Investigation into the toughening mechanism of epoxy reinforced with multi-wall carbon nanotubes

2.1 Materials

The chemical reagents and raw materials used in this study were epoxy (Epikote-828, supplied by Momentive Specialty Chemicals, USA), maleic anhydride 4,4′-diaminodiphenymethane (DDM) and hydrochloric acid supplied by State Drug Chemical Corporation (Shanghai, China). The MWCNTs (purity >95%, diameter 30–50 nm and length 10–20 μm) were purchased from Nanoport Co. Ltd. (Shenzhen, China).

2.2 Sample preparation

The MWCNTs were initially purified and then dispersed in 25 wt.% of hydrochloric acid by ultrasonication for 30 min. Thereafter, they were dispersed in hydrochloric acid and then diluted with deionised water to a pH value of 7, filtered and dried. Subsequently, the MWCNTs were calcinated in a muffle furnace at 600°C for 1.5 h to remove the amorphous carbon, followed by burning at 950°C for 2 h in N₂atmosphere to remove the surface oxide.

The epoxy control samples were prepared as follows: The epoxy was outgassed under vacuum at 100°C for 30 min. Then, the epoxy polymer was mixed with 28 phr (part per hundred parts of resin by weight) of DDM at 100°C under vacuum, outgassed and stirred for 15 min. Subsequently, the resulting mixture was poured into a silicon rubber mould and cooled down to 80°C for 2 h and then cured at 150°C for 4 h.

Furthermore, MWCNT/epoxy composites were prepared by the following method. Firstly, MWCNTs were added to epoxy resin and the resulting mixture was shear mixed for 1 h and ultrasonicated at 75°C for 4 h. Then, the mixture was outgassed for 30 min under constant stirring at 100°C. Subsequently, 28 phr of DDM was added and then outgassed in a vacuum pump for 30 min under stirring. The resulting mixture was then cast into a silicon rubber mould. The curing condition was the same as that adopted for the epoxy control sample.

2.3 Tensile measurement

Dumbbell-shaped samples of thickness 3.0±0.4 mm, working width of 13 mm and gauge length of 25 mm were obtained by pouring the composite mixture into a silicon rubber mould. Uniaxial tensile tests were conducted in accordance with the ASTM-D638-08 standard, using a Zwick/Roell 5-T universal testing machine, at a displacement rate of 5 mm/min and a test temperature of 20°C. The displacement over the gauge length of the samples was measured using an extensometer. The maximum tensile stress for each sample was recorded, and the elastic modulus (E) was calculated between strains of 0.05% and 0.25%. At least five replicate specimens were tested for each formulation.

2.4 Measurement of fracture toughness

The critical stress intensity (K_IC) of the hybrid and control samples was obtained by performing the single-edge-notch three-point bending (SEN-3PB) test, according to the ASTM-D5045-99 standard. The dimension of the test specimens was typically 127×12.7×6.35 mm. A sharp pre-crack, exhibiting a thumbnail-like crack front, was generated in each sample by tapping it with a fresh razor blade chilled in liquid nitrogen. The tests were performed at room temperature using a Zwick/Roell 2.5-T universal testing machine. The testing crosshead speed was chosen as 0.508 mm/min, and K_IC was calculated by the following equation:

where P is the critical load, B is the sample thickness, W is the specimen width, α is the average pre-crack length and f(α/w) is the non-dimensional shape factor (23). Similarly, the fracture energy, G_IC, was calculated as follows (23):

where E is the tensile modulus and ν is the Poisson’s ratio with a value of 0.35, which is typical for epoxy polymers (24).

2.5 Investigation of the toughening mechanisms

The fracture surfaces of the samples, obtained as a result of the SEN-3PB test, were analysed by scanning electron microscopy (SEM; Quanta FEG450, FEI Co., Ltd., USA) using an accelerating voltage of 3 kV. Prior to the analysis, all the samples were sputter-coated with a thin layer of gold.

The double-notch four-point bending (DN-4PB) technique was used to investigate the sequence of events that had occurred during the fracture process. The tests were conducted using the same parameters as those of SEN-3PB, according to the ASTM-D5045-99 standard. An ultra-thin section perpendicular to the fracture plane and parallel to the crack direction was obtained by ultramicrotomy and then placed on a 200-mesh holey Cu/carbon-coated grid for TEM analysis (TGF30, FEI), at an accelerating voltage of 300 kV.

3.1 Tensile properties

Table 1 shows the tensile strength and modulus of the epoxy control sample and the MWCNT/epoxy composites (MWCNT content varied from 0.1 to 1.25 wt.%), obtained by using the mean values calculated from reproducible results.

Neat epoxy possesses a tensile strength of about 60.0 MPa and an elongation at break of 4.3%, with an elastic modulus of about 2.67 GPa. As can be seen in Table 1, a small amount of MWCNTs (<1 wt.%) can improve the tensile strength, elastic modulus and elongation at break of the composites, suggesting the enhancement in strength and toughness simultaneously. With an MWCNT content of >0.7 wt.%, the tensile modulus showed a downward trend, which may due to the poor dispersion of the nanotubes in the composite. These nanotube bundles may serve as stress concentrators or defects, affecting the mechanical performance of the composite (25).

3.2 Fracture toughness

The quasi-static fracture toughness (K_IC and G_IC) of the epoxy with different amounts of MWCNTs is shown in Figure 1. It is apparent that the fracture toughness increased with increasing MWCNT content of up to 0.7 wt.%. The neat epoxy polymer is very brittle, with mean K_IC and G_IC values of 0.78 MPa·m^1/2 and 149 J/m², respectively, which is typical of a brittle thermosetting polymer (26). The fracture toughness of epoxy improved by adding MWCNTs; the highest K_IC and G_IC mean values of 0.99 MPa·m^1/2 and 235 J/m², which were observed for the composite with 0.7 wt.% of MWCNTs, increased by 27% and 58%, respectively. With further increase in MWCNT content, the fracture toughening effect decreased. As mentioned earlier, this might be due to the poor dispersion of MWCNTs in the epoxy matrix.

Figure 1:

Fracture toughness of epoxy plotted as a function of MWCNT content.

3.3 Morphology of the fracture surface

The reason underlying the aforementioned trend in fracture toughness was further investigated by analysing the fracture morphology in the broken samples obtained from the SEN-3PB test. A comprehensive understanding was gained by exploring the fracture surfaces by SEM (Figure 2) and the high-magnification SEM images (Figure 3).

Figure 2:

SEM images of the fracture surfaces of (A) an unmodified epoxy polymer, (B) an epoxy polymer containing 0.3 wt.% of MWCNTs and (C) an epoxy polymer with 0.7 wt.% of MWCNTs (crack propagation is in the direction of the arrow).

Figure 3:

High-magnification SEM images revealing the MWCNT pull-out and debonding in the crack tip of (A) an epoxy polymer with 0.1 wt.% of MWCNTs and (B) an epoxy polymer with 0.3 wt.% of MWCNTs [the pull-out is shown by arrows in (A), and the plastic void of debonded MWCNTs is circled in (B)].

As can be seen in Figure 2A, the fracture surface of neat epoxy is very smooth, except for several river-like lines, indicating that no large-scale plastic deformation had occurred during fracture. This reveals the brittle nature of neat epoxy and its weak resistance towards crack initiation and propagation. However, the composite containing 0.3 wt.% of MWCNTs (Figure 2B) showed the occurrence of crack branching, indicating the crack extension was obstructed by MWCNTs. The composite containing 0.7 wt.% of MWCNTs exhibited a more roughened surface with highly intense river-like lines (Figure 2C). Meanwhile, stress whitening zone could be found on the fracture surface. The surface roughness increased with increasing MWCNT content, suggesting that the crack propagation in the composites was suppressed. This is because more energy was needed for the segments of the primary crack front to bend between the nanotubes, leading to an increase in fracture toughness. Therefore, plastic deformation of the epoxy around the clusters of MWCNTs may be one of the toughening mechanisms in MWCNT/epoxy composites.

In Figure 3A, some MWCNTs were pulled out of the fracture surface. This indicates the bridging mechanism resulting from debonding and sliding resistance, which contributes to the overall strengthening and toughening of the composite. The debonding process is generally considered to absorb less energy when compared to the plastic deformation of the matrix. However, debonding is essential, as it reduces the constraint at the crack tip and allows the matrix to deform plastically via a void growth mechanism.

Nanoparticle debonding and subsequent plastic void growth have been considered to be two of the major toughening mechanisms for nanoparticle/epoxy composites (27). As circled in Figure 3B, the presence of voids was initiated by debonded MWCNTs. The measured diameter of those voids was between 60 and 90 nm, which was larger than that of MWCNTs (35–50 nm) prior to embedment. This demonstrates the occurrence of plastic void growth in the epoxy polymer, initiated by the pull-out of MWCNTs. A similar observation was also reported by Lachman and Wagner (14), who suggested that the plastic void growth can be due to epoxy coating. More elaborately, a cylindrical bulk of epoxy surrounding MWCNTs was also pulled out of the matrix, together with the MWCNTs. Consequently, the stress concentrations in the neighbouring MWCNTs allowed the dissipation of energy. These results imply that the plastic void growth mechanism may be partly responsible for the observed increase in the fracture toughness of the CNT-reinforced epoxy systems.

To gain further insight on the role of MWCNTs, the interior of the crack tip in the thin section of DN-4PB samples with sub-critical cracks was observed by TEM. In Figure 4A, statically bending the specimen resulted in two almost identical notches on one side. One of the two cracks would reach the critical state first and run unstably to failure, while the other crack, though initiated, would be unloaded and arrested with a sub-critical damage zone (Figure 4B). A convenient block size was divided into a section normal to the fracture surface but parallel to the crack direction and encapsulated in epoxy formulation cured at room temperature. The block was further trimmed down to a trapezoid shape, with the damage zone roughly placed at its centre. The small end of the block was approx. 0.3 ×0.3 mm. An ultra-thin section perpendicular to the fracture plane and parallel to the crack direction was obtained by ultramicrotomy (Figure 4C) and observed by TEM.

Figure 4:

DN-4PB geometry and the sub-fracture damage region for TEM.

Figure 5 shows the TEM image of a part of the DN-4PB sub-fracture damage zone of the epoxy with 0.3 wt.% of MWCNTs. The fractured MWCNTs, not pulled out during the crack propagation, could be observed, as indicated by arrows in Figure 5B. This signifies the strong bonding of the MWCNTs with the matrix. It has been reported (28) that stronger MWCNT/matrix interfaces do not necessarily imply that the overall fracture toughness of the composites is better. It is the optimal interface that makes the failure mode just like in the transition from nanotube pull-out to break. The bridging effect of fractured and pulled-out MWCNTs, which causes the plastic void growth phenomenon mentioned earlier, may be responsible for the epoxy toughening effects. Meanwhile, the bridging MWCNTs prevented the propagation of the crack tip, as can be seen in Figure 5D. MWCNT bridging mechanisms, including both pull-out and break, could favour energy consumption and thus contribute to the improvement in fracture toughness of the composites.

Figure 5:

TEM images of the sub-critical crack tip of the MWCNT/epoxy systems formed as a result of the DN-4PB test, indicating (A) a crack growth, (B) a crack propagation area, (C) a crack tip and (D) a high-magnification image of (C). White arrows indicate the direction of the crack growth; black arrow indicates the fractured MWCNTs.

3.4 Modelling fracture toughness

Till date, many authors have observed that the addition of particles to a brittle material increases the measured fracture energy. The fracture energy of a particle-modified polymer may be expressed (29) as

where G_ICc is the fracture energy of the modified polymer, G_ICm is the fracture energy of unmodified epoxy and ψ is the total additional energy dissipated per unit area. Based on the toughening mechanism analysis discussed previously (30), the value of ψ can be expressed as

where ΔG_s is the energy contribution from the localised shear banding, ΔG_v is the energy contribution from the plastic void growth and ΔG_r is the energy contribution from the MWCNT bridging mechanism. The terms ΔG_s and ΔG_v can be calculated as follows:

where μ_m is a material constant, Vfv is the volume fraction of the voids, Vfr is the volume fraction of the debonded particles, Vf is the volume fraction of the particles, σ_yc is the compressive yield stress, γ_f is the shear fracture strain of the epoxy matrix and r_yu is the radius of the plastic zone of the unmodified epoxy polymer.

The value of μ_m, which describes the pressure sensitivity of the material in the von Mises yield criterion, has been reported to be between 0.175 and 0.225, and is normally taken to be 0.2, as reported in Refs. (29, 31). γ_f is taken to be the same number as in Refs. (29, 31), which is 0.71. The maximum stress concentration factor, K_vm, can be set to 2.22 around a void in an epoxy matrix (29, 31). In order to obtain V_fv-V_fr, the size of the voids was estimated from the scanning electron micrograph, according to the method provided by Liang and Pearson (32); the average diameter of the voids was 80 nm (based on the voids circled in Figure 3B), and V_fv-V_fr was about 0.1V_f.

In addition, for the current investigation, the value of experimentally determined tensile yield stress σ_yt was substituted in the following equation to derive σ_yc (30):

The size of the plastic zone ahead of the crack tip can be calculated by assuming a linear-elastic fracture-mechanic behaviour. Under plane-strain conditions and assuming that the zone is circular, as proposed by Irwin (33), the radius of the plastic zone, r_yu, is given by

Table 2 summarises the volume content of MWCNTs, the corresponding compressive yield stress calculated by Equation 5 and the radius of the plastic zone calculated by Equation 6.

By comparing with experimental data, we can calculate the term ΔG_r as follows (27):

where σ_cnts is the strength of the MWCNTs (σ_cnts is approx. 130 MPa for typical pristine MWCNTs), r_cnts is the radius of the MWCNTs (in the present study, the average r_cnts was ~20 nm) and τ_i is the interfacial shear strength between the MWCNTs and the epoxy matrix (τ_i is approx. 30 MPa for typical pristine MWCNTs) (14).

Figure 6 shows the experimental values (symbols) of fracture energy compared with the model predicted values (lines) for MWCNT/epoxy composites. The energy contribution calculated from each toughening mechanism is also presented separately. The solid line represents the synergetic contribution of the three toughening mechanisms. The dashed lines represent the model prediction for the contribution of shear banding, bridging mechanism and plastic void growth. Considering the energy contributions from each of the toughening mechanisms, shear banding ΔG_s appears to be the dominating mechanism underlying the enhancement of fracture toughness in epoxy reinforced with low content of MWCNTs. In addition, the predicted values agreed well with the experimental data for the lower MWCNT contents (MWCNT contents <0.7 wt.%). The deviation between the model and the experimental results can be attributed to the following reasons. An underestimation in predicting the increased fracture toughness for a MWCNT content of <0.5 wt.% can be expected owing to the inaccurate measurements of the volume fraction particle debonding used in plastic void growth calculation. Meanwhile, the energy contribution from plastic void growth and bridging mechanism is much smaller when compared with that from shear banding mechanism. However, plastic void growth and bridging mechanisms are also important since the matrix shear yielding mechanism is induced by MWCNT debonding. With an increase in MWCNT content of up to 0.7 wt.%, the aggregation of MWCNTs becomes critical, which may become a defect causing the deterioration of toughness, similar to the MWCNT strengthening effect in the present study.

Figure 6:

Experimental data of fracture energy and the corresponding predicted values according to Equation 4, by considering shear banding, plastic void growth model and bridging mechanism.