About morphology of grafted ethylene-propylene(-diene) copolymers-based latexes : preparation, structure and properties Tillier, D.L.

Transcription

1 About morphology of grafted ethylene-propylene(-diene) copolymers-based latexes : preparation, structure and properties Tillier, D.L. DOI: /IR Published: 01/01/2005 Document Version Publisher s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication: A submitted manuscript is the author's version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. The final author version and the galley proof are versions of the publication after peer review. The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication Citation for published version (APA): Tillier, D. L. (2005). About morphology of grafted ethylene-propylene(-diene) copolymers-based latexes : preparation, structure and properties Eindhoven: Technische Universiteit Eindhoven DOI: /IR General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Users may download and print one copy of any publication from the public portal for the purpose of private study or research. You may not further distribute the material or use it for any profit-making activity or commercial gain You may freely distribute the URL identifying the publication in the public portal? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Download date: 04. Dec. 2017

2 About Morphology of Grafted Ethylene-Propylene(-Diene) Copolymers-Based Latexes - Preparation, Structure and Properties - Delphine L. Tillier

3 CIP-DATA LIBRARY TECHNISCHE UNIVERSITEIT EINDHOVEN Tillier, Delphine L. About morphology of grafted ethylene-propylene(-diene) copolymers-based latexes : preparation, structure and properties / by Delphine L. Tillier. Eindhoven : Technische Universiteit Eindhoven, Proefschrift. ISBN NUR 913 Trefwoorden: emulsiepolymerisatie ; latices / polymeermofologie ; kernschaaldeeltjes / EPDM rubber / deeltjesgrootteverdeling / vernetten / slagbestendige materialen / slagsterkte modificeerders / polymethylmethacrylate ; PMMA Subject headings: emulsion polymerization ; latexes / polymer morphology ; core-shell particles / EPDM rubber / particle size distribution / crosslinking / impact-resistant materials / impact modifiers / poly(methyl methacrylate) ; PMMA 2005, Delphine L. Tillier Printed by PrintService Ipskamp, The Netherlands. Cover designed by Delphine Tillier, Soazig Périn and Jan-Willem Luiten, JWL Producties This research was financially supported by the Foundation of Emulsion Polymerization (SEP) and the European Graduate School (EGS). An electronic copy of this thesis is available from the site of the Eindhoven University Library in PDF format (

4 About Morphology of Grafted Ethylene-Propylene(-Diene) Copolymers-Based Latexes - Preparation, Structure and Properties - PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de Rector Magnificus, prof.dr. R.A. van Santen, voor een commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op donderdag 17 februari 2005 om uur door Delphine Lucienne Tillier geboren te Domont, Frankrijk

5 Dit proefschrift is goedgekeurd door de promotoren: prof.dr. C.E. Koning en prof.dr. A.M. van Herk Copromotor: dr. J. Meuldijk

6 A mes parents

7

8 Table of Contents 1. General introduction Existing impact modifiers Preparation of impact modifiers Objective and outline of this thesis 4 References 7 2. Principles of emulsion polymerization Introduction Overall description of emulsion polymerization Emulsion polymerization processes Kinetics and mechanisms Seeded emulsion polymerization General description Production of core-shell particles by seeded emulsion polymerization Conclusions 19 References Production of colloidally stable latexes from low molecular weight ethylene-propylene(-diene) copolymers Introduction Experimental section Chemicals Preparation of an artificial latex from a low molecular weight EP(D)M Characterization of polymers Characterization of latexes Particle size distribution Results and discussion Optimization of the surfactant system 32

9 3.3.2 Influence of polymer viscosity on particle size distribution Conclusions 40 References About crosslinking of EP(D)M-based latexes General introduction General mechanism of crosslinking Aim of this chapter Experimental section Chemicals Preparation of a seed latex Chemically-initiated crosslinking Crosslinking initiated by a pulsed electron-beam Characterization of the latexes Results and discussion Chemically-initiated crosslinking Crosslinking initiated by a pulsed electron-beam Conclusions 77 References About morphology in EP(D)M-based latexes Introduction Experimental section Chemicals Preparation of the seed latex Seeded emulsion polymerization of MMA onto EP(D)M Characterization Results and discussion Grafting mechanism Efficiency of the grafting reaction Morphology of the EP(D)M-g-PMMA particles 101

10 5.4 Conclusions 112 References Toughening effect of EP(D)M-PMMA core-shell structures in a brittle PMMA matrix Introduction Experimental section Chemicals Synthesis of the composite latex particles Synthesis of a PMMA homopolymer latex Blending and specimen preparation Characterization Results and discussion Molecular characterization of PMMA Incorporation of the rubbery particles into a PMMA matrix Mechanical properties of PMMA/rubber blends Fracture mechanism Conclusions 127 References Highlights and technological assessment Highlights Technological assessment 133 References 135 Summary 137 Samenvatting 141 Acknowledgements 145 Curriculum vitae 149

11

12 General introduction Abstract The aim of this chapter is to introduce the background of the project. The developments in the field of impact modification are presented, followed by an overview of the research exposed in the next chapters.

13 2 Chapter Existing impact modifiers For many applications, high performance materials, e.g. coatings 1,2 and engineering plastics 3,4, require high impact strength. The paint of a car, for instance, should withstand being hit by a piece of gravel without film rupture. Thermoplastic materials, such as polycarbonate (PC), have a tendency to undergo brittle failure under environmental stress cracking conditions, e.g. by the presence of a sharp notch. Therefore such materials need some adjustments to extend their applications under high impact conditions, for example in the automotive industry. Improvement of impact strength may be achieved with tougheners, which consist of an elastomeric part, providing impact resistance, and a rigid part, providing good adhesion with a polymer matrix. Core-shell tougheners with a crosslinked core, as opposed to linear tougheners, have a fixed morphology and are the preferred impact modifiers especially in injection molded engineering plastics and in coatings. The first attempts to produce rubber-toughened materials were completed in the late 1940 s, leading to the high-impact grade of polystyrene (HIPS) 5. The wide range of high impact resistant polymers includes ABS (acrylonitrile/butadiene/styrene) and MBS (methyl methacrylate/butadiene/styrene). An early form of ABS consisted of a physical mixture of styrene/acrylonitrile (SAN) and acrylonitrile/butadiene copolymers. In the 1950 s, ABS was produced via emulsion polymerization 6, resulting in SAN grafted onto spherical polybutadiene latex particles. Those grafts greatly improved the compatibility of the elastomeric particles and the SAN matrix. Since then, the diversity and ease of processing of these copolymers turned ABS into one of the most popular engineering polymers. ABS polymers now find applications in coatings, adhesives, or impact modifiers. Unmodified, ABS copolymers are also used in automotive applications or electrical equipments, such as vacuum cleaners, phones or computers. An important application of both ABS and MBS resides in the impact modification of poly(vinyl chloride) (PVC) and polycarbonate (PC). MBS and ABS modifiers are

14 General introduction 3 mostly used in transparent applications where weather resistance is not required. Moreover, MBS and ABS modifiers are particularly efficient for low temperature impact performances. Although ABS and MBS have received a considerable industrial interest, the presence of many unsaturations in the main chain of polybutadiene results in a high susceptibility to thermal oxidation and UV degradation. This (partial) degradation leads to more brittle materials. Therefore, all-acrylic core-shell impact modifiers have been developed 7,8 for polymer outdoor applications, including window profiles, fences, gutters, or varnishes, adhesives and floor coverings. The low glass transition temperature acrylic core, usually consisting of a crosslinked poly(butyl acrylate), provides an outstanding low temperature impact performance similar to MBS, combined with the weatherability of acrylics. The shell, most commonly composed of poly(methyl methacrylate) (PMMA), sometimes copolymerized with functional monomers, has two major roles: - A rigid PMMA phase encapsulates the elastomeric core in order to facilitate product recovery and handling. Then, excellent powder properties, e.g. ability to flow, may be achieved even for high rubber contents. - Moreover, such a PMMA shell exhibits a fairly good compatibility with many polymers, e.g. PVC or polycarbonate, and therefore ensures the dispersion as well as adhesion of the primary impact-modifying particles into the polymer matrix. In order to circumvent the limited temperature stability of all-acrylic core-shell structures, a polysiloxane rubber may be used as core material 9,10. Indeed, polysiloxane rubbers exhibit an excellent stability towards UV irradiation and do not undergo oxidative degradation. However, such polysiloxane rubber core-shell structures are expensive and may suffer from hydrolytic instability.

15 4 Chapter Preparation of impact modifiers The size, volume fraction and morphology of rubber particles dispersed in a polymer matrix are of considerable importance. Emulsion polymerization 11 is the method of choice for the preparation of well-defined particles. The molecular weight of the obtained polymers may be adjusted to the desired value by varying the initiator concentration or with the help of chain transfer agents. Moreover, excessive heat evolvement and viscosity problems, as encountered during polymerizations in homogeneous media, i.e. bulk or solution polymerization, may be overcome by the use of water as dispersing medium. Desired core-shell particles are usually prepared using a two-stage emulsion polymerization 5. The first stage of the process consists of the synthesis of rubbery seed particles of controlled diameter, typically of the order of 0.1 to 0.5 m. The second stage is a seeded emulsion polymerization in which the second monomer is polymerized in the form of a shell onto the seed rubber latex. In the case of ABS, styrene and acrylonitrile are copolymerized onto polybutadiene seed particles, leading to SAN grafts. If desired, e.g. when high shear is required for processing the core-shell containing material, the rubbery core may be crosslinked to prevent flowing. Depending on the applications, the obtained dispersion may then be mixed with another latex, such as a paint for instance, or the particles may be harvested by coagulation or freeze-drying, and then dispersed at specific volume fractions in the material to be toughened. 1.3 Objective and outline of this thesis The overview of the commercially available impact modifiers demonstrated that a novel type of thougheners may be required for many applications, demanding high impact properties, durability and good weatherability.

16 General introduction 5 Therefore, the aim of the work described in this thesis is to investigate the preparation of a new generation of core-shell particles. A fully saturated rubber core, typically consisting of ethylene-propylene copolymer (EPM) or ethylene-propylene-diene copolymer (EPDM), will overcome the stability problems encountered with polybutadiene-based tougheners, since these saturated molecules exhibit an excellent UV, oxidative, thermal and hydrolytic stability. A glassy polyacrylate shell, being typically poly(methyl methacrylate) (PMMA), will provide a free flowing character to the obtained structures and will facilitate their incorporation into and compatibility with various polymer matrixes. Chapter 2 presents a brief theoretical overview of emulsion polymerization, to acquire a proper insight into the numerous parameters involved in the preparation of the targeted core-shell structures. Chapter 3 deals with the preparation of the rubber seed latex. The solutionemulsification technique was used to produce artificial latexes based on low molecular weight EPM and EPDM materials, without addition of organic solvent. For engineering plastics applications, crosslinking of the rubber core may be required in order to avoid flowing under the high demanding conditions of processing, e.g. injection molding. Several crosslinking methods are described in Chapter 4, including crosslinking of the latex using an electron-beam irradiation to produce the necessary radical species. Chapter 5 is dedicated to the grafting of the glassy shell onto the rubber core, by seeded emulsion polymerization of MMA onto the EP(D)M particles. The various obtained morphologies are discussed, taking into account the kinetics and the thermodynamics of the system. Chapter 6 emphasizes the effect of the obtained rubber-containing particles on the impact resistance or toughness of a brittle PMMA matrix. In addition, the distribution of

17 6 Chapter 1 the core-shell structures in the PMMA matrix was observed by Scanning Electron Microscopy (SEM) as well as Transmission Electron Microscopy (TEM). Part of the contents of this thesis have already been published 12 or submitted for publication 13,14.

18 General introduction 7 References 1. Mizoguchi, M.; Fuseya, Y.; Fujita, Y.; Ishino, Y.; Seki, M.; Miyawaki, T.; Mitsui Toatsu Chemicals Inc., Eur. Pat. Appl , Pascault, J.; Valette, L.; Magny, B.; Cray Valley S.A., PCT Int. Appl , Tanrattanakul, V.; Baer, E.; Hiltner, A.; Hu, R.; Dimonie, V. L.; El Aasser, M. S.; Sperling, L. H.; Mylonakis, S. G. J. Appl. Polym. Sci. 1996, 62, El-Aasser, M. S.; Segall, I.; Dimonie, V. L. Macromol. Symp. 1996, 101, Lovell, P. A.; Pierre, D. in "Emulsion Polymerization and Emulsion Polymers"; Lovell, P. A. and El-Aasser, M. S. Eds., Chichester: Wiley, 1997, Chapter Aerdts, A. PhD Thesis, University of Eindhoven, Guo, T. Y.; Tang, G. L.; Hao, G. J.; Wang, S. F.; Song, M. D.; Zhang, B. H. Polym. Advan. Technol. 2003, 14, Chung, J. S.; Choi, K. R.; Wu, J. P.; Han, C. S.; Lee, C. H. Korea Polym. J. 2001, 9, He, W. D.; Cao, C. T.; Pan, C. Y. J. Appl. Polym. Sci. 1996, 61, He, W. D.; Cao, C. T.; Pan, C. Y. Polym. Int. 1996, 39, El-Aasser, M. S.; Sudol, E. D. in "Emulsion Polymerization and Emulsion Polymers"; Lovell, P. A and El-Aasser, M. S. Eds., Chichester: Wiley, 1997, Chapter Tillier, D. L.; Meuldijk, J.; Koning, C. E. Polymer 2003, 44, Tillier, D. L.; Meuldijk, J.; Magusin, P. C. M. M.; van Herk, A. M.; Koning, C. E. Submitted for publication. 14. Tillier, D. L.; Meuldijk, J.; Höhne, G. W. H.; Frederik, P. M.; Regev, O.; Koning, C. E. Submitted for publication.

19

20 Principles of emulsion polymerization Abstract The so-called emulsion polymerization technique has been extensively used in this research. Therefore the present chapter provides an overview of the principles of emulsion polymerization. A fair understanding of the main processes, mechanisms and kinetics involved during the reaction will help in optimizing the conditions of our investigation, i.e. the seeded emulsion polymerization of methyl methacrylate onto artificial ethylene-propylene(-diene) copolymer latexes.

21 10 Chapter Introduction Emulsion polymerization has received a considerable industrial interest for the production of synthetic latexes since its first developments in the early 1930 s 1. Initially introduced for the manufacturing of synthetic rubber, emulsion polymerization has become an economically important process applied to a wide variety of monomers to produce elastomers, thermoplastics, and numerous specialty polymers, used for instance in drug delivery systems or chromatographic separations. Latexes may also find applications in adhesives, packaging, wall and ceiling coverings, flooring, consumer glues, and film laminates. An emulsion polymerization is a free-radical-initiated polymerization in which a monomer or a mixture of monomers is dispersed in water with the help of a surfactant. The reaction is usually initiated by a water-soluble initiator and leads to the formation of a colloidally stable dispersion of submicron polymer particles in an aqueous medium, i.e. a latex. The polymerization occurs in the latex particles that act as minireactors open for entry and exit of radicals. This compartmentalization allows the production of high molecular weight polymers at a polymerization rate higher than in polymerizations performed in homogeneous media, e.g. in bulk or solution. Moreover, the molecular weight of the polymer may easily be altered at will by the addition of chain-transfer agents, leading to more controlled properties of the final product, e.g. regarding mechanical strength or film-forming temperature. The expansion of the market for emulsion polymers may be explained by their good environmental profile. As water-based materials, their use results in lower emissions of volatile organic compounds during processing and/or application. The use of water as an inert and harmless continuous phase, contributes to maintain a relatively low viscosity of the end-products, and leads to a better controlled heat transfer than in homogeneous

22 Principles of emulsion polymerization 11 systems. Furthermore, an emulsion polymerization can often be carried out up to high conversion, thereby minimizing the amount of residual monomer. Unfortunately, a latex is a complex system containing various low molar mass compounds, e.g. surfactant and initiator fragments, that can not be removed easily. Therefore, the presence of these substances may affect the properties of the end-product, such as undesired color upon molding for instance. Finally, when the polymer needs to be recovered for further processing, the removal of the aqueous phase may induce additional expenses. The scope of this chapter is to introduce the characteristics of emulsion polymerization and latexes. After a brief description of the various existing processes, the mechanisms and kinetics involved in the fundamental reaction steps will be emphasized. A thorough understanding of these fundamental issues is a prerequisite for the optimization of the conditions of the seeded emulsion polymerization of methyl methacrylate (MMA) onto artificial ethylene-propylene(-diene) copolymer (EP(D)M) latexes, as described in Chapter 5 of this dissertation. 2.2 Overall description of emulsion polymerization Emulsion polymerization processes An emulsion polymerization may be performed using three different common processes 2, i.e. batch, semi-continuous (or semi-batch), and continuous. In a batch emulsion polymerization, all the ingredients, i.e. water, surfactant, monomer, and initiator, are charged into the reaction vessel before the start of the reaction. In a semi-continuous process, the ingredients are partially charged into the vessel, the rest being fed in a controlled way during the polymerization. As compared to batch operation, the semi-continuous method leads to an improved control of numerous

23 12 Chapter 2 variables, including polymer composition, particle concentration, and particle size distribution. Finally, in a continuous process, the ingredients are continuously fed to a reactor while the obtained latex product is simultaneously removed. Advantages of continuous methods include high production rates per unit equipment volume and steady heat removal Kinetics and mechanisms Traditionally, a batch emulsion polymerization can be subdivided into three timeseparated intervals 4,5, as depicted in Figure 2.1. M o M o o o o R o o o o o o o o o o o M o o o o M o o o o o o o o o o o o o o o o o o o o R o o o o o R o o o o o o o o o o o o o o o o Interval I Interval II Interval III Figure 2.1: Schematic representation of the three intervals of an emulsion polymerization, where M represents the monomer, R a radical generated by the initiator, and o a surfactant molecule. The reaction mixture initially consists of monomer droplets, stabilized by surfactant, and monomer-swollen surfactant micelles, dispersed in a continuous aqueous phase. The monomer is also dissolved in small amounts in the water phase.

24 Principles of emulsion polymerization 13 The reaction starts in Interval I, also referred to as nucleation stage. Radicals generated from the dissociation of a usually water-soluble initiator, e.g. sodium persulfate, react with the monomer present in the aqueous phase. This results in the formation of monomeric radicals, which will then undergo the different reactions related to conventional free-radical polymerizations, i.e. propagation with monomer, termination and transfer. Provided that termination and transfer do not occur for a given oligomeric radical, this species may grow, until a critical chain length, z, is reached 6. Above this critical value of z, typically in the order of 2 to 5 monomer units for styrene and methyl methacrylate, respectively 6, the oligoradical becomes surface active and enters a monomer-swollen micelle, thereby creating a particle. This mechanism of particle formation, suggested by Harkins 4, is known as micellar nucleation. The entered oligoradical then propagates in the micelle, until a second radical enters. This will instantaneously cause termination in the small latex particle. When oligoradicals grow until a critical chain length, j crit, higher than z, typically in the order of 4 to 11 monomer units for styrene and methyl methacrylate, respectively 6, they become insoluble in the aqueous phase and precipitate to form primary particles. Note that for more hydrophilic monomers, such as vinyl acetate, the critical chain length may consist of 18 units 6. The obtained precursor particles adsorb surfactant molecules to increase their colloidal stability and absorb monomer to propagate. This mechanism of particle formation is referred to as homogeneous nucleation However, when the stabilization of the new crop of particles is not sufficient, they tend to coagulate, giving rise to the so-called homogeneous-coagulative nucleation 1. Finally, droplet nucleation occurs when radicals, generated in the aqueous phase, enter monomer droplets as single radicals or oligoradicals and propagate to form polymer particles. This mechanism is mainly predominant in miniemulsion and microemulsion polymerizations 2. However, the chance that droplet nucleation occurs is low in view of much larger specific surface area of the micelles with respect to the monomer droplets.

25 14 Chapter 2 Although the three above-mentioned mechanisms of particle formation may occur simultaneously, their contribution mostly depends on the surfactant concentration and the monomer solubility in water. Interval I ends when all micelles have disappeared, either by nucleation or by dissolution to stabilize growing latex particles. The increase of the number of polymerization loci, and so of the polymerization rate, is the main characteristic of the nucleation stage, as depicted in Figure Rp (mol/m 3 water.s) Nparticles (1/m3 water ) Time (s) Figure : Rate of polymerization and particle concentration as a function of time in case of complete colloidal stability in Interval II. Possible gel effects in Interval III are not shown. After nucleation and throughout Interval II, a constant number of polymer particles grow at the expense of monomer droplets. The monomer consumption due to polymerization in particles is compensated by the transport of monomer from droplets to growing particles. As the time constant for monomer transport is much lower than that for reaction, the concentration of monomer within the polymer particles remains constant at its saturation value. As a result of the constant number of particles and the constant monomer concentration within these polymerization loci, the overall polymerization rate, R p, is constant and given by:

26 Principles of emulsion polymerization 15 R p = k p n C N Av m N p (2.1) In Equation 2.1, k p represents the propagation rate coefficient, n the average number of growing chains per particle, C m the overall monomer concentration in the particles, N p the number of latex particles per unit volume of aqueous phase, and Avogadro s number. N Av Once the monomer droplets have disappeared, the polymerization proceeds further on, throughout Interval III, until all the monomer dissolved in the aqueous phase or present in the growing particles is depleted. As a consequence, the overall monomer concentration within the particles decreases, resulting in a decreasing polymerization rate, provided that no gel effect occurs 12,13. Indeed, as more and more monomer is depleted, the viscosity inside the particles may raise. The termination rate may slow down and consequently, the number of radicals per particle may increase. This may lead to a sudden increase of the polymerization rate, also referred to as Trommsdorff-Norrish gel effect. In the case of the emulsion polymerization of styrene, this auto-acceleration is clearly represented by a second abrupt increase of the heat production rate, as depicted in Figure A 20 Q r [W] t [s] Figure 2.3: Rate of heat production during the emulsion polymerization of styrene.

27 16 Chapter Seeded emulsion polymerization General description As explained above, many nucleation mechanisms are involved in the Interval I of an emulsion polymerization. Therefore, in order to circumvent the complexity of this first stage, a preformed latex may be employed. The so-called seeded emulsion polymerization then starts in Interval II or III, depending on the presence or absence of monomer droplets, respectively, with the growth of the already existing polymer particles. Hence, a two-stage process allows a good control over particle size and particle size distribution. The type of monomer addition is an important parameter for the reaction. Under monomer-flooded conditions, the polymerization proceeds as in a batch reaction with saturation swelling. On the other hand, under monomer-starved conditions, the concentration of monomer in the particles is much lower than the saturation concentration and is mainly controlled by the addition rate of monomer Production of core-shell particles by seeded emulsion polymerization Traditionally, seeded emulsion polymerization is used to produce composite latex particles. Core-shell morphology is one of the most studied structures. The use of an oil-soluble initiator will result in grafting of the new growing polymer onto the seed polymer backbone. However, the presence of monomer droplets or surfactant micelles may lead to undesired phenomena, e.g. secondary nucleation and polymerization in this second generation of particles. Secondary nucleation gives rise to a broad particle size distribution and tremendously reduces grafting efficiency. Therefore a step forward would be made if this formation of new particles could be suppressed.

28 Principles of emulsion polymerization 17 The aim of the research described in this thesis is to graft MMA onto artificial EP(D)M latexes, as depicted in Figure 2.4. Our challenge consists of optimizing the reaction conditions in order to reduce the formation of a second crop of PMMA particles by secondary nucleation, and therefore to increase the grafting efficiency. EP(D)M latex particle Monomer EP(D)M latex particle grafted with PMMA + n CH2 CH3 C C O O CH3 Surfactant molecule Figure 2.4: Schematic drawing of the reaction studied. As suggested by Ferguson et al. 21 when they studied the formation of polystyrene/poly(vinyl acetate) core-shell particles, no secondary nucleation will occur when Condition 2.2 is obeyed: (overall rate of entry into existing particles) + (overall rate of termination) (rate of entry into micelles) + (rate of forming j crit -mers) (2.2) A starved-feed of monomer combined with a high initiator concentration should result in a lower monomer concentration in the aqueous phase. Therefore, a less significant amount of monomer should be available for propagation in the water phase. However, a low monomer concentration in water may also reduce the formation of z-mers and j crit - mers, resulting in a decreasing rate of entry into existing particles. As a consequence, it is essential to study other parameters, as described below, in order to minimize secondary nucleation, and so to optimize grafting efficiency.

29 18 Chapter 2 It has been demonstrated 22 that the use of an oil-soluble initiator also reduces secondary nucleation, since most of the reactions occur inside the hydrophobic seed particles. However, organic phase initiation may lead to a loss of morphology control, due to the homogeneous distribution of initiator radicals in the particles. If a redox initiation system is employed, radicals are generated at or near the particle surface. As a result, only a small fraction of these radicals is present in the aqueous phase, leading to limited homogeneous nucleation. Moreover, the polymerization locus, being near or at the particle surface, facilitates the formation of the preferred core-shell morphology. Working above the critical micelle concentration (CMC) of the surfactant will enhance micellar nucleation, resulting in a bimodal particle size distribution. As a consequence, a controlled addition of surfactant during the polymerization may help maintaining the surface coverage of the particles with emulsifier as high as possible without exceeding the CMC in the continuous phase. Hence, the growing particles 23 would remain colloidally stable. Finally, increasing the solids content, while keeping the same total reaction volume, may also contribute to decrease the secondary nucleation per unit volume of reaction mixture. Indeed, higher solids contents expand the particle surface per unit volume of reaction mixture. So a larger fraction of oligomeric radicals will enter the seed particles, resulting in a less significant fraction available for secondary nucleation. The effect of particle surface may be even more pronounced if smaller particles are used in combination with a high solids content. However, as reported by several groups 24,25, the viscosity and the amount of reactor fouling increase with solids content, and the rheology changes from Newtonian to pseudo-plastic behavior. Therefore, mixing quality and heat transfer are strongly hampered 26. In addition, transfer of a latex from a reactor to other parts of a plant may also be rendered difficult by a highly viscous fluid. Finally, as a large number of latex products are used in coatings applications 27,28, high viscosities, also leading to long drying times, are usually undesirable.

30 Principles of emulsion polymerization 19 As a consequence, secondary nucleation may be induced by deliberate addition of surfactant to obtain a bimodal particle size distribution and, hence, a lower viscosity of high solids latexes 29, Conclusions Emulsion polymerization is a very sophisticated polymerization technique, involving many complex features. Product properties, for coatings applications for instance, can be controlled by adjusting recipes and process operations. The choice of surfactant concentration, monomer feeding, as well as amount and type of initiator, becomes crucial when taking into account the occurrence of secondary nucleation. This will be further discussed in Chapter 5.

31 20 Chapter 2 References 1. Gilbert, R. G. Emulsion polymerization: a mechanistic approach; London: Academic Press, El-Aasser, M. S.; Sudol, E. D. in "Emulsion Polymerization and Emulsion Polymers"; Lovell, P. A and El-Aasser, M. S. Eds., Chichester: Wiley, 1997, Chapter Scholtens, C. A. PhD Thesis, University of Eindhoven, Harkins, W. J. Am. Chem. Soc. 1947, 69, Smith, W. V.; Ewart, R. H. J. Chem. Phys. 1948, 16, Maxwell, I. A.; Morrison, B. R.; Napper, D. H.; Gilbert, R. G. Macromolecules 1991, 24, Jacobi, B. Angew. Chem. 1952, 64, Priest, W. J. J. Phys. Chem. 1952, 56, Fitch, R. M. in "Polymer Colloids: A Comprehensive Introduction"; London: Academic Press, 1997, Chapter Hansen, F. K.; Ugelstad, J. J. Polym. Sci. Polym. Chem. Ed. 1978, 16, Meuldijk, J.; Kemmere, M. F.; de Lima, S.; Reynhout, X.; Drinkenburg, A.; German, A. L. Pol. React. Eng. 2003, 11, Norrish, R. G. W.; Smith, R. R. Nature (London) 1942, 150, Trommsdorff, E.; Kohle, H.; Lagally, P. Makromol. Chem. 1948, 1, Kemmere, M. F.; Meuldijk, J.; Drinkenburg, A. A. H.; German, A. L. Pol. React. Eng. 2000, 8, Merkel, M. P.; Dimonie, V. L.; El-Aasser, M. S.; Vanderhoff, J. W. J. Polym. Sci. Part A: Polym. Chem. 1987, 25, Berzinis, A. P.; Wills, W. L.; Rohm and Haas, Eur. Pat. Appl , Troy, E. J.; Rosado, A.; Rohm and Haas, Can. Pat. Appl , Cheng, S.; Chen, Y.; Chen, Z. J. Appl. Polym. Sci. 2002, 85, Ferguson, C. J.; Russell, G. T.; Gilbert, R. G. Polymer 2002, 43, 4557.

32 Principles of emulsion polymerization Landfester, K.; Rothe, R.; Antonietti, M. Macromolecules 2002, 35, Ferguson, C. J.; Russell, G. T.; Gilbert, R. G. Polymer 2002, 43, Sudol, E. D.; El-Aasser, M. S.; Vanderhoff, J. W. J. Polym. Sci. Part A: Polym. Chem. 1986, 24, Sudol, E. D.; El-Aasser, M. S.; Vanderhoff, J. W. J. Polym. Sci. Part A: Polym. Chem. 1986, 24, Mayer, M. J. J.; Meuldijk, J.; Thoenes, D. Chem. Eng. Sci. 1994, 49, Kemmere, M. F.; Meuldijk, J.; Drinkenburg, A. A. H.; German, A. L. Pol. React. Eng. 1998, 6, Meuldijk, J.; Van Den Boomen, F. H. A. M.; Kemmere, M. F.; Wijers, J. G. Entropie 1998, 34, Schneider, M.; Claverie, J.; Graillat, C.; McKenna, T. F. J. Appl. Polym. Sci. 2002, 84, Guyot, A.; Chu, F.; Schneider, M.; Graillat, C.; McKenna, T. F. Prog. Polym. Sci. 2002, 27, Chong, J. S.; Christiansen, E. B.; Baer, A. D. J. Appl. Polym. Sci. 1971, 15, Greenwood, R.; Luckham, P. F.; Gregory, T. J. Colloid Interf. Sci. 1997, 191, 11.

33

34 Production of colloidally stable latexes from low molecular weight ethylene-propylene(-diene) copolymers Abstract The solution-emulsification technique was used to produce artificial latexes based on low molecular weight EPM and EPDM materials. Conventional emulsification techniques as well as miniemulsification methods have been investigated. In both cases, a larger volume of polymer is reduced into smaller sub-units using the mechanical energy of comminution techniques, i.e. an Ultra-Turrax and a homogenizer operating at pressures ranging from 300 to 1100 bars and with corresponding shear rates of to s -1. The difference between conventional emulsification and miniemulsification resides in the stabilizing system. For the conventional emulsification method, an equimolar mixture of anionic (sodium dodecyl benzene sulfonate, SDBS) and nonionic (polyoxyethylene (100) stearyl ether, Brij 700) surfactants was found to be the optimal surfactant system. For the miniemulsification method, a combination of SDBS as surfactant and hexadecane or cetyl alcohol as costabilizer was the most suitable system. Both conventional emulsification and miniemulsification lead to latexes with monomodal particle size distributions and volume-average diameters ranging from 300 to 400 nm, determined with light scattering techniques. The low molecular weight elastomers, exhibiting viscosities lower than 2 Pa s at 20 C, were easily emulsified without addition of organic solvent. This chapter is based on: Tillier, D. L.; Meuldijk, J.; Koning, C. E. Polymer 2003, 44, 7883.

35 24 Chapter Introduction Increasing environmental concerns in the formulations of various coating compositions, such as paints or adhesives, has generated considerable interest in producing a wide variety of synthetic organic polymers in a latex form. The preparation of vinylic polymer latexes by emulsion polymerization techniques is well known. Typical products include polybutadiene, polystyrene, poly(styrene-acrylonitrile), and poly(methyl methacrylate) latexes. However, synthetic limitations, such as the need of water-sensitive catalysts, prevent the production of many polymers in aqueous dispersions. This is the case, for instance, for polyolefins such as polyethylene, polypropylene, and ethylene-propylene copolymers (EPM), as well as copolymers of one or more olefins with other monomers (e.g. non-conjugated dienes). An example of the latter is ethylene-propylene-diene monomer (EPDM), which is the preferred elastomer for application as impact modifier for coatings and engineering plastics. A dispersion of EPDM-based polymers in water is obtained through the preparation of an artificial latex, i.e. a preformed polymer colloidally dispersed in an aqueous medium 1. Among all the processes developed for the production of artificial latexes, two techniques are mostly used: the phase-inversion technique and the solutionemulsification technique. The phase-inversion technique 2 has been employed by Yang et al. 3 to prepare waterborne dispersions of epoxy resin. By dispersing, under stirring, increasing amounts of water into the emulsifier (or surface-active agent) containing polymer, at a temperature ranging from 60 to 80 C, a water-in-oil dispersion is initially formed. As more aqueous phase is incorporated, a phase inversion occurrence becomes increasingly likely, thereby creating the desired dispersion of the polymer in an aqueous medium. However, depending on the emulsifier concentration and on the temperature, a complex water-in-oil-in-water structure can be obtained by incomplete phase inversion. Therefore, a better controlled technique, i.e. the direct solution-emulsification technique, reported by Burton and O Farrell 4, can be used. It consists of the dissolution of the polymer in a volatile solvent to form a solution or cement. The resulting solution or

36 Production of colloidally stable latexes 25 cement is then emulsified in water with one or more emulsifiers to form a crude emulsion. The droplet size in the crude emulsion is then reduced to submicron size by application of a high shear rate (namely the homogenizing step). The final step of the procedure is the removal of the solvent from the emulsion, by reduced pressure or steam distillation. When producing artificial latexes, numerous parameters have to be taken into account. One of them is the emulsifier, which is critical for the formation of an emulsion that should remain stable at the relatively high temperature and under the mechanical forces of both the homogenizing step and the stripping operation. Colloidal particles are permanently subjected to the influence of the van der Waals- London attractive forces. Hence, in order to maintain colloidal stability and outweigh those attractive forces, an electrostatic and/or steric repulsion has to be introduced. Various types of surfactants have been employed for the synthesis and stabilization of polymer latexes 5. Electrostatic stability is provided by ionic surfactants as described in the Derjaguin-Landau-Verwey-Overbeek (DLVO) theory 6,7. The coverage of the particles by charged species creates an electric double layer, leading to electrostatic repulsion. Steric stabilization 8-10 is obtained with nonionic or polymeric surfactants in many cases based on poly(ethylene oxide) as the hydrophilic part. The robustness of the steric stabilization method can be exemplified by its relative insensitivity to high concentrations of electrolytes and its tolerance to temperature effects such as freezethawing 11. Note that, for nonionic surfactants, the critical coagulation temperature should not be exceeded. The combination of both stabilization mechanisms, by using anionic and nonionic surfactants, offers remarkable results. Colombié et al. 12 used for instance a mixture of sodium lauryl sulfate and Triton X-405 (octylphenoxy poly(ethylene oxide)) to stabilize submicron polystyrene particles dispersed in water. Another approach relies on the use of electrosteric surfactants, which consist of an ionic charge chemically anchored to the end of the hydrophilic nonionic moiety. An example of a surfactant providing both electrostatic and steric stabilization is the commercially available series Avanel S, C m H 2m+1 O(CH 2 CH 2 O) n SO 3 Na (with 12 m 15 and 3 n

37 26 Chapter 3 15) (PPG Industries) used by Sung and Piirma 13 in the emulsion polymerization of styrene. The term miniemulsion describes a submicron oil-in-water dispersion with colloidal stability ranging from hours to months 14. A practical application of the procedure to produce miniemulsions, i.e. miniemulsification i, is found in the preparation of artificial latexes from polymer solutions, as reported by El-Aasser et al. 15 Indeed, the miniemulsification method may be employed for the emulsification of non watermiscible polymer solutions in aqueous media containing the proper emulsifier system. After emulsification, the solvent can be removed by steam distillation under vacuum. However, as opposed to a conventional emulsion, a miniemulsion is stabilized by a combination of an efficient ionic surfactant and a costabilizer, i.e. a highly waterinsoluble low-molecular weight compound. This approach to emulsion stability issues was first reported by Higushi and Misra 16. The main reason for the destabilization of an emulsion is related to chemical potential differences. Indeed, the chemical potential of the monomer in small droplets is higher than that in large droplets or plane surfaces. Consequently, monomer diffuses from small to large droplets leading to larger droplets and partial emulsion degradation. This phenomenon is referred to as Ostwald ripening. Higushi and Misra 16 considered that both the rate of growth of large particles and the rate of dissolution of small particles were diffusion controlled. Therefore, the addition of a small amount of a water-insoluble compound would retard the emulsion degradation due to its slow diffusion rate, and the monomer distribution over the particles would not change during a long elapse of time. Typical costabilizers include long chain alkanes and fatty alcohols, more specifically hexadecane and cetyl alcohol. For example, Hansen and Ugelstad 17 used styrene as monomer and hexadecane as costabilizer, while Rodriguez 18 used styrene and methyl methacrylate with both hexadecane and cetyl alcohol. i The term miniemulsification does not exist as such in the miniemulsion terminology. However in the remainder of this chapter, it will refer, as suggested by El-Aasser et al. 15, to the process of dispersing a polymer in water, stabilized by a combination of an anionic surfactant and a costabilizer.

38 Production of colloidally stable latexes 27 Finally, in addition to the surfactant, the viscosity of the dispersed phase represents another key parameter for the emulsification of a polymer. According to Burton and O Farrell 4, the cement must exhibit a viscosity lower than 10 Pa s at 24 C to be properly dispersed, leading to particles with an average diameter below one micrometer. The viscosity of the cement is of course dependent on the molecular weight and molecular structure of the polymer, as well as on the polymer concentration in the solution. The direct solution-emulsification technique has been extensively used to produce high molecular weight ethylene-propylene-diene copolymers in a latex form. Numberaverage molecular weights of the polymers ranged from to g mol -1 and their viscosities were higher than 10 Pa s at room temperature. Therefore, in order to obtain low viscosity cements, the polymers had to be diluted with non water-miscible solvents, the organic phase containing up to 25 weight percent, preferably about 5 to 20 weight percent solid. As a consequence, in order to minimize the amount of solvent that has to be removed, polymer concentrations have to be brought to their maximum within the limits set for proper emulsification. The solvent (or the solvent/water azeotrope) must exhibit a boiling point lower than the boiling point of water. Thus, solvents of choice include aromatic hydrocarbons (e.g. toluene), but also aliphatic hydrocarbons (e.g. pentane, heptane). However, for environmental, economic, and time-saving reasons, a step forward would be made with the preparation of solvent-free artificial polymer latexes. Thus, the aim of the present work is to develop a procedure for the production of such latexes with submicron EP(D)M particles. The choice of EP(D)M is particularly relevant for its use in environments where weatherability is an important property. Indeed, as a saturatedmain chain rubber, it exhibits good stability not only towards oxygen and UV irradiation, but also towards hydrolysis and high temperatures. The submicron size of the particles is a prerequisite for the formation of core-shell structures, used as impact modifiers for coatings and engineering plastics. The artificial EP(D)M latex is the starting material for crosslinking and grafting reactions leading to core-shell structures. Crosslinking of the EP(D)M core will preserve its spherical shape upon injection

39 28 Chapter 3 molding. Grafting of e.g. methyl methacrylate by seeded emulsion polymerization onto the rubber core should generate a glassy shell, compatible with several targeted polymer matrixes. Moreover, a glassy PMMA shell will provide a free-flowing character to the impact modifier, which is crucial for easy processing. In this chapter, the emulsification process will be discussed in terms of conventional and miniemulsification processes. The choice of the surfactant system for an optimal stability of the produced EP(D)M-based latexes will be emphasized. Finally, the influence of the cement viscosity on the emulsification process will be discussed. 3.2 Experimental section Chemicals Surfactants, n-heptane (Biosolve, AR), hexadecane (HD, Aldrich, 99 %), and cetyl alcohol (CA, Aldrich, 99 %) were used as received. Deionized water was used for all latexes. The structures and characteristics of the surfactants and polymers used in this study are summarized in Tables 3.1 and 3.2, respectively. Table 3.1: Description of the surfactants. Name Chemical name Structure Supplier SDBS Brij 97 Sodium dodecyl benzene sulfonate Polyoxyethylene (10) oleyl ether HLB i Value M n (g mol -1 ) O CH 3 ( CH 2 ) 10 CH 2 S ONa Fluka O C 18 H 35 ( O CH 2 CH 2 ) 10 OH Aldrich Igepal CO-890 Polyoxyethylene (40) nonylphenyl ether C 9 H 19 O CH 2 CH 2 OH ( ) 40 Aldrich Brij 700 Polyoxyethylene (100) stearyl ether i HLB = Hydrophilic-lipophilic balance ( ) C 18 H 37 O CH 2 CH 2 OH 100 Aldrich

40 Production of colloidally stable latexes 29 Table 3.2: Properties of the polymers studied. Polymer No. P1 P2 P3 P4 P5 Polymer name Supplier Type of diene M w (g mol -1 ) Lucant HC-20 (EPM) EPM 01371L EPDM 99488K EPDM 99488L Trilene 67 (EPDM) Mitsui Chemicals, Inc. DSM Elastomers DSM Elastomers DSM Elastomers Crompton Corp. Ethylene content (mol%) Propylene content (mol%) Diene content (mol%) Ethylidene norbornene Ethylidene norbornene Ethylidene norbornene Preparation of an artificial latex from a low molecular weight EP(D)M Conventional emulsification Recipes are collected in Table 3.3. If necessary, the first step involved the dissolution of the polymer in n-heptane in order to reduce its viscosity. Surfactant was separately dissolved in water with a similar molar concentration of surfactant for all the systems studied. In a second step, the organic phase was brought into the aqueous phase, the resulting blend being stirred for one minute with a rotor-stator Ultra-Turrax T25 Basic at rpm. The size of the eventually swollen polymer particles was reduced by processing the emulsion product of the Ultra-Turrax in a Niro-Soavi Lab homogenizer NS1001L Panda operating at 300 bars, with a shear rate of approximately s -1. Processing times ranged from one to three hours, depending on the viscosity of the samples.

41 30 Chapter 3 Table 3.3: Recipes for conventional emulsifications (CE). CE No. E1 Type Lucant HC-20 (EPM) Polymer Amount (g) Weight fraction in the latex (%) Heptane (g) Water (g) Type Surfactants i Amount (g) SDBS 0.9 E2 E3 E4 E5 E6 Lucant HC-20 (EPM) Lucant HC-20 (EPM) Lucant HC-20 (EPM) Lucant HC-20 (EPM) Lucant HC-20 (EPM) Brij Igepal CO Brij E7 Lucant HC E8 Lucant HC-20 + EPDM 99488K i i SDBS + Brij 700 (90-10) SDBS + Brij 700 (50-50) SDBS + Brij 97 (50-50) SDBS + Brij 700 (50-50) i In all cases: (Emulsifier/EP(D)M) mol% and the total concentration of emulsifier in water is mol dm -3. i i Weight fraction of organic phase in the latex: 26.1 wt%. The calculations are based on M w of each polymer, given in Table Miniemulsification Recipes are collected in Table 3.4. Both hexadecane (HD) and cetyl alcohol (CA) were employed as costabilizers to produce artificial latexes. When HD was used, it was mixed with the organic phase before addition to the aqueous phase. However, as stated by Brouwer et al. 23, the initial presence of CA in the aqueous phase is a prerequisite for successful emulsification. Indeed, dissolution of CA in the oil phase before addition to the aqueous phase causes instantaneous destabilization of the emulsion after cessation of stirring. Therefore, CA was added in a first step to the aqueous solution of SDBS. In a subsequent step, the CA / water / SDBS mixture was stirred for two hours at 65 C to promote the dissolution of CA. After cooling down the mixture to room temperature, the solution was subjected to pulsed ultrasonification (sonicating probe 400W, Dr.

42 Production of colloidally stable latexes 31 Hielscher UP400S) for one minute with an amplitude of 50 %, in order to enhance the formation of mixed emulsifier liquid crystalline structures. The latter structures are believed to improve emulsifier adsorption and emulsion stability. In a last step, the organic phase, consisting of a neat polymer or a polymer diluted with n-heptane, was added to the aqueous phase. An emulsion with submicron polymer particles was obtained by stirring with the Ultra-Turrax at rpm, followed by particles shearing in the homogenizer at pressures ranging from 300 to 1100 bars, depending on the viscosity of the organic phase. Table 3.4: Recipes for miniemulsifications (ME). ME No i. M1 M2 M3 Type Lucant HC-20 + EPDM 99488K Lucant HC-20 + EPDM 99488K Lucant HC-20 + Trilene 67 Polymer Amount (g) Weight fraction in the latex (%) Heptane (g) Water (g) Surfactants ii i M1 and M2 were both prepared with a pressure of 300 bars, whereas a pressure of 1100 bars was used for M3. ii In water: [SDBS] = mol dm -3 and [costabilizer] = mol dm -3. iii Weight fraction of organic phase in latex M1: 26.6 wt%. iv Weight fraction of organic phase in latex M2: 26.7 wt%. Type Amount (g) iii SDBS + CA iv SDBS + HD SDBS + HD Characterization of polymers Molecular Weights Number- and weight-average molecular weights ( M and n M w ) as well as molecular weight distributions were determined using a Waters 2690 Alliance Size Exclusion Chromatograph (SEC) equipped with two Styragel HR 5E columns, a Waters 410 differential refractometer, and a Viscotek T50A differential viscosimeter. The eluent was THF, and the elution volumetric flow rate was maintained at 1 ml min -1. Absolute

43 32 Chapter 3 molecular weights were calculated using polystyrene standards, with the universal calibration principle and Mark-Houwink parameters [K = dl/g, a = 0.476]. Composition Copolymers compositions were determined by 1 H and 13 C nuclear magnetic resonance (NMR), on a Varian-500 spectrometer at 25 C, using TMS as internal standard and CDCl 3 as solvent. Viscosities All viscosities (Pa s) of pure polymers and polymer/n-heptane mixtures were measured at 20 C as a function of the shear rate (s -1 ) with an AR 1000-N rheometer from TA Instruments Characterization of latexes - Particle size distribution Particle size distributions and volume-average diameters of latex particles were determined with a Coulter LS 230. This analyzer uses the principles of light scattering, based on both Fraunhofer and Mie theories, to determine particle size distributions. Moreover, the range of detectable particle sizes is extended to the submicron region (lower size limit: 40 nm of diameter). 3.3 Results and discussion Optimization of the surfactant system The first experiments, concerning conventional emulsifications, were carried out with the Lucant HC-20 ethylene-propylene copolymer (EPM) and different types of

44 Production of colloidally stable latexes 33 emulsifier. Particle size distributions obtained for emulsions E1 to E4 are presented in Figure Intensity E 4 E 3 E 2 E Particle Diameter ( m) Figure 3.1: Influence of the nature of the surfactant on the particle size distribution of EPM HC-20 latexes: SDBS (E1), Brij 97 (E2), Igepal CO-890 (E3), and Brij 700 (E4). By using SDBS as surfactant, particles with an average diameter of 350 nm can be obtained (latex E1). However, a bimodal particle size distribution was observed, with a secondary peak around 2 m. This peak was present in all our studied systems, except in latex E4. In this formulation, Brij 700 was used as emulsifier. With Brij 700, particles with an average diameter of 430 nm were obtained. The better stabilization obtained with Brij 700 may be explained by its high hydrophilic-lipophilic balance (HLB), being 18.8 (Table 3.1). The HLB value represents the tendency of an emulsifier to act as an oil-soluble or as a water-soluble type of emulsifier 24. A low HLB, e.g. 1-9, indicates an oil-soluble substance, whereas a high HLB, e.g , suggests a water-soluble compound. Lipophilic emulsifiers are typically nonionic, such as sorbitan trioleate (HLB = 1.8) or propylene glycol monolaurate (HLB = 4.5), as well as the saturated and unsaturated fatty acids. On the other hand, hydrophilic emulsifiers are typically ionic, such as soaps of alkyl or aryl sulfuric acids, e.g. sodium lauryl sulfate or sodium

45 34 Chapter 3 dodecyl sulfate, or soaps of alkyl or aryl sulfonic acids, e.g. sodium dodecyl benzene sulfonate (HLB = 11.7). Moreover, for nonionic surfactants, the HLB value is also related to the ethoxylation level of the surfactant, namely the ethylene oxide content, which represents the watersoluble portion of the surfactant molecule. As a consequence, more ethylene oxide units lead to a higher water-solubility and a higher HLB value. In the present work, Brij 700 possesses the highest ethoxylation level of the surfactants used, hence the highest HLB value. Its use will enhance steric stabilization of the latex particles, compared to the other surfactants, since its long hydrophilic chain will generate the longest distances between particles. As explained earlier, Ostwald ripening often leads to the destabilization of a latex. To avoid this phenomenon, the miniemulsification principle may help. Thus, polymer diffusion from small to larger particles would be retarded due to the presence of a costabilizer. However, the principle of retardation of polymer transport from small to larger particles is only operative for systems with polymers that are slightly water soluble 25,26. Because EP(D)M can be regarded as completely insoluble in water, Ostwald ripening is not very likely the reason for the 2 m particles observed. Nevertheless, HD and CA were employed to further understand the mechanisms involved in the stabilization of the submicron particles. Figure 3.2 represents a comparison of the particle size distributions obtained for latexes based on mixtures of EPM HC-20 and EPDM 99488K (polymers P1 and P3), produced either with the conventional emulsification or with the miniemulsification route. Both costabilizers, i.e. HD and CA, lead to similar particle size distributions and to volume-average diameters of 360 nm. A good latex stability was provided by the surfactant combinations SDBS/HD and SDBS/CA, for a period exceeding at least three weeks, completely suppressing the peak at 2 m.

46 Production of colloidally stable latexes Intensity E8 M2 M1 0 0,01 0, Particle Diameter ( m) Figure 3.2: Particle size distribution obtained for latexes based on mixtures of EPM HC-20 and EPDM 99488K, and made by miniemulsification with SDBS/CA (M1) or with SDBS/HD (M2), or by conventional emulsification with an equimolar mixture of SDBS and Brij 700 (E8). It has been observed that an efficient surfactant, i.e. steric (Brij 700, Figure 3.1) or electrostatic (SDBS, Figure 3.2), is necessary to ensure the colloidal stability of submicron EP(D)M latexes. However, SDBS alone (Figure 3.1) is not able to avoid the existence of 2 m particles after homogenizing at 300 bars. Our results point to the length of the hydrophobic moiety to be a key parameter in the emulsification process. In order to emphasize the role of the aliphatic part of the stabilizing system, a study was carried out with Brij 97 as surfactant. The hydrophobic tail of both Brij 700 and Brij 97 consists of a succession of 18 carbon atoms. However, as observed in Figures 3.1 and 3.3, the use of Brij 97 did not lead to a colloidally stable latex with a monomodal particle size distribution. As explained earlier, the hydrophilic head of Brij 97 is too short to act as an efficient steric stabilizer. However as depicted in Figure 3.3, an equimolar mixture of Brij 97 and SDBS leads to a monomodal particle size distribution and a volume-average diameter of 420 nm. So, 2 m particles can be avoided using a combination of SDBS and Brij 97 (Figure 3.3), of SDBS and HD or CA (Figure 3.2), or with Brij 700 alone (Figure 3.1). In all those systems, an efficient colloidal stability was provided by either steric or

47 36 Chapter 3 electrostatic repulsions, and the presence of 2 m particles was avoided by a long hydrophobic chain Intensity E7 E2 E1 0 0,01 0, Particle Diameter ( m) Figure 3.3: Particle size distribution of EPM HC-20 latexes stabilized with: SDBS (E1), Brij 97 (E2), an equimolar mixture of SDBS and Brij 97 (E7). The long aliphatic part, consisting of a succession of at least 16 carbon atoms, may act as a co-solvent for the EP(D)M. Thus, polymer coils are partly swollen by the costabilizer or by the hydrophobic tail of the surfactant. So the viscosity of the particles is reduced, facilitating the formation of submicron particles. Since both Brij 700 and SDBS with HD or CA were able to avoid the existence of 2 m particles and to produce particles with an average diameter of 350 nm, mixtures of both surfactants were then investigated in order to achieve the most suitable surfactant system. The molar ratio of emulsifier and EP(D)M was kept constant; the molar ratio of SDBS and Brij 700 was the only variable parameter. The influence of the SDBS/Brij 700 molar ratio on the particle size distribution of EPM HC-20-based latexes is depicted in Figure 3.4. The best result for the studied mixtures of emulsifiers was obtained for latex E6 stabilized with an equimolar mixture of SDBS and Brij 700. A monomodal particle size distribution was obtained, the volume-average particle diameter being 360

48 Production of colloidally stable latexes 37 nm. This latex remained colloidally stable for at least two months. Note that latex E4, less stable, was characterized by a larger volume-average particle diameter, i.e. 430 nm. 40 Intensity E4 E6 E5 E1 0 0,01 0, Particle Diameter ( m) Figure 3.4: Influence of SDBS/Brij700 molar ratio on the particle size distribution of EPM HC-20 latexes. The following molar ratios were used: 100/0 (E1), 90/10 (E5), 50/50 (E6), and 0/100 (E4) Influence of polymer viscosity on particle size distribution According to literature 4, the main criterion for successful emulsification of a polymer is the low viscosity of the cement: above a polymer viscosity of 10 Pa s at room temperature, emulsification is hardly possible. Burton and O Farrell were able to produce a latex containing submicron particles with a cement exhibiting a viscosity of 1.5 Pa s. Unfortunately, no experimental details about shear rate or pressure during the homogenizing step were mentioned. For the present work, five polymers (Table 3.2) were used to determine the influence of the polymer viscosity on the particle size distribution of the obtained artificial latexes. As depicted in Figure 3.5, the apparent viscosity of the polymers does not significantly depend on the shear rate, indicating that all polymers behave like Newtonian fluids. Viscosities range from 0.5 Pa s to 10 6 Pa s at 20 C.

49 38 Chapter 3 Viscosity (Pa.s) Shear rate (s -1 ) Figure 3.5: Viscosity of the pure polymers measured at 20 C as a function of shear rate. The polymers used are P1 (, = 0.5 Pa s), P2 (, = 100 Pa s), P3 (, = 510 Pa s), P4 (, = Pa s), and P5 (, = 10 6 Pa s). Burton and O Farrell s statement 4 regarding the required viscosity of a polymer for its successful emulsification has been verified. As expected, the high viscosities of pure polymers P2 to P5 forbid their emulsification into stable dispersions of submicron particles in water. Polymer P1 was the only polymer which could be emulsified without addition of a low-viscosity solvent, i.e. n-heptane. Extensive studies on several cements based on mixtures of EPM HC-20, EPDM 99488K, EPDM Trilene 67, and n-heptane (Table 3.5) were carried out in order to determine the influence of the viscosity of the polymer solution on the emulsification process. All cements obeyed Newton s law of viscosity within experimental error.

50 Production of colloidally stable latexes 39 Table 3.5: Polymer solutions used for viscositiy measurements. Cement EPM HC-20 (wt%) EPDM 99488K (wt%) n Heptane (wt%) Viscosity (Pa s) C C C C C C The viscosities of cements C2 to C6 are lower than 10 Pa s. Thus, successful emulsifications are expected, leading to submicron particles. However, as depicted in Figure 3.6, only latexes made of cements with viscosities lower than 2 Pa s (C3 to C6) lead to submicron particles. 50 Intensity C2 C3 C4 C5 C Particle Diameter ( m) Figure 3.6: Influence of cement viscosity on particle size distribution of latexes based on cements C2 ( = 3.7 Pa s), C3 ( = 1.7 Pa s), C4 ( = 0.9 Pa s), C5 ( = 0.5 Pa s), and C6 ( = 0.3 Pa s). As can be concluded from Table 3.5 and Figure 3.6, particle size distributions and average diameters of the studied latexes are not significantly related to the composition

51 40 Chapter 3 of the polymer. In contrast, the cement viscosity plays a dominant role in the emulsification of the polymer. Indeed, a cement with a viscosity ranging from 2 to 10 Pa s can easily be emulsified. However, submicron particles will be hard to obtain for pressures below 1100 bars. Reduction of cement viscosities to values lower than 2 Pa s leads to proper formation of submicron particles with a volume-average diameter of 360 nm. This phenomenon may be explained by the decrease of the entanglement density 27 with decreasing viscosity and molecular weight. A higher molecular weight leads, for linear copolymers, to longer chains and to higher viscosities. These chains tend to form many physical entanglements that determine the fixed macroscopic morphology of the particles, counteracting deformation and further breaking up. The present work points out that EP(D)M viscosities in the range of 2 to 10 Pa s still result in too many entanglements per molecule. On the other hand, for similar polymers with viscosities below 2 Pa s, the homogenizing step of the emulsification, at pressures ranging from 300 to 1100 bars and with shear rates in the order of to s -1, respectively, is facilitated by a lower entanglement density. Note that higher viscosities probably require higher homogenization pressures that were not within reach with our equipment. 3.4 Conclusions The emulsification of low molecular weight EP(D)M requires the presence of two important stabilizing parts: an efficient surfactant and a species with a long alkyl chain. The surfactant, steric or electrostatic, ensures the colloidal stability of the obtained latex. The long aliphatic chain, i.e. the hydrophobic tail of the surfactant or the costabilizer, probably acts as a co-solvent to swell and disentangle the polymer, and therefore helps the breaking-up of the particles during the homogenizing step. Hence, low molecular weight EP(D)M can be successfully emulsified without addition of organic solvent, either by using a conventional method of preparation of artificial latexes or by using a miniemulsification procedure. For the conventional

52 Production of colloidally stable latexes 41 procedure, the best result has been obtained with a combination of electrostatic (anionic surfactant, SDBS) and steric (nonionic surfactant, Brij 700) stabilization. For miniemulsification, SDBS has been used in combination with hexadecane or cetyl alcohol. In all recipes leading to stable emulsions, a monomodal particle size distribution and an average diameter of 360 nm were obtained. An important parameter, in addition to the choice of the emulsifier system, is the viscosity of the polymer. It has been demonstrated that the viscosity of the elastomer has to be lower than 2 Pa s at 20 C for proper production of submicron particles by high shear stirring followed by homogenization at high pressure ( bars) and with high shear rates ( s -1 ).

53 42 Chapter 3 References 1. Johnsen, K. E.; Pelletier, R. R.; Dow Chemical, U.S , Blackley, D. C. in "Polymer Lattices: Science and Technology"; 2nd Ed., 1997, Chapter Yang, Z. Z.; Xu, Y. Z.; Zhao, D. L.; Xu, M. Colloid Polym. Sci. 2000, 278, Burton, G. W.; O'Farrell, C. P. J. Elastomers Plast. 1977, 9, Fitch, R. M. in "Polymer Colloids: A Comprehensive Introduction"; London: Academic Press, 1997, Chapter Derjaguin, B. V.; Landau, L. Acta Physicochim. URSS 1941, 14, Verwey, Evert J. W. and Overbeek, J. T. G. Theory of the Stability of Lyophobic Colloids; Elsevier, Bromley, C. W. A. Colloids Surf. 1986, 17, Davies, S. P.; Thompson, M. W.; Imperial Chemical Industries, UK , Napper, D. H.; Netschey, A. J. Colloid Interf. Sci. 1971, 37, Napper, D. H. Special Publication - Royal Society of Chemistry 1982, 43, Colombie, D.; Landfester, K.; Sudol, E. D.; El-Aasser, M. S. Langmuir 2000, 16, Sung, A. M.; Piirma, I. Langmuir 1994, 10, Asua, J. M. Prog. Polym. Sci. 2002, 27, El-Aasser, M. S.; Lack, C. D.; Choi, Y. T.; Min, T. I.; Vanderhoff, J. W.; Fowkes, F. M. Colloids Surf. 1984, 12, Higuchi, W. I.; Misra, J. J. Pharm. Sci. 1962, 51, Hansen, F. K.; Ugelstad, J. J. Polym. Sci. Part A: Polym. Chem. 1979, 17, Rodriguez, V. S. PhD Thesis, University of Lehigh, Burke, O. W., Jr.; Exxon Research and Engineering, U.S , Saunders, F. L.; Pelletier, R. R.; Dow Chemical, U.S , 1972.

54 Production of colloidally stable latexes Beerbower, A.; Burton, G. W.; Malloy, P. L.; Exxon Research and Engineering, U.S , Dekkers, M. E. J.; Adams, M. E.; General Electric, U.S , Brouwer, W. M.; El-Aasser, M. S.; Vanderhoff, J. W. Colloids Surf. 1986, 21, Becher, P. in "Emulsions: Theory and Practice"; 2nd Ed., 1965, Chapter Kabal'nov, A. S.; Pertsov, A. V.; Shchukin, E. D. Colloids Surf. 1987, 24, Tauer, K. in "Handbook of Applied Surface and Colloid Chemistry"; Holmberg, K Eds., Wiley, 2002, 1, Chapter Porter, R. S.; Johnson, J. F. Chem. Rev. 1966, 65, 1.

55

56 About crosslinking of EP(D)M-based latexes Abstract Crosslinking of EP(D)M-based latexes has not thoroughly been studied yet. Therefore, the aim of this chapter is to improve the general understanding of the chemistry involved in the crosslinking process. This chapter especially emphasizes the influence of the initiation method, i.e. by a peroxide or a pulsed electron-beam, on crosslinking efficiency. The efficiency of various peroxides is also reported. All crosslinking efficiencies were obtained after extraction of the soluble polymer by THF. The incorporation of the co-agent, i.e. divinylbenzene, into the EP(D)M phase, was studied on a microscopic level by solid-state 13 C and 1 H NMR. The distribution of co-agent in the latex particles was revealed by cryo-transmission Electron Microscopy (cryo-tem). Crosslinking of a low molecular weight EP(D)M latex usually requires the presence of a coagent, e.g. divinylbenzene, 1,6-hexanediol diacrylate or poly(1,2-butadiene). The efficiency of crosslinking initiated by a pulsed electron-beam was improved to a great extent by the presence, in the aqueous phase, of potassium nitrosodisulfonate, also referred to as Fremy salt. MALDI- ToF-MS was used to determine the influence of electron-beam irradiation on the chemical stability of surfactants. It was demonstrated that SDBS is not degraded by the irradiation, and is therefore the surfactant of choice for the stabilization of EP(D)M-based latexes subjected to electron-beam irradiation. Tillier, D. L.; Meuldijk, J.; Magusin, P. C. M. M.; van Herk, A. M.; Koning, C. E. Manuscript submitted for publication.

57 46 Chapter General introduction In the 1820 s, natural rubber's potential for the manufacture of waterproof fabrics and a wide range of other consumer and industrial goods was mostly appreciated at room temperature. In cold weather, rubber products froze stiff and cracked, whereas high summer temperatures resulted in rubber products melting down to a useless, glue-like material. Therefore, the rubber industry was about to expire when in 1839 Charles Goodyear invented the sulfur vulcanization of rubber 1. Confirmed in 1843 by Thomas Hancock, mixing rubber with sulfur and heating the mixture could give rise to a nontacky, waterproof, and weatherproof natural rubber 2. Nowadays, three-dimensional network structures also referred to as crosslinked polymers, e.g. elastomers, arouse a considerable industrial interest, mainly in automotive applications, building, and electrical sectors. More specialized grades may also be found in pharmaceutical, medical and food contact applications. Sulfur vulcanization is only one of the numerous methods for polymer crosslinking 3. This type of modifications in polymeric structure of plastic materials may occur either by conventional chemical procedures, usually involving silanes or peroxides, or by exposure to ionizing radiation from either radioactive sources, or highly accelerated electrons General mechanism of crosslinking Peroxide crosslinking Crosslinking of natural rubber initiated by benzoyl peroxide was first reported in the early 1910 s by Ostromislensky 4. However, peroxide crosslinking only became commercially available with the introduction of dicumyl peroxide in 1955.

58 About crosslinking of EP(D)M-based latexes 47 Nowadays, a wide variety of peroxides may be used to crosslink most elastomers. Peroxides have the ability to crosslink saturated elastomers including low-density polyethylene 5 (LDPE), ethylene-propylene copolymers 6 (EPM), and silicones 7, which can not be crosslinked with other types of vulcanizing agents. Peroxide crosslinking results in shorter crosslinking times than sulfur vulcanization, and is a relatively more simple process as well. The typical mechanism of peroxide crosslinking may be divided into three major reactions: - The homolytic thermal decomposition of the peroxide, R-O-O-R, results in the formation of two alkoxy free radicals, R-O. - Each radical is able to abstract a hydrogen atom from the polymer chain, P-H. - Combination reactions between radicals on adjacent polymer chains, P, then lead to the formation of a crosslinked polymer, P-P. The extent of peroxide decomposition depends on its thermal history, which includes parameters such as time and temperature. The relative stability of a peroxide is usually characterized by the so-called ten-hour half life temperature. This temperature corresponds to the temperature required for the decomposition of 50 % of a peroxide in ten hours. As exemplified in Table 4.1, the ten-hour half life temperature varies greatly for different peroxides 8. Most of the peroxides commonly used for elastomers crosslinking, e.g. dialkyl peroxides, exhibit a very high stability, and therefore require high temperatures for decomposition.

59 48 Chapter 4 Table 4.1: Commercially available peroxides. Types Structure Ten-hour half life temperature Diacyl peroxides t-alkyl peroxy esters O O R C O O C R' O R C O O R' 20 to 75 C 50 to 110 C Alkyl hydroperoxides R O O H 135 to 170 C Dialkyl peroxides R O O R' 120 to 135 C Dialkylperoxy ketals R O O R'' C R' O O R''' 90 to 115 C As mentioned above, the final step of peroxide crosslinking of elastomers consists of the coupling of two radicals on adjacent polymer chains. However, in the presence of oxygen, the polymeric radical may couple with an oxygen molecule, leading to the formation of a hydroperoxy radical. This phenomenon results in polymer degradation, leading to a sticky and partially cured rubber. In the case of peroxide crosslinking of diene-containing elastomers, e.g. polybutadiene, it has been proposed 9 that the polymeric radicals formed by allylic hydrogen abstraction may undergo both coupling and addition to a double bond. Moreover, higher crosslinking efficiencies have been observed for polybutadiene polymers containing high 1,2-double bond contents. This suggests that side chain double bonds exhibit a higher reactivity towards addition than those within the main chain. This higher reactivity of side chains as compared to main chains has also been confirmed by radiation-induced crosslinking 10. However, chemical crosslinking may involve the generation of toxic (gaseous) sideproducts of peroxide degradation. Therefore, the physical and chemical changes, induced by absorption of sufficiently high energy radiation to produce radicals, are of considerable interest. Moreover, high energy radiation may result in crosslinking at

60 About crosslinking of EP(D)M-based latexes 49 room temperature, whereas peroxide crosslinking requires relatively elevated temperatures for the decomposition of the initiator. Radiation-induced crosslinking The natural radioisotopes radium and radon were the first radiation sources used. Nowadays, the most common commercial sources of ionizing radiation are 60 Co and 137 Cs for gamma irradiation 11-13, and electron accelerators for electron-beam (beta) irradiation 14. The principle of radiation-induced crosslinking may be exemplified by polyethylene curing. The present mechanism 14 involves loss of two hydrogen atoms from adjoining chains, leading to a bond between two carbon-centered free radicals. Inter- and intrachain termination may then occur, the former giving rise to a three-dimensional network. Increased utilization of electron-beams, for modification and enhancement of polymer properties, has been well documented over the past forty years Electron-beam processing of crosslinkable plastics has yielded materials with improved dimensional stability, reduced stress cracking, improved heat resistance, reduced solvent and water permeability, and significant improvements in other thermomechanical properties. Enhancement of crosslinking Crosslinking of saturated elastomers, using either radiation or peroxide, may be limited by side reactions, e.g. chain scission. Therefore, a polymer may undergo crosslinking, resulting in an increased molecular weight, or chain scission 22 (or degradation), resulting in a reduced molecular weight. Note that, in the case of radiation-resistant polymers, no significant change in molecular weight is observed.

61 50 Chapter 4 Both types of reactions, i.e. crosslinking and chain scission, are currently of economical interest to add value to a wide variety of thermoplastics, elastomers, and other materials. For example, the beneficial changes observed in crosslinked polyethylene include increased modulus, tensile and impact strength, hardness, stresscrack resistance, and abrasion resistance. On the other hand, the chain scission effects observed in fluoropolymers 23,24 have been commercially exploited to produce fine micropowders exhibiting a chemically modified surface. For some polymers, such as poly(vinyl chloride) (PVC) 25, polypropylene (PP), poly(ethylene terephthalate) (PET), and polyethylene (PE) at elevated temperatures, crosslinking and chain scission may occur simultaneously. The crosslinking-to-chain scission ratio depends on factors including the total irradiation dose, the presence of oxygen, stabilizers and radical scavengers, as well as steric hindrances resulting from structural or crystalline forces. Crosslinking efficiencies may be enhanced by di- or trifunctional monomers, also referred to as co-agents 26. Examples of such crosslinking agents include ethylene dimethacrylate, divinylbenzene, trimethylolpropane trimethacrylate 27, or even poly(1,2- butadiene). Crosslinking in the presence of co-agents may occur either by addition of these co-agents to polymer radicals in competition with chain scission or by a transfer reaction, as described in Scheme 4.1 in the case of an ethylene-propylene copolymer (EPM). Note that the unsaturation present in EPM -H may further react with EPM for instance, leading to a divinylbenzene bridge between two EPM chains, when R 3 is a phenyl ring.

62 About crosslinking of EP(D)M-based latexes 51 Initiation By a peroxide: R 1 O R 2 CH 3 C CH 2 C CH 2 R 1 OH R 2 CH 3 C CH 2 C CH 2 H H H ( EPM H) ( EPM ) By radiation: EPM H Radiation EPM Propagation Chain scission: R 2 CH 3 C CH 2 C CH 2 R 2 C CH 3 CH 2 C CH 2 H H ( EPM ) Propagation with the co-agent leading to crosslinking: R 2 CH 3 R 2 CH 3 C H CH 2 C CH 2 ( EPM ) R 3 (CH CH 2 ) 2 C H CH 2 C CH 2 CH 2 CH R 3 ( EPM' ) CH CH 2 Termination Combination: EPM EPM EPM EPM EPM EPM' EPM EPM' EPM' EPM' EPM' EPM' Transfer to EPM-H EPM' EPM H EPM R 2 CH 3 C CH 2 C CH 2 H CH 2 CH 2 R 3 (EPM' H) CH CH 2 Scheme 4.1: Crosslinking of a polymer in the presence of a co-agent.

63 52 Chapter 4 Crosslinking in dispersed media So far, crosslinking of a polymer in water has very often been studied during the emulsion polymerization of the corresponding monomer. For instance, Bouvier-Fontes et al. 28 studied the seeded semi-batch emulsion polymerization of butyl acrylate (BA) in the presence of allyl methacrylate (AMA) or butanediol diacrylate (BDA) as crosslinking agents (Table 4.2), and initiated by potassium persulfate. Other groups considered the crosslinking occurring during the emulsion polymerization of styrene with divinylbenzene (DVB) 29,30, or the emulsion polymerization of methyl methacrylate with ethylene glycol dimethacrylate (EGDMA) 31,32. Table 4.2: Structure of various crosslinking agents. Crosslinking agent Structure O Allyl methacrylate CH 2 C C CH 3 Butanediol diacrylate O CH 2 CH CH 2 O ( CH 2 CH C O CH 2 CH 2 ) 2 CH 2 Divinylbenzene CH CH CH 2 O O Ethylene glycol dimethacrylate CH 2 C C O CH 2 CH 2 O C C CH 2 CH 3 CH 3 However, crosslinking of an already prepared latex, and more specifically peroxide crosslinking, has, to the best of the authors knowledge, hardly been studied. Although many peroxides are now commercially available, benzoyl peroxide is the most frequently used initiator for crosslinking of latexes. For instance, benzoyl peroxide has been employed in 1980 for the crosslinking of an artificial polyester latex 33, or more recently for the crosslinking of polyisoprene latexes 34. The use of a co-agent, e.g. divinylbenzene, has also been reported 35,36 to enhance crosslinking of latexes.

64 About crosslinking of EP(D)M-based latexes 53 As for radiation-induced crosslinking of latexes, it seems to have gained in interest. However so far, most of the studies were carried out on natural rubber latexes 37,38. The challenge consisted then of decreasing the irradiation dose while keeping the desired properties, as well as understanding the role of water in the crosslinking mechanism 38, Aim of this chapter As mentioned above, crosslinking of ethylene-propylene copolymer-based latexes suffers from a lack of investigations and knowledge. Therefore, the aim of the present chapter is to acquire a proper insight into the mechanisms and efficiencies involved in the crosslinking of such rubber-based latexes. Two processes have been investigated, i.e. crosslinking initiated by a peroxide and crosslinking initiated by a pulsed electron-beam irradiation. The results are described in this chapter. 4.2 Experimental section Chemicals Sodium dodecyl benzene sulfonate (SDBS, Fluka, technical grade, 80 %), polyoxyethylene (100) i stearyl ether (Brij 700, Aldrich), hexadecane (HD, Aldrich, 99%), pentane (Biosolve, AR), tetrahydrofuran (THF, Biosolve, AR), and the polymers were used as received. Relevant details of the polymers used in this study are collected in Table 4.3. Divinylbenzene (DVB, Aldrich, technical grade, 80 %), 1,6-hexanediol diacrylate (HDDA, UCB), and poly(1,2-butadiene) (Ricon 156, Cray Valley) were used without further purification. Peroxide crosslinking was performed with various initiators, see Table 4.4. i 100 repeating units of oxyethylene, see structure in Chapter 3.

65 54 Chapter 4 All other reagents were also used as received, viz. potassium nitrosodisulfonate (Fremy salt, FS, K 4 [(SO 3 ) 2 NO] 2, Aldrich), sodium formaldehyde sulfoxylate hydrate (SFS, Fluka, +98 %), disodium salt of ethylenediamine tetra-acetic acid (EDTA, Aldrich, +99 %), iron (II) sulfate heptahydrate (FeSO 4, Aldrich, +99 %). Deionized water was used in all recipes. Table 4.3: Characteristics of the polymers used. Polymer name Lucant HC- 20 (EPM) Supplier Mitsui Chemicals, Inc. Type of diene M w (g mol -1 ) Ethylene content (mol%) Propylene content (mol%) Diene content (mol%) EPDM 99488K DSM Elastomers Ethylidene norbornene Trilene 67 (EPDM) Crompton Corp. Ethylidene norbornene Table 4.4: Characteristics of the initiators used. Initiator Supplier, Grade Structure Di-t-butyl peroxyoxalate (DTBPO) t-butyl peroxypivalate (TBPP) Synthesized in the Polymer Chemistry laboratory of the Eindhoven University of Technology i Luperox 11M75, Atofina, 75 % in isododecane H 3 C CH 3 C O CH 3 CH 3 O O C C O CH 3 O C O O C CH 3 CH 3 Ten-hour half life temperature O O C CH 3 25 C CH 3 CH 3 C CH 3 58 C CH 3 Cumene hydroperoxide (CHP) Luperox CU90, Aldrich, 88 % in oil CH 3 C O OH 158 C CH 3 Benzoyl peroxide (BPO) Fluka, 75 % O C O O O C 73 C i According to the recipe described by Bartlett et al 40.

66 About crosslinking of EP(D)M-based latexes Preparation of a seed latex The preparation of the seed latex has been described in Chapter 3. Standard recipes used in the present work, for peroxide as well as for electron-beam crosslinking, are exemplified in Table 4.5 and Table 4.6, respectively. Table 4.5: Recipes of latexes used for peroxide-initiated crosslinking. Seed latex No. Lucant HC-20 EPDM 99488K Amount of product (g) Trilene 67 Heptane HD DVB i Ricon 156 Water SDBS i P P P P P P P i In all cases: wt DVB = 10 % wt EP(D)M and in water: [SDBS] mol dm -3. Table 4.6: Recipes of latexes used for electron-beam-initiated crosslinking and corresponding crosslinking efficiencies. Seed latex No. Lucant HC-20 Amount of product (g) Trilene 67 HD DVB HDDA Ricon 156 Water SDBS i FS i Gel content (%) EB EB EB EB i In all cases, in water: [SDBS] mol dm -3 and [FS] 10-2 mol dm -3. This concentration of FS implies that [K + ] mol dm -3.

67 56 Chapter 4 In a first step, SDBS was dissolved in water. A homogeneous mixture of EPM and EPDM was prepared with the help of pentane. After the evaporation of pentane, desired amounts of hexadecane and unsaturated materials, e.g. divinylbenzene or 1,6-hexanediol diacrylate and/or poly(1,2-butadiene) (Ricon 156), were added to the EP(D)Mcontaining mixture. In a second step, this organic phase was brought into the aqueous phase. The resulting blend was stirred for one minute with a rotor-stator Ultra-Turrax T25 Basic at rpm. The volume average diameter of the polymer particles was then reduced from 20 m to 360 nm by processing the emulsion product of the Ultra- Turrax in a Niro-Soavi Lab homogenizer Panda 2K, operating for 30 to 45 minutes at 1100 bars, with a shear rate of approximately s -1. Note that, when the seed latex was used in electron-beam-initiated experiments, after homogenization some Fremy salt (FS) was eventually added to the latex without colloidal destabilization Chemically-initiated crosslinking The crosslinking reactions initiated with peroxide were carried out in a 300 ml jacketed-reactor, equipped with a condenser and a downflow 45 pitched four-blade impeller. An EP(D)M seed latex was charged into the reactor, and the oxygen was removed by purging argon through the latex overnight. Before being charged into the reactor, the peroxide was dissolved in cyclohexane or toluene if necessary. The crosslinking reactions were performed at a stirring speed of 300 rpm and at a temperature ranging from 50 to 75 C, as reported in Table 4.7.

68 About crosslinking of EP(D)M-based latexes 57 Table 4.7: Recipes and reaction conditions for peroxide crosslinking and corresponding gel contents. No i. Seed latex Peroxide Additive Amount (g) Type Amount (g) Amount (mmol) Type Amount (g) T ( C) Reaction time (h) Gel content (%) P DTBPO Cyclohexane P TBPP P CHP SFS EDTA FeSO P CHP SFS EDTA FeSO P CHP SFS EDTA FeSO P BPO Toluene P BPO Toluene i See Table Crosslinking initiated by a pulsed electron-beam Set-up The linear electron accelerator (LINAC) of the Eindhoven University of Technology, used for the present work, is based on a modified 41,42 medical accelerator Philips SL As depicted in Figure , electrons are generated by an electron gun and are preaccelerated by a high voltage of the order of 30 kv. The energy of these electrons is then further increased in the acceleration tube with the help of an electromagnetic wave generated by a magnetron. The beam is focused and centered by magnets located around the tube. At the end of the acceleration tube, the electromagnetic wave is diverted into a dump while the electron beam, characterized by an energy of the order of 5 MeV, travels straight on through a 100 m-thick aluminum foil, separating the vacuum in the

69 58 Chapter 4 accelerator from the atmospheric air in the target holder. The scattered electrons then reach the sample, located 5 cm behind the foil. Figure 4.1: Schematic overview of the linear electron accelerator (from ref. 43). The irradiation is characterized by a pulse width of 4 s and a pulse repetition rate of 5, 10, 25 or 50 Hz. The experiments were carried out using two types of reactors 44, i.e. a batch reactor and a batchwise-operated loop reactor. As depicted in Figure 4.2a, the batch reactor consisted of a 5 ml, cylindrical cell, thermostated at 25 C. As for the loop reactor, presented in Figure 4.2b, it consisted of a 300 ml jacketed-vessel, equipped with a nitrogen inlet, and a downflow 45 pitched four-blade impeller. The feeding of the loop with latex was performed at a rate of 1 ml s -1, using a PTFE-peristaltic pump, from Cole-Palmer Instrument. The 5 ml-irradiation cell was connected to the vessel and to the pump using Iso-versinic tubes, characterized by a length of 1.5 m and a diameter of 0.8 cm.

70 About crosslinking of EP(D)M-based latexes 59 a- E-Beam b- E-Beam hg 2 cm 3.5 cm 1.5 cm 2.5 cm Figure 4.2: Schematic presentation of a: the batch reactor, and b: the loop reactor (from ref. 44). Procedure In a typical batch crosslinking procedure, 4.5 ml of a latex were charged into the reaction cell and irradiated at 25 C, with a frequency of 5, 10, 25, or 50 Hz. For crosslinking in a loop reactor, 150 g of a latex were charged into the 300 ml vessel. The oxygen dissolved in the latex was removed by purging nitrogen through the latex for at least one hour. The irradiation was carried out at 25 C, at a stirring speed of 100 rpm and with a frequency of 10 Hz Characterization of the latexes Particle size distribution A Coulter LS230 particle sizer was used to verify the colloidal stability of latexes after crosslinking. This analyzer uses the principles of light scattering, based on both Fraunhofer and Mie theories, to determine particle size distributions. Note that the lowest size limit of detectable particle is 40 nm. No significant destabilization was observed after crosslinking of the EP(D)M-based particles.

71 60 Chapter 4 Crosslinking efficiency The post-treatments included coagulation of the polymer latex using freeze-thawing cycles, filtration under vacuum over a Büchner funnel, and washing the coagulated product with deionized water. The polymer was then recovered and dried to constant weight in a vacuum oven at 60 C. Tetrahydrofuran (THF) was used for at least 10 h to extract the uncrosslinked EP(D)M. The crosslinking efficiency (CE) was then calculated with Equation 4.1: weight of polymer after extraction CE (4.1) weight of polymer before extraction Microstructure Incorporation of DVB into EP(D)M The microstructure of the crosslinked EP(D)M samples, i.e. the incorporation of DVB into EP(D)M, was analyzed using solid-state NMR spectroscopy. Proton-decoupled solid-state 13 C NMR spectra were recorded on a Bruker DMX500 spectrometer, operating at a 1 H and 13 C frequency of and MHz, respectively. A 4 mm magic-angle-spinning (MAS) probe head was used in a static mode as well as with a sample rotation rate of 6 khz. The radio-frequency power was adjusted to obtain 5 s 90 pulses both for the 1 H and 13 C nuclei. The ppm resonance of adamantane was used for external calibration of the 13 C chemical shift. Proton spin-lattice relaxation in the laboratory and in the rotating frame, T 1 ( 1 H) and T 1 ( 1 H), respectively, were measured for each component of the polymers separately both via cross-polarization (CP) to the 13 C nuclei and via direct excitation of the 1 H nuclei. Relaxation delays in CP-derived experiments, D1, were 1 and 3 s, and the number of experiments per relaxation data set, NE, was 20. Proton transverse relaxation, T 2, was determined at several temperatures ranging from -93 to +130 C, using a BVT-3000 variable temperature unit. The decay of the transverse magnetization was measured with the Hahn-echo pulse sequence (HEPS), (acquisition), where 2.5 s. An echo signal is formed after the second

72 About crosslinking of EP(D)M-based latexes 61 pulse in the HEPS with a maximum at time t = 2 after the first pulse. By varying the pulse spacing in the HEPS, the amplitude of the transverse magnetization, A(2 ), is measured as a function of time t = 2. T 2 was determined by computer fitting each 1 H NMR decay by a bi-exponential function. Distribution of DVB in the latex particles The distribution of DVB in the latex particles was analyzed using cryo-transmission Electron Microscopy (cryo-tem). For the present work, the use of cryo-tem was required because of the low glass transition temperature of the EP(D)M phase. To perform cryo-tem analysis, each latex sample was applied to a microscopy grid. The thin aqueous film obtained was then vitrified in liquid ethane before being transferred to a Philips CM12 microscope, for examination at liquid nitrogen temperature. Dehydration and major reorganization of the latex particles were prevented by the low temperature. Stability of the surfactants SDBS and Brij 700 were separately dissolved in deionized water, leading to 10 wt% solutions. Both solutions were subjected to a pulsed electron-beam irradiation, at 25 C, using the batch reactor. Both samples were irradiated at 25 Hz for 2 min, which corresponds to an estimated irradiation dose of 280 kgy 44. Note that this dose was the highest dose used for our latexes. The effect of pulsed electron-beam irradiation on surfactants was determined using Matrix Assisted Laser Desorption/Ionization Time of Flight Mass Spectrometry (MALDI-ToF-MS). Measurements were performed on a Voyager-DE-STR instrument (Applied Biosystems, Framingham, MA), equipped with a 337 nm nitrogen laser. Positive ion spectra were acquired in reflector mode. Sodium trifluoroacetate (Aldrich, 98 %) was used as cationic ionization agent. Dithranol and dihydroxybenzoic acid were chosen as the matrixes for SDBS and Brij

73 62 Chapter 4 700, respectively. Each matrix was dissolved in THF at a concentration of 40 mg ml -1. Sodium trifluoroacetate was added to THF at typical concentrations of 1 mg ml -1. The matrix and salt solutions were mixed in the ratio: 5 µl matrix: 0.5 µl salt. 0.5 µl of each aqueous solution of surfactant was hand-spotted on a target plate. 0.5µL of the matrix/salt mixture was then deposited on top of each sample after evaporation of water. 4.3 Results and discussion Chemically-initiated crosslinking Crosslinking efficiency Various peroxides were employed to evaluate their influence on EP(D)M crosslinking in emulsion. Note that the use of a co-agent was required to achieve significant crosslinking. As observed in Table 4.7, the highest crosslinking efficiency, i.e %, was obtained with the most reactive peroxide, i.e. DTBPO. However, the reactivity of this peroxide was so high that safety in the laboratory could no longer be warranted. Therefore, it was decided to prevent further utilization of DTBPO. The low efficiency of TBPP, i.e. 9.2 %, see Table 4.7, may be due to side reactions occurring during crosslinking. Indeed, as depicted in Equation 4.2, the decomposition of TBPP leads to the formation of two radicals, different towards hydrogen atom abstraction. The t-butoxy radical may initiate crosslinking as suggested in Scheme 4.1, whereas the less efficient t-butyl radical may terminate by combination. CH 3 CH 3 O C O O C CH 3 CH 3 CH 3 C CH 3 CH 3 C CO 2 CH 3 C CH 3 CH 3 CH 3 CH 3 O (4.2) Stable radical

74 About crosslinking of EP(D)M-based latexes 63 The use of a partially hydrophilic peroxide, e.g. CHP combined with a redox system, i.e. SFS, EDTA and FeSO4, leads to a gel content in the order of 15 to 20 %. Increasing the concentration of CHP does not further raise the amount of crosslinking. Moreover, as depicted in Figure 4.3a, cryo-tem pictures of latex particles crosslinked with CHP reveal a non-homogeneous distribution of DVB (dark regions), and therefore of crosslinked polymer over the latex particles (light grey regions). The use of CHP combined with a redox system as initiator enhances reactions near the surface of the latex particles, as suggested in more detail in Chapter 5, and therefore may explain the non-homogeneous distribution of DVB. In spite of a relatively poor efficiency of BPO, i.e. 13 to 20 %, cryo-tem pictures (Figure 4.3b) depicted a homogeneous distribution of DVB, as demonstrated by the absence of contrast inside the rubber particles. a- b- 100 nm 100 nm Figure 4.3: Cryo-TEM pictures of EP(D)M latex particles crosslinked with CHP (a) and BPO (b) in the presence of DVB. Both arrows point to the dark region of particles, where DVB is concentrated. Particles crosslinked with BPO exhibit a homogeneous distribution of DVB, resulting in the absence of dark domains. Incorporation of DVB into EP(D)M The incorporation of DVB into EP(D)M during crosslinking was studied using solidstate nuclear magnetic resonance. Solid-state 1H and 13 C NMR spectroscopy is a

75 64 Chapter 4 powerful technique to understand the molecular structure and chain dynamics of polymers 45. After soxhlet extractions with THF of latexes P1, P4, and P6, the obtained solids were dried to constant weight in a vacuum oven and submitted to solid-state NMR analysis. The differences observed in Figure 4.4, between the 13 C NMR spectra of the pure EP(D)M before crosslinking, a pure homopolymer of DVB and the EP(D)M crosslinked with DVB as co-agent and DTBPO, BPO, or CHP as initiators, show that it is possible to reasonably distinguish between the signals of EP(D)M and DVB. * * (e) (d) (c) (b) (a) Figure 4.4: 13 C NMR spectra obtained via cross-polarization for pure EP(D)M (a), pure poly(divinylbenzene) (b), and EP(D)M crosslinked in the presence of DVB with DTBPO (c), BPO (d), and CHP (e) as initiators. Asterisks indicate spinning side bands.

76 About crosslinking of EP(D)M-based latexes 65 Note that the 13 C spectra in Figure 4.4 were obtained via cross-polarization to the 13 C nuclei. This technique, as opposed to direct excitation, especially enhances the signal of rigid phases. Therefore, the signals of mobile EP(D)M appear fairly weak and nonquantitative in spectra obtained via cross-polarization, as observed in the spectra of pure EP(D)M in Figure 4.5. (b) (a) (ppm) 50 0 Figure 4.5: 13 C spectra obtained via cross-polarization to the 13 C nuclei (a) and via direct excitation of the 13 C nuclei (b) for pure uncrosslinked EP(D)M. Domains analysis by 1 H NMR T 1 and T 1 relaxation In the present investigation, proton spin-lattice relaxation in the rotating frame (T 1 ) and in the laboratory (T 1 ) was studied for the well-resolved signals in the 13 C NMR spectrum, i.e. signals of the aliphatic region (0-50 ppm) and signals of the aromatic region ( ppm) characteristic of DVB. 1 H T 1 and T 1 measurements yield mobility information of the considered nuclei at the nanosecond and millisecond timescale, respectively. In addition, 1 H T 1 and T 1

77 66 Chapter 4 measurements provide information about the miscibility and domain sizes of various phases inside a polymer. As a result of the proton-proton dipolar coupling, protons in a polymer continuously exchange their polarization ( nuclear magnetization ). This process, usually referred to as spin diffusion, tends to average out local differences in NMR properties, such as relaxation. Spin diffusion is fast in rigid polymers with closely interspaced protons, and slow in mobile polymers with low proton density. For a standard spin diffusion coefficient D of approximately 1 nm 2 ms -1, if the average domain size in a polymer blend is smaller than about 1 nm, proton-proton spin diffusion averages out any T 1 or T 1 relaxation difference. All protons then decay with the same effective T 1 and T 1. In contrast, if the domain size is larger than about 50 nm, spin diffusion is too slow to average out such differences, and each phase will decay with its intrinsic, probably different T 1 and T 1 values. Furthermore, since T 1 tends to be 10 to 100 times longer than T 1 the effective diffusion path length is longer in T 1 experiments. Therefore, in the intermediate range, i.e. for a domain size ranging from 1 to 50 nm, different effective T 1 values and a single effective T 1 are expected. Spin diffusion is still able to average out T 1 differences, although it is not able to homogenize T 1. Proton T 1 measurements, performed via cross-polarization to the 13 C nuclei on a homopolymer of DVB, as well as on EP(D)M samples crosslinked in the presence of DVB with DTBPO, BPO, or CHP as initiators, are depicted in Figure 4.6. The aromatic signals characteristic of poly(divinylbenzene) at 130 and 145 ppm exhibit a T 1 value similar, within experimental error, to that of the aliphatic signals of the EP(D)M crosslinked with DTBPO as initiator. Therefore, this points out that DVB does not form domains of poly(divinylbenzene) inside this EP(D)M sample, but is rather well incorporated into the rubber. As a result of spin diffusion between EP(D)M and DVB phases, DVB exhibits a significantly lower T 1 value in the rubber than in the

78 About crosslinking of EP(D)M-based latexes 67 homopolymer. This is consistent with the formation of small DVB segments between EP(D)M chains. 100 T 1 (ms) Shift (ppm ) Figure 4.6: Proton T 1 measurements performed on a homopolymer of DVB ( ) and on EP(D)M samples, crosslinked in the presence of DVB with DTBPO ( ), BPO ( ), and CHP ( ) as initiators, respectively. On the contrary, the analysis of T 1 via cross-polarization to the 13 C nuclei, for EP(D)M samples, crosslinked in the presence of DVB and using BPO or CHP as initiators, indicates different T 1 behaviors for the signals of EP(D)M and DVB, as depicted in Figure 4.6. Cross-polarization to the 13 C nuclei tends to enhance the signal originating from the rigid phase, as mentioned earlier, and therefore acts as a filter. As a consequence, to obtain quantitative information, a thorough investigation of T 1 was carried out by direct excitation of the protons. The existence of two components was then established. One component, referred to as the fast component, is characterized by a low T 1 value, i.e ms for the sample crosslinked with BPO and 0.53 ms for the one crosslinked with CHP, measured at 273 K. The other component, characterized by a higher T 1 value, i.e ms for the sample crosslinked with BPO and 5.48 ms for the one crosslinked with CHP, measured at 273 K, is referred to as the slow component.

79 68 Chapter 4 Comparison of this two-component behavior with chemically-resolved proton T 1 measurements obtained via cross-polarization to the 13 C nuclei, suggests that the slow component, i.e. the most rigid phase, mainly corresponds to DVB, whereas the fast component, i.e. the mobile phase, mainly consists of EP(D)M. The analysis of two 13 C spectra with different proton T 1 -filters, for the sample crosslinked with BPO as initiator, see Figure 4.7, indicates that the EP(D)M and the DVB phases decay similarly. Hence, both EP(D)M and DVB show identical T 1 behavior. The detection of two different T 1 values and a single T 1 value demonstrates the presence of distinct local DVB-rich domains in EP(D)M samples crosslinked with BPO. The sizes of these domains are in the order of 1 to 50 nm, as suggested by the above mentioned arguments s 1 H T 1 -filter (128 scans) 600 ms 1 H T 1 -filter (128 scans) ppm Figure 4.7: Superposition of two 13 C spectra obtained at different T 1 -filter times via cross-polarization to the 13 C nuclei, for the sample crosslinked in the presence of DVB and with BPO as initiator. A similar 1 H T 1 analysis was carried out for the sample crosslinked with CHP as initiator, see Figure 4.8. The EP(D)M signal is slightly less affected by the T 1 -filter than the DVB signal. This indicates that both the EP(D)M and the DVB phases relax at slightly different rates. The detection of two different T 1 values as well as two different T 1 values also demonstrates the presence of two distinct types of domains in EP(D)M

80 About crosslinking of EP(D)M-based latexes 69 samples crosslinked with CHP. The size of these domains is now larger than 50 nm, as suggested by the previously mentioned arguments s 1 H T 1 -filter (2048 scans) 400 ms 1 H T 1 -filter (6200 scans) ppm ppm Figure 4.8: Superposition of two 13 C spectra obtained at different T 1 -filter times via cross-polarization to the 13 C nuclei, for the sample crosslinked in the presence of DVB and with CHP as initiator. This difference in domains sizes, between samples crosslinked with BPO or CHP as initiators, may find an explanation in the distribution of DVB in the EP(D)M particles, as observed in Figure 4.3. Indeed, in the sample crosslinked with CHP as initiator, the non-homogeneous distribution of DVB within the EP(D)M particles may lead to the formation of larger domains than in the sample crosslinked with BPO as initiator. Network structure analysis by 1 H NMR T 2 relaxation T 2 relaxation was investigated for samples crosslinked with BPO and CHP as initiators, to collect information on network defects and on the heterogeneous distribution of network junctions. As reported by Litvinov 46,47, the T 2 relaxation time for elastomers networks, at temperatures about C above T g, is very sensitive to rotational motions of the polymer chains. In particular, T 2 at elevated temperatures depends on the amplitudes of collective chain motions faster than 1 ms. These chains motions may be affected by the presence of physical and chemical crosslink points. At

81 70 Chapter 4 low temperatures, the chain motion is determined by localized motions, and is naturally enhanced by increasing temperatures. At high temperatures, the chain mobility of a crosslinked material is limited by the constraints, originating from crosslinks. T 2 becomes, then, nearly independent of temperature. In our proton T 2 experiments, for EP(D)M samples crosslinked with BPO or CHP as initiators, the observed decays of the so-called Hahn-echo as a function of the echo time, 2, clearly deviated from a mono-exponential behavior. However, it was possible to fit these decays with a bi-exponential function, described by Equation : 2 2 A(2 ) A exp( ) B exp( ) (4.3) T T 2 A 2B where T 2A and T 2B represents the long and short T 2 values, respectively. A long T 2 value is characteristic of mobile polymer chains. The T 2A component is therefore assigned to the mobile EP(D)M segments relatively far away from the nearest crosslinks. The fast T 2B component is probably associated with the theoretically less mobile DVB-rich domains. To enhance the stability of the bi-exponential fit, a coupling between the various data sets was made, assuming that both fractions A/(A+B) and B/(A+B) are temperature independent. Physically, this implies that boundaries between mobile and immobile phases do not change upon temperature variation. This is a reasonable assumption for a system with clear-cut boundaries between EP(D)M and DVB domains. However, for real crosslinked rubbers, this model may be too simple. Figure 4.9 illustrates the T 2A values, collected for EP(D)M samples crosslinked with BPO and CHP as initiators, as a function of temperature. The temperature dependence of T 2B is less pronounced, and therefore not depicted here.

82 About crosslinking of EP(D)M-based latexes T 2A ( s) Temperature (K) Figure 4.9: Evolution of the slow transverse relaxation, T 2A, with calibrated temperature, for two samples crosslinked in the presence of DVB, with BPO ( ) and CHP ( ) as initiators, respectively. Although the gel contents, determined by soxhlet extractions, were very similar, i.e. 13 % and 15 % for the samples crosslinked with BPO and CHP, respectively, the different T 2 behaviors clearly indicate differences in network structure at high temperatures. At low temperatures, T 2 values do not depend on the crosslinking density, and are therefore comparable. The number of statistical segments between network junctions, Z, may be determined using Equation 4.4 as suggested by Litvinov 46,47 : where T Z (4.4) a. T p rl T 2 is the value of T 2 at the high temperature plateau, and T 2 is the value of T 2, independent of the crosslink density, in the glassy state, i.e s for the present system. The coefficient a depends on the angle between the segment axis and the internuclear vector between protons of the main chain. For polymers containing aliphatic protons in the main chain, the coefficient a is close to 6.2 ± The present systems consist of both aliphatic and aromatic protons. Therefore, a deeper NMR investigation is required to determine the exact number of segments p 2 rl 2

83 72 Chapter 4 between crosslink points. However, it is possible to estimate the ratio of the distances between crosslinks for both structures, using Equation 4.5: where Z i is the number of statistical segments between network junctions in a sample crosslinked with initiator i, and Z Z BPO CHP p 2 BPO p 2 CHP p T 2 is the value of i T (4.5) T p T 2 for the same sample. Using the values collected in Table 4.8, the distance between junctions in a sample crosslinked with BPO seems four times larger than in a sample crosslinked with CHP. This difference in network structure may also find an explanation in the distribution of DVB in the EP(D)M particles, as observed in Figure 4.3. Indeed, in the sample crosslinked using CHP as initiator, the non-homogeneous distribution of DVB within the EP(D)M particles may lead to a higher number of DVB bridges between EP(D)M chains, and therefore to a shorter distance between crosslinks in the extracted material. Table 4.8: T 2 values at the high temperature plateau for the samples studied in this chapter. Initiator p T 2 ( s) BPO 1400 CHP Crosslinking initiated by a pulsed electron-beam Stability of the surfactants As reported in Chapter 3, an EP(D)M-based latex may be stabilized using two different surfactant systems, i.e. an equimolar combination of SDBS and Brij 700, or a mixture of SDBS and hexadecane as in miniemulsion.

84 About crosslinking of EP(D)M-based latexes 73 The influence of high energy radiations on the chemical stability of both types of surfactant systems was analyzed using MALDI-ToF-MS. The MALDI-ToF-MS spectra obtained from an SDBS solution before and after irradiation (not given here) revealed no major modification of the SDBS molecule. Therefore, a maximum dose of 280 kgy does not alter the structure of the surfactant. This result is more or less confirmed by a study carried out by Pusch 44, on a similar anionic surfactant, i.e. sodium dodecylsulfate (SDS). According to the work of Pusch, a pulsed electron-beam irradiation under a relatively low dose of 380 kgy had no effect on SDS, whereas a high dose of 1140 kgy led to the destruction of the molecule. However, as observed in the high molar mass region of the Brij 700 MALDI-ToF-MS spectrum, see Figure 4.10, the 280 kgy-irradiation of the nonionic stabilizer leads to the complete destruction of the molecule. As a consequence, the aqueous solution of Brij 700 becomes hazy upon irradiation. This loss of transparency is accompanied by the formation of small fragments, observed in the low molar mass region of the MALDI- ToF-MS spectrum depicted in Figure 4.11b. The MALDI-ToF-MS results in Figures 4.10 and 4.11 point to the degradation of Brij 700 upon electron-beam irradiation. This degradation leads to a loss of colloidal stability of the latex particles, and explains the coagulation observed during electron-beam irradiation of latexes stabilized with an equimolar mixture of SDBS and Brij 700. As a result, all latexes used in the present chapter were stabilized using the miniemulsion surfactant system, i.e. SDBS and hexadecane.

85 74 Chapter 4 % Intensity a % Intensity Mass (m/z) b- Figure 4.10: High molar mass region of MALDI-ToF-MS spectra obtained from an aqueous solution of Brij 700 before (a) and after (b) electron-beam irradiation a- % Intensity % Intensity b Mass (m/z) Figure 4.11: Low molar mass region of MALDI-ToF-MS spectra obtained from an aqueous solution of Brij 700 before (a) and after (b) electron-beam irradiation.

86 About crosslinking of EP(D)M-based latexes 75 Effects of additives Several latexes, consisting of EPM Lucant HC-20, Trilene 67 and occasionally a co-agent, were used for electron-beam-initiated crosslinking experiments, carried out in the batch reactor, see Figure 4.2a. As mentioned above, latexes stabilized with an equimolar mixture of an anionic surfactant (SDBS) and a nonionic surfactant (Brij 700) suffered from coagulation upon irradiation. On the other hand, no destabilization was observed for latexes prepared with the combination of SDBS and hexadecane, also referred to as a miniemulsion surfactant system in Chapter 3. However, the electronbeam irradiation did not lead to any crosslinking either. This lack of crosslinking may be due to the reaction occurring between radicals formed by the irradiation in the water phase 38,39 and the radicals created on the polymer chains. Fremy salt (FS) 49 was then introduced, in order to act as a water radical scavenger. The anion of FS is a nitroxy radical, usually used as a selective oxidizing agent for organic compounds 50. The remarkable stability of FS makes it an efficient radical scavenger. So far, FS has been mostly used to investigate fundamental reaction steps in emulsion polymerization in the presence of a water-soluble initiator for instance, as suggested by Lacik et al. 51 FS is highly water-soluble but almost insoluble in most common organic solvents 52. Therefore, the advantages of FS also include its high selectivity towards aqueous phase radicals. As a consequence, in their work, Lacik et al. 51 compared the radical-loss rate coefficients in the presence and absence of FS, to determine whether desorbed radicals in absence of FS undergo termination in the aqueous phase or re-entry into another particle. Finally, no interaction between FS and anionic surfactants, i.e. sodium dodecylsulfate (SDS), has been observed so far 52,53. Therefore, potassium ions, originating from FS, do not lead to a loss of colloidal stability of the SDS-stabilized latexes. As a result, it seems reasonable to assume that FS will not enter SDBS stabilized latex particles, and therefore will not hamper any crosslinking by trapping polymeric radicals.

87 76 Chapter 4 So, in the present chapter, FS is employed for the first time to enhance a reaction, i.e. the crosslinking of EP(D)M-based latexes, by preventing termination of polymeric radicals by OH and H radicals formed in the aqueous phase. The orange crystals of FS lead to a purple solution when dissolved in water. Murib et al. 54 described the decomposition of FS in water according to Equation (SO 3 ) 2 NO 7H 2 O 2 (SO 3 ) 2 NOH N 2 O 4SO 4 4H 3 O (4.6) During high energy electron-beam irradiation, water gives rise to H and OH radicals, which may react with FS according to Equations 4.7 and 4.8, as suggested by Murib et al. 54 N 2 O (SO + 3 ) 2 NO 2- OH Intermediates HNO 2 SO 4 2- H 3 O + (4.7) (SO + 3 ) 2 NO 2- H (SO 3 ) 2 NOH 2- (4.8) The above assumption, concerning the crosslinking hampered by aqueous radicals, was confirmed when latexes containing FS were used. As observed in Table 4.6 (page 55) and Figure 4.12, crosslinking of EP(D)M-based latexes also occurred to a small extent, i.e. 7.6 %, when no co-agent, i.e. divinylbenzene (DVB) or poly(1,2-butadiene) (Ricon 156), was employed.

88 About crosslinking of EP(D)M-based latexes Crosslinking efficiency (%) Irradiation dose (kgy) Figure 4.12: Evolution of the crosslinking efficiency during pulsed electron-beam irradiation of seed latexes EB1 ( ), EB2 or EB3 ( ) and EB4 ( ). Note that both types of co-agent, i.e. DVB in seed latex EB2 and 1,6-hexanediol diacrylate in seed latex EB3, lead to similar crosslinking efficiencies, in the order of 20 %. The highest efficiency of 40 % was obtained for seed latex EB4, containing both DVB and poly(1,2-butadiene). Therefore, as expected, crosslinking is enhanced by the presence of unsaturated carbon-carbon bonds. However, as demonstrated by the results in Figure 4.12, an irradiation dose above 150 kgy leads to a reduced crosslinking efficiency, probably due to polymer chain scission. 4.4 Conclusions Crosslinking of a low molecular weight EP(D)M latex usually requires the presence of a co-agent, e.g. divinylbenzene, 1,6-hexanediol diacrylate and/or poly(1,2-butadiene). In the case of crosslinking initiated by a peroxide, a highly reactive peroxide, e.g. di-t-butyl peroxyoxalate (DTBPO), is essential to achieve a crosslinking efficiency of 42.5 %. The use of a partially hydrophilic peroxide, e.g. cumene hydroperoxide (CHP) combined with a redox system, leads to a gel content of 15 to 18 %. Moreover, cryo-tem pictures

89 78 Chapter 4 of latex particles crosslinked with CHP revealed a non-homogeneous distribution of DVB, and therefore of crosslinked polymer over the latex particles. Such a distribution of DVB within EP(D)M particles may be due to reactions occurring near the surface of these particles when both CHP and the redox initiation system are used. In spite of a similar efficiency for benzoyl peroxide (BPO) and for CHP, i.e. 20 % in the presence of both poly(1,2-butadiene) and DVB, an improved distribution over the particle volume was observed when BPO was used as crosslinking initiator. This difference in distribution of DVB, in samples crosslinked with BPO or CHP as initiators, was confirmed by a 1 H and 13 C solid-state NMR analysis. Moreover, the solid-state NMR analysis also demonstrated that the mechanism involved in crosslinking of EP(D)M-based latexes, in the presence of a co-agent, includes the formation of DVB bridges between polymer chains. The size of these bridges depends on the nature of the initiator used. Crosslinking initiated by a pulsed electron-beam requires the presence of a special salt, i.e potassium nitrosodisulfonate also referred to as Fremy salt (FS). This salt acts as OH and H scavenger in the aqueous phase, therefore preventing termination of polymer radicals. The use of FS combined with a crosslinking co-agent leads to reasonable crosslinking efficiencies, in the range of 20 to 40 %. However, although MALDI-ToF-MS revealed no major modification in the structure of the anionic surfactant, i.e. SDBS, upon high energy radiation, it proved that the nonionic surfactant, i.e. Brij 700, may be dramatically damaged. Therefore, particular attention must be devoted to the choice of surfactant to stabilize the latex.

90 About crosslinking of EP(D)M-based latexes 79 References 1. Tornqvist, E. G. M. in "Polymer Chemistry of Synthetic Elastomers"; Kennedy, Joseph P. and Tornqvist, Erik G. M. Eds., New York: Wiley-Interscience, 1968, Vol. 1, Chapter Stern, H. J. in "Rubber Technology and Manufacture"; Blow, C. M. Ed., London: Butterworths, 1971, Chapter Akiba, M.; Hashim, A. S. Prog. Polym. Sci. 1997, 22, Elliott, D. J. in "Developments in Rubber Technology"; Whelan, A. and Lee, K. S. Eds., London: Applied Science, 1979, Vol. 1, Chapter Krupa, I.; Luyt, A. S. S. Afr. J. Chem. 2002, 55, Loan, L. D. J. Polym. Sci. Part A: Polym. Chem. 1964, 2, Heiner, J.; Stenberg, B.; Persson, M. Polym. Test. 2003, 22, Dluzneski, P. R. Rubber World 2001, 224, Gonzalez, L.; Rodriguez, A.; Marcos, A.; Chamorro, C. Rubber Chem. Technol. 1996, 69, Von Raven, A.; Heusinger, H. J. Polym. Sci. Polym. Chem. Ed. 1974, 12, Abdel-Aziz, M. M.; Basfar, A. A. Polym. Test. 2000, 19, Rybak, A. S.; Grinyuk, E. V.; Klimets, T. G.; Polikarpov, A. P.; Brazhnikov, M. M.; Shadyro, O. I.; Krul', L. P. High Energ. Chem. 2003, 37, Zaharescu, T.; Setnescu, R.; Jipa, S.; Setnescu, T. J. Appl. Polym. Sci. 2000, 77, Singh, A.; Silverman, J. Progress in Polymer Processing, Vol. 3: Radiation Processing of Polymers; New York: Oxford University Press, International Standard Electric Corp., Neth. Appl , Pearson, D. S.; Skutnik, B. J.; Boehm, G. G. A. J. Polym. Sci. Polym. Phys. Ed. 1974, 12, Spenadel, L. Radiat. Phys. Chem. 1979, 14, Harnischfeger, P.; Kinzel, P.; Jungnickel, B. J. Angew. Makromol. Chem. 1990, 175, 157.

91 80 Chapter Bhowmick, A. K.; Majumder, P. S.; Banik, I. Macromol. Symp. 1999, Perraud, S.; Vallat, M. F.; Kuczynski, J. Macromol. Mater. Eng. 2003, 288, Chirinos, H.; Yoshii, F.; Makuuchi, K.; Lugao, A. Nucl. Instr. and Meth. B 2003, 208, Keller, R. C. Rubber Chem. Technol. 1988, 61, Forsythe, J. S.; Hill, D. J. T. Prog. Polym. Sci. 2000, 25, Tian, J.; Xue, Q. J. Appl. Polym. Sci. 1998, 69, Ratnam, C. T.; Zaman, K. Polym. Degrad. Stabil. 1998, 61, Palys, L. H.; Callais, P. A. Rubber World 2003, 229, Yanez-Flores, I. G.; Ibarra-Gomez, R.; Rodriguez-Fernandez, O. S.; Gilbert, M. Eur. Polym. J. 2000, 36, Bouvier-Fontes, L.; Pirri, R.; Arzamendi, G.; Asua, J. M.; Leiza, J. R. Macromol. Symp. 2004, 206, Cai, J. J.; Salovey, R. J. Mater. Sci. 2001, 36, Burford, R. P.; Vo, C. D. J. Appl. Polym. Sci. 1999, 74, Dullens, R. P. A.; Claesson, E. M.; Kegel, W. K. Langmuir 2004, 20, Pathmamanoharan, C.; Groot, K.; Dhont, J. K. G. Colloid Polym. Sci. 1997, 275, Falk, J. C.; Borg-Warner Corp., U.S , Cheong, I. W.; Fellows, C. M.; Gilbert, R. G. Polymer 2004, 45, Hiromoto, Y.; Matsuda, K.; Goto, H.; Ube Cycon Ltd., U.S , Tuncel, A.; Tuncel, M.; Salih, B. J. Appl. Polym. Sci. 1999, 71, Laizier, J.; Noel, M. T.; Veber, A.; Pordes, M. Large Radiat. Sources Ind. Processes, Proc. Symp. Util. Large Radiat. Sources Accel. Ind. Process. 1969, Sabharwal, S.; Das, T. N.; Chaudhari, C. V.; Bhardwaj, Y. K.; Majali, A. B. Radiat. Phys. Chem. 1998, 51, Allen, A. O. The Radiation Chemistry of Water and Aqueous Solutions; Princeton, N. J.: D. van Nostrand, 1961.

92 About crosslinking of EP(D)M-based latexes Bartlett, P. D.; Benzing, E. P.; Pincock, R. E. J. Am. Chem. Soc. 1960, 82, Van Duijneveldt, W.; Botman, J. I. M.; Timmermans, C. J.; De Leeuw, R. W. Nucl. Instr. and Meth. B 1993, B79, Derksen, A. T. A. M.; Timmermans, C. J.; Wintraeken, Y. J. E.; Botman, J. I. M.; Fiedler, H.; Hagedoorn, H. L. Proceedings of the 5th European Particle Accelerator Conference 1996, 3, Botman, J. I. M.; Derksen, A. T. A. M.; van Herk, A. M.; Jung, M.; Kuchta, F. D.; Manders, L. G.; Timmermans, C. J.; de Voigt, M. J. A. Nucl. Instr. and Meth. B 1998, 139, Pusch, J. PhD Thesis, University of Eindhoven, Vanhaecht, B.; Willem, R.; Biesemans, M.; Goderis, B.; Basiura, M.; Magusin, P. C. M. M.; Dolbnya, I.; Koning, C. E. Macromolecules 2004, 37, Litvinov, V. M.; Dias, A. A. Macromolecules 2001, 34, Litvinov, V. M. Macromolecules 2001, 34, Fry, C. G.; Lind, A. C. Macromolecules 1988, 21, Verdurmen, E. M.; Geurts, J. M.; German, A. L. Macromol. Chem. Phys. 1994, 195, Rozantsev, E. G. Free Nitroxyl Radicals; New York: Plenum Press, Lacik, I.; Casey, B. S.; Sangster, D. F.; Gilbert, R. G.; Napper, D. H. Macromolecules 1992, 25, Bratt, P. J.; Choudhury, H.; Chowdhury, P. B.; Gillies, D. G.; Krebber, A. M. L.; Sutcliffe, L. H. J. Chem. Soc. Faraday T. 1990, 86, Kotake, Y.; Janzen, E. G. Can. J. Chem. 1988, 66, Murib, J. H.; Ritter, D. M. J. Am. Chem. Soc. 1952, 74, 3394.

93

94 About morphology in EP(D)M-based latexes Abstract For the present chapter, EP(D)M-based polymers were dispersed in water using an Ultra-Turrax and a homogenizer, operating at a pressure of 1100 bars and with a shear rate of approximately s -1. The prepared latexes were then used, either without further treatment or after crosslinking, as seed latexes in the seeded emulsion polymerization of methyl methacrylate (MMA). Polymerizations were carried out using batch and semi-continuous processes. The free radical seeded emulsion polymerization of MMA was investigated in the presence of an oil-soluble initiator, i.e. cumene hydroperoxide (CHP), combined with a redox system, i.e. sodium formaldehyde sulfoxylate hydrate (SFS), disodium salt of ethylenediamine tetra-acetic acid (EDTA), iron (II) sulfate heptahydrate (FeSO 4 ). This initiation system promotes polymerization of MMA near the surface of the seed particles, partially suppressing homogeneous secondary nucleation and polymerization in the aqueous phase. Kinetic and thermodynamic considerations have been used to predict the particle morphology. The monomer type, the monomer-to-rubber ratio, the monomer feed type, and crosslinking of the seed latex were investigated, to optimize the polymerization kinetics and the properties of the resulting dispersions. The particle morphology was determined by cryo-transmission Electron Microscopy (cryo-tem). Monomer-flooded conditions led to the formation of inverted core-shell particles, whereas starved-fed MMA or MMA/styrene mixtures gave rise to partially engulfed structures, i.e. snowman-like. Finally, crosslinking of the EP(D)M seed particles was found to be required to provide the desired core-shell structures. A thorough analysis of these micrographs led to the conclusion that the core represents approximately ¾ of the volume of the obtained particles. This chapter is part of: Tillier, D. L.; Meuldijk, J.; Höhne, G. W. H.; Frederik, P. M.; Regev, O.; Koning, C. E. Manuscript submitted for publication.

95 84 Chapter Introduction The performance of composite latex particles in an end-use-application mostly depends on their morphology, and therefore on the process parameters used during emulsion polymerization, e.g. type and amount of initiator, second-stage monomer-topolymer seed ratio. The core-shell type of particles is widely employed in applications concerning impact modification 1-4 of plastics and coatings. Methods of preparation of core-shell polymers include controlled surface precipitation of the coating materials on a core material, or direct surface reactions. One of the most commonly used precipitation methods is controlled heterocoagulation. Controlled heterocoagulation has been applied, for instance, by Okubo et al. 5 to produce soft core-hard shell composite polymer particles. As depicted in Figure 5.1, see Li et al. 6, dispersions of oppositely charged large and small particles are first synthesized. The small particles should exhibit a glass transition temperature lower than the glass transition or melting temperature of the large core particles. When both dispersions are mixed, heterocoagulation occurs driven by electrostatic attraction. This heterocoagulation results in the formation of a monolayer of small beads on the surface of large cores. After controlled heat processing, a homogeneous shell is obtained around each core particle Figure 5.1: Preparation of core-shell particles by heterocoagulation (from ref. 6). However, traditionally, polymer-based core-shell particles are obtained by interfacial polymerization of the shell-forming monomer onto the core-forming polymer. This process is generally referred to as two-stage or seeded emulsion polymerization. The two-step polymer latex is produced by free radical polymerization of a monomer onto a

96 About morphology in EP(D)M-based latexes 85 seed latex. The use of an oil-soluble initiator or a water-soluble initiator will result in grafting due to the formation of radicals on the seed polymer backbone. However, the presence of monomer droplets or micelles may lead to side reactions, e.g. polymerization in secondary nucleated particles. Secondary nucleation not only affects the colloidal stability of the obtained latex product, but the grafting efficiency may also decrease tremendously. Moreover, seeded emulsion polymerizations lead to many different heterogeneous structures. Therefore, a number of parameters including monomer-to-core polymer ratio, polarities of core and shell polymers, and operational details of polymerization process, have to be taken into account to obtain the desired composite latex particles. The work presented in this chapter is based on the use of an artificial latex, containing ethylene-propylene(-diene) copolymers (EP(D)M), as seed latex in the two-stage emulsion polymerization of methyl methacrylate (MMA). The morphology of the obtained particles will be discussed, using kinetic and thermodynamic considerations for morphology predictions. The mechanism of grafting will be emphasized, taking into account the occurrence of secondary nucleation and its effect on grafting efficiency. 5.2 Experimental section Chemicals Methyl methacrylate (MMA, Aldrich, 99 %) and styrene (Aldrich, 99 %) were purified by passing through a column packed with basic aluminum oxide (Aldrich, Brockmann I, standard grade, 150 mesh, 58 Å). Sodium dodecyl benzene sulfonate (SDBS, Fluka, technical grade, 80 %), hexadecane (HD, Aldrich, 99 %), pentane (Biosolve, AR), divinylbenzene (DVB, Aldrich, technical grade, 80 %), poly(1,2- butadiene) (Ricon 156, Cray Valley), petroleum ether (Biosolve, boiling point ranging from 60 to 80 C) and acetonitrile (Biosolve, HPLC-S, %) were used as received.

97 86 Chapter 5 All other reagents were also used as received, viz. cumene hydroperoxide (CHP, Luperox CU90, Aldrich, 88 %), benzoyl peroxide (BPO, Fluka, 75 %), sodium formaldehyde sulfoxylate hydrate (SFS, Fluka, +98 %), disodium salt of ethylenediamine tetra-acetic acid (EDTA, Aldrich, +99 %), iron (II) sulfate heptahydrate (FeSO 4, Aldrich, +99 %). Deionized water was used in all recipes. Relevant details of the polymers used in this study are collected in Table 5.1. Table 5.1: Characteristics of the polymers used. Polymer Supplier Type of diene M w (g mol -1 ) Ethylene (mol%) Propylene (mol%) Diene (mol%) Lucant HC-20 (EPM) Mitsui Chemicals, Inc Trilene 67 (EPDM) Crompton Corp. Ethylidene norbornene Preparation of the seed latex The preparation of the seed latex has been extensively described in Chapter 3. Recipes used in the present work are collected in Table 5.2. In a first step, SDBS was dissolved in water. A homogeneous mixture of EPM and EPDM was prepared with the help of pentane. After pentane evaporation, hexadecane, DVB and/or poly(1,2-butadiene) (Ricon 156) were added to the EPM-EPDM mixture. In a second step, the organic phase was brought into the SDBS-containing aqueous phase. The resulting blend was then stirred for one minute with a rotor-stator Ultra-Turrax T25 Basic at rpm. The volume average diameter of the polymer particles was reduced from 20 m to 360 nm by processing the emulsion product of the Ultra-Turrax in a Niro-Soavi Lab homogenizer Panda 2K, operating for 45 minutes at 1100 bars and with a shear rate of approximately s -1. Latex S1 was then used without further treatment as seed latex for a seeded emulsion polymerization of MMA. Two homogenizer products were submitted to a crosslinking

98 About morphology in EP(D)M-based latexes 87 reaction, as described in Chapter 4, prior to the grafting experiment, giving rise to latexes S2 and S3. Crosslinking was performed using benzoyl peroxide (BPO) as initiator. Both DVB and Ricon 156 were used as crosslinking promoting co-agents in S2, whereas S3 was obtained in the presence of DVB only. Table 5.2: Recipe for seed latexes. Seed Amount of product (g) latex No. Lucant HC-20 Trilene 67 HD i DVB Ricon 156 Water SDBS i S S S i In water: [SDBS] mol dm -3 and [HD] mol dm Seeded emulsion polymerization of MMA onto EP(D)M The graft polymerizations were carried out in a 300 ml jacketed-reactor, equipped with a condenser and a downflow 45 pitched four-blade impeller. Two types of monomer addition were used, i.e. flooded and starved-fed, as indicated in Table 5.3. When monomer-flooded conditions were applied, both the EP(D)M seed latex and the monomer were charged into the reaction vessel. After swelling the EP(D)M particles with MMA overnight, the oxygen was removed by purging argon through the mixture for at least 30 minutes. An aqueous solution of additives, i.e. SFS, EDTA and FeSO 4, also referred to as redox system, was charged into the reactor, followed by the addition of the initiator, i.e. cumene hydroperoxide (CHP). In case of starved-feed conditions, the EP(D)M seed latex was charged into the reaction vessel and purged with argon for at least 30 minutes to remove the oxygen. MMA and CHP were mixed prior to the addition, to ensure the presence of radicals in the vessel during the complete addition of monomer. The monomer/chp mixture was

99 88 Chapter 5 continuously added to the reactor at a rate of 0.05 ml min -1, using an automatic burette (Metrohm, type 665 Multi Dosimat, refillable 10 ml-burette). The polymerization reactions were performed at 50 C, at a stirring speed of 300 rpm, for at least 5 h. The post-treatments included coagulation of the latex using freeze-thaw cycles, and washing the coagulated product with deionized water. The gross polymers were recovered and dried to a constant weight in a vacuum oven at 60 C. Table 5.3: Recipes for seeded emulsion polymerizations (SEP). SEP No. No i. Seed latex Amount (g) Amount of product (g) CHP SFS EDTA FeSO 4 MMA Styrene Monomerto-rubber ratio (g/g) t r iv (h) SEP1 S SEP2 S SEP3 ii S SEP4 iii S SEP5 S SEP6 S i See Table 5.2. ii In case of SEP3, the monomer-to-ep(d)m ratio was 1.3 instead of 0.8. Therefore, 25 ml of an aqueous solution of SDBS ([SDBS] mol dm -3 ) was regularly added during the polymerization to maintain the colloidal stability of the latex. iii SEP4 was carried out using monomer-flooded conditions instead of starved-feed conditions, as used in all the other cases. iv t r : reaction time.

100 About morphology in EP(D)M-based latexes Characterization Particle size distribution A Coulter LS230 particle sizer was used to verify the colloidal stability of latexes after grafting. This analyzer uses the principles of light scattering, based on both Fraunhofer and Mie theories, to determine particle size distributions. The lowest size detectable being 40 nm, most of the particles formed by secondary nucleation and observed by cryo-tem could not be identified. No significant destabilization occurred during grafting of PMMA onto the EP(D)Mbased particles. However, note that a monomer-to-rubber ratio larger than 0.8 required the addition of an aqueous solution of SDBS, to ensure the colloidal stability of the latex particles, as mentioned in Table 5.3. Grafting efficiency The gross polymers were separated into graft copolymers, free EP(D)M, and free PMMA by Soxhlet extractions. Petroleum ether, with a boiling point ranging from 60 to 80 C, and acetonitrile were used for at least 10 h to extract the free rubber and the free PMMA, respectively. The grafting efficiency (GE) was calculated with: weight of MMA grafted GE (5.1) weight of MMA polymerized Composition of the extracted materials The materials soluble in acetonitrile were analyzed, after drying, using 1 H nuclear magnetic resonance (NMR), on a Varian-300 spectrometer at 25 C, using TMS as internal standard and CDCl 3 as solvent.

101 90 Chapter 5 Thermal properties Thermal characterization of dried latexes was performed by Temperature Modulated- Differential Scanning Calorimetry (TM-DSC). Analyses were carried out on a modified Perkin-Elmer DSC-7 differential scanning calorimeter, equipped with a function generator for sinusoidal temperature modulation. The measurements were carried out in a nitrogen atmosphere to prevent degradation of the polymer samples. Standard aluminum pans of similar predetermined mass were used for all measurements. The pans masses on the reference side and sample side were balanced to give a zero signal for the baseline. Every sample typically consisted of approximately 2 mg of polymer. The experiments were carried out in heating cooling mode, using a temperature amplitude of 0.3 K and a frequency of 25 and 12.5 mhz, i.e. 40 and 80 s period, respectively. The equipment was operated at temperatures ranging from -100 C to 130 or 150 C, and with an underlying heating rate 0 of 2 K min -1. The measured heat flow rate consists of two parts 7 : a non-periodic underlying part u (t) and a periodic part per. (T,t). A so-called gliding integration of the signal over one period provides the underlying part u (t). Subtracting u (t) from the total measured signal yields the periodic part per. (T,t). The specific heat capacity c p (magnitude and phase shift) is then calculated using a mathematical procedure described in the literature 7,8. Glass transition temperatures and c p were determined using conventional methods 9. Minimum film formation temperature (MFFT) 200 m-thick films were drawn from various latexes on an MFFT Bar from Sheen Instruments. The minimum film formation temperature corresponds to the temperature at which a white, powdery, cracked film becomes clear and transparent.

102 About morphology in EP(D)M-based latexes 91 Molecular weight and molecular weight distribution Molecular weight (MW) and molecular weight distribution (MWD) were determined at ambient temperature, using a Waters Size Exclusion Chromatograph (SEC), equipped with a Waters model 510 pump, a Waters 410 differential refractometer operating at 40 C and a Waters model 486 UV detector operating at 254 nm. Samples were injected using a Waters WISP 712 autoinjector (50 L injection volume). The columns consisted of a PL gel guard (5 m particles) mm column, followed by two PL gel mixed- C or mixed-d (5 m particles) mm columns at 40 C in series. The eluent was THF, and the elution volumetric flow rate was maintained at 1 ml min -1. Calibration was carried out using narrow MWD polystyrene standards ranging from 580 to g mol -1. Particles morphology The morphology of the particles was examined using cryo-transmission Electron Microscopy (cryo-tem). For the present work, the use of cryo-tem was required because of the low glass transition temperature of the EP(D)M phase. To perform cryo-tem analysis, each latex sample was applied onto a microscopy grid. The thin aqueous film obtained was then vitrified in liquid ethane before being transferred to a Philips CM12 microscope, for examination at liquid nitrogen temperature. Dehydration and major reorganization of the latex particles were prevented by the low temperature. Surface and interfacial tension measurements Surface properties of both EP(D)M and PMMA as well as of the water phase are required for the prediction of particle morphology, as described in the next sections. Note that pure deionized water and surfactant-containing water exhibit different surface properties. Therefore, all measurements including water were carried out using an

103 92 Chapter 5 aqueous solution of SDBS, in which the SDBS concentration was similar to the one usually used in latexes, i.e mol dm -3. The interfacial tension between EP(D)M and SDBS-containing water ( EP ( D) M water / SDBS ) was determined with a Krüss G10 goniometer, using the pendant drop method. Other interfacial tensions, i.e. between PMMA and SDBS-containing water ), as well as between EP(D)M and PMMA ( EP( D) M PMMA ), were ( PMMA water / SDBS calculated using the modified Young s equation 10, also called geometric-mean equation 10,11 : (5.2) d d 1/ 2 p p 1/ ( 1 2 ) 2( 1 2 ) where 1 is the surface tension of phase 1, 2 is the surface tension of phase 2, and 12 d is the interfacial tension between phase 1 and phase 2. In the above equation, 1 and d 2 are the dispersive components of the surface tensions of materials 1 and 2, while p and 2 are the respective polar components. As suggested by Fowkes 12, these two components are related as (5.3) i d i p i p 1 Contact angles of water and methylene iodide were measured on EP(D)M and PMMA spin-coated films, using the Krüss G10 goniometer. Surface free energies of EP(D)M and PMMA were then evaluated using the Owens-Wendt method 13. Surface tension of SDBS-containing water was determined on a Krüss K100 tensiometer, equipped with a platinum plate, using the Wilhelmy plate method. Both polar and dispersive components were evaluated, using poly(tetrafluoroethylene) (PTFE) as standard reference surface, as reported in a Krüss technical note 14.

104 About morphology in EP(D)M-based latexes Results and discussion Grafting mechanism One of the aims of this work is to investigate the influence of recipe and process parameters, e.g. monomer type, monomer-to-rubber ratio or monomer feed type, on the outcome of graft polymerization. The effect of EP(D)M crosslinking on grafting was also investigated. Grafting of MMA onto EP(D)M-based latex particles was carried out using cumene hydroperoxide (CHP), depicted in Figure 5.2, combined with sodium formaldehyde sulfoxylate hydrate (SFS, H 2 O) / EDTA-chelated Fe 2+ as redox initiation system. CH 3 C O CH 3 OH Figure 5.2: Structure of cumene hydroperoxide. The polymerization occurs either on the EP(D)M backbone or on the monomer to be grafted. The radical formation on the monomer may result in homopolymerization followed by secondary homogeneous particle nucleation. Redox initiation systems have been extensively used in efficient emulsion graft polymerization Advantages of redox initiators compared to thermally dissociative initiators include a higher radical production rate, thus allowing a lower polymerization temperature. A mechanism of grafting has been reported by Arayapranee et al. 17 for the graft polymerization of the monomer pair styrene/mma onto natural rubber. Adapting their approach to our system, a mechanism for the graft polymerization of MMA onto EP(D)M is suggested in the present section.

105 94 Chapter 5 Alkoxy radicals (RO ) are produced by the decomposition of cumene hydroperoxide in a dilute aqueous solution, catalyzed by the redox system, as depicted in Scheme Ethylenediamine tetra-acetic acid (EDTA) is used as chelating agent and sodium formaldehyde sulfoxylate (SFS) as reducing agent, to reduce Fe 3+ into Fe 2+ during the redox cycle. Most of the free radicals are then produced at the monomer-swollen particle/water interface, since the peroxide is soluble in the organic phase, whereas the iron/edta complex is water-soluble. Therefore, the surface of the particle becomes the locus of polymerization. SFS -EDTA-Fe 2+ + CH 3 CH 3 C O OH C O OH -Fe 2+ EDTA + SFS CH 3 CH 3 CH 3 HO Fe 3+ EDTA + SFS O C + HO Fe 2+ EDTA + SFS CH 3 H ( CH 2 MMA CH 2 ) ( CH 2 C ) ( n m CH 2 CH 2 ) ( CH n 2 C ) CH 3 CH 3 m CH 3 CH 2 C C O O CH 3 Scheme 5.1: Schematic representation of the grafting reaction of MMA onto EP(D)M in the presence of an interfacial redox initiator system (from ref.18). For convenience, here, EPM is used. As suggested by Scheme 5.1, the alkoxy radical may react with the EP(D)M backbone, leading to a macroradical that will initiate grafting. However, the same alkoxy radical may also initiate MMA to form free PMMA radicals. These PMMA macroradicals may either recombine with EP(D)M radicals to terminate, leading to graft copolymers, or transfer to EP(D)M or monomer leading to free PMMA.

106 About morphology in EP(D)M-based latexes 95 The reaction scheme proposed for the free-radical graft copolymerization of MMA onto EP(D)M is given in Scheme 5.2, where RO is an alkoxy radical, M is MMA, M n is a PMMA radical, EPM-H is the rubber in which H corresponds to a hydrogen atom of a propylene unit, EPM is the rubber radical, EPM-M n is the growing graft polymer radical chain, and EPM-M n is the graft copolymer. Initiation Reaction with monomer: Reaction with EPM: Reinitiation: Propagation Propagation of free polymerization: RO + M RO + EPM-H EPM + M M 1 + M M n + M M 1 ROH + EPM EPM-M 1 M 2 M n+1 Propagation of graft polymerization: Chain transfer EPM-M 1 + M EPM-M n + M EPM-M 2 EPM-M n+1 Transfer to monomer: M n + M EPM-M n + M M n + M 1 EPM-M n + M 1 Transfer to EPM: M n + EPM-H M n H + EPM EPM-M n + EPM-H EPM-M n H + EPM Termination by disproportionation M n + M m M n + M m EPM-M n + EPM-M m EPM-M n + EPM-M m EPM-M n + M m EPM-M n + M m Scheme 5.2: Proposed mechanism for the free radical graft copolymerization of MMA (=M) onto EP(D)M. For convenience, here, EPM is used.

107 96 Chapter 5 In addition to the mentioned termination reactions, free polymer chain radicals M n will also undergo termination by combination. Finally, two rubber radicals EPM may also recombine leading to partial crosslinking of the starting seed latex particles. When homopolymerization of MMA occurs during the seeded emulsion polymerization of MMA onto EP(D)M, the system is described by homogeneous nucleation 19 : a radical generated in the aqueous phase will either enter a latex particle and undergo termination, or grow in the aqueous phase until precipitation occurs and a new particle is formed. This phenomenon may be explained by MMA s solubility in the aqueous phase (1.56 wt%) 20, which is sufficient to induce homogeneous secondary nucleation. Therefore, once an oligomeric radical is created in the aqueous phase, secondary nucleation and the monomer transport to those secondary nucleated PMMA particles become significant. Note that a macroradical may also react by direct addition to the double bonds present in the diene of the EP(D)M or in other additives, e.g. DVB or Ricon 156, leading to additional grafting. However, taking into account the low diene content of the EPDM and the high EPM-to-EPDM weight ratio, the formation of graft copolymer by direct addition to double bonds is assumed to be negligible. Moreover, Tsavalas et al. 21 reported that, due to steric hindrance at the methacrylate reactive center, grafting of methacrylates onto resins mostly occurs via allylic hydrogen abstraction instead of direct addition to double bonds. As exemplified by Scheme 5.2, this transfer to the resin not only leads to the formation of a relatively stable and unreactive radical on the resin backbone, but also to the termination of a propagating radical, thus to the formation of a free PMMA chain. Further evidence of allylic hydrogen abstraction instead of direct addition to double bonds as main grafting mechanism has been provided by Merkel et al. 22 for the seeded emulsion polymerization of MMA onto polybutadiene. So, in the present work, a step forward would be made if the homopolymerization of MMA via secondary nucleation could be minimized, if not avoided. As a consequence,

108 About morphology in EP(D)M-based latexes 97 conditions developed in Chapter 2 were applied to the present seeded emulsion polymerizations, i.e.: - The monomer was starved-fed in most experiments to minimize propagation in the aqueous phase. - The use of both cumene hydroperoxide and a redox initiation system, i.e. SFS, EDTA and FeSO 4, is expected to favor reactions inside the EP(D)M particles or near their surface. - The surfactant concentration was kept below its critical micelle concentration (CMC) in the latex to minimize homogeneous nucleation Efficiency of the grafting reaction Grafting efficiency The grafting efficiency of the various seeded emulsion polymerizations of MMA onto EP(D)M particles could not be determined using classical soxhlet extractions. Indeed, although acetonitrile leads to the extraction of pure PMMA obtained by secondary nucleation, this solvent also leads to the extraction of PMMA-g-EP(D)M copolymers containing divinylbenzene and poly(1,2-butadiene) (Ricon 156), as indicated by the 1 H NMR spectrum of an extracted product, see Figure 5.3.

109 98 Chapter 5 DVB Poly(1,2-butadiene) ppm (f1) ppm 8.0 (f1) Figure 5.3: 1 H NMR spectrum of PMMA extracted with acetonitrile. DVB and poly(1,2-butadiene) are also present in the extracted phase. Therefore, soxhlet extraction is not the right method to quantify the grafting efficiency. Thermal properties Since soxhlet extractions were not successful, Temperature Modulated-Differential Scanning Calorimetry (TM-DSC) was used to acquire more information about the grafted material, obtained by seeded emulsion polymerization of MMA onto the EP(D)M particles. Figure 5.4 shows an example of TM-DSC result, with at least two glass transitions, i.e. a sharp one at -80 C and a broad one at +90 C. A third very weak glass transition seems to appear at +26 C. As observed in this thermogram of SEP5 containing coreshell particles, the step-like change of the real part of the complex heat capacity c p ( ), in the glass transition region, is always coupled with a maximum of the imaginary part, i.e. the phase max. The maximum in the phase signal is usually more visible than the step change in the c p curve (or in conventional thermograms). This is one advantage of the TM-DSC method against conventional DSC.

110 About morphology in EP(D)M-based latexes 99 Following these observations, the glass transition at -80 C clearly indicates the presence of rubber domains. The broader glass transition in the range of 50 C to 120 C may be due to the overlapping of two consecutive glass transitions. Indeed, one T g at 106 C characterises PMMA. The presence of another T g of lower value may be explained by PMMA segments grafted onto EP(D)M chains. This may also explain the third glass transition observed in the region of 25 C. However, the broadness of a glass transition region may also find explanation in the existence of polymer chains exhibiting different molar masses. It is indeed recognized that small molecules may act as solvent or plasticizer, causing a decrease of the glass transition temperature. A distribution of molar mass or a distribution of various domains may lead to a continuous distribution of T g s, and therefore to a broad glass transition region as observed in Figure 5.4. For the present research, samples were investigated using size exclusion chromatography. A large polydispersity index of 2.6 confirmed the presence of small and long chains in the studied system. Moreover, the contribution of EP(D)M in the grafting process may be determined by a quantitative analysis of the thermogram. The TM-DSC analysis of pure EP(D)M showed a variation of c p, namely a c p,th, of 0.53 J g -1 K -1, as depicted in Figure 5.5. The T g of -80 C, observed in Figure 5.4 is characteristic of the fraction of EP(D)M that is not involved in the grafting process. The c p value corresponding to this transition, i.e. the average c p calculated from five different TM-DSC curves, is 0.22 J g -1 K -1. This c p step is less significant than the one observed for pure EPDM, since the fraction of EP(D)M involved in grafting does not contribute to this glass transition anymore. Equation 5.4 can then be used to determine the weight fraction of EP(D)M phase involved in the grafting process, i.e. approximately 22 %: c p % Grafted EP(D)M = (1 - ) 100 f c EP( D) M p, th (5.4) where f EP(D)M = 0.53 represents the weight fraction of EP(D)M in the sample, determined with the recipe, c p = 0.22 the measured step height, and c p,th = 0.53 the step height of pure EP(D)M.

111 100 Chapter c p (J.g -1.K -1 ) (rad) Temperature ( C) 1.20 Figure 5.4: Specific heat capacity (c p ) and phase angle ( ) of SEP5, after drying, obtained by TM-DSC c p (J.g -1.K -1 ) (rad) Temperature ( C) Figure 5.5: Specific heat capacity (c p ) and phase angle ( ) of the pure EPM/EPDM mixture used in SEP2, obtained by TM-DSC.

112 About morphology in EP(D)M-based latexes Morphology of the EP(D)M-g-PMMA particles Morphology predictions Two-stage emulsion polymerization produces heterogeneous structures such as coreshell 22,23, inverted core-shell 24,25 and phase-separated structures. Phase-separated morphologies include sandwich-like 26, snowman-like 27, salami-like 28, and raspberry-like 29 particles, as exemplified in Figure 5.6. Sandwich-like Snowman-like Salami-like Raspberry-like Figure 5.6: Possible phase-separated morphologies, where represents the first-stage polymer, and the second-stage polymer. The particle morphology may be affected by many polymerization parameters, e.g. initiator, surfactant, monomer-to-seed polymer ratio, as mentioned earlier. Therefore, predicting particle morphology using kinetic and thermodynamic considerations may help understanding the structure of the final composite latex particles. Kinetic considerations As reported by Chern et al , free radicals produced during an emulsion polymerization are partially hydrophilic. As a consequence, free radicals are preferably located near the particle surface, leading to a non-uniform distribution of free radicals in the latex particles. A core-shell structure may then be the result of such non-uniform distribution of radicals. The viscosity of the polymerization loci represents another key parameter in the determination of the final morphology of latex particles.

113 102 Chapter 5 As mentioned by van Zyl 33 and González-Ortiz et al , during a seeded emulsion polymerization, polymer chains are formed at various positions in the seed particles. However, in the case of a low local viscosity, incompatibility of the newly formed polymer and the seed polymer often leads to phase separation. The clusters obtained may then migrate towards the equilibrium morphology to minimize the Gibbs free energy. During the clusters migration, the size of the clusters may increase either by polymerization of monomer inside the clusters, or by diffusion of polymer chains into the clusters, or by coalescence with other clusters. Both diffusion of polymer chains and coalescence with other clusters are influenced by the particle viscosity. On the other hand, a high local viscosity prevents the polymer chains from diffusing through the seed particles, minimizing the occurrence of phase separation. The large influence of local viscosity on the particle morphology was further emphasized by Mills et al. in , when they studied the dependence of the particle structure on various parameters including rates of diffusion, propagation, termination, entry, transfer, and exit. Mills et al. reported significant inhomogeneities in latex particles at high monomer conversion, particularly for large particles. These heterogeneities may be explained by a slow diffusion of free radicals through the core material, due to a high viscosity of the polymerization loci, thereby leading to the formation of core-shell structures. In the present work, a combination of cumene hydroperoxide and a redox system has been used as initiator. The hydrophilicity of the redox part should prevent the peroxide from being buried inside EP(D)M seed particles, therefore favoring core-shell structures. Thermodynamic considerations A method for the thermodynamic approach of morphology prediction was first reported in 1970 by Torza and Mason 38, who considered two immiscible liquids dispersed in a third immiscible liquid, i.e. water. Since then, numerous researchers tried to extend this concept to various encapsulating systems. For instance, Sundberg et al. 39 studied the formation of a particle comprising a

114 About morphology in EP(D)M-based latexes 103 large size oil droplet encapsulated by a polymer. According to Sundberg, several morphologies (hemispherical, sandwich, multiple lobes) can coexist within a single (sub)micron dispersion, suggesting that they may simply represent different states of phase separation and are only metastable morphologies. In that case, the morphology is therefore governed by the diffusion of polymer chains through the more or less viscous core-forming material. Chen et al. 40 described the morphology changes occurring during the batch seeded emulsion polymerization of MMA onto polystyrene, in terms of free energy changes. The thermodynamically preferred morphology is the one that has the lowest interfacial free energy. Finally, Waters 41,42 investigated the evolution of interfacial energy from separated structures to fully engulfed particles and was able to determine the degree of engulfment corresponding to minimum interfacial energy. In the present work, the morphology of the particles produced during the seeded emulsion polymerization of MMA onto EP(D)M was predicted using both theories of Torza/Mason 38 and Waters 41,42. Torza and Mason 38 defined the spreading coefficients, S i, of a three-phase system as: S i ) (5.5) jk ( ij ik where is the interfacial tension and i, j, and k are the three phases considered. The different possible structures and their various spreading coefficients are depicted in Figure 5.7. By convention 38, phase 1 must be the phase for which 12 23, so that S 1 < 0. Then, complete engulfment, i.e. core-shell, occurs when S 2 < 0 and S 3 > 0. However if S 2 < 0 and S 3 < 0, the engulfment is only partial and leads to hemispherical, also called acorn or snowman morphologies. Finally, when S 2 > 0 and S 3 < 0, separated structures are preferred.

115 104 Chapter 5 FULLY ENGULFED ENGULFMENT NO ENGULFMENT CORE-SHELL S 1 < 0 S 2 < 0 S 3 > 0 INVERTED CORE-SHELL S 1 > 0 S 2 < 0 S 3 < 0 INTERMEDIATE STRUCTURE S 1 < 0 S 2 < 0 S 3 < 0 SEPARATED S 1 < 0 S 2 > 0 S 3 < 0 Figure 5.7: Engulfment and spreading coefficients. For the investigated system, the continuous phase (water, w), EP(D)M and PMMA were designated as phase 2, 1 and 3, respectively. The reason for this assignment is that it fulfills the requirement Surface tensions of SDBS-containing water and both polymers, collected in Table 5.4, were used for the determination of the various interfacial tensions. Table 5.4: Measured surface tensions of the various components of a latex. Material Surface tensions (mn m -1 ) d p Water + SDBS PMMA EP(D)M As demonstrated in Table 5.5, the calculated spreading coefficients predict a coreshell morphology for the particles produced during the seeded emulsion polymerization of MMA onto EP(D)M latex, since S 1 < 0, S 2 < 0, and S 3 > 0.

116 About morphology in EP(D)M-based latexes 105 Table 5.5: Interfacial tensions and spreading coefficients used in the Torza/Mason theory 35. EP( D) M water or Q W ) ( 12 Interfacial tensions (mn m -1 ) PMMA water or P W ) ( 23 EP( D) M PMMA or Q P ) ( 13 Spreading coefficients S 1 S 2 S The morphology predictions obtained with the Torza/Mason theory were verified using the quantification of relative surface energies, as suggested by Waters 41,42. Considering an engulfed polymer Q by a polymer P, fully engulfed structures are favored over any intermediate or separated structures if: Q W P W P Q 1 (5.6) where is the interfacial energy for the interfaces polymer Q-water, polymer P-water, and polymer Q-polymer P. Then, intermediate morphologies will be preferred over both fully engulfed and separated morphologies if: Q W P Q 1 1 P W (5.7) In case of Condition 5.6, it is possible to determine which morphology, either coreshell or inverted core-shell, is thermodynamically favored, by considering the fractional volumes, P and Q, of both polymers. Waters showed that a core-shell particle should be expected when the following condition is obeyed: Q W P Q P W 2 3 Q 2 3 P (5.8)

117 106 Chapter 5 In the present work, polymers Q and P correspond to EP(D)M and PMMA, respectively. The data in Table 5.6 confirm that, for our system, the core-shell morphology is to be expected, as was also predicted by the Torza/Mason theory. Table 5.6: Interfacial tensions and conditions for morphology predictions according to Waters 41,42. EP( D) M water ( Q W ) Interfacial tensions (mn m -1 ) PMMA water ( P W ) EP( D) M PMMA ) P W ( Q P Q W P Q Conditions Q W 3 2 / 3 P W 2 / Q P P Q Influence of various parameters on particle morphology Several parameters were studied in the present research to understand the chemistry involved in the procedure of producing core-shell impact modifiers, based on an EP(D)M core and a PMMA shell. In order to clearly differentiate EP(D)M from PMMA in cryo-tem pictures presented further on, a pure EP(D)M latex and a physical mixture of both an EP(D)M latex and a PMMA latex are depicted in Figure 5.8 and Figure 5.9, respectively.

118 About morphology in EP(D)M-based latexes 107 Figure 5.8: Cryo-TEM picture of a pure EP(D)M seed latex (S1). Figure 5.9: Cryo-TEM picture of a physical mixture of both an EP(D)M latex (S1) and a PMMA latex. EP(D)M particles exhibit a sharp edge and seem to exhibit a perfect spherical shape, while PMMA particles appear more fluffy. Moreover, the higher electron density of PMMA compared to EP(D)M emphasizes the darkness of the PMMA phase, as depicted in Figure 5.9. Monomer type Seeded emulsion polymerization SEP1 was carried out under starved-feed conditions and using a monomer-to-rubber ratio of 0.8 (wt/wt), as presented in Table 5.3. SEP1 leads to the formation of two types of particles, as depicted in Figure A minority of small particles of pure PMMA was formed by secondary nucleation, whereas grafting of monomer onto the EP(D)M seed latex gave rise to snowman-like particles. In terms of the Torza/Mason theory, this unpredicted morphology reveals that PMMA does not spread as expected on EP(D)M. This behavior may be due to the higher hydrophilicity of PMMA compared to EP(D)M. Therefore, styrene was introduced in small quantities in order to form more hydrophobic copolymers of MMA and styrene 43, hopefully resulting in an improved spreading over EP(D)M particles. Unfortunately, using a combination of MMA and styrene in SEP2 does not produce core-shell structures either,

119 108 Chapter 5 as depicted in Figure From this, we conclude that the different hydrophilicity of both the homopolymer PMMA and the styrene/mma copolymer as well as that of rubber is not the driving force to obtain core-shell morphology. Figure 5.10: Cryo-TEM picture of SEP1 (PMMA-to-EP(D)M ratio = 0.8). Figure 5.11: Cryo-TEM picture of SEP2, consisting of P(MMA-co-Styrene) (90/10 wt/wt) grafted onto EP(D)M particles. Monomer-to-rubber ratio The quantity of monomer, i.e. MMA only or a mixture of styrene and MMA, may be too small to completely cover the EP(D)M particles, which may explain the formation of unexpected snowman-like structures. The monomer-to-rubber ratio was therefore raised from 0.8 to 1.3, as indicated in Table 5.3. Note the presence of a larger amount of small particles, i.e. characterized by a diameter of 20 to 50 nm, produced by secondary nucleation, see Figure Snowman-like morphologies, exhibiting a much larger PMMA phase (dark and fluffy) than in Figure 5.10, can clearly be observed. Complete coverage of the EP(D)M particles by PMMA was still not achieved.

120 About morphology in EP(D)M-based latexes 109 Figure 5.12: Cryo-TEM picture of SEP3, consisting of PMMA-g-EP(D)M particles. The PMMA-to-EP(D)M ratio is 1.3. Small particles of pure PMMA were formed by secondary nucleation. Monomer feed type In addition to the monomer type and to the monomer-to-rubber ratio, the influence of the monomer feed type, i.e. flooded or starved-feed, was investigated. As observed in Figure 5.10, starved-feed conditions lead to the formation of snowman-like particles. Monomer-flooded conditions, on the other hand, gave rise to inverted core-shell particles, as depicted in Figure 5.13 for SEP4.

121 110 Chapter 5 a- b- Figure 5.13: Cryo-TEM pictures of SEP4, containing EP(D)M-g-PMMA particles, obtained in monomer-flooded conditions. The arrows pointing to dark clouds indicate non-polymerized monomer. In Figure 5.13b, small particles of pure PMMA formed by secondary nucleation can be distinguished. These inverted core-shell structures consist of a dark PMMA core and a relatively sharp-edged EP(D)M shell, see Figure 5.13a. Many secondary nucleated PMMA particles were also produced, as depicted in Figure 5.13b. During this experiment, most of the MMA entered the EP(D)M particles while a smaller fraction, typically 1.56 wt% 20, was present in the aqueous phase. Moreover, although the initiatior, i.e. CHP, is mostly water-insoluble, partitioning occurs, leading to a gradient of radicals inside the EP(D)M particles. As a consequence, both monomer and initiator radicals coexist in the rubbery particles, leading to the polymerization of MMA within these rubbery particles, and therefore to the formation of inverted core-shell structures. In this specific case, the morphology of the obtained composite particles is therefore determined by the kinetics of the reaction. Crosslinking of the EP(D)M seed particles Finally, the last parameter investigated consisted of the crosslinking of the EP(D)M seed particles, in the presence or absence of poly(1,2-butadiene) (Ricon 156). Two

122 About morphology in EP(D)M-based latexes 111 seeded emulsion polymerizations (SEP) of MMA were carried out. SEP5 was performed on an EP(D)M seed latex, crosslinked with benzoyl peroxide (BPO) as initiator, in the presence of both divinylbenzene (DVB) and Ricon 156 as crosslinking co-agents. SEP6 was completed in the same conditions, using DVB as the only co-agent. The latexes used as seed for SEP5 and SEP6 exhibited a gel content of approximately 21 % and 13 %, respectively. The structures obtained for SEP5 and SEP6 are presented in Figure 5.14 and Figure 5.15, respectively. Figure 5.14: Cryo-TEM picture of SEP5. The EP(D)M latex was crosslinked with BPO in the presence of DVB and poly(1,2- butadiene) prior to the grafting reaction (gel content 21 %). Figure 5.15: Cryo-TEM picture of SEP6. The EP(D)M latex was crosslinked with BPO in the presence of DVB only prior to the grafting reaction (gel content 13 %). The fluffy outer shell is a typical representation of PMMA in cryo-tem pictures, as previously stated. Therefore, Figure 5.14 and Figure 5.15 clearly demonstrate that a crosslinked EP(D)M seed latex is required for the formation of core-shell structures. Note that pendant polybutadiene unsaturations are not necessary to promote grafting of PMMA onto EP(D)M particles, as depicted in Figure 5.15.

123 112 Chapter 5 Furthermore, a thorough analysis of various cryo-tem pictures leads to the conclusion that particles from SEP5 consist of approximately 76 vol% of EP(D)M and 24 vol% of PMMA shell. The formation of snowman-like structures, as in Figures 5.10 and 5.12, as well as the formation of core-shell structures, as in Figures 5.14 and 5.15, may be explained by the relative difference of viscosity between EP(D)M and PMMA. Indeed, when an noncrosslinked EP(D)M latex is used, the soft and flowing EP(D)M chains tend to deform and are then expelled from the (hard) growing PMMA shell. This escape may be exemplified, for instance, by the behavior of a barely cooked egg pressed between two slices of hard bread. EP(D)M crosslinking hampers the mobility of the EP(D)M chains, therefore preventing the particles from deformation. However, during a seeded emulsion polymerization carried out with monomerflooded conditions, the PMMA chains grow within the soft EP(D)M particles, hence forming a core inside the EP(D)M shell. Note that the formation of core-shell structures was confirmed by minimum film formation temperature (MFFT) measurements. Thus, a latex based on snowman-like or inverted core-shell particles has a much lower MFFT, i.e. 7 C, than a latex containing core-shell particles with EP(D)M as core, i.e. 74 C. 5.4 Conclusions The present chapter reports about the various morphologies that may be obtained during seeded emulsion polymerizations of MMA onto EP(D)M. The morphology of the obtained particles was studied upon variation of several parameters, i.e. monomer type, monomer-to-ep(d)m ratio, monomer feed type, as well as crosslinking of the EP(D)M seed latex.

124 About morphology in EP(D)M-based latexes 113 Although kinetic and thermodynamic considerations allowed the prediction of coreshell structures, there remain some discrepancies between theory and experiment. A thorough experimental investigation demonstrated that the seeded emulsion polymerization of MMA onto EP(D)M can not only result in partial engulfment of the EP(D)M seed particles by the growing PMMA chains, leading to snowman-like particles, but can also result in complete engulfment, i.e. core-shell or inverted coreshell. The present chapter demonstrates that EP(D)M crosslinking is required to obtain coreshell structures. Therefore the viscosity of the seed particles should be taken into account to a more significant extent in the kinetic considerations. However, a gel content of 13 % (SEP6), obtained with divinylbenzene as the only co-agent, see Chapter 4, is sufficient to limit the mobility of the EP(D)M chains, and therefore to prevent the rubber phase from being expelled from the hard PMMA phase.

125 114 Chapter 5 References 1. Berzinis, A. P.; Wills, W. L.; Rohm and Haas, Eur. Pat. Appl , Troy, E. J.; Rosado, A.; Rohm and Haas, Can. Pat. Appl , Aerdts, A. M.; Groeninckx, G.; Zirkzee, H. F.; van Aert, H. A. M.; Geurts, J. M. Polymer 1997, 38, Lovell, P. A. Macromol. Symp. 1995, 92, Okubo, M.; Lu, Y.; Wang, Z. Colloid Polym. Sci. 1998, 276, Li, H.; Han, J.; Panioukhine, A.; Kumacheva, E. J. Colloid Interf. Sci. 2002, 255, Höhne, G.; Hemminger, W.; Flammersheim, H. J. in "Differential Scanning Calorimetry"; 2nd Ed., Berlin, Germany: Springer, 2003, Chapter Höhne, G. W. H. Thermochim. Acta 1997, 304/305, Höhne, G.; Hemminger, W.; Flammersheim, H. J. in "Differential Scanning Calorimetry"; Berlin, Germany: Springer, 2003, Chapter Wu, S. Polymer Interface and Adhesion; New York: Marcel Dekker, Inc., van Krevelen, D. W. in "Properties of Polymers"; 3rd Ed., Amsterdam: Elsevier, 1990, Chapter Fowkes, F. M. J. Phys. Chem. 1962, 66, Owens, D. K.; Wendt, R. C. J. Appl. Polym. Sci. 1969, 13, Rulison, C. Technical Note No. 306, Krüss, USA, Aerdts, A. M.; Theelen, S. J. C.; Smit, T. M. C.; German, A. L. Polymer 1994, 35, Wang, C. C.; Yu, N. S.; Chen, C. Y.; Kuo, J. F. Polymer 1996, 37, Arayapranee, W.; Prasassarakich, P.; Rempel, G. L. J. Appl. Polym. Sci. 2002, 83, van Zyl, A. J. P.; Sanderson, R. D.; Wet-Roos, D.; Klumperman, B. Macromolecules 2003, 36, 8621.

126 About morphology in EP(D)M-based latexes El-Aasser, M. S.; Sudol, E. D. in "Emulsion Polymerization and Emulsion Polymers"; Lovell, P. A and El-Aasser, M. S. Eds., Chichester: Wiley, 1997, Chapter Lide, D. R. CRC Handbook of Chemistry and Physics; 84, Boca Raton, FL ; London : CRC, Tsavalas, J. G.; Luo, Y.; Schork, F. J. J. Appl. Polym. Sci. 2003, 87, Merkel, M. P.; Dimonie, V. L.; El-Aasser, M. S.; Vanderhoff, J. W. J. Polym. Sci. Part A: Polym. Chem. 1987, 25, Merkel, M. P.; Dimonie, V. L.; El-Aasser, M. S.; Vanderhoff, J. W. J. Polym. Sci. Part A: Polym. Chem. 1987, 25, Ferguson, C. J.; Russell, G. T.; Gilbert, R. G. Polymer 2003, 44, Lee, S.; Rudin, A. J. Polym. Sci. Part A: Polym. Chem. 1992, 30, Cho, I.; Lee, K. W. J. Appl. Polym. Sci. 1985, 30, Van Doremaele, G. H. J. PhD Thesis, University of Eindhoven, Omi, S.; Senba, T.; Nagai, M.; Ma, G. H. J. Appl. Polym. Sci. 2001, 79, Okubo, M.; Yamada, A.; Matsumoto, T. J. Polym. Sci. Polym. Chem. Ed. 1980, 18, Chern, C. S.; Poehlein, G. W. J. Polym. Sci. Part A: Polym. Chem. 1987, 25, Chern, C. S.; Poehlein, G. W. J. Polym. Sci. Part A: Polym. Chem. 1990, 28, Chern, C. S.; Poehlein, G. W. J. Polym. Sci. Part A: Polym. Chem. 1990, 28, van Zyl, A. J. P. PhD Thesis, University of Stellenbosch, Gonzalez-Ortiz, L. J.; Asua, J. M. Macromolecules 1995, 28, Gonzalez-Ortiz, L. J.; Asua, J. M. Macromolecules 1996, 29, Gonzalez-Ortiz, L. J.; Asua, J. M. Macromolecules 1996, 29, Mills, M. F.; Gilbert, R. G.; Napper, D. H. Macromolecules 1990, 23, Torza, S.; Mason, S. G. J. Colloid Interf. Sci. 1970, 33, Sundberg, D. C.; Casassa, A. P.; Pantazopoulos, J.; Muscato, M. R.; Kronberg, B.; Berg, J. J. Appl. Polym. Sci. 1990, 41, 1425.

127 116 Chapter Chen, Y. C.; Dimonie, V.; El-Aasser, M. S. Macromolecules 1991, 24, Waters, J. A. Colloids Surf. A: Physicochem. Eng. Aspects 1994, 83, Waters, J. A. in "Colloidal Polymer Particles"; Goodwin, J. W. and Buscall, R Eds., London: Academic Press Limited, 1995, Chapter Davis, T. P.; O'Driscoll, K. F.; Piton, M. C.; Winnik, M. A. J. Poly. Sci. Part C: Polym. Lett. 1989, 27, 181.

128 Toughening effect of EP(D)M-PMMA core-shell structures in a brittle PMMA matrix Abstract The core-shell structured particles produced by seeded emulsion polymerization of methyl methacrylate (MMA) onto ethylene-propylene-(diene) copolymers (EP(D)M) were used to toughen a poly(methyl methacrylate) (PMMA) matrix. The tensile properties of the modified PMMA matrix were investigated. The micro-morphology of modified PMMA was studied by both Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM). Both tensile tests and SEM analysis demonstrated that the main mechanism of deformation operating in the EP(D)M-toughened PMMA matrix is shear yielding, accompanied by debonding and cavitation processes. This chapter is part of: Tillier, D. L.; Meuldijk, J.; Höhne, G. W. H.; Frederik, P. M.; Regev, O.; Koning, C.E. Manuscript submitted for publication.

129 118 Chapter Introduction As already discussed in Chapter 1, applications of both engineering materials and coatings may be limited by the propensity of these materials for brittle fracture under environmental stress cracking conditions. A method to enhance the toughness of such materials, under stringent conditions such as low temperatures or in the presence of a sharp notch, consists of incorporating a rubbery phase 1. Interfacial adhesion between this rubber phase and the polymer matrix is regarded as playing a significant role in the toughening of the brittle matrix. Hence, an improved adhesion leads to enhanced toughness. Therefore, the nature of the matrix considered in the present chapter was chosen similar to that of the shell, i.e. poly(methyl methacrylate) (PMMA), in order to assure adhesion between the impact modifier and the fragile matrix. PMMA is a typical example of a brittle polymer. In the past, attempts have been made to increase PMMA s toughness by incorporation of a rubbery phase 1-6. In analogy with these works 1-6, results obtained with mechanical tests, i.e. tensile tests, as well as scanning electron microscopy (SEM), are presented in the present chapter. Tensile tests and SEM have been used to evaluate the properties of the structured particles, of which the preparation has been described in Chapter 5, and which have been incorporated into a PMMA matrix. 6.2 Experimental section Chemicals The chemicals used for the preparation of the core-shell-based latex are presented in Chapters 4 and 5.

130 Toughening effect of EP(D)M-PMMA core-shell structures 119 The pure PMMA latex, forming the basis for the matrix, was prepared using sodium dodecyl sulfate (SDS, Aldrich, 98 %) as surfactant, sodium persulfate (SPS, Aldrich, reagent grade, +98 %) as initiator, and sodium hydrogen carbonate (NaHCO 3, Aldrich, 99 %) as ph buffer. Deionized water was used in all recipes Synthesis of the composite latex particles The preparation of a structured latex has been extensively described in Chapter 5. For the present chapter, an EP(D)M seed latex was crosslinked in the presence of DVB as the only co-agent, i.e. latex S3 in Table 5.2 (page 87), and was then subjected to the seeded emulsion polymerization of MMA, i.e. SEP6 in Table 5.3 (page 88) Synthesis of a PMMA homopolymer latex The emulsion homopolymerization of MMA was carried out in a 300 ml jacketedreactor, equipped with a condenser and a downflow 45 pitched four-blade impeller. SDS, NaHCO 3, and water were first charged into the vessel. Oxygen was removed by purging argon through the mixture for at least 30 minutes. The addition of monomer was then followed by its emulsification at 60 C and at a stirring speed of 400 rpm, during 45 minutes. Finally, the polymerization was initiated by introducing an aqueous solution of SPS into the reaction mixture. A typical recipe is exemplified in Table 6.1. Table 6.1: Typical recipe of an emulsion homopolymerization of MMA. Products Amount (g) Total weight fraction (wt%) Water SDS NaHCO SPS MMA

131 120 Chapter Blending and specimen preparation Mixing the composite rubber latex and the PMMA homopolymer latex in different weight ratios led to the formation of two new mixed latexes, containing a total EP(D)M weight fraction of 5 wt% and 15 wt%, respectively. After stirring, the homogeneous mixtures were submitted to freeze-drying, leading to white powders. These PMMA/rubber blends were compression molded at 180 C, under pressures ranging from 40 to 100 bars. The obtained films were then transformed into tensile bars, characterized by a thickness of 0.7 mm Characterization Molecular weight and molecular weight distribution The molecular weight (MW) and molecular weight distribution (MWD) of the PMMA matrix were determined at ambient temperature, using a Waters Size Exclusion Chromatograph (SEC), equipped with a Waters model 510 pump, a Waters 410 differential refractometer operating at 40 C and a Waters model 486 UV detector operating at 254 nm. The sample was injected using a Waters WISP 712 autoinjector (50 L injection volume). The columns consisted of a PL gel guard (5 m particles) mm column, followed by two PL gel mixed-c or mixed-d (5 m particles) mm columns at 40 C in series. The eluent was THF, and the elution volumetric flow rate was maintained at 1 ml min -1. Calibration was carried out using narrow MWD polystyrene standards ranging from 580 to g mol -1. Tensile properties Tensile tests were performed on a Zwick Z010 tensile-testing machine, based on ISO standard 527, at room temperature. The strain speed was 1 mm min -1. The elongation

132 Toughening effect of EP(D)M-PMMA core-shell structures 121 was measured directly on the sample by a tensometer. The tensile curves represent the average data points of five measurements. Micro-morphology The dispersion of the rubbery particles in the PMMA matrix was observed on nonelongated samples by Transmission Electron Microscopy (TEM). Samples were trimmed at low temperature to achieve a smooth undeformed surface and subsequently treated during 20 hours with a ruthenium tetraoxide (RuO 4 )-solution 7. Ultra-thin sections were obtained at room temperature using a Reichert Ultracut E microtome, equipped with a diamond knife. TEM was performed using a Jeol JEM 2000 FX microscope, operated at 80 kv. After tensile tests, the fracture surfaces of the specimens were examined with a Philips XL30 Field Emission Gun-Environmental Scanning Electron Microscope (FEG-ESEM). Samples were coated with a thin gold layer, using an Emitech K575X sputter-coater. 6.3 Results and discussion Molecular characterization of PMMA Both number and weight average molecular weights, as well as the polydispersity index of the PMMA used for the matrix are summarized in Table 6.2. Table 6.2: Molecular characterization of the PMMA used for the matrix, obtained by SEC. M n (g mol -1 ) M w (g mol -1 ) M M w n

133 122 Chapter 6 Note that this PMMA, synthesized by emulsion polymerization, not only exhibits a tailing towards the low molecular weight region, as demonstrated by Figure 6.1, but also include small quantities of short molecules, such as surfactant and ph buffer. Similarly, the EP(D)M-PMMA latex obtained after seeded emulsion polymerization of MMA onto EP(D)M, contains the redox initiation system, as well as pure PMMA formed by secondary nucleation. Therefore, the mechanical properties of the modified PMMA matrix will be very likely affected by the presence of these low molar mass compounds. As a consequence, in the future, a thorough purification of the latexes may be of great significance for the complete characterization of the mechanical properties. DRI signal (a.u.) Retention time (min) Figure 6.1: SEC chromatogram of PMMA used for the matrix Incorporation of the rubbery particles into a PMMA matrix TEM pictures of the samples containing 5 wt% and 15 wt% of rubber are depicted in Figures 6.2 and 6.3, respectively. These pictures demonstrate that the rubbery particles seem more homogeneously dispersed in the sample containing 5 wt% of rubber. Not only single particles but also agglomerates can be distinguished in the sample containing 15 wt% of rubber, see Figure 6.3. However, as reported in the next sections, these

134 Toughening effect of EP(D)M-PMMA core-shell structures 123 agglomerates do not seem to affect the tensile properties of the studied samples to a large extent. Figure 6.2: TEM section of a sample containing 5 wt% of EP(D)M dispersed in a PMMA matrix. Figure 6.3: TEM section of a sample containing 15 wt% of EP(D)M dispersed in a PMMA matrix Mechanical properties of PMMA/rubber blends Characterization of pure PMMA samples with tensile tests could not be performed, since samples either slipped between the clamps of the machine or were damaged by them. Therefore, it is only possible to compare the data of the present PMMA/rubber blends with those of pure PMMA obtained from literature2. Obviously, the blends containing various amounts of rubber can also be compared with each other. Note that the stress-strain curve of PMMA obtained from literature most likely corresponds to the upper stiffness limit for our pure PMMA sample, whereas the curve obtained for the sample containing 5 wt% of rubber, corresponds approximately to the lower limit. As demonstrated by the stress-strain curves in Figure 6.4, the ductility of a brittle PMMA matrix is significantly enhanced by the incorporation of an increasing amount of rubber into the PMMA matrix, in the form of core-shell particles. It should be

135 124 Chapter 6 mentioned here that the incorporation of snowman-like EP(D)M-g-PMMA particles, as opposed to core-shell particles, gives rise to sticky tensile specimens. Therefore, encapsulation of the EP(D)M phase by a PMMA shell is required to obtain satisfactory surface properties of the samples. As reported by Schneider et al. 8, the Young s modulus, E, is controlled by the rubber content of the blends. This statement is confirmed by Figure 6.4, where the slope of the stress-strain curves, i.e. the Young s modulus, significantly decreases with increasing amounts of EP(D)M in the PMMA matrix. Simultaneously, the tensile strain increases from approximately 2 % for pure PMMA to 4 % for a 5 % EP(D)M-based sample, and to 6 % for an EP(D)M content of 15 %. Moreover, the surface area under the stress-strain curve, which is a measure for the toughness of the material, is clearly enlarged by the increasing rubber content, see Figure PMMA 50 Tensile stress, (MPa) % EPDM content 15% EPDM content Tensile strain, (%) Figure 6.4: Stress-strain curves of PMMA toughened by different amounts of EP(D)M-PMMA core-shell particles. Note that the pure PMMA curve was obtained from literature 2.

136 Toughening effect of EP(D)M-PMMA core-shell structures 125 Finally, tensile stress and bending of the test bars lead to the formation of white shear bands, as observed in Figure 6.5. Stress-whitening, which results from plastic deformation, is usually regarded as a characteristic for deformation of tough materials 1,9. Hence, the ductility of the material clearly increases with an increasing rubber content. The designed core-shell structures thus provide the desired toughness enhancement. Figure 6.5: Tensile bars after bending. The pure PMMA specimen (top) was broken without any stresswhitening. The toughened material (bottom), containing 5 % of EP(D)M, shows stress-whitening in its center, where bending had taken place Fracture mechanism Scanning electron microscopy (SEM) was performed to relate the results of the tensile tests to the morphology of the multiphase materials produced. Figures 6.6 and 6.7 present the fracture surface of two PMMA/rubber blends containing 5 wt% and 15 wt% of EP(D)M, respectively.

137 126 Chapter 6 Figure 6.6: SEM of the fracture surface of a PMMA matrix blended with EP(D)M-g-PMMA core-shell particles. The total EP(D)M content in the sample is 5 wt%. Figure 6.7: SEM of the fracture surface of a PMMA matrix blended with EP(D)M-g-PMMA core-shell particles. The total EP(D)M content in the sample is 15 wt%. The roughness of the fracture surfaces, including holes and dome-like features, can clearly be observed in the SEM micrographs. These holes correspond to the space