3D-Printed Fiber-Reinforced Polymer Composites by Fused Deposition Modelling (FDM): Fiber Length and Fiber Implementation Techniques

Fused Deposition Modelling (FDM) is an actively growing additive manufacturing (AM) technology due to its ability to produce complex shapes in a short time. AM, also known as 3-dimensional printing (3DP), creates the desired shape by adding material, preferably by layering contoured layers on top of each other. The need for low cost, design flexibility and automated manufacturing processes in industry has triggered the development of FDM. However, the mechanical properties of FDM printed parts are still weaker compared to conventionally manufactured products. Numerous studies and research have already been carried out to improve the mechanical properties of FDM printed parts. Reinforce polymer matrix with fiber is one of the possible solutions. Furthermore, reinforcement can enhance the thermal and electrical properties of FDM printed parts. Various types of fibers and manufacturing methods can be adopted to reinforce the polymer matrix for different desired outcomes. This review emphasizes the fiber types and fiber insertion techniques of FDM 3D printed fiber reinforcement polymer composites. A brief overview of fused deposition modelling, polymer sintering and voids formation during FDM printing is provided, followed by the basis of fiber reinforced polymer composites, type of fibers (synthetic fibers vs. natural fibers, continuous vs. discontinuous fiber) and the composites’ performance. In addition, three different manufacturing methods of fiber reinforced thermoplastics based on the timing and location of embedding the fibers, namely ‘embedding before the printing process (M1)’, ‘embedding in the nozzle (M2)’, and ‘embedding on the component (M3)’, are also briefly reviewed. The performance of the composites produced by three different methods were then discussed.


Introduction
Additive manufacturing (AM) or 3-dimensional printing (3DP) technology is one of the most promising areas in component manufacturing. AM has paved its way into application areas ranging from automotive [1], construction [2], aerospace [3] and consumer products to biomedical products such as prosthetics [4]. AM refers to a group of fabrication techniques where parts are fabricated layer-by-layer directly from a computer-aided design (CAD) file. AM technology is a very broad term that encompasses many methods such as Stereolithography (SLA) of a photopolymer liquid [5], Laminated Object Manufacturing (LOM) from plastic laminations [6], Selective Laser Sintering (SLS) from plastic or metal powder [7] and Fused Deposition Modelling (FDM) from plastic filaments [8].
Since 1980, many studies have been conducted to maximize the potential applications of these technologies, as AM is well-known and still a far more cost-effective alternative to subtractive manufacturing technologies such as milling, drilling and turning [9]. FDM, also called Fused Filament Fabrication (FFF), is one of the most popular techniques due to its relatively low cost, low material wastage and ease of use. Nowadays, most people can even purchase and use this technique at home. However, FDM 3D print is yet to replace conventional manufacturing in producing functional parts. FDM 3D print parts are weaker than conventionally manufactured counterparts due to their layer-by-layer fabrication method. Research has been carried out to improve the mechanical properties of FDM printed parts by using various methods, such as optimizing printing parameters, annealing, snap-fitting [10], printing in an oxygen free environment [11], mechanical pressing [12] and fiber reinforced thermoplastics. Of all the methods, fiber reinforced polymer composites (FRPC) are known to have high stiffness, strength, damage tolerance, fatigue resistance and corrosion resistance. FRPCs are produced by adding fibers or particles into the thermoplastic matrix to improve the mechanical strength of the printed components [13]. This method reduces voids and increases interlaminar bonding between the deposited filaments. There are two types of fiber reinforcement: continuous and discontinuous, depending on fiber length. Fiber reinforced composites have a long history and are traditionally produced by techniques like hand lay-up, molding, etc. FDM is a relatively new technique for manufacturing fiber reinforced polymer composites. Research in FDM 3D printed fiber reinforced polymer composites has flourished recently. Previous works in FDM 3D printed fiber reinforced polymer composites have been reviewed by several state-of-the-art review papers with a different emphasis [14][15][16][17][18][19][20][21]. Additive manufactured fiber reinforced polymer composites produced by different AM techniques such as FDM, DIW, SLS, SLA, and Laminated Object Manufacturing (LOW) and their mechanical behavior were summarized by Li et al. [14]. The matrix and materials used in fiber reinforced additive manufacturing and the mechanical behavior were reviewed by Fidan et al. [15]. Understanding and optimizing the printing parameters of the FDM process to achieve the desired mechanical properties of the printed part was summarized by Krajangsawasdi et al. [16] and Shanmugam et al. [17]. The FDM printed natural fiber reinforced polymer produced was reviewed by Mazzanti et al. [18], whereas FDM printed discontinuous fiber polymer composites were summarized by Hu et al. [19]. Lastly, the fabrication techniques and application of 3D printed anisotropic polymer materials were presented by Chen et al. [20] and Xu et al. [21]. However, there has not been an in-depth discussion on fabrication techniques, and fiber types of FDM 3D printed fiber reinforced polymer. Therefore, this review aims to summarize recent progress in FDM 3D printed fiber reinforced polymers and emphasize fiber type and fabrication techniques. The following section summarizes the working principle for FDM 3D printing, especially the polymer sintering and forming of voids. Previous attempts to improve print quality by optimizing printing parameters are presented in (Section 3). Since this review's focus is on fiber as reinforcement, the basis of fiber reinforced polymer composites and the type of fibers (synthetic fibers vs. natural fibers, continuous vs. discontinuous fiber) will be given in Section 4. We highlight the manufacturing methods of fiber reinforced polymer composites in Section 5, and then propose opportunities for future development in Section 6.

Fused Deposition Modelling Process
The process of creating an object with an FDM printer begins with the product design using CAD software such as CATIA and SOLIDWORKS, which is saved in a Surface Tessellation Language (STL) file. Before such a file can be printed, it must be converted into a format that the 3D printer can understand, namely a G-code file. Slicer software such as Cura, Ideamaker and Simpliy 3D are used to convert the STL file into a G-code file. The G-code contains commands for moving parts within the printer. It consists of G-and M-commands that have assigned actions and movements in x-, y-and z-directions of the nozzle and bed of the FDM printer. Figure 1 shows a schematic diagram of a typical FDM printer setup and filament deposition process. nozzle and bed of the FDM printer. Figure 1 shows a schematic diagram of a typical FDM printer setup and filament deposition process.
A 3D geometry is produced in the FDM process by building up an extruded thermoplastic filament layer-by-layer. The filament is fed through the extrusion head (nozzle), which is heated to a semi-liquid state and applied to the build platform through a nozzle in layers. Each layer is bonded to the adjacent layers in the semi-liquid state. Thus, it is crucial to control the feed rate of the printer to ensure that the previous layer does not solidify too early. Feed rate can be easily adjusted in the slicer software. Figure 1b shows a schematic diagram of the FDM extrusion head and filament deposition process. The filament is first driven into the print head by rollers. As it passes through a liquefier, the feedstock is heated by a heater to a viscous melt and pushed out of the print nozzle by the incoming still-solid filament.

Polymer Sintering of Deposited Thermoplastics
The FDM process uses a heated nozzle to melt and extrude thermoplastic filaments such as Acrylonitrile-butadiene-styrene (ABS), poly-lactic acid (PLA), nylon, polypropylene (PP), polyethylene (PE), and so on. These materials are common thermoplastics used in 3D printing. Each material has a different melting point, and the printer must be set accordingly. An error in setting up the temperature of the feedstock material will affect the cosmetic and strength of 3D printed products. During the FDM process, each filament extruded through the heated nozzle solidifies and forms a cross-bond with the adjacent filaments extruded previously. These filaments form a bridge between them, known as the "neck" by the process of polymer sintering [23]. This bond, which is responsible for growing the necks within a layer, may be termed as the "intra-layer bonding". Since the temperature of the previously solidified layer is still high, there is a good tendency for similar bonds to form between the filaments of the two successive layers, which can be termed as "inter-layer bonding". Figure 2 describes the surface contact stage, neck growth and molecular diffusion of the deposited filaments [24].  [18] (b) schematic of the FDM extrusion head and filament deposition process, reprinted/adapted with permission from [22], 2008, Emerald Publishing, Bingley, UK.
A 3D geometry is produced in the FDM process by building up an extruded thermoplastic filament layer-by-layer. The filament is fed through the extrusion head (nozzle), which is heated to a semi-liquid state and applied to the build platform through a nozzle in layers. Each layer is bonded to the adjacent layers in the semi-liquid state. Thus, it is crucial to control the feed rate of the printer to ensure that the previous layer does not solidify too early. Feed rate can be easily adjusted in the slicer software. Figure 1b shows a schematic diagram of the FDM extrusion head and filament deposition process. The filament is first driven into the print head by rollers. As it passes through a liquefier, the feedstock is heated by a heater to a viscous melt and pushed out of the print nozzle by the incoming still-solid filament.

Polymer Sintering of Deposited Thermoplastics
The FDM process uses a heated nozzle to melt and extrude thermoplastic filaments such as Acrylonitrile-butadiene-styrene (ABS), poly-lactic acid (PLA), nylon, polypropylene (PP), polyethylene (PE), and so on. These materials are common thermoplastics used in 3D printing. Each material has a different melting point, and the printer must be set accordingly. An error in setting up the temperature of the feedstock material will affect the cosmetic and strength of 3D printed products. During the FDM process, each filament extruded through the heated nozzle solidifies and forms a cross-bond with the adjacent filaments extruded previously. These filaments form a bridge between them, known as the "neck" by the process of polymer sintering [23]. This bond, which is responsible for growing the necks within a layer, may be termed as the "intra-layer bonding". Since the temperature of the previously solidified layer is still high, there is a good tendency for similar bonds to form between the filaments of the two successive layers, which can be termed as "inter-layer bonding". Figure 2 describes the surface contact stage, neck growth and molecular diffusion of the deposited filaments [24]. Gurrala and Regalla et al. [23] also investigated the effects of inter-layer bonding, intra-layer bonding and neck formation between adjacent filaments on the tensile strength of FDM products, both experimentally and theoretically. They found that in the FDM sam-   Gurrala and Regalla et al. [23] also investigated the effects of inter-layer bonding, intra-layer bonding and neck formation between adjacent filaments on the tensile strength of FDM products, both experimentally and theoretically. They found that in the FDM sample with 0 • raster angle, the failure of a specimen was due to inter-layer fracture, whereas at 45 • raster angle, the specimens failed due to both inter-layer and intra-layer fracture. This research has shown that inter-layer and intra-layer bonds play an important role in the mechanical properties of FDM products. Figures 3 and 4 are schematic diagrams of multi-layer extruded filaments. The strength of printed parts depends on these two interlaminar bonds. To improve these interlaminar bonds, much research has been done focusing on the printing parameters. They believe that an optimal setup results in high strength FDM products with great interlaminar bonding. An overview of these research will be presented in the next section (Section 3).

Figure 2.
Formation of the neck between filaments: (1) surface contacting, (2) neck growth, and (3) diffusion at interface, Reprinted with permission from [24], 2004, Elsevier, Amsterdam, The Netherlands. Gurrala and Regalla et al. [23] also investigated the effects of inter-layer bonding, intra-layer bonding and neck formation between adjacent filaments on the tensile strength of FDM products, both experimentally and theoretically. They found that in the FDM sample with 0° raster angle, the failure of a specimen was due to inter-layer fracture, whereas at 45° raster angle, the specimens failed due to both inter-layer and intra-layer fracture. This research has shown that inter-layer and intra-layer bonds play an important role in the mechanical properties of FDM products. Figures 3 and 4 are schematic diagrams of multi-layer extruded filaments. The strength of printed parts depends on these two interlaminar bonds. To improve these interlaminar bonds, much research has been done focusing on the printing parameters. They believe that an optimal setup results in high strength FDM products with great interlaminar bonding. An overview of these research will be presented in the next section (Section 3).    Formation of the neck between filaments: (1) surface contacting, (2) neck growth, and (3) diffusion at interface, Reprinted with permission from [24], 2004, Elsevier, Amsterdam, The Netherlands. Gurrala and Regalla et al. [23] also investigated the effects of inter-layer bonding, intra-layer bonding and neck formation between adjacent filaments on the tensile strength of FDM products, both experimentally and theoretically. They found that in the FDM sample with 0° raster angle, the failure of a specimen was due to inter-layer fracture, whereas at 45° raster angle, the specimens failed due to both inter-layer and intra-layer fracture. This research has shown that inter-layer and intra-layer bonds play an important role in the mechanical properties of FDM products. Figures 3 and 4 are schematic diagrams of multi-layer extruded filaments. The strength of printed parts depends on these two interlaminar bonds. To improve these interlaminar bonds, much research has been done focusing on the printing parameters. They believe that an optimal setup results in high strength FDM products with great interlaminar bonding. An overview of these research will be presented in the next section (Section 3).   Multiple attempts have been made to numerically model the sintering process of polymers based on heat transfer calculations. Early work by Yardimci et al. [25,26] presented different modelling approaches to capture the heat transfer between printed beads, but did not consider the polymer flow dynamics. Bellehumeur et al. [24] used a model based on a polymer sintering model described by Pokluda et al. [27]. Pokluda et al. performed an energy balance between surface tension and viscous dissipation [27], and Bellehumeur et al. incorporated temperature-dependent surface tension and viscosity into Pokluda et al.'s model. Although they did not model molecular diffusion, they found that the extruded material cooled too quickly for complete bonding. They also reported that the convective heat transfer coefficient greatly affects bond formation and neck growth, with less heat transfer leading to better neck formation. However, they modelled isothermal polymer sintering and did not consider heat transfer from the hot extruded material to the surrounding material. Bellini [28] performed extensive modelling of the entire FDM process with ceramic-filled filament using four different numerical simulations focusing on: the liquefier, the nozzle contraction, deposition on the printing bed and on stacked layers. This enabled the tracking of material temperature, swelling and filling as a function of the various printing parameters. It was found that the higher thermal conductivity of the filled material increases heat transfer from the liquefier to the printed material and improve the flow behavior.

Voids in FDM Printed Components
The strength of components produced by the FDM process differs from that of parts fabricated by conventional injection molding. The presence of voids and gaps between the individual layers reduces the layer-to-layer bond strength. The strength of fabricated components by FDM is compromised by significant voids and weak interlaminar bonding between layers. The percentage of void is depends on printing parameters and typically ranged from 4% to 18.5% [29][30][31]. Although the deposition filaments can be integrated into the adjacent deposition filaments by their gravity and the force of the printer's stepper motor, the presence of significant voids between them greatly affects the mechanical properties of fabricated components. In addition, the extruded filament cools rapidly from the melting temperature to the chamber temperature, developing inner stresses responsible for a weak bond between two deposition filaments. This leads to a deformation between layer (inter-layer) and within layer (intra-layer) in the form of cracks, delamination or even part fabrication failure [32].
Voids in FDM printed part can be classified into five categories according to their formation mechanism: raster gap voids, partial neck-growth voids, sub-perimeter voids, intra-bead voids and infill voids [29]. Raster gap voids are formed by gaps between the raster surfaces and are visible, as shown in Figure 5. Partial neck growth voids are internal voids formed by incomplete neck growth between adjacent intra-and inter-layer rasters ( Figure 6). This occurs when rasters solidify before coalescence is complete. Partial neck growth voids are a major contributor to voids in FDM prints. Due to physical limitations, sub-perimeter voids form in between turning rasters along the perimeter of the FDM layer. Even when the printer is set into 100% infill density, voids form between the blue wall lines and the infill zones, as shown in Figure 7. Intra-bead voids are specific to composites due to fiber loading, as shown in Figure 8. Finally, infill voids are voids in the infill depending on the infill pattern selected for printing the parts and can be controlled/adjusted.  . Raster gap voids [29]: (a) FDM process parameter [18] and (b) cross-sectional view of ABS printed part with 0/90 rasters. Reprinted with permission from [33], 2003, Emerald Publishing, Bingley, UK. Figure 6. Partial neck growth voids, [29]: cross sectional view of an FDM laminate, where 2y is the neck growth between adjacent rasters. Reprinted with permission from [22], 2008, Emerald Publishing, Bingley, UK.  . Partial neck growth voids, [29]: cross sectional view of an FDM laminate, where 2y is the neck growth between adjacent rasters. Reprinted with permission from [22], 2008, Emerald Publishing, Bingley, UK.

Quantification of Voids
Density measurement, imaging technique, optical microscope (OM), scanning electron microscope (SEM) and CT scan are commonly used to study voids [29]. OM is widely used to study the meso-structures of printed parts, while SEM is often used to analyze microstructures. OM and SEM can both capture 2D images, with the right angle proportional to the FDM layers, to capture valuable data for the analysis, such as raster gap voids, partial neck growth voids, sub-perimeter and intra-bead voids. In contrast, a CT scan is a valuable tool for observing and investigating the effects of FDM voids in 3D. Moreover, Figure 8. Intra-beads voids. Reprinted with permission from [35], 2020, Elsevier, Amsterdam, The Netherlands.

Quantification of Voids
Density measurement, imaging technique, optical microscope (OM), scanning electron microscope (SEM) and CT scan are commonly used to study voids [29]. OM is widely used to study the meso-structures of printed parts, while SEM is often used to analyze microstructures. OM and SEM can both capture 2D images, with the right angle proportional to the FDM layers, to capture valuable data for the analysis, such as raster gap voids, partial neck growth voids, sub-perimeter and intra-bead voids. In contrast, a CT scan is a valuable tool for observing and investigating the effects of FDM voids in 3D. Moreover, CT scans can also be used to reconstruct 3D models of scanned specimens in great detail [29].

Printing Parameters
Four main factors affect the print quality of FDM printed parts: material, machine, printing process and environment [9]. Matrix and reinforcement materials are the main determining properties of the printed parts. Machine factors are mainly related to printer productivity. Nozzle temperature, heating mechanism, diameter and geometry influence the quality of the printer. The temperature is set according to the materials to be processed. The nozzle diameter affects the print resolution. Selecting the right temperature based on the material is tedious and significantly influences productivity and print quality. A high temperature results in better interlamellar bonding and less void space occupancy due to lower viscosity and better rheological properties, improving mechanical properties [36]. However, a too high temperature may affect printing quality, especially the dimensional accuracy. This section discusses the main parameters that are set before the printing process.
The FDM processed component primarily depends on three important control factors: namely the extruder, the processing and the structural, which are defined in Table 1 and illustrated in Figure 9. Selecting the optimal process parameters for printing will significantly improve mechanical performance, surface roughness and geometric accuracy. Based on previous research, it was found that the mechanical strengths such as tensile and flexural of parts printed by FDM are highly anisotropic [20,37], i.e., the performance of a material depends on the direction of printing. The printing speed has a minor effect on tensile strength but a significant effect on the production cost. A low feed rate increases interlamellar bonding but negatively affects productivity and increases production costs. The angle between the deposed material and the x-axis Raster gap The distance between two contiguous paths on the same layer Build orientation Basic print build either upright, on-edge and flat Air gap or raster-to-raster gap is the distance between two adjacent rasters on the same layer. The default value is usually zero, meaning the beads are touching. A positive gap means a gap between adjacent rasters or a negative gap, which means that the bead tracks overlap. With many slicing software that control the 3D printers, only the infill rate can be set, which represents the density of the printed pattern. Therefore, the actual positive or negative air gap must be determined or estimated, even if some specimens were printed with a setting of 100 infill. This affects the part as the air gap has been shown to be an important contributing factor in tensile strength [38].
Onwubolu et al. [39] investigated five significant process parameters: layer thickness, part orientation, raster angle, raster width and air gap. According to their results, the tensile strength was highest when the layer thickness is lowest, and the part orientation was printed parallel to the direction of the applied tensile force. Furthermore, a high raster angle with a low raster width and a negative air gap contributes to the increase in tensile strength, which was also observed by Dawoud et al. [40]. As shown in Figure 9, the raster angle, i.e., the angle of the deposited raster or bead relative to the horizontal direction, has a greater influence on tensile strength, toughness and ductility. Ning et al. [41] studied fiber orientations at 0 • , 90 • and ±45 • . They found the latter had low tensile strength but better toughness and ductility due to poor interfacial adhesion between the matrix and reinforcing filament. tive or negative air gap must be determined or estimated, even if some specimens were printed with a setting of 100 infill. This affects the part as the air gap has been shown to be an important contributing factor in tensile strength [38].
Onwubolu et al. [39] investigated five significant process parameters: layer thickness, part orientation, raster angle, raster width and air gap. According to their results, the tensile strength was highest when the layer thickness is lowest, and the part orientation was printed parallel to the direction of the applied tensile force. Furthermore, a high raster angle with a low raster width and a negative air gap contributes to the increase in tensile strength, which was also observed by Dawoud et al. [40]. As shown in Figure 9, the raster angle, i.e., the angle of the deposited raster or bead relative to the horizontal direction, has a greater influence on tensile strength, toughness and ductility. Ning et al. [41] studied fiber orientations at 0°, 90° and ±45°. They found the latter had low tensile strength but better toughness and ductility due to poor interfacial adhesion between the matrix and reinforcing filament.  Garg et al. [32] also studied the effects of layer thickness and raster angles on the failure of FDM parts under tensile loading. They found that to increase the mechanical strength, the printed part should be oriented so that the longest contours align the tension/stress force, which is confirmed by Onwubolu et al. [39]. As the layer thickness increases, the number of layers required to fabricate the specimen decreases, further decreasing the number of air voids in the interstitices. However, Onwubolu believes that a lower layer thickness increases the adhesion force in the layer-to-layer bond and hence improves the tensile strength. Figure 10 illustrates the presence of air voids in the deposited filaments. These air voids are the major cause of crack initiation and propagation in the specimen. Therefore, a lower number of air voids contributes to a higher strength of the printed parts. Garg et al. [32] also studied the effects of layer thickness and raster angles on the failure of FDM parts under tensile loading. They found that to increase the mechanical strength, the printed part should be oriented so that the longest contours align the tension/stress force, which is confirmed by Onwubolu et al. [39]. As the layer thickness increases, the number of layers required to fabricate the specimen decreases, further decreasing the number of air voids in the interstitices. However, Onwubolu believes that a lower layer thickness increases the adhesion force in the layer-to-layer bond and hence improves the tensile strength. Figure 10 illustrates the presence of air voids in the deposited filaments. These air voids are the major cause of crack initiation and propagation in the specimen. Therefore, a lower number of air voids contributes to a higher strength of the printed parts. Another investigation was carried out by Carneiro et al. [43] on the effects of build orientation, layer thickness, infill degree, and inclusion of fiberglass as reinforcement fiber. In this work, they also compared the results of FDM and specimens fabricated by compression molding. With an optimal selection of printing parameters and fiber reinforcement of the printed parts, the result shows that the strength is almost the same as the specimens fabricated by compression molding with less than 20%. They believed this result is not due to poor adhesion between the filaments but due to the existence of voids in the printed samples, which supports the results of Garg et al. [32].
On the other hand, infill pattern options such as honeycomb, hexagonal, triangular and solid structures significantly alter interfacial adhesion and interlamellar bonding. This parameter also affects the tensile, flexural, impact and compression properties. For instance, a solid infill pattern may have lower impact resistance than other pattern types because the free air gaps in these structures help absorb large impact shock [44]. Hui et al. [45] investigated how the different infill types, together with layering, affect the tensile strength and elastic modulus of carbon fiber (CF), glass fiber (GF) and Kevlar-reinforced Nylon composites. It was found that rectangular fill had the highest tensile strength, followed by hexagonal and triangular infill, respectively, but different behavior was noticed Another investigation was carried out by Carneiro et al. [43] on the effects of build orientation, layer thickness, infill degree, and inclusion of fiberglass as reinforcement fiber. In this work, they also compared the results of FDM and specimens fabricated by compression molding. With an optimal selection of printing parameters and fiber reinforcement of the printed parts, the result shows that the strength is almost the same as the specimens fabricated by compression molding with less than 20%. They believed this result is not due to poor adhesion between the filaments but due to the existence of voids in the printed samples, which supports the results of Garg et al. [32].
On the other hand, infill pattern options such as honeycomb, hexagonal, triangular and solid structures significantly alter interfacial adhesion and interlamellar bonding. This parameter also affects the tensile, flexural, impact and compression properties. For instance, a solid infill pattern may have lower impact resistance than other pattern types because the free air gaps in these structures help absorb large impact shock [44]. Hui et al. [45] investigated how the different infill types, together with layering, affect the tensile strength and elastic modulus of carbon fiber (CF), glass fiber (GF) and Kevlar-reinforced Nylon composites. It was found that rectangular fill had the highest tensile strength, followed by hexagonal and triangular infill, respectively, but different behavior was noticed by another group of researcher [46].
Based on previous research and investigations, although optimization of processing parameters, such as build orientation, raster angle, infill degree, layer thickness, or feed rate to improve the mechanical performance of thermoplastics has been studied, the results are still poor due to the used of pure thermoplastics and the presence of air voids in each layer. Therefore, combining different materials to achieve the desired mechanical and functional properties is a promising way to solve these problems. Table 2 shows a compilation of previous studies on printing parameters' effect on the mechanical properties of printed parts.

Fiber Reinforced Polymer Composite (FRPC)
The development of composites that are compatible with FDM printers has attracted a lot of attention. This is because composites promise better mechanical properties and performances compared to neat polymers. Many results in the development of new printable composites reinforced with particles, fibers or nanomaterials have already been demonstrated. Carbon black, platelets, chopped fibers and polymer fibrils are mixed with the polymer matrix and then extruded together during printing. However, the performance of these composites depends largely on the fiber orientation in the plastic and the fibervolume-fraction (FVF). Parts manufactured with FDM from neat polymer have shown insufficient strength in load tests. This limits the range of applications in which FDM technology can be used for functional parts and not for prototypes.
Researchers have used fiber reinforced polymer composites (FRPCs) to overcome the aforementioned limitations in their work. In FRPCs, the material properties of a component are enhanced by combining reinforcing fibers and polymer matrix. Various fibers have been used for reinforcement, including chopped carbon fibers, carbon nanotubes, glass fibers, natural fibers etc. [72][73][74][75]. There are certain requirements that FRPC materials must meet in order to be processed by AM, namely: 1.
Types of reinforcement and matrices; 2.
The fiber reinforcement must be matched in size, shape and length to the part's intended use. Both the matrix material, which holds the fibers in place, and the reinforcement must be compatible with the selected AM technique. A good bond between fibers and matrix is required at the fiber-matrix interface to transfer loads efficiently from the matrix, resulting in composites that follow the "rule of mixtures". Fiber loading is also crucial to obtain AM composites with good mechanical properties. Mechanical properties such as elastic modulus increase with fiber loading at a low loading ratio but degrade after reaching an optimum value [76]. This phenomenon generally occurs due to poor wettability of the fiber with the thermoplastic, which results in a poor fiber-matrix interface.
Higher loading leads to an increase in viscosity and a decrease in flowability, leading to processability problems such as clogging of the nozzle. Furthermore, fiber reinforcement may cause negative effects on interlaminar bonding and the properties of printed parts. Based on previous research, interlamellar matrix regions between the reinforced fiber layers are critical regions that are highly prone to delamination when subjected to mechanical stress. Delamination can result from weak fiber-matrix bonding, which often leads to internal damage in composites, potentially leading to global failure of the component with reduced strength and stiffness [77]. Furthermore, porosity and weak interface bonding between fibers and matrix have been cited as a major problem for 3D printed fiber reinforced polymer composites [78]. Understanding the mechanism of filament bonding is important to further investigate how FRPC works to reduce voids and increase the strength of the interlaminar bond between the deposited filaments.

Synthetic Fibers vs. Natural Fibers
Various fibers were used as reinforcement for polymer composites and can be grouped under two categories: synthetic fibers and natural fibers [13]. Natural fibers were first used as reinforcement for polymers since 1936 [79] and were slowly replaced by synthetic fibers because synthetic fibers are usually much stronger than natural fibers. However, natural fibers re-emerged as reinforcing materials for polymer composites when environmental issues became more important in engineering applications. In FDM 3D printed polymer composites, both synthetic and natural fibers were used to reinforce polymers, although synthetic fibers are a more popular choice.
Natural fibers are used as reinforcement to reduce the inorganic content in thermoplastic composites without compromising mechanical strength, ultimately improving biodegradability and reducing costs [18]. Common natural fibers used in FDM 3D printed polymer composites are jute [87], wood [99], harakeke/flax [100,101], bamboo [101], sugarcane and many more. Recent and systematic reviews on natural fiber reinforced polymer composites as feeders in FDM-Based 3D Printing were reported by researchers [18,102,103]. Natural fibers are a cheaper and greener alternative to reinforce polymer matrix during the FDM 3D printing process, but challenges such as fiber agglomeration, clogging in the nozzle, poor fiber-matrix interface, non -homogenous mixing etc., have to be investigated further. Furthermore, various treatments such as chemical and thermal are required to be applied to natural fibers to enhance the performance of natural fibers reinforced polymers. In addition, a different combination of polymer matrix and natural fibers requires different treatments and processes. As such, more research works are required to improve the performance of natural fibers reinforced polymers. Environmentally friendly engineering materials are getting more attention recently. Therefore, polymer composites produced by bio-based polymers such as PLA [99], soy-based resin [104,105] etc., and reinforced with natural fibers have great potential because they are biodegradable and environmentally friendly.
The advantages and limitations of synthetic and natural fibers as reinforcing materials for FDM printed polymer are summarized in Table 3. Table 3. The advantages and limitations for synthetic fibers and natural fibers as reinforced material for FDM printed polymer [18,19,106].

Type of Fibers Advantages Limitation
Synthetic fibers

Continuous vs. Discontinuous Fiber
Fiber reinforced polymer composite is a subcategory of fiber reinforced composites. Generally, fiber reinforcement can be categorized into discontinuous and continuous fibers according to critical fiber length [107]. Critical length l c is the fiber length that allows applied load transfer to the reinforced fibers by the matrix, and depends on fiber's ultimate strength σ f , fiber diameter d, and fiber-matrix bond strength or shear yield strength of the matrix τ c . Continuous fibers are referred to fiber with length more than 15 l c , and discontinuous fibers are fibers with length less than 15 l c [107]. Nevertheless, some other researchers have slightly different definition. Krajangsawasdi et al. further classified short and discontinuous fiber, where short fibers are fibers shorter than critical length l c , and discontinuous fiber are those with length above critical length l c [16]. Pruß and Vietor defined discontinuous fibers as fibers with fiber length less than 1 mm (0.04 in.), while continuous fibers are fibers with a length above 50 mm (2 in.) [108].
Besides the obvious motivation of improving mechanical properties, reinforcement can also be used to provide the material with additional functions such as electro-conductivity, thermal conductivity or biocompatibility. Kalsoom et al. [109] and Wang et al. [110] have provided a general overview of 3D printable composites; this paper instead focuses in more detail on the engineering aspects of FDM as a composite manufacturing method.
Conventionally, fiber-reinforced composites can be classified into: (a) continuous and aligned fiber composites, (b) discontinuous and aligned-fiber composites, and (c) discontinuous and randomly oriented-fiber composites, depending on the length and alignment of the fibers [107]. The major advantages and disadvantages are listed in Table 4. Table 4. Brief comparison of fiber reinforced composites, according to length and orientation of fiber [107,111].

Discontinuous and Randomly Oriented-Fiber Composites
Properties of the composite are highly anisotropic The continuous and aligned fibers can reinforce composites in the intended direction but have no significant effect in the transverse direction. Conventional methods for producing continuous and aligned fiber composites are pultrusion, prepreg, and filament winding [107]. In terms of additive manufacturing, the FDM 3D printed 'continuous and aligned fiber composites' are being investigated by various researchers [80][81][82]86,88,94,97,112]. Previously, 3D printed continuous and aligned fiber composites were mostly printed using inhouse developed or modified 3D printers [80,86]. The first commercial 3D printer capable of printing continuous and aligned fiber composites was developed by MarkForged. With the availability of commercial machines such as Markforged's Markone, Marktwo 3D printers, research on FDM printing of continuous fiber reinforced thermoplastics (CFRT) composites is booming. Most of the recent research on FDM 3D printed continuous and aligned fiber composites uses Markforged's Markone, Marktwo 3D printers [81,82,88,94,97,112]. Various types of continuous fibers, such as carbon fibers [80][81][82]86,88], glass fibers [81,94], and Kevlar fibers [81], have been used as reinforcement. In general, the FDM printed continuous and aligned fibers can have better electrical properties [97] and mechanical properties, such as tensile strength [81,82,86,88,94], flexural strength [86] if the printing parameters are properly selected. A systematic review of 3D printed continuous fiber polymer composites is presented by [113]. However, 3D printed continuous and aligned fiber composites are limited in terms of design freedom, as fiber placement is challenging and more voids are created, especially when printing complex shapes [19,88]. Design freedom is one of the main advantages of additive manufacturing over conventional manufacturing, and incorporating continuous fibers into FDM 3D printing, negates this advantage.

Discontinuous and Randomly Oriented-Fiber Composites
Discontinuous fiber composites have a long history, and the first scientific publication dates back to 1936 [79]. Due to the nature of reinforced fibers and conventional fabrication methods, such as hand lay-up, resin transfer molding, etc., early fiber reinforced composites are mainly discontinuous and randomly oriented. FDM 3D printed discontinuous fiber composites are manufactured using composite filaments by commercial FDM 3D printers. Generally, discontinuous fibers were premixed with the polymer matrix as composite filament, and the composite filaments were then used in FDM 3D printing to produce discontinuous fiber composites. To date, more than 10,000 published papers have been found in Scopus using the keywords "additive manufacture" and "short fiber reinforced polymers", and it is not possible to discuss them all here. However, most of these research papers focused on the mechanical or thermal properties of the composites. They did not report on the orientation of the fibers in FDM 3D printed discontinuous fiber composites. Nevertheless, research with FDM 3D printed discontinuous and randomly oriented-fiber composites have been reported by several researchers [91], although not all of them emphasized the orientation of the fibers.
One of the recent works with FDM 3D printed discontinuous and randomly orientedfiber composites was reported by Zhao et al. [91]. They compared the tensile properties of 3D printed CNT-short glass fiber (SGF) reinforced PLA composite with the tensile properties of 3D printed PLA, SGF/PLA, and found that both composites are better than neat PLA in terms of tensile strength and tensile modulus. In addition, CNT-SGF /PLA composite has a higher tensile strength than SGF/PLA composite. From the SEM images of the fracture surfaces of the composite specimens, they found that the fibers in the composites are randomly oriented. Su et al. reinforced polyamide with reclaimed carbon fiber in four different weight percentages (10%, 20%, 30%, 40%). They found that the fibers were better aligned at low fiber contents (10-20%) and had no significant alignment at 40%. They concluded that the tensile performance of the reclaimed carbon fiber reinforced polyamide composites (rCF/PA) largely depended on the fiber content and orientation, with higher fiber content and aligned fiber being able to improve tensile strength. All composites, including rCF/PA with 40 wt% and non-aligned fiber performed better than neat PLA [114].

Discontinuous and Aligned-Fiber Composites
Discontinuous and aligned fibers are an alternative to continuous fibers in 3D printing of polymer composites, with the advantage of better design freedom. Early research on discontinuous and aligned-fiber composites (also named as aligned discontinuous fiber thermoplastic) produced by non-additive manufacturing processes was summarized by Such et al. [115]. Although the manufacturing methods for 3D printed FDM 3D printed discontinuous and aligned-fiber composites are different from the conventional make discontinuous and aligned-fiber composites, the motivations for reinforcing polymers with discontinuous and aligned-fiber are similar. In general discontinuous and aligned-fiber are added to polymers for three main reasons: (1) to improve mechanical, thermal, or electrical properties in the desired direction, (2) to reduce the cost and complexity of manufacturing compared to composites with continuous fibers, and (3) enabling design freedom or complex geometries [20,79,115,116]. FDM 3D printed discontinuous and aligned-fiber composites are mainly manufactured using composite filaments by commercial FDM 3D printers. The discontinuous fibers were aligned by shear (referred to as shear-induced alignment or flow-induced alignment), where the shear force between a nozzle and the molten material forces the fibers to align in the direction of extrusion or flow [20,117]. Furthermore, the orientation of fibers is affected by experimental extrusion width, where experimental extrusion width depends on extrusion temperature, speed and width. Fibers were more aligned in a narrow extruder than in a wider extruder [118], as shown in Figure 11. One of the first published papers on FDM 3D printed discontinuous, and aligned-fiber composites was by Tekinalp et al. [76]. They fabricated the carbon fiber reinforced ABS filament and used the filament with a commercial FDM 3D printer. They applied the method of Bay and Tucker [119] to characterize the fiber orientation in the printed part and found that the carbon fibers in the printed parts are mainly oriented in the load-bearing direction. They concluded that the carbon fibers could increase the strength and modulus of both the FDM printed and compression molded samples, but the FDM samples have significant voids [76].
composite, they were able to improve the thermal conductivity of the polymer [120]. However, they also pointed out that the presence of voids in FDM-printed composites affects the through-plane thermal conductivity of the composites. Papon and Haque investigated fracture toughness of 3D printed carbon fiber reinforced PLA composites with different fiber content (3 wt.%, 5 wt.%, 7 wt.% and 10 wt.%), manufactured by two different nozzle shapes (circular and square) [121]. The square shape nozzle was custom-made to improve the contact area and inter-bead void. The fibers are mostly aligned in the extrusion direction, but they did not report how nozzle shape affects the fiber orientation. Their experimental results show that the fracture toughness increased with fiber content from 0% to 5%, at both layer orientation of 45°/−45° and 0°/90°. The print layer orientation of 45°/−45° and 0°/90° has no major different in fracture properties. Furthermore, parts printed by a square nozzle have better fracture toughness than parts printed by a circular nozzle because less void is produced in parts produced by the square nozzle.  Jia et al. fabricated graphite flakes reinforced PA6/POE-g-MAH/PS composite with an FDM 3D printer and verified by microscopy that the graphite flakes were aligned along the through-plane direction (parallel to the x-y plane) via microscopy. With this designed composite, they were able to improve the thermal conductivity of the polymer [120]. However, they also pointed out that the presence of voids in FDM-printed composites affects the through-plane thermal conductivity of the composites. Papon and Haque investigated fracture toughness of 3D printed carbon fiber reinforced PLA composites with different fiber content (3 wt.%, 5 wt.%, 7 wt.% and 10 wt.%), manufactured by two different nozzle shapes (circular and square) [121]. The square shape nozzle was custom-made to improve the contact area and inter-bead void. The fibers are mostly aligned in the extrusion direction, but they did not report how nozzle shape affects the fiber orientation. Their experimental results show that the fracture toughness increased with fiber content from 0% to 5%, at both layer orientation of 45 • /−45 • and 0 • /90 • . The print layer orientation of 45 • /−45 • and 0 • /90 • has no major different in fracture properties. Furthermore, parts printed by a square nozzle have better fracture toughness than parts printed by a circular nozzle because less void is produced in parts produced by the square nozzle.
Researchers at the University of Bristol developed a method named High Performance Discontinues Fiber (HiPerDIF) to manufacture discontinuous and aligned-fiber composites [122] and investigated the performance of composites produced with this method [116]. Generally, the fibers were suspended in a liquid medium (water), and the orientation of fiber was controlled by the orientation head [122]. With this method, they fabricated discontinuous and aligned-fiber epoxy composites using carbon fiber [122] and recycled carbon fibers [123]. They reported that the mechanical properties of composites are proportional to the fiber lengths [123]. To expand the HiPerDIF technology to additive manufacturing/FDM, Blok et al. have identified 4 different polymers (ABS, PLA, Nylon, PETG) as the potential polymer matrix materials to be reinforced with high performance discontinues and formed the feedstock materials for FDM. The four polymers were selected based on 14 factors. They fabricated the composite tapes using an in-house consolidation method, where the HiPerDiF fiber was sandwiched between two layers of polymer matrix films of 0.125 mm.
They proofed that aligned discontinuous fiber composites produced using HiPerDIF technology are better than currently available short fiber thermoplastic. Furthermore, the composite fabricated with HiPErDIF technology has comparable mechanical behavior compared with continuous fiber composite but with better manufacturing flexibility [89]. Krajangsawasdi et al. recently extended their work by fabricating 3D printer filament using ADFRC fiber to reinforce PLA thermoplastic. They managed to produce HiPerDiF-PLA filament and also identified the optimal printing parameters of their newly developed filament. They compared the mechanical properties of the HiPerDiF-PLA printed parts with PLA, PLA-short carbon fiber, PLA-continuous carbon fiber, and Markforged continuous carbon fiber [116], and they concluded that HiPerDiF-PLA outperformed other PLA composites in terms of mechanical performance.

Manufacturing Techniques of Fiber Reinforced Polymer Composites
When manufacturing fiber-reinforced polymer composites, the techniques used to embed the fiber into a thermoplastic matrix influence the mechanical properties of the printed parts. There are at least three different ways of embedding fibers, taking into account the timing and location of the embedding of fibers [108]. Figure 12 illustrates the three techniques for embedding fibers in the matrix used by researchers to reinforce continuous fibers.
They proofed that aligned discontinuous fiber composites produced using HiPerDIF technology are better than currently available short fiber thermoplastic. Furthermore, the composite fabricated with HiPErDIF technology has comparable mechanical behavior compared with continuous fiber composite but with better manufacturing flexibility [89]. Krajangsawasdi et al. recently extended their work by fabricating 3D printer filament using ADFRC fiber to reinforce PLA thermoplastic. They managed to produce HiPerDiF-PLA filament and also identified the optimal printing parameters of their newly developed filament. They compared the mechanical properties of the HiPerDiF-PLA printed parts with PLA, PLA-short carbon fiber, PLA-continuous carbon fiber, and Markforged continuous carbon fiber [116], and they concluded that HiPerDiF-PLA outperformed other PLA composites in terms of mechanical performance.

Manufacturing Techniques of Fiber Reinforced Polymer Composites
When manufacturing fiber-reinforced polymer composites, the techniques used to embed the fiber into a thermoplastic matrix influence the mechanical properties of the printed parts. There are at least three different ways of embedding fibers, taking into account the timing and location of the embedding of fibers [108]. Figure 12 illustrates the three techniques for embedding fibers in the matrix used by researchers to reinforce continuous fibers.

Method 1 (M1): Embedding before the Printing Process
M1 is the method most used by researchers because it is the most straight forward approach and does not require major modifications to the machine. The polymer matrix and reinforcement are premixed before 3D printing in the form of 3D printer composite filaments. Commercial composite filaments such as carbon fiber/ PETG, carbon fiber/PLA, carbon fiber/Nylon etc., are available in the market. However, only a limited type of fiber (mostly carbon fiber) was used to reinforce commercial composite filaments, and other fibers' potential is under investigation. Various researchers studied combinations of different matrix and different reinforcements. Usually, the composite filaments were fabricated in-house, following the processes described by [41,125] and shown in Figure 13. Although the exact procedures used by different researchers might not be identical, the treated fibers and polymer matrix are generally mixed in a mixer or blender and then fed into an extruder (twin-screw extruder or extruder). The composites normally go through the second extrusion process to obtain better matrix distribution and reinforcement by supplying fiber matrix. The prefabricated composite filament has a constant fiber volume ratio.
fibers' potential is under investigation. Various researchers studied combinations of different matrix and different reinforcements. Usually, the composite filaments were fabricated in-house, following the processes described by [41,125] and shown in Figure 13. Although the exact procedures used by different researchers might not be identical, the treated fibers and polymer matrix are generally mixed in a mixer or blender and then fed into an extruder (twin-screw extruder or extruder). The composites normally go through the second extrusion process to obtain better matrix distribution and reinforcement by supplying fiber matrix. The prefabricated composite filament has a constant fiber volume ratio. Figure 13. Preparation process of composite filaments, for M1 embedding before the printing process Reprinted with permission from [125], 2020, Elsevier, Amsterdam, The Netherlands.
For nanofiber-reinforced polymers made with FDM, Shofner et al. [127] combined vapor-grown carbon fibers (VGCFs) with ABS copolymer to create a composite filament for use with FDM. The dispersion, porosity and fiber alignment issues were also investigated to prevent agglomeration when mixing VGCFs with ABS. The results show that the tensile strength and modulus of VGCF-filled ABS are, on average, 39% and 60% higher, respectively, than those of the neat ABS. On the other hand, the storage modulus measurements from the dynamic mechanical analysis indicated that the stiffness increased by 68%.
Zhong et al. [129] conducted experiments to investigate the processability of short glass fiber reinforced ABS matrix composites with three different glass fiber contents used as feedstock filaments in FDM. The results showed that glass fiber could significantly improve the ABS filament's tensile strength and surface rigidity. The effect of fiber content on the mechanical properties of printed parts is another interesting research topic. Similar but not identical work was also reported by Wang et al., where carbon fiber and glass fiber Figure 13. Preparation process of composite filaments, for M1 embedding before the printing process Reprinted with permission from [125], 2020, Elsevier, Amsterdam, The Netherlands.
For nanofiber-reinforced polymers made with FDM, Shofner et al. [127] combined vapor-grown carbon fibers (VGCFs) with ABS copolymer to create a composite filament for use with FDM. The dispersion, porosity and fiber alignment issues were also investigated to prevent agglomeration when mixing VGCFs with ABS. The results show that the tensile strength and modulus of VGCF-filled ABS are, on average, 39% and 60% higher, respectively, than those of the neat ABS. On the other hand, the storage modulus measurements from the dynamic mechanical analysis indicated that the stiffness increased by 68%.
Zhong et al. [129] conducted experiments to investigate the processability of short glass fiber reinforced ABS matrix composites with three different glass fiber contents used as feedstock filaments in FDM. The results showed that glass fiber could significantly improve the ABS filament's tensile strength and surface rigidity. The effect of fiber content on the mechanical properties of printed parts is another interesting research topic. Similar but not identical work was also reported by Wang et al., where carbon fiber and glass fiber with different fiber contents were mixed with PEEK to form composite filaments [125]. Both CF/PEEK and GF/PEEK composites were reported to have better tensile and flexural strengths than the neat PEEK, and GF/PEEK was stronger than CF/PEEK in terms of tensile and flexural strengths. They also highlighted that composite with higher fiber contents had more porosity (voids) and lower mechanical properties (tensile strength, flexural strength, impact strength and ductility). Tekinalp et al. [76] demonstrated that the short carbon fibers (0.2-0.4 mm) reinforced ABS composites produced by FDM showed an increase in tensile strength and modulus with increasing fiber content, with a maximum increase of 115% and 700%, respectively, at a fiber content of 40 wt%. However, the tensile strength of the FDM printed specimen is still lower than the tensile strength of compression molded specimen, especially at higher fiber loading. Tensile strength and modulus measurement of specimens prepared by both FDM and compression molding are shown in Figure 14.
carbon fibers (0.2-0.4 mm) reinforced ABS composites produced by FDM showed an in-crease in tensile strength and modulus with increasing fiber content, with a maximum increase of 115% and 700%, respectively, at a fiber content of 40 wt%. However, the tensile strength of the FDM printed specimen is still lower than the tensile strength of compression molded specimen, especially at higher fiber loading. Tensile strength and modulus measurement of specimens prepared by both FDM and compression molding are shown in Figure 14. Ning et al. [41] studied the use of thermoplastic ABS composites with different percentages of added carbon fibers (CF) of various sizes in an FDM printer. Their observation showed that the addition of CF increased the tensile strength and Young's modulus of the plastic compared to pure plastic specimens. Moreover, the CFRP specimens enriched with longer carbon fibers (150 µm) had higher tensile strength and Young's modulus as well as lower toughness and ductility than the specimens with shorter carbon fibers (100 µ m). They also studied the effect of fiber content on the mechanical properties of the FDM printed ABS/carbon fiber composites. The best performance of the printed parts was obtained at a fiber content of 5 wt%. Higher fiber content deteriorated the performance of printed parts due to the higher porosity. Figure 15 shows that the highest mean value was found at 5 wt% CF with 42 MPa, and the lowest at 10 wt% with 24 MPa, which was almost the same with neat ABS plastic. Ning et al. [73] also reported the fabrication of M1 short CFRP specimens using an FDM machine (Creatr, Leapfrog Co., Alphen aan den Rijn, Netherlands) and a composite filament (FilaBot Co., Montpelier, USA) with a diameter of 1.75 mm that contained 5wt% chopped CF in ABS thermoplastic matrix. Their work evaluated the mechanical properties of parts printed at four different process parameters: Nozzle temperature, infill speed, raster angle and layer thickness. The objective was to find the best parameters to improve the tensile strength of the parts. Ning et al. [41] studied the use of thermoplastic ABS composites with different percentages of added carbon fibers (CF) of various sizes in an FDM printer. Their observation showed that the addition of CF increased the tensile strength and Young's modulus of the plastic compared to pure plastic specimens. Moreover, the CFRP specimens enriched with longer carbon fibers (150 µm) had higher tensile strength and Young's modulus as well as lower toughness and ductility than the specimens with shorter carbon fibers (100 µm). They also studied the effect of fiber content on the mechanical properties of the FDM printed ABS/carbon fiber composites. The best performance of the printed parts was obtained at a fiber content of 5 wt%. Higher fiber content deteriorated the performance of printed parts due to the higher porosity. Figure 15 shows that the highest mean value was found at 5 wt% CF with 42 MPa, and the lowest at 10 wt% with 24 MPa, which was almost the same with neat ABS plastic. Ning et al. [73] also reported the fabrication of M1 short CFRP specimens using an FDM machine (Creatr, Leapfrog Co., Alphen aan den Rijn, The Netherlands) and a composite filament (FilaBot Co., Montpelier, WT, USA) with a diameter of 1.75 mm that contained 5 wt% chopped CF in ABS thermoplastic matrix. Their work evaluated the mechanical properties of parts printed at four different process parameters: Nozzle temperature, infill speed, raster angle and layer thickness. The objective was to find the best parameters to improve the tensile strength of the parts.  , respectively. To achieve maximum improvement of mechanical properties, the proportion of fibers varies considerably in the different cases of the study, mostly because the conditions for fiber distribution and interfacial bond strength differ greatly from case to case. So far, the added content of fibers is up to 40 wt%, and the composites with more fibers cannot be printed due to nozzle clogging problem. In addition, composites with higher fiber loading are difficult to process into continuous filaments for FDM due to loss of toughness. Therefore, the properties of the resulting composites are limited by the low fiber content. Applying plasticizers and compatibilizers could be a possible way to improve the feedstock processability [129].
Most of the reported research utilized short fiber in producing polymer composites by using method 1 (Embedding before the printing process) and reported research on FDM 3D printed composites with continuous fiber is relatively scarce. FDM 3D printed composites with continuous fiber are mostly produced with self-manufactured continuous carbon fiber reinforced filament [132][133][134]. Hu et al. designed a device to manufacture continuous fiber reinforced thermoplastic (CFRTP) filaments, where molten resin was squeezed into continuous carbon fiber to form continuous carbon fiber prepreg filament [132]. They optimized the printing parameters (printing temperature, printing speed, layer thickness) during printing continuous fiber reinforced PLA and found the optimized composite had better flexural strength than the neat PLA. Similar manufacturing technique was utilized by Zhang et al. [134] and Uşun et al. [133] to produce continuous fiber reinforced filament. Zhang et al. produced two different composites, the continuous carbon fiber reinforced PLA (CCF-PLA) and the continuous carbon fiber reinforced Nylon (CCF-Nylon). They found that CCF-PLA had higher tensile and bending strength than neat PLA and short carbon fiber reinforced PLA. Similarly, CCF-Nylon had higher bending and tensile strength than neat Nylon. Uşun et al. fabricated continuous fiber-reinforced thermoplastic (CFRTP) filaments with a melt impregnation line and later printed 3D parts with the CFRTP filaments. Their experiments showed CFRTP CF-PLA composites with 40% CF have higher tensile and flexural strength than composites with 22% CF and 33% CF [133]. Recently, a refined manufacturing technique was proposed to manufacture continuous fiber reinforced thermoplastic filament [135] and the manufactured filament has better prepreg quality and volume fraction. However, they did not report the mechanical properties of 3D printed part by using the new filament.

Method 2 (M2): Embedding in the Nozzle
In method 2, both polymer matrix and reinforcing fiber are mixed in the printing nozzle during the printing process. Embedding the fiber to the polymer matrix in the nozzle during the printing process requires two material supplies, one for the polymer matrix and another for the reinforcement (fiber). The extrusion nozzle receives both the thermoplastic polymer and the continuous fiber, and the continuous fiber is fed through the core of the nozzle, as shown in Figure 16. When the nozzle is heated, infusion of the matrix occurs, and the molten thermoplastic material is deposited along with the reinforcing filament. Research machines mostly produced polymer composites fabricated with this method, as no commercial 3D printer that provides a similar function is available in the market. Fidan et al. [15] mention in their review that in the fabricating of composites, the selection of matrix and reinforcement must be compatible physically (good adhesion), chemically (matrix and fiber must not react chemically) and thermally (similar thermal behavior; coefficient of thermal expansion). occurs, and the molten thermoplastic material is deposited along with the reinforcing filament. Research machines mostly produced polymer composites fabricated with this method, as no commercial 3D printer that provides a similar function is available in the market. Fidan et al. [15] mention in their review that in the fabricating of composites, the selection of matrix and reinforcement must be compatible physically (good adhesion), chemically (matrix and fiber must not react chemically) and thermally (similar thermal behavior; coefficient of thermal expansion). In the case of continuous fibers, the reinforcing fibers are then supplied as a dry roving. The thermoplastic matrix is added separately. This allows adjusting the fiber volume ratio by machine control during the printing process. This enables the same nozzle to deposit neat plastic. However, these advantages can be offset by additional challenges. For example, the infiltration process of dry roving without air inclusions must be carefully controlled, while the extrusion and deposition processes must run simultaneously. Furthermore, handling a dry roving with its non-rigid properties is much more difficult than handling a pre-impregnated one. Fabrication of specimen by using method M2 also was reported by Yang et al. [86] and Tian et al. [80]. They presented a 3D printing equipment with a novel composite extrusion head that can continuously process CF with ABS and PLA, respectively. Yang et al. developed a novel composite extrusion head in which dry carbon fiber is fed through a melt pool of ABS. This increased the in-plane mechanical properties by a factor of 205, but a limiting factor was the interlaminar shear properties of the printed part. Research about the implementation of continuous fibers inside the nozzle with different machine setups was reported by Prüß and Vietor [108], where they In the case of continuous fibers, the reinforcing fibers are then supplied as a dry roving. The thermoplastic matrix is added separately. This allows adjusting the fiber volume ratio by machine control during the printing process. This enables the same nozzle to deposit neat plastic. However, these advantages can be offset by additional challenges. For example, the infiltration process of dry roving without air inclusions must be carefully controlled, while the extrusion and deposition processes must run simultaneously. Furthermore, handling a dry roving with its non-rigid properties is much more difficult than handling a pre-impregnated one. Fabrication of specimen by using method M2 also was reported by Yang et al. [86] and Tian et al. [80]. They presented a 3D printing equipment with a novel composite extrusion head that can continuously process CF with ABS and PLA, respectively. Yang et al. developed a novel composite extrusion head in which dry carbon fiber is fed through a melt pool of ABS. This increased the in-plane mechanical properties by a factor of 205, but a limiting factor was the interlaminar shear properties of the printed part. Research about the implementation of continuous fibers inside the nozzle with different machine setups was reported by Prüß and Vietor [108], where they designed and fabricated an adapted FDM print head with 1 port for fiber and 2 ports for polymers.
A novel technique called continuous lattice fabrication (CLF) was proposed by Eichenhofer et al. [136]. The CLF head consists of a two-stage, pultrusion-extrusion system. They reported an increase in tensile properties of carbon fiber reinforced PA12 composites, which can achieve a tensile strength of 560 MPa and elastic moduli of 83 GPa along the fiber direction. The material properties obtained are comparable to those of high-performance aluminum alloys, which have a tensile strength of 540 MPa and tensile modulus of 72 GPa [137]. The technique involves a softening cycle procedure [137] that attempts to maximize the mechanical properties of the printed composites by minimizing the residual void content. Matsuzaki et al. [87] printed continuous fibers (straight carbon fibers or twisted jute fiber yarns) by feeding them through a nozzle simultaneously with a thermoplastic filament (PLA) serving as a matrix. They reported a strength and stiffness of 195 MPa and 10.5 GPa, respectively, due to a low V f of 6.6%. This technique also showed uneven fiber distribution as the fiber was not pre-impregnated into the matrix.

Method 3 (M3): Embedding on the Component
Composites produced by method 3 require two or more independent extruders, each with an independent nozzle, to deposit both polymer matrix and reinforce fiber on the printer bed, as shown in Figure 17. Several research that implemented the 'fibers after the nozzle method' used 3D printers manufactured by Markforged [46,81,97,138,139].
Markforced, with its MarkOne, MarkTwo and MarkX printers, is the first (and only, to the best of our knowledge) company that produced commercially available printers which can manufacture continuous FRPC [140]. Research with M3 process was presented by Dickson et al. [81], who report the fabrication of continuous carbon, fiberglass and Kevlar fiber reinforced polymer composites using a Markforged Markone 3D printer. Naranjo-Lozada et al. [46] compared the tensile performance of Nylon composites fabricated by two different fiber insertion methods; M1 with chopped carbon fiber and M3 with continuous carbon fiber by a Markforged Marktwo printer. Three test setups were carried out to compare the tensile strength of pure nylon, chopped carbon fibers and continuous carbon fibers. In addition, printing parameters such as infill density, infill pattern, fiber volume fraction and build orientation were also investigated. Naranjo-Lozada et al. concluded that increasing the fiber concentration and length at optimal printing parameters improves the tensile strength of printed parts. The results suggest that the initial point of application of the reinforcement fiber affects the tensile properties of the specimen. Mei  In addition, they also reported that hot-pressed composites have higher strength compared to their non hot pressed counterpart [138].
uneven fiber distribution as the fiber was not pre-impregnated into the matrix.

Method 3 (M3): Embedding on the Component
Composites produced by method 3 require two or more independent extruders, each with an independent nozzle, to deposit both polymer matrix and reinforce fiber on the printer bed, as shown in Figure 17. Several research that implemented the 'fibers after the nozzle method' used 3D printers manufactured by Markforged [46,81,97,138,139]. Markforced, with its MarkOne, MarkTwo and MarkX printers, is the first (and only, to the best of our knowledge) company that produced commercially available printers which can manufacture continuous FRPC [140]. Research with M3 process was presented by Dickson et al. [81], who report the fabrication of continuous carbon, fiberglass and Kevlar fiber reinforced polymer composites using a Markforged Markone 3D printer. Naranjo-Lozada et al. [46] compared the tensile performance of Nylon composites fabricated by two different fiber insertion methods; M1 with chopped carbon fiber and M3 with continuous carbon fiber by a Markforged Marktwo printer. Three test setups were carried out to compare the tensile strength of pure nylon, chopped carbon fibers and continuous carbon fibers. In addition, printing parameters such as infill density, infill pattern, fiber volume fraction and build orientation were also investigated. Naranjo-Lozada et al. concluded that increasing the fiber concentration and length at optimal printing parameters improves the tensile strength of printed parts. The results suggest that the initial point of application of the reinforcement fiber affects the tensile properties of the specimen. Mei et al. printed carbon reinforced Nylon composites with Marktwo printer at different fiber angles, and they concluded that the sample printed with mixed isotropic fiber angle [0°/45°/90°] 2 is stronger than samples printed at fiber angles [30°/45°/60°] 2 and [15°/45°/75°] 2 . In addition, they also reported that hot-pressed composites have higher strength compared to their non hot pressed counterpart [138]. Non-commercial printers or self-developed methods also produced composites produced by method 3. Implementing the fibers after the nozzle directly into the print job is done by Mori et al. with their 'dieless forming' method [141,142], where carbon fibers were placed manually on FDM printed acrylonitrile butadiene styrene (ABS) plate. Mori et al. have conducted experiments with manually deposited continuous fibers between plastic layers made with FDM [141]. Their research shows that the carbon fibers were entirely pulled out from a ruptured tensile test specimen, and the fibers had little effect on the resulting tensile strength due to weak fiber adhesion. They recommend a thermal postprocess to bond the fibers to the matrix. This enables the specimens to reach almost twice the strength. With an additional thermal bonding, the test specimen reached about twice the tensile strength of unreinforced specimens [142]. Since the tensile strength without fibers was only 11 MPa, about a quarter of common ABS values, the FLM process quality, in this case, is more than questionable.
Baumann et al. [124] used three different fiber implementation concepts (direct overprint, hypodermic needle, solvent) to fabricate polymer composites. They found out that direct overprint is the best method among the three. They also reported that the M3 technique significantly increases tensile strength and elastic modulus for different cases of continuous carbon fiber reinforced polymers. This study showed their processes' potential for producing functional parts for engineering applications.
This classification is important because the properties of the part depend not only on the amount, often measured as the volume fraction of the reinforcing fiber, but also on the manner in which the fibers are integrated into the matrix material. Currently, FDM 3D printers that can fabricate composite with the M3 method are commercially available for continuous fibers, whereas only M1 is available for short fibers, as summarised in Table 5. To give an idea of the current performance in terms of fiber reinforced composites and the mechanical performances for parts produced with FRPC using the FDM process, this section summarizes a large collection of data on the mechanical properties and methods of fiber implementation for various fiber reinforced material systems. The mechanical characterization of printed fiber-reinforced polymers compared to neat polymer material is the primary focus of researchers publishing research in the field of FDM with composites. Table 6 summarizes previous research on different fiber implementations and the effect on the mechanical performances of FDM printed parts.    Melting point, thermal decomposition temperature and crystallization temperature of both composites are higher than neat PEEK GF/PEEK has better interfacial bonding than CF/PEEK Both composites have better mechanical strengths (tensile, flexural, impact) than neat PEEK Composites with 5 wt.% fiber content are the best in terms of mechanical strength. The increase of fiber content from 5% to 15% reduced the strengths

Opportunities for Future Developments
Currently, three methods (M1, M2, M3) are commonly used to fabricate fiber reinforced polymer composites. As discussed in the previous section and Table 5, most researchers have produced composites where the short fibers have been reinforced before the printing process (M1), and the short fibers added in this way are unable to reinforce intralayer and interlayer adhesion strength effectively. Embedding short fibers on the component (M3) may affect the intralayer and interlayer adhesion of deposited parts, but no previous work in embedding short fibers on the component is reported. Hence, embedding short fibers on the component is worth further investigation, and a new fabrication technique/machine is required to achieve this.
Furthermore, due to better environmental awareness, green composites such as fiberreinforced bio-polymer composites produced by bio-based polymers and natural fibers, require further studies. Examples of popular bio-based polymers are PLA, biodegradable polybutylene adipate terephthalate (PBAT), polybutylene succinate (PBS), etc. [153,154]. Previously, most research focused on synthetic fiber reinforced PLA, and currently, some works on natural fiber reinforced PLA are reported [103,152]. The potential of FDM printed fiber reinforced composites formed by other bio-based polymers or bio-degradable biopolymers (as matrix) and natural fiber (as reinforcement) should be explored. Similarly, bioactive particles integration within the FDM can be explored [155,156].There is also a need to develop better numerical models with a focus on: (i) the manufacturing process of printed parts to reduce development costs, optimize the deposition parameters and provide a better scientific understanding of the fiber reinforced FDM process, based on current research in simulation of FDM manufacturing process [157,158]; and (ii) developing models to understand the deformation and failure mechanism of printed parts under mechanical and/or thermal loading conditions [32,151,159]. Some of these models can incorporate hardening, pore closure, creep and the plasticity of polymeric materials [160][161][162].
Next, most of the previous research focused on the fundamental aspect such as mechanical, thermal, or electrical properties of FDM 3D printed fiber reinforced polymer in the form of standard test specimens instead of the actual product, despite some works on actual products being reported [64,163,164]. Further research on the performance of actual fiber reinforced composite products fabricated via FDM 3D printing is necessary before FDM 3D printed fiber reinforced composite can be adopted widely in product design.

Conclusions
The Fused Deposition Modelling method can significantly affect manufacturing industries and additive manufacturing technology. The ability to produce functional parts directly from a commercial 3D printer with controllable properties has created a huge rush for new developments and research in this field. This paper provides an overview of FDM 3D printing and the formation of voids in 3D printed parts. FDM printing of fiber reinforced polymer composites can be an ideal method to improve the mechanical properties, thermal properties and electrical properties of FDM-printed parts. Both synthetic fibers and natural fibers of different lengths can be used as reinforcement for polymer matrix to produce different outcomes. On top of fiber type and length, fiber implementation technique is another factor that can affect the performance of composites. Currently, three techniques (M1, M2 and M3) are commonly used to fabricate fiber reinforced polymer composites. Embedding short fibers on the component is not explored and required further investigation. In addition, 3D printed green composites with bio-based polymer or biodegradable polymer and natural fibers are also worth for further investigation. Lastly, to optimize the application of FDM printed fiber reinforced composites, numerical models for both printing process and failure mechanism have to be developed, and testing on 3D printed actual products have to be conducted. Acknowledgments: The first author would like to thank the university for financial support.

Conflicts of Interest:
The authors declare no conflict of interest.