Diffusion Phenomena in the Fused Deposition Modeling Process

DIFFUSION PHENOMENA IN THE FUSED DEPOSITION MODELING PROCESS

OF FLUORINATED THERMOPLASTIC BINDERS

PAU SALDANA BAQUE

B.S., Universitat de Girona, 2019

A thesis submitted to the Graduate Faculty of the

University of Colorado Colorado Springs

in partial fulfillment of the

requirements for the degree of

Master of Science

Department of Mechanical and Aerospace Engineering

2021

PAU SALDANA BAQUE

This thesis for the Master of Science degree by

Pau Saldana Baque

has been approved for the

Department of Mechanical and Aerospace Engineering

Jena M. McCollum, Chair

Scott T. Iacono

James Stevens

Date: 04/20/2021

ii Saldana Baque, Pau (M.S., Mechanical Engineering)

Diffusion Phenomena in the Fused Deposition Modeling Process of Fluorinated Thermoplastic Binders

Thesis directed by Assistant Professor Jena M. McCollum

ABSTRACT

Motivated by the lack of polymer welding information in the additive manufacturing (AM) field, this work studies the effect of diffusion phenomena in fused deposition modeling (FDM) of fluorinated thermoplastic binders. In particular, this study compares the anisotropic mechanical response of 3D-printed binary blends of poly(vinylidene fluoride) (PVDF) and poly(methyl methacrylate) (PMMA) with the isotropic mechanical response of these blends fabricated via molding techniques.

PVDF/PMMA filaments were produced by twin-screw extrusion and, subsequently, injection-molded or 3D printed into dog-bone shapes. Specimen mechanical and thermal properties were evaluated by tensile testing and differential scanning calorimetry, respectively. Results show that higher PMMA concentration not only improved the tensile strength and increased specimen brittleness but prevented the crystallization. As expected, injection-molded samples revealed better mechanical properties compared to 3D printed specimens. Interestingly, 3D printed blends with lower PMMA content demonstrated better diffusion (adhesion) across interfaces than those with a higher amount of PMMA. The present study provides new data to improve the description of the effect of PMMA in processed PVDF/PMMA blends. This will be useful to optimize the PVDF/PMMA mixture composition when producing energetic thermoplastics through FDM.

iii ACKNOWLEDGEMENTS

First of all, I would like to express my sincere gratitude to the Balsells Foundation for providing me with financial support during the final year of undergraduate studies and throughout the master's degree. Further, I would like to thank my advisor, Dr. McCollum, and my best friend and coworker, Jared Strutton, for their consistent support and guidance during the running of this project. I am also thankful to Dr. Iacono for letting us use the laboratory facilities at the Air Force Academy. To conclude, I cannot forget to thank my family and girlfriend for all the unconditional support in these very intense academic years.

iv TABLE OF CONTENTS

CHAPTER

1. INTRODUCTION ...... 1

2. BACKGROUND ...... 4

2.1 Thermoplastics vs thermosets ...... 4 2.2 Adhesion theory ...... 5 2.2.1 Wetting and contact angle ...... 5 2.2.2 Surface tension and surface energy...... 7 2.2.3 Crystallinity...... 9 2.2.4 Weak boundary layer theory, and adhesion and transition zones ...... 9 2.3 Polymer adhesion mechanisms in thermoplastic printing ...... 10 2.3.1 Diffusion and entanglement ...... 12 2.3.2 Mechanical interlocking (anchoring) ...... 15 2.3.3 Dispersive or adsorptive adhesion ...... 16 2.3.4 Electrostatic interactions (contact charging) ...... 17 2.4 Bond formation process among thermoplastic printed layers ...... 18 2.5 Theoretical model of the bonding formation process ...... 21 3. MATERIALS AND METHODS ...... 23

3.1 Materials description ...... 23 3.1.1 Poly(vinylidene fluoride) (PVDF) ...... 23 3.1.2 Poly(methyl methacrylate) (PMMA) ...... 24 3.2 Blend preparation ...... 24 3.3 Apparent viscosity measurements ...... 25 3.4 Dog-bone production ...... 25 3.4.1 Fused deposition modeling (FDM) ...... 26 3.4.2 Injection molding (IM) ...... 26 3.5 Performed experiments ...... 27 3.5.1 Tensile testing ...... 27 3.5.2 Thermal analysis ...... 28 3.5.3 Attenuated total reflectance spectroscopy (ATR-FTIR) ...... 29 3.5.4 Gel permeation chromatography (GPC) ...... 29 4. RESULTS AND DISCUSSION ...... 32

4.1 Apparent viscosity measurements ...... 32

v 4.2 Tensile testing ...... 33 4.3 Thermal analysis ...... 37 4.4 Attenuated total reflectance spectroscopy (ATR-FTIR) ...... 42 4.5 Gel permeation chromatography (GPC) ...... 46 5. CONCLUSIONS ...... 49

6. FUTURE WORK ...... 52

REFERENCES ...... 54

vi LIST OF TABLES

TABLE

Table 1. Glass transition (Tg), melting temperature (Tm), melting enthalpy (Hm), and crystallization (Xc) for the first and second heating cycles of various 3D printed PVDF/PMMA compositions...... 41

Table 2. Glass transition (Tg), melting temperature (Tm), melting enthalpy (Hm), and crystallization (Xc) for the first and second heating cycles of various twin-screw extruded (filament) PVDF/PMMA compositions...... 41

Table 3. Glass transition (Tg), melting temperature (Tm), melting enthalpy (Hm), and crystallization (Xc) for the first and second heating cycles of various injection molded PVDF/PMMA compositions...... 41 Table 4. Number ( ) and weight ( ) averaged molecular weights and dispersity index (Ð) for virgin and processed PMMA samples...... 47 푀푛 푀푤 Table 5. Number ( ) and weight ( ) averaged molecular weights and dispersity index (Ð) for virgin and processed PVDF samples...... 48 푀푛 푀푤

vii LIST OF FIGURES

FIGURE

Figure 1. Schematic illustration of the adhesion quality importance during FDM...... 3 Figure 2. Illustration of different contact angles formed between the cross-section of printed thermoplastic filaments (liquid) and the build platform (solid). (a) Perfect wetting (b) High wettability (c) Low wettability...... 6 Figure 3. Interfacial tensions and equilibrium contact angle formed between the cross-section of an extruded thermoplastic filament (liquid) and the build platform (solid). , , and represent훾 the liquid-gas (air), solid-gas휃푐 (air), and solid-liquid surface tensions, respectively...... 8 훾퐿퐺 훾푆퐺 훾푆퐿 Figure 4. Schematic illustration of polymer chains inter-diffusion across an interface. .. 15 Figure 5. Schematic illustration of the mechanical interlocking mechanism between two surfaces...... 16 Figure 6. Schematic illustration of a) dipole-dipole interactions and b) dipole-induced dipole interaction...... 17 Figure 7. Schematic illustration of the contact charging phenomenon...... 18 Figure 8. Bond formation process between the extruded polymer filament and the preceding printed layer. (a) surface contact, (b) wetting and, (c) diffusion and randomization...... 21 Figure 9. Schematic illustration of dog-bone fabrication for tensile testing...... 28 Figure 10. Schematic illustration of GPC analysis...... 30 Figure 11. Viscosity response of PVDF/PMMA blends during compounding as a function of PMMA content. Error bars represent the standard error of the mean values...... 32 Figure 12. Mechanical properties of injection molded (IM) and 3D printed (3D) PVDF/PMMA dog-bones as a function of PMMA content: (a) tensile strength, (b) elongation at break, and (c) Young’s modulus. In each panel, error bars represent the standard error of the mean values of each mechanical property...... 33 Figure 13. DSC traces for 3D printed samples undergoing the (a) first heating cycle and (b) second heating cycle. Data are summarized in Table 1...... 40 Figure 14. DSC traces for twin-screw extruded samples (filament) undergoing the (c) first heating cycle and (d) second heating cycle. Data are summarized in Table 2...... 40 Figure 15. DSC traces for injection molded samples undergoing the (e) first heating cycle and (f) second heating cycle. Data are summarized in Table 3...... 40

viii Figure 16. FTIR spectra of PVDF/PMMA blends processed by (a) injection-molding, (b) 3D printing and (c) twin-screw extrusion (filament)...... 43 Figure 17. β-phase percent ratio in PVDF/PMMA processed blends as a function of PMMA content...... 45 Figure 18. GPC curves of PMMA and PVDF for virgin (pellet) and processed (filament, injection-molded, and 3D printed) samples. Data are summarized in Tables 4 and 5. .... 46

ix CHAPTER 1 INTRODUCTION

Just as a sculptor carves a stone to create a statue, traditional or subtractive manufacturing allows us to manipulate the raw material to make a specific shape. However, this approach changed radically with the appearance of additive manufacturing (AM), the key to which is elements adhesion. Instead of removing, the material is added to achieve the desired piece. This disruptive technology allows the production of complex geometries that are not achievable by traditional methods [1]. Still, existing studies of additive manufacturing (AM) technologies have shown that, despite their considerable advantages over traditional manufacturing methods, the realization of their full potential is stifled by challenges with understanding the processing-structure-property relationship [1, 2]. In the particular case of the fused deposition modeling (FDM) technique, the deep understanding of the adhesion phenomena that occur during thermoplastic printing remains an open problem. Adhesion does not only play a very important role in AM techniques but also in many aspects of traditional construction techniques [3]. The adhesion of a brick with the mortar (cement) is a clear example of it [4].

In any extrusion-based printing method, the basic idea behind a good polymer interlayer welding is to have the best possible interfacial adhesion during the printing process. In other words, finding the ideal printing conditions for each material to adhere to as well as possible for minimizing anisotropic static and dynamic properties caused by microstructural heterogeneity [1]. This implies that polymer chains must reach a high enough temperature, above the glass-transition temperature (Tg), so that they are mobile and can interact and diffuse through the boundaries. Ultimately, the mechanical response

1 of a printed thermoplastic (multilayered specimen) is desired to be similar to the isotropic mechanical response of the same thermoplastic fabricated via molding techniques

(homogeneous specimen). In this way, by performing mechanical testing on these specimens, one can probe interface diffusion by observing behavior in the bulk.

The FDM process consists of a heated print nozzle through which a polymer filament passes undergoing melting. The extruded filament is then deposited onto a build platform in a chosen xy-plane, while building layer-upon-layer in the vertical z-dimension, allowing the production of complex geometries (see Figure 1). In this process, the adhesion quality among thermoplastic interfaces depends on the type of polymer and grade used, convective and conductive exposure, blend composition, if any, and on the printing parameters employed, such as speed, temperature, etc. Being able to quantify this quality is of utmost importance because it will determine the integrity and mechanical properties of the manufactured materials [5]. Thus, the main objective of this study is to understand and probe interface diffusion to know the factors that affect layer adhesion.

This work aims to describe the diffusion phenomena that occur in the FDM process of fluorinated thermoplastic binders. Particularly, it focuses on 3D printed binary blends of poly(vinylidene fluoride) (PVDF) and poly(methyl methacrylate) (PMMA). At this point, the question may arise of why someone should be interested in the diffusion of

PVDF/PMMA binders. A good contextualization to answer that question could be novel energetic thermoplastics [6]. In general, when a person hears the word combustion, a redox chemical reaction between a fuel (the reductant) and oxygen (the oxidant) is the first process that comes to mind. Still, there are many other types of oxidants. For instance, fluorinated thermoplastic oxidizers, which are widely used in pyrotechnic applications [7],

2 are a good alternative as an oxidizing agent. Nevertheless, the fluoropolymer of study

(PVDF) alone has very low surface energy, which makes it difficult to directly 3D print with it. This is where the use of PMMA plays an essential role because it has high surface energy and presents good miscibility with PVDF [8–10], increasing the overall surface energy and miscibility of the blend. Hence, it enables the production of energetic thermoplastics through FDM employing PVDF/PMMA blends as binders.

Figure 1. Schematic illustration of the adhesion quality importance during FDM.

3 CHAPTER 2 BACKGROUND

2.1 Thermoplastics vs thermosets

Properties of polymers depend on various factors such as monomer units, type of linkage between monomers, and intermolecular and intramolecular forces among polymers. Furthermore, depending on their response to the application of heat, they can be classified into thermoplastics and thermosetting polymers (well-known as thermosets).

Thermoplastics, the subject of this study, are polymers that soften when heated and hardens when cooled without causing any chemical change. Thus, they can be remolded and reshaped. This behavior is only possible because this type of polymers, which generally have a high molecular weight, are not chemically linked together. Their macromolecule chains are only entangled. The latter explains why these chains, which can only interact by weak intermolecular forces, move over one another with a snake-like motion (reptation motion) when heated [11–13].

On the other hand, thermosetting polymers do not soften and retain their shape when heated. In its uncured form, the material is usually malleable or liquid. When it undergoes the curing process, normally induced by heat, it forms irreversible chemical bonds, i.e., polymer chains are permanently joined by cross-links. It is the presence of these cross-links that increases the resistance of thermosets to heat and gives them a permanently set shape. Once thermosets cool and harden, they cannot be melted again. Instead, they burn on heating.

4 2.2 Adhesion theory

Adhesion is defined as the propensity of dissimilar surfaces of the same or different substances to cling to one another when they come into contact [14], as a result of intermolecular forces acting between them. In contrast, cohesion is distinctly different from adhesion. Cohesion is the force of attraction between adjacent particles within the same body, while adhesion is the interaction between the surfaces of different bodies. Hence, the overall bond effectiveness of printed material depends on the combination of adhesion and cohesion forces. Still, assuming that thermoplastic cohesion is generally going to be stronger than interlayer adhesion, one should focus much of his effort on controlling adhesion among printed layers.

Adhesion quality depends on the intermolecular interactions between surfaces, and the contact area. Thus, there are several factors involved in the overall adhesion phenomenon, such as wetting, contact angle, surface tension, surface energy, crystallinity, and adhesion and transition zones.

2.2.1 Wetting and contact angle

Wetting is the capability of a liquid to form an interface with a solid surface. Its quality depends on the balance between adhesive and cohesive forces of the material. For example, if the adhesive forces are greater, understood as attractive forces between molecules of dissimilar surfaces, the melted thermoplastic tends to spread across the surface of contact, improving wetting quality. On the contrary, if the cohesive forces are greater, understood as intermolecular forces within the liquid, the melted thermoplastic tends to form droplets on the surface of contact, worsening wetting quality.

5 The degree of wetting (wettability) is determined by measuring the contact angle

formed between the liquid and the solid surface (see Figure 2). This angle is characteristic휃푐 of each solid-liquid-environment system [15]. When wetting is perfect, corresponding to a contact angle of 0°, the liquid spreads and forms a thin film on the solid surface. Conversely, when the angle is close to 180°, no wetting (dewetting) occurs because the fluid minimizes its contact with the solid surface and forms a bead on it. Between these two limit cases, wettability can be high or low. The first occurs when the contact angle is less than 90°, while the second occurs when it is greater than 90° (see Figure 2). The goniometer is the typical instrument used to measure these contact angles.

Figure 2. Illustration of different contact angles formed between the cross-section of printed thermoplastic filaments (liquid) and the build platform (solid). (a) Perfect wetting (b) High wettability (c) Low wettability.

Heat transfer and time play a very important role in surface wetting. During the

FDM process, the extruded thermoplastic has to be maintained at a temperature above the glass transition temperature for a specific time period so that it has enough time to wet either the build platform or the previously printed layer. In other words, the time factor and molecular mobility must be taken into account when studying wetting [16]. To do so, it is essential to know the rheological properties of the material, such as viscosity and thixotropy [17]. These depend on the molecular structure of the material. Therefore, when working with thermoplastics, it is of utmost importance to know the length of the main

6 chains and the presence of side chains. Lastly, it is necessary to note that wetting also depends on thermodynamic surface energies, which will be discussed below.

2.2.2 Surface tension and surface energy

The intermolecular attractive force that contracts a liquid surface is called surface tension, which together with the gravity force causes the droplet shape. In particular, this tension is caused by the unbalanced cohesive forces of liquid molecules at the surface [18].

Unlike bulk molecules, those exposed at the liquid surface do not have neighboring molecules in all directions. This creates an unbalanced net force which in turn generates internal pressure that pulls the surface molecules inward. Consequently, the liquid surface area contracts to maintain the lowest surface energy. Thus, a small surface tension implies a small contact angle, that is, a high degree of wetting. This tension can also be expressed as the energy required to increase the surface area of a liquid by a unit of area. However, when working in terms of energy, surface energy is an interconnected concept that is more commonly used. The latter is generally used only when referring to solid surfaces. The surface energy can be defined as the work needed to create an area of a specific surface

(J/m2), and it is proportional to the adhesiveness of the material. It can also be defined as the excess energy at the material surface compared with its bulk [19]. Therefore, it can be interpreted as the surface tension of a solid since both concepts refer to the same physical quantity.

As previously stated, adhesion depends on the intermolecular interactions between surfaces. These interactions are determined by the surface energy of the solid and the surface tension of the applied liquid. Hence, if known, the behavior of any liquid on a solid surface can be predicted. The surface tension of a liquid is typically measured directly

7 using tensiometry, while the surface energy of a solid is determined indirectly using the wetting behavior of test liquids with known surface tensions [20]. More specifically, the surface energy, which quantifies the wetting characteristics of solid materials, is obtained through calculations based on the measured contact angle values. The basis for these calculations was first introduced by Thomas Young [21]: 휃c

(1) where , , and represent훾퐿퐺 푐표푠 the ( liquid-gas휃푐) = 훾푆퐺 (air),− 훾푆퐿 solid-gas (air), and solid-liquid

퐿퐺 푆퐺 푆퐿 surface훾 tensions/energies,훾 훾 respectively (see Figure 3). Eq. (1) is referred to as Young’s equation. In the case of perfect wetting (i.e., and ), this equation yields , which implies 휃푐 =. This0° means푐표푠 ( that휃푐) = wetting 1 will occur provided훾퐿퐺 that= 훾the푆퐺 −surface 훾푆퐿 energy of the solid훾퐿퐺 ≤ 훾푆퐺 is higher than the surface tension of the liquid . In other words, a polymer melt will훾푆퐺 spontaneously coat a high-energy surface as long훾 퐿퐺as the surface tension of the liquid is lower than the surface energy of the solid substrate [16,20]. It can be noted that the value of is zero when the same materials are in contact. 훾푆퐿

Figure 3. Interfacial tensions and equilibrium contact angle formed between the cross-section of an extruded thermoplastic filament (liquid)훾 and the build platform (solid). 푐 휃 , , and represent the liquid-gas (air), solid-gas (air), and solid-liquid surface tensions, respectively. 훾퐿퐺 훾푆퐺 훾푆퐿

8 To summarize, the higher the surface energy of a thermoplastic, the better the wetting and hence the contact area. Thus, it can be inferred that adhesion has a clear dependence on wetting. Indeed, a good wetting implies that polymer chains can interact more efficiently because the contact area is larger. This is why low surface energy thermoplastics do not wet and are more difficult to bond. Fluoropolymers, such as poly(vinylidene fluoride) (PVDF) and poly(tetrafluoroethylene) (PTFE), are good examples of it. In particular, PTFE is widely used as a non-stick coating for cookware [22].

2.2.3 Crystallinity

When a semi-crystalline polymer crystallizes, it loses some of its amorphous characteristics, thereby preventing the movement of polymer chains. In other words, the molecular mobility is restricted in crystalline areas. As a result, diffusion worsens and hence the adhesion strength. It is worth mentioning that since crystalline domains usually form on surfaces, they do not only affect interface adhesion but also surface wetting.

Surface energy values have been shown to increase as crystallinity decreases (or amorphous content increases) [23]. Moreover, by controlling the rate of solidification during cooling and the chain structure, it is possible to achieve desired degrees of crystallinity [24]. Consequently, crystallization enhancement is a way to probe interface, as the more interfaces there are, the more likely it is for the crystalline polymer to start crystallizing.

2.2.4 Weak boundary layer theory, and adhesion and transition zones

Bikerman [25] first introduced the idea that at the vicinity of the adhesion interface may exist a weak boundary layer, which contributes to cohesive failure within the material.

This layer can be caused by flaws near the adhesion interface that undermine the overall

9 bond strength. Hence, when a bond failure occurs, it can be the weak boundary layer that fails, although failure may seem to occur at the interface (i.e., adhesive failure) [26].

When printing with thermoplastics, once the adhesion process is complete, there may be a region between printed layers called the adhesion zone. This interfacial region is characterized by containing changes in the molecular structure/orientation caused by the bonding interactions that occur during the process. Furthermore, between the bonding interface and the bulk material, there could also exist a region referred to as the transition zone where mechanical properties differ from those of the bulk material. The thickness of both zones depends on the type of thermoplastic and the printing process. Thus, the overall bond effectiveness of a printed polymer may also depend on the properties of this transition zone, as these properties may be impaired because of inadequate cohesion within the same body [27]. Therefore, if mechanical testing is performed and a printed prototype fails within the transition zone, the failure mechanism will be considered cohesive instead of adhesive.

To recap, most printed layers can undergo physical or property changes caused by the heat transfer from the applied top layer. These layers may experience remelting, change in molecules orientation, and possible melt welding. Moreover, it is important to note that the transition zone and the theory of the weak boundary layer and are two interconnected concepts.

2.3 Polymer adhesion mechanisms in thermoplastic printing

There is no single theory that explicitly explains the adhesion phenomenon [26,28].

Instead, the latter should be described through the sum of several simultaneously occurring bonding forces, which are all based on different adhesion mechanisms [29]. The accuracy

10 of the choice of mechanism(s) will largely depend on the type of thermoplastic and the printing process. It is important to keep in mind that, in most cases, these mechanisms and their reciprocal effects have not yet been fully understood [29]. To optimize and understand adhesion in thermoplastic printing, it is essential to examine the role that different adhesion mechanisms play during the FDM process. Hence, it is important to quantify the effect that each of these mechanisms has on the layer adhesion strength, that is, on how well the individual layers of the printed material bond together. To do so, it must be borne in mind that adhesion quality is highly dependent on the 3D printing parameters employed.

The mechanism for adhesion when working with polymers may be physical, chemical, and/or mechanical. In the case of thermoplastics, adhesion is caused by weak intermolecular interactions between surfaces and not by strong chemical bonds

(crosslinking). Their macromolecule chains are entangled, as these weak intermolecular forces can easily be broken by applying heat. Owing to this, their chains move over one another with a snake-like motion (reptation motion) when heated. In addition to the intermolecular forces, mechanical interlocking can also be involved in the overall adhesion phenomenon [27]. Thermosetting polymers, which include chemical adhesion mechanisms, will not be discussed here.

Most of the literature considers diffusion and entanglement as the only adhesion mechanism when working with thermoplastics [2,5,13,30]. While it is true that thermoplastics generally have high molecular weights and diffusion plays a very important role in their layer adhesion, it is also necessary to take into account all other adhesion mechanisms. Others have demonstrated that, in some cases, mechanical interlocking plays a more important role in layer adhesion than diffusion and entanglement [31]. This said,

11 and to the best of our knowledge, the adhesion mechanism of diffusion and entanglement plays the most important role in this study. Neither adsorption nor mechanical interlocking theory can provide a satisfactory interpretation of the increasing adhesion strength with time (i.e., hysteresis), and only the diffusion theory easily explains such dependence by the slow penetration of the macromolecules across the interface [30].

Next, the adhesion mechanisms that usually play a role in thermoplastic printing have been presented.

2.3.1 Diffusion and entanglement

In general, molecular interdiffusion and entanglement from one material to another across the interface will be the main adhesion mechanism since thermoplastics usually have high molecular weights and are not cross-linked. In its most abstract sense, diffusion is the spontaneous thermal motion of soluble molecules in each other caused by a difference in molecular concentration between different phases. Ultimately, the system wants to be completely balanced with a uniform concentration. Regarding polymer diffusion, surfaces brought into contact must be mutually soluble and be above their glass-transition temperature (Tg) or crystalline melting temperature (Tc) to achieve sufficient molecular mobility. In this way, polymeric chains can diffuse, intermingle, and tangle into one another across the interface. In particular, the increase in temperature weakens the secondary bonding (bonds between chains) via the thermal vibration of the long molecules which, in turn, enables more free movement of adjacent chains [24]. Thus, thermoplastic molecules migrate from the bulk towards the interface (interphase) as a result of micro-

Brownian motion, thereby establishing an entangled molecular network.

12 The diffusion and entanglement formation of polymer chains determines much of the adhesion strength. In other words, adhesion depends on the number of molecular chain segments that have penetrated the interface, and on the length and density of the entanglements formed. Moreover, penetration depth and time are two correlated factors that greatly influence adhesion quality. The higher the values, the stronger the adhesion. In an optimum adhesion between surfaces, a failure will occur due to the breaking of molecular bonds instead of the disentanglement of polymer chains. Indeed, adhesion at the interface should be as imperceptibly as possible and contain as many molecular segments as possible of the bonded bodies. It should also be noted that there are many factors involved in the diffusion and entanglement mechanism. Temperature, contact time, molecular weight, degree of crystallinity, the strength of polar groups - just to mention a few - are very influential factors. For instance, the higher the temperature, the greater the mobility and diffusion capacity of molecule chains. In contrast, high polarity and/or crystallinity normally induces lower molecular mobility and thus inhibits diffusion processes [29]. Furthermore, contact time (hysteresis) between surfaces must be long enough to achieve sufficient diffusion and entanglement. Yet, the timeframe for diffusion processes can be expected to be very low [29]. This diffusion time also has a limit above which the mechanism shows no improvement.

The diffusion probability of the free ends (i.e., end segments) of macromolecules into a bulk polymer is usually more than that of the middle segments [30]. In this way, as the molecular weight (chain length) of a polymer increases, the number of end segments of its macromolecules decreases, thereby hampering partial diffusion. Moreover, the length and density of entanglements also increase, which reduces the degree of molecular mobility

13 and thus the diffusion tendency. Here, adhesion strength improves and worsens at the same time, as diffusion is hindered but entanglement formation is enhanced. On the other hand, there is also a critical minimum molecular weight below which chains are too short for sufficient diffusion and entanglement processes [29]. Therefore, increasing or decreasing molecular weight does not have to imply an improvement or deterioration in adhesion strength. Indeed, there is an ideal molecular weight for each polymer which optimizes bonding strength. This ideal weight can be found by examining correlations between molecular weight and adhesion strength of thermoplastics.

Typically, Fick’s laws of diffusion have been used to mathematically describe the diffusion process. However, thermoplastics usually have very long, linear, and entangled macromolecules. Hence, they require more energy during the diffusion process than materials with shorter and less entangled molecules, that is, with fewer mobility restrictions. This explains why Fick’s laws of diffusion cannot be directly applied to describe a thermoplastic’s tendency to diffuse and diffusion speed in entangled systems

[32]. Instead, reptation theory [33,34] must be employed. The latter describes the dynamics of entangled polymers melts. More specifically, it explicates the dependence of the mobility of a macromolecule on its length [11–13]. It does so through the concept of reptation relaxation time Tr and its relationship with molecular weight, which justifies why shorter chains diffuse faster than longer chains. Theoretically, if the contact time between surfaces is higher than the reptation time, then the adhesion strength at the interface will reach the cohesion strength of the bulk.

Figure 4. Schematic illustration of polymer chains inter-diffusion across an interface.

2.3.2 Mechanical interlocking (anchoring)

The basic idea behind this type of adhesion is that molecules fill the micro-voids, crevices, and pores of the surfaces in contact and hold them together by interlocking as they harden. This process can take place on different length scales. For instance, to understand more accurately the nature of this mechanism, one can imagine how velcro works. The latter is a good example to visualize this mechanical process from a macroscopic viewpoint. Contrarily, in the context of thermoplastics, adhesion is produced by micromechanical bonds, that is, occurs on a microscale. These bonds are caused by the mechanical interlocking of molecules with the roughness and irregularities of the surfaces in contact. Thus, some thermoplastics could exhibit poor adhesion in that they do not easily stick onto surfaces unless mechanical effects intervene.

The effectiveness of micromechanical adhesion depends on the surface wetting, the surface roughness, the material rheological properties, and the electrostatic forces, if any, between surfaces in contact [27]. In addition, trapped air in the interface must be avoided

15 at all costs by good surface wetting, as it undermines the bond strength. Last, it is important to bear in mind that this theory is not universally applicable, since good adhesion also takes place between smooth surfaces [26].

Figure 5. Schematic illustration of the mechanical interlocking mechanism between two surfaces.

2.3.3 Dispersive or adsorptive adhesion

Dispersive adhesion is caused by weak secondary bonding forces, called van der

Waal forces, developed between molecules in close vicinity. These molecular attractive interactions result from molecular dipoles (i.e., polar molecules) and are often classified according to the nature of the interacting dipoles [35]. Thus, they can be divided into

Keesom forces, London dispersion forces, Debye forces, and hydrogen bonding. Keesom forces, also known as dipole-dipole interactions, occur due to permanent dipoles.

Contrarily, London dispersion forces are due to temporary (instantaneous) dipoles between nonpolar adjacent molecules. These latter are present in all polymers and depend on the polarizability of the molecules. In particular, thermoplastics can have significant London dispersion forces because larger molecules are often more polarizable. On the other hand,

Debye forces take place between a permanent and an induced dipole. Lastly, hydrogen bonding is the strongest of the four interactions and occurs between molecules when a hydrogen atom is bonded to an atom of high electronegativity (e.g., F, O, or N), resulting in a polarization of the bond.

16 Although not being the main thermoplastic adhesion mechanism, adsorption theory has its importance due to its presence in every type of adhesion process. However, unlike primary chemical bonds (i.e., covalent and ionic bonds), the explained intermolecular forces are relatively weak as they do not require high amounts of energy to break. Indeed, the real work of adhesion is several magnitudes higher than that expected from molecular forces [30]. Furthermore, close proximity of polymer chains is a fundamental requirement, as these interactions only take place over very small distances. About 99% of the work required to break van der Waals bonds is performed once the joined surfaces are separated by more than a nanometer and, as a result, the effectiveness of adhesion due to dispersive bonding is limited [27]. Hence, intimate molecular contact between surfaces is absolutely necessary.

Figure 6. Schematic illustration of a) dipole-dipole interactions and b) dipole-induced dipole interaction.

2.3.4 Electrostatic interactions (contact charging)

Although not very common, an electrical double layer can sometimes form across the adhesion plane as a result of a charge transfer. Thus, part of the adhesion between surfaces could occur due to electrostatic attractions induced by these potential differences.

This is a reasonable explanation for polymer-metal adhesion bonds. Nevertheless, such attraction forces only act over very small distances and estimates of the energy associated with this process are generally small compared with adhesion fracture energies [35].

17 Moreover, this mechanism fails to explain adhesion between similar materials, since electrostatic attractions would be close to zero [29]. Consequently, when working with thermoplastics, electrostatic interactions between surfaces are usually disregarded.

Figure 7. Schematic illustration of the contact charging phenomenon.

2.4 Bond formation process among thermoplastic printed layers

The bond formation process between the extruded polymer filament and the preceding printed layer can be divided into 4 steps [36]:

• Step 1: Surface rearrangement

When a polymer filament passes through the nozzle, it experiences relatively high shear stress. Inside the nozzle, this shear stress aligns polymer chains that are in a very low entropy state; they have very few degrees of freedom. Once the material exits the nozzle, forces acting on it are relieved, causing the well-known die swell phenomenon [2]. Hence, this shear stress may affect the phase domains, if any, of the extruded surface of thermoplastic blends. It is important to note molecular orientation close to the surface of printed layers is usually parallel to the adhesion interface (i.e., aligned towards the flux direction), whereas bulk molecules are less oriented [29]. This alignment of molecular chains facilitates the formation of PVDF crystals.

18 The interaction between a filament and the nozzle wall can not only affect the superficial distribution of the extruded material but also the weight of its molecular chains

(i.e., chain scission). Thus, in case molecular chains are being cut, the polymers that will diffuse on this interface could be weaker than the rest of the part, that is, if a failure occurs, it would be adhesive instead of cohesive. This approach is very important because drives the rest of the process.

• Step 2: Surface approach or contacting

This approach sets the boundary conditions for the wetting and diffusion approaches, which will be explained below. It tells how many points of contact exists and if the FDM process is happening on a flat or rounded surface. For instance, a print can be staggered, which implies that the extruded filament interacts with two filaments of the previously printed layer. In contrast, the extruded filament of an aligned print only interacts with one filament of the preceding printed layer (see Figure 8.a).

It is important to highlight that the first printed layer will be the surface approach of the extruded material on the build plate. There will be no diffusion happening, however, wetting will occur. Hence, to summarize, the extruded filament can be printed staggered, aligned, or on the build plate.

• Step 3: Wetting

As already mentioned, the higher the surface energy of a material, the better the wetting and hence the adhesion (diffusion) at the interface (see Figure 8.b). Therefore, the use of PMMA in the PVDF/PMMA blends is essential due to the low surface energy of

PVDF, which makes direct printing with it extremely difficult. Since fluoropolymers have

19 low surface energy, they do not easily wet surfaces. Fluorine is the most electronegative element known. It is one electron away from being a noble gas and it is the smallest of the halogens. Hence, the carbon-fluorine bond in PVDF, disrupts further molecular interaction.

Conversely, PMMA has high surface energy and is paritally miscible with PVDF [8–10], thus helping to blend and increasing overall surface energy, improving wetting.

It bears mentioning that earlier authors attempted to relate adhesion to wetting and incorrectly approached the bonding as a thermodynamically reversible phenomenon [30].

Thus, rather than relating adhesion to wetting, the testing of bond strength by analyzing the type of failure through the determination of the type of separation is a more practical approach for quantifying adhesion bond strength [24].

• Step 4: Diffusion and randomization

This last step is a multivariable step that occurs as long as the filament temperature is higher than the glass transition temperature of the thermoplastic. Diffusion is a function of the first three steps. It depends on wetting which, at the same time, depends on the existing phase domains and the contact area. The dependency of diffusion on wetting is that a good wetting implies that the polymer chains can diffuse and randomize a little bit more efficiently because the contact area is larger (see Figure 8.c).

The diffusion dependency on phase domains is that depending on the type of polymer domains there is, there will be a better or worse diffusion. For instance, it is known that PMMA will diffuse into itself and PVDF relatively well. However, in the case of

PVDF, the diffusion will not be as good. Therefore, by extruding at different shears, it can be known if changing the shear stress of the nozzle wall affects the resulting phase domains

20 if any. Lastly, the dependency of diffusion on the surface approach is that one can get more contact area depending on how many points of contact exists during wetting.

Figure 8. Bond formation process between the extruded polymer filament and the preceding printed layer. (a) surface contact, (b) wetting and, (c) diffusion and randomization.

2.5 Theoretical model of the bonding formation process

Bellehumeur and colleagues proposed a Newtonian polymer-sintering model Eq.

(2), which is a modification of Frenkel’s model, which can be used to help probe interface diffusion. In particular, it predicts the neck growth formation (wetting) between polymer filaments as a nonlinear function of time, t, material viscosity, , initial filament radius,

, and surface tension, . The sintering model is as follows [5]:휇

푎표 훾

(2) −5/3 1/3 푑휃 훾 2 푐표푠휃푠푖푛휃(2 − 푐표푠휃) −1 푦 = 1/3 , 푤ℎ푒푟푒 휃 = 푠푖푛 푑푡 푎표휇 (1 − 푐표푠휃)(1 + 푐표푠휃) 푎 where is the neck radius, ao represents the initial radius of the filament, and a denotes the radius of푦 the filament at a specific time t (see Figure 8). Among other assumptions, Eq. (2)

21 is limited to describing a process at constant temperature and viscosity, whereas these parameters will typically vary throughout the FDM process [2]. It can be inferred from this model that to achieve good wettability (i.e., increase y) thermoplastics should have high surface tension ( ) and low viscosity (µ). Thus, this model helps to visualize the diffusion between layers by훾 providing some information about the mobility of the molecules at the interface. It is important to note that, although this model provides information about the degree of wetting achieved at the interface of the filaments, it does not give direct information about the interface diffusion.

The sintering model was used in conjunction with a 1D transient heat transfer model

Eq. (3) which allows determining the time at which filament temperature dropped below the polymer glass transition temperature and thus chain motion ceased. This heat transfer model is as follows [5]:

(3) 휕푇 휕푇 휕(푘 ) 휌퐶퐴 = 퐴 휕푥 − ℎ푃(푇 − 푇∞), 0 < 푥 < 푣푡, 푡 > 0 휕푡 휕푥 where the term h covers the effects of both heat convection with air and conduction with foundation, A is the filament cross-section area, P is the filament cross-section perimeter, ρ is the polymer density, k is the polymer thermal conductivity, C is the polymer heat capacity, and v is the velocity of the print nozzle (x=vt).

22 CHAPTER 3 MATERIALS AND METHODS

3.1 Materials description

Both polymers used in this study were obtained from commercial sources and dried at 50°C under vacuum for 8 hours to remove any moisture before use. Poly(vinylidene fluoride) (PVDF) high flow homopolymer grade Kynar 705 with an average molecular weight of 82,700 g/mol was generously donated by Arkema and arrived in pellet form.

Poly(methyl methacrylate) (PMMA) with an average molecular weight of 141,000 g/mol was purchased from A ALDRICH Chemistry and arrived in powder form. “PMMAXX” is the nomenclature structure for samples used in this study where “XX” indicates PMMA content. For instance, a sample made with 30 wt% PMMA and 70 wt% PVDF is designated as “PMMA30”.

3.1.1 Poly(vinylidene fluoride) (PVDF)

PVDF is a fluorinated semicrystalline thermoplastic produced by the polymerization of vinylidene difluoride. It has a glass-transition temperature of approximately -40°C, a relatively low melting point of around 177°C, and a low density of

1.78 g/cm3. Depending on its conformation, PVDF can crystalize into five different crystalline phases (polymorphs): , β, , and [37,38]. The first three phases ( , β, ) are the most commonly studied,훼 with훾 the훿 -phase휀 being the most thermodynamically훼 훾 preferred [39]. When PVDF crystallizes into훼 β and/or phases, it exhibits piezoelectric properties. 훾

23 Finally, PVDF can be used as an oxidizer, as a sensor, or actuator due to its piezoelectric properties, as a hydrophobic material due to its low surface energy, etc.

3.1.2 Poly(methyl methacrylate) (PMMA)

PMMA, also known as acrylic, is an amorphous transparent thermoplastic produced by the polymerization of methyl methacrylate. Its glass-transition temperature is approximately 118°C, which means that at room temperature it is an organic glass. Thus,

PMMA has a brittle behavior in response to applied loads and relatively high strength due to its extremely high molecular mass. Manufactured products of PMMA are characterized by strength, rigidity, high surface glossiness, and scratch resistance.

Although PMMA does not have a clear melting point, it softens when heated to around 160℃ and starts to flow at around 180℃. Furthermore, it has a low density of 1.188 g/cm3 at room temperature (25°C).

3.2 Blend preparation

Before compounding, both thermoplastics were dried at 50°C under vacuum for 8 hours to remove any moisture. Then, blends were prepared by melt blending PVDF and

PMMA in a twin-screw extruder (Compounder HAAKE Minilab II, Thermo Scientific).

Compounding was carried out at 190℃ and a screw speed of 100 rpm for 5 minutes. Lastly, different PVDF/PMMA blends were extruded at 100 rpm to form filament approximately

3 mm in diameter, containing 0%, 15%, 25%, 50%, 75%, 85%, and 100% of PMMA (by weight).

24 3.3 Apparent viscosity measurements

Blends apparent viscosity was measured by monitoring the two pressure transducers located in the inside wall of the compounder chamber. Pressure data (60 values) was taken at each transductor for 1 minute at a screw speed of 100 rpm using the

HAAKE minilab software. In this way, viscosity calculations were made using the pressure difference between the two transducers. Apparent viscosity can be calculated as follows:

(4) 휏 휂 = where denotes apparent viscosity, represents훾̇ shear rate, and τ is the shear stress at the wall. The휂 latter is given by: 훾̇

(5) ℎ푤∆푃 휏 = where h and w are the height and width2(ℎ of + the 푤)∆퐿 channel (i.e., 1.5 and 10 mm, respectively),

ΔL is the distance between the pressure transducers, and ΔP is the pressure differential.

Similarly, the shear rate can be expressed as:

(6) 2 (ℎ + 푤)푤 훾̇ = where w represents screw speed. 60∆퐿

3.4 Dog-bone production

HAAKE Type 3 specimen (557-2290) dog-bone tensile bars were produced via

FDM and injection molding. All samples of each composition were fabricated under identical processing conditions to eliminate sample-to-sample errors.

25 3.4.1 Fused deposition modeling (FDM)

PVDF/PMMA melt-extruded filaments were around 0.5-0.7 m in length and 2.9-

3.1 mm in diameter. These filaments were used in an Ultimaker 2 to create the 3D printed dog-bones. Solidworks 2019 and Simplify3D were used to model the dog-bones prior to the printing process. Printing of the specimens was done on a glass build plate heated to

90℃ through a 0.4 mm nozzle heated to 190-210℃. For blends from PMMA0 to

PMMA25, the print nozzle was heated to 190℃. For blends from PMMA50 to PMMA100, the print nozzle was heated to 210℃. The layers were printed with 100% infill density, with an extrusion multiplier of 1, with high quality, with a layer height of 0.1 mm, at a print speed of 30 mm/s, and a x/y movement speed of 55 mm/s. Unsurprisingly, PMMA0 was the most difficult mixture to 3D print due to its low surface energy (i.e., layering not settling right away once cooling began). Consequently, a 3D printing adhesive was used to make prints stick more firmly.

3.4.2 Injection molding (IM)

PVDF/PMMA melt-extruded filaments were pelletized and introduced into a preheated cylinder (210°C). A HAAKE Minijet Pro (Thermo Scientific) was then used to form the dog-bone via injection molding. The cylinder was placed on top of a heated mold

(90°C) and fitted with a plunger. The plunger was pushed by a hydraulic press to extrude the molten blend into the closed mold. For blends from PMMA0 to PMMA25, the press was programmed to deliver an initial pressure of 420 bar for 20s, and a secondary pressure of 200 bar for 15s before releasing. For blends from PMMA50 to PMMA100, the press was programmed to deliver an initial pressure of 600 bar for 40s, and a secondary pressure

26 of 250 bar for 20s before releasing. Upon completion, the mold was then removed from the machine and opened to reveal the injected dog-bone.

When injection molding PMMA (or blends with high wt% in PMMA), there exists a wide range of adjustable injection temperatures (usually between 180 to 240°C) since its decomposition temperature is around 270°C [40]. Furthermore, this material is hygroscopic and has high viscosity (i.e., poor fluidity) making it difficult to injection mold. As a result, during this process, high temperature and injection pressure are required. It is important to note that, although the improvement of fluidity through injection temperature is greater, the increase of injection pressure helps to decrease specimen shrinkage. Regarding the mold, its temperature should be raised (usually between 60 to 90°C) to improve the solidification process, since PMMA has poor impact and abrasion resistance, and scratches and breaks easily. Simultaneously, the mold temperature may also significantly influence

PVDF crystallization, either positively or negatively (see section 2.1.3).

3.5 Performed experiments

3.5.1 Tensile testing

Bulk tensile testing of dog-bones was carried out using an Instron 5569 Tensile

Tester with a 50 kN load cell. A dog-bone was placed in a pair of manual dual screw press clamps which were torqued to 30 ft ·lb before testing. The displacement of the sample was measured using the crosshead of the machine frame. Tests were performed at room temperature using a crosshead speed of 50 mm/min, while axial strain was continuously recorded. The Young´s modulus (E) was calculated from the initial linear portion of the stress-strain curves. Maximum tensile stress (σ) and elongation at break (ε) were also reported. At least eight dog-bones specimens were tested for each composition and the

27 average values are presented. The software used to measure and calculate stress-strain data was Bluehill Universal.

The main objective is to examine the mechanical response of 3D printed thermoplastics (multiple interface specimens) and of thermoplastics fabricated via injection molding (homogeneous specimens) to study adhesive versus cohesive failure mechanism (i.e., study the adhesion/diffusion quality).

Figure 9. Schematic illustration of dog-bone fabrication for tensile testing.

3.5.2 Thermal analysis

Thermal analysis was carried out using a TA Instruments model Q20 differential scanning calorimeter (DSC). PVDF/PMMA samples of approximately 5 mg were placed into an alumina crucible and loaded into the DSC. Thereupon, these samples were heated at a rate of 20 ºC/min, from 30 to 250 °C, and then cooled to 30 °C under a nitrogen atmosphere. This process was carried out twice (two cycles). The glass-transition temperature (Tg), melting temperature (Tm), crystallization temperature (Tc), and heat of melting (∆Hm) were determined for the different processed blends. To assess the impact of

28 PMMA on PVDF crystalline domains, the degree of crystallinity (Xc) of each sample was calculated from the following equation [41]:

(7) ∆퐻푚 푐 ∗ 푋 (%) = 푚 100 where ∆Hm is the experimental melting∆퐻 enthalpy휙 of the blend (obtained through DSC analysis), represents the theoretical melting enthalpy for 100% crystalline PVDF (i.e., ∗ 푚 104.5 J/g ∆퐻[42–44]), and is the weight fraction of PVDF. DSC analysis was performed from the derivative of heat휙 flow with respect to temperature. The derivative was used to establish the initial and final points of the exothermic/endothermic peaks.

As already mentioned, crystalline domains affect interface adhesion by preventing the diffusion of polymer chains. Thus, DSC was used to locate when crystallization enhancement occurs.

3.5.3 Attenuated total reflectance spectroscopy (ATR-FTIR)

Attenuated total reflectance ATR sampling technique was used in conjunction with infrared spectroscopy to determine the presence and relative amount of PVDF β-phase within the blends. Thus, to identify the PVDF crystalline phases of PVDF/PMMA mixtures, absorption spectra were collected using a Thermo Scientific FTIR-ATR Nicolet iS20. The analysis was performed at room temperature in the range of 500 to 3000 cm-1 averaged from 32 scans with a spectral resolution of 2 cm-1. A background spectrum was collected before each sample scan.

3.5.4 Gel permeation chromatography (GPC)

The basic idea behind this chromatographic method is that the polymer in solution passes through columns that absorb and filter it, while a detector measures the size of its

29 molecular chains over time (see Figure 10). The average size corresponds to the polymer molecular weight (Mw). The number of molecules per size corresponds to the molecular weight distribution. It is worth noting that, apart from the aforementioned effect that Mw has on diffusion, most polymer properties also depend on the molecular weight and its distribution.

Figure 10. Schematic illustration of GPC analysis.

The molecular weight and polydispersity of PMMA were determined via size exclusion chromatography (SEC). GPC was conducted using a Waters Alliance 2695 separations module, an online multiangle laser light scattering (MALLS) detector fitted with a gallium arsenide laser (20 mW) operating at 658 nm, an interferometric refractometer operating at 65 °C, and 685 nm, and two PLgel mixed-E columns in series

(pore size 50-103 Å, 5 μm bead size). The mobile phase was freshly distilled tetrahydrofuran (THF) delivered at a flow rate of 1.0 mL/min. Sample concentrations were

6 mg of polymer/mL of THF, and the injection volume was 100 μL. The detector signals

30 were simultaneously recorded using ASTRA software, and absolute molecular weights were determined by MALLS using a dn/dc value obtained from the interferometric refractometer response, assuming 100% mass recovery from the columns.

For PVDF, the molecular weight and polydispersity were also determined via the same procedure. However, the mobile phase was 0.02 M LiBr/DMF (dimethylformamide) delivered at a flow rate of 0.5 mL/min. Furthermore, the polymer samples were pre- dissolved in 0.02 M LiBr/DMF by stirring for 3 h at 80 °C. The injection volume was set at 100 μL.

31 CHAPTER 4 RESULTS AND DISCUSSION

4.1 Apparent viscosity measurements

As shown in Figure 11, a strong rise in apparent viscosity can be observed with increasing blends PMMA content. This increasing trend in viscosity reflects the high molecular weight (Mw) of PMMA, which causes the melt blending process to consume more energy due to high PMMA Mw (i.e., fluid has a higher difficulty flowing inside the compounder chamber relative to shear rate). In contrast, the relatively low molecular weight of PVDF causes it to act as a plasticizer and soften mixtures. More precisely, at low

PMMA concentrations, the viscoelastic behavior suggests the blends have a high degree of partial miscibility, whereas at PMMA50 a more significant phase separation of the two polymers can be inferred. For higher PMMA concentrations, blends exhibit responses resembling pure PMMA.

Figure 11. Viscosity response of PVDF/PMMA blends during compounding as a function of PMMA content. Error bars represent the standard error of the mean values.

32 4.2 Tensile testing

Figure 12. Mechanical properties of injection molded (IM) and 3D printed (3D) PVDF/PMMA dog-bones as a function of PMMA content: (a) tensile strength, (b) elongation at break, and (c) Young’s modulus. In each panel, error bars represent the standard error of the mean values of each mechanical property.

33 Blends mechanical properties as a function of weight fraction of PMMA are displayed in Figure 12. 3D printed specimens show a similar trend as injection molded

(IM) ones with respect to tensile strength and Young's modulus as the PMMA weight percent increases. The overall increase in tensile strength by having more PMMA content in the mixture can be explained by dividing the mechanical response into three ranges. The first range is from PMMA0 to PMMA25, the second one is PMMA50, and the third one is from PMMA75 to PMMA100. In the first range, crystallization slowly decreases (see section 4.3) but the addition of PMMA, which has a much higher overall tensile strength response than PVDF, toughens the material. By looking at the tensile strength responses of the two processing techniques, it can be observed how these responses get closer together while the overall tensile strength increases with PMMA content. This decrease in the differences between the tensile strength response of IM and 3D printed samples is due to

PMMA acting as an adhesive. The latter enables higher mobility and diffusion of polymer chains at lower temperatures owing to its high surface energy, which raises blends overall surface energy. Therefore, this adhesive mechanism helps to increase the bond formation of 3D-printed samples.

At PMMA50, a notable deterioration of mechanical properties can be observed. In particular, the tensile strength response of both IM and 3D printed specimens drops off due to loss of crystallization. As explained in section 4.3, PMMA50 is the limit concentration beyond which the PVDF crystallization event is lost. The PMMA concentration is too high to allow for PVDF crystallization, which increases the tensile strength in the polymer. More specifically, PMMA chains impede PVDF crystallization due to the high glass-transition temperature (Tg) of PMMA, which causes PVDF chains to not have enough space to move

34 and form crystals. This loss of the crystallization impact on mechanical strength is responsible for the drop in PMMA50 tensile strength value. Note that the latter is not a bond formation issue because the drop in mechanical properties occurs for both processed samples. Thus, what changes as a function of the composition (i.e., it is not processing technique dependent) is the fact that PMMA50 is no longer able to crystallize and, hence, it has lost part of its ability to bear a load.

From PMMA75 to PMMA100, the increase in PMMA content raises tensile strength basically because the overall PMMA tensile strength response is much higher than that of PVDF. In this way, the mechanism for improving this mechanical property is simply the fact that PMMA itself has a higher tensile strength than PVDF. Within this range, blends tensile strength responses are approaching the PMMA response. However, the difference between the tensile strength response of IM and 3D printed specimens is noticeably greater than it is for lower PMMA concentrations. These differences are greater because blends glass-transition temperatures have notably increased close to the value of virgin PMMA Tg (i.e., 118 °C). The relatively high glass-transition of PMMA creates diffusion limitations as the PMMA network becomes rigid below Tg. These limitations hinder domain evolution at the printed filament boundary, effectively seizing polymer welds at high temperatures thus limiting the degree of diffusion from layer to layer.

Concerning the elongation at break, a similar but decreasing trend can be observed with the addition of PMMA weight ratios. Starting from PMMA0, ductility decreases as the PMMA content increases. The same drop as in tensile strength analysis can then be seen at PMMA50. This drop in ductility can be explained by the fact that PMMA50 samples can only withstand a much lower load for a shorter time than the other bends due

35 to the loss in crystallization. Hence, they cannot be stretched as long as the other compositions do. From PMMA75 to PMMA100, ductility follows the same decreasing trend that it had before PMMA50 until reaching PMMA100. High PMMA concentrations exhibit brittle behavior in response to applied loads mainly because they have high Mw and were tested below their Tg (see next section).

Finally, Young's modulus trend matches the trend with tensile strength as it slowly increases proportionally to the amount of PMMA. In addition, the same drop in property can be observed at PMMA50. This intermediate composition does not have the crystallization impacts that help to increase Young’s modulus and does not have enough

PMMA content either to allow having Young's modulus close to that of PMMA. Unlike the differences of injection molded and 3D printed samples with respect to tensile strength and elongation properties, Young’s modulus has very small changes between the two types of production. The latter can be explained because Young’s modulus is a material- dependent property, whereas the former properties depend more on production. This material dependence can be visualized in that even if the geometry of the sample were to be changed, Young's modulus would not change. In contrast, the first two properties, which are more process-dependent, are very sensitive to voids and weld quality (i.e., stress concentrations and interlayer adhesion effects). As explained later in the GPC and FTIR analysis results, the internal properties (i.e., Mw and crystalline phases) of the material have not changed and remain fairly constant between processed specimens.

36 4.3 Thermal analysis

Differential scanning calorimetry (DSC) was performed to study the effect of the addition of PMMA on the thermal behavior of both processed thermoplastic and their respective mixtures. The DSC results of the investigated PVDF/PMMA blends are listed in Table 1, 2, and 3. Moreover, the heat flow curves of the blends are represented in Figure

13, 14, and 15. It can be observed from the second heating cycles that as the PMMA concentration increases, the melt temperature, the melting enthalpy, and the degree of crystallinity decrease markedly for all processed blends. This behavior is caused by the partial miscibility of PVDF in PMMA, as the PMMA is largely amorphous and does not contribute to the heat of fusion [8,43]. Thus, the increase in PMMA content disrupts PVDF crystalline domains. More specifically, PMMA chains impede PVDF crystallization due to the high glass-transition temperature (Tg) of PMMA, below which the overall chain mobility significantly decreases by forcing PVDF chains to diffuse through the rigid

PMMA network to form crystals, thus altering melting behavior. In this way, crystallization indicates that there is still mobility in the system (i.e., interdiffusion of PMMA chains through amorphous PVDF regions). These results suggest that PMMA50 is the limiting concentration beyond which the PVDF crystallization event is lost.

As mentioned in the previous section, the addition of PMMA raises the glass- transition temperature of the mixtures (see Table 1, 2, and 3). Thus, by controlling and maintaining the temperature between the glass transition of PMMA and the melt temperature of PVDF, blends diffusion and crystallization may be optimized [9].

Moreover, pure PVDF exhibits a single endothermic peak around 173 °C which could be

37 ascribed to the presence of α-phase crystals [8,43]. This was confirmed by the results shown in Figure 17.

The thermal behavior of injection molded blends presents a similar profile with that of 3D printed and compounded (filament). However, the melt temperature of the PMMA15 concentration increases with respect to pure PVDF. This shift could be ascribed to the increase of α-phase crystals of PVDF. Similarly, the degree of crystallinity of the PMMA25 concentration increases with respect to PMMA15, which could be attributed to the slight increase of β-phase content. This was confirmed by the results shown in Figure 17.

Furthermore, it can be seen from the second heating of Table 3 that low PMMA injection molded samples experienced significantly less overall crystallinity than the same concentrations of the other processing techniques.

The above reasonings are extrapolated from the second heating-cooling cycle of the

DSC analysis because the first cycle was used as an annealing process to relieve part of the residual stresses induced during production (i.e., it is not representative of the blend composition). Differences between the crystallization-melting behavior of the first and second cycles are processing-related phenomena and not composition-related phenomena.

In fact, the changes observed between the first and second cycles could simply be a polymer chain readjustment to relieve the internal stresses (i.e., residual stress relaxation) caused during production. Hence, the first DSC cycle shows all the processing effects while the effects in the second cycle are more representative of the material composition. That is, samples act more like virgin material in the second cycle, whereas some of the interactions and phase transitions in the first cycle (and some of the mechanical responses) could simply be a residual stress relaxation issue.

38 The surface of the material experiences relatively high shear stress during production resulting in internal residual stress. Inside the nozzle, this shear stress aligns polymer chains that are in a very low entropy state; they have very few degrees of freedom.

Ideally, once the material exits the nozzle, forces acting on it are relieved. However, if the material cools too quickly, the polymer chains do not have enough time to reorder at the surface, resulting in strain-induced crystallization or other types of stress phenomena. This shear stress is more impactful for high PMMA concentrations due to their high glass- transition temperature, which hinders blends readjustment. Moreover, the high surface energy of PMMA increases the tacticity (i.e., adhesive properties) of the system, facilitating interactions between the polymer and nozzle (or mold) surfaces. Therefore, both events demonstrate that higher PMMA concentrations add more shear stress to the material. Similarly, 3D printed components are more like to experience higher residual stresses and process effects than the other two processes based on the surface area that is formed during production (i.e., surface area to volume ratio).

It is then clear that in the 50% PMMA concentration composition does not promote crystallization, processing does by the strain-induced crystallization (see Table 1, 2, and

3).

Lastly, it is important to note that the Flory-Fox Equation for Tg calculation of polymer blends does not apply in this case due to the approx. 160 °C difference between the PVDF and PMMA glass transition temperatures.

Figure 13. DSC traces for 3D printed samples undergoing the (a) first heating cycle and

(b) second heating cycle. Data are summarized in Table 1.

Figure 14. DSC traces for twin-screw extruded samples (filament) undergoing the (c) first heating cycle and (d) second heating cycle. Data are summarized in Table 2.

Figure 15. DSC traces for injection molded samples undergoing the (e) first heating cycle and (f) second heating cycle. Data are summarized in Table 3.

40 Table 1. Glass transition (Tg), melting temperature (Tm), melting enthalpy (Hm), and crystallization (Xc) for the first and second heating cycles of various 3D printed PVDF/PMMA compositions.

3D Printed First Heating Second Heating Tg (°C) Tm (°C) Hm (J/g) Xc Tg (°C) Tm (°C) Hm (J/g) Xc PMMA0 164.2 45.0 43% 157.9 57.8 55% PMMA15 53.6 157.0 35.5 40% 157.0 44.6 50% PMMA25 56.5 159.5 28.8 37% 151.9 36.7 47% PMMA50 60.9 146.4 16.5 32% 0% PMMA75 88.4 0% 94.7 0% PMMA85 98.1 0% 105.5 0% PMMA100 111.4 0% 118.0 0%

Filament First Heating Second Heating Tg (°C) Tm (°C) Hm (J/g) Xc Tg (°C) Tm (°C) Hm (J/g) Xc PMMA0 161.0 38.4 37% 158.1 54.4 52% PMMA15 56.9 159.7 30.2 34% 158.0 41.8 47% PMMA25 60.7 157.4 21.0 27% 153.7 35.4 45% PMMA50 62.9 156.0 0.6 1% 0% PMMA75 98.6 0% 98.5 0% PMMA85 110.2 0% 104.8 0% PMMA100 111.9 0% 117.8 0%

Table 3. Glass transition (Tg), melting temperature (Tm), melting enthalpy (Hm), and crystallization (Xc) for the first and second heating cycles of various injection molded PVDF/PMMA compositions.

IM First Heating Second Heating Tg (°C) Tm (°C) Hm (J/g) Xc Tg (°C) Tm (°C) Hm (J/g) Xc PMMA0 166.3 37.8 36% 158.6 44.4 42% PMMA15 57.5 162.1 30.6 34% 159.1 34.5 39% PMMA25 60.4 163.7 32.1 41% 152.8 31.7 40% PMMA50 61.7 151.1 11.9 23% 149.1 2.0 4% PMMA75 85.6 0% 92.9 0% PMMA85 101.3 0% 105.9 0% PMMA100 105.5 0% 115.3 0%

41 4.4 Attenuated total reflectance spectroscopy (ATR-FTIR)

To study the phase domains of PVDF/PMMA mixtures (i.e., PVDF crystalline phases), processed compositions underwent ATR-FTIR analysis. The latter measures the infrared spectrum of absorption of the sample over a range of wavelengths (see Figure 16).

In this way, absorbance spectra allow the determination of the morphology of the polymer blend by observing which bonds and phases are present in the material. Therefore, PVDF characteristic peaks, as well as PMMA frequency bands of interest, need to be examined.

Figure 16 shows the infrared spectra of the different PVDF/PMMA processed blends from 600 to 1800 cm-1. As previously mentioned, PVDF can crystalize into five different crystalline phases (polymorphs): , β, , and [37,38]. The first three phases

( , β, ) are the most commonly studied,훼 훾 훿 with 휀 the -phase being the most thermodynamically훼 훾 preferred [39]. Only the representative peaks훼 of - (614, 762, 795, 855, and 976 cm-1) and β- (840 and 1279 cm-1 essentially) phases are present훼 in the spectrum of blends. More specifically, only concentrations up to 25% in PMMA content display these peaks. Beyond this limit concentration, and β peaks diminish in intensity or vanish, which agrees with the crystallinity results discussed훼 in the previous section. Moreover, pure

PMMA displays at 1720 cm-1 the stretching frequency of its carbonyl group (C=O).

However, for blends containing both PVDF and PMMA, this absorption band is shifted to higher wavelengths (1727-1729 cm-1) due to the interaction between the carbonyl groups of PMMA and the CH2 groups of PVDF, which indicates the formation of blends [43,45].

It can also be observed that blends containing high PMMA content display at 990 cm-1 the bending frequency of CH3-O [46].

Figure 16. FTIR spectra of PVDF/PMMA blends processed by (a) injection-molding, (b) 3D printing and (c) twin-screw extrusion (filament).

43 Investigation of the crystalline domains present in each mixture provides information about the diffusion environment. For instance, if PMMA interacts strongly with PVDF through its carbonyl group, then the β-phase should be more dominant than the

-phase. Consequently, the higher the β-phase fraction, the greater the intermolecular interaction훼 between PVDF and PMMA, which will subsequently negatively affect diffusion. Alternatively, if it is mainly the -phase that is present within the blends, there will be fewer PVDF-PMMA interactions and훼 thus a higher possibility of diffusion across the polymer interfaces. Thus, the β-phase content of blends containing only and β polymorphs can be calculated using the following equation [47]: 훼

(8) 퐴훽 퐹(훽) = 훽 × 100 퐾 훼 훽 훼 퐴 + 퐴 where F(β) represents the amount of퐾 β-phase; Aα and Aβ are the absorbance intensities for

-1 - and β-phase at 762 and 840 cm , respectively; Kα and Kβ are the absorption coefficients

4 2 4 2 at훼 the respective wavenumbers (i.e., Kα = 6.1·10 cm /mol and Kβ = 7.7·10 cm /mol) [48].

Surprisingly, the results presented in Figure 17 show a decrease in the relative fraction of β-phase for the processed samples as the PMMA content increases, which runs counter to the findings of the existing literature [8]. Only the transition from PMMA15 to

PMMA25 of the injection-molded sample shows a slight increase in the β-phase content.

Still, these trends are consistent with the previously reviewed results. Hence, the increase in PMMA content not only worsens PVDF crystallization but also hinders nucleation of the β-phase PVDF polymorph. The total amount of β-phase in absolute terms can be calculated by combining these results with the DSC crystallinity calculations.

Figure 17. β-phase percent ratio in PVDF/PMMA processed blends as a function of PMMA content.

It can be noted that injection-molded specimens present higher absorbance intensities than 3D-printed specimens, which can be explained by the level of transparency of the samples. As shown in Figure 16, injection-molded sample is more translucent than a 3D printed one, which is opaquer due to its interfaces. Thus, when the machine detects the reflected infrared light from the multilayer sample, this return light is more scattered, and the detector cannot capture it as well as with the injection-molded sample. Even so, this issue does not affect the determination of the β-phase content since Eq. (8) works with relative amounts of the same spectrum. It is worth mentioning that ATR-FTIR spectroscopy was used instead of FTIR transmission spectroscopy because samples were very opaque, and light could not pass through properly. Reflectance is a superficial measurement, while transmittance is a bulk measurement.

45 4.5 Gel permeation chromatography (GPC)

GPC was run to study the processing effects of injection molding, 3D printing, and twin-screw extrusion on the molecular weight Mw and polydispersity of PMMA and PVDF.

GPC curves are displayed in Figure 18, while its number and weight averaged molecular weights and dispersity values are shown in Table 4-5. From GPC curves it can be observed that the molecular weight (Mw) is higher in PMMA than PVDF since its narrow peak occurs early in the range of 10-14 min whereas the PVDF peak takes place later between 24 and

30 min. Moreover, PVDF exhibits a lower intensity peak around 40 min, which suggests that its low Mw chains act as a plasticizer and soften it. Therefore, PVDF chains require less energy (i.e., temperature) to experience motion than PMMA chains. In this way, PVDF chains can diffuse into PMMA and reduce large chain interactions with one another, improving blend plasticity.

Figure 18. GPC curves of PMMA and PVDF for virgin (pellet) and processed (filament, injection-molded, and 3D printed) samples. Data are summarized in Tables 4 and 5.

46 Results show that both processed thermoplastics did not experience degradation because they maintain a fairly constant molecular weight distribution throughout the eluded time for each production technique. This can also be inferred from the data presented in

Table 4-5, as the number and weight average molecular weights values are within an acceptable range of variation for PMMA and PVDF. Therefore, for the tensile test performed, differences between the responses of IM and 3D printed specimens can be argued from a diffusion point of view (i.e., no change in material properties). Similarly, it can be verified that the thermal analysis was carried out in the appropriate temperature range because there is no change in thermal behavior due to the decomposition of polymer chains. It is important to note that degradation would imply that polymer chains are being cut (chain scissoring) during blends production and hence there would be a clear difference in the molecular weight distribution of the different samples.

Table 4. Number ( ) and weight ( ) averaged molecular weights and dispersity index (Ð) for virgin and processed PMMA samples. ̅푀̅̅푛̅ 푀̅̅̅̅푤̅ Sample (g/mol) (g/mol) Dispersity (Ð = ) Virgin 푴̅8.95̅̅풏̅ x 104 푴̅̅1.41̅̅풘̅ x 105 1.57 (± 2.21 %) (± 1.33 %) (± 2.58̅푴̅̅̅풘̅⁄ %)푴̅̅̅ 풏̅

Filament 9.08 x 104 1.46 x 105 1.61

(± 1.37 %) (± 1.08 %) (± 1.74 %) PMMA

(Aldrich) Injection 8.29 x 104 1.33 x 105 1.59

Molded (± 4.76 %) (± 2.12 %) (± 5.21 %)

3D Printed 8.62 x 104 1.37 x 105 1.62 (± 3.71 %) (± 1.73 %) (± 4.09 %)

47 Table 5. Number ( ) and weight ( ) averaged molecular weights and dispersity index (Ð) for virgin and processed PVDF samples. ̅푀̅̅푛̅ 푀̅̅̅̅푤̅ Sample (g/mol) (g/mol) Dispersity (Ð = ) Virgin 푴̅4.57̅̅풏̅ x 104 푴̅̅8.27̅̅풘̅ x 104 1.81 (± 7.11 %) (± 1.19 %) (± 7.21̅푴̅̅̅풘̅⁄ %)푴̅̅̅ 풏̅

Filament 4.41 x 104 8.24 x 104 1.87 PVDF (± 9.72 %) (± 1.77 %) (± 9.88 %) (K705) Injection 4.13 x 104 8.25 x 104 1.99 Molded (± 8.38 %) (± 1.31 %) (± 8.49 %)

3D Printed 4.09 x 104 8.02 x 104 1.96 (± 8.66 %) (± 1.27 %) (± 8.75 %)

48 CHAPTER 5 CONCLUSIONS

In this study, adhesion mechanisms involved in thermoplastic printing were reviewed. Influencing factors such as wetting, contact angle, surface tension, surface energy, crystallinity, and adhesion and transition zones were also introduced. To the best of our knowledge, the adhesion mechanism of diffusion and entanglement plays the most important role in PVDF/PMMA printing. Neither adsorption nor mechanical interlocking theory can provide a satisfactory interpretation of the increasing adhesion strength with time (i.e., hysteresis), and only the diffusion theory easily explains such dependence by the slow penetration of the macromolecules across the interface [30].

Therefore, PVDF/PMMA thermoplastic welding is considered to be the result of the diffusion and entanglement mechanism.

By blending PVDF with PMMA, the overall surface energy increased, as well as viscosity, which allowed the production of the fluoropolymer through FDM. Although qualitatively an increase in apparent viscosity should worsen wetting (based on the bond formation model explained (Eq. (2)), it was concluded that the increase in global surface energy produced by the addition of PMMA played a more dominant role than the increase in viscosity.

PVDF/PMMA dog-bone specimens were fabricated via FDM and injection molding. Specimen mechanical and thermal properties were evaluated by tensile testing and DSC, respectively. Tensile testing results proved PMMA to be the strongest material with the highest Young's modulus. In contrast, results revealed that PVDF has the highest

49 elongation at break. Mechanical properties of PVDF/PMMA blends varied almost linearly within these two limiting cases imposed by the two materials. When PMMA content ratios increased for each mixing ratio, the strength increased as well as Young's Modulus, whereas elongation at break decreased. However, PMMA50 was demonstrated to differ from this trend to be the weakest and most brittle composition due to the loss of crystallization. Low PMMA concentrations exhibited plastic mechanical properties, whereas high PMMA concentrations behaved mainly elastically from the observed elapsed time to take before fracture. Fracture of high PMMA blends had a clean break and longer force exposure (before fracture) than lower PMMA blends. Even though high PMMA concentrations had higher solid quality in final prints, high PVDF concentrations had a malleable feature allowing for necking of the specimens once tested on the Instron.

Unsurprisingly, injection molding proved to be the processing technique with the highest mechanical performance for all tensile strengths recorded. Elongation results showed that dog-bones processed through FDM had more brittle structures than injection- molded ones. Differences between injection-molded and 3D printed specimens (i.e., bulk mechanical anisotropy of 3D printed components) provided information on molecular diffusion through the layer interface. It was found that low PMMA blends had interactions with the chain mobility, that is, chains could pass through amorphous PVDF before diffusing into one another. It can be visualized as a diffusion mechanism competition where

PMMA can diffuse into itself or PVDF. In contrast, it was found that at high PMMA concentrations, PMMA was able to diffuse mainly into itself, and, hence, diffusion worsened. These results help to assess the impact of PMMA and the FDM process on

PVDF/PMMA blends behavior and processability.

50 DSC results revealed the impact of PMMA on blends crystallinity ratio. The increase in PMMA content disrupted PVDF crystalline domains. The results suggested that

PMMA50 is the limiting concentration beyond which the PVDF crystallization event is lost. Furthermore, PVDF crystalline phases were examined using the ATR-FTIR technique. PVDF/PMMA characteristic absorption bands that were present in FTIR spectrums were identified by comparison with PVDF/PMMA FTIR studies from existing literature. Results revealed that for high concentrations of PVDF, an increase in PMMA content hindered nucleation of the β-phase PVDF polymorph.

GPC results showed that both processed polymers did not experience degradation as specimens examined maintained a fairly constant molecular weight distribution throughout the eluded time.

Finally, this work provides insight into the multiscale mechanical behavior of

PVDF/PMMA structures processed through FDM. Being able to 3D print PVDF and understand its behavior will allow sensing equipment such as sensors and actuators to be

3D printed. Hence, interesting technical advances can be made in different fields by understanding these fundamental behaviors slightly better.

51 CHAPTER 6 FUTURE WORK

It is interesting to mention that Weibull's weakest link model [49] could have been addressed in this study. Weakest link theory states that the survival probability of a solid is the product of the survival probabilities of each element within the solid [50]. In the context of this work, and because the cohesion of printed materials is generally stronger than their interlayer adhesion, it could be argued that the height of the printed layers could play a very important role in overall adhesion strength. This is because the more layers are printed, the higher the probabilities of failure due to not achieving the desired adhesion. In this way, this fascinating topic could be studied as future work.

Future work will investigate events that alter the adhesion property based on the layer interface. One event could be the effect of print orientation between layers on weld quality. Another one could be the effect of stress concentrations, created due to voids and cracks presents in 3D printed specimens, on bonding quality. Through the application of fracture mechanics, a better analysis could incorporate the effects of stress concentrations on the adhesion strength of the samples.

Although crystallization hinders diffusion, it provides piezoelectric properties to the PVDF/PMMA blends. Thus, the next step for this work could be to learn to control diffusion and crystallization phenomena simultaneously (i.e., manufacturing aspects versus crystallization aspects). For example, to examine what type of sensors can be 3D printed while maintaining good processability. The final goal could be to understand how processing effects change the piezoelectric properties of PVDF.

52 The results of the thermal analysis revealed that it would be promising to reproduce some of the mixtures and anneal them for various amounts of time to relieve residual internal stresses introduced during manufacture. In this way, the stress-strain response of the annealed specimens would be measured and compared with the findings obtained in this work. Annealing samples to remove internal stresses could enhance mechanical performance during tensile tests, as well as improve mechanical durability.

Finally, a computation model of diffusion could be developed to compare experimental results to theoretical predictions and to customize specific properties for design.

53 REFERENCES

[1] S.S. Babu, L. Love, R. Dehoff, W. Peter, T.R. Watkins, S. Pannala, Additive manufacturing of materials: Opportunities and challenges, MRS Bull. 40 (2015) 1154–1161. https://doi.org/10.1557/mrs.2015.234. [2] D.P. Cole, J.C. Riddick, H.M. Iftekhar Jaim, K.E. Strawhecker, N.E. Zander, Interfacial mechanical behavior of 3D printed ABS, J. Appl. Polym. Sci. 133 (2016) 1–12. https://doi.org/10.1002/app.43671. [3] J.M. Arenas, J.J. Narbón, C. Alía, Optimum adhesive thickness in structural adhesives joints using statistical techniques based on Weibull distribution, Int. J. Adhes. Adhes. 30 (2010) 160–165. https://doi.org/10.1016/j.ijadhadh.2009.12.003. [4] B.K. Larson, L.T. Drzal, P. Sorousian, Carbon fibre-cement adhesion in carbon fibre reinforced cement composites, Composites. 21 (1990) 205–215. https://doi.org/10.1016/0010-4361(90)90235-O. [5] C. Bellehumeur, L. Li, Q. Sun, P. Gu, Modeling of bond formation between polymer filaments in the fused deposition modeling proceBellehumeur, C., Li, L., Sun, Q., & Gu, P. (2004). Modeling of bond formation between polymer filaments in the fused deposition modeling process. Journal of Manufac, J. Manuf. Process. 6 (2004) 170– 178. https://doi.org/10.1016/S1526-6125(04)70071-7. [6] H.A. Miller, B.S. Kusel, S.T. Danielson, J.W. Neat, E.K. Avjian, S.N. Pierson, S.M. Budy, D.W. Ball, S.T. Iacono, S.C. Kettwich, Metastable nanostructured metallized fluoropolymer composites for energetics, J. Mater. Chem. A. 1 (2013) 7050–7058. https://doi.org/10.1039/c3ta11603d. [7] S.K. Valluri, M. Schoenitz, E. Dreizin, Fluorine-containing oxidizers for metal fuels in energetic formulations, Def. Technol. 15 (2019) 1–22. https://doi.org/10.1016/j.dt.2018.06.001. [8] S. Aid, A. Eddhahak, S. Khelladi, Z. Ortega, S. Chaabani, A. Tcharkhtchi, On the miscibility of PVDF/PMMA polymer blends: Thermodynamics, experimental and numerical investigations, Polym. Test. 73 (2019) 222–231. https://doi.org/10.1016/j.polymertesting.2018.11.036. [9] J.A. Bencomo, S.T. Iacono, J. McCollum, 3D printing multifunctional fluorinated nanocomposites: Tuning electroactivity, rheology and chemical reactivity, J. Mater. Chem. A. 6 (2018) 12308–12315. https://doi.org/10.1039/c8ta00127h. [10] S.J. Kang, Y.J. Park, I. Bae, K.J. Kim, H.C. Kim, S. Bauer, E.L. Thomas, C. Park, Printable ferroelectric PVDF/PMMA blend films with ultralow roughness for low voltage non-volatile polymer memory, Adv. Funct. Mater. 19 (2009) 2812–2818. https://doi.org/10.1002/adfm.200900589. [11] M. Bousmina, H. Qiu, M. Grmela, J.E. Klemberg-Sapieha, Diffusion at polymer/polymer interfaces probed by rheological tools, Macromolecules. 31 (1998) 8273–8280. https://doi.org/10.1021/ma980562r.

54 [12] R.J. Composto, E.J. Kramer, D.M. White, Reptation in polymer blends, Polymer (Guildf). 31 (1990) 2320–2328. https://doi.org/10.1016/0032-3861(90)90319-T. [13] A. Indrakanti, N. Ramesh, J.L. Duda, S.K. Kumar, Modeling diffusion in miscible polymer blend films, J. Chem. Phys. 121 (2004) 546–553. https://doi.org/10.1063/1.1760078. [14] J.N. Israelachvili, Intermolecular and surface forces, Third Edit, Academic press, 2011. [15] I. Moutinho, M. Figueiredo, P. Ferreira, Evaluating the surface energy of laboratory- made paper sheets by contact angle measurements, Tappi J. 6 (2007) 26–32. [16] J. Izdebska, Printing on Polymers: Theory and Practice, Elsevier Inc., 2015. https://doi.org/10.1016/B978-0-323-37468-2.00001-4. [17] R.G. Larson, Y. Wei, A review of thixotropy and its rheological modeling, J. Rheol. (N. Y. N. Y). 63 (2019) 477–501. https://doi.org/10.1122/1.5055031. [18] G. Bracco, B. Holst, Surface science techniques, 2013. https://doi.org/10.1007/978- 3-642-34243-1. [19] J.P. Allain, M. Echeverry-Rendón, Surface treatment of metallic biomaterials in contact with blood to enhance hemocompatibility, Elsevier Ltd., 2018. https://doi.org/10.1016/B978-0-08-100497-5.00008-2. [20] D.L. Williams, T.M. O’Bryon, Cleanliness Verification on Large Surfaces: Instilling Confidence in Contact Angle Techniques, Elsevier, 2013. https://doi.org/10.1016/B978-1-4377-7879-3.00005-4. [21] Y. Thomas, An essay on the cohesion of fluids, Philos. Trans. R. Soc. (1805) 65– 87. https://doi.org/10.4324/9781315761183-4. [22] S. Rossi, G. Gai, R. De Benedetto, Functional and perceptive aspects of non-stick coatings for cookware, Mater. Des. 53 (2014) 782–790. https://doi.org/10.1016/j.matdes.2013.07.079. [23] D.J. Burnett, F. Thielmann, R.A. Ryntz, Correlating thermodynamic and mechanical adhesion phenomena for thermoplastic polyolefins, J. Coatings Technol. Res. 4 (2007) 211–215. https://doi.org/10.1007/s11998-007-9017-0. [24] K.M. Praveen, C.V. Pious, S. Thomas, Y. Grohens, Relevance of Plasma Processing on Polymeric Materials and Interfaces, Elsevier Inc., 2019. https://doi.org/10.1016/b978-0-12-813152-7.00001-9. [25] J.J. Bikerman, Causes of Poor Adhesion: Weak Boundary Layers, Ind. Eng. Chem. 59 (1967) 40–44. https://doi.org/10.1021/ie51403a010. [26] S. Ebnesajjad, A.H. Landrock, Introduction and Adhesion Theories, Adhes. Technol. Handb. (2015) 1–18. https://doi.org/10.1016/b978-0-323-35595-7.00001- 2.

55 [27] J.A. Von Fraunhofer, Adhesion and cohesion, Int. J. Dent. (2012). https://doi.org/10.1155/2012/951324. [28] E.M. Petrie, Adhesive bonding of textiles: Principles, types of adhesive and methods of use, 2013. https://doi.org/10.1533/9780857093967.2.225. [29] F. Mieth, M. Tromm, Multicomponent Technologies, 2016. https://doi.org/10.1016/B978-0-323-34100-4.00001-8. [30] V. Voyutskii, SS and Vakula, The role of diffusion phenomena in polymer‐to‐ polymer adhesion, J. Appl. Polym. Sci. 7 (1963) 475–491. https://doi.org/https://doi.org/10.1002/app.1963.070070207. [31] F. Tamburrino, S. Graziosi, M. Bordegoni, The influence of slicing parameters on the multi-material adhesion mechanisms of FDM printed parts: an exploratory study, Virtual Phys. Prototyp. 14 (2019) 316–332. https://doi.org/10.1080/17452759.2019.1607758. [32] H. Zhang, K. Lamnawar, A. Maazouz, Rheological modeling of the diffusion process and the interphase of symmetrical bilayers based on PVDF and PMMA with varying molecular weights, Rheol. Acta. 51 (2012) 691–711. https://doi.org/10.1007/s00397-012-0629-7. [33] P.G. De Gennes, Reptation of a polymer chain in the presence of fixed obstacles, J. Chem. Phys. 55 (1971) 572–579. https://doi.org/10.1063/1.1675789. [34] M. Doi, S. Edwards, The theory of polymer dynamics, Clarendon, Oxford. (1986) 189–234. https://doi.org/https://doi.org/10.1002/pol.1989.140270706. [35] D.E. Packham, Handbook of Adhesion Technology, 2017. https://doi.org/10.1007/978-3-319-42087-5. [36] R.P. Wool, K. M. OʹConnor, A theory crack healing in polymers, J. Appl. Phys. 52 (1981) 5953. [37] Z. Cui, N.T. Hassankiadeh, Y. Zhuang, E. Drioli, Y.M. Lee, Crystalline polymorphism in poly(vinylidenefluoride) membranes, Prog. Polym. Sci. 51 (2015) 94–126. https://doi.org/10.1016/j.progpolymsci.2015.07.007. [38] B.E. El Mohajir, N. Heymans, Changes in structural and mechanical behaviour of PVDF with processing and thermomechanical treatments. 1. Change in structure, Polymer (Guildf). 42 (2001) 5661–5667. https://doi.org/10.1016/S0032- 3861(01)00064-7. [39] F.C. Chiu, C.C. Chen, Y.J. Chen, Binary and ternary nanocomposites based on PVDF, PMMA, and PVDF/PMMA blends: Polymorphism, thermal, and rheological properties, J. Polym. Res. 21 (2014). https://doi.org/10.1007/s10965-014-0378-7. [40] Y.H. Hu, C.Y. Chen, C.C. Wang, Viscoelastic properties and thermal degradation kinetics of silica/PMMA nanocomposites, Polym. Degrad. Stab. 84 (2004) 545–553. https://doi.org/10.1016/j.polymdegradstab.2004.02.001.

[41] L. Wang, S. Chen, Crystallization behaviors of poly(vinylidene fluoride) and poly(methyl methacrylate)-block-poly(2-vinyl pyridine) block copolymer blends, J. Therm. Anal. Calorim. 125 (2016) 215–230. https://doi.org/10.1007/s10973-016- 5364-3. [42] W. Ma, J. Zhang, X. Wang, Crystallizaion and surface morphology of poly(vinylidene fluoride)/poly(methylmethacrylate) films by solution casting on different substrates, Appl. Surf. Sci. 254 (2008) 2947–2954. https://doi.org/10.1016/j.apsusc.2007.10.037. [43] N.A.H. I.S. Elashmawi, Effect of PMMA addition on characterization and morphology of PVDF, Polym. Eng. Sci. (2008) 1–10. https://doi.org/10.1002/pen.21032. [44] J.H.K. Minho Lee, Taesang Koo, Seongho Lee, Byong Hun Min, Morphology and physical properties of nanocomposites based on poly(methyl methacrylate)/poly(vinylidene fluoride) blends and multiwalled carbon nanotubes minho, Polym. Compos. 16 (2015) 101–113. https://doi.org/10.1002/pc. [45] W. Li, H. Li, Y.M. Zhang, Preparation and investigation of PVDF/PMMA/TiO2 composite film, J. Mater. Sci. 44 (2009) 2977–2984. https://doi.org/10.1007/s10853-009-3395-x. [46] X. Zhao, S. Chen, J. Zhang, W. Zhang, X. Wang, Crystallization of PVDF in the PVDF/PMMA blends precipitated from their non-solvents: Special orientation behavior, morphology, and thermal properties, J. Cryst. Growth. 328 (2011) 74–80. https://doi.org/10.1016/j.jcrysgro.2011.06.036. [47] R. Gregorio,, M. Cestari, Effect of crystallization temperature on the crystalline phase content and morphology of poly(vinylidene fluoride), J. Polym. Sci. Part B Polym. Phys. 32 (1994) 859–870. https://doi.org/10.1002/polb.1994.090320509. [48] P. Martins, A.C. Lopes, S. Lanceros-Mendez, Electroactive phases of poly(vinylidene fluoride): Determination, processing and applications, Prog. Polym. Sci. 39 (2014) 683–706. https://doi.org/10.1016/j.progpolymsci.2013.07.006. [49] W. Weibull, A statistical theory of strength of materials, IVB-Handl. (1939). [50] F.W. Zok, On weakest link theory and Weibull statistics, J. Am. Ceram. Soc. 100 (2017) 1265–1268. https://doi.org/10.1111/jace.14665.