Multi-scale interlaminar fracture mechanisms in woven composite laminates reinforced with aligned carbon nanotubes

Composites Science and Technology 100 (2014) 128–135

Contents lists available at ScienceDirect

Composites Science and Technology journal homepage: www.elsevier.com/locate/compscitech

Multi-scale interlaminar fracture mechanisms in woven composite laminates reinforced with aligned carbon nanotubes Sunny S. Wicks, Wennie Wang 1, Marcel R. Williams, Brian L. Wardle ⇑ Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge, MA 02139, United States

a r t i c l e

i n f o

Article history: Received 16 March 2014 Received in revised form 19 May 2014 Accepted 5 June 2014 Available online 13 June 2014 Keywords: A. Carbon nanotubes A. Laminate A. Hybrid composites B. Fracture toughness D. Fractography

a b s t r a c t Aligned nanoscale fibers have been shown to provide inter and intralaminar reinforcement of fiber-reinforced polymer composites. In one hierarchical architecture, aligned carbon nanotubes (CNTs) grown on advanced microfibers in a woven ply creates a ‘fuzzy fiber’ reinforced plastic (FFRP) laminate. Here, the mechanisms of Mode I fracture toughness enhancement for such laminates are elucidated experimentally by varying the type of epoxy and aligned CNT length. Reinforcement effectiveness is found to vary from reduced initiation toughness to over 100% increase in steady-state fracture toughness, depending upon the interlaminar fracture mechanisms. Fracture-surface morphology investigations reveal that interlaminar toughness enhancement is significantly less for an aerospace infusion resin than that for a much tougher hand lay-up marine epoxy. Long (20 micron) aligned CNTs add significant toughness in steady state (>1 kJ/m2 increase for marine epoxy) via CNT pullout and by driving the crack through tortuous paths around and through microfiber bundles/tows, whereas shorter CNTs produce less toughening (or even reduced toughness in aerospace epoxy), which is attributed to increased microfiber–matrix cohesive failure. These findings reveal the multi-scale nature of the aligned-CNT reinforced composite ply interface, and the mechanisms at work at the micro and nanoscales that influence the macroscopic behavior. These new insights provide impetus for using aligned CNTs to tailor microfiber polymer composites by adjusting microfiber orientation, steering cracks through tortuous paths, and maximizing fracture surface area through both microfiber and nanofiber pullout. Ó 2014 Published by Elsevier Ltd.

1. Introduction Aerospace structural components are increasingly being made of (micro)fiber reinforced plastics (FRPs) due to their excellent mass-specific mechanical properties. While properties in the plane of the microfibers are strong, stiff, and tailorable, the interlaminar region is relatively weak and compliant, leading to failure due to delamination and other damage modes. To address these deficiencies at the interlaminar interface, several architectures have been developed to enhance strength and/or toughness in the throughthickness direction of the laminate, including Z-pinning, stitching, and 3-D weaving [1–3]. These processes result in a significant reduction in in-plane properties of the laminate due to damage from insertion of through-thickness direction reinforcements that are hundreds of microns in diameter, and loss of microfiber volume fraction in the in-plane direction [1,4]. Alternative routes to increasing interlaminar toughness include modification of matrix ⇑ Corresponding author. Tel.: +1 617 252 1539. 1

E-mail address: [email protected] (B.L. Wardle). Current address: University of California, Santa Barbara, United States.

http://dx.doi.org/10.1016/j.compscitech.2014.06.003 0266-3538/Ó 2014 Published by Elsevier Ltd.

properties through added toughening agents, compliant or reinforced interlayers [5,6] or particles [7–9] and nanofibers [10,11], including in particular carbon nanotubes (CNTs) that possess advantageous specific stiffness and strength [12]. Their excellent electrical and thermal conductivity make them attractive as a nanofiller for aerospace composites due to their ability to add multifunctional properties and capabilities [13–16]. While CNTs possess significant potential for composite reinforcement, only small improvements in composite properties have been realized due to difficulties in processing and controlling CNT orientation during manufacturing. Many approaches to CNT incorporation have been explored in industry and in the literature, dominated by mixing CNTs in polymer matrices before infiltration [17–23] in low loading to minimize influence on matrix viscosity. CNTs tend to agglomerate into bundles, and techniques to separate the CNTs by sonication [24], surface modification [25,26] and/or shear mixing [27] can easily damage the CNTs. Alignment of the CNTs in the matrix is also difficult to obtain even with severe processing techniques like extrusion [28] that are not compatible with laminated composites, though some orientation has been controlled through directional

S.S. Wicks et al. / Composites Science and Technology 100 (2014) 128–135

infusion [29]. Grafting CNTs onto microfiber surfaces prior to polymer infiltration allows precise control over CNT organization and distribution, though care must be taken to avoid new manufacturing challenges that accompany the grafting process by preventing damage to the microscale fibers [30] and understanding changes to microfiber packing through decreased compressibility [31,32]. CNT-grafted microfibers show potential for mechanical reinforcement and have been demonstrated to provide interlaminar reinforcement [33,34], increase interfacial shear strength [35–37], and locally reinforce the matrix [38]. These main approaches to reinforce composites with CNTs are reviewed in the literature [39]. In this paper, aligned CNTs are implemented in a three-dimensionally reinforced nano-engineered composite. The micron-scale diameter fibers are alumina, serving as a model system for this study, because the microfiber tensile strength and microfiber– polymer interfacial shear strength are preserved following the CNT growth process [40,41], unlike (tensile) strength losses observed by many researchers for the industrially more relevant carbon fibers due to the carbon fiber surface sensitivity to high temperatures and/or metal catalysts [30,35,36]. Other researchers have noted interfacial shear strength enhancement for nanostructure-modified microfibers [42,43], usually with concomitant microfiber strength loss in the carbon fibers. The aligned carbon nanotubes in this work are grown in situ on the surfaces of microfibers, termed ‘fuzzy fibers’ as shown in Fig. 1a, and extend across interlaminar and intralaminar regions to provide mechanical reinforcement [44–46]. By controlling the CNT alignment and dispersion in fuzzy fiber reinforced plastics (FFRP), facile manufacturing with polymer matrix systems is enabled that is compatible with existing manufacturing processes [31,47]. Preliminary work on the same alumina fiber system in a thick coarse weave had demonstrated a 76% improvement in steady-state toughness for

129

laminates selectively reinforced along the midplane [45], a geometry that can lead to jumping of the crack from the tough midplane to adjacent ply interfaces that do not have CNT reinforcement. The work herein remedies this situation with laminates reinforced throughout the thickness with aligned CNTs. Mode I double cantilever beam (DCB) laminates were manufactured and tested with a brittle aerospace epoxy and a tough marine epoxy to understand epoxy interaction with both microfibers and CNTs as it relates to toughening behavior. 2. Materials and methods FFRP laminates are manufactured by producing ‘fuzzy fiber’ plies and then infiltrating with an epoxy matrix either by vacuum assisted resin infusion (VARI) or hand lay-up. Alumina cloth (Cotronics Ultra-Temp 391, Dutch weave tape, areal density of 371 g/m2) is first dipped in a 50 mM solution of iron nitrate in isopropanol, distributing catalyst to all alumina fiber surfaces even within the cloth, and then dried in an oven for 8 h at 30 °C. The cloth is cut into 200  700 plies onto which aligned CNTs are grown by chemical vapor deposition in a 200 tube furnace (Lindberg Blue M). The cloth is heated to a furnace setpoint of 650 °C under 1040 sccm of hydrogen, then stabilized at temperature for 5 min for the hydrogen pre-treatment step to create catalyst particles of iron on the microfiber surface [45,46]. Ethylene is introduced at 350 sccm to grow aligned CNTs, the lengths of which are controlled by varying the duration of ethylene exposure. For example, the bulk of the CNTs in this set are made with approximately 3 min of ethylene exposure, yielding CNTs that are 19 lm long, which cause the annular aligned CNT forests to split into ‘mohawks’ along the microfiber length (see Fig. 1c). The short CNT plies are made with 1 min of ethylene exposure to yield 6 lm long annular ACNTs around the microfiber as shown in Fig. 1d [46]. Tuning the

Fig. 1. Fuzzy fiber reinforced plastics (FFRP) (a) Illustration of FFRP architecture. (b) FFRP DCB test specimen, with dimensions of 20  180 mm (width  length), with red chopped-fiber plate bending arm reinforcement. (c) Fibers in long CNT fuzzy architecture before polymer infiltration with 20 lm long aligned CNTs on the surface of fibers. (d) Fibers in short CNT fuzzy architecture before polymer infiltration with 6 lm CNT fuzz on fiber surface. Insets show arrangement of aligned CNTs around fibers. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

130

S.S. Wicks et al. / Composites Science and Technology 100 (2014) 128–135

Fig. 2. Manufacturing routes for FFRP production. (a) Side schematic of infusion setup showing primary epoxy (Hexcel RTM6) flow path. (b) Hand lay-up for viscous epoxy system (West Systems 105/206).

A-CNT length also affects the microfiber geometrical arrangement while also reinforcing the matrix. A-CNTs on the microfiber surfaces effectively push apart the microfibers in the tows, with the long CNTs inducing the greatest degree of tow swelling that can alter the local tow geometry and, importantly for this work, how the neighboring plies nest together at the interlaminar interfaces. The A-CNTs are not functionalized after growth, i.e., they are used as-grown, and the A-CNTs do not affect the degree of cure of the matrix, as determined by comparing the energy to cure via Differential Scanning Calorimetry (DSC) on epoxy and A-CNT reinforced epoxy. Laminates are manufactured 8 plies thick by infusion or hand lay-up depending on polymer type. Both matrix systems are epoxy, though they have very different characteristics: the aerospace epoxy is based on a tetrafunctional resin, tetraglycidylether of 4,40 -diaminodiphenyl methane (TGDDM), while the marine epoxy is based on a difunctional resin, bisphenol-A epoxy resin (BPA). The aerospace resin is a highly crosslinked polymer compared to the marine epoxy due to the chemical structure, and as suggested by the high Tg for the aerospace resin; the differing chemical structure leads to different mechanical behavior, where the aerospace epoxy has a tensile strength of 75 MPa with a strain to failure of 3.4%, while the marine epoxy has a tensile strength of 50 MPa and a strain to failure of 4.5%. VARI is used to manufacture both FFRP and baseline (without CNTs) laminates with the unmodified aerospace infusion resin (Hexcel RTM6, 33 cPs at processing temperature), while hand lay-up is used with the marine epoxy. Fig. 2a shows the VARI setup, where spiral tubes and distribution mesh sheets draw the epoxy along and into the laminate. The laminate is layered with porous Teflon peel ply and flow media on both surfaces, before being sandwiched between steel and glass plates held apart by spacers to ensure a flat and uniform part with some control over volume fraction. A 50.8 mm (200 ) wide steel plate that

matches the width of the laminate is placed on top of the stack to minimize ‘racetracking’ of the epoxy along the edges during infusion. This allows the vacuum bag to conform to the top plate and the edges of the laminate. Prior to infusion, the plates and laminate stack are heated under vacuum to 150 °C (300 °F) for 2 h to displace any residual moisture. Then, the aerospace epoxy is degassed at 80 °C, and slowly drawn into the laminate at 90 °C at a vacuum of 29 in Hg (740 mmHg). The inlet and outlet of the system is then closed off and cured at 160 °C and post cured at 180 °C. The microfiber volume fractions of laminates made by infusion are all comparable: 41–48% for baseline, 40–51% for the short CNT, and 40–48% for the long CNT FFRP laminates, with resulting CNT weight fractions in the total laminate at 0.5% and 2% for the short and long CNT systems, respectively. Successful testing was performed on 11, 3, and 14 specimens of baseline, short CNT, and long CNT laminates, respectively. Due to challenges in growing consistent short CNT plies, only an exploration into short CNT laminates is performed. Hand lay-up is used with viscous epoxy systems to produce both baseline and FFRP laminates as shown in Fig. 2b. In hand lay-up, a 2-part marine epoxy (West Systems Resin 105 and Hardener 206, 725cPs) is mixed and poured onto a flat surface. As shown in Fig. 2b, a ply is then laid down in the epoxy pool, allowing the epoxy to wick into the interior of the ply. More epoxy is poured on top, and the process is repeated until the entire laminate stack is wet. The stack is then compressed between two plates wrapped in vacuum bagging film held apart by spacers, and left to cure for 24 h. The microfiber volume fractions of laminates made by hand lay-up were 35–43% for baseline, 44–50% for the short CNT, and 33–41% for the long CNT FFRP laminates, with resulting CNT weight fractions in the total laminate at 0.5% and 2% for the short and long CNT systems, respectively. The ranges of microfiber volume fractions are evidence of the difficulty to control volume fraction even when using spacers due to the variations

Fig. 3. Mode I fracture toughness of double cantilever beam laminates. Strain energy release rate, GIc, of baseline, short CNT, and long CNT curves of laminates made with aerospace and marine epoxy. Initiation toughness based on visual onset of propagation, and percentages are changes from baseline of each set.

S.S. Wicks et al. / Composites Science and Technology 100 (2014) 128–135

131

Fig. 4. Mode I fracture of aerospace epoxy laminates. (a) Representative R-curves of baseline and FFRP tests. (b) Smooth epoxy cleavage in epoxy of baseline laminates in steady-state portion of R-curve. (c) Fiber–matrix adhesive failure and resin-rich regions in short CNT FFRP laminates in steady-state portion of R-curve. (d and e) Rough epoxy fracture along fiber tows in long CNT FFRP laminates in steady-state portion of R-curve. (f) CNTs pulled out of fracture surface in FFRP laminates.

Fig. 5. Example fractography of aerospace epoxy laminate with long A-CNTs. (a) Fiber–matrix interface adhesive failure in initiation region of crack propagation. (b) Evolution of strain energy release rate along a long CNT aerospace laminate. (c) Rough matrix cohesive failure in peak toughness regions of FFRP.

in cloth compressibility, and large scale implementation of FFRP laminates should employ a closed mold manufacturing system that would allow microfiber volume fraction to be more tightly controlled. For both hand lay-up and infusion, aligned CNTs have been previously demonstrated to remain attached to the microfibers within the FFRP laminate [47,48]. Fracture testing was performed on 10, 5, and 17 specimens of baseline, short CNT, and long CNT laminates, respectively, and in all cases the extracted values for fracture show sufficiently low variability to allow comparison. To fabricate double cantilever beam (DCB) laminates, a 25 lm fluoropolymer release film is placed between the middle two plies on the last 5 cm on one end of the laminate during lay-up, to form a starter precrack at the laminate centerline. The resulting laminates are cut into 20 mm-wide specimens. To prevent excessive bending during DCB testing, 3.2 mm thick chopped fiberglass plate tabs (see Fig. 1b), cut to the same width of the laminate, are used to sandwich the laminate [45]. The surfaces of the laminate and fiberglass are roughened with sandpaper and cleaned with acetone, isopropanol, and methanol to prepare the surfaces for bonding. A 30 min epoxy is used to glue the fiberglass plates and hinges for test grips to the samples. The tabbed DCB specimens are tested in a Zwick Roell mechanical tester with a 10 kN load cell following ASTM D5528 [49]. The samples are loaded at 0.5 mm/min and load, stroke, and crack length are recorded. The strain energy release rate, GIc, for each laminate type is calculated using the Compliance

Calibration Method, as outlined in ASTM D5528 [49]. Initiation toughness, steady-state toughness, and the maximum observed toughness are computed.

3. Results and discussion Mode I fracture toughness is characterized by a resistance (Rcurve) behavior that includes an initiation value as well as a region of increasing toughness (resistance) as the crack extends until a steady-state value is reached. Exemplary R-curves are shown in Figs. 4a and 6a. The initiation toughness, GIc, initiation, was assessed by three different methods in following the ASTM standard: onset of visual propagation (VIS GIc, initiation), deviation from linearity (NL GIc, initiation), and a 5% offset from linearity (NL GIc, initiation). The steady-state toughness was taken as the average of the entire R-curve excluding at least the first 5 points (where toughness is increasing). Because the standard provides no method to calculate steady-state toughness when the R-curves do not plateau, the comparison of the maximum GIc value (GIc,max) is also presented. Computed toughness values are summarized in Table 1, and the results for visual initiation onset and steady-state are included in Fig. 3. The trends observed for each epoxy type will be discussed in detail next. Representative GIc R-curves for the aerospace epoxy laminates showing the evolution of toughness with increased crack length

132

S.S. Wicks et al. / Composites Science and Technology 100 (2014) 128–135

Fig. 6. Mode I fracture of marine epoxy laminates. (a) Representative R-curves of baseline and FFRP tests. (b) Individual microfibers pulled out of fracture surface in baseline laminates. (c) microfiber and tow pullout across crack tip, where vertical lines demarcate millimeters of crack growth. (d and e) Broken tow surface revealing broken fibers and holes from microfiber fracture and pullout. (f) Rough CNT-filled fracture of epoxy between microfibers showing long CNT pullout.

Table 1 Summary data for Mode I fracture tests. 5% GIc, Aerospace Epoxy Baseline Short CNT Long CNT Marine Epoxy Baseline Short CNT Long CNT

initiation

0.32 ± 0.08 0.14 ± 0.02 0.29 ± 0.17 0.57 ± 0.26 0.74 ± 0.33 0.77 ± 0.25

(kJ/m2)

{56%} {9%} {+31%} {+37%}

NL GIc,

initiation

(kJ/m2)

0.21 ± 0.08 0.095 ± 0.035 {55%} 0.14 ± 0.06 {33%} 0.27 ± 0.09 0.52 ± 0.15 {+90%} 0.31 ± 0.06 {+15%}

are summarized in Fig. 4a, and differences in the shape of the R-curves can be seen, notably between the baseline and long CNT FFRP. The baseline curves exhibit minimal toughening after crack initiation while the long CNT FFRP curve exhibits a lower initiation toughness followed by clear toughening behavior. A moderate reduction (12–36% across all three measures) in initiation toughness is observed for these long CNT FFRP laminates, whereas there is a moderate, but larger, increase in both steady-state and maximum toughness of 32.9% and 39.0%, respectively. Laminates made with short CNTs in the same aerospace epoxy lead to significantly reduced toughness, with all three measures of initiation as well as steady-state toughness dropping by over 50%. SEM fractography of the fractured laminates shows that the CNTs in the FFRP induce microscale changes to the fracture surface. In aerospace epoxy baseline laminates, as shown in Fig. 4b, fracture surfaces are dominated by smooth epoxy regions with some exposed microfibers. The drop in toughness in the short CNT FFRPs is counterintuitive, but can be understood by considering the fracture surface in Fig. 4c where exposed microfiber surfaces are adjacent to resin-rich regions with no CNTs. In contrast, the fracture surfaces of long FFRP laminates, shown in Fig. 4d and e, show a very different morphology comprised of a rougher fracture surface as the crack follows the tows, exposing some microfiber surfaces with a mixture of fiber–epoxy interface adhesive failure and matrix cohesive failure. Exposed in the rough surfaces of the matrix cohesive failure are CNTs that were pulled out, up to a few microns, during crack propagation as shown in the SEM image in Fig. 4f, suggesting CNT pullout as a mechanism for toughening the matrix [50]. Thus, the drop in toughness for the short CNTs is primarily

VIS GIc,

initiation

0.36 ± 0.08 0.14 ± 0.01 0.25 ± 0.10 0.58 ± 0.21 0.71 ± 0.25 0.83 ± 0.32

(kJ/m2)

{61%} {29%) {+22%} {+43%}

GIc,

max

(kJ/m2)

0.47 ± 0.08 0.23 ± 0.01 0.66 ± 0.15 1.33 ± 0.12 2.65 ± 0.33 2.67 ± 0.42

{51%} {+40%} {+99%} {+101%}

GIc,

steady-state

(kJ/m2)

0.38 ± 0.07 0.19 ± 0.003 {50%} 0.50 ± 0.11 {+32%} 1.19 ± 0.09 2.25 ± 0.19 {+89%} 2.38 ± 0.32 {+100%}

attributed to the short CNTs leaving resin rich regions between plies combined with the crack path being driven to the fiber– matrix interface. The short CNTs are not long enough to effectively bridge the interlaminar region. Comparatively, longer A-CNTs fully bridge the interlaminar region and contribute toughness from CNT pullout and increased cloth nesting. Fracture morphology evolution within a given laminate, as shown in Fig. 5, provides further support for these observations. The fracture surfaces corresponding to the toughest regions of the laminate are dominated by matrix cohesive failure, creating rough, jagged surfaces as seen in Fig. 5c. Conversely, bare exposed microfibers are observed in the initiation fracture surface in Fig. 5a where toughness is lowest. The causes for evolution of fracture morphology that govern whether the fracture remains in the matrix or follows the microfiber surface might include nesting prevention of the woven cloth at initiation by the release film precrack, or variation in the CNT morphology along the laminate, among others. Representative R-curves of Mode I fracture of marine epoxy laminates are shown in Fig. 6a. Unlike the R-curves in the aerospace resin laminates above, both the baseline and FFRP laminates form a typical R-curve, i.e., classic toughening behavior. Changes in initiation, steady-state, and maximum fracture toughnesses are summarized in Table 1, where large gains in all measures are seen for both short and long CNTs. The baseline laminates consistently fracture, reaching steady-state within 10 mm of crack advance. The initiation toughness for long CNT laminates increase above the baseline 15– 43% across all three initiation measures, while the steady-state and max toughnesses increase by 100% and 101%, respectively. The long CNT FFRP laminates behave in two ways, either toughening

S.S. Wicks et al. / Composites Science and Technology 100 (2014) 128–135

133

Fig. 7. Example fractography of a marine epoxy laminate with long A-CNTs. (a) Minimal fiber pullout in fracture surface of initiation region. (b) Evolution of strain energy release rate along a long-CNT laminate with gradual increasing toughening (c) Tow breakage and microfiber pull-out in FFRP laminates.

for 15 mm of crack advance until peaking, or toughening over the entire length of the sample. Samples in the latter category were excluded from the steady-state average as those laminates did not reach steady-state before the end of the test, so GIc,ss is conservatively based on specimens that did not have an increasing toughness trend to the end of the test. If the steady-state values were evaluated to include the peak value reached by these specimens at the end of the specified crack length of 50 mm, the GIc,ss would have been 2.41 kJ/m2, corresponding to a 102% increase over the baseline. The short CNT laminates are nearly as tough as the long CNT laminates, with the initiation toughness at 22–90% above baseline for the 3 initiation measures, and the steady-state and max toughness values increase by 89% and 99%, respectively. The fracture surfaces of the marine epoxy laminates highlight a major difference between the epoxy systems. All marine epoxy laminate fracture surfaces show widespread microfiber–matrix interface failure and microfiber pullout (see Fig. 6b and d, contrasted to the morphology of the toughest regions of aerospace epoxy FFRP as discussed above in Fig. 5c). Fiber–matrix interface failure leaves portions of the fracture surface exposed and smooth, indicating that the prevention of debonding at the microfiber surface is not required for interlaminar reinforcement. Instead, increased toughness in marine epoxy laminates manifests in increased microfiber and tow breakage which can be observed during crack propagation as these fibers are pulled free from the surrounding matrix and bridge across

the crack tip, as seen in the side view of the crack wake in Fig. 6c. In addition to pull-out and bridging of microscale fibers as seen in Fig. 6d and e, the broken epoxy between microfibers reveals pullout and bridging of nanoscale CNTs, indicating toughening at multiple length scales. The CNTs pulled out of the marine epoxy in Fig. 6f are up to a few microns in length, similar to the CNTs in the aerospace epoxy surfaces. Increasing toughness in the FFRP laminates as seen in Fig. 7 corresponds directly to increased microfiber breaks in the fracture surface. This fracture morphology suggests that the CNTs increase the toughness of the interlaminar region by creating a tortuous path, even to the extreme such that the crack propagates through the surface of the weave rather than around each tow. The CNTs grown on the microfiber surfaces within each tow help push apart the microfibers causing the tow to swell, allowing for tows on opposite plies to nest and create a tortuous path for the crack to follow. The monotonic increase from initiation to steady-state toughness (limited by the laminate specimen test length in Fig. 7b) may be caused by variations in weave nesting along the crack length, which is prevented at initiation by the release film in all baseline and FFRP laminates, but could vary along the length of the sample depending on how adjacent plies register to one another. Example fractography on short CNT marine epoxy is shown in Fig. 8, which demonstrates that while the fracture surfaces show limited microfiber fracture, other toughening

Fig. 8. Example fractography of marine epoxy laminate with short A-CNTs. (a) Tow breakage and fiber pull-out as crack jumps from main crack path to one ply interface away. (b) Minimal fiber pullout and flat fracture surface one ply interface away from the centerline main DCB crack path. (c) Flat fracture and fiber–matrix interface failure at initiation region. (d) Strain energy release rate along a short CNT laminate.

134

S.S. Wicks et al. / Composites Science and Technology 100 (2014) 128–135

mechanisms come into play such as crack jumping across plies to another interface. Therefore, the grown A-CNTs change the microscale fracture behavior not only via pullout and altering mechanical properties of the matrix (simple nanofiber reinforcement changing, e.g., moduli), but also by altering the failure path and the fracture surface texture, adding surface area through both microfiber and CNT pullout and failure. Other mechanisms of interlaminar toughening also need to be considered in the context of the above findings. The A-CNTs do not quantitatively alter the interlaminar region (thickness) in the woven laminate interface, nor do the A-CNTs cause any measurable change in the polymer matrix degree of cure. Hard nanostructures such as CNTs have been shown to change local crystallinity or alter polymer orientation in some cases, particularly for thermoplastics [51,52]. However, this is not expected in these epoxies as confirmed through prior studies on the aerospace epoxy showing no change in polymer orientation for a wide range of CNT volume fractions [53] and by preliminary DSC investigations which again show no measurable change in the either epoxy’s characteristics with and without CNTs. Thus, the two reinforcement mechanisms of either increased interlaminar thickness (adding a compliant interlayer) or polymer morphology alteration are not contributing substantially to toughness trends observed in this work. 4. Conclusions and recommendations The addition of aligned CNTs has been shown to improve the steady-state Mode I fracture toughness of laminated composites. Fractography reveals toughening mechanisms for both aerospace and marine epoxy laminates at several length scales, from the pull-out of carbon nanotubes to inducing tow breakage. The toughening behavior and magnitude of steady-state toughness improvement, however, is dependent on the type of epoxy. In the more brittle aerospace epoxy laminate, modest improvement in steady-state toughness with long CNTs is made possible by maintaining cohesive interlaminar matrix failure around woven tow features while increasing fracture surface area through CNT pullout and rough epoxy fracture. The tougher marine epoxy allows large improvements with CNTs by significantly adding instances of microfiber breakage and pullout along with CNT pullout and epoxy fracture. Varying the CNT length begins to reveal how the geometrical arrangement of microfibers through tow swelling and changes in nesting affect the crack propagation path and subsequent interlaminar toughness. When applying the fuzzy fiber architecture to a chosen fiber and matrix system, the dependence of toughness on the CNT length and polymer type will need to be established for tailoring interlaminar toughness. The clarification of multi-scale toughening modes in this work will be supplemented through future studies establishing the extent to which nanoscale reinforcement toughens the matrix only (i.e. no microfibers) and isolating the role and mechanisms of the CNT-matrix interface in resisting crack growth, leading to an understanding of how the choice of epoxy and nanofiber type dictate the magnitude of improvement possible. The alumina FFRP system studied here is a useful model system to understand features of aligned CNT property tailoring that are isolated from microfiber damage due to CNT growth. However, a more engineering relevant system is carbon FFRP, particularly for aerospace applications. Future work will include study of carbon FFRP laminates, based on recent work demonstrating synthesis of CNTs on carbon fibers that avoids deleterious damage to the fibers [30]. Acknowledgments This work was supported by Airbus Group, Boeing, Embraer, Lockheed Martin, Saab AB, ANSYS Inc., Hexcel, and TohoTenax

through MIT’s Nano-Engineered Composite aerospace STructures (NECST) Consortium. Sunny is supported by the NASA Space Technology Research Fellowships (NSTRF) and MIT’s Linda and Richard (1958) Hardy Fellowship. The authors thank Dale Lidston, Soraya Kalamoun, and Henna Jethani for fabrication and testing assistance, and Hexcel for materials donation. This work made use of the MRSEC Shared Experimental Facilities at MIT, supported by the NSF under award number DMR-08-19762, and core facilities at the Institute for Soldier Nanotechnologies at MIT supported by the U.S. Army Research Office under contract W911NF-07-D-0004.

References [1] Tong L, Mouritz AP, Bannister MK. 3D Fibre reinforced polymer composites. Oxford: Elsevier; 2002. [2] Mouritz AP. Review of z-pinned composite laminates. Compos A 2010;38(12):709–28. [3] Dransfield KA, Jain LK, Mai Y-W. On the effects of stitching in CFRPs—1. Mode I delamination toughness. Compos Sci Technol 1998;58:815–27. [4] Mouritz AP, Cox BN. A mechanistic interpretation of the comparative in-plane mechanical properties of 3D woven, stitched, and pinned composites. Compos A 2010;41(6):709–28. [5] Yasaee M, Bond IP, Trask RS, Greenhalgh ES. Mode I interfacial toughening through discontinuous interleaves for damage suppression and control. Compos A 2012;43(1):198–207. [6] Sager TJ, Klein PJ, Davis DC, et al. Interlaminar fracture toughness of woven fabric composite laminates with carbon nanotube/epoxy interleaf films. J Appl Polym Sci 2011;121(4):2394–405. [7] Böger L, Sumfleth J, Hedemann H, Schulte K. Improvement of fatigue life by incorporation of nanoparticles in glass fibre reinforced epoxy. Compos A 2010;41(10):1419–24. [8] Hayes BS, Seferis JC. Modification of thermosetting resins and composites through preformed polymer particles: a review. Polym Compos 2001;22(4):451–67. [9] Siddiqui NA, Woo RSC, Kim J-K, Leung CCK, Munir A. Mode I interlaminar fracture behavior and mechanical properties of CFRPs with nanoclay-filled epoxy matrix. Compos A 2007;38(2):449–60. [10] Zhou Y, Jeelani S, Lacy T. Experimental study on the mechanical behavior of carbon/epoxy composites with a carbon nanofiber-modified matrix. J Compos Mater 2013. 0021998313512348. [11] Bortz DR, Merino C, Martin-Gullon I. Mechanical characterization of hierarchical carbon fiber/nanofiber composite laminates. Compos A 2011;42(11):1584–91. [12] Coleman JN, Khan U, Blau WJ, Gun’ko YK. Small but strong: a review of the mechanical properties of carbon nanotube–polymer composites. Carbon 2006;44(9):1624–52. [13] Guzmán de Villoria R, Yamamoto N, Miravete A, et al. Multi-physics damage sensing in nano-engineered structural composites. Nanotechnology 2011;22(18):185502. [14] Gibson RF. A review of recent research on mechanics of multifunctional composite materials and structures. Compos Struct 2010;92(12):2793–810. [15] Wu AS, Na W-J, Yu W-R, et al. Carbon nanotube film interlayer for strain and damage sensing in composites during dynamic compressive loading. Appl Phys Lett 2012;101(22):221909. [16] Gao L, Chou T-W, Thostenson ET, et al. In situ sensing of impact damage in epoxy/glass fiber composites using percolating carbon nanotube network. Carbon 2011;49(10):3382–5. [17] Seyhan AT, Tanoglu M, Schulte K. Mode I and mode II fracture toughness of Eglass non-crimp fabric/carbon nanotube (CNT) modified polymer based composites. Eng Fract Mech 2008;75(18):5151–62. [18] Gojny FH, Wichmann MHG, Köpke U, et al. Carbon nanotube-reinforced epoxycomposites: enhanced stiffness and fracture toughness at low nanotube content. Compos Sci Technol 2008;64(15):2363–71. [19] Thakre PR, Lagoudas DC, Riddick, et al. Investigation of the effect of single wall carbon nanotubes on interlaminar fracture toughness of woven carbon fiber– epoxy composites. J Compos Mater 2011;45(10):1091–108. [20] Godara A, Mezzo L, Luizi F, Warrier A, et al. Influence of carbon nanotube reinforcement on the processing and the mechanical behaviour of carbon fiber/epoxy composites. Carbon 2009;47(12):2914–23. [21] Chou T-W, Gao L, Thostenson ET, Zhang Z, Byun J-H. An assessment of the science and technology of carbon nanotube-based fibers and composites. Compos Sci Technol 2010;70(1):1–19. [22] Bekyarova E, Thostenson ET, Yu A, et al. Functionalized single-walled carbon nanotubes for carbon fiber epoxy composites. J Phys Chem C 2007;111(48):17865–71. [23] Lee S-B, Choi O, Lee W, et al. Processing and characterization of multi-scale hybrid composites reinforced with nanoscale carbon reinforcements and carbon fibers. Compos A 2011;42(4):337–44. [24] Chandrasekaran V, Advani S, Santare M. Role of processing on interlaminar shear strength enhancement of epoxy/glass fiber/multi-walled carbon nanotube hybrid composites. Carbon 2010;48(13):3692–9.

S.S. Wicks et al. / Composites Science and Technology 100 (2014) 128–135 [25] Fan Z, Santare MH, Advani SG. Interlaminar shear strength of glass fiber reinforced epoxy composites enhanced with multi-walled carbon nanotubes. Compos A 2008;9(3):540–54. [26] Ashrafi B, Guan J, Mirjalili V, et al. Enhancement of mechanical performance of epoxy/carbon fiber laminate composites using single-walled carbon nanotubes. Compos Sci Technol 2011;71(13):1569–78. [27] Gojny FH, Wichmann MHG, Fiedler B, et al. Influence of nano-modification on the mechanical and electrical properties of conventional fibre-reinforced composites. Compos A 2005;36(11):1525–35. [28] Isayev AI, Kumar R, Lewis TM. Ultrasound assisted twin screw extrusion of polymer-nanocomposites containing carbon nanotubes. Polymer 2009;50(1):250–60. [29] Qiu J, Zhang C, Wang B, Liang R. Carbon nanotube integrated multifunctional multiscale composites. Nanotechnology 2007;18(27):275708. [30] Steiner SA, Li R, Wardle BL. Circumventing the mechanochemical origins of strength loss in the synthesis of hierarchical carbon fibers. ACS Appl Mater Interfaces 2013;4(11):4892–903. [31] Lomov SV, Gorbatikh L, Kotanjac Zˇ, et al. Compressibility of carbon woven fabrics with carbon nanotubes/nanofibres grown on the fibres. Compos Sci Technol 2011;71(3):315–25. [32] Lomov SV, Wicks S, Gorbatikh L, et al. Compressibility of nanofibre-grafted alumina fabric and yarns: Aligned carbon nanotube forests. Compos Sci Technol 2014;90:57–66. [33] Veedu VP, Cao A, Li X, et al. Multifunctional composites using reinforced laminae with carbon-nanotube forests. Nat Mater 2006;5(6):457–62. [34] Kepple KL, Sandborn GP, Lacasse PA, et al. Improved fracture toughness of carbon fiber composite functionalized with multi walled carbon nanotube. Carbon 2008;46(15):2026–33. [35] Qian H, Bismarck A, Greenhalgh ES, Shaffer MSP. Carbon nanotube grafted carbon fibres: a study of wetting and fibre fragmentation. Compos A 2010;41(9):1107–14. [36] Sager RJ, Klein PJ, Lagoudas DC, Zhang Q. Effect of carbon nanotubes on the interfacial shear strength of T650 carbon fiber in an epoxy matrix. Compos Sci Technol 2009;69(7–8):898–904. [37] Zhang FH, Wang RG, He XD, Wang C, Ren LN. Interfacial shearing strength and reinforcing mechanisms of an epoxy composite reinforced using a carbon nanotube/carbon fiber hybrid. J Mater Sci 2009;44(13):3574–7. [38] Qian H, Kalinka G, Chan KL, et al. Mapping local microstructure and mechanical performance around carbon nanotube grafted silica fibres: methodologies for hierarchical composites. Nanoscale 2011;3(11):4759–67. [39] Qian H, Greenhalgh ES, Shaffer MSP, Bismarck A. Carbon nanotube-based hierarchical composites: a review. J Mater Chem 2010;20(23):4751–62.

135

[40] Garcia EJ, Hart J, Wardle BL. Long carbon nanotubes grown on the surface of fibers for hybrid composites. AIAA J 2008;46(6):1405–12. [41] Lachman N, Wiesel E, Guzmán de Villoria R. Interfacial load transfer in carbon nanotube/ceramic microfiber hybrid composites. Compos Sci Technol 2012;72(12):1416–22. [42] Jin S-Y, Young RJ, Eichhorn SJ. Hybrid carbon fibre–carbon nanotube composite intefaces. Compos Sci Technol 2014;95:114–20. [43] Allen RJ, Ghita O, Farmer B, et al. Mechanical testing and modelling of a vertically aligned carbon nanotube composite structure. Compos Sci Technol 2013;77:1–7. [44] Garcia EJ, Wardle BL, Hart AJ, Yamamoto N. Fabrication and multifunctional properties of a hybrid laminate with aligned carbon nanotubes grown In Situ. Compos Sci Technol 2008;68(9):2034–41. [45] Wicks SS, Guzmán de Villoria R, Wardle BL. Interlaminar and intralaminar reinforcement of composite laminates with aligned carbon nanotubes. Compos Sci Technol 2010;70(1):20–8. [46] Yamamoto N, Hart AJ, Garcia EJ, et al. High-yield growth and morphology control of aligned carbon nanotubes on ceramic fibers for multifunctional enhancement of structural composites. Carbon 2009;47(3): 551–60. [47] Yamamoto N, Guzmán de Villoria R, Wardle BL. Electrical and thermal property enhancement of fiber-reinforced polymer laminate composites through controlled implementation of multi-walled carbon nanotubes. Compos Sci Technol 2012;72(16):2009–15. [48] Wicks SS, Kalamoun S, Williams MR, et al., Effect of manufacturing route on mode I fracture toughness of aligned carbon nanotube reinforced composites. In: 53rd AIAA Structures, Structural Dynamics, and Materials (SDM) conference, Honolulu, HI, April 23–26; 2012. [49] ASTM D 5528-01. Standard Test Method for Mode I Interlaminar Fracture Toughness of Unidirectional Fiber-Reinforced Polymer Matrix Composites, ASTM International. [50] Wagner HD, Ajayan PM, Schulte K. Nanocomposite toughness from a pull-out mechanism. Compos Sci Technol 2013;83:27–31. [51] Ma Q, Mao B, Cebe P. Chain confinement in electrospun nanocomposites: using thermal analysis to investigate polymer–filler interactions. Polymer 2011;52(14):3190–200. [52] Meng J, Zhang Y, Song K, Minus M. Forming crystalline polymer-nano interphase structures for high-modulus and high-tensile/strength composite fibers. Macromol Mater Eng 2014;299(2):144–53. [53] Wardle BL, Saito DS, Garcia EJ, et al. Fabrication and characterization of ultrahigh volume fraction aligned carbon nanotube–polymer composites. Adv Mater 2008;20(14):2655–796.