Towards Self-Reinforced PLA Composites for Fused Filament Fabrication

Abstract

Aligned with the Sustainability Development Goals (SDGs), we present the complete methodology of preparing bio-based polymer filaments to be used in additive manufacturing, specifically by means of so-called Fused Filament Fabrication (FFF) in 3D printing. Filament production and 3D printing were both developed and optimised in this work. First, we focused on the steps of producing and optimising the extrusion process of unreinforced polylactic acid (PLA) composite filaments. Second, we studied the resulting material properties by discussing the selection of a specimen geometry and the international standards adequate for FFF 3D printing. Moreover, we investigated the process parameters in order to achieve reliable structures. Based on the reinforcement material (stereocomplex fibres (Sc-PLA fibre) and bi-component fibres (bi-co PLA fibre), base-matrices were selected for producing un-reinforced filaments. In this way, we present the complete preparation approach by identifying problems and pitfalls for fostering studies of bio-based polymer filaments.

1. Introduction

In the beginning of the 20th century, efforts to replicate natural polymeric materials such as natural rubber, silk and similar, were mainly achieved by trial and error, with lower qualities than their natural counterparts. Since the introduction of the concept of macro-molecules in “Über Polymerisation” (“On Polymerisation”) in 1920 [1], Hermann Staudinger gave polymer chemistry its theoretical foundations, achieving the design of high-value polymers comparable to the properties of natural materials. This achievement marked the beginning of a new era in tailor-made molecular designs of structural and functional polymeric materials, including synthetic polymer materials [2], fibres, rubbers, adhesives, floorings, etc., but also a boom in its practical application in almost all sectors. The rise of synthetic polymers is owed to a variety of reasons, including the low cost-to-performance ratio, low energy consuming production and processing techniques, as well as the vast variety of chemical and physical properties, such as elasticity (elastomers), impermeability, and electrical conductivity, to name a few [3,4]. Driven by the focus on economical optimisation, the polymer community has been working towards producing plastics with reduced costs, increased strength, versatility, and durability [5]. Despite their advantages, many polymers are still based on finite fossil fuel resources [3,6], microplastics are now present in almost all organisms [2,7], and marine litter and microplastics are accumulating in landfills or the natural environment [2]. In order to deal with the negative repercussions of the petroleum-based polymers, potential concepts, such as bioplastics and, more recently, the United Nation’s Sustainable Development Goals (SDGs), have increasingly motivated the industry to utilize different materials [7]. Bioplastics are more environmentally compatible in comparison to petroleum and coal-derived synthetic polymers owing to the biodegradability, eco-friendly manufacturing processes and their wide range of applications [8] as they are derived from renewable resources such as starch, gelatin [9], cellulose [10], chitin [11], and lignin [12], thus reducing our dependency on petroleum resources [13].

Polylactic acid is a bio-based polymer synthesised from starch of corn, maize, potato, and cassava root. Figure 1 shows the repeating unit so-called monomer known as lactic acid, which is obtained by fermenting sugar or starch. Figure 2 shows the two optically active stereoisomers compounds that lactic acids exist in: D−Lactide, L−lactide, and a third form where the compound exists in a combination of one L−lactyl and one D−lactyl unit [14]. Depending on the mixture of the stereoisomers, a range of different PLA may be synthesised with varying viscosities, crystallinity, and melt temperatures.

Bioplastic is an umbrella term that consists of (a) bio-based plastic from renewable resources and/or (b) biodegradable plastic. A bioplastic may be based on a renewable resource but may not be biodegradable, or a biodegradable polymer may not be based on a bio-based resource. Here, it is important to note that the source of the carbon is either bio-based or petroleum-based. The trait of biodegradability is a result of the chemical structure of a bioplastic. PLA, owing to it being an agriculture-based resource, is highlighted as a carbon-neutral bioplastic [14]. Moreover, PLA is a thermoplastic biodegradable polyester that offers high functionality (flame-retardant, bacteriostatic, weather resistant) through intrinsic properties [15].

Recent engineering developments in additive manufacturing (AM) have not only actualised numerous advantages such as improved automation, independent production of spare/main parts [16,17,18], but have also enabled the production of multi-scale materials [19,20].

The raw material used for the process of 3D printing is based on the type of the employed 3D printing process. Here, we discuss Fused Filament Fabrication (FFF), which is an extrusion-based process, where a thermoplastic polymer in the form of a filament is melted in a hot-end and pushed through a nozzle (orifice). In this way, it is laid onto a building platform in single lines. Multiple of these singles lines at one height make up a layer [21]. Most of the time, multiple layers then add up to a final structure [22].

The most common issue with conventional composite material is recycling, owing to the heterogeneous hybrid structure of the matrix and the reinforcement. The major barriers for the commercialisation of composite recycling include high recycling costs relative to the raw material costs. Yet, there is also a lack of adequate research. Therefore, for promoting composite recycling, there is a need for extensive research on the design of materials, the manufacturing process and end-of-life cycle scenarios for easier recycling of composite materials [23,24]. Similarly, other natural fillers such as wood sawdust [25], walnut shell [26], jute, flax and hemp [27], cotton [28], kenaf fibres [29], rice straw [30], lignin [31], and cellulose [31] are preferred choices because of their potential mechanical properties. There is limited research ongoing where studies investigate PLA fibres incorporated in a PLA matrix. The recent studies carried out on the reinforcement used for the current study are discussed in Section 2.1 and Section 2.2.

Figure 3 shows the main targeted research areas of the underlying work. Methods were developed for incorporating the bio-based polymer composite in FFF 3D printing and achieving a reproducible 3D printed test specimen for experimental characterisation. For that, the filament was first produced and optimised to an adequate quality such as being bubble-free, uncontaminated, and having an even oval cross-section. Then, the filament was used in a commercial FFF 3D printer for printing the test specimens. We demonstrate the challenges faced during the polymer filament production and the employed optimisation approaches, then strategically elaborate on the reasons for the chosen geometry based on the standards for the FFF printing, and the work-flow preparation needed for the production of a specimen. The materials used for the unreinforced filament production are explained in Section 2.

2. Materials

Herein, we used composite materials as bio-plastic materials, consisting of (i) the matrix and (ii) a filler/reinforcement phase. The matrix material surrounded the filler phase. Based on the individual material properties of the fibre reinforcements used in the study, two different base matrix materials were utilised as mentioned in

Table 1. Materials used in the study. The base matrices and the corresponding reinforcement for the bio-based polymer composite production. Both matrix materials are procured from Total Corbion [32] and NatureWorks [33], respectively.

	Matrix	Reinforcement	Composite
Matrix Material A	PLA Luminy L130	Stereocomplex PLA	Unreinforced
Matrix Material B	PLA Ingeo 6302D	Bi-component PLA	Unreinforced

.

Since PLA is a hygroscopic material, it is recommended to pre-dry (dehumidify) the material before processing. This step not only prevents unwanted foaming during the filament processing but also prevents possible changes of chain structure because of the chemical reactions in the presence of water [34]. The two fibre types—stereocomplex (Sc) and bi-component (bi-co) fibres—used in the study are hereafter referred to as material types A and type B, respectively. The base matrix material for type A and type B were pre-dried at 60 °C and 40 °C, respectively, in a vacuum oven.

PLA is a crystalline thermoplastic polymer capable of being spun into numerous fibre types such as monofilaments, multifilaments, bulked continuous filaments, staple fibres, short-fibres and spun bound fabrics [35]. However, there are still difficulties in using PLA to its full potential in a variety of applications, which is constrained by sub-optimal material properties such as weak thermal and hydrolytic stability [36].

Figure 2 shows the two optically active states of the monomer (lactic acid). The ratio of L− and D−lactic acid in the PLA polymer determines the crystallinity resulting in an increase in the mechanical, thermal and electrical properties of the polymer. Stereocomplex crystallisation occurs when the enantiomeric pair of PLA, i.e., L−lactide (PLLA) and D−Lactide (PDLA) are taken in racemic mixture (1:1). As a result of multicentre hydrogen bonding interactions, an alternative arrangement of helical chains is formed between PLLA and PDLA units possessing an opposite chiral formation [36].

The stereocomplex method of synthesising highly ordered structures was reported as early as the 1980s, stressing several challenges [37]. The conventional method of synthesising Sc-PLA involves utilising high temperatures of 220 °C to 260 °C, usually yielding low Sc content (low molecular weight) [38,39,40]. Many studies have been conducted to enhance the formation of high molecular weight polymers within a PLLA/PDLA blend [41]. However, for industrial applications, it is more relevant that the enhancement offers advantages of scalability and the low cost of raw materials as compared to molecular level engineering [36]. PLA-Sc’s wide application is limited as there are still some challenges that need to be resolved:

-

One major obstacle is the high cost of D−lactide (PDLA). As stated before, the crystal formation in the Sc-PLA decreases with an increase of PDLA homo crystal. Therefore, for industries, the development of high-performance PLA-Sc with low PDLA content is beneficial;

-

The spinning of the stereocomplex fibres needs further research, especially the parameters (pre-treatment, dyeing, clearing, post-processing/ finishing treatment) that impart the physical and chemical effects of the Sc-PLA fibre [35,36];

-

So far, the literature is focused on the controlled structure and properties of Sc-PLA blends, but not enough studies have been conducted to understand the biodegradation mechanism and the kinetics [36];

-

The need to develop engineering solutions for producing on-demand optimised crystal morphology of PLA-Scs. We refer to [42] for an application demonstrating degradable bio-electronics to be used in health monitoring.

Ample studies are available in the literature on PLA reinforcement in a PLA base-matrix. Using non-wovens made out of PLA bi-component fibre (ratio of PLLA65:PDLA35), in [43], a single polymer fibre composite was produced and characterised by means of its mechanical properties. A self-reinforced composite was produced in [44] based on high stiffness PLA yarns. Single Polymer Composites (SPC) made out of PLA were prepared in [45], where the polymer composites were fabricated by laminating two amorphous PLA sheets and a layer of crystalline PLA fibre, followed by compression moulding to form the SPC. The preparation and properties of self-reinforced PLA composites were investigated in [46] based on oriented tapes. A self-reinforced composite (SRC) and PLA reinforced PBS (Poly Butylene Succinate) composite via a film-stacking method was demonstrated in [47].

2.1. Matrix Material A and Stereocomplex PLA Fibre

Table 2. Material properties of the stereocomplex fibres.

Reinforcement: Stereocomplex Fibre	Properties
Uni-axial tensile Modulus	6 ± 1 GPa
Melting temperature	220 °C
Fibre fineness	2 dtex
Strength	35 cN/tex
Elongation at break	20%

shows the reinforcement properties of Sc-PLA fibres.

Table 3. Material properties of the type A base-matrix material.

Base-Matrix: PLA Luminy L130	Properties
State	Crystalline polymer
Melting temperature	175 °C
Glass transition temperature	60 °C
Uni-axial tensile modulus	3.5 GPa
Tensile strength	50 MPa

shows the properties of the base-matrix used for the Sc-PLA fibre reinforcement.

2.2. Matrix Material B and Bi-Component PLA Fibre

The bi-component PLA fibre consists of an inner-core and a mantel or outer sheath (surrounding the inner-core). In this study, a semi-crystalline PLA formed the high strength inner-core component of the fibre by means of a core fraction 65% crystallinity and an amorphous low-melting PLA comprised the outer-sheath.

Table 4. Material properties of the bicomponent fibre (inner-core and the outer-sheath matrix).

Reinforcement: Bicomponent Fibre	Properties
Uni-axial tensile Modulus	3.5 GPa
Inner-core	PLA Ingeo 6100D
State	Semi-crystalline polymer
Glass transition temperature	55–60 °C
Melting temperature	165–180 °C
Outer-sheath matrix	PLA Ingeo 6302D
State	Amorphous polymer
Glass transition temperature	55–60 °C
Melting temperature	125–135 °C

shows the reinforcement properties of bi-co fibres.

Table 5. Material properties of the type B base-matrix material.

Base-Matrix: PLA Ingeo 6302D	Properties
State	Semi-crystalline polymer
Melting temperature	125–135 °C
Glass transition temperature	55–60 °C
Elongation at break	50%

shows the properties of the base-matrix used for the bi-co fibre reinforcement.

Melting temperatures differ in the semi-crystalline inner and amorphous outer cores of the bi-co fibres. This difference allows the matrix material to adhere homogeneously to the fibre and the filament, while the highly oriented semi-crystalline PLA core in the fibres remain intact.

When the bi-component fibres are pressed, the low-melting sheath component forms the matrix, while the high-strength core component, a semi-crystalline PLA, remains as a reinforcing fibre.

Composite rule of mixture: To predict the material properties of the fibre reinforced composite, such as the effective uni-axial tensile Young’s modulus, the rule of mixture is used for composite materials [48].

The PLA composites used in the current study are visualised in Figure 4 with two different reinforcements. The effective uni-axial tensile modulus is given by

$$E_{\mathrm{specimen}}=E_{\mathrm{m}}(1+f_{\mathrm{f}}(E_{\mathrm{f}}/E_{\mathrm{m}}-1)),$$

(1)

where $E_{\mathrm{m}}$ is the uni-axial tensile modulus of the matrix, $E_{\mathrm{f}}$ is the uni-axial tensile modulus of the reinforcement and $f_{\mathrm{f}}$ is the fibre fraction. The diameters of the Sc-PLA and bi-co PLA fibre are 0.03 mm and 0.08 mm. For the 1.75 mm filament, the fibre fraction stands at 1.7% and 4.5%. The effective uni-axial tensile modulus of the Sc-PLA composite filament is predicted to have a maximum and minimum of 3560 MPa and 3525.5 MPa, respectively. The effective uni-axial tensile modulus of the bi-co PLA composite filament is predicted to be 3500 MPa. According to the theory of highest possible packing density, the tensile modulus was calculated with a max. of about 30–40 monofil fibres for lower and upper bound values. For the same standard size of monofil diameter 0.03/0.08 mm, the fibre fraction stands at 51% for the number of monofils, n = 30. In the case of different monofil diameters, the fibre fraction stands at 68% for the number of monofils, n = 40, which is still less than the highest possible packing density [49,50]. Theoretically, the effective uni-axial tensile modulus at a fibre fraction of 51% for the Sc-PLA filament is predicted to be 4265 MPa and 5285 MPa, according to the uni-axial tensile modulus lower and upper values, respectively. The effective uni-axial tensile modulus for the bi-co PLA filament, at the fibre fraction of 51%, is predicted to be 3500 MPa.

3. Methods

3.1. Processing and Optimising the PLA Composite Filament for the FFF 3D Printer

Herein, the fabrication of matrix material A and matrix material B from pellets to filament is described. The matrix material A was PLA Luminy L130, which was selected to process the Sc-PLA fibre. The matrix material B was PLA Ingeo 6302D, corresponding to the bi-co PLA fibre.

Production set-up: the work-flow of the filament production was established on an assembly of units as shown in Figure 5 and the schematic figure of the extrusion line for the production of the filament is shown in Figure 6. The heart of the set-up was the polymer extruder machine from Axon AB plastic machinery [51]. The matrix pellets were fed into the hopper (station 1), from where the pellets moved into the screw (station 3), changing the solid to viscoelastic. For the production of the future continuous fibre reinforced filament, we used the self-fabricated nozzle suiting the production capacity of the extruder as shown in Figure 7 (station 2). The molten polymer exited the nozzle. From the nozzle, the filament was drawn into a water-bath path (station 4), providing uniform cooling. The solidified filament passed to the conveyor belt (station 5), accessorised with a roller on top. For controlling, the filament passed a laser diameter controller (station 6) and wound up on the spool-winding station (station 7). The matrix A pellets PLA Luminy L130 were pre-dried in a vacuum oven at 60 °C. Luminy L130 has a ${\mathrm{T}}_{\mathrm{g}}$ of 60 °C and a ${\mathrm{T}}_{\mathrm{m}}$ of 175 °C.

Table 6. Extrusion temperature profile for the extruder-screw. The entire screw is temperature controlled in each segment/zone separately. The matrix material A, PLA Luminy L130. Luminy L130 has ${\mathrm{T}}_{\mathrm{g}}$ 60 °C and ${\mathrm{T}}_{\mathrm{m}}$ 175 °C.

Extrusion Zone	Temperature
1	130 °C
2	175 °C
3	176 °C
4	170 °C
5	170 °C
6	180 °C

presents the processing temperatures T at the extrusion machine for matrix material A, Luminy L130.

The matrix B pellets PLA Ingeo 6302D were pre-dried at 40 °C. Ingeo 6302D has a ${\mathrm{T}}_{\mathrm{g}}$ between 55 °C and 65 °C and a ${\mathrm{T}}_{\mathrm{m}}$ between 125 °C and 135 °C. Ingeo 6302D was fabricated at the following processing temperatures at the extrusion machine. The optimisation of the filament during the extrusion production is explained in detail in Section 3.2

3.2. Optimisation of the Filament

Matrix material A and matrix material B were processed via the polymer extrusion process. In order to avoid the swelling and change of structure in the filament [52], the base-matrix PLA pellets were pre-dried in a vacuum oven at 60 °C and at 40 °C for matrix material A (PLA Luminy L130) and matrix material B (Ingeo 6302D), respectively.

If not pre-dried properly, mostly there is a change in filament structure. Because of hydrolysis, there is chain thinning process during extrusion [53]. To counterpart this issue, the screw was degassed through the back of the screw when the resin was in the compression zone [54]. Figure 8 and Figure 9 show the microscopic images of the optimised filament and the defects such as unmelted polymer and voids for the base-matrix materials A and B used in the study. Figure 10 shows the challenge of unmelted polymers is caused by the contamination with other polymers or processing lower than the material’s melt temperature, creating agglomerations. To make sure the polymer pellets are melting correctly, the temperature of the zone-heaters must be increased, resulting in melting of the pellets. Another approach to give more melt time to the pellets in the barrel is to reduce the screw speed. For amorphous polymers, the reduction of the processing temperature of zone-heaters results in a more homogeneous melt.

When the stream of the hot melted polymer is forced through a nozzle (die, orifice), the extruded material diameter is usually greater than the channel size. This is known as extrudate-swell or the Barus effect [55]. As the melted macromolecules leave the nozzle, their diameter increase is two or three times the capillary diameter. The ratio of the extrudate diameter to the diameter of the capillary is called a swelling ratio. It is dependent on the extrusion rate and the viscoelastic properties of the fluid, along with the geometry of the extrusion system [56].

As the material leaves the nozzle, the shape might be oval owing to the temperature gradient within the composite material. The difference in temperature generates a drop in forces to a higher velocity, as the cold entrance makes the fluid velocity zero. For hot liquid, the viscosity is low, meaning that the flow is at a higher speed at the centre than at the wall.

For controlling the die-swell phenomenon, several optimisation strategies were employed by controlling the rotation speed (RPM), temperature and flow rate, and winder speed. For material A and material B, the properties are presented in Table 6 and

Table 7. Extrusion temperature profile for the extruder-screw. The entire screw is temperature controlled in each segment/zone separately. The matrix material B, PLA Ingeo 6302 D. Ingeo 6302D has ${\mathrm{T}}_{\mathrm{g}}$ between 55 °C and 65 °C and ${\mathrm{T}}_{\mathrm{m}}$ between 125 °C and 135 °C.

Extrusion Zone	Temperature
1	115 °C
2	130 °C
3	133 °C
4	135 °C
5	136 °C
6	140 °C

[57].

The melted polymer exits the nozzle-die to enter the cooling-bath as shown in Figure 5. The high temperature melt when cooled may cause cavitation in the filament because of polymer shrinkage.

During the process of cooling, at a certain stage the molten material inside the thick area is unable to pull the solidified outer-surface. As a result, a void Figure 8 and Figure 9 (top) is created as the shrinkage causes the molten material on the inside to shrink towards the outer surface.

In order to counterpart this challenge, the filament needs to be cooled uniformly at a lower rate, thus allowing the shrinkage to happen from the outside to the centre [58]. When the material exits the nozzle, it starts to cool down as the atmospheric temperature is lower than the extruder temperature. Here, it is important to set up a cooling system for regulating temperature of the extruded filament. As hot filaments are soft and thus deformable, the puller mechanism easily flattens the final shape, thus yielding the challenge of an oval shape instead of round filaments. The cooling system consists of a path circulated with a water flow. The filaments, after passing through the cooling bath, lead to the conveyor belt station. Another reason for oval-shaped filaments is often the high polymer crystallinity leading to flat, distorted or oval filaments upon leaving the nozzle. In this study, we used semi-crystalline polymers, and they are known to have high domains of crystals in comparison to amorphous regions, which normally exhibit deformation problems in the form of oval or flat filaments. This formation is a process-related phenomenon.

The non-uniform polymer shrinkage causes the filament to become oval or flat. Non-uniform shrinkage is influenced by the molecular orientation, reinforcing fibres and the fibre direction, uniformity in cooling, and melt temperature uniformity. For counteracting the oval or flat or distorted filament shape, the following approaches are employed:

• Cleaning the extrusion channel and nozzle of contaminants—a clean channel helps the molecular orientation to be kept in order, thus reducing the oval shaping effect;
• Reducing the filament output speed—to decrease the flow speed difference, the RPM (Rotation Per Minute) of the extruder is decreased, thus reducing the potential difference in molecular orientation and thus the shrinkage;
• Setting up a more uniform cooling—a possible external improvement is providing a perfectly uniform cooling as the filament has a tendency to form an oval or distorted shape as a result of the temperature gradient on the surface.

Another difficulty of filament production is the filament thickness deviation. The deviation in the thickness is influenced by various factors such as:

• External factors;
• Temperature settings;
• Extruder RPM;
• Polymer degradation;
• Material contamination;
• Non-uniform cooling;
• Winder mechanism; and
• Holes or bubbles in the filament.

For better control over the filament deviation, we report that the right balance between the extruder RPM, the winding speed, and a uniform cooling setup is important.

3.3. Geometry of Tensile Test Specimens

For determining the mechanical properties of thermoplastic polymers, there exist standards for polymers but they are not specifically designed for 3D printing. They even date back before 3D printing such that their applicability is possible but not guaranteed. Therefore, it is important to select and interpret the recommendations adequately for the intended research and use case. In this paper, the main priority was to 3D print the tensile specimens to determine the tensile properties of the material(s), similar to in [59]. The different test standards for conventional specimen production and their application to additive manufacturing are compared in [60]. From the various norms recommended in [60], the ASTM D3039 was chosen as the testing standard with the following justifications:

• The ASTM D3039 is specified for polymer matrix composite materials, which applies to self-reinforced PLA as well;
• Although not specified in the ASTM D3039 itself, the standard has been used in various studies for unreinforced FFF 3D printed tensile specimens [61,62];
• Since the specimen geometry of the ASTM is a simple rectangular geometry, the slicer fine-tuning compared to dog bone geometries is less extensive. For dog bone geometries, the area of the curvature is critical. Abrupt endings of the layer paths in the area of the curvature can cause premature failure due to shear stress concentrations, we refer to [61]. This problem is eliminated with a rectangular specimen geometry;
• The rectangular specimen geometry is also well-suited for printing continuous fibre reinforced specimens.

The ASTM D3039 only provides concrete information on the form tolerance of the specimen geometry, but not on the dimensions. The ASTM describes that different studies have found different width to thickness ratios to work, but ultimately this ratio needs to be adapted according to the individual experiments (we refer to Section 8.2.2 in the ASTM D3039).

The first step was to determine the cross-sectional area of the specimen regarding the tensile machine’s maximum force. Here, the machine used for tensile testing was the MTS Tytron 250 (MTS Systems Corporation, Eden Prairie, USA), which is capable of applying a maximum force of 250 N. To be able to test until failure with an adequate accuracy, the maximum force used for sizing the specimens F was set to 200 N. In the end, two geometries were established; one for testing the matrix material and one for testing the composite material where the load capacity was expected to be higher. For the unreinforced specimens (printed only with matrix material), the cross-section A was set to 4 mm². This cross-section allows (in combination with the previously mentioned testing machine) for the following maximum stress:

$$\sigma =\frac{F}{A}=\frac{200\mathrm{N}}{4{\mathrm{mm}}^2}=50\mathrm{MPa}.$$

(2)

As a preliminary estimation, the tensile cross-section for the reinforced, composite specimens was set to 2 mm² to compensate for the probable increase in load capacity with fibres.

With the cross-section being set, the width w and thickness t of the specimens needed to be sized. Due to the rectangular geometry of the specimen and the clamping devices at the chair, friction clamps were employed. Therefore, it is crucial to have a sufficient gripping surface, for which the width of the specimen had to be quite high. From previously conducted experiments on commercially available PLA filament, a thickness to width ratio of 1:16 was chosen. The final geometry is shown in Figure 11.

3.4. Methods of 3D Printing

The geometry defined in Figure 11 has been modelled in FreeCAD and exported into a STereoLithography (STL) file for slicing. Slicing is a process in 3D printing where the Computer-Aided Design (CAD) model of the final part is processed into machine instructions known as G-Code, which determine the printer head’s (thus nozzle’s) movement and operation settings, such as temperature and feeding speed.

We use the open-source slicing software Ultimaker Cura 4.13.1, released by Ultimaker (Ultimaker B.V., Utrecht, The Netherlands). The software also supports the download of user-made add-ons or plugins which help enhance both the user and Cura’s FFF 3D printing capabilities. We have used cura backups and settings guide as plugins. The cura backups allow access to the settings from all devices and the settings guide explains the different settings more in depth.

A possible quality problem is due to so-called over/under extrusion, resulting in dimensional inaccuracy of the specimens. During slicing, the height has been approximated by the amount of layers. Depending on the height of these layers, the approximation may lead to over or under extrusion. Since the specimens are so thin regarding the extrusion thickness, this over/under extrusion has dominated as the main quality problem. Analogously, the same challenge has been addressed for the specimen’s width. This slicer approximation has led to a new specimen geometry, which is dependent on the slicer settings in use.

In order to meet the concurrency tolerances defined in the ASTM, this new geometry should be determined and used for quality control purposes; we refer to Section 4.2.

We have used the Original Prusa i3 MK3S+ (Prusa Research a.s., Prague, Czech Republic) as the 3D printer with a smooth heat bed steel sheet.

3.5. Established 3D Printing Routines and Methods of Optimisation

Studies report that different stacking sequences exhibit higher penetration threshold capacity [63] and influence the subsequent mechanical properties of the final structure [64].

Thus, printing routines for the following layer orientations have been established for unidirectional in 0° as well as 90° and angle-ply of −45°/45°.

4. Results and Discussion

The results and discussion section summarises the crucial parameters needed for (a) the filament production and (b) the crucial slicing print parameters needed for achieving an adequate quality in 3D printed specimens, fulfilling the requirements of the testing standards.

4.1. Filament Production

Filament quality is of utmost importance in additive manufacturing. Figure 12 shows the manual quality control of the filament after production. This procedure is achieved by testing the filament with a borehole gauge. The filament is pulled through the 1.9 mm hole. By testing the filament in this way, small filament thickness deviations or agglomerations are located precisely. This method is crucial since these relatively rare local errors cause the FFF 3D printer to clog up and waste an entire print.

Below, we list the challenges that we encountered during filament production and how we approached them:

a.

Filament thickness deviation;

We observed that the speed of the extruder and the conveyor belt had a tremendous effect on the filament thickness deviations. A possible explanation for a thinner filament is that if the conveyor pulls on the filament faster than the extruder supplies new filament out of the nozzle, the filament is thereby stretched thinner. We fine-tuned the conveyor belt speed and extruder speed to work in synchronisation. For the extruder, we used a speed of 40 rev/min and for the conveyor belt a value of 5.65 rev/min. Although the filament diameter is measured and regulated during filament production between fixed values of 1.65 mm to 1.85 mm via cross-axis laser micrometer, it is additionally verified and re-structured (see: Figure 12).

b.

Air bubbles, unmelted particles in the extrudate (see top: Figure 8 and Figure 9);

Several ellipsoidal shaped single defects were observed, which consisted of partially molten pure PLA granules (Figure 10). However, they were not found to be gelled or dirt particles, and the following experiments with modified extrusion settings were successful by allowing more time for melting the PLA granules in the extruder as well as increasing to higher temperatures in the later zones. Through testing, we observed that the holes in the filament were mostly linked to the temperature in the compression and metering zones of the extruder screw (see Figure 5 and Figure 6)]. Therefore, we employed optimised temperature extrusion profiles for the two filaments to match their specific respective material properties; refer to Table 6 and Table 7.

c.

Agglomerations/particles (see Figure 10);

For limiting the formation of agglomerations in the filament, as explained in Section 3.2, and a wavy filament geometry, we utilised a more uniform cooling set-up and maintained an extrusion temperature profile where the agglomerates and particles were entirely melted. This feature was achieved by aligning the cooling path with the extruder outlet and experimenting with the extrusion temperature profile.

d.

Filament slipping off the conveyor belt;

A 3D printed in-house tool was used to keep the filament from falling off the conveyor belt (see Figure 13).

Table 6 and Table 7 show the setting of the temperature profile adjusted to achieve clear, melted filaments free of agglomerations and unmelted particles with a uniform diameter for matrix material A, PLA Luminy L130 and matrix material B, Ingeo 6302D, respectively.

4.2. FFF 3D Printing

One key issue during printing was layer adhesion and delamination. Often, the first or second layer detaches from the build plate while printing the layers above. The z offset of the printer needs to be calibrated accurately. For the Prusa printer, this offset is calibrated manually, following the instructions provided on their website. It should be noted that the offset for the tailor-made filament is different from the offset utilised for PLA from a commercial supplier. Apart from that, the following slicer settings were also found to improve the layer adhesion:

• The line width is increased to enlarge the surface area between different layers and so prevent delamination or peeling off. However, especially for the 0° orientation, the line width has to be chosen so that the sum of the lines converges precisely to the specified 8 mm specimen width;
• The cooling is disabled on the first two layers to ensure an increased adhesion to the build plate. This choice prevents the layers from shrinking too fast and allows for a higher surface area with the build plate;
• In order to give the filament more time to attach to the build plate and the surface below, the print speed was also reduced for the first two layers.

The equalize filament flow feature was used on the 90° and −45°/45° oriented specimens, considering the frequent change in print directions for these specimens.

Table 8. Shared slicer settings.

Setting	Value
Layer height	0.09 mm
Wall count	0
Top/bottom count	0
Infill density	100%
Printing temperature Mat. A	200 °C
Printing temperature Mat. B	180 °C
Build plate temperature	60 °C
Flow	100%
Number of slower layers	2
Fan speed	100%
Initial fan speed	0%
Regular fan speed at layer	3

,

Table 9. Slicer settings 0°orientation.

Setting	Value
Line width	0.38 mm
Infill line distance	0.38 mm
Infill line directions	[90]
Print speed	25 mm s−1
Initial layer print speed	15 mm s−1

,

Table 10. Slicer settings 90° orientation.

Setting	Value
Line width	0.28 mm
Infill line distance	0.28 mm
Infill line directions	[0]
Print speed	60 mm s−1
Initial layer print speed	20 mm s−1
Equalize Filament Flow	enabled

and

Table 11. Slicer settings −45°/45° orientation.

Setting	Value
Line width	0.55 mm
Infill line distance	0.55 mm
Infill line directions	[−45,45,−45,−45,45,−45]
Print speed	50 mm s−1
Initial layer print speed	15 mm s−1
Equalize Filament Flow	enabled

provides information on the optimised slicer settings used to 3D-print the specimens. We stress the importance of fine-tuning the process parameters [65].

Apart from the slicer settings, the placement of the specimens on the build plate is crucial. Not only does printing multiple specimens at once save time, but it is also more efficient and helps to prevent defects on the specimens.

Figure 14 shows a frequent issue that was encountered during 3D printing. After finishing one layer, the print head would travel over the previously printed layer diagonally, leaving excess material in its path and potentially damaging the layer.

For preventing such a failure, different measures were taken. In the case of the 90° and 0° specimens, it was sufficient to print two specimens next to each other, so that the longer sides align, as seen in Figure 15.

However, the error correction on the −45°/45° specimens requires a different method. For these specimens, auxiliary STL-files were created. These auxiliary models are cuboids which have the same height as the specimen and a $4\mathrm{mm}\times 4\mathrm{mm}$ base area. They have to be placed with an x and y offset with respect to the actual specimen to prevent corrupted travel paths, as described in Figure 14. The optimised placement of the specimen with the auxiliary .stl files can be seen in Figure 16.

We explain in the following the obtained dimensional accuracy and visual print quality. The total number of layers, $i_{\mathrm{l}}$, was six across all specimens. Multiplied by the layer height, $h_{\mathrm{l}}$ (see Table 8), the real specimen thickness, $t_{\mathrm{r}}$, was determined,

$$t_{\mathrm{r}}=i_{\mathrm{l}}·h_{\mathrm{l}}=0.54\mathrm{mm}.$$

(3)

For the 0° oriented specimens, the same can be done for the width of the specimen. The number of printer lines next to each other $i_{\mathrm{pl}}$ was counted in Cura to be 21 lines. Multiplied by the line width, $w_{\mathrm{l}}$, defined in Table 9, the real specimen width, $w_{\mathrm{r},0°}$, reads

$$w_{\mathrm{r}},0{}^{\circ }=i_{\mathrm{pl}}·w_{\mathrm{l}}=7.98\mathrm{mm}.$$

(4)

Here, the deviation from the desired width of $8\mathrm{mm}$ is $0.02\mathrm{mm}$ because the line width was chosen accordingly. However, for a different line width, this deviation may even increase to $0.4\mathrm{mm}$.

The minimum and maximum values for width, $w_{\min }$, $w_{\max }$, and thickness, $t_{\min }$, $t_{\max }$, as given by tolerances in ASTM D3039, is calculated using the parallelism tolerances shown in Figure 11 and the previously calculated real thickness (and real width for the 0° specimens $w_{\min ,0°}$, $w_{\max ,0°}$), as follows:

$$t_{\min }=t_{\mathrm{r}}-0.08\mathrm{mm}=0.46\mathrm{mm}$$

(5)

$$t_{\max }=t_{\mathrm{r}}+0.08\mathrm{mm}=0.62\mathrm{mm}$$

(6)

$$w_{\min }=w-0.08\mathrm{mm}=7.92\mathrm{mm}$$

(7)

$$w_{\max }=w+0.08\mathrm{mm}=8.08\mathrm{mm}$$

(8)

$$w_{\min ,0°}=w_{\mathrm{r},0°}-0.08\mathrm{mm}=7.9\mathrm{mm}$$

(9)

$$w_{\max ,0°}=w_{\mathrm{r},0°}+0.08\mathrm{mm}=7.06\mathrm{mm}$$

(10)

After printing the final specimens, they were removed from the build plate and measured using a digital calliper.

Table 12. Measured dimensional accuracy of the 3D printed specimens using a digital calliper.

Specimen	Mean Width	Mean Thickness	Within Tolerance	Length
mat. A: 0°	7.96 mm	0.54 mm	True	80.38 mm
mat. A: 90°	8.05 mm	0.54 mm	True	79.94 mm
mat. A: −45°/45°	8.04 mm	0.54 mm	True	79.94 mm
mat. B: 0°	7.95 mm	0.52 mm	True	80.18 mm
mat. B: 90°	8.02 mm	0.54 mm	True	79.58 mm
mat. B: −45°/45°	8.01 mm	0.54 mm	True	79.78 mm

displays the measured geometry values.

The specimens’ print quality was improved to the point where there were no longer any indications of faults in the gauge portion. The remaining imperfections were limited to the ends of the specimens, which were of secondary importance since they were covered by the clamps during tensile testing. The print quality in these sections could potentially be increased further by utilizing longer retraction moves, but this risks damaging the material and loosening the feeders grip on the filament.

The final print quality of all specimens can be seen in Figure 17 and Figure 18.

5. Conclusions

The implementation of the optimised processing parameters for materials A and B, as described in Section 3.2, resulted in the extrusion of tailor-made PLA filaments of adequate quality. The examination of the specimens reveals that they comply with the quality assurance requirements specified in ASTM D3039. We emphasise that the process parameters’ optimisation is the key for improved tolerances.

We have developed our own special die design suiting the in-house output capacities with necessary geometric opening, resulting in adequate quality for the future processing of continuous PLA-fibre reinforced PLA composite. By making the design publicly available, we invite research labs to repeat and amend our studies herein. Our future work will to focus on assessing and improving the mechanical properties of the tensile specimens, while maintaining a simple approach to implementing the printing routine. Furthermore, we aim to transfer the established procedure in the production of PLA/PLA self-reinforced composite. We stress that the material is composed of two different types of PLAs, in other words, the same type of material is reinforced to the base matrix material. Self-reinforced composites offer advantages in terms of sustainability as compared to composites made of different materials, they have a better interface between matrix and reinforcement and they are easier to recycle than traditional heterogeneous composite materials [66]. Hence, for recycling purposes, the separation of matrix and fibres has been eliminated, simplifying the recycling procedure and reducing costs. From the perspective of achieving high mechanical properties, we show possible routes to achieving improved and higher mechanical properties compared to standard/commodity PLA, especially with regard to higher strength and modulus of the final 3D printed FFF specimen.