CHARACTERIZATION OF COMPOSITES FABRICATED FROM DISCONTINUOUS RANDOM CARBON FIBER THERMOPLASTIC MATRIX SHEETS PRODUCED BY A PAPER MAKING PROCESS By Martin Paul Ducote, Jr. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Mechanical Engineering Ð Master of Science 2015 ABSTRACT CHARACTERIZATION OF COMPOSITES FABRICATED FROM DISCONTINUOUS RANDOM CARBON FIBER THERMOPLASTIC MATRIX SHEETS PRODUCED BY A PAPER MAKING PROCESS By Martin Paul Ducote, Jr. In this thesis, a papermaking process was used to create two randomly oriented, high performance composite material systems. The primary objective of this was to discover the flexural properties of both composite systems and compare those to reported results from other studies. In addition, the process was evaluated for producing quality , randomly oriented composite panels. Thermoplastic polymers have t he toughness and necessary strength to be alternatives to thermosets, but with the promise of lower cycle times and increased recyclability. The wet -lay papermaking process used in this study produces a quality, randomly oriented thermoplastic composite at low cycle times and simple production. The materials chosen represent high performance thermoplastic s and carbon fibers. Short chopped carbon fiber filled Nylon 6,6 and PEEK composites were created at varying fiber volume fractions. Ten nylon based panels and five PEEK based panels were subjected to 4 -point flexural testing. In several of the nylon -based panels, flexural testing was done in multiple direction to verify the in -plane isotropy of the final composite. The flexural strength performance of both systems showed promise when compared to equivalent products currently available. The flexural modulus results were less than expected and further research should be done into possibly causes. Overall, this research gives good insight into two high performa nce engineering composites and should aid in continued work. Copyright by MARTIN PAUL DUCOTE, JR. 2015 iv ACKNOWLEDGMENTS There were many people involved with this process and I am grateful to everyone who helped. I would like to say thanks to Dr. Dahsin Liu and Dr. Mahmood Haq for serving on my thesis committee. Mr. Timothy Hinds for allowing me to work as a teaching assistant and lea rn from him. The department of mechanical e ngineering and everyone involved with my study while at Michigan State University. I would like to say a special thanks to Tim Luchini and Steve Sommerlot for always taking time to h elp me whenever I asked, Colby W oods for his endless support, and finally Dr. Alfred Loos for being an excellent mentor and professor. v TABLE OF CONTENTS LIST OF TABLES ........................................................................................................................ vii LIST OF FIGURES ..................................................................................................................... viii KEY TO SYMBOLS AND ABBREVIATIONS ........................................................................... x Chapter 1. Introduction ................................................................................................................... 1 Chapter 2. Literature Review .......................................................................................................... 4 2.1. Thermoplastics ..................................................................................................................... 4 2.2. Common Thermoplastic Polymers ....................................................................................... 5 2.2.1 Nylon 6,6 ........................................................................................................................ 6 2.2.2 PEEK .............................................................................................................................. 6 2.3. Carbon Fiber Reinforcements .............................................................................................. 7 2.3.1 General Structure ........................................................................................................... 8 2.3.2 Manufacturing Carbon Fibers ........................................................................................ 8 2.4. Strength of Thermoplastics Based Composites .................................................................... 9 2.5. Manufacturing with Thermoplastic s .................................................................................. 10 2.5.1 Pre -Melt Mixing ........................................................................................................... 11 2.5.2 Melt Mixing ................................................................................................................. 13 2.5.3 Low Viscosity Precursors ............................................................................................ 14 2.6. Compression Molding of Thermoplastic Composites ........................................................ 15 2.7. Tensile Strength versus Flexural Strength ......................................................................... 16 2.8. Nylon Composite Data from Literature ............................................................................. 17 2.9. PEEK Composite Data from Literature ............................................................................. 18 2.10. Other Comparable C omposite Strengths ......................................................................... 20 2.11. Calculating Volume Fraction from Weight Fractions ..................................................... 20 2.12. Scanning Electron Microscopy ........................................................................................ 21 Chapter 3. Panels and Materials Tested ........................................................................................ 23 3.1. Panels Tested ..................................................................................................................... 23 3.2. Materials Used ................................................................................................................... 24 3.2.1 Reinforcing Fibers ....................................................................................................... 24 3.2.2 Thermoplastic Matrix ................................................................................................... 25 3.2.3 White Water ................................................................................................................. 25 3.2.3.1 Thickener .............................................................................................................. 25 3.2.3.2 Surfactant .............................................................................................................. 26 3.2.3.3 Anti -Foam ............................................................................................................. 26 3.2.4 Mold Release ............................................................................................................... 26 vi Chapter 4. Production of Test Composite Plates .......................................................................... 28 4.1. White Water ....................................................................................................................... 28 4.2. Mixing ................................................................................................................................ 29 4.3. Water Removal ................................................................................................................... 30 4.4. Drying ................................................................................................................................ 32 4.5. Molding Preparation ......................................................................................................... 33 4.6. Compression Molding ........................................................................................................ 34 Chapter 5. Testing ......................................................................................................................... 37 5.1. Flexural Testing ................................................................................................................. 37 5.2. Adjusted Fiber Volume Fractions ...................................................................................... 38 5.3. Processing .......................................................................................................................... 39 5.4. Microscopy ......................................................................................................................... 40 Chapter 6. Results ......................................................................................................................... 41 6.1. Fiber volume fraction ........................................................................................................ 41 6.2. Flexural Strength ............................................................................................................... 42 6.3. Flexural Modulus ............................................................................................................... 44 6.4. Planar Isotropy .................................................................................................................. 47 6.5. Microscopy ......................................................................................................................... 48 Chapter 7. Discussion ................................................................................................................... 54 7.1. Fiber Volume Fraction ...................................................................................................... 54 7.2. Comparison of Flexural Strength ...................................................................................... 54 7.3. Comparison of Flexural Modulus ...................................................................................... 57 7.4. Planar Isotropy .................................................................................................................. 60 7.5. Microscopy ......................................................................................................................... 61 7.6. Manufacturing Process ...................................................................................................... 63 7.6.1 Surfactan t in White Water ........................................................................................... 63 7.6.2 Separation of Carbon Fiber .......................................................................................... 64 7.6.3 Inconsistent Drying of Mats ......................................................................................... 65 7.6.4 Cutting testing coupons ................................................................................................ 67 Chapter 8. Conclusion ................................................................................................................... 68 8.1. Conclusions ........................................................................................................................ 68 8.2. Future Work ....................................................................................................................... 69 APPENDIC ES .............................................................................................................................. 71 Appendix 1: Individual Sample Results from Flexural Testing ................................................ 72 Appendix 2: MATLAB code ...................................................................................................... 74 BIBLIOGRAPHY ......................................................................................................................... 79 vii LIST OF TABLES Table 2.1: Nylon based composite data from the literature .......................................................... 18 Table 2.2: PEEK based composite data from the literature .......................................................... 19 Table 3.1: Nylon based panels produced for testing ..................................................................... 24 Table 3.2: PEEK based panels produced for t esting ..................................................................... 24 Table 4.1 Materials used for white water ...................................................................................... 29 Table 5.1: Specifications of flexural testing ................................................................................. 38 Table 5.2: Sample coupons used for SEM microscopy ................................................................ 40 Table 6.1: Predicted versus final calculated fiber volume fraction for nylon based panels ......... 41 Table 6.2: Target and adjusted fiber volume fract ions for PEEK based plates ............................ 42 Table 6.3: Flexural strength results for nylon based panels .......................................................... 43 Table 6.4: Flexural strength results for PEEK based panels ......................................................... 44 Table 6.5: Flexural modulus results for nylon based panels ......................................................... 45 Table 6.6: Flexural modulus results for PEEK based panels ........................................................ 46 Table 6.7: Flexural testing results for panel 007 ........................................................................... 48 Table A.1: Flexural strength data for PEEK panels tested. All values are in MPa. ..................... 72 Table A.2: Fl exural modulus data for nylon panels tested. All values in GPa. ............................ 72 Table A.3: Flexural strength data for nylon panels tested. All values are in MPa. ...................... 73 Table A.4: Flexural modulus data for nylon panels tested. All values in GPa. ............................ 73 viii LIST OF FIGURES Figure 4 -1: Overview of production method used in panel production ........................................ 28 Figure 4 -2: FORMAX maelstrom 8 L pulper ............................................................................... 30 Figure 4 -3: Formax sheet former. Shown with deckle box and couching plate rotated up .......... 31 Figure 4 -4: Thermoplastic and carbon fibers mixed with white water in deckle box .................. 32 Figure 4 -5: Fibrous mat after couching step ................................................................................. 32 Figure 4 -6: Steel compression mold ............................................................................................. 34 Figure 4 -7: PHI press used in molding nylon based panels .......................................................... 35 Figure 4 -8: TMP press used for molding PEEK based panels ...................................................... 36 Figure 6-1: Typical stress -strain curve from flexural testing. Diamonds and circles represent slope locations for modulus calculations. Stars represent peak strength. ................. 42 Figure 6 -2: Flexural strength results for Nylon based panels ....................................................... 43 Figure 6 -3: Flexural strength results for PEEK based panels ....................................................... 44 Figure 6 -4: Flexural modulus results for nylon based panels ....................................................... 46 Figure 6 -5: Flexural modulus results for PEEK based panels ...................................................... 47 Figure 6 -6: Photomicrograph of panel 012 coupon 06 (Nylon). Surface pit circled in black. ..... 49 Figure 6 -7: Photomicrograph of panel 13 coupon 1 (PEEK) ....................................................... 50 Figure 6 -8: Photomicrograph of panel 007 coupon 1 (Nylon) ..................................................... 50 Figure 6 -9: Photomicrograph of panel 007 coupon 1 (Nylon) ..................................................... 51 Figure 6 -10: Photomicrograph of panel 023 coupon 02 (PEEK) ................................................. 52 Figure 6 -11: Photomicrograph from panel 013 coupon 01 (PEEK) ............................................. 52 Figure 6 -12: Photomicrograph of panel 007 coupon 01 (Nylon) ................................................. 53 ix Figure 6 -13: Photomicrograph of Panel 004 coupon 6 (Nylon). Shows failure surface (washed out scale bar reads 100 µm) ....................................................................................... 53 Figure 7 -1: Comparison of Nylon based panelÕs flexural strength in current study with literature results. Source of information showed in brackets. Dotted line references current study. ......................................................................................................................... 55 Figure 7 -2: Comparison of PEEK based panelÕs flexural strength in current study with literature results. Source of information showed in brack ets. Dotted line references current study. ......................................................................................................................... 56 Figure 7 -3: Flexural strength comparision of current study to previous work by Lu and Caba at close fiber volume fra ction. References are in brackets. ........................................... 57 Figure 7 -4: Comparison of Nylon based panelÕs flexural modulus in current study with literature results. Source of informati on showed in brackets. Dotted line references current study. ......................................................................................................................... 58 Figure 7 -5: Comparison of PEEK based panelÕs flexural modulus in current study with literature results. Source of information showed in brackets. Dotted line references current study .......................................................................................................................... 59 Figure 7 -6: Flexural Modulus comparision of current study to previous work by Lu and Caba at close fiber volume fraction. References are in brackets. ........................................... 60 Figure 7 -7: Flexural strength results from individual coupons of Panel 007 ............................... 61! x KEY TO SYMBOLS AND ABBREVIATIONS d Specimen Depth (Length) D Deflection (Length) b Specimen Width (Length) !" Mass of Fibers (Mass) P Load (Force) S Support Span (Length) #" Fiber Volume Fraction (Dimensionless) #"$% Adjusted Fiber Volume Fraction (Dimensionless) &" Fiber Weight Fraction (Dimensionless) '" Flexural Strain (Length/Length) (" Fiber Density (Mass/Length 3) () Matrix Density (Mass/Length 3) *" Flexural Stress (Force/Length 2) 1 Chapter 1. Introduction The concept of composite materials is by no means new to engineering. Throughout history, there are numerous examples of combining materials to enhance the properties of one or both of them. This has continued all the way through to the steel reinforced co ncrete used in todayÕs bridges and structures. As a concept, the method of composite making is simple; combine two or more materials in the correct quantit y until the desired properties are accomplished [1]. However, in todayÕs ever evolving and complex world, the subject of composites continues to provide a substantial challenge for engineers. A composite material can be defined as ÒThe combination of a reinforcement material (such as a particle or fiber) in a matrix or binder material [1].Ó It is also important to understand that the matrix and binder materials remain identifiable on the macroscopic level and continue to have their own unique properties. Although the matrix material and fibers can be made from an infinite combination of materials, the focus of this research has been to observe fiber -reinforced thermo plastics , a subset of polymers . Specifically, these will be composites made using carbon -fiber reinforcement with a thermoplastic polymer matri x. The differences between thermoplastic and thermosetting polymers produces the interest to perform this research. The polymers used in composites can generally be separated into two broad categories : Thermoplastics and Thermosets . These tw o categories ar e separated by a fundamental difference in the curing process. Thermosets undergo a chemical cross -linking process while thermoplastics do not. This leads to a number of differences with regards to storage, process ing, and recycling . The crosslinking react ion in thermosets is increased by the addition of heat, which leads to refrigeration requirements for these polymers as well as limited shelf lives . In addition, the cross - 2 linking renders a chemically changed material that cannot be reversed. Thermosets ar e well known to have good tensile, shear , and compressive strength s but are lacking in damage tolerance and environmental stability [1Ð4]. The lack of a chemical curing step in thermoplastics give s them an unlimited shelf life , inherently good toughness c haracteristics that reduce initial damage and resist crack propagation, and the ability to be remolded after processing . In addition, high performance thermoplastics have characteristically stiff aromatic chains that produce high glass transition temperatu res. This allows the use of thermoplastics in high temperature designs where the rmosets would be inappropriate [1, 3]. For the reasons stated above, thermoplastics as a matrix material in composites are becoming more popular. The purpose of this thesis is to measure the flexural properties of a high performance engineering composite materials made using a wet -lay paper making process. The process was originally patented by James E. Geary, Jr. and Gregory P. Weeks in 1993 under the name ÒMethod of Making Fib er Reinforced Porous SheetsÓ (Patent # 5194106) [5]. This method has been adapted for the equipment available and is explained thoroughly in Chapter 4. The advantage of using a paper making process is the uniform porous mat of reinforcement and thermoplastic fibers that are produced that can then be compre ssion molded. This process was first investigated by Lu [6] with carbon and glass filled polyethylene terephthalate (PET ) and polypropylene (PP) , two common polymers . This thesis is a continuation of that research , using higher performance thermoplastics w ith the same process. The matrix materials used were Nylon 6,6 and PEEK, both of which are considered higher performance thermoplastics [1-3]. In this study, 15 composite panels were created using the wet -lay and compression molding techniques. These were then subjected to flexural testing to determine their 3 strength and modulus. Following that, selected tested samples were subjected to scanning electron microscopy. The entire process was also critiqued for sources of error and future recommendations for co ntinued research. The results showed promise for both the material systems and the process. The flexural strengths found are competitive with products currently found on the market. However, the flexural modulus values were found to be lower than expected and f urther research should be done into possible causes and recommendations for future work have been given. This thesis offers an in depth literature review, providing background information on thermoplastics, carbon fibers, manufacturing methods and sel ected data from similar studies. A detail of the materials used in this research as well as a list of composite panels follows. Chapters 4 and 5 detail the production and testing of the panels. Chapters 6 and 7 follow with results of testing as well as a d iscussion. Finally, Chapter 8 gives a conclusion and summary of recommendations future work. 4 Chapter 2. Literature Review The purpose of this section is to give a background of the concepts used in this thesis. This begins with an overview of thermop lastics and carbon fibers, the two materials used in these composites. More detailed information is given on the two thermoplastics used in this research: Nylon 6,6 and PEEK. An overview of common manufacturing techniques with thermoplastics is provided following the introductions of the materials. This includes pre -melt mixing, melt mixin g, and low viscosity precursors. As the process detailed in this thesis belongs to the latter, more focus is given to pre -melt mixing. Sections 2.4 to 2.6 report recent r esearch results that are of interest to the work done in the current study. This includes flexural testing data from carbon -nylon and carbon -PEEK composites as well as other appropriate composite systems. These results will be directly compared in the disc ussion in Chapter 7. Finally, in order to understand the testing methods being performed, brief introductions of flexural testing, fiber volume and weight fractions and scanning electron microscopy are given. These sections introduce concepts that were use d during the testing phase of this research. 2.1. Thermoplastics Thermoplastics differ with thermosets because they do not require a reactive cure step. As opposed to thermosets, which have chemical crosslinking, there are no chemical bonds that hold thermopla stic molecules together. Instead, secondary bonding, such as hydrogen bonds or van der Waals bonds, is present. Secondary bonds are easily broken by the addition of heat , allowing molecules to freely move around for a temporary amount of time. An in depth 5 discussion of this chemistry is outside the scope of this thesis. However, from a design and manufacturing standpoint, the production and mechanical performance of thermoplastic composites rely on the chemistry . In general, thermoplastics can be formed to suit whatever structural shape is required [3, 7, 8, 9]. For that reason, thermoplastics have been gaining popularity in the composites industry. As an example, i n 2007, 35% of composites were being produced with thermoplastics with the remaining still com ing from thermosets [10]. 2.2. Common Thermoplastic Polymers Thermoplastic polymers are known for characteristically good toughness and environmental resistance properties . There are 4 broad categories of thermoplastics: fully polymerized amorphous polymers, liquid crystalline polymers, extendable polymers, and semi -crystalline polymers. A background of common thermoplastic polymers is necessary because, the thermoplastics used in this study, Nylon 6,6 and PEEK, are both semi -crystalline polymers. As these semi -crystalline polymers are processed at elevated temperatures, some areas solidify in repeated ordered units that are crystalline. Amorphous regions , or areas where the polymer chains have a somewhat random arrangement , surround these pockets of crystallinity . There are advantages and disadvantages to the presence of these crystalline regions , often called crystallites [11, 12]. These regions can act as crosslinks betwee n polymer chains; protecting the polymer from solvents and increasing mechanical performance at elevated temperatures. However, these crystalline areas can also limit the energy absorption of the polymer, causing it to act more brittle . Generally, semi -cry stalline thermoplastics can be 5% to 50% crystalline with an optimum range being between 20% to 35% [3, 11]. 6 2.2.1 Nylon 6,6 Nylon 6,6 is a matrix material that belongs to the polyamides family of thermoplastics. Nylon 6,6, referred to as simply as nylon in remainder of this thesis, is one of the best known, accounting for 70% of the polyamides used today. Polyamides are commonly combined to form composites , with about 60% of them being filled with glass, carbon , or mineral s [10]. Generally, these n ylon based composites are used for automotive parts in the transportation industry. There are a few drawbacks that must be overcome when manufacturing with nylon. High water absorption, as much as 1 Ð 3%, means that composite parts must be protected fr om w eather and humidity [10]. Meanwhile, geometric tolerancing is difficult due to high shrinkage after molding and can lead to processing complications and costs associated with complex molds or post processing. The high shrinkage is due to its semi -crystalli ne nature and the development of the cryastallites during cooling. However, these drawbacks are manageable when the beneficial material properties such as low strains at yield and high elongations at failure are required . Neat nylon generally has a maximum density of 1.15 g/cm 3, a maximum yield stress of 85 MPa and a flexural modulus of 3.0 GPa [10]. 2.2.2 PEEK Polyether Ether Ketone (PEEK) is considered a high -performance engineering thermoplastic belonging to the Polyarylketone family. PEEK is attractive in hi gh temperature applications because of a 335 ¡C melting point and a 143 ¡C glass transition temperature ( Tg) [10]. Furthermore, PEEK has a continuous use temperature of 250 ¡C. Moisture content is less of an issue with PEEK when compared to some other thermop lastics because, it has low water absorption at less than 0.5% and is resistant to most solvents with the exception of methylene 7 chloride. In 2007, the global consumption of PEEK was between 2000 and 4000 tons per year with a majority being used in industr y and automotive applications [10]. Also, as a semi -crystalline polymer, PEEK thermoplastics have a crystallinity between 30% to 35% when cooled at a normal rate, avoiding a rapid cool from an amorphous state. However, with the introduction of fibers to PE EK resin systems, the crystallinity will tend to increase. The fracture toughness of PEEK is about 50 -100 times greater than standard thermosetting epoxys [3]. PEEK can have a density between 1.27 to 1.32 g/cm 3 and shrinkage of 1.1%. Drawbacks to working w ith PEEK include sensitivity to UV light and the excessive cost. As an example to other thermoplastics, PEEK can cost as much as 20 times more than nylon [10]. 2.3. Carbon Fiber Reinforcements Although composite materials can be made using a variety of fiber types and styles, carbon fibers have been considered the most promising for high performance materials. Carbon fibers can be produced from three different precursors: PAN (polyacrylonitrile) , Pitch, or Rayon. The different precursors, coupled with different manufacturing methods provide a wide variety of carbon fibers and performances providing tensile moduli ranging from 207 GPa to 1035 GPa [3]. The most common precursor used is PAN due to i ts relatively low cost. The properties of Pitch based fibers make it useful for applications with extreme temperature changes and harsh environments. The least used precursor is Rayon, which is only used in ablative functions [13]. 8 2.3.1 General Structure The h igh tensile modulus of carbon fibers comes from the graphitic organization of their carbon molecules. Carbon atoms are arranged in repeated hexagonal patterns that form planes. The planes are aligned in the longitudinal direction of the fiber, providing th e high tensile strength. The interactions and arrangement of the carbon rings, which are connected through weak van der Waals forces, govern the properties in the radial and circumferential directions. Several common structural types arise from the orien tations of these planes including circumferential, radial, or random. The orientations as well as the purity of the graphitization are all controlled by both the precursor and the manufacturing process used [3]. 2.3.2 Manufacturing Carbon Fibers Carbon fibers p roduced from PAN are the most commonly used and produced, accounting for 90% of all commercial carbon or graphite fibers [13]. The process begins with spinning a PAN solution into fibers. The fibers are then heated and stretched, aligning the filament chai ns and creating PAN filaments. The alternative to PAN filaments would be creating pitch filaments. Pitch is relatively low cost by -product of petroleum refining. However, the purifying and processing of these fibers can make them more expensive when compar ed to the PAN based fibers [13]. The pitch is heated to 300 ¡C - 500¡C and the hot spun into fibers and then stabilized by a second heat treatment [3]. Once a filament of either pitch or PAN has been made, the next stage is carbonization. This involves stre tching the fibers while heating in an inert atmosphere to 1000 ¡C - 2000¡C. This process removes the oxygen and nitrogen atoms, leaving only carbon atoms arranged in the hexagonal plane patterns previously discussed. However, there is no order between paral lel 9 planes, giving these fibers high strength, but relatively low -modulus, between 200 GPa and 300 GPa. The final stage for the fibers is to graphitize or order the planes. The fibers are heated above 2000 ¡C, increasing the tensile modulus to between 500 G Pa and 600 GPa. Although this graphitization lowers the strength of the fibers, some of the strength can be recovered by stretching the fibers during the process, aligning the planes with the longitudinal axis of the fibers. The overall purity of the fiber and crystallinity control the electrical and thermal conductivity as well as the resistance to oxidation [3]. 2.4. Strength of Thermoplastic s Based Composites From a strength aspect, thermoplastics are somewhat lacking when compared to the conventional thermoset matrix composite s. However, thermoplastics have the advantages of damage tolerance and environmental stability, which led to a surge in research and use since the early 1970s [8]. Thermoplastics can be roughly divided into two categories: convent ional and high -performance. Conventional thermoplastics have relatively good properties but generally lack strength when compared to their engineered counterparts. Most conventional thermoplastics are used extensively in everyday objects as neat plastics w ithout fiber reinforcement. Some examples include polyethylene, nylon, polystyrene, and polyester. For improved mechanical properties, all of these can incorporate fibers or fillers. Normally glass fiber is added to enhance certain aspects of a parts respe ctive mechanical property for a relatively small cost [1, 3]. Originally, thermoplastic based composites were created using solvent impregnation with amorphous polymers. However, other impregnation processes were used to develop PEEK and PPS (Polyphenylene sulfide) [2]. These are two examples of what are considered high - 10 performance thermoplastics. Although more expensive than conventional thermoplastics, the higher glass -transition temperature, T g, allows for much higher mechanical properties at higher oper ating temperatures. Also, an added benefit characteristic of these thermoplastics is increased toughness. 2.5. Manufacturing with Thermoplastics Thermoplastics can be classified as non -reactive solids or linear polymers at lower temperature and pressures. The most evident outcome of this is that they are re -moldable and re -formable. Thi s has both advantages and disadvantages to cost an d cycle times for manufacturing [1, 2, 10]. Transporting and storing thermoplastics is much simpler than that for thermosets . This is because they have, unless limited by some additive or other substance, an unlimited shelf life. A decrease in cycle times also makes them more attractive since thermoplastics do not require an extra mixing step or the long curing times of thermoset s. In these cases, the cycle time is determined by the time required to heat, disperse and cool the thermoplastic. Where a process involving thermosets is dependent on the chemical crosslinking, the complication with processing thermoplastics comes from co ntrolling viscosity. This is due to the fact that thermoplastics are fully polymerized and therefore have much higher molecular weights [8]. Typical injection viscosity of thermosets is less than 100 P a-s, however, thermoplastics will generally be 5 to 50 times greater [2]. This leads to manufacturing issues with full fiber wet out, fiber movement , and consolidation rate. The literature [8, 14, 15] showed multiple, innovative processes to effectively mix the fiber and resin to create a pre -impregnated mater ial for either immediate processing or for forming and then processing . These processes can be broken down into three general practices: pre -melt mixing, melt mixing, and low viscosity precursors. 11 2.5.1 Pre -Melt Mixing One possible solution to the high melt vis cosity of thermoplastics is found by mixing the solid thermoplastic material with the reinforcement material before melting and consolidation. The advantage of this is that the distance the melted matrix must travel to coat the fiber is lowered. The extens ion of this is what solid form should the thermoplastic resin take to minimize the distance and give the best possible impregnation results. Powder impregnation has been used as early as 1973 [14]. This method involves pulling bundles of fibers in tows thr ough a resevoir of thermpolas tic powder. The powder can be either dry or in an aqueous solution or slurry depending on the process. The advantage coating fibers with powdered resin is that the flow of thermplastic will now be parallel to the fiber tows as opposed to transverse, giving good impregnation results. This arises from the fact that permeability is generally at a minimum one order of magnitude less in the transverse directi on when c ompared to the parallel fiber direction [14]. Another obvious advantage arises from the fact that the viscosity of the melted resin is less of an issue during the impregnation process. This allows higher temperature thermoplastics to be impregnate d in the systems. BASF claimed in the late 1980s to be able, by using powder impregnation methods, to create composites with PEEK [2]. Some other possiblities that arise with powder impregenations is the option of multiple thermopolastic polymers being use d. Recent research has gone into finding cost -effective methods of combining multiple polymer systems into one finished composite [15]. Some challenges involved with powder impregnation arise from fully impregnated the fiber tows. This has been achieved, i n some cases, by including several pin rollers in the reservoir of resin powder. In the pin rolling process, fiber tows are drawn over th e pins, opening 12 the fiber tows and allow ing powder between the filaments as well as adding pressure to force the resin between the fibers . Polymer particles diamaters should be smaller than the reinfocment fibers . However, due to production costs associated with producing particles that small, this option is normally not available. For example, it has been found that havin g the thermoplastic be of the same diameter as the fibers could increase the production cost of resin by as much as 50%. It has been generally found that using powders with 15 Ð 150 µm diameters have produced good results [14]. Achieving the correct fiber volume fraction of during this process can also cause problems for manufacturers, leading to research into the correct number of pins, tension of fibers and form the powder takes, either dry, wet, or slurry. The final step of powder impr egnation is dependent on the desired performance or production method of the composite . The impregnated fiber tows can be heated, allowing the resin to melt and coat the fibers. The fiber tows are then dried and chopped to produce pellets of pre-impregnate d composite material for production. This creates a pelletized material that can be more rapidly processed into structural components. Another option would be to form the fiber and powder system before melting creating a powder coated tow or prepreg . This can be problematic as well as the fibers have little strength in the transverse direction [14]. An alternative to coating the powder process would be a commingling of fibers. In this process, the thermoplastic resin takes the form of fibers, either chopped or continuous. These thermoplastic fibers are then twisted with the reinforcement filaments to create yarn . The yarn, of both thermoplastic matrix and fiber reinforcement, can then be woven into fabrics as required and then subjected to heat and pressure to produce the composite part. These fabrics are generally drap -able and easy to handle for most geometry . A final advantage, as with the powder process, is the distance that the thermoplastic must impregnate is minimized [8, 14]. 13 One drawback to this is t hat the thermoplastic must first be in a fiber form before commingling; sometimes adding to the cost of the thermoplastic manufacturing. There is also the fear that the commingling, if not perfect, could create areas of excess resin or resin -dry areas during the cure phase of the process. The heating and molding process must accommodate release of air to avoid voids. Film stacking is a final example of resin and reinforcement material before melting. The process is easily described by its name, wherein alte rnating layers of resin films and reinforcement plies are stacked to certain specifications. Once stacked, the layup is subjected to heat and pressure to consolidate into a final prepreg sheet. The simplicity of this process makes it attractive for laborat ory use; however, long cycle times and simple geometries have kept this from becoming a viable industrial process [14]. 2.5.2 Melt Mixing Melt mixing methods involve mixing the resin and reinforcement material above the melt temperature of the thermoplastic. Na turally, the high viscosities of thermoplastic resins previously discussed now become a great concern. Extrusion compounding is one of the most common and most effective methods. The thermoplastic polymer is extruded using a single barrel or twin screw ext ruder. while the reinforcement fibers are introduced. The final product is cut to size for use in processes such as compression molding. Some drawbacks of this process are potential damage to the fibers during extrusion, and limited fiber lengths. This res tricts the products of this process to granules or small pellets of fiber reinforced thermoplastics. These later need to be melted for final forming [14]. 14 The next evolution of the extrusion compounding process is continuous strand impregnation. This proce ss removes the limitations on fiber length that are caused by extrusion processing. Resin is introduced under pressure into the reinforcement tows, which are being drawn through a mold or die. The tows are then cooled and the final product is chopped or possibly even wound for later processing. The resin saturation is governed by the rate tows pass though the cross -head die or the infusion time. However, the pressure from the extruders could also transfer into the fiber tows, packing them and decreasing the ir permeability. This causes problems with fiber impregnation. A natural progression of this limitation leads to the development of continuous strand impregnation that is similar to a resin bath. This resin bath is generally designed with pin rollers. Similar to aqueous or slurry powder impregnation, pin rollers flatten tows to separate filaments and allow resin to impregnate before cooling [14]. 2.5.3 Low Viscosity Precursors As seen with melt mixing, the high viscosities of the thermoplastic resin require more heat and pressure to fully impregnate the reinforcement fibers. This final option for mixing thermoplastic resin and fibers removes the problem of high viscosities altogether, allowing for conventional methods to be used with low viscosity materials. Chain extension of thermoplastic polymers is an area researched for applying traditional low viscosity production methods. In this case, a very low molecular weight thermoplastic resin is used during the impregnation stage of manufacturing. The thermoplastic resin will have a much lower viscosity, aiding in the impregnation, because of the low molecular weight. Once impregnated, the molecular weight will be raised by the introduction of chain extensions through 15 a reagent. Low viscosity processing has been adva nced by sizing the fibers with such a reagent before impregnation [14]. Another method of lowering resin viscosity is introducing a solvent. This process of impregnating fibers with the matrix material involves dissolving the thermoplastic into a solvent solution. This lowers the viscosity of the resin/solvent solution, which aids in the impregnation process. The solvent is then recovered in a later step. For example, the fibers can be dried by leaving the fibers fully impregnated with the thermoplastic re sin and degassing. Naturally, matching a solvent with a particular thermoplastic becomes another issue. Higher performance thermoplastics , PEEK, will generally be solvent resistant , due to its semi -crystalline structure, rendering this process useless. However, for an amorphous thermoplastic that is easily dissolved into a solvent, there remains the problem of completely removing th e solvent from the final system [2]. 2.6. Compression Molding of Thermoplastic Composites Compression molding is a common metho d of producing composite materials. The addition of heat and pressure to a composite causes the matrix to flow, removing air and filling the mold cavity. The process is relatively quick and can handle complex geometries, making it especially good for the a utomotive industry where a quick process time is desired. However, there are a number of factors that control the results of a compression molding process [16Ð18]. The heating and cooling rates of the mold are two major process parameters that must be cont rolled. Time must be allowed for the mold to be brought to temperature so that the material being processed is uniformly heated to the appropriate thermoplastic melt temperature. The cooling rate can be even more critical to the final mechanical performanc e of the composite part 16 due to the formation of crystallites. Different cooling methods can be applied including air cooling, air/water mist cooling, and dump water cooling. The effects of pressure are also important to any composite compression molding pr ocess. Pressure should be applied to the mold to consolidate the composite. However, a large increase in pressure before the melt temperature could damage the mold and restrict proper flow of materials. Pressure should also be held during the cooing cycle [16]. Anisotropy can become a problem in some thermoplastic molded parts. This is due to the long chains of polymers becoming aligned at some point in the process and cooling too fast to become entangled again. This problem is commonly seen in injection mo lding when the polymer is aligned in the direction of injection. Movement of melting resin in compression mold can also produce the same effect. This phenomena can cause variations in strength with preferred directions performing better than on other angle s [16] Semi-crystalline polymers, such as Nylon and PEEK, have an added level of complexity during compression molding cycles. As has been discussed, the advantage of compression molding thermoplastics is the lower cycle times. However, due to the high tem peratures that these thermoplastics require to crystallize, the cooling time increases [16, 12]. These thermoplastics are also susceptible to shrinkage, which means high pressure must be applied to counteract this effect. It is important to have pressure a pplied at the moment that it transitions back into a solid. 2.7. Tensile Strength versus Flexural Strength Flexural testing is generally not accepted as a true test of failure strength for design projects. This is because flexural testing generally produces hi gher strength results when 17 compared to standard tensile testing. This is thought to come from the stress gradient that occurs in the flexure testing [19, 20]. However, there are a number of difficulties that arise from running standard tensile testing on p olymer composites. In order to ensure the failure occurs at the center of the specimen in tensile testing, a dog bone coupon is generally required. This creates a uniform stress in the center of the coupon, away from the grippers. For molded composites, th is would require a special mold and fiber preform or expensive post processing [19]. There have been attempts to predict tensile testing results from 3 and 4 point flexural bending testing using a two -parameter Weibull model with some success [19, 21]. Th is requires a large database and constructing one for these particular materials is outside the scope of this research. For these reasons, only flexural test data will be taken into account and compared to this work. 2.8. Nylon Composite Data from Literature Although nylon has been used extensively for some time now, there was little direct results from previous work that were exactly comparable to the type of composite produced in this research. Some information could be found on aligned and discontinuous glas s fiber reinforced nylon, for example [22], but little of carbon reinforced. However, there were some examples of flexural properties from products currently on the market , as well as , a few studies on those products. Fiberforge Corporation ( Glenwood Sprin gs, CO ) creates aligned carbon fiber prepregs in several polymers. A comparable product sold today by them is 65% fiber by weight and has a quasi -isotropic layup. The strengths reported by Fiberforge show a flexural strength of 520 MPa 18 and modulus of 40 GP a [23]. As a reference, the unidirectional layup offered by Fiberforge has a flexural strength of 1700 MPa and modulus of 190 GPa at 65% fiber by weight [24]. Another study by Thomason [25] was performed on randomly oriented glass fibers at 30% fiber by weight in nylon. The results showed a flexural strength of 286.5 MPa and flexural modulus of 9.20 GPa. The glass used was E -type with a nomin al fiber diameter of 10 microns . The data from the literature has been compiled in Table 2.1 below. Table 2.1: Nylon based composite data from the literature Composite Tested Flexural Strength (MPa ) Flexural Modulus (GPa ) Fiber Forge Aligned at 65 % by weight carbon fibers [24] 1700 190 Fiber Forge Quasi -isotropic at 65 % by weight carbon fibers [23] 520 40 30 % by weight glass fibers [25] 286.5 9.2 2.9. PEEK Composite Data from Literature The literature review showed a lack of good data from randomly oriented PEEK composites. However, some information could be found for aligned carbon fiber PEEK composites and laminate layups. A study published by McGrath et al. [18] indicated results for standard, continuously reinforced P EEK composites at 68% fiber by weight . In addition to [0]16, [0/90] 4S, and [+45/90/ -45/0] s laminate layups, a possible prepreg version of the PEEK composite was tested. In this case, an aligned carbon fiber PEEK composite was ground to a specific specification , using a ÒmonomuncherÓ, two parallel shafts with teeth that rotate inward to tear and shear the composite apart. The resulting, fragmented composite particles are then compression molded . All samples were tested for f lexural strength and modulus. 19 Although the continuously aligned laminatesÕ strengths and modulus values are much higher than those thought to be attainable by the composite system produced in this thesis , the [+45/90/ -45/0] S , also known as a quasi -isotrop ic laminate, shows reasonably attainable values. This system acts similarly to an isotropic material in the plane, which are the expected results from a randomly oriented composite. The performance of the quasi -isotropic laminate should still be higher tha n a randomly oriented composite because the laminate is produced from layers of aligned fibers. In another study by Kurokawa et al. [26], PEEK -carbon fiber extruded composites at 15% fiber by volume were tested for use in gears. As part of the work, flexur al test specimens were created and reported flexural strengths of 305 MPa and a flexural modulus of 14.3 GPa. Naturally, at such a low volume fraction, the samples produced in this study should exceed these values. All of these examples are summarized in Table 2.2.The flexural strength data is assumed to be along the 0 ¡ orientation for the aligned fiber composites. Table 2.2: PEEK based composite data from the literature Composite Flexural Strength (MPa) Flexural Modulus (GPa) UD Prepreg layup [0] 16 [18] 1876 130 UD Prepreg layup [0/ 90]4s [18] 1163 68.8 UD Prepreg layup [+45/90/ -45/0]s [18] 616 40.9 Ground UD - 3mm [18] 439 33.5 15% Carbon Fiber by Volume [26] 305 14.3 20 2.10. Other Comparable Composite Strengths Sheet molding compounds (SMC) are a quick and cheap solution to cutting down production times for thermosetting composites. Used often in the automotive industry, these are thermosetting vinyl esters with either short glass or carbon fibers added. The mixture is then placed in a mold and compressed with heat to form products. Quantum Composites ( Bay City, MI) has several products currently available at hig h carbon fiber weight percent. A 53% by weight carbon fiber SMC product has a flexural strength of 421 MPa a nd flexural modulus of 26.2 GPa [27]. Lu [6] performed flexural testing on composites made with the same wet lay process containing different polymers. In his research, carbon fiber filled PET at about 0.34 fiber volume fraction had a flexural strength of 240 MPa and a flexural modulus of 22 GPa. Caba [28] also performed similar studies with carbon fiber in PET and PP. The results indicated that a carbon PP based composite at 0.36 fiber volume fraction produced a flexural strength of 138.7 MPa and flexural modulus of 11.1 Gpa. Finally, from unpublish ed data, the same papermaking process was used to create nylon based composites but with reclaimed carbon fiber. In this case, an average flexural strength of 205 MPa and average flexural modulus of 17 GPa was reported at 0.347 fiber volume fraction. 2.11. Calc ulating Volume Fraction from Weight Fractions In this thesis a fiber volume fraction will always be given for comparison. However, during production, mass measurements are used to mix materials. Equation 2.1 below can be used to convert between fiber volum e fraction and fiber weight percent: 21 +,-./01023./010234567./8102/ (2.1) Where +, is volume fraction of fiber, 9, is weight fraction of fiber, :; is density of the matrix material and :, is the density of the reinforcement fibers. 2.12. Scanning Electron Microscopy Due to the ease of sample preparation, good resolution and high depth of field, scanning electron microscopy (SEM) has become a very popular method of studying fracture surfaces. In SEM, a focused electron beam is swept across a solid object, producing a n umber of responses. One effect of this beam is the release of secondary electrons, coming from the inelastic excitation of the electrons on the specimenÕs surface. These secondary electrons have energies in the range of 1 -50 eV. Because electron movement i s limited in air, the testing chamber is run under a vacuum [29]. Failure in short -fiber reinforced semi -crystalline thermoplastics can be seen in several examples in the literature. In the study , the failure of short fiber composites was shown to depend on fiber orientation and environmental conditions. It was found in short glass fiber filled PET composites that fibers are aligned parallel to fracture planes. This type of failure is attributed to de-bonding between the fibers and matrix due to poor adhesi on. This was indicated by clean fiber surfaces in the case of poor adhesion and rough surfaces in cases with good adhesion. The rough fiber surfaces were created by remains of matrix material on the fiber surface, indicating that the failure was due to the matrix and not the fiber matrix interface. In the same composite systems, but with loading transverse to the fracture plane, other mechanisms become responsible 22 for crack propagations. Shear loading can cause de -bonding of the fibers and matrix leading to fiber pull out. Furthermore, high stress concentrations at fiber ends can also lead to crack initiation and failure. Roulin -Moloney et al. and Purslow et al. [29, 30] performed fracture surface investigations for glass filled PEEK. Failures were seen aro und fiber ends at temperatures less than 90¡C and low loading rates. Failures occur following ductile matrix failure after fiber failures or from the matrix crack growth around fiber ends. The same failure can be found following brittle matrix failure with higher loading rates . As these samples will be loaded in flexure at room temperature, there will be some differences between the results found in this and other literature sources [30]. However, two things should be observed on the fracture surfaces: fibe r ends and matrix surfaces. Clean fiber pullout with no cracking would indicate poor bonding betw een fibers and matrix material. 23 Chapter 3. Panels and Materials Tested This section introduces the panels produced and the materials used in the p roduction. The panels tested, shown in S ection 3.1, lists the number of panels used in flexural testing, not the total number of panels produced. Section 3.2 follows with a full description of the materials used in the production process. 3.1. Panels Tested During this study, 15 panels were produced for testing and are listed in Table 3.1 and Table 3.2 below . The p anel number field indicates the universal label of the panel, numbered sequ entially by time produced with no regard for material or specifications. The tables also indicate the target fiber volume fraction and fiber weight fraction used in production. This indicates the amount of reinforcement and matrix material used in the proc ess outlined in the following chapter. The tables have been ordered by target fiber volume fraction. These volume fractions were chosen to represent a collection of the composites for a range of volume fractions. Although there were more panels produced, n ot all were included in flexural testing and have not been included in this thesis. 24 Table 3.1: Nylon b ased panels produced for testing Pane l # Material Target Fiber Volume Fraction Target Fiber Weight Fraction 011 Nylon 0.312 0.417 021 Nylon 0.340 0.449 022 Nylon 0.358 0.468 008 Nylon 0.369 0.480 009 Nylon 0.369 0.480 006 Nylon 0.410 0.523 004 Nylon 0.410 0.523 003 Nylon 0.410 0.523 007 Nylon 0.410 0.523 012 Nylon 0.425 0.539 Table 3.2: PEEK based panels produced for testing Panel Material Target Volume Fraction Target Fiber Weight Fraction 025 PEEK 0.330 0.405 023 PEEK 0.341 0.417 013 PEEK 0.390 0.470 026 PEEK 0.410 0.490 010 PEEK 0.440 0.521 3.2. Materials Used 3.2.1 Reinforcing Fibers The carbon fibers used in this study were Toray T700S -C-50-C. The fibers were chopped to 25 mm and were provided by Crosslink Technologies (Nebraska) [31]. They were delivered without modification to sizing or surface chemistry. The manufacturerÕs data sheet for these fibers claims a tensile strength of 4900 MPa and a tensile modulus of 230 MPa . The density was reported as 1.8 0 g/cm 3. The data sheet also noted that the general -purpose sizing used on these fibers performs well for epoxy, phenolic, polyester, and vinyl ester resins. 25 3.2.2 Thermoplastic Matrix The thermoplastic nylon used was supplied from Minifibers, Inc (Johnson City, Tennessee). The Nylon 6,6 was described as regular tenacity nylon. It was rough ly cut to 6.25 mm (0.25 inches) and was 3.0 dpf filament size. From the manufactureÕs data sheet, the nylon had a density of 1.14 g/cm 3, a breaking tenacity of 2.3 -6.0 gpd, and a melting temperature between 244 ¡C and 260 ¡C [32]. The thermoplastic PEEK was supplied from Zyex Ltd. (United Kingdom). The information came from both Zyex Ltd website [33] and through contact with company representatives. It was cut form stretched, oriented multifilament yarn. The density was reported as 1.3 g/cm 3. The glass transi tion temperature was reported as 143 ¡C with a melting temp of 343 ¡C. The tensile strength of the material was stated as 600 MPa. 3.2.3 White Water The white water was a water -based solution with thickener, surfactant and anti -foam added. The materials and prope rties are given in the sections below. 3.2.3.1 Thickener The thickener used was Nalco 7590. The information sheet provided by Nalco Company (Naperville, Illinois) indicated that this particular thickener could increase drainage, improve formation of mats and show ed good retention of sizing agents. The thickener was an opaque white color and had a density of 1.03 - 1.07 g/cm 3. This sometimes made it difficult weigh properly . 26 3.2.3.2 Surfactant Rhodameen VP -532/SPB was used as the surfactant and was provided, as a free sa mple, from Rhodia. However, this company was purchased by Solvay who provided the second free sample. Rhodameen is a mixture of 20 -23 %wt Tallow Amine and 77 -80%wt water. The mixture is a viscous liquid, is water soluble and slightly yellow. The specific g ravity is 1.01 at 25 C. While the Rhodameen was unav ailable, a solution of its compo nets was mixed in house. The tallow ammine was ordered from Chem Service Inc. under the name POE (15) tallow amine (CAS: 61791 -26-2). The data sheet gave the density as 0.966 g/cm 3. The tallow amine was amber colored at room temperature and, due to it being only slightly soluble in water, was difficult to mix properly to form the surfactant. 3.2.3.3 Anti -Foam This chemical was also provided by Nalco and was called PP06 -3586. The o ff-white liquid had a mild odor and a density of 0.994 g/cm3 . The information sheet indicated that it was dispersible in water. 3.2.4 Mold Release There were two mold release agents used depending on the thermoplastic being processed. In both cases, the mold release was delivered as an aerosol and sprayed on the mold before closing at room temperature. Due to the nature of the mold being used and the large volume of the dry fibers before processing, it was impossible to apply the mold release to a heated mold. 27 The mold release used for the nylon based panels was MS -122AD from Miller -Stephenson (Danbury, Connecticut). This was a PTFE based mold releas e that indicated good results in previous work performed. The release agent sprayed on as a milky white liquid, slightly visible on the mold surface. The density was given as 1.2 g/cm 3. Because of the toughness of the PEEK, an alternative mold release was ordered from Miller Stephenson, MS -K1206A. This Boron Nitride release agent was slightly harder to handle, spraying on as a white powder at room temperature. The density was given as 1.38 g/cc. Although working marginally better than MS -122AD on PEEK panel s, the mold release seemed mostly ineffective, leaving PEEK residue on the surface of the mold. 28 Chapter 4. Production of Test Composite Plates The production of composite plates was a multi -step process limited by equipment size and availability. Due to these constraints, there was some unavoidable variation in the process from panel to panel. However, special attention and care was taken to remove most of these small changes and to produce panels using the same method every time. The overall process remains the same for all panels apart from changes in processing temperature and mold release due to the thermoplastic being used. In all cases, the overall method can be seen in Figure 4-1.The specific fiber volume fraction for each panel was used to calculate, using equation (2.1) , the weight of matrix and reinforcement fibers to be added during the mixing step. For this chapt er, no specific amounts of fibers are mentioned and the exact amounts of fibers used for each respective plate are included in t he A ppendix. Figure 4-1: Overview of production method used in panel productio n 4.1. White Water For each 15.24 cm square panel produced, 20 L of white water were required. The pulper used was a FORMAX Malestrom 8 L laboratory pulper , shown in Figure 4-2, from the Adirondack Machine Corporation (Queensbury, NY). Due to the limited size of the pulper, the white water was mixed in 5 L batches and stored in two, 10 L plastic buckets with lids. The "#$%&!"'%&( )$*$+, "'%&(! -&./0'1 2(3$+, )/14$+,! 5(&6&('%$/+ 7/.6(&88$/+! )/14$+, 29 amount of materials used per 5 L batch by weight is giv en in Table 4.1. Each material batch was mixed for 5 minutes at 600 rpm and then stored in one 10 L plastic bucket. Table 4.1 Materials used for white water Material Amount Thickener 1.25 g Surfactant 1.25 g Anti -Foam 0.49 g For a given panel, all 20 L of white water was mixed continuously to ensure the same consistency. As the thickener used appeared to be extremely hydrophobic, a batch of white water was used within 2 days of mixing. This ensured the thickener was still prop erly mixed for the fibrous mat production. 4.2. Mixing Fibrous mats were produced using a semi -automatic sheet former. The FORMAX semi -automatic sheet former was model number G -300W also from the Adirondack Machine Corporation. Again, the number of batches wa s controlled by the size of the pulper. For all of the mats produced, 4 L of white water were placed in the pulper and stirred for several seconds to ensure no thickener had collected out of solution . The thermoplastic fiber was measured out and added dire ctly to the white water. The mixture was allowed to mix at 250 rpm until the thermoplastic had dispersed throughout the solution. 30 Figure 4-2: FORMAX maelstrom 8 L pulper In the same way, the carbon fiber was weighed and added to the mixture. The carbon fiber immediately clumped together on the lower rpm setting. The mixing speed was then increased to 600 rpm and run for 10 minutes to allow the carbon fiber to properly disp erse in the white water. After 10 minutes, the mixture was collected and transferred to the sheet former. From observations, white water that was lacking surfactant or that was improperly mixed resulted in fiber agglomeration or fiber collection surface of the solution. This was noted specifically for one panel and is discussed in the Chapter 7. 4.3. Water Removal The sheet former , shown in Figure 4-3, allows for 5 batches of solution to be added at once. Each successive batch of solution is poured over the next and allowed to sit while the successive ones are being mixed. Once the 5 batches have been added, the remaining volume in 31 the deckle box is filled with tap water. The full deck le box can be seen in Figure 4 -4. The mixture is then agitated for 30 seconds. During this time, air is introduced at 12 points around the base of the deckle box. The resulting air bubbles travel up through the solution, agitating the fibers an d attempting to disperse them evenly in the box. After 30 seconds of agitation, the solution is allowed to sit for 3 seconds and then the drop leg valve was opened. Water is removed through the drain while the fibers collect on the 30.45 cm square forming wire. Once done, the deckle box is removed and the mat is couched using an air bladder to remove additional moisture. Two blotter sheets are added between the mat and the bladder to absorb further moisture. Figure 4-3: Formax sheet former. Shown with deckle box and couching plate rotated up 32 Figure 4-4: Thermoplastic and carbon fibers mixed with white water in deckle box Figure 4-5: Fibrous mat after couching step This process results in a damp, fibrous mat of both carbon fibers and thermoplastic fibers , shown in Figure 4-5. The mat is 30.54 cm square. It is then removed from the forming wire and blotter sheets are placed above and below. 4.4. Drying The mats were then removed from the sheet former and allowed to air dry on an open lab table. In order to minim ize the number of reinforcement and matrix fibers, the blotter sheets were 33 allowed to remain on the mats while they air -dried. The wet blotter sheets creased and yellowed, allowing the fibers to easily pull away. The mats were deemed sufficiently dried whe n the blotter sheets could be removed. In some circumstances, a heat gun was used to decrease the drying time and aid in the removal of the blotter sheets from the damp mats. Once sufficiently dried, the blotter sheets are easily removed, taking extra prec aution to ensure no fibers are pulled away from the mat. At that point, only the fiber network loosely held the mat together. This made handling the mat difficult as lightweight fibers are easily pulled from the surface. The mats were carefully transferred to a large air -convection oven. Nylon based mats were pl aced on porous Teflon sheets while PEEK based mats were placed on aluminum foil covered with a light layer of mold release. The oven temperature was set to the respective thermoplastic melting temperature and allowed to heat. Once the desired temperature w as reached for 5 minutes, the oven was opened and the mats were checked. Indications that the mat has properly dried include slight variations in color, visible spheres of resin and a ÒstifferÓ feel to the mat. Some variation in this process was observed for all panels made. Hot spots in the oven were found to create regions of some mats that would be over heated, indicated by large areas of color change and brittle areas. 4.5. Molding Preparation Once dried, the mats were quartered using a fabric cutter . The mats were cut on a polyethylene surface. Moderate pressure was required to cut through the carbon fibers, sometimes resulting in lose of fibers on the cutting surface. The resulting, 4 smaller mats are 34 each about 15.24 cm square. These are collected and placed in a plastic bag until mold assembly is ready. The 15.24 cm, steel compression mold is shown in Figure 4-6. Because the final volume of the panel is known a nd the x and y dimensions are set, the required open thickness of the mold can be found. The Z dimension is achieved by including shims in the molding process. Figure 4-6: Steel compression mold All mold parts were then coated with the appropriate mold release before assembly. The injection ports were plugged and the lower and middle segments were bolted together. The 15.24 cm square fibrous mats were then individually weighed and stacked inside the cavity . Finally, the spacer was placed and the mold closed. Because of the large volume of the fibrous mats, the mold could not be closed completely by hand. 4.6. Compression Molding The mold was then transported to one of two presses. For nylon based panels, a 40 T on PHI press was used and is shown in Figure 4-7. This manual press allowed for a heating range of 35 up to 315¡ C, well within the range for the materials used. For PEEK based panels, a large 140 Ton TMP press was used, allowing for temperatures of up to 427¡ C. This press is shown in Figure 4-8. For all materials, the mold is placed on the cold platen surface. The press is closed and a force of 20 kN is applied . This ensures the mold is properly aligned and the material is contained within. This pressure is held constant while the temperature is i ncreased to the melting temperature of the thermoplastic being used. The ramp rate is set at the maximum allowed or possible by the press. A portable type -k thermocouple was placed on a small port on the mold to indicate the temperature of the mold itself. This thermocouple was used as the driving temperature of the process, not the platen temperature reading of the press itself. Figure 4-7: PHI press used in molding nylon based panels 36 Figure 4-8: TMP press used for molding PEEK based panels Once, the melting temperature of the material was reached, also indicated by a drop in force as the material compressed, the press was set to 200 -250 kN. The mold closed easily and was allowed to sit under pressure at the melt temperature for 10 minutes. In the case of nylon based panels, some nylon always leaked out of the mold. Once the panel was processed at melting temperature and 250 kN of force for 1 0 minutes, the heating was turned off and the mold allowed to cool under pressure. For both presses, forced cooling was implemented. In the case of the manual PHI press, a large fan was placed in front of the mold. The larger, TMP press included forced air cooling within the platens as well as air/water mist cooling and flood water cooling at lower temperatures. 37 Chapter 5. Testing 5.1. Flexural Testing The flexural testing was done according to ASTM 6272: Standard Test Method for Flexural Properties of Unreinforce d and Reinforced Plastics. For all panels, coupons were cut using a diamond -tipped, circular wet saw mounted on a table. Coupons were cut according to the ASTM standard with most being 60 mm x 12.8 mm. However, due to the inaccuracy of t he wet saw, the cou pon widths varied by 0.50 mm . After being cut, all coupons were dried using a paper towel or heat gun. The heat gun was set to relatively high temperatures, between 400 ¡F Ð 500¡F. However, it was only used indirectly on the coupons. The heat gun became nec essary to dry the nylon samples, which the literature review revealed absorb water. After being cut all coupons were allowed to dry for at least one day before being tested. Special care was taken to label all coupons with the correct plate number, coupon number, and in the case of same panels, the local material direction. The width and depth measurements of all coupons were taken using electronic calipers and recorded. Loading and support spans for the 4 -point bending tests were calculated using the ASTM standard. The support and loading spans are given in the Table 5.1 as well as the number of test coupons. Once run, Equations 5.1 and 5.2 were used to convert the load Ð displacement curves to stress -strain curves: <,-=>?@ABC (5.1) D,-EFBGC (5.2) 38 Where < is flexural stress, P is load, S is support span, b is with of test specimen, d is depth of specimen and D is maximum deflection at center of the beam. The maximum flexural stress could then be found as the highest stress value recorded before failure. The flexural strength w as found by finding the slope of the initial portion of the stress -strain curve. Table 5.1: Specifications of flexural testing Panel Support Span (mm) Load Span (mm) Load Rate (mm/min) Samples Tested 003 26.09 12.48 0.77 10* 004 25.4 12.7 0.70 10* 006 25.4 8.47 0.78 10* 007 25.4 8.47 0.78 10* 008 26.4 8.8 0.73 10* 009 25.4 8.47 0.74 10* 010 25.4 8.47 0.71 5 011 25.4 8.47 0.79 10* 012 25.4 8.47 0.77 10* 013 25.4 8.47 0.74 5 021 25.4 8.47 0.76 5 022 25.4 8.47 0.78 5 023 25.4 8.47 0.77 5 025 25.4 8.47 0.80 5 026 25.4 8.47 0.78 5 * starred panels included two sets of 5 coupons orthogonal to the other. In order to consistently find the flexural strength and flexural modulus from the data, a MATLAB code , given in the appendix, was written and implemented to process the raw data. 5.2. Adjusted Fiber Volume Fractions Because carbon fibers were used as the reinforcement, no burn off tests were possible to discover the actual fiber volume fraction. Acid digestion tests could have been implemented, 39 however, due to the cost and dangers associated, these tests were also rul ed out. As a method to discover the average ÒadjustedÓ fiber volume fraction of the panels, the final depth of the panel sufficed. It was assumed that the only material loss during the compression -molding step would be matrix. By using this assumption, the adjusted fiber volume fraction could be calculated by using known fiber weight use in production and the final volume of the panel. The volume of the panel was found by multiplying the know area, 232.3 cm 2, by the measured thickness. The adjusted fiber vo lume fraction can then be found by using the following equation: +,$H-I/0J0B0K6LM6K (5.3) where M f is the total mass in grams of carbon fiber in the plate and d is the depth in mm of the final panel. The constant 41.814 comes from a collection of constants and conversions. 5.3. Processing For all methods of production, extensive notes were kept to locate all sources of error and to indicate the practicality of the pr oduction method. This included all materials mixed to create white water and fibrous mats. The fibrous mats were weighed before the compression -molding step. During the molding stage, regular time, temperature and pressure measurements were recorded. These results will take the form of possibly sources of error during the production and testing of these composite panels. 40 5.4. Microscopy Selected coupons were observed using a high vacuum scanning electron microscope. These were selected to give images of coupo ns with a range of materials, performance, or interesting phenomena. Twelve coupons were studied under high magnification to gain insight from their surfaces. This included looking for voids that may appear, bundles of aligned fib ers, interesting fracture areas, and bonding between reinforcement and matrix. The panels tested are listed in Table 5.2. Table 5.2: Sample coupons used for SEM microscopy Panels Coupons 004 06 007 01, 03 012 06 010 03, 05 013 01, 02 023 02, 04 41 Chapter 6. Results This chapter reports the results and observations of the production and testing of panels. The volume fraction, flexural strength, flexural modulus, and microscopy results follow. A more detailed discussion of the results follows in Chapter 7. 6.1. Fiber volum e fraction Through the production process, the fiber volume fraction tended to increase when comparing the predicted versus final adjusted volume fraction values. The volume fraction results for nylon and PEEK based panels can be seen in Table 6.1 and Table 6.2. The target volume fraction indicates the material amounts that were used in the production of the panels. The final depth was the measured depth before flexural testing and from that; the adjusted average volume fraction calculated using the equation given in Chapter 5. The percent change is the percentage increase or decrease between the target and final volume fraction. Table 6.1: Predicted versus final calculated fiber volume fraction for nylon based panels Panel Material Target Volume Fraction Final Depth Adjusted Average Volume Fraction Percent Change 011 Nylon 0.312 0.51 0.325 0.042 021 Nylon 0.340 0.53 0.349 0.026 022 Nylon 0.358 0.54 0.365 0.020 008 Nylon 0.369 0.76 0.329 -0.108 009 Nylon 0.369 0.60 0.362 -0.019 006 Nylon 0.410 0.55 0.415 0.013 004 Nylon 0.410 0.57 0.410 - 003 Nylon 0.410 0.56 0.412 0.005 007 Nylon 0.410 0.56 0.412 0.005 012 Nylon 0.425 0.57 0.425 - 42 Table 6.2: Target and adjusted fiber volume fractions for PEEK based plates Panel Material Target Volume Fraction Final Depth Adjusted Average Volume Fraction Percent Change 025 PEEK 0.330 1.53 0.339 0.027 023 PEEK 0.341 1.55 0.346 0.015 013 PEEK 0.390 1.62 0.378 -0.031 026 PEEK 0.410 1.54 0.418 0.020 010 PEEK 0.440 1.66 0.416 -0.055 6.2. Flexural Strength Raw data was collected from flexural tests and processed using a MATLAB code that is provided in the Appendix . All calculations relating displacement and load to strain and stress were according to ASTM 6272 and the equations shown in Chapter 5. A typical stress strain curve can be viewed in Figure 6-1. This particular data set comes from panel 006 and displays the results from all 10 test coupons. Figure 6-1: Typical stress -strain c urve from flexural testing . Diamonds and circles represent slope locations for modulus calculations. Stars represent peak strength. 43 The strengths and adjusted volume fractions of the nylon based panels are shown in Table 6.3 and Figure 6 -2 with the standard deviations for each individual panel Õs coupons. The standard deviation represents the variation in data from the coupons tested for each panel. Table 6.3: Flexural strength results for nylon based pa nels Panel Adjusted Volume Fraction Flexural Strength (MPa) Std. Dev (MPa) 011 0.325 217.84 23.49 021 0.349 167.65 16.00 022 0.365 202.73 11.69 008 0.329 188.60 14.13 009 0.362 193.35 21.57 006 0.415 193.12 19.38 004 0.410 315.53 43.98 003 0.412 327.15 32.55 007 0.412 365.74 54.18 012 0.425 206.58 29.82 Figure 6-2: Flexural strength results for Nylon based panels 0501001502002503003504004500.250 0.300 0.350 0.400 0.450 0.500 Flexural Strength (MPa) Adjusted Fiber Volume Fraction 44 The results for the PEEK based panels are shown in Table 6.4 and Figure 6-3 and are presented in the same manner as the nylon based panel data. Table 6.4: Flexural strength results for PEEK based panels Panel Adjusted Volume Fraction Flexural Strength (MPa) Std. Dev (MPa) 025 0.339 532.82 34.77 023 0.346 640.39 53.51 013 0.378 553.63 66.04 026 0.418 542.39 37.39 010 0.416 619.72 82.34 Figure 6-3: Flexural strength results for PEEK based panels 6.3. Flexural Modulus The same data from the flexural testing was again processed in a MATLAB code. A linear curve fit was used on the initial increase of the stress -strain curve to determine the flexural modulus. The curve fit was performed on each set of data multiple times to find the best fit. The 9:99;99<99=99>99?99@99A999B;>9 9B<99 9B<>9 9B=99 9B=>9 9B>99 Flexural Strength (MPa) Adjusted Fiber Volume Fraction 45 MATLAB code indicated the selected start and e nd points of the curve fit with a diamond and star. These can be seen in the sample results in Figure 6-1. This was done by comparing the R -squared values for each fit . The full results and R -squared values for each test can be found in the Appendix. The flexural modulus results for nylon based panels can be seen in Table 6.5 and Figure 6-4. Table 6.5: Flexural modulus results for nylon based panels Panel Adjusted Volume Fraction Flexural Modulus (GPa) Std. Dev (GPa) 011 0.325 18.90 4.86 021 0.349 13.78 6.45 022 0.365 19.12 2.59 008 0.329 19.48 2.38 009 0.362 21.07 1.96 006 0.415 23.05 3.12 004 0.410 20.23 2.38 003 0.412 22.63 3.59 007 0.412 28.87 5.65 012 0.425 22.71 3.66 46 Figure 6-4: Flexural modulus results for nylon based panels The flexural modulus results for PEEK based panels can be seen below in Figure 6-5 and Table 6.6 and are found and displayed in the same was as described for the nylon based panels. Table 6.6: Flexural modulus results for PEEK based panels Panel Adjusted Volume Fraction Flexural Modulus (Gpa) Std. Dev (GPa) 025 0.339 24.81 3.49 023 0.346 30.24 3.17 013 0.378 30.47 4.68 026 0.418 31.27 3.48 010 0.416 28.95 7.93 9B99 >B99 :9B99 :>B99 ;9B99 ;>B99 <9B99 <>B99 =9B99 9B;>9 9B<99 9B<>9 9B=99 9B=>9 9B>99 Flexural Modulus (GPa) Adjusted Fiber Volume Fraction 47 Figure 6-5: Flexural modulus results for PEEK based panels 6.4. Planar Isotropy As discussed in Chapter 5, this production method should result in in -plane isotropic material properties. In an attempt to validate that this was indeed the case, several nylon based panels were tested in two directions. In these cases, ten total flexural testing coupons were cut, five being orthogonal to the remaining five. Comparison of all ten coupons could show variations in the performance based on two directions. Table 6.7 shows the flexural testing results for each coupon sample from panel 007, the only panel that showed a strong directional dependence. In Table 6.7, the st rengths and moduli are given as two directions, 0 ¡ and 90 ¡. These degree values should only be considered relative to each other and not to the panel as a whole or connected to any part of the production method. The data presented in Table 6.7 shows that the flexural strength in the 0 ¡ direction had an average value of 410.21 MPa, 89 MPa higher than the average in the 90 ¡ direction at 321.26 9>:9:>;9;><9<>=99B;>9 9B<99 9B<>9 9B=99 9B=>9 9B>99 Flexural Modulus (GPa) Adjusted Fiber Volume Fraction 48 MPa. The flexural modulus also showed marginally better results in the 0 ¡ direction at 33.69 GPa compared to 24.04 GPa in the 90 ¡ direction. Table 6.7: Flexural testing results for panel 007 Panel -Coupon Relative Orientation Flexural Stress (MPa) Flexural Modulus (GPa) 007-01 0¡ 416.65 36.26 007-02 0¡ 344.21 28.82 007-03 0¡ 433.66 35.02 007-04 0¡ 416.63 31.79 007-05 0¡ 439.92 36.57 007-06 90¡ 301.27 21.92 007-07 90¡ 325.55 25.83 007-08 90¡ 317.12 25.59 007-09 90¡ 323.12 23.40 007-10 90¡ 339.22 23.47 6.5. Microscopy A scanning electron microscope was used at high vacuum to take micrographs of the surface of selected sample coupons. No attempt was made to look at morphology or the degree of crystallinity of the thermop lastics; however, some insights can be gained by looking at fiber orientation and overall cracking. All test coupons were photographed after flexural testing had been completed. The first n oticeable observation from the SEM images was to look for easily de finably layers from the manufacturing process. This can be seen in Figure 6-6, an edge view of a coupon from panel 012, a nylon based panel . These layers appear to be vertical lines on the image. The individual mats that are stacked and then consolidated probably caused this. The image also shows good consolidation with no apparent voids. To the right of the image, a small p it can be 49 seen, and is indicated, on what woul d be the large face of the panel. This could be caused by poor resin flow or a problem with the mold release on the surface of the mold. Figure 6-6: Photo micrograph of panel 012 coupon 06 (Nylon). Surface pit circled in black. Figure 6-7 shows another edge view of the surface of a test coupon, this one from panel 13, coupon 1, which was a PEEK based panel. Again, the consolidation looks good along with a good randomization of fiber directions. Fibers appear to be covered well with thermoplastic with no apparent voids visible. The glowing white areas in the images are polymer areas that are becoming Òcharged.Ó This occurs in poorly conductive materials and represents a buildup of negative charge. 50 Figure 6-7: Photomicrograph of panel 13 coupon 1 (PEEK) Figure 6-8: Photomicrograph of panel 007 coupon 1 (Nylon) Figure 6-8 again shows good consolidation but shows some collections of aligned fibers. This image is from the n ylon based panel 007, coupon 1. A detail of the cracking surface of the sa me 51 panel and coupon is shown in Figure 6 -9. The crack was probably allowed to grow easily by the alignment of the fibers near the surface of the panel. Figure 6-9: Photomicrograph of panel 007 coupon 1 (Nylon) The images below in Figure 6 -10 and Figure 6 -11 show both good and poor bonding of matrix to fibers. Both of these panels 023 and 013 were both PEEK based panels. The crack shown in Figure 6.10 shows fibers that still appear to be bonded with som e thermoplastic. However, in Figure 6 -11, the carbon fibers appear to have pulled clean fr om the matrix material, clearly causing a weak area in the composite. 52 Figure 6-10: Photomicrograph of panel 023 co upon 02 (PEEK) Figure 6-11: Photomicrograph from panel 013 coupon 01 (PEEK) 53 Figure 6-12: Photomicrograph of panel 007 coupon 01 (Nylon) The image in Figure 6-12 shows repeated lines that are thought to be left from the cutting process. Finally , in Figure 6-13, another failure surface can be seen with larger, unconnected voids opening up through the thickness of the coupon. Figure 6-13: Photomicrograph of Panel 004 coupon 6 (Nylon ). Shows failure surface (wa shed out scale bar reads 100 µm ) 54 Chapter 7. Discussion 7.1. Fiber Volume Fraction Overall, a comparison of the target fiber volume fraction to the final calculated fractions only shows slight increases in Table 6.1 and Table 6.2. This is good from a manufacturing perspective, making it easy to attain a particul ar volume fraction through production. The slight increases can be attributed to the shrinkage inherent in thermoplastics and which was mentioned in Chapter 2. The slight shrinkage in volume increases the volume fraction of carbon fiber present in the pane ls. Panels 008, 009, 010, and 013 all showed decreases in their respective fiber volume fractions due to their large final depth measurements. This would indicate a lack of consolidation during the molding process, however, there was no correlation between these panels and lowered flexural properties. 7.2. Comparison of Flexural Strength A direct comparison between the nylon and PEEK composites confirms the strength benefits of the more expensive PEEK. The flexural strength results in Table 6.3 and Table 6.4 show that the PEEK flexural strength performance is easily 300 MPa greater than nylon. Comparing the results of only the nylon panels, the data seems to move away from the expected. From previous work, a peak value of flexural strength should be present somewhere in the range of fiber volume fractions tested. However, Figure 6-2 does not show any particular trend. There appears to be a maximum strength around 0.42 fiber volume fraction, or 0.53 fiber weight fraction, at 365.7 MPa . However, panels 006 and 012, which a re around the same volume fraction as panels 004, 003, and 007, show much lower performance at closer to 200 MPa. 55 A performance comparison of the current studyÕs nylon based panels to those found in the literature is shown in Figure 7-1. Clearly, the nylon and PEEK provides an increase in strength over the PET of LuÕs research. However, the comparison to Fiber ForgeÕs quasi -isotropic layup is somewhat disappointing, fa lling well below. However, a continuously aligned layup, even in a quasi -isotropic layup would give a definite strength advantages over the truly randomly oriented fiber alignment used in the current study, in this case over 155 MPa improved flexural stren gth. Finally, the continuously aligned Fiber forge material was expected and did outperform the current study, in this case by 4.5 times the flexural strength. Figure 7-1: Comparison of Nylon based panelÕs flexural strength in current study with literature results . Source of information showed in brackets. Dotted line references current study. The PEEK panelÕs flexural strength results also show no indication of a peak fiber volume fraction in the range tested. The strength values displayed graphically in Figure 6-3 show 9;99=99?99A99:999:;99:=99:?99:A9965 % Weight Carbon Aligned -Nylon [24] 65% Weight Carbon Quasi -isotropic -Nylon [23] 30% Weight Glass Random -Nylon [25] 53% Weight Carbon Random -Nylon Current Study Flexural Strength, MPa Material System 56 relatively close results for the fiber volume fraction range, particularly when the standard deviation is included. The results of this study were compared directly to data from the literature in Figure 7-2 below. Again, this comparison shows first the limitations of using a randomly oriented reinforcement. The aligned laminate is clearly over double the strength of the maxim um strength found in this study in the directi on tested. However, the current study shows promise, outperforming the quasi -isotropic laminate by 14 MPa. Figure 7-2: Comparison of PEEK based panelÕs flexural strength in current study with literature r esults . Source of information showed in brackets. Dotted line references current study. 9;99=99?99A99:999:;99:=99:?99:A99;99968% Weight Carbon Aligned -PEEK [18] 68 % Weight Carbon Quasi -isotropic PEEK [18] 68% Weight Carbon Ground -PEEK [18] 20% Weight Carbon Random -PEEK [26] 42 % Weight Carbon Random -PEEK Current Study Flexural Strength, MPa Material System 57 Figure 7-3: Flexural strength c omparision of current study to previous work by Lu and Caba at close fiber volume fraction. References are in brackets. As a final comparison of flexural strength, the strengths of several material systems from Lu [6], Caba [28] and the current study are all compared in Figure 7 -3. The obvious performance of PEEK can be seen. However, there does not seem to be an increase in strength from the carbon reclaim and nylon to the new carbon and nylon. This, again, could be a result of processing techniques between the two studies. 7.3. Comparison of Flexural Modulus A comparison of the flexural modulus data between the nylon and PEEK based panels was much more similar than the strengths. Figure 6-4 and Figure 6-5 in Chapter 6 show that, for both material systems, the flexural modulus values are around 18 GPa Ð 30 GPa. However, as with the ;9>;9<010020030040050060070036% Volume Carbon Random -PP [28] 34% Volume Carbon Random -PET [6] 34.7% Volume Reclaim Carbon Random -Nylon 36.5% Volume Carbon Random -Nylon Current Study 35.4% Volume Carbon Random -PEEK Current Study Flexural Strength, MPa Material System 58 case of the flexural strength, no indication of a peak modulus can be found. There is instead, for both material systems, a trend of increased stiffness with increasing fiber volume fraction. The results of the nylon based panelÕs show, between fiber volume fractions of 0.32 to 0.42, the flexural modulus incr eased by as much as 10 GPa. There were a few exceptions to the trend, namely panel 021. However, this particular panel had a much larger standard deviation indicating a wide range of values around the panel. A comparison to the flexural moduli found in the literature for nylon based panels is compared to the current study. This showed the same results as for the flexural strength. The stiff carbon fibers increased the modulus of the nylon based composite. However, the current study again failed to attain , a s was expected, the modulus of the quasi -isotropic material created by fiber forge. Figure 7-4: Comparison of Nylon based panelÕs flexural modulus in current study with literature results . Source of information showed in brackets. Dotted line references current study. 02040608010012014016018020065 % Weight Carbon Aligned -Nylon [24] 65% Weight Carbon Quasi -isotropic -Nylon [23] 30% Weight Glass Random -Nylon [25] 53% Weight Carbon Random -Nylon Current Study Flexural Modulus, GPa Material System 59 The results for PEEK panels also showed a general increase with fiber volume fraction. The best results, found at 0.35 fiber volume percent, produced a flexural modulus of 30.25 GPa, which is comparable to the literature. A comparison of the literature results to the current study can be found in Figure 7-5 below. Again, this shows the limitation of randomly oriented composites. Figure 7-5: Comparison of PEEK based panelÕs flexural modulus in current study with literature results . Source of information showed in brackets. Dotted line references current study Again, the results are compared with the previous work at approximately the same fiber volume fraction in Figure 7 -6. This shows again the similar performance of both the reclaimed carbon fiber and new carbon fiber in nylon. 02040608010012014068% Weight Carbon Aligned -PEEK [18] 68 % Weight Carbon Quasi -isotropic PEEK [18] 68% Weight Carbon Ground -PEEK [18] 20% Weight Carbon Random -PEEK [26] 42 % Weight Carbon Random -PEEK Current Study Flexural Modulus, GPa Material System 60 Figure 7-6: Flexural Modulus comparision of current study to previous work by Lu and Caba at close fiber volume fraction. References are in brackets. 7.4. Planar Isotropy All panels performed equally in two directions apart from pan el 007. In this case an marked decrease of properties was observed when the testing angle was changed by 90 ¡. As indicated by Table 6.7, the flexural strength decreased by almost 90 MPa in the 90 ¡ reference direction. The results have also been shown graphically in Figure 7-7. From the literature and knowledge of composites, an alignment of fibers in the 0 ¡ would give that direction an advantage in strength. Some visible examples of this fiber alignme nt can also be seen in Figure 6 -8. 9>:9:>;9;><9<>C!D/1E.&!7'(F/+! -'+4/.! GM31/+!!!!!!!!!!!!!!!!!!!! 7E((&+%!N%E43 <>B=C!D/1E.&!7'(F/+! -'+4/.! G5JJO!!!!!!!!!!!!!!!!! 7E((&+%!N%E43 Flexural Modulus, GPa Material System 61 Figure 7-7: Flexural strength results from individual coupons of Panel 007 7.5. Microscopy The microscopy results showed several interesting insights into the micro -scale layers of the composite panels. All panels showed good consolidation with few voids present. This points toward good consolidation pressures and lack of excess moisture and air in processing. Particularly, Figure 6-6 shows the entire edg e of a test coupon, with little void content , no apparent gaps in the matrix and fibers . One interesting image, shown in Figure 6-13, shows a failure where voids might have been a possible cause. The fractured surface appears to be at multiple, unconnected planes moving through the depth of the composite. It is possible that this area of the panel did not reach the compaction necessary , leaving voids between the fibrous mat layers. This allowed them to open during testing, causing failure. Other causes of these voids ;99;>9<99<>9=99=>9>9901020304050607080910Flexural Strength, MPa Coupons Tested 62 could be from the sample preparation before testing. In Figure 6 -13, the surface appears rough where it was in contact wi th the wet -saw. Because no polishing was done before the SEM was used, the effects of the cutting process can clearly be seen. This also becomes clear in Figure 6-12 with the repeated diagonal cut lines. As was discussed in Chapter 4 and later in Section 7 .6, special effort was taken to fully separate the carbon fiber tows before being mixed in the white water. However, there are several obvious indications where these methods were not fully successful. The image given in Figure 6-10 shows a large bundle of fibers, closely aligned with little matrix present coming o ut of a fracture. These bundles could indicate areas tows that were not completely separated during the mixing process, which should result in weaker areas of the panels. Indeed, in Figure 6 -9 the image shows the fracture of a particularly strong panel. Th e area deeper in the panel seems to show a good randomization of fiber orientation, at least what can be seen from this surface. The area around the fracture, however, shows two major fiber bundles, angled at different directions. This could have been the cause of this particular failure spot. Reinforcement and matrix bonding were also observed from the SEM micrographs. Two contrasting images can be seen in Figure 6-10 and Figure 6-11. In the case of Figure 6-10, the carbon fibers showing in the fracture surface appear to be covered in small amounts of resin, indicating that it bonded well to carbon fiber surface. Contrasting that, the image in Figure 6-11 shows very clear, clean fibers that pulled out during fracture. This would indicate fibers that never bonded well to the matrix and were able to pull out easil y. 63 7.6. Manufacturing Process This process was originally used in an automated machine where both high volume and consistency could be achieved. This was reflected in the results LuÕs [6] results . Due to the limiting sizes of the mixer/pulper and sheet form er, the batch size was relatively small and introduced small, unintended variations into the data. In addition, the processing parameter differences between PEEK and Nylon also added complexity to the process. The following sections attempt to report all o bservations during the production of panels as a method of critiquing the production process. There are also possible sources of error from the testing process. This comes from test setup and the testing process itself. Attempts were made to minimize these , such as preloading the specimens during flex testing to reduce movement and false readings. 7.6.1 Surfactant in White Water Originally, Rhodameen VP -532/SPB was used as the surfactant in making the white water for panel production. However, this product had b een discontinued which lead to several panels being created with an alternate surfactant. In these cases, the surfactant was made Òin -houseÓ by mixing 20% by weight POE tallow ammine deionized water. For the nylon based panels affected, there appeared to b e no difference between the two surfactants. However, for the PEEK fibers, which are considerably less dense, the concentration of POE tallow ammine appeared to be too low. In these cases, the thermoplastic PEEK fibers floated to the surface of the white w ater. A new batch of 20% by weight POE tallow ammine and deionized water was mixed and gave better results. No cause for why this re -mixed solution worked better could be 64 found. This was used until Rhodameen VP -532/SPB was found and used. For all pa nels us ing Rhodameen VP -532/SPB , the solution had passed its expiration date. 7.6.2 Separation of Carbon Fiber As was previously introduced in Chapter 3, the chopped carbon fiber tows were separated into smaller fibers before being added to the white water during the mixing step. As one of the most time consuming and physically draining steps, the method used to separate carbon fiber was altered several times. The initial method used several panels was to simply hand separate the fibers directly into the white water im mediately after adding the thermoplastic fibers. This process would take 20 -25 minutes per batch and led to inconsistencies in the quality of separation. This method required the operator to hold their arms above the pulper for extended periods of time whi le squeezing and twisting the chopped tows to pull them apart. As the carbon fiber was being separated above white water, being mixed at low rpms, it was necessary to complete this step while wearing a mask and standing. Because of fatigue this method intr oduced, the level of separa tion decreased with each batch. As a way to combat this time intensive step during the mixing stage, the weight of carbon fiber for a particular panel could be pre -separated at an earlier date. This would allow the operator to s it while separating for several hours which produced more uniform separation of fibers. Unfortunately, there was no method of characterizing the amount of separation. To test to effects of fiber separation, panel 003 was made with non -separated fibers, while 004 was hand separated. The strength d ata through testing of panels 003 and 004 indicated that there was no relationship between simply adding the chopped tows versus separating them 65 before. This increased the productivity of the process by removing t he separation step. During this step, carbon fiber tows were simply added to the white water after the thermoplastic. The mixture was then pulped for 10 minutes at high rpms resulting in the separation of the carbon fiber. Again, the only method used was a visual inspection of the effectiveness of this process change. However, the strength results indicated that the properties remained the same. A new hybrid method was also used where carbon fiber s would be separated in the pulper alone, in tap water. Generally, 30 grams of chopped carbon fiber tow were mixed with 2 liters of tap water and mixed for 10 minutes at a high rpm. The mixture was then drained and the resulting fibers were dried in an oven, 200 ¡C for 10 minutes. This ensu red all water was removed and the appropriate fiber weight was correct. This pre -separated fiber could then be added in the normal way immediately after the thermoplastic fiber. This hybrid method combines the self-separation involved in the mixing step tw ice, allowing for full fiber separation before the final product was created. Some issues that could arise from the hybrid method are the possibility of washing off fiber sizing. Additionally, the extra time in the pulper could break some fibers or create surface damage that could lower the bonding properties in the composite resulting in decreased strength. 7.6.3 Inconsistent Drying of Mats During the production of mats, a large TPS Blue M oven was used for the drying step. Again, comparing back to previous work, which used a continuous heating system, some inconsistencies were introduced in the process. Because of the large scale of the oven compared with the smaller mats, it became difficult to achieve uniform heating and drying. In some cases, this became so extreme that an individual mat would be one half over dried while the other side 66 seemed untouched. The large temperature gradient made it difficult to precisely flash dry the mats which is dependent on introducing the mats to the thermoplastic melt temperature for a specific amount of time. Again, in order to combat this problem, the method was altered several times although with less f requency than the fiber separation and mixing steps. In order to control and better regulate the temperature gradient in the oven, the outer dampers were used to restrict airflow. However, this sometimes resulted in fault from a pressure switch that remove d heater power. This problem became more acute when the oven was pushed to higher temperatures due to the high melting temperature of thermoplastic PEEK. Finally, the oven requires that the blower motor must not be turned off, which occurs when the door is open, above 300 ¡C. This introduced problems when drying the PEEK based mats, whose melting temperature was 250 ¡C. Because of this limitation, the correct drying time became tedious to discover and achieve. Possible variations in performance due to these i ssues could be large loss of fibers before consolidation or from re -melting of some thermoplastic. The loss of fibers primarily comes from under drying of the mat. The drying step is primarily to remove remaining water but also to lightly melt the thermopl astic to bind all fiber together before stacking in the mold. By not melting the matrix fibers , both reinforcement and matrix, are lost during transport, cutting and placement in the mold. To combat this , all mats were weighed immediately before being plac ed in the mold. Conversely, over drying the mat results in a lack of matrix material in some areas, as the thermoplastic had a tendency to run toward the bottom of the mat during overheating . This would also be indicated by a change in color after drying. 67 Both of these phenomenon sometimes occurred on the same mat, making it difficult to control. As a best practice, this study finds that locating the mats toward the center -left of the oven produced the most even results. Leaving space to increase airflow ar ound the mats also removed some of these problems. As these problems occurred in several of the mats, there was no data to correlate these to strength results. 7.6.4 Cutting testing coupons The diamond tipped wet saw used to cut panels made straight cuts diffi cult and also introduced large volumes of water. The imprecise nature of cuts made measuring the coupons difficult, and sizes are therefore averages of widths for a given coupon. The ASTM standard also requires perfectly rectangular coupons. Although the c oupons were towel dried immediately cutting, some moisture absorption was suspected and could have altered strength characteristics and results. In order to avoid this, coupons were allowed to air dry 24 hours before being tested. 68 Chapter 8. Conclusion This thesis revealed some important insight into the wet-lay paper making process used with thermoplastic nylon and PEEK. The goal of this thesis was to build off the work of Lu [6] by branching into different matrix material systems: carbon -nylon and carbon -PEEK. Although the results did not indicate the optimal fiber volume fractions of either material system, several other conclusions have been found. In particular, this research has given unique insight into the production process itself, including sources of error and possible ways to responding to them. The flexural strengths and moduli of both nylon and PEEK based composites made with this method have been obtained for a narrow band of fiber volume fractions and compared to the literature. 8.1. Conclusions The overall process produced quality panels with good in -plane isotropic behavior. Apart from one panel, all those tested show ed little variation in flexural properties measured orthogonally. The process itself was limited by the size and availability of the e quipment and materials. However, this process could easily be automated to reduce cycle times and error introduced in small batch processes. Only small percent changes in target fiber volume fractions versus final fiber volume fractions are also encouragin g for future research. The nylon based panels show some promise with regard s to flexural strength. When compared to the literature results, the current study , which produced a flexural strength of 365 MPa in one case, performed only marginally less than s ome current market products. The modulus results were less than satisfactory, attaining a maximum of only 28 GPa. This was 69 below the results found in research. Further study , with the comments made in Section 7.5, will no doubt yield higher quality panels with better performance. PEEK carbon fiber composites show great promise, both in the literature and in the results of this thesis . Given the lack of a trend in flexural strength for the volume fractions produced, more volume fractions tested. The flexural strength results were as high as 640 MPa. However, as with the nylon based panels, the flexural modulus resul ts were somewhat disappointing attaining only 30 GPa. The photomicrographs collected using the scanning electron microscope showed some problem ar eas with bonding. Large bundles of aligned fibers around failure areas indicated that fiber separation may be a problem. In addition, clean fiber pull outs could indicate poor bonding between the carbon fibers and nylon. This could be linked to the mixing step and pos sible loss of the fibers sizing or nylon absorbing water. Finally, the overall process yielded good results. Through the progress of this thesis, several problem causing manufacturing steps have been observed and solutions found. Due to the availability of equipment, such as drying ovens and pulper size, some pro blems continued throughout the research, possibly producing bias in results. However, from a manufacturing perspective, this process could easily be automated and used in high volume composite production. 8.2. Future Work Overall, more work should be done on t his process and materials. In order to remove many sources of inconsistencies with panels, the whole process should be streamlined, or e ven automated. Proper equipment, such as a smaller oven or larger pulper, would produce more 70 consistent mats and final c omposite panels. This would remove inconsistencies and make data comparison more meaningful. Another area of research that should be performed next would be morphology of the PEEK and nylon composites. The research pointed toward large dependence of semi-crystalline polymer compositesÕ performance depended on the structure formed while cooling. Due to the nature of randomly oriented composites, research should be done to obtain the effects a random orientation of fibers h as on the morphology of these polymers . This could point to problems with cooling rates in the production process. Research should also be done into the effects of water absorption by these nylon systems. The literature review revealed nylons strong water absorption. The data from thi s thesis also indicated possible loss of flexural strength from water absorption during sample cutting. Finally, the range of fiber volume fractions should be expanded to find an optimal fraction for both strength and modulus performance. This process cou ld also be aided by the addition of tensile testing. This will add complications during testing but will make the data more comparable with other research results. 71 APPENDI CES 72 Appendix 1: Individual Sample Results from Flexural Testing Table A. 1: Flexural strength data for PEEK panels tested. All values are in MPa. Sample # Plate Number 010 013 023 025 026 1 684.91 460.97 598.70 506.87 564.05 2 635.67 640.88 607.81 558.02 560.64 3 532.48 565.50 638.62 578.33 479.07 4 536.22 573.69 716.45 495.34 538.20 5 709.33 527.13 - 525.56 570.00 Average 619.72 533.53 640.39 532.82 542.39 Std. Dev 82.34 66.04 53.51 34.77 37.39 Table A. 2: Flexural modulus data for PEEK panels tested. All values in GPa. Sample # Plate Number 010 013 023 025 026 1 26.89 25.30 27.93 24.26 29.22 2 34.27 26.27 27.55 25.33 30.83 3 23.05 33.00 31.16 30.17 27.51 4 20.78 31.28 34.33 20.55 32.15 5 39.77 36.48 - 23.72 36.66 Average 28.95 30.47 30.24 24.81 31.27 Std. Dev 7.93 4.68 3.17 3.49 3.48 73 Table A. 3: Flexural strength data for nylon panels tested. All values are in MPa. Sample # Plate Number 003 004 006 007 008 009 011 012 021 022 1 308.03 326.32 196.80 416.66 179.19 222.86 230.99 224.44 156.67 213.26 2 358.97 267.16 180.53 344.21 158.76 210.47 227.58 179.90 174.79 196.68 3 486.33 269.36 188.16 433.66 198.32 197.55 256.46 190.20 169.33 189.55 4 343.94 253.42 186.22 416.63 192.92 208.96 206.24 203.20 148.12 197.40 5 353.70 343.81 214.24 439.92 186.71 186.31 245.93 166.77 189.30 216.76 6 313.83 325.70 180.68 301.27 180.79 187.88 208.46 252.32 - - 7 315.40 372.14 210.36 325.55 194.97 169.90 195.46 231.13 - - 8 277.39 307.30 191.74 317.12 184.90 169.54 178.07 220.81 - - 9 359.14 384.07 224.60 323.12 198.99 217.79 207.84 230.76 - - 10 413.79 306.04 157.82 339.22 210.49 162.24 221.33 166.27 - - Average 353.05 315.53 193.12 365.74 188.60 193.35 217.84 206.59 167.64 202.73 Std. Dev 59.92 43.98 19.39 54.18 14.13 21.57 23.49 29.82 16.00 11.69 Table A.4: Flexural modulus data for nylon panels tested. All values in GPa. Sample # Plate Number 003 004 006 007 008 009 011 012 021 022 1 22.88 18.14 20.20 36.26 21.55 24.07 18.41 23.34 14.04 23.47 2 16.85 16.85 19.96 28.82 18.95 21.74 26.44 19.15 2.73 18.83 3 8.30 19.62 20.52 35.02 21.31 23.03 22.85 21.21 18.21 18.91 4 22.56 18.72 21.38 31.79 22.32 21.64 14.20 21.27 15.56 17.71 5 32.77 21.24 22.76 36.57 17.62 19.85 24.40 20.52 18.39 16.70 6 23.34 19.00 24.26 21.92 17.27 22.98 18.80 31.39 - - 7 23.91 20.33 28.70 25.83 17.63 19.55 14.33 25.82 - - 8 21.65 20.29 22.72 25.59 17.03 17.74 10.97 20.23 - - 9 27.95 23.42 28.11 23.40 17.85 19.67 17.98 23.92 - - 10 18.47 24.68 21.89 23.47 23.29 20.48 20.62 20.31 - - Average 21.87 20.23 23.05 28.87 19.48 21.07 18.90 22.72 13.78 19.12 Std. Dev 6.54 2.38 3.12 5.65 2.38 1.96 4.86 3.66 6.45 2.60 74 Appendix 2: MATLAB code 75 76 77 78 79 BIBLIOGRAPHY 80 BIBLIOGRAPHY [1] A. B. Strong, Fundamentals of Composites Manufacturing: Materials, Methods, and Applications . Dearborn, MI: Society of Manufacturing Engineers, 1989. [2] I. Y. Chang and J. K. Lees, ÒRecent Development in Thermoplastic Composites: A Review of Matrix Systems and Processing Methods,Ó J. Thermoplast. Compos. Mater. , vol. 1, no. 3, pp. 277Ð296, Jul. 1988. [3] P. K. Mallick, Fiber -Reinforc ed Composites: Materials, Manufacturing, and Design , Third. Boca Raton: CRC Press, 2008. [4] G. R. Belbin, P. A. Staniland, M. Bhatt, and S. F. Bush, ÒAdvanced Thermoplastics and their Composites [and Discussion],Ó Philos. Trans. R. Soc. Lond. Ser. Math. Phys. Sci. , vol. 322, no. 1567, pp. 451Ð464, Jul. 1987. [5] J. E. G. Jr and G. P. Weeks, ÒMethod of making fiber reinforced porous sheets,Ó US5194106 A, 16 -Mar-1993. [6] Y. Lu, ÒMechanical Properties of Random Discontinuous Fiber Composites Manufactured from Wetlay Process,Ó 22 -Aug -2002. [Online]. Available: http://scholar.lib.vt.edu/theses/available/etd -08132002-163349/. [Accessed: 25 -Jun -2015]. [7] C. Soutis, ÒCarbon fiber reinforced plastics in aircraft construction,Ó Mater. Sci. Eng. A , vol. 412, no. 1Ð2, pp. 171Ð176, Dec. 2005. [8] M. Balasubramanian, Composite Materials and Processing . CRC Press, 2014. [9] C. A. Mahieux, ÒCost effective manufacturing process of thermoplastic matrix composites for the traditional industry: the example of a carbon -fiber reinforced thermoplastic flywheel,Ó Compos. Struct. , vol. 52, no. 3 Ð4, pp. 517Ð521, May 2001. [10] M. Biron, Thermoplastics and Thermoplastic Composites . William Andrew, 2012. [11] F. N. Cogswell, Thermoplastic Aromatic Polymer Composites: A Study o f the Structure, Processing and Properties of Carbon Fibre Reinforced Polyetheretherketone and Related Materials . Elsevier, 1992. [12] N. Klein, G. Marom, and E. Wachtel, ÒMicrostructure of nylon 66 transcrystalline layers in carbon and aramid fibre reinf orced composites,Ó Polymer , vol. 37, no. 24, pp. 5493 Ð5498, Nov. 1996. 81 [13] M. S. A. Rahaman, A. F. Ismail, and A. Mustafa, ÒA review of heat treatment on polyacrylonitrile fiber,Ó Polym. Degrad. Stab. , vol. 92, no. 8, pp. 1421 Ð1432, Aug. 2007. [14] A. G . Gibson and J. -A. M„nson, ÒImpregnation technology for thermoplastic matrix composites,Ó Compos. Manuf. , vol. 3, no. 4, pp. 223 Ð233, 1992. [15] P. Mitschang, M. Blinzler, and A. Wıginger, ÒProcessing technologies for continuous fibre reinforced thermoplastics with novel polymer blends,Ó Compos. Sci. Technol. , vol. 63, no. 14, pp. 2099Ð2110, Nov. 2003. [16] B. A. Davis, P. J. Gramann, and A. C. Rios, Compression Molding . Henser, 2003. [17] L. Ye, K. Friedrich, J. K−stel, and Y. -W. Ma i, ÒConsolidation of unidirectional CF/PEEK composites from commingled yarn prepreg,Ó Compos. Sci. Technol. , vol. 54, no. 4, pp. 349Ð358, 1995. [18] G. C. McGrath, D. W. Clegg, and A. A. Coll Yer, ÒThe mechanical properties of compression moulded reconsti tuted carbon fibre reinforced PEEK (APC -2),Ó Composites , vol. 19, no. 3, pp. 211 Ð216, May 1988. [19] J. M. Whitney and M. Knight, ÒThe relationship between tensile strength and flexure strength in fiber -reinforced composites,Ó Exp. Mech. , vol. 20, no. 6, pp. 211Ð216, Jun. 1980. [20] C. Zweben, ÒThe Flexural Strength of Aramid Fiber Composites,Ó J. Compos. Mater. , vol. 12, no. 4, pp. 422Ð430, Dec. 1978. [21] R. E. Bullock, ÒStrength Ratios of Composite Materials in Flexure and in Tension,Ó J. Compos. Mate r., vol. 8, no. 2, pp. 200 Ð206, Apr. 1974. [22] W. H. Bowyer and M. G. Bader, ÒOn the re -inforcement of thermoplastics by imperfectly aligned discontinuous fibres,Ó J. Mater. Sci. , vol. 7, no. 11, pp. 1315 Ð1321, Nov. 1972. [23] Fiberforge, ÒFiberforge Ca rbon/Nylon Quasi -Isotropic Blank Representative Material Data.Ó . [24] Fiberforge, ÒFiberforge Carbon/Nylon Unidirectional Blank Representative Material Data.Ó [25] J. L. Thomason, ÒThe influence of fibre properties of the performance of glass -fibre -rei nforced polyamide 6,6,Ó Compos. Sci. Technol. , vol. 59, no. 16, pp. 2315 Ð2328, Dec. 1999. [26] M. Kurokawa, Y. Uchiyama, and S. Nagai, ÒPerformance of plastic gear made of carbon fiber reinforced poly -ether -ether -ketone: Part 2,Ó Tribol. Int. , vol. 33, no . 10, pp. 715Ð721, Oct. 2000. 82 [27] Quantum Composites, ÒQuantum Composites Technical Data Sheet for AMC 8592,Ó Quantum Composites , 07-Sep-2013. [Online]. Available: http://www.quantumcomposites.com/pdf/datasheets/amc/Quantum -AMC-8592%20126-76-86.pdf. [28] A. Caba, ÒCharacterization of Carbon Mat Thermoplastic Composites: Flow and Mechanical Properties,Ó 14 -Sept -2005, [29] A. C. Roulin -Moloney, Ed., Fractography and Failure Mechanisms of Polymers and Composites . Essex England: Elsevier Science Publishers LTD, 1989. [30] D. Purslow, ÒFractography of fibre -reinforced thermoplastics, Part 3. Tensile, compressive and flexural failures,Ó Composites , vol. 19, no. 5, pp. 358 Ð366, Sep. 1988. [31] Toray Carbon Fibers America, Inc, ÒTorayCA T700S Data Sheet.Ó . [32] Minifibers, Inc., ÒMiniFibers, Inc. Technical Data Sheet,Ó MiniFibers, Inc. [Online]. Available: http://www.minifibers.com/techdata/TD -04_NYT66 -0302RR.pdf. [33] ÒWelcome to ZYEX World leader in PEEK fibres.Ó [Online]. Available: http://www.zyex.com/i ndex.html#. [Accessed: 16 -Jul -2015].