Enhancement of interface strength

Many studies have been conducted to determine the type of bonds formed between CNT and epoxy. The general conclusion is that CNT bond in three main ways: micromechanical interlocking, chemical bonding, and van der Waals bonding. While the CNT surface is quite smooth, it has been proposed that there are local non-uniformities in the CNT such as kinks, bends, and changes in diameter. It is at these local non-uniformities where micromechanical interlocking occurs [Wong, at al., 2003]. Chemical bonding is possible, but it is not guaranteed [Shadler, et al., 1998]. Finally, van der Waals bonding certainly occurs, but a relatively weak bond forms. One study also proposes the effects of thermal properties. The


Enhancement of interface strength

coefficient of thermal expansion of CNT is much higher than that of the polymer matrix. As a result, residual compressive thermal stress is present after the polymer matrix hardens. This thermal stress results in closer contact between the CNT and polymer, which in turn increases micromechanical interlocking and non-bond interactions [Wong, et al., 2003].

In an attempt to improve the joint interface strength, CNTs are introduced at the interface. Since the scarf joint is constructed using the two-step process, CNTs are applied to the joint surface area after the base portion is cured, sanded, and cleaned, i. e. after Fig. 2(b). The CNTs are mixed with acetone and the solution is stirred until the CNTs are dispersed in the solution. The CNT solution is manually spread out over the joint interface area and acetone evaporates completely leaving CNTs behind. Therefore, it is not possible to have very uniform distribution of CNTs over the joint surface area. First, two different surface concentrations of CNTs are used. For this part, 95% pure Multi-Wall Caron NanoTubes (MWCNTs) with a length of 1-5 microns and a diameter of 15 +/-5nm are selected. The tested surface concentration levels are 11.5g/m2 and 7.5g/m2, respectively. Compressive strength tests are conducted for the specimens without CNTs as well as specimens with two different surface concentrations of MWCNTs. The result is plotted in Fig. 16. The data shows CNTs improve the joint strength when compared to the no CNT case. In addition, the CNT surface concentration of 7.5g/m2 yields a greater strength than the higher concentration case.

Enhancement of interface strength

Fig. 16. Comparison of failure strength with two different surface concentrations of CNTs

For subsequent studies, the surface concentration of 7.5g/m2 is selected. The next study uses various kinds of CNTs at the joint interface. Single-Wall Carbon NanoTubes (SWCNTs), MWCNTs with two different diameters and two different lengths, and bamboo-structure MWCNTs with two different lengths are considered as listed in Table 1. Figure 17 compares the strength of scarf joints made of different CNTs.

Each CNT group provides a joint strength increase, compared to the non-reinforced specimens, with the exception of group C. The CNTs from Group C is the economical one which has less quality control. The greatest strength increase is observed by Groups D, E, and F. All three of those groups demonstrate an average strength increase of greater than 11 percent. Of these three groups, it appears that Group D possesses the best strength enhancement characteristics. It has greater than an 11 percent increase in strength and possesses the most consistent data of the three top reinforcements. This consistency can be seen by observing the standard deviations shown in Fig. 17.

Groups E and F are bamboo-type CNTs. They have regularly occurring compartment-like graphitic structures inside the nanotubes similar to the bamboo plant [Ding, et al., 2006]. These types of CNTs are used with the notion that the compartment-like graphitic structures may provide additional support when used for reinforcement. The open ended molecular structure of the multi-walled bamboo CNT may increase wettability and functionalization as well. This may allow for increased interfacial bonding which in turn increases the load transfer between the resin and the CNT so that the joint interface strength of the composite structure may be improved. The strength increase confirms that the bamboo structure has better strength characteristics compared to conventional CNTs of similar size and purity. Group B, the economic option, has some samples that provide strong reinforcement and others that are actually weaker than the non-reinforced specimens. As a result, the average strength is greater than the non-reinforced samples, but the standard deviation is quite large. The standard deviation of group B is almost 30 percent larger than any other group. All MWCNT groups are 95% pure, but perhaps the economic option encounters a higher frequency of defects.



Multiwall carbon nanotubes, outer diameter 30 +/-15nm, Length 1-5 microns, Purity > 95%


Multiwall carbon nanotubes, outer diameter 25 +/-5nm, Length 10-30 microns, Purity > 95%


Multiwall carbon nanotubes, outer diameter 15 +/-5nm, Length 5-20 microns, Purity > 95%


Multiwall carbon nanotubes, outer diameter 30 +/-15nm, Length 5-20 microns, Purity > 95%


Bamboo structure multiwall carbon nanotubes, outer diameter 30 +/-15nm, Length 1-5 microns, Purity > 95%


Bamboo structure multiwall carbon nanotubes, outer diameter 30 +/-15nm, Length 5-20 microns, Purity > 95%

Table 1. Different types of multi-wall carbon nanotubes

The majority of test samples fracture at the expected location along the diagonal step interface of the joint as shown in Fig. 6(a). Those test joints initiates cracks at either the base of the bottom step or at the center of the joint and propagate diagonally along the joint interface. There is another type of fracture that rarely occurs, where the crack propagation does not follow the path of the joint interface. The crack initiate at the base but instead of propagating along the interface, it propagates at a 45 degree angle away from the interface as shown in Fig. 6(b).

Figure 18 shows another line of fracture under compression. In this case the fracture follows the path of the undulated fiber section instead of the scarf interface. Figure 18 shows a test specimen that failed along the alternative fracture line. Only one group has consistently these types of fractures. Group D has every test joint failed along this line of fracture. This group also happens to have the most consistent strength enhancement and the highest elastic modulus of the three top CNT reinforcements.

Enhancement of interface strength

Fig. 17. Comparison of failure strength with different CNTs

A potential explanation for the consistency of this failure in Group D is that the CNTs provide enough enhancements in strength along the joint interface that the interface ceases to be the weakest portion of the specimen. Instead the samples fail along second weakest portion of the joint, the undulated section of the overlap construction. The mode of this type of failure is localized fiber buckling. Normally this type of failure is intermittent. The consistency in Group D suggests the joint may be reinforced enough to make it stronger than the stress required to cause the localized buckling failure at the location of the fabric down step.

Line of fracture

Enhancement of interface strength

Enhancement of interface strength

Fig. 18. Another facture mode under compression

In order to further understand the effect of CNTs on the strength of the scarf joints, individual fracture testing of Mode I and Mode II, as shown in Figs. 4 and 5, is conducted, respectively. This set of tests explains on what fracture mode the CNTs affects to improve the interface joint strength [Faulkner, et al., 2009].

Crack opening mode (i. e. Mode I) testing results shows a modest improvement in the critical energy release rate GI when the joint interface is reinforced with CNT. Figure 19 compares the average values of normalized Gi for resin only samples (i. e. without CNT reinforcement) and CNT reinforced samples. Standard deviation is also shown in the figure. The average GI value increases about 10% with CNT reinforcement. However, Mode I crack propagation characteristics were observed with no discernable difference between the CNT reinforced and non-reinforced samples. The Digital Image Correlation (DIC) System was used to observe the crack growth in both CNT reinforced and non-reinforced specimens and their images are very similar. After testing, the samples are fully broken to inspect the cracked surface. Mode I samples reveal little difference between CNT reinforced and non-reinforced samples. Both CNT reinforced and non-reinforced samples have crack growth through the resin layers where the initial cracks are located.

Enhancement of interface strength

Fig. 19. Normalized GI values for Mode I

Mode II (i. e. shearing mode) testing results in a significant increase in the critical energy release rate GII for the samples reinforced with CNT. Figure 20 shows the normalized average values of GII for the specimens. Again, standard deviation is also shown in the figure. As shown by the standard deviation, the lowest CNT reinforced value is higher than the highest non-reinforced value. The average CNT reinforced GII value is 32% higher than the average resin only GII value. The GII values are computed from the compliance of the load vs. displacement curves. Representative plots of load-displacement are shown in Figs. 21 and 22. The point of crack propagation is marked with an X.

Qualitatively, the observed crack propagation for Mode II is significantly different between the CNT reinforced and non-reinforced samples. For the non-reinforced samples, crack propagation begins at the tip of the initial crack and continues through the interface resin material. However, for CNT reinforced samples, a crack begins to nucleate away from the initial crack tip, perhaps in an area of lower CNT concentration, i. e., a weaker strength zone. Eventually, this newly formed crack grows to be connected to the initial crack. This result is widely observed in the CNT reinforced samples.

After testing, the samples are fully broken to inspect the cracked surface. For the non­reinforced samples, the joint interface bond is broken through the resin while in others the resin is pulled away from the fibers. However, the CNT reinforced samples fail much differently. The CNTs reinforce the resin at the interface, making it stronger. The CNTs themselves do not fracture. The CNTs bond to the resin, blocking crack propagation. As a result, at some locations, the crack propagates through the fibers rather than through the resin. The critical energy release rate for CNT reinforced samples become higher because the crack propagates through the carbon fibers vice resin.

Representative images from the DIC system are shown in Fig. 23. Without CNT reinforcement at the joint interface, the initial crack propagates through the interface all the way as shown in Fig. 23(a). On the other hand, CNT reinforcement resulted in a tougher joint interface so that the crack path deviates away from the joint interface as seen in Fig. 23(b).

Enhancement of interface strength

Fig. 20. Mode II Normalized Gn Values

Enhancement of interface strength

Fig. 21. Representative load versus extension plot for Mode II testing of non-reinforced sample (The point of crack propagation is marked with an X.)

CNT reinforcement is more significant for Mode II fracture than for Mode I. A possible explanation is given below. The CNTs are not believed to have a strong chemical bond with the resin material. Instead, CNTs are considered to be entangled with resin polymer chains, called a mechanical interlocking. Such a mechanical interlocking is more effective to resist the shearing force of Mode II than the normal force of Mode I. Therefore, the fracture toughness of Mode II becomes much higher with CNT reinforcement.

Enhancement of interface strength

Enhancement of interface strength

(b) With CNT reinforcement

Fig. 23. Images of crack growth: (a) Without CNT reinforcement, the crack propagated through the joint interface plane. (b) With CNT reinforcement, the crack path showed deviation away from the joint interface.


Another purpose of testing is to optimize the surface concentration of CNT, i. e. the mass of CNT per unit CNT-reinforced surface area of interface. To achieve this goal, three concentrations of CNT are used: 5 g/m2, 7.5 g/m2, and 10 g/m2. As with all sample sets, non-reinforced samples are constructed and tested as a reference point. Mode II testing is completed since prior phases determined CNT reinforcement significantly affected Mode II fracture toughness. The results of Mode II testing are shown in Fig. 24 along with standard deviation. As shown in the figure, 7.5 g/m2 of CNT is the optimal concentration, which is consistent with the previous study [Kwon, et al., 2008]. Again, the lowest value of Gn for samples reinforced with 7.5 g/ m2 CNT is higher than the highest value of non-reinforced samples.

The higher concentration of 10 g/m2 results in slightly lower critical energy release rate than the 7.5 g/m2 concentration. On the other hand, interface toughness with the CNT concentration of 5 g/m2 is even lower than that of non-reinforced specimens. This result suggested that a lower amount of CNT at the interface does not provide proper mechanical interlocking while serving as a localized defect because of a lower bonding between CNT and polymers.

The additional purpose of testing is to determine the effect of "banding" CNT. "Banding" refers to only reinforcing a part of the interface area on the sample. All other sample sets involved using CNT to reinforce the entire secondary bond between the top and bottom plates. However, samples for the present tests are only reinforced in the area extending 6 cm from the initial crack tip. "Banding" CNT may be applicable to repair of carbon fiber composite components when only a localized area requires reinforcement. The Mode II critical energy release rate results in 19% increase due to CNT reinforcement with 7.5 g/ m2 CNT concentration. The drop from roughly 30% found in previous sample sets is due to "banding" the CNT vice reinforcing the entire secondary bond.

Enhancement of interface strength

Fig. 24. Normalized Gn values for different concentration of CNT