Research Article, Biomater Med Appl Vol: 4 Issue: 1

A Production Method for Standardized Continuous Fiber Reinforced FFF Filament

Mohamed Aburaia^1*, Christoph M. Bucher¹, Maximilian Lackner¹, Joamin Gonzalez-Gutierrez², Haiguang Zhang³ and Herfried Lammer⁴

¹Competence Center Digital Manufacturing and Robotics, UAS Technikum Wien, Vienna, Austria

²Polymer Processing, Montan University Leoben, Leoben, Austria

³Rapid Manufacturing Engineering Center, Mechatronic Engineering and Automation of Shanghai University, Shangha, China

⁴Wood K Plus KompetenzzentrumHolz GmbH, Linz, Austria

*Corresponding Author : Mohamed Aburaia
Competence Center Digital Manufacturing and Robotics, Faculty of Industrial Engineering, UAS Technikum Wien, Vienna, Austria
Tel: 431333407770
E-mail: aburaia@technikum-wien.at

Received: April 28, 2020 Accepted: May 13, 2020 Published: May 20, 2020

Citation: Aburaia M, Bucher CM, Lackner M, Gonzalez-Gutierrez J, Zhang H, et al. (2020) A Production Method for Standardized Continuous Fiber Reinforced FFF Filament. Biomater Med Appl 4:1. doi: 10.37532/bma.2020.4(1).123

Abstract

Keywords: Additive Manufacturing; 3D Printing, Filament; Continuous fiber; Carbon fiber reinforcement; FFF; FDM; 3D printing

Introduction

The world of additive manufacturing (AM) has experienced an unprecedented growth over the past few years. This growth is mainly because of affordable machines that became commercially available. Industries are also starting to use additive manufacturing methods as SLS, SLA and the newer MJF, for the lower production costs, compared to traditional methods such as CNC machining, casting or injection molding, but also due to the technical and material advanced, that lead to usable mechanical characteristics. The most common machine is still the FFF desktop printer but is mostly used by consumers or for prototyping purposes. This is due to the lower mechanical properties and poorer surface finish compared to other AM methods.

Technological advances have also been made in the field of FFF manufacturing. Machines are more compact, have better software and are affordable, but the most important advance is in the field of compatible materials. FFF filaments are readily available, ranging from plain thermoplastics, to advanced materials with special characteristics such as thermal, electric or magnetic conductivity, food grade, or composites with fillers such as wood, stone or carbon fiber, with improved mechanical properties. Even though a lot of research has been made in the field of materials for FFF printing, which lead to improved materials, print quality and its popularity, a new material with a high potential for further increasing mechanical properties is still in its early stage of development: continuous fiber composite filaments. Research has proven the high potential of this composite, yet few variations are commercially available at a high cost. There still is a need for research and development in order to make these materials market-ready: methods for producing such continuous fiber reinforced filaments are still not perfect, the methods of printing with such filaments and the needed slicing software are different from those used for usual filaments and are still not fully developed or available.

Research showed that carbon fiber composite filaments are either only compatible with the manufacturers own FFF printers, such as those produced by Markforged [1] or Anisoprint [2], coming at a high price point, or they are compatible, such as the one produced by Wood [3], with a standard diameter of 1.75 mm, but still have flaws and are not yet market-ready.

This paper proposes a new method for producing continuous carbon fiber reinforced thermoplastic filament using a combination of impregnation, coating and coextrusion of the carbon fiber, in order to manufacture 1.75 mm continuous carbon (CCF) filament, with improved fiber bonding and mechanical properties. The goals of this work include the prototype production method of such CCF filaments, but also the development of a prototype apparatus to produce the filament. Furthermore, the method is adaptable and can be used to manufacture different composites and is not be limited to those used for in this work.

State of the Art

Recent publications thoroughly compare, analyze and review possible materials used in reinforced FFF printing: short-fiber composite filaments, alternating extrusion of fiber and matrix and coextrusion of fiber and matrix material [4,5].

Firstly, the most common commercially available fiber reinforced filament is the short/chopped fiber composite filament. Available from multiple producers, both in 1.75 mm and 3 mm filament format, these composites combine thermoplastics (ABS, PLA, PC, PETG, PA) with chopped fibers (glass, carbon, Kevlar) 0.1 to 1 mm long. These composites have better mechanical properties than plain thermoplastics, yet also cause voids in the extrusion due to the reorientation of the fibers during extrusion [6].

The second method of FFF fiber reinforcement is the alternating extrusion of continuous fibers and polymers to create a composite. Commercially available FFF printers, which can reinforce parts with continuous fibers, use a method of sequential printing of thermoplastic and fiber composite. Markforged [1] was the first commercially available desktop FFF printer. It uses two extruders to reinforce parts. One that extrudes a layer of thermoplastic polymer, a second one that embeds carbon, glass or Kevlar composite fiber in the anterior layer. The fibers used as reinforcement are a fiber – nylon (PA6) composite, 400 μm thick. The fibers are coated in a thin PA6 layer in order to bond and stiffen the fibers, so that they can be reliable extruded, but also to create an interphase between the two materials [7].

Another newer commercially available printer is the Anisoprint FFF printer [2], which uses two extruders to manufacture parts: one for polymer filament and a second one that co-extrudes carbon or basalt composite fiber and standard polymer filament. Using this method, the manufactured parts contain about 25% fiber reinforcement.

The fiber composite filaments used by these printers are usually 400 μm thick, containing 1.5k fibers [2]. They are very useful in reinforcing certain parts and details, but the costs and duration to reinforce fully reinforce a larger part increase drastically. This method is not only time consuming, having to print each reinforced layer twice: once with thermoplastic, once with fiber. For these reasons, other methods to reinforce FFF manufactured parts were researched. Experimental setups using modified FFF extruders have been used to simultaneously print thermoplastics and fiber [8-11].By using the movement of the extruded thermoplastic filament; the fibers are pulled into the extruder and extruded together. This method allows the use of untreated fiber tows of multiple widths and comes with lower costs. Furthermore, compared to the inlay of fiber into prints, coextrusion leads to better mechanical properties and shorter manufacturing times.

With many studies proving the mechanical improvements of simultaneously extruding fibers and thermoplastic it is safe to say, that they have a great potential to change the way of FFF printing and used materials. Some problems currently remain unsolved, without any commercially available solution. The studies discussed used custommade extruders that pull the fibers with the melt. As already discovered in own research [11].

This method is not ideal: it cannot be reliably automated, is prone to slippage and tear of the fiber bundle and often results in uncentered fiber distribution.

Another way of co extruding thermoplastic and fibers is to already have them in one composite, but in standard FFF formats (1.75 mm or 2.85 mm) and with lower production cost, unlike those produced by Markforged and Anisoprint: costly and proprietary diameters. In cooperation with Wood [3] different types of fiber reinforced thermoplastic filaments were manufactured and tested. The aim was to eliminate the problems caused by in-situ fiber-polymer melt composite creation, by having the continuous fiber already embedded in the thermoplastic matrix. A continuous fiber filament, 1.75mm thick was manufactured, as it is the most commonly used filament for FFF printing. Different thermoplastics such as PLA, ABS and PA6 were combined with carbon and jute continuous fibers to produce the composite filament. These filaments were successfully used to print reinforced parts.

A recent study [13] investigated the use of natural fibers such as flax fiber as reinforcement for polymer-based filament to be used in FFF manufacturing. These bio composites have improved mechanical properties, compared to plain biopolymers, but are also environmentally friendly, as they use regenerative sources as reinforcement and are not as abrasive on the 3D printer’s nozzle, as carbon or glass fiber. The bio composite filament to be used in FFF printing was produced by extrusion coating of continuous twisted flex yarn with polylactic acid polymer. The printed bio composite showed a uniform distribution of the natural fibers in the cross section and led to the increase of the tensile modulus and strength of more than 4.5 times. The tensile properties of the tested composite were comparable to that of glass fiber – polyamide composite. It was observed, that due to the twisting of the used yarn, a full impregnation of the fibers with thermoplastic was not achieved.

With the goal to increase the polymer-resin interface and reduce voids in the composite, researchers at the Jiaotong University [14] used an independently designed and built apparatus to pre impregnate raw carbon fiber tow with a thermosetting resin. Using multiple rollers submerged in the liquid resin, to direct the fibers, the tow was held longer in contact with the resin before curing, to ensure proper penetration of the polymer in the fibers. This composite filament was then used to print parts. Due to the used matrixpolymer, E-20, a thermosetting resin, the printed parts had to be thermally cured, while buried in sodium chloride powder, to fully harden the resin in the desired shape.

As already discovered by multiple researches [5,14-17], voids in the composite material are also a problem that occur during manufacturing and printing of the material. These residual voids are often a cause for deconsolidation and decompaction of the fiberthermoplastic composite. The Laboratory of Composite Materials and Adaptive Structures of ETH Zurich [18] investigated methods to minimize the void content that results from the filament being reheated above its melting point in the process of printing. With a method that uses cycling softening of the composite, the void content in the material was reduced by 80%. In order to achieve this result, a supplementary pultrusion module, consisting of 4 chambers, each gradually compacting and decompacting the composite was used before the actual 3D printer extruder. Not only voids in the composite are problematic for FFF, but also the poor adhesion of the fibers to each other and to the matrix material. By simply coextruding fiber and thermoplastic, there is not enough contact time and pressure for the thermoplastic to seep through the fibers and fully impregnate them. Because of this, the innermost fibers are free to move in relation to the matrix material, as only the outer layer of the bundle is merged to the matrix. This only increases the mechanical properties of the matrix material slightly, but could have a far higher impact, if all the fibers were merged.

A method to counter this problem is to modify the surface of the carbon fiber bundle, in order to increase infiltration rate and the interfacial strength [10]. With the use of matrix compatible solvents, surface active, emulsifying and antifoaming agents, a solution can be prepared, in which the fibers are soaked. Using this method for carbon fiber and PLA composite, few voids remained as confirmed by SEM micrographs, which lead to a 164% improvement compared to untreated, raw carbon fiber tow.

A similar experiment was internally conducted using carbon fiber and ABS as a matrix. ABS has an excellent solubility in acetone, which leads to an easier process of infiltration and impregnation of the fibers, than with other solvent- polymer solution. In this experiment, the raw carbon tow was submerged multiple times in the solution to deposit more material with each pass. A stream of hot air was used to evaporate the acetone and harden the fiber composite between each submerge. After a certain number of submerges, the composite fiber was pulled through a heated nozzle, to smooth out the surface of the composite fiber and achieve more consistent diameter. The designed apparatus can be seen in (Figure 1).

Even though the goal was to produce 1.75 mm filament by multiple coating of the fiber that was not achieved, as the surface of the fiber would start to become very irregular after a certain number of passes. The work did however succeed in showing the potential of this method of producing fiber composite. No further testing, regarding fiber content, distribution, impregnation, voids or mechanical properties, was conducted.

Methods

This work will focus on producing ABS continues carbon fiber composite filament, with a diameter of 1.75 mm. For the matrix material, acrylonitrile butadiene styrene (ABS) was chosen. This thermoplastic is commonly used for a variety of consumer products, has good mechanical properties, but is also soluble in acetone, a widely available, common and stable solvent. The main reason this material solvent pair was chosen, was its relative safety compared to other solvents needed for other thermoplastics, such as hot benzene, dichloromethane, rdioxane, PLA solvent, which are hazardous and toxic, and may be a health risk, as this research involves being in prolonged contact with the uncontained solvent that was used.

Even though the ABS infusion would provide the best adhesion to ABS extrusion coating, another problem/goal to keep in mind is sustainability and plastic pollution. The last few years have shown us, that we have been abusing plastics, which lead to a global crisis. Of course, plastics have had a great impact on the development of technology, but they are also being unnecessarily misused and improperly disposed of. Most thermoplastics can be recycled in a certain degree, if free from contaminants, but composites are currently not recycled on a large scale and contribute to environmental pollution. In order to reduce the environmental impact of the composites researched, biodegradable plastics will be used for most of the composition of the filament. The bio plastic used is polylactic acid, made from renewable resources such as corn starch.

Furthermore, it is food safe, suitable for medical applications and can be composted. It also has a higher melt flow index, compared to other common thermoplastics, which makes the test production of filament easier. PLA also comes with some downsides: it has lower mechanical properties that other engineering or high-performance thermoplastics, which might make it unsuited for certain applications. In such cases, the use of other non-biodegradable thermoplastics may be necessary. For mass production, even the ABS infusion used in this work can be substituted to PLA infusion. In an industrial setup, the process can be automated and contained, so that even health hazardous solvents needed for PLA can be used without health risks.

Mentioned in the state-of-the-art chapter, continuous fiber reinforced filament was already successfully produced in previous trials by WoodK. In that case, PLA matrix filament with 1.5K raw fiber tow reinforcement was produced. The filament had better mechanical properties than regular PLA and was successfully printed using a standard FFF desktop printer with a 2mm nozzle. But then some of its problems were discovered. Firstly, because of the continuous fiber reinforcement in the filament, certain printing parameters have to respected, as later described in chapter C Printing parameters for CFR filament. Because of these constraints, the width of extrusion while printing is around 2 mm, varying with the layer height. Because of this, details and sharp corners cannot be printed. Oftentimes, parts only need reinforcement in certain areas. Using this filament, parts can either be completely printed with the reinforcement, which is not needed everywhere and just increases the cost of the part, or it can be used to reinforce certain areas of the part, the rest being printed using PLA. The problem with reinforcing is that the ratio of fibers to matrix is very small: a line of 2 mm width has to be printed in order to have a 1.5k fiber bundle embedded. This could also be accomplished by the filament produced by Markforged or Anisoprint, which also has 1.5k fiber but a far smaller diameter, thus conserving details.

Because printing a whole part with fiber reinforcement is almost never needed, a goal of this work is to produce a filament used for reinforcing areas of a part. Because the width of the reinforcement will be at least 2 mm, as dictated by the necessary nozzle diameter, it should at least maximize the fiber content and the strength increase. For this reason, further testing was conducted with 12k carbon fiber tow.

To improve the fiber reinforced filament, multiple studies [16-19] tested pretreatments of the fibers by infusing the fiber bundle to increase the adhesion between fibers and matrix material. Following these discoveries, an already presented work, tried to produce 1.75mm carbon fiber reinforced filament by repeatedly coating the fiber bundle in an acetone ABS solution. The goal of the method was to also increase the strength of the final filament by infusion and produce a standard FFF 1.75 mm filament at the same time, using the same process and apparatus.

The apparatus built for that purpose (Figure 1) was designed to coat the fiber multiple times and dry the previous layer in- between submerges. The result of the work showed that by coating the fiber for up to 20 times was not sufficient to achieve the desired thickness. In principle the method would work with some modifications such as longer drying times in-between coating and possibly higher densities of ABS solution. Even with those modifications, one problem would persist: the acetone has a low boiling point of 56 °C. That means that even at room temperature, the solvent evaporates quickly, thus increasing the concentration of ABS in the solution if not adjusted. Even if that would be adjusted, there would be a high loss of acetone through evaporation. The evaporated acetone could be recovered by condensation processes, but that would further increase the complexity and cost of the production process.

Another method and process are needed in order to produce a filament of 1.75mm diameter with a pretreated continuous fiber core. The method developed consists of two production steps. The first one is to infuse and coat the fiber bundle with the desired solution. This fiber is then dried and used for the second step. This second step uses a thermoplastic extruder to coat the fiber core with the desired thickness of thermoplastic. Using these two steps, a multitude of materials can be used, both for the infusion and the extrusion coating. Different diameters of fiber bundle can be used, as the extrusion coating will compensate for the different core thicknesses.

Impregnation and Coating of Raw Fibers

For the impregnation and coting of the fibers, a test apparatus was developed and build. The schematic of this apparatus can be seen in Figure 2.

The fiber tow is firstly pulled through the ABS acetone solution. Test showed that a 10% ABS concentration is a good starting point, as it can be absorbed quickly enough into the bundle while it is submerged. During this phase, the fiber bundle is kept flat to increase its contact area with the solution and promote total penetration of the solution into the fiber bundle. The bundle is then redirected to a funnel which serves two purposes. Firstly, it removes excess solution, returning it to the rest of the solution and secondly, it bundles and compresses the fibers into a round bundle. At this point, the fiber bundle is still wet and can easily stick to anything. Previous tests have shown that in order to avoid damaging the fiber bundle, it should not be touched while wet. Because of that, it must be dried before redirecting it to the winder. For this purpose, the wet fiber bundle travels through a drying tunnel. The drying tunnel consists of a heated aluminum tube of 1m length. Through this tube using a ventilator, air is blown against the movement of the fiber. The tube is heated with a 30W heating band clamped on the top of the tube. This ensures that the temperature in the tube is the highest at the end of drying process and lowest at the beginning.

This temperature gradient is needed to ensure proper drying of the fiber. The goal is not to dry the outer surface of the fiber to fast, as that would not allow the still wet inner core to dry. If the coated fiber is still wet during the second step, the acetone that is still present will quickly evaporate during the extrusion process, which produces voids and bubbles in the final filament. Voids in the filament are a well-known flaw, causing reduced mechanical properties. By also forcing air through the tube, the acetone vapors are pushed out of the drying tunnel, reducing the air acetone vapor concentration and increasing the evaporation rate.

The drying tunnel should be longer, as the used 1m tunnel was at its drying limits, even at the lowest speed of the used winder. The out coming air, at the bottom of the tube was cold, due to the evaporation of the acetone and the infused fiber bundle was still malleable, but not sticking. Therefore, it was allowed to dry out completely at room temperature, before being used in the next step.

The same process could also be designed horizontal. This would reduce the needed height and would improve handling and operation. The length of drying tunnel could also be increased easily. The downside of the horizontal method is the slacking of the fiber bundle. The drying zone would either have to be of a larger diameter or could consist of two vertical plates, allowing the fiber to slack without touching anything.

Co-extrusion process

The second part of the process involves co-extruding the pretreated fiber with thermoplastic melt. This step coats the fiber bundle with a defined thickens of thermoplastic (Figure 3).

To melt and inject the thermoplastic melt into the coextrusion die, a Noztek Pro HT screw extruder was used. The fabricated die was connected to the extruder.

After exiting the die, the produced filament is still hot and over its glass transition temperature. In this state it cannot be handled or wound onto a spool, as it would fuse to itself. Therefore, a cooling step is needed after the extrusion. For this purpose, a water bath was used. The filament is pulled through recirculation cold water, which cools it done and hardens the filament. The filament is then wound onto spools. For this step, a Noztek Filament Winder was used.

Ideally a tolerance puller would also be used, to ensure constant diameter. While that is needed for commercial production, for this test setup, it was not necessary, as the diameter of the resultant filament was constant enough to be compatible with desktop FFF printers and could be used to prove the concept.

Printing parameters for CFR filament

For the continuous fiber reinforced filament to be printed using FFF machines, a large nozzle is needed. The nozzle should be at least the diameter of the filament or larger. With plastic only filaments, a smaller nozzle can be used, as the plastic melt can be reshaped and redirected through the smaller nozzle. The continuous fiber reinforced filament cannot be squeezed through a smaller nozzle, as the fibers still remain solid. Even if the nozzle would allow the fiber bundle to pass without constraint, the melt surrounding the fibers would not have enough space to exit, thus being pushed back, against the movement of the filament. The plastic melt will reach the cool side of the extruder, where it will quickly harden and jam the extruder.

To avoid this problem, the whole filament has to be extruded simultaneously. This is only possible with a nozzle that has on orifice with the same diameter as the filament. Because the filament is not perfectly constant, a larger nozzle should be used, to avoid any backflow. In this case, a 2 mm nozzle was used.

Another parameter that is specific to printing this type of filament is the extrusion rate and speed. Because the fiber reinforcement in the filament is not flexible or stretchable, it has to be laid onto the build surface at the same rate as the movement speed of the print head. If these two speeds are not synchronized, two problems can occur. The first one is depicted on the left side of Figure 4. When the extrusion speed is too high, the filament and the fiber within will start to curl or wave. If the extrusion speed is too low, the fiber will be working against the printing head, either stalling it and ruining the print, or it will cut and torn by the edge of the nozzle. This phenomenon is seen on the right side of the figure.

Often times when printing sharp corners, the fiber bundle would curl up. This happens because the thermoplastic jacket surrounding the fibers is in a molten state just after extrusion and cannot resist the tension in the fibers. This problem was solved by pausing at each sharp corner for 500 ms.

This pause gives the thermoplastic enough time to harden and keep the fiber bundle in its place.

After each complete print of a test specimen, the fiber filament still connecting the print to the print head was cut using wire cutters. This process should be automated in further work to allow printing different parts or perimeters continuously without manual intervention.

Considering these parameters, specimens for tensile tests were printed. Because of the current software limitations, the specimens printed were not designed after any norm or standard. They are rectangular, 100 mm long, 10 mm wide and1.2 mm thick with one layer, or 2.4 thick with 2 layers.

Results and Discussion

In order to assess the quality and the properties of the filament produced with the proposed method, multiple tests and analyses were conducted.

Filament composition analysis

Light microscopy is helpful to analyze the composition and interior distribution of the fibers in the filament. For this reason, transparent PLA was used as filler for the filament. Both optical and digital microscopes were used to study the filament and its composition, but also to determine if the fiber bundle was correctly treated with the ABS solution.

The first inspection was of the produced carbon fiber with ABS infusion and coating. A micrograph of the outer surface and longitudinal cut through the fiber can be seen in Figure 5. The outer surface was successfully coated, covering all fibers in the bundle. The longitudinal cut shows that the treatment was also successful in penetrating the fiber bundle. This can be identified by the white deposits between the individual carbon fibers.

A micrograph (Figure 6) of a diagonal section cut of the final PLA filament with pre-treated carbon fiber reinforcement shows that the fiber bundle was successfully penetrated by the ABS solution. Furthermore, it shows that the treatment of the fibers is not affected by the extrusion process. Another factor that might play a role in the cohesion of the fibers is the amount of thermoplastic between the fibers, and thus, the fiber density in the bundle. In this case, the distance between the fibers, that was filled by the ABS solution ranged from 5 to 10 μm. This density is most probably related to the ABS concentration of the used solution and the compression of the bundle as it travels through the funnel after the solution bath.

A frequent problem with continuous fiber reinforcement is the formation of voids between the fibers. The formation of voids between fibers was also identified in the printed test specimens, as seen in Figure 7. These voids could by cause the expansion of the thermoplastic during printing, by the evaporation of residual moisture, or by unequal tensions in the fiber bundle during printing.

Another possibility is that they have been created by the cross section cut during preparation of the microscope specimen. This possibility is taken into consideration, because of the high difference in void density compared to the bundle before printing (Figure 6).

Examining multiple cross sections of the filament (Figure 8) revealed that the geometry and positioning of the fiber bundle are not constant throughout the filament. Furthermore, it is visible that the thermoplastic sheath tends to follow the geometry of the fiber bundle. A possible solution to improve the roundness of the filament is to increase the extrusion pressure.

By measuring area of the fibers in the cross section, the fiber content could be determined. In this case, the fiber bundle accounted for 22% to 28% of the filaments area. The fluctuation is mainly caused by inconstant filament diameter but may also be caused by voids. Voids content could not be accurately measured using this method.

Mechanical properties – Tensile testing

In order to assess the change in performance, tensile tests of both unprinted filament and printed test specimens were conducted. To compare the results of the proposed production method, filaments with both untreated and pre-treated carbon fiber were produced. For the untreated fiber filament, the coating step was skipped, and raw fiber tow was coextruded with thermoplastic, like the treated fiber bundle.

The first tensile testing was of the unprinted filaments. For this purpose, multiple 100mm long pieces of commercial PLA, WoodK’s 1.5k raw CF/PLA, 12k raw CF/PLA and pretreated 12k CF/PLA were tested. The results can be seen in Figure 9.

As expected, all filaments containing fiber reinforcement outperformed the plain PLA filament.

Both composite filaments with raw, untreated fibers had similar behaviors: the fibers started slipping and pulling out before the total failure, as seen in the stress-strain diagram. That was not the case for the pretreated, coated fiber, which also increased both the tensile modulus and the yield stress of the filament.

The results of the tensile test can be seen in Table 1. The additions of 12k carbon fibers in the PLA matrix lead to the increase of the elastic modulus by 461% and of the yield stress by 274%. The pretreatment of the fibers further increased the elasticity modulus by 12.8% and the yield stress by 70%.

Table 1: Mechanical properties of filaments.

This results in an increase of the elastic modulus of 532% and of the yield stress of 535%, from plain PLA filament to treated CFR filament.

The second tensile test was of specimens printed with the two produced CFR filaments: with raw CF and with treated CF.

The results of the stress-strain test of the printed specimens can be seen in Figure 10 and in Table 2. Compared to the unprinted filament, all test specimens showed lower yield stress but higher elongation. This effect may be caused by the printing process. It is possible that the fiber bundle is affected by the sharp turn experienced when it is laid onto the build surface or may also be caused by the extruder gear or hot end nozzle.

Table 2: Mechanical properties of printed specimens.

An increase from the untreated, raw fiber filament to the treated fiber filament could still be seen, yet not as high as with the unprinted filaments. The yield stress increased for one-layer specimen by 48.3% and for the two layered one by 40.2%.

The pre-treatment of the fibers still was a significant improvement, as one layered pre-treated the two layered specimens with double the amount of untreated fibers. This shows that the pre-treatment of the fibers is a great method of improving strength and of reducing the necessary fiber amount.

A further phenomenon could be observed with the printed test specimens. The fiber bundle was consistently located at the top of the printed layer, with minimal PLA covering the fibers, as depicted in Figure 11.

Because of this, the thin PLA layer over the fibers failed first. Because of this, the top of the fiber bundle was no longer encased and was not contributing to the strength of the part as much. This could also be seen in the tested specimens, as seen in Figure 12. The bottom of the specimen was not as damaged as the top side of the specimen. Even though the PLA failed on both sides, the fibers on the top were severely affected, and damaged, compared to the still intact lower layers of the fiber bundle.

Most probably, because of this effect, the two layered test specimens did not have double the tensile strength compared to the one layered one. The bottom layer was topped by the PLA of the second layer, protecting and encasing the fibers of that layer, but the top one was still exposed.

The layer height also has an impact on the strength of the fiber reinforcement. A smaller layer height leads to the compression and flattening of the fiber bundle, which increases the area of contact surface between fibers and thermoplastic, thus increasing adhesion and probably its mechanical properties. This is similar to the tests conducted with flat fiber bundles but only after printing, therefore avoiding the increased stiffness of the filament.

Alternative fiber geometry experiment

An alternative concept of the carbon reinforced filament composition was also tested, with the goal of increasing the adhesion between the fibers and the thermoplastic. As discovered from previous tests with the fibers bundled in the center of the filament, only the outer layers of the fiber bundle are bonded to the thermoplastic. The interior of the fiber bundle is consolidated by the ABS infusion, but the adhesion in not as strong as that between fibers and the extruder thermoplastic. This can be clearly seen in the broken tensile test specimens: Some of the fibers remain bonded to the thermoplastic and broke together, but the center of the bundle did not break, and was pulled out, even when infused with the ABS solution. If all the fibers in the bundle would be bonded to the thermoplastic, the mechanical properties would increase further.

In order to achieve that, the contact surface between the fiber bundle and the thermoplastic would have to be increased. Ideally, the fibers would not be bundled, but uniformly distributed in the filament, as seen in the following illustration on the right side. The left drawing shows the current case: the fibers are bundled and centered (Figure 13).

As there is no current method to achieve the uniform distribution, another method is to flatten the bundle, thus exposing more surfaces to the thermoplastic. The first method tested was to remove the fiber entrance nozzle of the filament extruder, as that would be bundling the flat fiber. The carbon fiber tow already comes as a flat, so that was directly fed into the extruder. That resulted in a filament similar to the one described by the following illustration, on the left side. The problem was that the fiber was separating the thermoplastic into two halves, as the plastic-plastic contact was limited to two small areas (marked with red circles). This may lead in some cases to the filament splitting along the fibers, as there is lower adhesion between the fibers. A better shape of the fiber bundle would be something similar to the right one, as it would still leave enough contact for the thermoplastic (Figure 14).

The actual results were a mixture of the two described distributions, as seen in the next micrograph of a cross section of the filament.

The used flat fiber bundle was wider than the filament. Because of that, it was forced together in two layers during extrusion. The micrographs (Figure 15) can also be used to identify defects in the filament. The left micrograph was taken using incident light, from above the specimen. This revealed two air bubbles in the middle of the filament. The right one was taken using transmitted light from below. Some voids large were also formed in the filament, marked by the red contours.

This method of extrusion increases the contact surface between fiber and thermoplastic, but it also leads to another problem. The extruded filament is much stiffer than the previous one, where the fiber was bundled in the center. The stiffness was high enough to cause the filament to break when spooled on a standard commercial 1kg 1.75mm filament spool. Because of this reason and of the voids created, this method, of flattening the fibers, was considered infeasible and was not further investigated. The uniform distribution of the fibers within the filament (Figure 13) would probably also cause the same behavior. In such a case, the composite would have to be produces in the form of rods instead of spools, to avoid fiber break.

Conclusion

This work focused on developing a reliable method of producing continuous fiber reinforced filaments with a standardized diameter. Furthermore, the process was designed to improve existing methods by pre-treating the fibers with a polymer solution. For this purpose, a prototype apparatus for the production of such filament was successfully build and tested. Filament with both treated and raw fibers was successfully produced and could be used to compare its mechanical properties and composition to other existing filaments.

Key parameters for the FFF printing of such filaments were identified and described. Respecting these parameters, test specimens could be printed and used for tensile testing. The results showed an increase in yield stress of 535% compared to plain PLA filament. The results of the printed specimens were lower than expected, showing an averaged decrease of strength of 55% compared to the unprinted filaments. The causes for this decrease could only be speculated and need further investigation.

Along with that, further work is needed to optimize the production process and the printing process of fiber reinforced filaments. For the fiber treatment process, the concentration of the solution used, most probably has an impact on the mechanical properties of the filament. As discussed, the concentration of the solution, but also the squeezing amount of the excess and their effects have not yet been investigated. By varying these factors and testing their effects, an ideal concentration and resulting fiber density within the bundle could be identified.

The used fiber tow is not raw carbon fiber, as the fibers are coated with a sizing agent to improve adhesion. As this sizing is a trade secret of each producer, it could not be established what its effect are and if it also aided in the adhesion to the PLA and ABS. Most commonly, fiber tow is used with epoxy resin, so the used sizing is probably optimized to be compatible with commonly used epoxy resin. The effects of sizing agents on the adhesion and their compatibility with thermoplastics should also be explored.

Other properties, including compressive and flexural strength, impact resistance and creep should also be studied to fully investigate the effects of each parameter on the final product.

Further work is also needed to optimize the printing process. Firstly, a cutting mechanism for the fibers is needed. This will allow the printing process to be automated. Secondly, the design and geometry of the nozzle may have a significant impact on the strength of the fibers. The nozzle should not have any sharp edges, as it may damage the fibers and should also have a large, flat surface to aid in the spreading and compression of the fiber bundle and the thermoplastic.

References

1. Mark Two Desktop Carbon Fiber 3D Printer for Manufacturing (2019) Markforged.
2. Anisoprint (2019) Stop metal thinking - start anisoprinting.
3. WoodK plus (2018) KompetenzzentrumHolz GmbH - Annual Report.
4. Kabir SMF, Mathur K, Seyam AFM (2019) A critical review on 3D printed continuous fiber- reinforced composites: history, mechanism, materials and properties. Compos Struct 232: 11476.
5. Naranjo-Lozada J, Ahuett-Garza H, Orta-Castañón P, Verbeeten WMH, Sáiz-González D (2019) Tensile properties and failure behavior of chopped and continuous carbon fiber composites produced by additive manufacturing. AdditManuf 26: 227–241.
6. Ivey M, MelenkaGw, Carey JP, Ayranci C (2017) Characterizing short-fiber-reinforced composites produced using additive manufacturing. Adv ManufPolym Compos Sci 3.
7. Araya-Calvo M, López-Gómez I, Chamberlain-Simon I, León-Salazar JL, Guillén-Girón T, et al. (2018) Evaluation of compressive and flexural properties of continuous fiber fabrication additive manufacturing technology. AdditManuf 22: 157–164.
8. Shama Rao N, Simha TGA, Rao KP, Ravi kumar GVV (2018) Carbon Composites Are Becoming Competitive And Cost Effective. Infosys Sustainability Report.
9. Mori K, Maeno T, Nakagawa Y (2014) Dieless Forming of Carbon Fibre Reinforced Plastic Parts Using 3D Printer. Procedia Eng 81: 1595–1600.
10. Li  N, Li Y, Liu S (2016) Rapid prototyping of continuous carbon fiber reinforced polylactic acid composites by 3D printing. J Mater Process Technol 238: 218– 225.
11. Liu S, Li Y, Li N (2018) A novel free-hanging 3D printing method for continuous carbon fiber reinforced thermoplastic lattice truss core structures. Mater Des 137: 235–244.
12. Bucher CM, Aburaia M, Markl E, Engelhardt NC, Gonzalez GJ, et al.(2018) Continuous Fibre Reinforced Freeform 3D Printing. Add IT Additive Manufacturing Symposium Steyr.
13. Le Duigou A, Barbé A, Guillou E, Castro M (2019) 3D printing of continuous flax fibre reinforced biocomposites for structural applications. Mater Des 180: 107884.
14. Ming Y, Duan Y, Wang Y, Xiao H, Zhang X (2019) A Novel Route to Fabricate High-Performance 3D Printed Continuous Fiber-Reinforced Thermosetting Polymer Composites. Materials 12.
15. Oztan CY, Karkkainen R, Fittipaldi M, Nygren G, Roberson L, et al. (2019) Microstructure and mechanical properties of three dimensional-printed continuous fiber composites. J Compos Mater 53: 271–280.
16. Shan-ShanY, Fan-Long Jin, Kyong YR, David H, Soo-Jin P (2018) Recent advances in carbon-fiber-reinforced thermoplastic composites: A review. Compos 142: 241–250.
17. Goh GD, Yap YL, Agarwala S, Yeong WY (2019) Recent Progress in Additive Manufacturing of Fiber Reinforced Polymer Composite. Advanced Materials Technologies 4.
18. Eichenhofer M, Wong JCH, Ermanni P (2018) Exploiting cyclic softening in continuous lattice fabrication for the additive manufacturing of high performance fibre-reinforced thermoplastic composite materials. Compos Sci Technol 164: 248–259.
19. Vijayakumar D (2016) Manufacturing Carbon Nanotube Yarn Reinforced Composite Parts by 3D Printing. University of Cincinnati.