact stabilization material. For each edge card/connector pair, the edge card was disconnected from the connector and then re-connected after coated with stabilization material. In order to test the reliability of the stabilization material, this process was repeated ten times and the contact resistance was measured each time after re-connecting to the connector. It was observed that the contact resistance decreased and remained at the reduced value after being coated regardless of the number of re-connecting processes.

141

Fig. 6.9. Contact resistance results for the formulations M-9-3.8, M-9-5.7, and M-9-8.6. The Ref. was measured without any contact stabilization materials.

It is important to note that by adding non-functional PPG-b-PEG-b-PPG and decreasing the amount of PPG-b-PEG-b-PPG dimethacrylate and reactive diluents in the formulations, softer gels could be produced. For example, a formulation based on PPG-b-

PEG-b-PPG/PPG-b-PEG-b-PPG dimethacrylate/MPEGMA with the molar ratios of 3/1/6 showed a very soft gel characteristic after being UV cured and was able to be re-inserted multiple times without any scratching.

6.4. Discussion

Rheological measurements were performed to investigate the exact effect of reactive diluents on the final structure of cured materials. In general, the viscosity and modulus of the UV-curable materials developed in this study were increased by several orders of magnitude at the end of curing process, namely the sample going through a transition from a liquid state to a solid one. However, the steady shear might disrupt the

142 gel-like structure during this transition [212]. Hence, the dynamic rheological measurements, where a small sinusoidal strain deformation is applied to the sample, were adopted here to monitor the real time development of rheological properties during UV curing [101].

As the curing process further proceeded, the increasing trend of G’ values started to decrease gradually, indicating that the reaction rate decreased, which can be ascribed to the reduction of active free radicals [207, 213]. In comparison with the pre-cured sample, there was an over six orders of magnitude increase in the G’ values at the end of the UV curing process, suggesting a significant change in the elasticity of the sample.

Reactive diluents were added to all formulations to reduce the viscosity of the liquid precursor, hence improving their processability. It is also well known that reactive diluents can significantly affect the coating materials in terms of appearance, curing kinetics, rheological properties, and mechanical properties [94]. Consequently, the effects of reactive diluents, including the functionality, molecular weight, and content were investigated and discussed in this study.

The significant upshift of the reaction rate from mono-functional to di-functional reactive diluents could be explained by the fact that the reactive diluents with higher degree of functionalities showed faster responses to the UV-radiation process [92, 211].

In a UV curing reaction, where the methacrylated oligomers are polymerized, the average crosslinking equals to the average degree of polymerization of the vinyl double bonds.

Consequently, a multilayer and star-shaped cross-linked structure is formed at the end of reaction [93]. In contrast to the mono-functional reactive diluents, the addition of di- functional reactive diluents greatly increased the crosslink density [95], which gave rise

143 to almost ten times larger values of the elastic and viscous modulus at the end of the UV curing process. This was demonstrated by the plateau values of G’ and G” curves for M-

5-3.8 and D-5-3.8 formulations as shown in Fig. 6.6a. In fact, mono-functional methacrylated monomers are considered as viscosity and crosslinking reducers owing to the depletion of active functionalities compared to those of di-functional monomers

[214]. Fig. 6.10 shows a schematic representation of the crosslink density in formulations

M-5-3.8 and D-5-3.8.

Fig. 6.10. Schematic representation of the crosslink density in formulations M-5-3.8 and D-5-3.8. Compared to the mono-functional reactive diluent MPEGMA, the addition of di- functional reactive diluent PEGDMA greatly increases the crosslink density.

144

It has been already discussed that the addition of mono-functional reactive diluents decreased the crosslink density, while the addition of multi-functional reactive diluents resulted in the increased crosslinking [95]. Thus, the more mono-functional reactive diluents added, the lower crosslink density could be obtained, leading to decreased modulus level. Besides, owing to the lower crosslink density in the presence of higher reactive diluent content, the gel point was postponed, as indicated by the delayed crossover of G’ and G” curves at higher contents of mono-functional MPEGMA reactive diluent. The overlap of all G’ curves at the early stage after being scaled with the gel time revealed that the gelation processes showed a universal behavior independent of molecular weight, functionality, and content of the reactive diluents utilized in these systems. In general, a single dimensionless time is able to sufficiently describe the gelation mechanism for materials that are radically different in structure and properties

[215].

An ideal electrical contact must perform its role of passing electric current reliably without any change in its contact resistance for the whole duration of its life.

Hence, an ideal contact stabilization material should not have any kind of interaction with the metallic surfaces. In this research, the surface interactions of different metal surfaces, such as copper, silver, and gold in the presence of this contact stabilization material was studied by different microscopic and spectroscopic techniques. It was found that the developed material did not show any corrosion activity after six months at room temperature. In addition, there was no likelihood of physical damage or degradation to circuit boards when the connectors were coated with the contact stabilization material.

145

Electron transport through a dielectric material can be divided theoretically into two conditions, depending on whether the voltage applied across the film is large or small. When the voltage is small, the dielectric is a classically forbidden area for electrons at the Fermi level. Therefore, at absolute zero temperature, the electron transport would be via direct tunneling from one metallic electrode to the other one. In the case of large voltages, tunneling would generally occur from the cathode into the conduction band of the dielectric material (electron injection), through which the electrons move on the remainder of their path to the anode [216, 217]. Basically, this distinction implies a division of film thicknesses into two classes, because at a given voltage the current through the film is strongly dependent on the thickness. To get a measurable current at voltages of the order of a volt, the film thickness should be of the order of 100 Å, whereas for thicker films of the order of 1000 Å the current is difficult to measure in a simple manner unless the applied voltage is more than about 10 V.

Consequently, in the thicker films the observed electron transport will depend not only on the mode of electron injection but also on electron scattering in the conduction band of the dielectric, since the mean free path can be expected to be much less than 1000 Å.

Holm et al. first applied tunneling theory to the conduction mechanism of thin insulating layers between metal contacts over a small area, and explained the independence of the conductivity on temperature [218, 219]. The space-charge-limited conduction mechanism was also recognized in somewhat thicker films between two plane parallel electrodes, and Mott-Gurney law was given to describe the non-Ohmic current- voltage behavior (Eq. 6.4) [220-222].

9 푉2 퐽 = 휇휀휀 (Eq. 6.4) 8 0 푥3

146 where 휇 is the free carrier mobility, 휀 is the relative dielectric constant of the material, 휀0 is the dielectric constant of vacuum, 푥 is the film thickness, 퐽 is current density and 푉 is the applied voltage. Then, Schottky [223, 224] and Poole-Frenkel [225, 226] emission were also observed on the current transfer in an insulating polymer layer between two metal electrodes under applied electric field, depending on if the conductivity is electrode-limited or bulk-limited. For Schottky effect, the electrons are emitted from the metal electrode due to the lowering of the coulombic potential barrier under applied electric field, whereas in the case of Poole-Frenkel emission, the electrons are emitted from the bulk of the insulating material [227]. Intensive attention has then been attracted to the electrical properties of polymers arising from both practical and fundamental aspects [228]. In fact, large number of polymers were found to exhibit different conduction mechanisms under various conditions or combinations of more than one conduction mechanisms [216, 229-237]. The conduction mechanism of these gel-like materials was performed by our coworkers and their results indicated that the conductivity of these systems is more likely to be based on Schottky emission.

6.5. Conclusions

The rheological properties of UV-curable electrical contact stabilization materials were investigated by the real time small amplitude oscillatory rheology. It was found that

Winter-Chambon criteria were applicable in the evolution of dynamic modulus for the materials studied. The gel point values were evaluated as the G’-G” crossover, validated by the complex viscosity vs. oscillation frequency curves as well as tanδ vs. UV-exposure time at a series of oscillation frequencies. The real time FT-IR spectroscopy was

147 successfully utilized to study the crosslinking kinetics of the UV-curable materials. A combination of real time rheology and FT-IR spectroscopy was exploited to study the effect of reactive diluents. It was found that the use of di-functional reactive diluent or a decreased content of mono-functional reactive diluents both cause higher crosslinking density, hence higher modulus and accelerated reaction rate. The molecular weight of mono-functional reactive diluents barely affected the conversion rate or ultimate modulus due to the fact that they could only form the dangling ends in the cross-linked network.

Based on these findings, UV-curable contact stabilization coating materials which could be applied in electronics to reduce the contact resistance were successfully developed. It was shown that these materials were safe for normal processing. Additionally, their ability to be cured by UV radiation into soft gels resulted in producing materials which were not runny or leaky and could be utilized for varieties of applications. The UV- curable coatings developed in this work could deliver much more effective and environmentally friendly alternatives to the commercially available contact stabilization materials. They had no VOCs because of their 100% solids content, and could be cured almost instantaneously. These materials can be used for all electronic parts which are highly temperature-sensitive.

148

CHAPTER VII

SUMMARY

The purpose of this dissertation was to demonstrate that nonlinear polymers based on methacrylated PEGs combine a unique set of advantages and can be utilized in modern technological applications, such as superplasticizers for cement and concrete as well as UV-curable contact stabilization materials.

In the first part, well-defined comb-like polycarboxylate-ether based superplasticizers (PCEs) were synthesized using controlled RAFT polymerization and studied in cement suspensions to explain the working mechanisms of cement hydration in molecular detail. Adsorption and retardation of cement setting depend on the density and length of side chains and correlate with the amount of carboxylate groups per unit volume. A lower density of side chains and shorter side chains lead to a maximum retardation and highest adsorbed mass.

The adsorption mechanism involves migration of Ca2+ ions in the acrylate backbone to the calcium silicate hydrate surface and subsequent ion pairing of the anionic polymer backbone with the positively charged C-S-H surface. PCE adsorption thereby neutralizes the negative zeta potential of native cement paste towards positive values.

High density of intermediately long PEG side chains leads to increased conformation strain on the acrylate backbone and flat-on adsorbed conformations with periods of detachment from the surface. Lower density of intermediately long side chains leads to

149 more conformational flexibility on the surface, stronger adsorption, and partially tilted (or upright) conformations. Short side chains increase the ionic content further and maximize adsorption in tilted and upright conformations due to strong inter-chain interactions.

Fluidity, water reduction, and zeta potential neutralization follow a different trend as a function of polymer architecture in comparison to adsorption and retardation of hydration. Best dispersion of cement particles and greatest water reduction is achieved for low density of PEG side chains and intermediate length of side chains. Strong adsorption to the surface is still required yet the presence of non-ionic PEG side chains, or perhaps other nonpolar side chains, is also needed to mitigate attractive interparticle interactions. These interactions would be stronger in the presence of a strongly ionic polymer thin film (very short or no side chains). The zeta potential also shows a maximum for low density of side chains with intermediate chain length as upright- trending conformations of more highly charged polymer backbones do not support as much interfacial charge separation in an electric field.

This systematic study of PCEs demonstrates for the first time the adsorption mechanism, conformations, and orientation of the adsorbed polymer layers, which are driven by the ionic interactions with the C-S-H surface and by the geometric balance between the main chain of the polymer and the PEG side chains. Macroscopic and nanoscale measurements are consistently explained from the molecular scale, enabling specific interpretations and design criteria to tailor the mechanism of cement setting.

Limitations of prior models such as the Flatt-Gay-Raphaël model were shown and overcome by systematic experiments along with atomistic simulations.

150

The complexity and multi-scale nature of interactions in cement minerals shall nevertheless not be underestimated. The size of cement particles, the chemistry of surface regions, and the polymer networks give rise to a variety of interactions that also involve hydrodynamic, gravitational, and Brownian forces. Multiscale understanding still leaves many open questions that may be answered by a suitable combination of experiment and modeling approaches.

In the second part of this dissertation, UV-curable contact stabilization coating materials based on methacrylated PEGs and PPGs were developed. These materials could be applied in electronics to decrease the contact resistance between co-operating surfaces.

The capability of these materials for being cured by UV radiation into soft gels resulted in materials which were not leaky or runny and could be utilized for many applications.

In order to better understand the effects of reactive diluents on the final structure of these systems, the rheological properties of these materials were investigated by the real time small amplitude oscillatory rheology. It was demonstrated that Winter-Chambon criteria were applicable in the evolution of dynamic modulus. In addition, the gel points were evaluated as the crossover of G’-G”.

The real time FT-IR spectroscopy was used to investigate the crosslinking kinetics of these UV-curable systems. A combination of real time rheology and FT-IR spectroscopy showed that the use of both lower content of mono-functional reactive diluents or di-functional reactive diluents cause higher crosslinking density, therefore higher modulus and enhanced reaction rate. Moreover, the molecular weight of mono- functional reactive diluents did not affect the ultimate modulus or conversion rate because they could only create dangling ends in the cross-linked networks.

151

Development of polymers based on (meth)acrylated PEGs for industrial applications is a growing field of research having great promise for the future. With the plethora of PEG-based monomers available today, high extent of polymerization methods, and choice of mechanisms, the ever growing field of nonlinear polymers derived from (meth)acrylated PEGs has broadened the options for industrial applications during the last decade. The increasing requirements for these applications mean an ongoing search for novel polymeric materials with improved properties, which can expand the range of upcoming applications. In addition, with the new reagents and techniques in hand, the exploration of macromolecules based on (meth)acrylated PEGs is at its height, and scientists can be excited regarding the developments still to come.

152

REFERENCES

1. Akers, M.J., Excipient–Drug Interactions in Parenteral Formulations. Journal of

Pharmaceutical Sciences, 2002. 91(11): p. 2283-2300.

2. Gullapalli, R.P. and C.L. Mazzitelli, Polyethylene glycols in oral and parenteral

formulations—A critical review. International Journal of Pharmaceutics, 2015. 496(2): p.

219-239.

3. Strickley, R.G., Solubilizing Excipients in Oral and Injectable Formulations.

Pharmaceutical Research, 2004. 21(2): p. 201-230.

4. Richards, D.H. and M. Szwarc, Block polymers of ethylene oxide and its analogues with

styrene. Transactions of the Faraday Society, 1959. 55(0): p. 1644-1650.

5. Topchieva, I.N., A.I. Kuzaev, and V.P. Zubov, Modification of polyethylene glycol.

European Polymer Journal, 1988. 24(9): p. 899-904.

6. Li, J. and W.J. Kao, Synthesis of Polyethylene Glycol (PEG) Derivatives and

PEGylated−Peptide Biopolymer Conjugates. Biomacromolecules, 2003. 4(4): p. 1055-

1067.

7. Hutanu, D., et al., Recent Applications of Polyethylene Glycols (PEGs) and PEG

Derivatives. Modern Chemistry and Applications, 2014. 2(2).

8. Myers, B.K., et al., The characterization of dendronized poly(ethylene glycol)s and

poly(ethylene glycol) multi-arm stars using matrix-assisted laser desorption/ionization

time-of-flight mass spectrometry. Analytica Chimica Acta, 2014. 808: p. 175-189.

9. Zhu, M. and G. Carta, Adsorption of polyethylene-glycolated bovine serum albumin on

macroporous and polymer-grafted anion exchangers. Journal of Chromatography A,

2014. 1326: p. 29-38.

153

10. Bai, M.-Y. and S.-Z. Liu, A simple and general method for preparing antibody-PEG-

PLGA sub-micron particles using electrospray technique: An in vitro study of targeted

delivery of cisplatin to ovarian cancer cells. Colloids and Surfaces B: Biointerfaces,

2014. 117: p. 346-353.

11. Baker, D.W., et al., Development of optical probes for in vivo imaging of polarized

macrophages during foreign body reactions. Acta Biomaterialia, 2014. 10(7): p. 2945-

2955.

12. Proulx, S.T., et al., Use of a PEG-conjugated bright near-infrared dye for functional

imaging of rerouting of tumor lymphatic drainage after sentinel lymph node metastasis.

Biomaterials, 2013. 34(21): p. 5128-5137.

13. Sun, G., et al., Mono-PEGylated radix ophiopogonis polysaccharide for the treatment of

myocardial ischemia. European Journal of Pharmaceutical Sciences, 2013. 49(4): p. 629-

636.

14. Peng, F., et al., PEGylation of G-CSF in organic solvent markedly increase the efficacy

and reactivity through protein unfolding, hydrolysis inhibition and solvent effect. Journal

of Biotechnology, 2014. 170: p. 42-49.

15. Tian, L., et al., Bioplasmonic calligraphy for multiplexed label-free biodetection.

Biosensors and Bioelectronics, 2014. 59: p. 208-215.

16. Moreno, M., E. Zurita, and E. Giralt, Delivering wasp venom for cancer therapy. Journal

of Controlled Release, 2014. 182: p. 13-21.

17. Santi, D.V., E.L. Schneider, and G.W. Ashley, Macromolecular Prodrug That Provides

the Irinotecan (CPT-11) Active-Metabolite SN-38 with Ultralong Half-Life, Low Cmax ,

and Low Glucuronide Formation. Journal of Medicinal Chemistry, 2014. 57(6): p. 2303-

2314.

154

18. Saville, S.L., et al., The formation of linear aggregates in magnetic hyperthermia:

Implications on specific absorption rate and magnetic anisotropy. Journal of Colloid and

Interface Science, 2014. 424: p. 141-151.

19. Zhang, Z., et al., Highly sensitive visual detection of copper (II) using water-soluble

azide-functionalized gold nanoparticles and silver enhancement. Biosensors and

Bioelectronics, 2014. 59: p. 40-44.

20. Gajbhiye, V., et al., Drug-loaded nanoparticles induce gene expression in human

pluripotent stem cell derivatives. Nanoscale, 2014. 6(1): p. 521-531.

21. Miller, D.J., et al., Constant flux crossflow filtration evaluation of surface-modified

fouling-resistant membranes. Journal of Membrane Science, 2014. 452: p. 171-183.

22. Petek, N.A. and R.D. Mullins, Chapter Two - Bacterial Actin-Like Proteins: Purification

and Characterization of Self-Assembly Properties, in Methods in Enzymology, D.V.

Ronald, Editor. 2014, Academic Press. p. 19-34.

23. Hoehlig, K., et al., A Novel C5a-neutralizing Mirror-image (l-)Aptamer Prevents Organ

Failure and Improves Survival in Experimental Sepsis. Mol Ther, 2013. 21(12): p. 2236-

2246.

24. van de Manakker, F., et al., Self-Assembling Hydrogels Based on β-

Cyclodextrin/Cholesterol Inclusion Complexes. Macromolecules, 2008. 41(5): p. 1766-

1773.

25. Maitz, M.F., et al., Bio-responsive polymer hydrogels homeostatically regulate blood

coagulation. Nature Communications, 2013. 4: p. 2168.

26. Hao, Y., et al., Visible light cured thiol-vinyl hydrogels with tunable degradation for 3D

cell culture. Acta Biomaterialia, 2014. 10(1): p. 104-114.

27. Xue, X., et al., Heat treatment increases the bioactivity of C-terminally PEGylated

staphylokinase. Process Biochemistry, 2014. 49(7): p. 1092-1096.

155

28. Kang, T., et al., iNGR-modified PEG-PLGA nanoparticles that recognize tumor

vasculature and penetrate gliomas. Biomaterials, 2014. 35(14): p. 4319-4332.

29. Bai, J., Y. Liu, and X. Jiang, Multifunctional PEG-GO/CuS nanocomposites for near-

infrared chemo-photothermal therapy. Biomaterials, 2014. 35(22): p. 5805-5813.

30. Kong, X., et al., A novel multifunctional poly(amidoamine) dendrimeric delivery system

with superior encapsulation capacity for targeted delivery of the chemotherapy drug 10-

hydroxycamptothecin. International Journal of Pharmaceutics, 2014. 465(1–2): p. 378-

387.

31. Kim, K.S., et al., A cancer-recognizable MRI contrast agents using pH-responsive

polymeric micelle. Biomaterials, 2014. 35(1): p. 337-343.

32. Song, W., et al., Anti-tumor efficacy of c(RGDfK)-decorated polypeptide-based micelles

co-loaded with docetaxel and cisplatin. Biomaterials, 2014. 35(9): p. 3005-3014.

33. Xu, J., E. Feng, and J. Song, Bioorthogonally Cross-Linked Hydrogel Network with

Precisely Controlled Disintegration Time over a Broad Range. Journal of the American

Chemical Society, 2014. 136(11): p. 4105-4108.

34. Zhang, Z., et al., Synthesis of Poly(ethylene glycol)-based Hydrogels via Amine-Michael

Type Addition with Tunable Stiffness and Postgelation Chemical Functionality.

Chemistry of Materials, 2014. 26(12): p. 3624-3630.

35. McKinnon, D.D., et al., Design and Characterization of a Synthetically Accessible,

Photodegradable Hydrogel for User-Directed Formation of Neural Networks.

Biomacromolecules, 2014. 15(7): p. 2808-2816.

36. Dimitrov, I. and C.B. Tsvetanov, 4.21 - High-Molecular-Weight Poly(ethylene oxide) A2

- Matyjaszewski, Krzysztof, in Polymer Science: A Comprehensive Reference, M. Möller,

Editor. 2012, Elsevier: Amsterdam. p. 551-569.

156

37. Knop, K., et al., Poly(ethylene glycol) in Drug Delivery: Pros and Cons as Well as

Potential Alternatives. Angewandte Chemie International Edition, 2010. 49(36): p. 6288-

6308.

38. Pelegri-O’Day, E.M., E.-W. Lin, and H.D. Maynard, Therapeutic Protein–Polymer

Conjugates: Advancing Beyond PEGylation. Journal of the American Chemical Society,

2014. 136(41): p. 14323-14332.

39. Dingels, C. and H. Frey, From Biocompatible to Biodegradable: Poly(Ethylene Glycol)s

with Predetermined Breaking Points, in Hierarchical Macromolecular Structures: 60

Years after the Staudinger Nobel Prize II, V. Percec, Editor. 2013, Springer International

Publishing: Cham. p. 167-190.

40. Herzberger, J., et al., Polymerization of Ethylene Oxide, Propylene Oxide, and Other

Alkylene Oxides: Synthesis, Novel Polymer Architectures, and Bioconjugation. Chemical

Reviews, 2016. 116(4): p. 2170-2243.

41. Lutz, J.-F., Ö. Akdemir, and A. Hoth, Point by Point Comparison of Two

Thermosensitive Polymers Exhibiting a Similar LCST: Is the Age of Poly(NIPAM) Over?

Journal of the American Chemical Society, 2006. 128(40): p. 13046-13047.

42. Lutz, J.-F. and A. Hoth, Preparation of Ideal PEG Analogues with a Tunable

Thermosensitivity by Controlled Radical Copolymerization of 2-(2-Methoxyethoxy)ethyl

Methacrylate and Oligo(ethylene glycol) Methacrylate. Macromolecules, 2006. 39(2): p.

893-896.

43. Han, S., M. Hagiwara, and T. Ishizone, Synthesis of Thermally Sensitive Water-Soluble

Polymethacrylates by Living Anionic Polymerizations of Oligo(ethylene glycol) Methyl

Ether Methacrylates. Macromolecules, 2003. 36(22): p. 8312-8319.

44. Zhao, B., et al., Synthesis of Thermosensitive Water-Soluble Polystyrenics with Pendant

Methoxyoligo(ethylene glycol) Groups by Nitroxide-Mediated Radical Polymerization.

Macromolecules, 2005. 38(23): p. 9509-9517.

157

45. Neugebauer, D., Graft copolymers with poly(ethylene oxide) segments. Polymer

International, 2007. 56(12): p. 1469-1498.

46. Lutz, J.-F., Polymerization of oligo(ethylene glycol) (meth)acrylates: Toward new

generations of smart biocompatible materials. Journal of Polymer Science Part A:

Polymer Chemistry, 2008. 46(11): p. 3459-3470.

47. Aïtcin, P.C., 3 - Portland cement, in Science and Technology of Concrete Admixtures.

2016, Woodhead Publishing. p. 27-51.

48. Aïtcin, P.-C., Binders for Durable and Sustainable Concrete. 2007: CRC Press. 528.

49. Marchon, D. and R.J. Flatt, 8 - Mechanisms of cement hydration, in Science and

Technology of Concrete Admixtures. 2016, Woodhead Publishing. p. 129-145.

50. Tenoutasse, N. The hydration mechanism of C3A and C3S in the presence of calcium

chloride and calcium sulfate. in Proceedings of the 5th International Symposium on the

Chemistry of Cement. 1968.

51. Tsubakimoto, T., M. Hosoido, and H. Tahara, Copolymer and method for manufacture

thereof. 1984, EP Patent.

52. Flatt, R.J., et al., Conformation of Adsorbed Comb Copolymer Dispersants. Langmuir,

2009. 25(2): p. 845-855.

53. Marchon, D., et al., Molecular design of comb-shaped polycarboxylate dispersants for

environmentally friendly concrete. Soft Matter, 2013. 9(45): p. 10719-10728.

54. Rinaldi, D., et al., RAFT copolymerization of methacrylic acid and poly(ethylene glycol)

methyl ether methacrylate in the presence of a hydrophobic chain transfer agent in

organic solution and in water. Journal of Polymer Science Part A: Polymer Chemistry,

2009. 47(12): p. 3045-3055.

55. Pourchet, S., et al., Effect of the repartition of the PEG side chains on the adsorption and

dispersion behaviors of PCP in presence of sulfate. Cement and Concrete Research,

2012. 42(2): p. 431-439.

158

56. Hirata, T., et al., Cement additive. 2006, Google Patents.

57. Amaya, T., et al., Cement dispersant and concrete composition containing the dispersant.

2002, Google Patents.

58. Gelardi, G., et al., 9 - Chemistry of chemical admixtures, in Science and Technology of

Concrete Admixtures. 2016, Woodhead Publishing. p. 149-218.

59. Miao, C., et al., Preparation method of hyperbranched polycarboxylic acid type

copolymer cement dispersant. 2013, US Patent 0102749 A1.

60. Liu, M., et al., Synthesis and properties of methacrylate-based and allylether-based

polycarboxylate superplasticizer in cementitious system. Journal of Sustainable Cement-

Based Materials, 2013. 2(3-4): p. 218-226.

61. Winnefeld, F., et al., Effects of the Molecular Architecture of Comb-Shaped

Superplasticizers on their Performance in Cementitious Systems. Cement and Concrete

Composites, 2007. 29(4): p. 251-262.

62. Plank, J., et al., Synthesis and performance of methacrylic ester based polycarboxylate

superplasticizers possessing hydroxy terminated poly(ethylene glycol) side chains.

Cement and Concrete Research, 2008. 38(10): p. 1210-1216.

63. Zingg, A., et al., Interaction of polycarboxylate-based superplasticizers with cements

containing different C3A amounts. Cement and Concrete Composites, 2009. 31(3): p.

153-162.

64. Ran, Q., et al., Effect of the length of the side chains of comb-like copolymer dispersants

on dispersion and rheological properties of concentrated cement suspensions. Journal of

Colloid and Interface Science, 2009. 336(2): p. 624-633.

65. Plank, J. and C. Hirsch, Impact of zeta potential of early cement hydration phases on

superplasticizer adsorption. Cement and Concrete Research, 2007. 37(4): p. 537-542.

66. Büyükyağcı, A., G. Tuzcu, and L. Aras, Synthesis of copolymers of methoxy polyethylene

glycol acrylate and 2-acrylamido-2-methyl-1-propanesulfonic acid: Its characterization

159

and application as superplasticizer in concrete. Cement and Concrete Research, 2009.

39(7): p. 629-635.

67. Alonso, M.M., M. Palacios, and F. Puertas, Compatibility between polycarboxylate-based

admixtures and blended-cement pastes. Cement and Concrete Composites, 2013. 35(1):

p. 151-162.

68. Lange, A., T. Hirata, and J. Plank, Influence of the HLB value of polycarboxylate

superplasticizers on the flow behavior of mortar and concrete. Cement and Concrete

Research, 2014. 60: p. 45-50.

69. Liu, X., et al., Synthesis, characterization and performance of a polycarboxylate

superplasticizer with amide structure. Colloids and Surfaces A: Physicochemical and

Engineering Aspects, 2014. 448: p. 119-129.

70. Kong, X., Z. Shi, and Z. Lu, Synthesis of novel polymer nano-particles and their

interaction with cement. Construction and Building Materials, 2014. 68: p. 434-443.

71. Houst, Y.F., et al., Design and function of novel superplasticizers for more durable high

performance concrete (superplast project). Cement and Concrete Research, 2008. 38(10):

p. 1197-1209.

72. Dalas, F., et al., Fluidizing efficiency of comb-like superplasticizers: The effect of the

anionic function, the side chain length and the grafting degree. Cement and Concrete

Research, 2015. 71: p. 115-123.

73. Dalas, F., et al., Tailoring the anionic function and the side chains of comb-like

superplasticizers to improve their adsorption. Cement and Concrete Research, 2015. 67:

p. 21-30.

74. Zhang, Y.-R., et al., Effects of the charge characteristics of polycarboxylate

superplasticizers on the adsorption and the retardation in cement pastes. Cement and

Concrete Research, 2015. 67: p. 184-196.

160

75. Barret, P. and D. Bertrandie, Saturated Solutions of Anhydrous Phases in the System

Lime-Silica-Water: Example of β-C2S. Journal of the American Ceramic Society, 1990.

73(11): p. 3486-3492.

76. Labbez, C., et al., Surface Charge Density and Electrokinetic Potential of Highly

Charged Minerals: Experiments and Monte Carlo Simulations on Calcium Silicate

Hydrate. The Journal of Physical Chemistry B, 2006. 110(18): p. 9219-9230.

77. Sowoidnich, T., et al., Calcium complexation and cluster formation as principal modes of

action of polymers used as superplasticizer in cement systems. Cement and Concrete

Research, 2015. 73: p. 42-50.

78. Antler, M., Effect of Surface Contamination on Electric Contact Performance, in Treatise

on Clean Surface Technology, K.L. Mittal, Editor. 1987, Springer US. p. 179-203.

79. DeGarmo, E.P., J.T. Black, and R.A. Kohser, Materials and processes in manufacturing.

7th ed. 1988: New York : Macmillan Pub. Co., c1988.

80. Holm, R., Introduction. A simplified summary of the theory of stationary electric

contacts, in Electric Contacts. 1967, Springer Berlin Heidelberg. p. 1-7.

81. Hall, P.M., Resistance calculations for thin film patterns. Thin Solid Films, 1968. 1(4): p.

277-295.

82. Zhang, P., Effects of Surface Roughness on Electrical Contact, RF Heating and Field

Enhancement. 2012, University of Michigan: Ann Arbor. p. 167.

83. Timsit, R.S., Electrical contact resistance: properties of stationary interfaces.

Components and Packaging Technologies, IEEE Transactions on, 1999. 22(1): p. 85-98.

84. Chen, X., et al., Protective coating on stainless steel interconnect for SOFCs: oxidation

kinetics and electrical properties. Solid State Ionics, 2005. 176(5–6): p. 425-433.

85. Greenwood, J.A., Constriction resistance and the real area of contact. British Journal of

Applied Physics, 1966. 17(12): p. 1621.

161

86. Wright, W.M.D., Contact stabilization coating material for electrical contact surfaces

and method. 1987, Google Patents.

87. Ostendorf, F., T. Wielsch, and M. Reiniger. The Effects of Lubrication on Electroplated

Tin Surfaces: A Systematic Approach by DOE-Methodology. in Electrical Contacts

(Holm), 2012 IEEE 58th Holm Conference on. 2012.

88. Antler, M. Corrosion control and lubrication of plated noble metal connector contacts. in

Electrical Contacts, 1995., Proceedings of the Forty-First IEEE Holm Conference on.

1995.

89. Abbott, W.H. and J. Whitley, The Lubrication and Environmental Protection of

Alternatives to Gold for Electronic Connectors. Parts, Hybrids, and Packaging, IEEE

Transactions on, 1976. 12(1): p. 51-56.

90. Javadi, A., et al., Cure-on-command technology: A review of the current state of the art.

Progress in Organic Coatings, 2016. 100: p. 2-31.

91. Tyagi, A.K., V. Choudhary, and I. Varma, Effect of reactive diluents on curing behaviour

and thermal stability of urethane methacrylate. Die Angewandte Makromolekulare

Chemie, 1991. 189(1): p. 105-115.

92. Oraby, W. and W.K. Walsh, Elastomeric electron beam‐cured coatings: Structure–

property relationships. II. Chain transfer agents. Journal of Applied Polymer Science,

1979. 23(11): p. 3243-3254.

93. Oraby, W. and W.K. Walsh, Elastomeric electron beam‐cured coatings: Structure–

property relationships. I. Oligomer structure. Journal of Applied Polymer Science, 1979.

23(11): p. 3227-3242.

94. Speckhard, T., et al., Properties of UV‐curable polyurethane acrylates: Effect of reactive

diluent. Journal of Applied Polymer Science, 1985. 30(2): p. 647-666.

95. Schmidle, C.J., ULTRAVIOLET CURABLE FLEXIBLE COATINGS. Journal of Coated

Fabrics, 1978. 8(1): p. 10-20.

162

96. Zong, Z., et al., Cationic photopolymerization of epoxynorbornane linseed oils: The effect

of diluents. Journal of Polymer Science Part A: Polymer Chemistry, 2003. 41(21): p.

3440-3456.

97. Wang, J. and V.M. Ugaz, Using in situ rheology to characterize the microstructure in

photopolymerized polyacrylamide gels for DNA electrophoresis. Electrophoresis, 2006.

27(17): p. 3349-3358.

98. Khan, S.A., I.M. Plitz, and R.A. Frantz, In situ technique for monitoring the gelation of

UV curable polymers. Rheologica Acta, 1992. 31(2): p. 151-160.

99. Chiou, B.-S., S.R. Raghavan, and S.A. Khan, Effect of colloidal fillers on the cross-

linking of a UV-curable polymer: gel point rheology and the Winter-Chambon criterion.

Macromolecules, 2001. 34(13): p. 4526-4533.

100. Chiou, B.-S., R.J. English, and S.A. Khan, Rheology and Photo-Cross-Linking of

Thiol−Ene Polymers. Macromolecules, 1996. 29(16): p. 5368-5374.

101. Chiou, B.-S. and S.A. Khan, Real-Time FTIR and in Situ Rheological Studies on the UV

Curing Kinetics of Thiol-ene Polymers. Macromolecules, 1997. 30(23): p. 7322-7328.

102. Decker, C. and K. Moussa, A new method for monitoring ultra‐fast photopolymerizations

by real‐time infra‐red (RTIR) spectroscopy. Die Makromolekulare Chemie, 1988.

189(10): p. 2381-2394.

103. Decker, C. and K. Moussa, Real-time kinetic study of laser-induced polymerization.

Macromolecules, 1989. 22(12): p. 4455-4462.

104. Winter, H.H. and F. Chambon, Analysis of linear viscoelasticity of a crosslinking

polymer at the gel point. Journal of Rheology (1978-present), 1986. 30(2): p. 367-382.

105. Winter, H., Can the gel point of a cross‐linking polymer be detected by the G′–G

″crossover? Polymer Engineering & Science, 1987. 27(22): p. 1698-1702.

163

106. Jacobine, A., et al., Photocrosslinked norbornene–thiol copolymers: Synthesis,

mechanical properties, and cure studies. Journal of Applied Polymer Science, 1992.

45(3): p. 471-485.

107. Udagawa, A., F. Sakurai, and T. Takahashi, In situ study of photopolymerization by

fourier transform infrared spectroscopy. Journal of Applied Polymer Science, 1991.

42(7): p. 1861-1867.

108. Yoshioka, K., et al., Role of Steric Hindrance in the Performance of Superplasticizers for

Concrete. Journal of the American Ceramic Society, 1997. 80(10): p. 2667-2671.

109. Ollitrault-Fichet, R., et al., Microstructural aspects in a polymer-modified cement.

Cement and Concrete Research, 1998. 28(12): p. 1687-1693.

110. Uchikawa, H., S. Hanehara, and D. Sawaki, The role of steric repulsive force in the

dispersion of cement particles in fresh paste prepared with organic admixture. Cement

and Concrete Research, 1997. 27(1): p. 37-50.

111. Yamada, K., et al., Effects of the chemical structure on the properties of polycarboxylate-

type superplasticizer. Cement and Concrete Research, 2000. 30(2): p. 197-207.

112. Borget, P., et al., Microstructural characterisation and behaviour in different salt

solutions of sodium polymethacrylate-g-PEO comb copolymers. Colloids and Surfaces A:

Physicochemical and Engineering Aspects, 2005. 260(1–3): p. 173-182.

113. Braunecker, W.A. and K. Matyjaszewski, Controlled/living radical polymerization:

Features, developments, and perspectives. Progress in Polymer Science, 2007. 32(1): p.

93-146.

114. Chiefari, J., et al., Living Free-Radical Polymerization by Reversible

Addition−Fragmentation Chain Transfer: The RAFT Process. Macromolecules, 1998.

31(16): p. 5559-5562.

164

115. Barner-Kowollik, C. and S. Perrier, The future of reversible addition fragmentation chain

transfer polymerization. Journal of Polymer Science Part A: Polymer Chemistry, 2008.

46(17): p. 5715-5723.

116. Mertoglu, M., et al., Stimuli responsive amphiphilic block copolymers for aqueous media

synthesised via reversible addition fragmentation chain transfer polymerisation (RAFT).

Polymer, 2005. 46(18): p. 7726-7740.

117. Fournier, D., et al., Tunable pH- and Temperature-Sensitive Copolymer Libraries by

Reversible Addition−Fragmentation Chain Transfer Copolymerizations of

Methacrylates. Macromolecules, 2007. 40(4): p. 915-920.

118. Bouhamed, H., S. Boufi, and A. Magnin, Dispersion of alumina suspension using comb-

like and diblock copolymers produced by RAFT polymerization of AMPS and MPEG.

Journal of Colloid and Interface Science, 2007. 312(2): p. 279-291.

119. Zhang, L., et al., Shell-Cross-Linked Micelles Containing Cationic Polymers Synthesized

via the RAFT Process: Toward a More Biocompatible Gene Delivery System.

Biomacromolecules, 2007. 8(9): p. 2890-2901.

120. Chen, Y., et al., Poly(vinylidene fluoride) with Grafted Poly(ethylene glycol) Side Chains

via the RAFT-Mediated Process and Pore Size Control of the Copolymer Membranes.

Macromolecules, 2003. 36(25): p. 9451-9457.

121. Cheng, et al., Brush-Type Amphiphilic Diblock Copolymers from “Living”/Controlled

Radical Polymerizations and Their Aggregation Behavior. Langmuir, 2005. 21(16): p.

7180-7185.

122. Becer, C.R., et al., Libraries of methacrylic acid and oligo(ethylene glycol) methacrylate

copolymers with LCST behavior. Journal of Polymer Science Part A: Polymer Chemistry,

2008. 46(21): p. 7138-7147.

165

123. Khousakoun, E., J.-F. Gohy, and R. Jérôme, Self-association of double-hydrophilic

copolymers of acrylic acid and poly(ethylene oxide) macromonomer. Polymer, 2004.

45(25): p. 8303-8310.

124. Perrier, S. and P. Takolpuckdee, Macromolecular design via reversible addition–

fragmentation chain transfer (RAFT)/xanthates (MADIX) polymerization. Journal of

Polymer Science Part A: Polymer Chemistry, 2005. 43(22): p. 5347-5393.

125. Favier, A. and M.-T. Charreyre, Experimental Requirements for an Efficient Control of

Free-Radical Polymerizations via the Reversible Addition-Fragmentation Chain Transfer

(RAFT) Process. Macromolecular Rapid Communications, 2006. 27(9): p. 653-692.

126. Moad, G., E. Rizzardo, and S.H. Thang, Radical addition–fragmentation chemistry in

polymer synthesis. Polymer, 2008. 49(5): p. 1079-1131.

127. Baussard, J.-F., et al., New chain transfer agents for reversible addition-fragmentation

chain transfer (RAFT) polymerisation in aqueous solution. Polymer, 2004. 45(11): p.

3615-3626.

128. Ito, K., et al., Poly(ethylene oxide) macromonomers. 7. Micellar polymerization in water.

Macromolecules, 1991. 24(9): p. 2348-2354.

129. Stein, H.N. and J.M. Stevels, Influence of silica on the hydration of 3 CaO,SiO2. Journal

of Applied Chemistry, 1964. 14(8): p. 338-346.

130. Garrault, S., et al., Study of C-S-H growth on C3S surface during its early hydration.

Materials and Structures, 2005. 38(4): p. 435-442.

131. Garrault, S., T. Behr, and A. Nonat, Formation of the C−S−H Layer during Early

Hydration of Tricalcium Silicate Grains with Different Sizes. The Journal of Physical

Chemistry B, 2006. 110(1): p. 270-275.

132. Bazzoni, A., Study of early hydration mechanisms of cement by means of electron

microscopy. 2014: Lausanne: EPFL.

166

133. Nkinamubanzi, P.C., S. Mantellato, and R.J. Flatt, 16 - Superplasticizers in practice, in

Science and Technology of Concrete Admixtures. 2016, Woodhead Publishing. p. 353-

377.

134. Roussel, N., 4 - From industrial testing to rheological parameters for concrete, in

Understanding the Rheology of Concrete, N. Roussel, Editor. 2012, Woodhead

Publishing. p. 83-95.

135. Hot, J., et al., Adsorbing polymers and viscosity of cement pastes. Cement and Concrete

Research, 2014. 63: p. 12-19.

136. Marchon, D. and R.J. Flatt, 12 - Impact of chemical admixtures on cement hydration, in

Science and Technology of Concrete Admixtures. 2016, Woodhead Publishing. p. 279-

304.

137. Ng, S. and J. Plank, Interaction mechanisms between Na montmorillonite clay and

MPEG-based polycarboxylate superplasticizers. Cement and Concrete Research, 2012.

42(6): p. 847-854.

138. Roussel, N., A Theoretical Frame to Study Stability of Fresh Concrete. Materials and

Structures, 2006. 39(1): p. 81-91.

139. Yamada, K., Basics of analytical methods used for the investigation of interaction

mechanism between cements and superplasticizers. Cement and Concrete Research,

2011. 41(7): p. 793-798.

140. Hanehara, S. and K. Yamada, Rheology and early age properties of cement systems.

Cement and Concrete Research, 2008. 38(2): p. 175-195.

141. Marchon, D., et al., 10 - Adsorption of chemical admixtures, in Science and Technology

of Concrete Admixtures. 2016, Woodhead Publishing. p. 219-256.

142. Vickers Jr, T.M., et al., Influence of dispersant structure and mixing speed on concrete

slump retention. Cement and Concrete Research, 2005. 35(10): p. 1882-1890.

167

143. Flatt, R.J., Towards a prediction of superplasticized concrete rheology. Materials and

Structures, 2004. 37(5): p. 289-300.

144. Flatt, R.J., Dispersion forces in cement suspensions. Cement and Concrete Research,

2004. 34(3): p. 399-408.

145. Yoshioka, K., et al., Adsorption characteristics of superplasticizers on cement component

minerals. Cement and Concrete Research, 2002. 32(10): p. 1507-1513.

146. Zingg, A., et al., Adsorption of polyelectrolytes and its influence on the rheology, zeta

potential, and microstructure of various cement and hydrate phases. Journal of Colloid

and Interface Science, 2008. 323(2): p. 301-312.

147. Kirby, G.H. and J.A. Lewis, Comb Polymer Architecture Effects on the Rheological

Property Evolution of Concentrated Cement Suspensions. Journal of the American

Ceramic Society, 2004. 87(9): p. 1643-1652.

148. Li, C.-Z., et al., Effects of polyethlene oxide chains on the performance of

polycarboxylate-type water-reducers. Cement and Concrete Research, 2005. 35(5): p.

867-873.

149. Lei, L. and J. Plank, Synthesis, working mechanism and effectiveness of a novel

cycloaliphatic superplasticizer for concrete. Cement and Concrete Research, 2012. 42(1):

p. 118-123.

150. Lv, S., et al., Synthesis of Modified Chitosan Superplasticizer by Amidation and

Sulfonation and Its Application Performance and Working Mechanism. Industrial &

Engineering Chemistry Research, 2014. 53(10): p. 3908-3916.

151. Li, Y., et al., Synthesis of copolymers with cyclodextrin as pendants and its end group

effect as superplasticizer. Carbohydrate Polymers, 2014. 102: p. 278-287.

152. Plank, J., B. Sachsenhauser, and J. de Reese, Experimental determination of the

thermodynamic parameters affecting the adsorption behaviour and dispersion

168

effectiveness of PCE superplasticizers. Cement and Concrete Research, 2010. 40(5): p.

699-709.

153. Roussel, N., C. Stefani, and R. Leroy, From mini-cone test to Abrams cone test:

measurement of cement-based materials yield stress using slump tests. Cement and

Concrete Research, 2005. 35(5): p. 817-822.

154. Lv, S., et al., Effects of connection mode between carboxyl groups and main chains on

polycarboxylate superplasticizer properties. Journal of Applied Polymer Science, 2013.

128(6): p. 3925-3932.

155. Flatt, R., J., Analysis of superplasticizers used in concrete. Analusis, 1998. 26(2): p. 28-

34.

156. Fleer, G., Polymers at Interfaces. 1993: Springer Science & Business Media.

157. Kirby, G.H., et al., Poly(acrylic acid)–Poly(ethylene oxide) Comb Polymer Effects on

BaTiO3 Nanoparticle Suspension Stability. Journal of the American Ceramic Society,

2004. 87(2): p. 181-186.

158. Papo, A. and L. Piani, Effect of various superplasticizers on the rheological properties of

Portland cement pastes. Cement and Concrete Research, 2004. 34(11): p. 2097-2101.

159. Mishra, R.K., R.J. Flatt, and H. Heinz, Force Field for Tricalcium Silicate and Insight

into Nanoscale Properties: Cleavage, Initial Hydration, and Adsorption of Organic

Molecules. Journal of Physical Chemistry C, 2013. 117(20): p. 10417-10432.

160. Heinz, H., R.A. Vaia, and B.L. Farmer, Interaction Energy and Surface Reconstruction

Between Sheets of Layered Silicates. The Journal of Chemical Physics, 2006. 124(22): p.

224713.

161. Fu, Y.T. and H. Heinz, Cleavage Energy of Alkylammonium-Modified Montmorillonite

and Relation to Exfoliation in Nanocomposites: Influence of Cation Density, Head Group

Structure, and Chain Length. Chemistry of Materials, 2010. 22(4): p. 1595-1605.

169

162. Mishra, R.K., et al., A Force Field for Tricalcium Aluminate to Characterize Surface

Properties, Initial Hydration, and Organically Modified Interfaces in Atomic Resolution.

Dalton Transactions, 2014. 43(27): p. 10602-10616.

163. Russel, W.B., Saville, D.A., Schowalter, W.R., Colloidal Dispersions. 1992: Cambridge

University Press.

164. Gelardi, G. and R.J. Flatt, 11 - Working mechanisms of water reducers and

superplasticizers, in Science and Technology of Concrete Admixtures. 2016, Woodhead

Publishing. p. 257-278.

165. Yang, M., C.M. Neubauer, and H.M. Jennings, Interparticle potential and sedimentation

behavior of cement suspensions. Advanced Cement Based Materials, 1997. 5(1): p. 1-7.

166. Flatt, R.J. and P. Bowen, Electrostatic repulsion between particles in cement

suspensions: Domain of validity of linearized Poisson–Boltzmann equation for nonideal

electrolytes. Cement and Concrete Research, 2003. 33(6): p. 781-791.

167. Pellenq, R.J.M., J.M. Caillol, and A. Delville, Electrostatic Attraction between Two

Charged Surfaces: A (N,V,T) Monte Carlo Simulation. The Journal of Physical

Chemistry B, 1997. 101(42): p. 8584-8594.

168. Lesko, S., et al., Investigation by atomic force microscopy of forces at the origin of

cement cohesion. Ultramicroscopy, 2001. 86(1–2): p. 11-21.

169. Plank, J. and B. Sachsenhauser, Impact of Molecular Structure on Zeta Potential and

Adsorbed Conformation of α-Allyl-ω-Methoxypolyethylene Glycol -

Maleic Anhydride Superplasticizers. Journal of Advanced Concrete Technology, 2006.

4(2): p. 233-239.

170. Plank, J. and M. Gretz, Study on the interaction between anionic and cationic latex

particles and Portland cement. Colloids and Surfaces A: Physicochemical and

Engineering Aspects, 2008. 330(2–3): p. 227-233.

170

171. Plank J, S.R., Vlad D, Brandl A, Chatziagorastou P., Kolloidchemische Aspekte zur

Verflüssigung von Zementleimen mit Polycarboxylaten. GDCh-Monographie Band, 2004.

31: p. 58-69.

172. Neubauer, C.M., M. Yang, and H.M. Jennings, Interparticle Potential and Sedimentation

Behavior of Cement Suspensions: Effects of Admixtures. Advanced Cement Based

Materials, 1998. 8(1): p. 17-27.

173. Suzuki, K., et al., Approach by zeta-potential measurement on the surface change of

hydrating C3S. Cement and Concrete Research, 1981. 11(5): p. 759-764.

174. Singh, N.B. and P.N. Ojha, Changes in Zeta Potential During Hydration of C3S With and

Without CaCl2 Additions. Journal of the American Ceramic Society, 1981. 64(7): p. C-

99-C-99.

175. Heinz, H., et al., Thermodynamically Consistent Force Fields for the Assembly of

Inorganic, Organic, and Biological Nanostructures: The INTERFACE Force Field.

Langmuir, 2013. 29(6): p. 1754-1765.

176. Zheng, H., et al., Data Mining of Metal Ion Environments Present in Protein Structures.

Journal of inorganic biochemistry, 2008. 102(9): p. 1765-1776.

177. Feng, J., et al., Influence of the Shape of Nanostructured Metal Surfaces on Adsorption of

Single Peptide Molecules in Aqueous Solution. Small, 2012. 8(7): p. 1049-1059.

178. Emami, F.S., et al., Prediction of Specific Biomolecule Adsorption on Silica Surfaces as a

Function of pH and Particle Size. Chemistry of Materials, 2014. 26(19): p. 5725–5734.

179. Gay, C. and E. Raphaël, Comb-Like Polymers Inside Nanoscale Pores. Adv. Colloid

Interface Sci., 2001. 94(1–3): p. 229-236.

180. Kauppi, A., K.M. Andersson, and L. Bergström, Probing the Effect of Superplasticizer

Adsorption on the Surface Forces Using the Colloidal Probe AFM Technique. Cement

and Concrete Research, 2005. 35(1): p. 133-140.

171

181. Giraudeau, C., Organo-aluminate interaction in cement. Intercalation of

polymethacrylate-g-PEO into hycrocalumite. 2009, Université Pierre et Marie Curie -

Paris VI.

182. Pedersen, J.S. and C. Sommer, Temperature dependence of the virial coefficients and the

chi parameter in semi-dilute solutions of PEG. Progress in Colloid and Polymer Science,

2005. 130: p. 70-78.

183. Flatt, R. and I. Schober, 7 - Superplasticizers and the Rheology of Concrete, in

Understanding the Rheology of Concrete, N. Roussel, Editor. 2012, Woodhead

Publishing. p. 144-208.

184. Cheung, J., et al., Impact of Admixtures on the Hydration Kinetics of Portland Cement.

Cement and Concrete Research, 2011. 41(12): p. 1289-1309.

185. Schennach, R., et al., A Review of Cement–Superplasticizer Interactions and their

Models. Advances in Cement Research, 2000. 12(4): p. 153-161.

186. Flatt, R. and I. Schober, 7 - Superplasticizers and the rheology of concrete A2 - Roussel,

Nicolas, in Understanding the Rheology of Concrete. 2012, Woodhead Publishing. p.

144-208.

187. Kjeldsen, A.M., R.J. Flatt, and L. Bergström, Relating the Molecular Structure of Comb-

Type Superplasticizers to the Compression Rheology of MgO Suspensions. Cement and

Concrete Research, 2006. 36(7): p. 1231-1239.

188. Mastrangelo, S.V.R., Insulative composition for forming polymeric electric current

regulating junctions. 1982, US4359414 A.

189. Cuddy, W.A., B. Thir, and S.E. Eisenstein, Printed circuit board soldering. 1982,

US4360144 A.

190. Angell, C.A., K. Xu, and C. Liu, Lithium ion conducting ionic electrolytes. 1996, US

08/262,158.

172

191. Wright, W.M.D., Contact stabilization coating material for electrical contact surfaces

and method. 1987, US4696832 A.

192. Koleske, J.V., Radiation Curing of Coatings. 2002: ASTM International

193. Kinstle, J.F., Radiation polymerization. Paint and Varnish Production, 1973. 63(6): p. 17-

23.

194. Lake, R.T., Ultraviolet Curing of Organic Coatings. Radiation Curing, 1983. 10(2): p.

18.

195. Lawson, K. Status of UV/EB Curing in North America-2000. in RadTech 2000. 2000.

Baltimore, MD.

196. Schwalm, R., CHAPTER 2 - The UV Curing Process, in UV Coatings, R. Schwalm,

Editor. 2007, Elsevier: Amsterdam. p. 19-61.

197. Stowe, R.W. Some economic factors of UV curing. in RadTech. 1994. North America:

Orlando, FL.

198. Hume, P.S., C.N. Bowman, and K.S. Anseth, Functionalized PEG hydrogels through

reactive dip-coating for the formation of immunoactive barriers. Biomaterials, 2011.

32(26): p. 6204-6212.

199. Fairbanks, B.D., et al., Photoinitiated polymerization of PEG-diacrylate with lithium

phenyl-2,4,6-trimethylbenzoylphosphinate: polymerization rate and cytocompatibility.

Biomaterials, 2009. 30(35): p. 6702-6707.

200. Stafford, C.M., et al., Generating thickness gradients of thin polymer films via flow

coating. Review of Scientific Instruments, 2006. 77(2): p. 023908.

201. and, S.J.W. and K.J. Ruschak, COATING FLOWS. Annual Review of Fluid Mechanics,

2004. 36(1): p. 29-53.

202. Ruschak, K.J., Coating Flows. Annual Review of Fluid Mechanics, 1985. 17(1): p. 65-

89.

203. Doi, Y., et al., Rheology: principles, measurements, and applications. 1994.

173

204. Hiemenz, P.C. and T.P. Lodge, Polymer chemistry. 2007: CRC press.

205. Cox, W. and E. Merz, Correlation of dynamic and steady flow viscosities. Journal of

Polymer Science, 1958. 28(118): p. 619-622.

206. Morgan, C., F. Magnotta, and A. Ketley, Thiol/ene photocurable polymers. Journal of

Polymer Science: Polymer Chemistry Edition, 1977. 15(3): p. 627-645.

207. Odian, G., Principles of polymerization. 2004: John Wiley & Sons.

208. Harkal, U.D., A.J. Muehlberg, and D.C. Webster, UV curable glycidyl carbamate based

resins. Progress in Organic Coatings, 2012. 73(1): p. 19-25.

209. Allen, C.C., et al., The Radiation-Induced Copolymerization of Methyl Methacrylate with

Di- and Trimethacrylates. Journal of Macromolecular Science: Part A - Chemistry, 1974.

8(5): p. 965-980.

210. Stansbury, J.W., Dimethacrylate network formation and polymer property evolution as

determined by the selection of monomers and curing conditions. Dental Materials, 2012.

28(1): p. 13-22.

211. Allen, C., et al., The Radiation-Induced Copolymerization of Methyl Methacrylate with

Di-and Trimethacrylates. Journal of Macromolecular Science—Chemistry, 1974. 8(5): p.

965-980.

212. Bonino, C.A., et al., Real-time in situ rheology of alginate hydrogel photocrosslinking.

Soft Matter, 2011. 7(24): p. 11510-11517.

213. Bowman, C.N. and C.J. Kloxin, Toward an enhanced understanding and implementation

of photopolymerization reactions. Aiche Journal, 2008. 54(11): p. 2775-2795.

214. Schulz, V.G.V. and G. Harborth, Über den mechanismus des explosiven

polymerisationsverlaufes des methacrylsäuremethylesters1. Die Makromolekulare

Chemie, 1947. 1(1): p. 106-139.

215. Cao, X., H. Cummins, and J. Morris, Structural and rheological evolution of silica

nanoparticle gels. Soft Matter, 2010. 6(21): p. 5425-5433.

174

216. Mann, H., Electrical properties of thin polymer films. Part I. Thickness 500–2500 Å.

Journal of Applied Physics, 1964. 35(7): p. 2173-2179.

217. Chynoweth, A.G., Progress in Semiconductors, ed. A.F. Gibson. Vol. 4. 1960, New

York: John Wiley & Sons, Inc.

218. Holm, R. and W. Meissner, Einige Kontaktwiderstandsmessungen bei tiefen

Temperaturen. Zeitschrift für Physik, 1933. 86(11): p. 787-791.

219. Holm, R. and W. Meissner, Kontaktwiderstand zwischen supraleitern und

nichtsupraleitern. Z. Physik, 1932. 74: p. 715.

220. Mott, N.F. and R.W. Gurney, Electronic Processes in Ionic Crystals 1940, London:

Oxford University Press.

221. Murgatroyd, P., Theory of space-charge-limited current enhanced by Frenkel effect.

Journal of Physics D: Applied Physics, 1970. 3(2): p. 151.

222. Lampert, M.A., Simplified theory of space-charge-limited currents in an insulator with

traps. Physical Review, 1956. 103(6): p. 1648.

223. Emtage, P. and W. Tantraporn, Schottky emission through thin insulating films. Physical

Review Letters, 1962. 8(7): p. 267.

224. Gregor, L., Electrical conductivity of polydivinylbenzene films. Thin Solid Films, 1968.

2(3): p. 235-246.

225. Yeargan, J. and H. Taylor, The Poole‐Frenkel effect with compensation present. Journal

of Applied Physics, 1968. 39(12): p. 5600-5604.

226. Simmons, J.G., Poole-Frenkel effect and Schottky effect in metal-insulator-metal systems.

Physical Review, 1967. 155(3): p. 657.

227. Simmons, J., Conduction in thin dielectric films. Journal of Physics D: Applied Physics,

1971. 4(5): p. 613.

228. Von Hippel, A.R., Dielectric materials and applications. 1954: Artech House.

175

229. Gregor, L., Polymer dielectric films. IBM Journal of Research and Development, 1968.

12(2): p. 140-162.

230. Gazso, J., Electrical behaviour of thin layer Au-polyethylene-Al sandwiches. Thin Solid

Films, 1974. 21(1): p. 43-49.

231. Karanja, P. and R. Nath, Electrical conduction in biaxially-oriented polypropylene.

Journal of electrostatics, 1993. 31(1): p. 51-63.

232. Christy, R., Electrical properties of thin polymer films. Part II. Thickness 50–150 Å.

Journal of Applied Physics, 1964. 35(7): p. 2179-2184.

233. Tyczkowski, J., M. Zieliński, and M. Kryszewski, On the mechanism of electrical

conduction in glow discharge polysilazane films. Thin Solid Films, 1978. 55(2): p. 253-

259.

234. Miyoshi, Y. and K.i. Chino, Electrical properties of polyethylene single crystals.

Japanese Journal of Applied Physics, 1967. 6(2): p. 181.

235. Christy, R.W., Formation of Thin Polymer Films by Electron Bombardment. Journal of

Applied Physics, 1960. 31(9): p. 1680-1683.

236. Phadke, S.D., K. Sathianandan, and R. Karekar, Electrical conduction in polyferrocene

thin films. Thin Solid Films, 1978. 51(1): p. L9-L11.

237. Vodenicharova, M. and C. Vodenicharov, Electrode‐and bulk‐limited currents in thin

films of aliphatic unsaturated poly (1, 3, 4)‐oxadiazoles. physica status solidi (a), 1977.

41(2): p. 487-491.

176

APPENDIX

Fig. S1. FT-IR spectrum of PCE 3:1-19.

Fig. S2. 1H NMR spectrum of PCE 3:1-19.

177

a b c Copolymer code P N n Conformation Rac (nm) msat (mg/g) PCE 1:1-19 19.65 2 25 SBW 2.73 1.094 PCE 2:1-19 19.65 3 16.7 FBW 2.62 0.969 PCE 3:1-19 19.65 4 12.5 FBW 2.55 0.889 PCE 4:1-19 19.65 5 10 FBW 2.49 0.831 PCE 5:1-19 19.65 6 8.3 FBW 2.45 0.787 PCE 6:1-19 19.65 7 7.1 FBW 2.41 0.751 PCE 6:1-5 4.87 7 7.1 DC 0.91 0.653 PCE 6:1-9 9.43 7 7.1 FBW 1.44 0.698 PCE 6:1-43 43.52 7 7.1 FBW 4.21 0.813 a Theoretical polymer conformation in solution according to Gay and Raphaël diagram. b Calculated radius of theoretical adsorbed spherical core using Eq. 5.3. c Calculated amounts of the PCEs adsorbed at saturation plateau using Eq. 5.4.

178

PERMISSIONS

179

180

181

182

183

184

185

186

187

188

189

190

191