Flexural and shear properties of CFRP laminates reinforced with functionalized multiwalled CNTs

Abstract This study focuses on mechanical characterization of carbon fiber reinforced polymer (CFRP) laminates reinforced with non-treated and treated multiwalled carbon nanotubes (CNTs) using nitric acid. The CNTs were treated using nitric acid to obtain carboxylic functional group. The epoxy resins are mixed with 0.3%wt of multiwalled CNTs at a constant mixing speed of 2000 rpm and mixing times varied from 24 to 96 h. Laminates reinforced with treated multiwalled CNTs show an increase in the flexural strength by 17.4 and 15.3% at mixing times of 24 and 96 h as compared to control laminates. The test results indicated that laminates reinforced with treated multiwalled CNTs have improved interlaminar shear failure stress which is 14 and 7% higher at mixing times of 24 and 96 h as compared to control specimen. Improvement in behavior was achieved for functionalized CNTs based laminates which prevents agglomeration. Longer mixing time (96 h) is not beneficial for enhancing the mechanical properties due to the break-up of small aggregates by overcoming the effect of van der Waals forces. Graphical Abstract


Introduction
Carbon fiber reinforced polymer (CFRP) composites have been extensively used as structural materials for aircraft, automotive, aeronautical and space engineering components and for wind turbine blades. Application of CFRP in these industries reduces the weight of components, thereby leading to energy savings [1]. To broaden the use of CFRP composites in many engineering applications, it is important to improve the mechanical properties of the binder material. Polymers have excellent dimensional stability, good adhesion, compatibility with most fibers, wear resistance and low cost. However, these materials have inherent brittleness and poor mechanical properties which limits their wider use. Recently, researchers have carbon nanotubes (CNTs) to reinforce polymer matrix to improve its mechanical properties [2]. Both experimental and theoretical studies indicated that CNTs have exceptional stiffness of up to 1 TPa and strength of order 100 GPa. This made CNTs an ideal reinforcing material to improve not only the mechanical properties but also thermal, magnetic and electrical properties of composites [3][4][5]. Fawad Tariq et al. [6] developed a composite material by combing carbon fabric and MWCNTs in epoxy matrix. Tensile and flexural strength were conducted to determine the effect of MWCNTs on mechanical behavior of CFRP material. Improvement in tensile and flexural strength was attributed to the microcrack bridging effect of MWCNTs. Because of its excellent properties, CNTs can be used as ideal reinforcing agents for composite structures [7][8][9][10].
Since the discovery of CNTs in 1991 by Ijima, the material has received considerable attention for use in several applications such as superconductors, nanoelectronics, nanowires, electrochemical capacitors and nanocomposite structures [11,12]. The properties that can be improved with the addition of CNTs include tensile modulus, tensile strength, flexural strength, interlaminar shear strength (ILSS), toughness, damping behavior and glass transition temperature [13,14]. Achieving good dispersion and aligning of CNTs in a polymer matrix are critical concerns as well as a challenging task due to strong attractive forces leading to agglomeration and aggregation in the form of bundles [15]. These issues are main obstacles for developing high-performance polymer/CNT composites by improving the load transfer capacity of polymer/CNT interface. Improvement in mechanical, electrical and thermal properties on composite structures has been shown by growing of CNT onto carbon fiber (CF) surface. Particularly, the interfacial adhesion between the CF and the matrix will be improved due to the presence of CNT [16]. Chemical vapor deposition (CVD) is one of the techniques used in the production of hybrid CF-CNT, which had involved the in-situ growth of the CNT on the CF surface. In this method, the CF surface used as a substrate exposed to a volatile precursor material, which react and decomposes on CF surface to form the desired CNT deposit. However, CF substrate may be contaminated by the catalysts transition metals for growth of the CNT during in-situ growth [17]. Electrospray deposition (ESD) method is used to produce uniform dispersion of non-agglomerating nano or micro-droplets, which are suitable for producing micrometer-thin or thinner layers of spray and uniform coatings into the fibers, which improved interfacial bonding in composite structures [18]. Functionalization of CNTs has been considered as a better technique to prevent agglomeration and to achieve better interfacial characteristics between CNTs and the polymer matrix. During covalent functionalization, CNTs are oxidized using acids in order to obtain hydroxyl or carboxylic groups to the end caps to form direct chemical bonding between nanotubes [19].
Chemical functionalization of CNTs is an effective way to prevent nanotubes from aggregation. This approach supports a better dispersion and stabilizes the CNTs within the polymer matrix [20,21]. Hu et al. [22] studied the effect of curing process and mixing time on electrical properties for a mixture of polymer matrix and multiwalled CNTs. They identified the curing temperature and mixing conditions as key factors for the formation of a conducting network.
Jaemin Cha et al. [23] investigated the changes in mechanical properties of CNT/epoxy nanocomposites through functionalization of CNTs. The results indicated improvements in tensile strength, stiffness and fracture toughness with the addition of 2 wt% of functionalized CNTs. In addition, the behavior of crack propagation was studied as shown an improvement in fracture toughness.
Researchers have been using mechanical adhesion/compatibility methods to disperse the CNTs in a polymer matrix. For example, Kim et al. [24] used high-energy mechanical sonication to disperse CNTs in a polymer matrix to fabricate and characterize the behavior of CFRP epoxy composites. Sonication time has been observed as affecting the mechanical, electrical and thermal properties of the nanocomposites. Optimized sonication time maximized the CNTs dispersion, stabilization and damage minimization [25].
In another study, Gojny et al. [26] investigated the importance of surface functionalization on the mechanical performance of epoxy nanocomposites for different CNTs and noted that functionalized nanocomposites had improved strength, stiffness and fracture toughness. A sufficient stress transfer from polymer matrix to the CNTs is essential to achieve the potential of CNTs as reinforcement. The interfacial bonding between CNT and polymer matrix can be enhanced by functionalizing the CNT surface.
Almuhammadi et al. [27] worked on enhancing the delamination resistance of unidirectional CF composite laminates using functionalized MWCNTs. It was noted that the introduction of functionalized MWCNTs affected the interlaminar fracture toughness of the laminates. To improve the fracture toughness of quasi-isotropic CFRP composite laminates, Kostopoulos et al. [28] used modified MWCNTs and subjected the laminates to low-velocity impact using a drop tower. The results indicated that the inclusion of MWCNTs into the polymer matrix improved the fracture toughness under Mode I and Mode II loadings. Moreover, incorporation of MWCNTs, graphene oxide and CF into the epoxy enhanced the Modes I and II interlaminar fracture toughness [29,30]. Double cantilever beam (DCB) and End notched flexure (ENF) tests were conducted. The experimental results showed improvement of the interlaminar fracture toughness of the laminates.
Guo et al. [31] studied the electrical conductivity of laminated composites used in aircraft industries. They prepared butyl glycidyl ether (BGE) and MWCNTs, and then dispersed it in an epoxy resin. The mixture of MWCNT-BGE/epoxy was transferred into the interlayer of CF/epoxy composite laminates using spray techniques. The electrical conductivity of the through-thickness has improved by 283%. Additionally, authors recommended that this method of laminate preparation is important for improving the ILSS. These results indicated that functionalization of CNTs, mixing speed, curing temperature and mixing rate of polymer matrix containing multiwalled nanotubes need further research to investigate their effect on the mechanical properties of the nanocomposites and also to optimize these properties.
In this study, the improvements in short beam shear (SBS) and flexural properties of multiwalled CNTs/CF/epoxy laminates are studied experimentally as a function of constant mixing speed and variable mixing time. An epoxy resin is reinforced with modified and non-modified MWCNTs using overhead stirrer at 2000 rpm for 24 and 96 h in steps of 24 h. All laminates contain 0.3 wt% of MWCNTs with respect to the epoxy resin. ILSS, flexural strength, flexural modulus, failure strain and fracture behavior of the laminates are examined using three-point bending tests, short beam tests and scanning electron microscope (SEM). Dispersion states of MWCNTs in the polymer matrix are morphologically characterized by the Fourier transform infrared (FTIR) and transmission electron microscopy (TEM).

Materials
The commercial unidirectional CF (Toray-300), prime 27 LV, epoxy resin and prime 27 LV slow hardener were purchased from AMT composites, South Africa (SA). The physical and mechanical properties of the materials are shown in Table 1.
Commercially available multiwall CNTs (outer diameter: 10-20 nm, length: 3-8 mm, 99% purity) were purchased from Capital Lab Supplies, Cape Town, SA. Dimethylformamide (DMF) was purchased from United Scientific, Cape Town, SA. The effect of non-modified (untreated) and modified (treated) MWCNTs on the properties of CFRP composites is the subject of this work. The production process of acid-modified multiwall CNTs is described next. Five grams of multiwall CNTs were refluxed at 100 C for 7 h in HNO 3 (200 mL, 55% concentration). After cooling to the ambient temperature, the carboxylic functionalized (MWCNTs-COOH) was washed several times using distilled water until neutralization and dried in an oven at 80 C for 12 h before use. Centrifugation at 3000 rpm for 5 min was applied for solid-liquid separation. The refluxing process is presented in Figure  1(a). Silicon oil bath was used during the refluxing process to maintain a proper heat transfer. Subsequently, the modified and non-modified MWCNTs were dispersed in 30 mL DMF and 100 mL acetone solution in a flask for 120 min and finally, the MWCNT solution was centrifuged at 2000 rpm for 5 min.
FTIR spectroscopy is the best technique to characterize the functionalities present in the structure of MWCNTs and epoxy resin. Treated and untreated MWCNTs were mixed with epoxy resin at 2000 rpm for 24, 48, 72 and 96 h to examine the chemical structures with FTIR spectrum.

Dispersion procedures of MWCNT and preparation of laminates
Dispersion agents were used for dispersion of nontreated and treated multiwall CNTs in the epoxy as shown in Figure 2. Treated (MWCNTs-COOH) and non-treated (MWCNTs) were prepared and drawn into a beaker to mixed with the epoxy. The epoxy was heated at 60 C for 30 min to reduce viscosity before adding the multiwall CNTs. The non-modified MWCNTs and epoxy subjected to overhead stirrer with the rotational speed of 2000 rpm for 24, 48, 72 and 96 h. Similarly, the modified MWCNTs and epoxy were subjected to overhead stirrer at a rotational speed of 2000 rpm for 24, 48, 72 and 96 h. Later, the dispersed material was degassed for about 30 min to remove the entrapped air from the inside of the mixtures. The hardener and the epoxy were mixed using an overhead stirrer with a rotational speed of 500 rpm for about 15 min, then degassed for 10 min.
CFRP laminates were prepared using resin transfer molding (RTM) method. Fourteen layers of unidirectional carbon fabric were laid up in the unidirectional orientation considering testing standards. The epoxy material to hardener ratio was 10:2.8. The MWCNTs/epoxy mixtures were infused Characterization SEM and FTIR analyses were conducted to study the fracture behavior and microstructure of laminates. For SEM images, all specimens were coated three times with gold sputter. TEM was used to examine the dispersion quality and distribution of treated and untreated MWCNTs in the epoxy. The  ILSS, flexural strength (r f ), flexural modulus (E f ) and strain at failure (e f ) values were measured using a Lloyd LR test machine based on ASTM D 2344 and ASTM D7264/D7264M-07 standards [33,34].
At least five samples were tested for SBS and flexural tests at a constant crosshead speed of 1 mm/ min. The setups to characterize the laminates are shown in Figure 3. In the SBS test, the ILSS was determined based on classical (Bernoulli-Euler) beam theory. The maximum ILSS for rectangular laminate occurs at the mid-thickness and calculated to be where P max is the maximum load and b and h are the width and thickness of the laminates. The spanto-depth ratio for the SBS test for the samples of L=h ¼ 4 was considered. The average span, width and thickness of the samples were around 18, 9.4 and 4.5 mm. The flexural strength and strain at failure on the composite laminates can be estimated by where D is the deflection of the laminates. We selected a span to thickness ratio for the laminates of L=h ¼ 32: The span, width and thickness of the flexural specimens were around 145, 13 and 4.5 mm, respectively. The stiffness of the laminates was estimated from r f and e f values, respectively.

Results and discussion
Characteristics of treated and nontreated MWCNTs Figure 4 shows the FTIR spectrum of non-treated MWCNTs/epoxy at different mixing times. In the case of non-treated MWCNTs, sharp peaks and weak signal at 3459 cm À1 corresponded to 0 À H stretching bands. Further, the peaks at 2971 and 2868 cm À1 contributed to the presence of C À H stretching vibrations. In this case, broader signals appeared. The peaks at 2134 cm À1 linked to the CC stretching bands. The spectrum of non-treated MWCNTs/epoxy shows the highest intensity and sharp peaks at about 1739 cm À1 , corresponding to C ¼ O stretching vibration. The peaks at 1604 cm À1 correspond to C ¼ C stretching bands. In all cases, the absorption bands within each functional group slightly varied as a function of the mixing times as shown in Figures 4 and 5. This indicated that mixing time might affect conducting network between each atom and may lead to more stretching [35]. The treated surface of the MWCNTs-COOH mixed with the epoxy resin was examined with FTIR spectra as shown in Figure 5. The band at 3457 cm À1 is related to stretching vibration of O À H groups. As can be seen from treated FTIR spectra, the stretching vibrations of C À H group appear between 2927 and 2867 cm À1 . Sharp signal and medium intensity at 1606 cm À1 linked to C ¼ C stretching vibration. The FTIR observations on treated MWCNTs confirm that the shape and signal intensity could be linked to bonding of MWCNTs-COOH with the epoxy resin. Functionalization of MWCNTs improved the bonding with functional groups and this may contribute to better behavior and interfacial strength.

Behavior of MWCNTs with respect to dispersibility in epoxy resin
Dispersing of nanotubes in a polymeric matrix is a challenging process. In order to homogeneously disperse the treated and non-treated MWCNTs in the epoxy, high sonication and stirring times, including a large energy input, are needed to reduce the van der Waals forces of attraction between the nanotubes [26]. High energy due to sonication may cause reduction in the effective length and may also cause damage to the nanotube structures. In this study, we use the same sonication methods for treated and non-treated multiwall CNTs, but different stirring times to disperse epoxy evenly. TEM is used to characterize the distribution of treated and nontreated multiwall CNTs within the matrix. TEM images of non-treated MWCNT/epoxy are shown in Figure 6(a-d). As it can be seen from Figure 6(a-c), MWCNTs in epoxy have a low dispersibility. The figures indicate a high level of agglomeration of multiwall CNTs. This may have occurred due to strong forces of attraction between MWCNTs. The dispersion of multiwall CNTs is observed to be slightly better in Figure 6(d). This may be due to the higher stirring time.
TEM images of the treated MWCNTs/epoxy are shown in Figure 7(a-d). Functionalization of MWCNTs leads to an improved mixing with the matrix. As shown in Figure 7, better dispersion is obtained on treated MWCNTs/epoxy as a function of the stirring times. This may be due to better bonding of MWCNTs-COOH/epoxy which leads to better-conducting networks. The levels of dispersion using sonication and using overhead stirring methods varied the dispersibility of treated multiwall CNTs as a function of the mixing rate. As shown in the TEM images (Figure 7), the dispersion and nanotube damage (Figure 7(d)) increased as the mixing time increased.

Short beam shear (SBS) test
The variation of apparent ILSS for CFRP, untreated MWCNTs-CFRP and treated MWCNTs-CFRP laminates at different stirring times is shown in Figure 8 and Table 2. The ILSS of the control laminate is 58.7 MPa. Addition of MWCNTs to epoxy improves the matrix-dominated behavior of the material. For the case of untreated MWCNTs-CFRP, the ILSS remains the same at 61.9 MPa with stirring times of 24 and 96 h. However, the ILSS of treated MWCNTs-CFRP increases to 66.7 and 62.8 MPa at the specified stirring times. The maximum increase in ILSS was obtained for treated MWCNTs-CFRP laminates as compared to the other cases. The change of ILSS with CFRP laminates was about 14 and 7%. The ILSS of treated MWCNTs-CFRP show increases of 8 and 2% as compared to untreated MWCNTs-CFRP laminates at stirring times of 24 and 96 h. As indicated by the experimental results, incorporation of MWCNTs into the epoxy matrix improved the ILSS of the laminates. However, the ILSS of the laminates decreased with stirring time. This may be due to nanotube damage and loss of nanotube network formation. In this study, the optimal stirring time is specified as 24 h.

Three-point bending test
For a span-to-depth ratio of L=h ¼ 32, the stressstrain plots of control CFRP, untreated MWCNTs- CFRP and treated MWCNTs-CFRP laminates are shown in Figure 9. The effect of the incorporation of MWCNTs on flexural properties of unidirectional CFRP laminates can be observed in Table 2. CFs have high strength and stiffness but low strain properties as compared to glass fibers [36]. After the incorporation of MWCNTs into the epoxy matrix, the strain properties of MWCNTs-CFRP laminates improve by about 21.6% compared to control CFRP laminates. Thus, reinforcing by multiwalled       Figure 11. From Figure 11(a), it can be seen that longitudinal delamination between fibers and rupture of epoxy matrix occurs in the case of the control laminate. In the case of treated MWCNT-CFRP laminate (Figure 11(b)), voids with discontinuous cracks are observed. Discontinuous cracks may have happened due to the incorporation of treated MWCNTs in the epoxy matrix which can lead to changes in ILSS properties between the layers. SEM images of control and treated MWCNTs-CFRP laminates after failure under three-point bending test are shown in Figure 12. The cross-sectional image of the fracture surface in Figure 12(a) shows that a straight crack propagation can be observed in the case of the control laminate. Intermittent voids and cracks were detected in the case of treated MWCNTs-CFRP laminates ( Figure  12(b)). This may have happened due to the improvement in the bonding strength between the fibers and the epoxy matrix as a result of addition of MWCNTs to the epoxy matrix.

Conclusions
Properties of CNT-based composites are important in many fields of engineering. This work presents a study on flexural and shear properties of untreated and treated MWCNTs-CFRP with a control CFRP composite laminates. The mechanical properties are studied using three-point bending and SBS tests with the flexural strength, modulus and strain of the test specimens compared to assess the effect of MWCNT reinforcement. The test specimens were prepared with the incorporation of CFs into a multiwall CNTs reinforced epoxy matrix. The epoxy resin was modified with 0.3% wt of multiwalled CNTs at a mixing speed of 2000 rpm and at mixing times of 24, 48, 72 and 96 h. MWCNT/epoxy was infused into unidirectional carbon fabric using RTM process. Based on the test results of the nanotube enhanced laminates and control CFRP laminates, Figure 11. SEM images of control (a) and treated MWCNT-CFRP (b) laminates after SBS test.
the following observations and conclusions can be drawn: