Tensile properties, void contents, dispersion and ... - SAGE Journals

0 downloads 0 Views 3MB Size Report
Tensile properties, void contents, dispersion and fracture behaviour of 3D printed carbon nanofiber reinforced composites. Easir Arafat Papon and Anwarul ...
Original article

Tensile properties, void contents, dispersion and fracture behaviour of 3D printed carbon nanofiber reinforced composites

Journal of Reinforced Plastics and Composites 2018, Vol. 37(6) 381–395 ! The Author(s) 2018 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0731684417750477 journals.sagepub.com/home/jrp

Easir Arafat Papon and Anwarul Haque

Abstract In this paper, additive manufacturing technology has been used in processing carbon nanofiber (CNF) reinforced thermoplastic composites through fused deposition modeling (FDM). The effects of nanofiber concentrations, nozzle geometries on void contents, and tensile properties of FDM printed nanocomposites have been studied. The contact angles and void geometries have been characterized as a function of bead spreading orientations. Such measurements were carried out using micrographs of optical and scanning electron microscopy (SEM) examinations. The dispersion and orientation of nanofiber in polylactic acid (PLA) matrix have been studied using transmission electron microscopy observations. Finally, the fracture surfaces of both neat PLA and CNF/PLA nanocomposites have been investigated in order to understand the failure mechanism. The results show improved modulus, strength, and strain up to elastic limit in CNF/PLA nanocomposites in comparison to neat PLA matrix. The square nozzle geometry shows increased tensile strength and reduced void geometry in comparison to circular shaped nozzle. Three types of voids such as inter-bead, intra-bead, and interfacial bead were observed with total voids ranging from 24% to 30% with circular nozzle. The level of void content is reduced to about 7% for square nozzle. Contact angles and bead orientation play an important role in forming void contents. CNF/PLA nanocomposites exhibit decreased contact angle and higher void contents in comparison to neat PLA resin. Transmission electron microscopy study of CNF/PLA nanocomposites show uniform dispersion of carbon nanofibers with aligned orientation in both composite filament and three-dimensional printed specimen. Fractured surfaces of CNF/PLA nanocomposites show fiber pullout mechanism and comparatively coarse surface topography indicating higher fractured energy than neat polylactic acid resin. Keywords Additive manufacturing, composites, voids, contact angles, tensile properties

Introduction Additive manufacturing (AM) technologies have been an important area of research in processing composites with higher structural integrity in comparison to traditional plastics. It is a process of joining materials layer by layer to build an object from a computer-aided design (CAD) model.1–3 In recent years, many AM methods have been developed for composites but most of them require high-cost equipment.4 Fused deposition modeling (FDM)5–7 which produces functional parts by sequential deposition of fused materials is considered to be a cost-effective and most widely used AM technique due to its process simplicity in comparison to others such as stereolithography

(SLA), selective laser sintering (SLS),8 and laminated object manufacturing (LOM).9 In general, FDM is used for printing prototype plastic parts in a costeffective manner but recently it shows great potentials in fabricating stronger fiber (micro and nano-sized) reinforced composite materials with potential

Department of Aerospace Engineering and Mechanics, The University of Alabama, Tuscaloosa, AL, USA Corresponding author: Anwarul Haque, The University of Alabama, Tuscaloosa, Box 870280, Tuscaloosa, AL 35406, USA. Email: [email protected]

382 application in aerospace, automotive, and biomedical applications.10–12 The future thrust is to move FDM technology from prototype parts production to actual products. In FDM, a material filament is fed into a liquefier using a pinch roller mechanism and feedstock is melted in a temperature controlled liquefier as shown in Figure 1. The incoming filament acts as a piston to push the melt through a print nozzle whose diameter is in the order of a few millimeters. A support control system moves the nozzle in the x–y plane as the material is deposited on a build surface. The build surface is moved in vertical z-direction via another motor control mechanism to print complex three-dimensional (3D) design prototype parts.13 Usually, thermoplastic material is used as feedstock for FDM and the most commonly used thermoplastics are acrylonitrile butadiene styrene (ABS), polycarbonate (PC), polylactic acid (PLA), and polyamide (PA).13 However, the final products made of pure thermoplastics have limited mechanical properties and lack of strength and stiffness in producing fully load-bearing parts. Therefore, to make this technology appropriate for producing functional and applicable parts, FDM protocols need to be modified for composite materials. Recently, FDM has been used to make composite materials by adding reinforcing materials like glass fibers,4 carbon fibers,14,15 carbon nanotubes,16,17 carbon nanofiber (CNF),18 nano-clays,19 and graphene20,21 nano-platelets into the thermoplastic resin materials. It is to be noted that both micron-sized short fibers and nano-sized reinforcing materials are used in these studies. Ning et al.,1,14 has observed an increased ultimate strength (24% with 5 wt.% CF) and increased stiffness (32% with 7.5 wt.% CF) in short carbon fiber reinforced ABS composites. Although

Figure 1. Schematic of fused deposition modeling process.

Journal of Reinforced Plastics and Composites 37(6) yield strength, toughness, and ductility of CF/ABS composites are not seen to be promising in comparison to neat ABS. Tekinalp et al.,15 also studied FDM processed CF/ABS short fiber reinforced composites and reported comparable tensile strength and modulus in relation to traditional compression molding process. Although large porosity of around 20% were reported in short fiber reinforced FDM specimens. Improved tensile strength, modulus, and thermal stability have also been reported in FDM processed CNF/ABS, carbon nanotube/ABS, and montmorillonite/ABS nanocomposites in comparison to neat ABS resin.16– 19,22,23 Chuang et al.,24 have studied mechanical properties of FDM processed Ultem-1000 and Ultem 9085 resin-based composites and quantified porosity which ranges from 25% to 31%. Recently, Klift et al.,25 have studied about printing the continuous carbon fiber reinforced thermoplastics and performed tensile testing on FDM processed specimens. It appears that FDM investigations have been carried out with both micron-sized and nano-sized reinforcing materials mostly with ABS resin but similar studies with PLA resin are rarely observed in open literature. Also, the effects of nozzle geometry and bead orientations in tensile properties and void contents suggest further investigation. The properties of FDM printed components still lack qualification in comparison to injection molding and compression molding parts due to higher void contents and dimensional instability. Void formations in FDM parts are mostly related to bead spreading orientations, contact angles, and thermodynamic parameters related to cooling and phase transformation. Bead spreading architecture and contact angle are further related to the characteristic of melt flow and nozzle geometry. Thus, experimental investigations relating to void contents, nozzle geometry, and bead spreading architecture, dispersion and mechanical properties in FDM are important in understanding process, structure, and performance relationship. Such information is critical in developing an optimum FDM process model with less void contents and improved structural performance. In the present study, CNF/PLA nanocomposite filaments suitable for FDM process were produced using a laboratory mixing extruder. The effects of CNF concentrations and nozzle geometries on the tensile properties and void contents of 3D printed PLA and CNF/PLA nanocomposites have been studied. Bead spreading geometries, contact angles, void geometry, fracture, and dispersion are extensively studied as a function of CNF concentration, layer orientation and nozzle geometry using optical, scanning, and transmission electron microscopic examinations.

Papon and Haque

Experimental work Materials Polylactic acid-biopolymer (PLA) granulates supplied by Goodfellow Cambridge Ltd. with nominal size of 3 mm, glass transition temperature of 55 C and melting temperature 175 C were used as matrix material. CNF with a maximum length of 400 nm and average diameter of 50 nm was used as reinforcing material in making CNF/PLF composite filament suitable for FDM.

Manufacturing The complete experimental set-up of filament manufacturing process for FDM and 3D printing of a carbon nanofiber reinforced polymer nanocomposites (CFRP) coupon are shown in Figure 2. It consists of a two-step process, (1) making a filament with a constant diameter suitable for FDM and (2) printing the tensile test coupon using the filament. Details about the filament processing method and printing tensile specimens are described below.

Filament preparation A laboratory mixing extruder CSI-LE-075, a take-up apparatus CSI-194T and mini pelletizer CSI-194C (Figure 2) were used in processing CNF/PLA composite filament. PLA pellets and CNF powder were mixed in the hopper with different CNF weight percentages (0.5 wt.% and 1 wt.%) which then fed into the extruder melting chamber. A temperature range of 165 C–185 C was used for melting PLA. CNF and PLA powder were shear mixed in space between a stationary and rotating

383 rod through a gap control mechanism. Extruder speed was set between 50% and70% of its maximum range. A take-up apparatus speed was set at 30%–40% of its maximum speed for maintaining filament diameters in the range of 1.70–1.80 mm. The composite filaments are then passed through the orifice of a mini pelletizer to get the CNF/PLA pellets. After pelletizing it twice, the small pieces of filament were fed into the extruder hopper to make the filaments with more homogenous distribution of carbon nanofibers. The process was repeated twice and a well dispersed CNF reinforced polymer composite filaments with a constant diameter of 1.8 mm were produced for feeding into 3D printer to print tensile specimen.

Printing specimens The MakerBot Replicator 2X 3D printer based on FDM principle was used to make CNF/PLA tensile coupons. A tensile test specimen was designed using a CAD software based on ASTM D638–10 designation. Afterward, the file was transferred as a .slt file to MakerBot Desktop software. During the printing process, it was necessary to set and monitor the fabrication process parameters. The nozzle diameter of the liquefier section was 0.4 mm. The liquefier nozzle and build plate temperatures were set at 230 C and 110 C, respectively. There were nine layers in each specimen and the deposition orientation were ½45 ; 45 ; 45 ; 90 ; 0 s respectively. A printing velocity was set at 1.2 mm/s for the first layer and 1.5 mm/s for the subsequent layers. Three sets of FDM manufactured specimens were printed such as neat PLA, 0.5% CNF/PLA, and 1% CNF/PLA.

Figure 2. FDM process for CFRP nanocomposite specimen. FDM: fused deposition modeling; CFRP: carbon nanofiber reinforced polymer nanocomposites.

384

Journal of Reinforced Plastics and Composites 37(6)

Test methods

Results and discussion

The tensile test was conducted using an MTS QTest10 machine under displacement control according to ASTM D638–10 standard.26 The crosshead speed was maintained at 5 mm/min and the distance between two grips was 115 mm. Four specimens were tested in each category. The load vs. displacement plots were recorded using a data acquisition system for further analysis. The effect of carbon fiber content on the tensile properties such as strength, modulus, and failure strain was studied using standard stress vs. strain plots.

The tensile properties of all the 3D printed specimens with different CNF concentration were determined. The void contents, the geometry of voids and bond formation quality were investigated using both optical and SEM examinations. An attempt has been taken to identify the variations in void contents, void geometry, and contact angles with different lay-up sequence and fiber concentrations. Fractured surfaces of tensile test specimens were observed in SEM in order to identify failure modes. Finally, TEM have been conducted in both extruded filament and 3D printed single bead to observe size, orientation, and dispersion of CNF in PLA matrix.

Optical and electron microscopic examination The samples were cut for optical, scanning, and transmission electron microscopic examinations. An optical microscope (OM) manufactured by Olympus and connected to a desktop computer was used to quantify void contents. QCapture Pro, a scientific imaging software was used to measure length in the micrometer range. OM samples were cut transverse to loading direction and the surface was grinded using sandpapers ranging 320–1200 grits. The polishing was carried out using a polishing disc with one-micron diamond particles. Both the polished and fractured samples were viewed under scanning electron microscope (SEM). Polished samples were examined to approximate void geometries and contact angles. The fractured samples were studied to understand the fracture type and fiber pullout. SEM micrographs of the surfaces were taken with Hitachi SU3500 SEM applying an acceleration voltage of 5 kV. To eliminate charging, samples were coated with thin conductive film before placing into the high-vacuum SEM chamber. In this case, the samples were coated with gold/palladium using Hummer 6.6 Sputter coater. The coating was approximately 5 nm thick and overall coating process continued for 120 seconds with an applied current of 15 mAmp. To study CNF dispersion, size and orientation in PLA matrix, transmission electron microscopic (TEM) examinations were carried out using a Hitachi TEM H-7650. Two different types of samples (0.5% CNF/PLA and 1% CNF/PLA) taken from extruded filament and single bead spread in each nanocomposites were prepared for TEM observation. The samples were cut at a speed of 0.6 mm/s using a Leica EM UC6 ultramicrotome. Thickness of each section was 90 nm. The cut sections were then placed in EMS2010-Cu grids (size 2 mm1 mm) and then transferred to a 0.5% FORMVAR coated copper slot grid. Two magnifications (8000 and 25,000) with an acceleration voltage of 60 kV were used for CNF quantification.

Tensile properties Figure 3 shows the stress vs. strain plots of neat PLA and PLA/CNF nanocomposites with 0.5% and 1% CNF concentrations. The plots shown in Figure 3 are very close to the average data set among four test specimens in each category. All the plots show almost linear trend up to very high stress level and then become nonlinear prior to final failure. Failure strength and Young’s modulus of 0.5% CNF/PLA nanocomposites are seen to be higher in comparison to neat PLA resin. But the failure strain of neat PLA is observed to be higher than CNF/PLA nanocomposites. Figure 4 (a) to (c) shows plots of tensile properties as a function of CNF concentrations. The plots show increased Young’s modulus as CNF concentrations are increased from 0.5% to 1%. The modulus is seen to be, 1822 MPa (12% enhancement) for the nanocomposite with 1% CNF whereas the modulus of neat PLA is observed to be, 1670 MPa. An increased trend in yield strength is also observed for nanocomposites with 0.5% and 1% CNF concentration. But the yield strength

Figure 3. Stress–strain plots for different CNF contents. CNF: carbon nanofiber.

Papon and Haque

385

Figure 5. Micrographs showing effects of lay-up sequence in void geometry (a) PLA, (b) 0.5% CNF/PLA and (c) 1% CNF/ PLA. CNF: carbon nanofiber; PLA: polylactic acid.

CNF/PLA nanocomposites with 0.5% and 1% CNF show decreased failure strains such as 3% and 3.32% in comparison to 3.95% observed for neat PLA resin. Figure 4. (a) Modulus vs. CNF contents. (b) Tensile strength vs. CNF contents. (c) Strain vs. CNF contents. CNF: carbon nanofiber.

enhancement in nanocomposites is not seen to be significant as compared to Young’s modulus. The ultimate strength is also observed to be increased with 0.5% CNF but it showed a decreased value with 1% CNF. This is possibly due to the high percentage of CNF agglomeration at 1% CNF concentration.

Void contents The printed specimens were studied through optical microscopic examinations in order to quantify the void contents and identify the void geometries as a function of CNF concentrations and lay-up sequence. Void contents in FDM are functions of many parameters such as liquefier nozzle size and shape, melt flow characteristics like pressure and velocity gradients, bead orientations and thermodynamic parameters

386 such as bed temperature and cooling rate related to melt solidifications. These factors contribute primarily to form three types of voids: inter-bead, intra-bead, and interfacial voids. Inter-bead voids are observed between individual beads resulting primarily from bead spreading architecture. The shape and size of these inter-bead voids may vary at different locations depending upon bead orientation, shape, size and thermodynamic parameters related to bead solidification. It is to be noted that bead shape and size are further related to nozzle geometry and melt flow characteristics. The interfacial voids are usually seen at poor bonding regions between individual beads. The bonding between the individual bead of the same layer and of neighboring layer is driven by the thermal energy of the semi-molten material. Thus, the temperature history of successive layers is important in determining the bond quality. Once the material is deposited and comes in contact with a previously deposited layer, the interface temperature should remain above the material glass transition temperature prior to bonding. Again, non-uniform melt flow during bead spreading also results in interfacial voids. The variations in any

Journal of Reinforced Plastics and Composites 37(6) required thermodynamic parameters such bed temperature and cooling rate related to melt solidification may result in interfacial voids. The intra-bead voids are observed inside the individual beads and these voids mostly relate to melt flow characteristics and melt solidification parameters. These intra-bead voids play an important role in developing crack initiation inside the bead due to stress concentration. Figure 5(a) to (c) shows micrographs of polished FDM printed specimens of neat PLA, 0.5% CNF/PLA and 1% CNF/PLA nanocomposites. Each specimen consists of nine layers with bead orientations [þ45 =  45 = þ 45 =90 =0 =90 = þ 45 =  45 = þ 45 ]. The OM micrographs in Figure 5(a) to (c) are taken from the same location of each specimen containing all the nine layers. Figure 5(b) and (c) represents CNF reinforced PLA nanocomposites samples which show solid dark regions as deposited layers and white regions as voids. It is to be noted that the nanocomposites with CNF reinforcement develop dark black contrast. As a result, the void contents in such specimens are not easily detectable from OM micrographs. So, the image surface of CNF reinforced specimen were embedded

Figure 6. Schematic of inter-bead void geometry and correction factors: (a) Triangular void (b) Diamond shaped void (c) Correction factor.

Papon and Haque

387

with white powder particles and later cleaned in order to detect voids in OM micrographs. It is observed in Figure 5(a) to (c) that 0 layer is sandwiched between two 90 layers at the midsection followed by þ45 ,45 , þ45 layers at the top and bottom. The geometry and size of inter-bead voids vary at different locations due to a mismatch in bead layer orientations. Mainly two types of inter-bead void geometries similar to diamond and triangular shaped were observed in the

2D representation. These inter-bead void contents were quantified using a scaled grid pattern marker attached to the digital OM. Figure 6(a) to (c) shows schematic views of observed inter-bead voids which do not exactly resemble perfect triangle or diamond-shaped pores. As a result, a correction factor was introduced after physical measurement of the corner points of the triangles and diamonds using a digitized scale. Figure 6(c) shows a schematic of the correction factor region which

Table 1. Inter-bead void contents in F DM printed specimens.

Sample

Specimen surface area (mm2 )

Largest void (mm2 )

Smallest void (mm2 )

Void contents without correction factor ()

Void contents with correction factor (15%)

Neat PLA 0.5% CNF/PLA 1% CNF/PLA

12 12 12

0.05 0.05 0.06

0.0025 0.0034 0.0036

18.46% 20.76% 24.37%

15.70% 17.70% 20.70%

CNF: carbon nanofiber; FDM: fused deposition modeling; PLA: polylactic acid.

Figure 7. Micrographs of interfacial bead voids and interfacial bonding. (a) Schematic of interfacial bead voids. (b) SEM micrograph of top layer. (c) Micrograph of interfacial bonding (enlarged view of boxed section in b). SEM: scanning electron microscopy.

388

Journal of Reinforced Plastics and Composites 37(6)

Figure 8. Schematic of inter-bead void geometry as a function of lay-up sequence.

Figure 9. SEM micrograph of inter-bead void geometries as a function of lay-up sequence. SEM: scanning electron microscopy.

Figure 10. Contact angle representation as a function of lay-up sequence. (a) Triangular void. (b) Diamond void.

Papon and Haque

389

was calculated after subtracting a triangular area from the relevant quarter circle. The area of the quarter circle is easily determined from the bead diameter approximated from nozzle geometry. This correction factor is seen to be approximately 15%. Table 1 shows approximate inter-bead void size and percentage of void contents in PLA, 0.5% CNF/PLA, and 1% CNF/PLA specimens. The micrographs of the specimens show inter-bead void contents in the range 18%–24% without correction factor but they are reduced to 16%–21% with correction factor. The results show some enhancement in inter-bead pore contents in CNF/PLA nanocomposites in comparison to neat PLA. Similar results are also reported by some investigator for short fiber reinforced composites.15 Intra-bead voids were observed in Figure 5(a) to (c) as small black dots in PLA bead and white dots in CNF/PLA nanocomposite beads. The resolution of these intra-bead voids was not very clear in OM micrographs but it was estimated approximately in the range of 3%–4%. Such quantification was also verified in SEM micrographs shown in Figure 12(d). Finally, interfacial voids between beads due to poor bonding are clearly shown in Figure 7(a) and (b). Layers with 45 or 45 orientation show better bonding in comparison to 90 bead orientations. It is not seen like a channel all along the bead path due to prescribed orientation. Such interfacial voids are approximated in the range 3%–4%. Figure 7(c) shows area with good interfacial bonding. Finally, the results of this study show a total 22% and 29% void contents on FDM printed neat PLA and CNF/PLA nanocomposites, respectively. It is observed that inter-bead void is significantly high (21%) in comparison to intra-bead (4%) and interfacial (4%) voids in FDM processed nanocomposites. Although such high volume of intra-bead voids is not considered to be as critical in reducing load-bearing capacity like intra-bead voids. Eventually, overall loadbearing capacity of FDM samples should not be very detrimental if specific strength and specific stiffness are considered while comparing with compression and injection molded parts.

Void geometry and contact angle The deposited bead geometry and contact angle play an important role in inter-bead void formation. These parameters further depend on many factors such as melt flow characteristics, bed spreading orientations, and bead temperature. A breakdown of layer-wise print orientation and schematic of corresponding predicted bead cross section are shown in Figure 8. The geometry of these beads influences contact angle and void geometry within the specimen. There are three types of bead lay-up sequence found in each specimen like ð45 =  45 Þ; ð45 =90 Þ, ð90 =0 Þ and vice versa. Depending on predetermined bead lay-up sequence two types of inter-bead void geometry approximately resembles to triangle and diamond shape were observed. Diamond shaped voids were visible in the upper and lower regions whereas triangular shaped voids were observed in the midsection. Diamond shaped voids were formed when successive printed layer orientations were 45 =  45 or 45 =45 and triangular shaped voids were observed when orientation of successive layers was 0 =90 ,90 =0 , 45 =90 , and 90 =45 orientations. A schematic representation of predicted triangular and diamond-shaped (approximate) inter-bead voids and SEM micrographs of actual inter-bead void geometries are shown in Figures 8 and 9. A very close resemblance of predicted and experimental observation of inter-bead void geometry are clearly observed in Figures 8 and 9. The results of this study indicate that 45 =90 and 90 =0 orientations are preferred over 45 =  45 orientation due to reduced void contents resulting from triangular void geometries. Figure 10 shows a schematic representation of contact angle (ht, hd) measurement for both approximate triangle and diamond-shaped voids. It is represented as the angle between the solid build surface and incoming molten drop. Initially, major axis and minor axis of 90 and 45 bead cross sections were measured using a digitized scale attached to the OM. Later, contact angles were directly measured using a scaled protractor from

Table 2. Contact angle variation with lay-up sequence and CNF percentage. Layer orientation 90

45

Parameters

PLA

0.5% CNF/PLA

1% CNF/PLA

PLA

0.5% CNF/PLA

1% CNF/PLA

Major axis (lm) Minor axis (lm) Contact angle (h)

390 335 128

381 337 125

381 334 128

519 320 153

665 319 138

660 321 134

CNF: carbon nanofiber; PLA: polylactic acid.

390 SEM micrographs shown in Figure 9 for both triangular and diamond-shaped voids. It is to be noted that the triangular shape voids are always formed with 90 bead orientation and diamond shape void are developed with 45 =  45 bead orientation. The data for the major and minor axis for both 90 and 45 bead geometry and corresponding contact angles for PLA, 0.5% CNF/ PLA, and 1% CNF/PLA are provided in Table 2. It is seen that void content size is reduced with increased contact angle based on contact angle notation used in this study. The contact angles for 45 bead is observed to

Journal of Reinforced Plastics and Composites 37(6) be higher than the corresponding value for 90 bead. The results mostly show that adding CNF in PLA reduce the contact angle. The void content data shown earlier also justify such trend of contact angle. Although data with 1% CNF/PLA nanocomposite show some discrepancy which may need further investigation.

Effects of nozzle geometry Effects of square nozzle have been investigated to study the geometry effect on the bead spreading architecture.

Figure 11. Effect of nozzle geometry on inter-bead voids. (a) Voids: circular nozzle-PLA. (b) Voids: square nozzle-PLA. (c) Voids: circular nozzle-0.5% CNF/PLA. (d) Voids: square nozzle-0.5% CNF/PLA. (e) Schematic of inter-bead void size for circular nozzle. (f) Schematic of inter-bead void size for square nozzle. CNF: carbon nanofiber; PLA: polylactic acid.

Papon and Haque

391

Table 3. Effect of nozzle geometry on mechanical properties for PLA. Nozzle geometry Bead orientation (degree) Young’s modulus (MPa) 0.2% offset yield strength (MPa) Ultimate tensile strength (MPa) Failure strength (MPa) Strain at yield (%) Strain at break (%)

Circular 

45 1994 21.50 29.50 30.50 1.28 2.10

Square 

45 2103 27.00 36.10 35.50 1.50 3.00

PLA: polylactic acid.

It was observed from Figure 11(a), (c) and (e) that the circular nozzle leads to elliptical cross-sectional beads resulting higher inter-bead void contents. It is to be noted that a square cross-sectional bead geometry would reduce or remove the inter-bead voids in the specimen and improve the mechanical properties. As a result, a nozzle geometry is desirable to achieve an extruded bead with square cross-section. To determine the effect of the nozzle geometry on bead formation, a square-shaped nozzle have been used and compared to a circular nozzle (0.5 mm dia.). The square nozzle used in this study was prepared in-house as a replacement of the circular nozzle. The geometry effect on the mechanical properties and void contents are studied on the CNF/PLA specimens for both square and circular nozzle. Each specimen were printed with 100% infill density consisting of 10 layers with 45 bead orientation. The other printing setup (layer height, layer thickness, layer orientation, print temperature, bed temperature) were maintained identical for both circular and square nozzle. Table 3 shows comparison of mechanical properties for PLA printed using circular and square shaped nozzle. An increasing trend of tensile properties are observed for square nozzle. Young’s modulus is found to be improved by almost 5.5%; whereas, yield strength, ultimate tensile strength, and failure strength are increased by almost 25.5%, 22. 4%, and 16%, respectively. The samples were examined under the microscope to study bead bonding and inter-bead voids. The OM micrographs shown in Figure 11 are taken from the same location of each specimen. It is observed that the shapes and sizes of inter-bead voids vary significantly for circular and square nozzle. Circular nozzle (0.5 mm dia.) is seen to develop almost 20% total voids (inter-bead, intra-bead, and interfacial) which is observe to be almost same as calculated earlier with the 0.4 mm circular nozzle. The number of total voids for square nozzle is found to be much less with

comparatively smaller void size as shown in Figure 11 (b) and (c). In Figure 11(b), black dots represent the inter-bead voids; whereas, white dots represent the same in Figure 11(d). The inter-bead voids developed by square nozzle varies in sizes and shows mostly triangular shape with very few in diamond shape. The approximated inter-bead void areas are almost 2% of the entire area. The total void percentage including intra-bead and interfacial voids for square nozzle is quantified to be approximately 7% which is significantly less than the same developed by circular nozzle. Moreover, the melt deposition volume is observed to be larger for square nozzle. The high melt volume smoothed out the top surface of each deposited bead allowing almost a flat surface for the next deposited layer. Again, the flat-like surface area of the deposited bead lead to higher contact angle. As the contact angle is increased, the successive bead contact surface is also improved. This led to better bead bonding among the subsequent layers. Therefore, the geometric shape of each bead cross section is not clearly visible from the OM image. It is to be noted that, the quantitative values of mechanical properties slightly varies from previously reported data. This is mostly due to use of different nozzle geometry and layer orientations.

Fracture analysis Figure 12 shows the SEM image of the fractured surface of neat PLA, 0.5% CNF/PLA, and 1% CNF/PLA nanocomposites. The micrograph of neat PLA clearly shows very smooth fractured surface in 0 bead at the midsection in Figure 12(a). Interfacial fractured surface of two 90 beads surrounding 0 bead show signs of good bonding. It is to be noted that 45 PLA bead also shows a smooth fractured surface with few undulation due to orientation effect. In contrast, fractured surface of both 0.5% CNF/PLA and 1% CNF/PLA show coarse topography with a larger extent of undulation in Figure 12(b) to (d). The presence of CNF pullout in fractured surface of nanocomposites was also observed in Figure 12(d). This explains extra fracture energy required in CNF/PLA nanocomposites in comparison to neat PLA. The presence of inter-bead and intra-bead voids were also observed in fractured surface shown in Figure 12(a) to (d).

Dispersion and orientation Figures 13 and 14 show the TEM images of 0.5% CNF/ PLA and 1% CNF/PLA nanocomposites for both extruded and FDM printed specimens at two magnifications. The cross sections shown in the images are apparently perpendicular to filament and bead

392

Journal of Reinforced Plastics and Composites 37(6)

Figure 12. Fracture surface. (a) Neat PLA, (b). 0.5% CNF/PLA, (c) enlarged view of boxed section in b, (d). 1% CNF/PLA (e). CNF pullout and intra-bead voids. CNF: carbon nanofiber; PLA: polylactic acid.

deposition direction. The TEM samples were cut by a microtome with a thickness of 90 nm. A uniform dispersion of CNF was clearly evident in all the specimens as shown in Figures 13(a) and (b) and 14 (a) and (b). The results show no significant variation in CNF distribution between extruded and FDM printed single bead specimens. A larger concentration of CNF distribution is clearly observed in 1% CNF/PLA specimens (Figures 14 a-b) in comparison to 0.5% CNF/PLA specimens

(Figures 13 a-b). The CNF distribution mostly shows transverse view rather than longitudinal sections. It is seen in Figures 13(c) and (d) and 14(c) and (d) that CNF fibers are mostly aligned in extruded and bead orientation directions. This indicates that extruder orifice and printer nozzles assist in orienting the fibers. Such control of CNF orientation is beneficial in selecting bead layup sequence for improved performance. The crosssection of individual single fiber diameter is seen to be

Papon and Haque

393

Figure 13. CNF dispersion in PLA matrix for 0.5% CNF/PLA nanocomposite samples (a). extruded filament (b) 3D printed single bead (c) higher magnification of extruded filament (d) higher magnification of samples from 3D printed single bead. CNF: carbon nanofiber; PLA: polylactic acid.

Figure 14. CNF dispersion in PLA matrix for 1% CNF/PLA nanocomposite samples (a) extruded filament (b) 3D printed single bead (c) higher magnification of extruded filament (d) higher magnification of samples from 3D printed single bead. CNF: carbon nanofiber; PLA: polylactic acid.

394 around 40 nm in Figure 13(d). A misaligned single fiber is also seen in longitudinal orientation in Figure 14(c) and (d) which indicates approximate CNF length equals to 400 nm. A number of similar misaligned fibers and few fiber agglomeration are also evident in Figures 13 and 14.

Conclusion 1. CNF/PLA nanocomposites with 0.5% CNF show improved modulus of elasticity and ultimate strength in comparison to neat PLA. Although such improvement was not observed in failure strain. The factors such as agglomeration, low CNF aspect ratio, and poor interfacial bonding between CNF and PLA are critical factors for further improvement in tensile properties of CNF/PLA nanocomposites. 2. Three types of void such as inter-bead (21%), intrabead (4%), and interfacial (4%) voids were observed in FDM printed nanocomposites. Such level of void content is significantly high in comparison to the specimen prepared through compression molding and injection molding process. 3. SEM micrographs reveal two dominant void geometries approximately resemble to triangular and diamond configuration. It is observed that bead lay-up orientation plays an important role in shape and size of void contents. The results show similar to triangular voids with 0 =90 , 90 =45 bead orientations whereas 45 =  45 bead orientation develops close to diamond shape void. 4. The contact angle of neat PLA shows higher values resulting reduced void size in comparison to CNF reinforced nanocomposites. Bead orientation also influences contact angles which eventually effects void size and geometry. The contact angle is seen to be comparatively higher for angle ply (45 =  45 ) bead than cross-ply 0 =90 bead. 5. The geometry of the nozzle significantly influences the bead architectures, void contents, and mechanical properties. It is seen that the nozzle with square geometry develops significantly low inter-bead voids and improved mechanical properties in comparison to circular nozzle. 6. The fractured surface of neat PLA is seen to be smooth in comparison to CNF/PLA nanocomposites. The SEM micrograph of bead cross-section shows sign of CNF pullout mechanism and coarse fractured surface indicating higher fracture energy in CNF/PLA nanocomposites specimen. 7. TEM investigation reveals uniform dispersion and aligned orientation of CNF in both extruded and 3D printed specimen. The extruder orifice and

Journal of Reinforced Plastics and Composites 37(6) printer nozzle assist in orienting CNF in extrusion and bead orientation direction. Acknowledgements The authors would like to thank Dr Kim Lackey for her support in conducting SEM and TEM inspections at Optical Analysis Facility, Department of Biological Sciences, The University of Alabama. The assistance of summer intern Mr Ivan Cesar from Brazil in tensile specimen preparation and testing is also acknowledged with appreciation.

Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research is sponsored by ALEPSCoR National Science Foundation (NSF) Grant (# 1158862) and ALEPSCoR GRSP Grant.

References 1. Ning F, Cong W, Qiu J, et al. Additive manufacturing of carbon fiber reinforced thermoplastic composites using fused deposition modeling. Compos Part B Eng 2015; 80: 369-378. 2. ASTM F2792-12a. Standard terminology for additive manufacturing technologies. West Conshohocken: ASTM International, 2012. 3. Villalpando L, Eiliat H and Urbanic RJ. An optimization approach for components built by fused deposition modeling with parametric internal structures. Procedia CIRP 2014; 17: 800-805. 4. Zhong W, Li F, Zhang Z, et al. Short fiber reinforced composites for fused deposition modeling. Mater Sci Eng A 2001; 301: 125–130. 5. Too MH, Leong KF, Chua CK, et al. Investigation of 3D non-random porous structures by fused deposition modeling. Int J Adv Manuf Technol 2002; 19: 217–223. 6. Carneiro OS, Silva AF and Gomes R. Fused deposition modeling with polypropylene. Mat and Des 2015; 83: 768–776. 7. Turner B, Strong R and Gold SA. A review of melt extrusion additive manufacturing processes: I. Process design and modeling. Rap Prototyp J 2014; 20: 192–204. 8. Kruth JP, Mercelis P, Vaerenbergh JV, et al. Binding mechanisms in selective laser sintering and selective laser melting. Rap Prototyp J 2005; 11: 26–36. 9. Mueller B and Kochan D. Laminated object manufacturing for rapid tooling and patternmaking in foundry industry. Comput Ind 1999; 39: 47–53. 10. Hofstatter T, Paderson D, Tosello G, et al. State-of-theart of fiber reinforced polymers in additive

Papon and Haque

11.

12.

13.

14.

15.

16.

17.

18.

manufacturing technologies. J Reinf Plast Compos 2017; 36: 1061–1073. Wang X, Jiang M, Zhou Z, et al. 3D printing of polymer matrix composites: a review and prospective. Compos Part B 2017; 110: 442–458. Zeina I, Hutmacher DW, Tan KC, et al. Fused deposition modeling of novel scaffold architectures for tissue engineering applications. Biomaterials 23: 1169–1185. Bellini A. Fused deposition of ceramics: a comprehensive experimental, analytical and computational study of material behavior, fabrication process and equipment design. PhD dissertation, Drexel University, Philadelphia, PA, 2002. Ning F, Cong W, Wei J, Wang S, et al. Additive manufacturing of CFRP composites using fused deposition modeling: Effects of carbon fiber content and length. In: Proceedings of the ASME 2015 International Manufacturing Science and Engineering Conference, 8 June 2015, pp. 1–7. Tekinalp HL, Kunc V, Velez-Garcia GM, et al. Highly oriented carbon fiber–polymer composites via additive manufacturing. Compos Sci and Tech 2014; 105: 144–150. Dudek P. FDM 3D printing technology in manufacturing composite elements. Arch Metall Mater 2013; 58: 1415–1418. Farahani RD, Dalir H, Le Borgne V, et al. Direct-write fabrication of freestanding nanocomposite strain sensors. Nanotechnology 2012; 23: 085502 Postiglione G, Natale G, Griffini G, et al. Conductive 3D microstructures by direct 3D printing of polymer/carbon

395

19.

20.

21. 22.

23.

24.

25.

26.

nanotube nanocomposites via liquid deposition modeling. Compos Part A Appl Sci Manuf 2015; 76: 110–114. Shofner ML, Lozano K, Rodrı´ guez-Macı´ as FJ, et al. Nanofiber-reinforced polymers prepared by fused deposition modeling. J Appl Polym Sci 2003; 89: 3081–3090. Compton BG and Lewis JA. 3D-printing of lightweight cellular composites. Adv Mater Weinheim 2014; 26: 5930–5935. Wei X, Li D, Jiang W, et al. 3D printable graphene composite. Sci Rep 2015: 5. Dul S, Fambri L and Pegoretti A. Fused deposition modelling with ABS–graphene nanocomposites. Compos Part A Appl Sci Manuf 2016; 85: 181–191. Gray IVRW, Baird DG and Bohn JH. Thermoplastic composites reinforced with long fiber thermotropic liquid crystalline polymers for fused deposition modeling. Polym Compos 1998: 19: 383–394. Chuang KC, Grady JE and Draper RD. Additive manufacturing and characterization of Ultem polymers and composites. In: CAMX conference proceedings, Dallas, TX, October 26–29, 2015. Klift FVD, Koga Y, Todoroki A, et al. 3D printing of continuous carbon fibre reinforced thermo-plastic (CFRTP) tensile test specimens. Open Journal of Composite 2015; 6: 18. ASTM D638-10. Standard test method for tensile properties of plastics. West Conshohocken: ASTM International; 2010.