Flexure Behaviors of ABS-Based Composites Containing Carbon and Kevlar Fibers by Material Extrusion 3D Printing

Short-fiber-reinforced thermoplastics are popular for improving the mechanical properties exhibited by pristine thermoplastic materials. Due to the inherent conflict between strength and ductility, there are only a few successful cases of simultaneous enhancement of these two properties in polymer composite components. The objective of this work was to explore the feasibility of simultaneous enhancement of strength and ductility in ABS-based composites with short-carbon and Kevlar fiber reinforcement by material extrusion 3D printing (ME3DP). Microstructure characterization and measurement of thermal and mechanical properties were conducted to evaluate the fiber-reinforced ABS. The influence of printing raster orientation and build direction on the mechanical properties of material extrusion of 3D-printed composites was analyzed. Experimental results demonstrated that the reinforcement of the ABS-based composites by short-carbon and Kevlar fibers under optimized 3D-printing conditions led to balanced flexural strength and ductility. The ABS-based composites with a raster orientation of ±45° and side build direction presented the highest flexural behaviors among the samples in the current study. The main reason was attributed to the printed contour layers and the irregular zigzag paths, which could delay the initiation and propagation of microcracks.


Materials and Processing
In this study, ABS, ABS/CF, ABS/KF, and ABS/CF/KF filament spools were purchased from Nanovia (Saint Paul, France). The materials were supplied in the form of filament tows with a standard diameter of 1.75 mm. Table 1 presents mean length, mono diameters, and population of fibers acquired from the datasheet of supplier. Before processing, the filaments were kept in a drying oven to reduce the relative humidity. The filament tows were utilized to fabricate ABS and ABS composite samples using a desktop 3D printer (Raise3D N2 Plus, Raise3D INC., Costa Mesa, CA, USA). A diagram of the printing process is shown in Figure 1. The printing process consisted of melting and extrusion. During melting, the filaments were rolled into the heated die. When the extrusion head was heated to the preset melting temperature of the materials, the printing filaments were extruded. During the extrusion process, the deposition lines were arranged in an orderly fashion on the printing platform and were superimposed to form the work piece layer by layer. In the current study, the specimen was printed from a nozzle with an extrusion temperature of 250 • C onto a plate preheated to 100 • C with a printing speed of 60 mm/min, as suggested by the supplier. The rectangular specimens with dimensions of 9.5 mm × 80.0 mm × 4.2 mm were printed with different raster orientations and build directions, as shown in parts b and c of Figure 1, respectively. Each specimen had 42 layers with a controlled layer thickness of 0.1 mm.
Polymers 2019, 11, x FOR PEER REVIEW 3 of 12 In this study, ABS, ABS/CF, ABS/KF, and ABS/CF/KF filament spools were purchased from Nanovia (Saint Paul, France). The materials were supplied in the form of filament tows with a standard diameter of 1.75 mm. Table 1 presents mean length, mono diameters, and population of fibers acquired from the datasheet of supplier. Before processing, the filaments were kept in a drying oven to reduce the relative humidity. The filament tows were utilized to fabricate ABS and ABS composite samples using a desktop 3D printer (Raise3D N2 Plus, Raise3D INC., Costa Mesa, CA, USA). A diagram of the printing process is shown in Figure 1. The printing process consisted of melting and extrusion. During melting, the filaments were rolled into the heated die. When the extrusion head was heated to the preset melting temperature of the materials, the printing filaments were extruded. During the extrusion process, the deposition lines were arranged in an orderly fashion on the printing platform and were superimposed to form the work piece layer by layer. In the current study, the specimen was printed from a nozzle with an extrusion temperature of 250 °C onto a plate preheated to 100 °C with a printing speed of 60 mm/min, as suggested by the supplier. The rectangular specimens with dimensions of 9.5 mm × 80.0 mm × 4.2 mm were printed with different raster orientations and build directions, as shown in parts b and c of Figure 1, respectively. Each specimen had 42 layers with a controlled layer thickness of 0.1 mm.

Characterization
Thermogravimetric analysis (TGA) in the current study was performed using a Mettler TGA/DSC 3+ simultaneous thermal analyzer (Mettler Toledo Co., Switzerland). Specimens with a mass of 6.8 ± 2.1 mg were cut from the raw filament tows. The materials were heated from room temperature to 900 °C at a heating rate of 10 °C/min in a pure nitrogen atmosphere. The evolution of the weight of the materials as a function of temperature was recorded using the STARe software.
The microstructure of the cryofractured filament tows and the fractured printed composites was studied using scanning electron microscopy (Hitachi S-4800 FE-SEM, Japan). The morphology

Characterization
Thermogravimetric analysis (TGA) in the current study was performed using a Mettler TGA/DSC 3+ simultaneous thermal analyzer (Mettler Toledo Co., Greifensee, Switzerland). Specimens with a mass of 6.8 ± 2.1 mg were cut from the raw filament tows. The materials were heated from room Polymers 2019, 11, 1878 4 of 12 temperature to 900 • C at a heating rate of 10 • C/min in a pure nitrogen atmosphere. The evolution of the weight of the materials as a function of temperature was recorded using the STARe software.
The microstructure of the cryofractured filament tows and the fractured printed composites was studied using scanning electron microscopy (Hitachi S-4800 FE-SEM, Hitachi, Japan). The morphology study focused on the adhesion states, feature dimensions, distribution, and dispersion of the short-carbon fibers and/or Kevlar fibers in the filament tows and printed components.
Three-point bending tests were performed using an MTS universal mechanical testing machine (MTS E44, Eden Prairie, MN, USA) with a 30 kN loading cell. The bending tests were conducted at room temperature with a constant cross-head speed of 2.1 mm/min and according to ISO14125. The deflection of the samples during the mechanical tests was measured by a digital micrometer (Mitutoyo, ID-SX, Kawasaki, Japan). The flexure moduli of the materials were calculated from the initial slope of stress-strain curves over a strain range of 0.05-0.25%. The maximum stress was taken as the flexural strength. The flexural toughness of the materials was calculated by integrating the area under the stress-strain curve. Each measurement was averaged from the results of five repeated and recorded tests, in order to investigate the effects of fiber addition, raster orientations, and building directions on the flexure behaviors of ABS/CF/KF composites.

Thermal and Morphological Behaviors of the Filament Tows
TGA curves of the pristine ABS, ABS/CF, ABS/KF, and ABS/CF/KF composites are presented in Figure 2. In this figure, ABS and the ABS/CF composite show only one mass loss step in the temperature range of 300-500 • C. However, the degradation process of the ABS/KF and ABS/CF/KF composites is a two-step process with a substantial weight loss in the range from 350-500 • C and a second weight loss in the range from 500-600 • C. In addition, the initiated decomposition temperatures of the ABS/CF, ABS/KF, and ABS/CF/KF composites are slightly higher than that of pristine ABS in the range of 300-400 • C. In the temperature range from 400-500 • C, the ABS/KF composite and pristine ABS exhibit similar decomposition behaviors. However, the addition of CFs accelerates the degradation processes of the ABS/CF and ABS/CF/KF composites, resulting in an elevated slope of the weight loss. Upon increasing the temperature to 900 • C, ABS degrades completely and other materials retain carbon residues. Three-point bending tests were performed using an MTS universal mechanical testing machine (MTS E44, Eden Prairie, MN, USA) with a 30 kN loading cell. The bending tests were conducted at room temperature with a constant cross-head speed of 2.1 mm/min and according to ISO14125. The deflection of the samples during the mechanical tests was measured by a digital micrometer (Mitutoyo, ID-SX, Kawasaki, Japan). The flexure moduli of the materials were calculated from the initial slope of stress-strain curves over a strain range of 0.05-0.25%. The maximum stress was taken as the flexural strength. The flexural toughness of the materials was calculated by integrating the area under the stress-strain curve. Each measurement was averaged from the results of five repeated and recorded tests, in order to investigate the effects of fiber addition, raster orientations, and building directions on the flexure behaviors of ABS/CF/KF composites.

Thermal and Morphological Behaviors of the Filament Tows
TGA curves of the pristine ABS, ABS/CF, ABS/KF, and ABS/CF/KF composites are presented in Figure 2. In this figure, ABS and the ABS/CF composite show only one mass loss step in the temperature range of 300-500 °C. However, the degradation process of the ABS/KF and ABS/CF/KF composites is a two-step process with a substantial weight loss in the range from 350-500 °C and a second weight loss in the range from 500-600 °C. In addition, the initiated decomposition temperatures of the ABS/CF, ABS/KF, and ABS/CF/KF composites are slightly higher than that of pristine ABS in the range of 300-400 °C. In the temperature range from 400-500 °C, the ABS/KF composite and pristine ABS exhibit similar decomposition behaviors. However, the addition of CFs accelerates the degradation processes of the ABS/CF and ABS/CF/KF composites, resulting in an elevated slope of the weight loss. Upon increasing the temperature to 900 °C, ABS degrades completely and other materials retain carbon residues. The one weight-loss step that occurs for ABS and the ABS/CF composite in Figure 2 could be attributed to the radical degradation process in ABS under an inert atmosphere [26]. The second degradation process of the ABS/KF and ABS/CF/KF composites is due to the decomposition of the The one weight-loss step that occurs for ABS and the ABS/CF composite in Figure 2 could be attributed to the radical degradation process in ABS under an inert atmosphere [26]. The second degradation process of the ABS/KF and ABS/CF/KF composites is due to the decomposition of the Kevlar main chain [27]. The introduction of CFs could improve the thermal stability of the CF, ABS/CF, and ABS/CF/KF composites due to a higher degradation temperature than those of ABS and the ABS/KF composite [28]. On the other hand, the addition of CFs also improves the thermal conductivity of the ABS/CF and ABS/CF/KF composites, causing steeper weight reduction slopes than those for ABS and the ABS/KF composite [29]. Finally, the residues at the end of the heating process can be used to prove the presence of the reinforcements. Taking the TGA curve of ABS as a baseline, the CF and KF contents in the ABS/CF and ABS/KF composites are 2.48 wt % and 4.42 wt %, respectively. Also, the CF and KF contents in the ABS/CF/KF composites are 3.34 wt % and 3.05 wt %, respectively.
Scanning electron microscopy (SEM) was used to investigate the cryofractured morphology of pristine ABS and the ABS/CF, ABS/KF, and ABS/CF/KF composite filaments employed in this study. As shown in Figure 3, both the CFs and KFs are well dispersed and distributed within the ABS matrix (arrow 1). In Figure 3d,f, the small grayish cylinders, indicated by arrow 2, indicate the short CFs with an average diameter of 5.8 ± 0.5 µm oriented along the filament length [30]. In Figure 3e,f, the large stick-like fillers, indicated by arrow 3, are KFs. Note that the diameter of the short KFs is 12 ± 0.6 µm, twice that of the CFs. The fibers that were pulled out during the cryofracture process lead to some voids (black dots) in the ABS matrix. Most of the fibers demonstrate a good interface with the ABS matrix. Kevlar main chain [27]. The introduction of CFs could improve the thermal stability of the CF, ABS/CF, and ABS/CF/KF composites due to a higher degradation temperature than those of ABS and the ABS/KF composite [28]. On the other hand, the addition of CFs also improves the thermal conductivity of the ABS/CF and ABS/CF/KF composites, causing steeper weight reduction slopes than those for ABS and the ABS/KF composite [29]. Finally, the residues at the end of the heating process can be used to prove the presence of the reinforcements. Taking the TGA curve of ABS as a baseline, the CF and KF contents in the ABS/CF and ABS/KF composites are 2.48 wt % and 4.42 wt %, respectively. Also, the CF and KF contents in the ABS/CF/KF composites are 3.34 wt % and 3.05 wt %, respectively. Scanning electron microscopy (SEM) was used to investigate the cryofractured morphology of pristine ABS and the ABS/CF, ABS/KF, and ABS/CF/KF composite filaments employed in this study. As shown in Figure 3, both the CFs and KFs are well dispersed and distributed within the ABS matrix (arrow 1). In Figure 3d,f, the small grayish cylinders, indicated by arrow 2, indicate the short CFs with an average diameter of 5.8 ± 0.5 μm oriented along the filament length [30]. In Figure 3e,f, the large stick-like fillers, indicated by arrow 3, are KFs. Note that the diameter of the short KFs is 12 ± 0.6 μm, twice that of the CFs. The fibers that were pulled out during the cryofracture process lead to some voids (black dots) in the ABS matrix. Most of the fibers demonstrate a good interface with the ABS matrix.

Effect of Raster Orientations
The flexural behaviors of pristine ABS and the composites of ABS/CF, ABS/KF, and ABS/CF/KF with different raster orientations are shown in Figure 4. In Figure 4a,b, the stress-strain curves of all the studied materials show a linear initial stage. At the end of this linear stage, the stress-strain curves display a nonlinear transition to a yielding state and then a viscoplastic state, followed by the rupture of the materials. The extracted flexural moduli for the composites are summarized in Table

Effect of Raster Orientations
The flexural behaviors of pristine ABS and the composites of ABS/CF, ABS/KF, and ABS/CF/KF with different raster orientations are shown in Figure 4. In Figure 4a,b, the stress-strain curves of all the studied materials show a linear initial stage. At the end of this linear stage, the stress-strain curves display a nonlinear transition to a yielding state and then a viscoplastic state, followed by the rupture of the materials. The extracted flexural moduli for the composites are summarized in Table 2. These results show that the flexural strengths and moduli of the composites increase substantially with the addition of a small amount of short CFs, regardless of the raster orientation (Table 2). Specifically, in Figure 4b, the flexural strength and modulus increase from 48. 38 16.7% and 58.7%, respectively. However, the ABS/CF composite exhibits lower elongation than pristine ABS for both the raster orientations in our investigated range (Figure 4a,b). With the reinforcement of the KFs, the ductility of the ABS/KF composite shows a significant improvement for both the cases with different raster orientations. However, the addition of the KFs obviously decreases the flexure strength of ABS. In the ABS/CF/KF composite, the decrease in the strength induced by the addition of the KFs could be compensated by the reinforcement of the CFs, resulting in a simultaneous enhancement of both the rigidity and ductility. The stress-strain curves for the materials with raster orientations of 0 • /90 • show lower ductility than those for the composites with raster orientations of ±45 • . In Figure 4c,d, the energy absorption properties of the materials as a function of strain show that the composites with raster orientations of ±45 • present higher energy absorption capabilities than the materials with raster orientations of 0 • /90 • at a given strain. The ABS/CF/KF composite show the highest flexural toughness obtained herein for both the cases with different raster orientations.
4a,b). With the reinforcement of the KFs, the ductility of the ABS/KF composite shows a significant improvement for both the cases with different raster orientations. However, the addition of the KFs obviously decreases the flexure strength of ABS. In the ABS/CF/KF composite, the decrease in the strength induced by the addition of the KFs could be compensated by the reinforcement of the CFs, resulting in a simultaneous enhancement of both the rigidity and ductility. The stress-strain curves for the materials with raster orientations of 0°/90° show lower ductility than those for the composites with raster orientations of ±45°. In Figure 4c,d, the energy absorption properties of the materials as a function of strain show that the composites with raster orientations of ±45° present higher energy absorption capabilities than the materials with raster orientations of 0°/90° at a given strain. The ABS/CF/KF composite show the highest flexural toughness obtained herein for both the cases with different raster orientations.   SEM images of the fractured surfaces of the ABS-based composites with raster orientations of ±45 • after flexural tests are shown in Figure 5. Other SEM images for the fractured composites with raster orientations of 0 • /90 • are not shown here because they led to similar conclusions. In Figure 5, it can be observed that short CFs and KFs are aligned with the line deposition direction. Compared to pristine ABS, the increased strength and modulus of the ABS/CF composite could be attributed to the addition of increasingly rigid CFs as well as the good interfaces between fillers and matrix [31]. The decreased ductility of the ABS/CF composite could be explained by the brittle CFs, which demonstrate smooth fractured CF surfaces with little plastic deformation, while the fractured KF surfaces show dimples that are characteristic of microvoid coalescence with a large amount of plastic deformation [19,32]. As shown in Figure 5b,c, the short KFs present relatively rough wavy fractured surfaces with microvoids, suggesting a large amount of plastic deformation. Such phenomenon could be linked to the toughening effects of short KFs in the composites of ABS/KF and ABS/CF/KF [33,34]. As the ABS/CF/KF composite in the current study demonstrated a good balance between rigidity and ductility, as shown in Figure 4, the following study focuses on ABS/CF/KF composites. In the following sections, we analyze the cross-sectional graphs of both undeformed and deformed samples of the ABS/CF/KF composite built with raster orientations of 0 • /90 • and ±45 • .  SEM images of the fractured surfaces of the ABS-based composites with raster orientations of ±45° after flexural tests are shown in Figure 5. Other SEM images for the fractured composites with raster orientations of 0°/90° are not shown here because they led to similar conclusions. In Figure 5, it can be observed that short CFs and KFs are aligned with the line deposition direction. Compared to pristine ABS, the increased strength and modulus of the ABS/CF composite could be attributed to the addition of increasingly rigid CFs as well as the good interfaces between fillers and matrix [31]. The decreased ductility of the ABS/CF composite could be explained by the brittle CFs, which demonstrate smooth fractured CF surfaces with little plastic deformation, while the fractured KF surfaces show dimples that are characteristic of microvoid coalescence with a large amount of plastic deformation [19,32]. As shown in Figure 5b,c, the short KFs present relatively rough wavy fractured surfaces with microvoids, suggesting a large amount of plastic deformation. Such phenomenon could be linked to the toughening effects of short KFs in the composites of ABS/KF and ABS/CF/KF [33,34]. As the ABS/CF/KF composite in the current study demonstrated a good balance between rigidity and ductility, as shown in Figure 4, the following study focuses on ABS/CF/KF composites. In the following sections, we analyze the cross-sectional graphs of both undeformed and deformed samples of the ABS/CF/KF composite built with raster orientations of 0°/90° and ±45°. The crack initiation and propagation in the composite of ABS/CF/KF with different raster orientations and build directions during the bending test are illustrated in Figure 6. With increasing strain up to 0.027, microcracking initiates and white cracks appear at the bottom for all specimens. At a strain of 0.038, cracks eventually climb along the loading directions and additional cracks are created. Specifically, after crack initiation, the cracks of the composites with a ±45° orientation propagate along the initial crack direction with a shorter distance than those of the composites with a 0°/90° orientation at a given strain. Finally, the composites of ABS/CF/KF with a raster orientation of 0°/90° fracture with a planar cross section, while those with a raster orientation of ±45° exhibit a very jagged fractured surface. The differences in the crack propagation behavior of the composites may be due to the different relationships between the layers and crack planes in the present work. As schematically presented in Figure 7a, in the case of a raster orientation of 0°/90°, the 0° layers acting as strong barriers are pulled under tension along the filaments that are perpendicular to the propagation plane, delaying crack propagation [35]. However, the 90° layers are parallel to the propagation plane. The breakage of interfilaments creates paths that facilitate the propagation of cracks [36]. Therefore, only half of the filaments for the composites could resist crack propagation, resulting in a relatively low fracture toughness [37]. In the case of a raster orientation of ±45°, all filaments have an angle of 45° against the loading direction. The filaments act as strong barriers that retard the crack propagation along the weak interfilament interfaces. The cracks propagate through The crack initiation and propagation in the composite of ABS/CF/KF with different raster orientations and build directions during the bending test are illustrated in Figure 6. With increasing strain up to 0.027, microcracking initiates and white cracks appear at the bottom for all specimens. At a strain of 0.038, cracks eventually climb along the loading directions and additional cracks are created. Specifically, after crack initiation, the cracks of the composites with a ±45 • orientation propagate along the initial crack direction with a shorter distance than those of the composites with a 0 • /90 • orientation at a given strain. Finally, the composites of ABS/CF/KF with a raster orientation of 0 • /90 • fracture with a planar cross section, while those with a raster orientation of ±45 • exhibit a very jagged fractured surface. The differences in the crack propagation behavior of the composites may be due to the different relationships between the layers and crack planes in the present work. As schematically presented in Figure 7a, in the case of a raster orientation of 0 • /90 • , the 0 • layers acting as strong barriers are pulled under tension along the filaments that are perpendicular to the propagation plane, delaying crack propagation [35]. However, the 90 • layers are parallel to the propagation plane. The breakage of interfilaments creates paths that facilitate the propagation of cracks [36]. Therefore, only half of the filaments for the composites could resist crack propagation, resulting in a relatively low fracture toughness [37]. In the case of a raster orientation of ±45 • , all filaments have an angle of 45 • against the loading direction. The filaments act as strong barriers that retard the crack propagation along the weak interfilament interfaces. The cracks propagate through irregular zigzag paths, increasing the fracture toughness by dissipating the fracture energy over a larger volume (Figure 7b). irregular zigzag paths, increasing the fracture toughness by dissipating the fracture energy over a larger volume (Figure 7b).   irregular zigzag paths, increasing the fracture toughness by dissipating the fracture energy over a larger volume (Figure 7b).

Effect of Building Directions
The flexural behaviors of the ABS/CF/KF composites with different build directions and raster orientations are shown in Figure 8a. The stress-strain curves for the composites with a side build Polymers 2019, 11, 1878 9 of 12 direction show higher initial linear slopes than those for the composites with a horizontal build direction. As shown in Table 3, moduli of the ABS/CF/KF composites horizontally printed at 0 • /90 • and ±45 • increased from 2167.17 ± 36.69 MPa and 2110.08 ± 29.79 MPa to 2683.33 ± 36.17 MPa and 2513.33 ± 43.12 MPa for the composites with a side-build direction, representing increases of 23.8% and 19.1%, respectively. In addition, the flexural strengths of composites with a side build direction showed a significant improvement in both the raster orientations in our studied ranges. The ultimate flexural strain of the composite with a side build direction and a raster orientation of ±45 • is much higher than that of the composite with a horizontal build direction and with the same raster orientation. (Figure 8 (a)). In Figure 8 (b), the energy absorption properties of the materials as a function of strain show that the composites with a side build direction presented higher energy absorption capabilities than those for the materials with a horizontal direction at a given strain. The composites with a raster orientation of ±45 • and side build direction showed the highest energy absorption capacity obtained in this work. The flexural behaviors of the ABS/CF/KF composites with different build directions and raster orientations are shown in Figure 8a. The stress-strain curves for the composites with a side build direction show higher initial linear slopes than those for the composites with a horizontal build direction. As shown in Table 3, moduli of the ABS/CF/KF composites horizontally printed at 0°/90° and ±45° increased from 2167.17 ± 36.69 MPa and 2110.08 ± 29.79 MPa to 2683.33 ± 36.17 MPa and 2513.33 ± 43.12 MPa for the composites with a side-build direction, representing increases of 23.8% and 19.1%, respectively. In addition, the flexural strengths of composites with a side build direction showed a significant improvement in both the raster orientations in our studied ranges. The ultimate flexural strain of the composite with a side build direction and a raster orientation of ±45° is much higher than that of the composite with a horizontal build direction and with the same raster orientation. (Figure 8 (a)). In Figure 8 (b), the energy absorption properties of the materials as a function of strain show that the composites with a side build direction presented higher energy absorption capabilities than those for the materials with a horizontal direction at a given strain. The composites with a raster orientation of ±45° and side build direction showed the highest energy absorption capacity obtained in this work.  As shown in Figure 6, the deformation phenomena of the composites with side build directions are different from those of composites with horizontal build directions. Specifically, even though the microcracks initiated at the bottom for all composites, they formed on the deposited layers for composites in a horizontal build direction rather than on the contour layers for composites with a side build direction. The multiple contour layers constructed with a raster orientation of 0° might offer an elevated resistance to bending due to the effective long raster lengths, resulting in elevated flexure moduli and strengths [38]. The raster direction also plays an important role in the crack propagation rate. The side build composites stretch to a strain of 0.038, the cracks extend to 1.82 ± 0.09 mm for the composites printed with the 0°/90° raster orientation and only extend to 1.12 ± 0.05 mm for the composites printed with the ±45° raster orientation, showing a different crack propagation velocity. The crack propagation of the composites with the ±45° raster orientation show a zigzag path, but the composite with the 0°/90° raster orientation exhibits a flat crack path. Similar to the composites with a horizontal build direction, the elevated fracture toughness for the  As shown in Figure 6, the deformation phenomena of the composites with side build directions are different from those of composites with horizontal build directions. Specifically, even though the microcracks initiated at the bottom for all composites, they formed on the deposited layers for composites in a horizontal build direction rather than on the contour layers for composites with a side build direction. The multiple contour layers constructed with a raster orientation of 0 • might offer an elevated resistance to bending due to the effective long raster lengths, resulting in elevated flexure moduli and strengths [38]. The raster direction also plays an important role in the crack propagation rate. The side build composites stretch to a strain of 0.038, the cracks extend to 1.82 ± 0.09 mm for the composites printed with the 0 • /90 • raster orientation and only extend to 1.12 ± 0.05 mm for the composites printed with the ±45 • raster orientation, showing a different crack propagation velocity. The crack propagation of the composites with the ±45 • raster orientation show a zigzag path, but the composite with the 0 • /90 • raster orientation exhibits a flat crack path. Similar to the composites with a horizontal build direction, the elevated fracture toughness for the composites with a raster orientation of ±45 • can be attributed to the zigzag path for crack propagation (Figure 7c,d).

Conclusions
In this work, we analyzed the mechanical and thermal properties of 3D-printed ABS-based composite specimens with different reinforcements (CFs and/or KFs). The samples were manufactured via a material extrusion 3D-printing technique using two different raster orientations and different build directions. The morphological structures at different tensile strains were investigated by digital camera and scanning electron microscopy (SEM). It was found that both CFs and KFs were well dispersed and distributed within the ABS matrix. Thermal testing showed that the introduction of CFs improved both the thermal conductivity and stability of the composites. The mechanical testing results revealed that the addition of short CFs increased the flexure modulus and strength, but decreased the ultimate flexure strain of the composites. However, the introduction of KFs increased the ductility of 3D-printed ABS/KF specimens because of the toughening effect induced by the short KFs. In addition, the ABS-based composites containing both carbon and Kevlar fibers simultaneously showed improved rigidity and ductility, which is suitable for energy absorption applications. Moreover, the deformation and failure modes of the composite specimens printed with different raster orientations indicated that the irregular zigzag crack propagation paths for the composites printed with a raster orientation of ±45 • delayed the breakage of the materials. The contour of the printed composites further postponed the initiation and propagation of cracks. The ABS/CF/KF composite specimens printed at a ±45 • raster orientation with a side build direction had the highest flexure toughness with an improved strength and ductility among all specimens investigated in this study.