Characterization of Mode I and Mode II Interlaminar Fracture Toughness in CNT-Enhanced CFRP under Various Temperature and Loading Rates

This study investigates the influence of temperature and loading rate on the Mode I and Mode II interlaminar fracture behavior of carbon-nanotubes-enhanced carbon-fiber-reinforced polymer (CNT-CFRP). CNT-induced toughening of the epoxy matrix is characterized by producing CFRP with varying loading of CNT areal density. CNT-CFRP samples were subjected to varying loading rates and testing temperatures. Fracture surfaces of CNT-CFRP were analyzed using scanning electron microscopy (SEM) imaging. Mode I and Mode II interlaminar fracture toughness increased with increasing amount of CNT to an optimum value of 1 g/m2, then decreased at higher CNT amounts. Moreover, it was found that CNT-CFRP fracture toughness increased linearly with the loading rate in Mode I and Mode II. On the other hand, different responses to changing temperature were observed; Mode I fracture toughness increased when elevating the temperature, while Mode II fracture toughness increased with increasing up to room temperature and decreased at higher temperatures.


Introduction
Carbon-fiber-reinforced polymer (CFRP) composites are widely utilized as structural materials in engineering applications due to their exceptional properties, such as high specific strength, corrosion resistance, high stiffness, and fatigue resistance [1][2][3][4]. Carbon fiber reinforced in an epoxy polymer matrix is known to exhibit a high in-plane tensile strength (up to 1400 MPa); however, applications where the peel stresses exceed the matrix's tensile strength can result in delamination failure within composite layers [5][6][7]. In realworld applications, harsh environmental conditions and loading rates can also contribute to premature delamination. In order to improve interlaminar strength, it is necessary to understand the propagation of cracks and characterize its critical energy release rate under various environmental and loading conditions [8,9].
The delamination of carbon-fiber-reinforced polymer (CFRP) components can result in significant material damage and shorten their service life. Several approaches have been proposed to improve the interlaminar fracture toughness of these components, including stitching, braiding [10], matrices toughening [11,12], and z-pinning in the through-thickness direction [13]. However, these techniques may be expensive, difficult to implement, or negatively affect the performance of individual fibers. As an alternative solution, using nanofillers with superior mechanical properties to strengthen the carbon fiber polymer matrix and enhance fiber-to-matrix interaction has been proposed [14,15]. Various nanofillers, such as ceramic nanotubes, graphene nanotubes, carbon nanotubes, and natural fiber, have been used in engineering applications [16]. Several types of carbon fiber, such as quartz and glass fibers, have been utilized due to their respective advantages [17,18]. Studies have shown that composite laminates containing CNT in the delamination plane exhibit

Interlaminar Fracture Toughness in Mode I and Mode II
The fracture toughness of composite materials is quantified through the measurement of the cracking strain energy release rate, which is determined through interlaminar fracture tests [33][34][35]. Interlaminar fracture tests are named based on Mode I and Mode II loading modes. In the Mode I test, also known as the double cantabile beam test (DCB), loading is applied in a direction normal to a crack plane. In the Mode II test, also known as the end-notched flexure (ENF) test, loading is applied as a three-point bending, parallel to the crack plane and perpendicular to the crack front. Different loading modes are shown in Figure 1. the fracture surface of CNT-DFRP, revealing that the interplay between these factors has a significant and noticeable effect.

Interlaminar Fracture Toughness in Mode I and Mode II
The fracture toughness of composite materials is quantified through the measurement of the cracking strain energy release rate, which is determined through interlaminar fracture tests [33][34][35]. Interlaminar fracture tests are named based on Mode I and Mode II loading modes. In the Mode I test, also known as the double cantabile beam test (DCB), loading is applied in a direction normal to a crack plane. In the Mode II test, also known as the end-notched flexure (ENF) test, loading is applied as a three-point bending, parallel to the crack plane and perpendicular to the crack front. Different loading modes are shown in Figure 1.

Materials
Unidirectional (UD) carbon fiber-epoxy prepreg laminates, VTC401-C150-24, were supplied by SHD Composites. These UD laminates have an areal weight of 150 g/m 2 and a 38% epoxy resin content by weight. Multi-walled carbon nanotubes (MWCNTs) with an average diameter of 9.5 nm, NC7000, were used as nanofillers. Ethanol was used to make a spraying solution of 0.2 wt% CNT. A non-adhesive polytetrafluoroethylene (PTFE), with a required thickness of 13 µm according to ASTM D5528-01, was used as an insert in the mid-plane of the laminate during layup to form an initiation crack for the delamination.

Samples Preparation
First, the CFRP prepregs were cut with dimensions of 150.81 mm × 152.4 mm and 150.81 mm × 203.2 mm for Mode I and Mode II interlaminar fracture tests, respectively. Then, PTFE was inserted on the surface of the mid-layer. At the same time, CNT and Ethanol solutions were prepared at 0.2 wt%. A mechanical stirrer and ultrasonic cleaner, shown in Figure 2a, were operated for 75 min to disperse CNT effectively in ethanol. Once the solution was prepared, it was sprayed on the surface of the mid-layer using an airbrush. After drying, the other half of the lamina, without CNT or insert, was attached to CNT-sprayed and PTFE-inserted lamina ( Figure 2b). Then, multiple samples and glass molds were placed on top of each other to cure multiple samples efficiently. Once multiple glass molds were stacked, the glass mold and uncured laminates were wrapped in a breather fabric, as shown in Figure 2c. Since the wrapping procedure was completed, the glass mold with the uncured samples was placed inside the vacuum bag and sealed entirely on all sides. Slight vacuum pressure was applied to the vacuum bag to prevent air leakage ( Figure 2d). Finally, the vacuum bag was placed into the oven to cure samples for 1 h at 100 °C. After 1 h of curing, the temperature was increased to 135 °C, and samples were cured for two more hours.

Materials
Unidirectional (UD) carbon fiber-epoxy prepreg laminates, VTC401-C150-24, were supplied by SHD Composites. These UD laminates have an areal weight of 150 g/m 2 and a 38% epoxy resin content by weight. Multi-walled carbon nanotubes (MWCNTs) with an average diameter of 9.5 nm, NC7000, were used as nanofillers. Ethanol was used to make a spraying solution of 0.2 wt% CNT. A non-adhesive polytetrafluoroethylene (PTFE), with a required thickness of 13 µm according to ASTM D5528-01, was used as an insert in the mid-plane of the laminate during layup to form an initiation crack for the delamination.

Samples Preparation
First, the CFRP prepregs were cut with dimensions of 150.81 mm × 152.4 mm and 150.81 mm × 203.2 mm for Mode I and Mode II interlaminar fracture tests, respectively. Then, PTFE was inserted on the surface of the mid-layer. At the same time, CNT and Ethanol solutions were prepared at 0.2 wt%. A mechanical stirrer and ultrasonic cleaner, shown in Figure 2a, were operated for 75 min to disperse CNT effectively in ethanol. Once the solution was prepared, it was sprayed on the surface of the mid-layer using an airbrush. After drying, the other half of the lamina, without CNT or insert, was attached to CNTsprayed and PTFE-inserted lamina ( Figure 2b). Then, multiple samples and glass molds were placed on top of each other to cure multiple samples efficiently. Once multiple glass molds were stacked, the glass mold and uncured laminates were wrapped in a breather fabric, as shown in Figure 2c. Since the wrapping procedure was completed, the glass mold with the uncured samples was placed inside the vacuum bag and sealed entirely on all sides. Slight vacuum pressure was applied to the vacuum bag to prevent air leakage ( Figure 2d). Finally, the vacuum bag was placed into the oven to cure samples for 1 h at 100 • C. After 1 h of curing, the temperature was increased to 135 • C, and samples were cured for two more hours.  Afterward, the samples were cooled to room temperature and cut with a water jet according to the ASTM standards. Mode I and Mode II interlaminar fracture tests sample dimensions are shown in Figure 3.

Experimental Procedure
An MTS Universal test machine with 500 N and 10 kN load cells was used to perform DCB and Mode II interlaminar fracture tests. Crack propagations were recorded using a LaVision digital image correlation device, and temperatures were controlled using Ther-moStream machine (Thermal Temptronic ATS-545-M-6), which delivers dry air.
Mode I and Mode II interlaminar fracture tests were carried out using double cantilever beam (DCB) and end notch flexure (ENF) samples, shown in Figure 3, prepared according to ASTM D5528 and ASTM D7905/D7905M, respectively. All tests were repeated five times to reduce variation. DCB and ENF tests were conducted in 3 groups, as shown in Table 1. The first group of tests was performed to determine the optimum CNT quantity in CFRP. The DCB and ENF tests were conducted using samples with different CNT densities at constant room temperature and loading rate of 1 mm/s. Once optimum CNT density was determined as 1 g/m 2 , the second group of tests was performed to evaluate the effect of loading rates on the interlaminar fracture toughness. The DCB and ENF tests were conducted using CNT-CFTP with a CNT density of 1 g/m 2 at a constant room temperature and four different loading rates: 0.001 mm/s, 0.01 mm/s, 0.1 mm/s, and 1 mm/s. The final group of tests was performed to examine the effect of temperature on interlaminar fracture toughness. The DCB and ENF tests were conducted on CNT-CFTP specimens with a CNT density of 1 g/m 2 at a constant loading rate of 1 mm/s and four different temperatures: −60 °C, −20 °C, 20 °C, and 60 °C. Afterward, the samples were cooled to room temperature and cut with a water jet according to the ASTM standards. Mode I and Mode II interlaminar fracture tests sample dimensions are shown in Figure 3. Afterward, the samples were cooled to room temperature and cut with a water jet according to the ASTM standards. Mode I and Mode II interlaminar fracture tests sample dimensions are shown in Figure 3.

Experimental Procedure
An MTS Universal test machine with 500 N and 10 kN load cells was used to perform DCB and Mode II interlaminar fracture tests. Crack propagations were recorded using a LaVision digital image correlation device, and temperatures were controlled using Ther-moStream machine (Thermal Temptronic ATS-545-M-6), which delivers dry air.
Mode I and Mode II interlaminar fracture tests were carried out using double cantilever beam (DCB) and end notch flexure (ENF) samples, shown in Figure 3, prepared according to ASTM D5528 and ASTM D7905/D7905M, respectively. All tests were repeated five times to reduce variation. DCB and ENF tests were conducted in 3 groups, as shown in Table 1. The first group of tests was performed to determine the optimum CNT quantity in CFRP. The DCB and ENF tests were conducted using samples with different CNT densities at constant room temperature and loading rate of 1 mm/s. Once optimum CNT density was determined as 1 g/m 2 , the second group of tests was performed to evaluate the effect of loading rates on the interlaminar fracture toughness. The DCB and ENF tests were conducted using CNT-CFTP with a CNT density of 1 g/m 2 at a constant room temperature and four different loading rates: 0.001 mm/s, 0.01 mm/s, 0.1 mm/s, and 1 mm/s. The final group of tests was performed to examine the effect of temperature on interlaminar fracture toughness. The DCB and ENF tests were conducted on CNT-CFTP specimens with a CNT density of 1 g/m 2 at a constant loading rate of 1 mm/s and four different temperatures: −60 °C, −20 °C, 20 °C, and 60 °C.

Experimental Procedure
An MTS Universal test machine with 500 N and 10 kN load cells was used to perform DCB and Mode II interlaminar fracture tests. Crack propagations were recorded using a LaVision digital image correlation device, and temperatures were controlled using ThermoStream machine (Thermal Temptronic ATS-545-M-6), which delivers dry air.
Mode I and Mode II interlaminar fracture tests were carried out using double cantilever beam (DCB) and end notch flexure (ENF) samples, shown in Figure 3, prepared according to ASTM D5528 and ASTM D7905/D7905M, respectively. All tests were repeated five times to reduce variation. DCB and ENF tests were conducted in 3 groups, as shown in Table 1. The first group of tests was performed to determine the optimum CNT quantity in CFRP. The DCB and ENF tests were conducted using samples with different CNT densities at constant room temperature and loading rate of 1 mm/s. Once optimum CNT density was determined as 1 g/m 2 , the second group of tests was performed to evaluate the effect of loading rates on the interlaminar fracture toughness. The DCB and ENF tests were conducted using CNT-CFTP with a CNT density of 1 g/m 2 at a constant room temperature and four different loading rates: 0.001 mm/s, 0.01 mm/s, 0.1 mm/s, and 1 mm/s. The final group of tests was performed to examine the effect of temperature on interlaminar fracture toughness. The DCB and ENF tests were conducted on CNT-CFTP specimens with a CNT density of 1 g/m 2 at a constant loading rate of 1 mm/s and four different temperatures: −60 • C, −20 • C, 20 • C, and 60 • C. DCB samples with 20 layers were prepared and cut to a specific dimension of 21.68 mm × 132.40 mm × 3.9 mm. Piano hinges were attached to each side of the sample to set the initial crack length to 50 mm from the PTFE film. Before conducting the DCB test, the samples were cured for 24 h. In order to allow visual crack detection during the test, white paint was sprayed on the side of the sample, and a ruler paper was attached to the painted surface. An environmental chamber and ThermoStream machine (Thermal Temptronic ATS-545-M-6) were used for the temperature tests. The piano hinges were loaded into MTS grips, and DCB tests were conducted in tension according to ASTM D5528-13. The DCB test setup is shown in Figure 4.  DCB samples with 20 layers were prepared and cut to a specific dimension of 21.68 mm × 132.40 mm × 3.9 mm. Piano hinges were attached to each side of the sample to set the initial crack length to 50 mm from the PTFE film. Before conducting the DCB test, the samples were cured for 24 h. In order to allow visual crack detection during the test, white paint was sprayed on the side of the sample, and a ruler paper was attached to the painted surface. An environmental chamber and ThermoStream machine (Thermal Temptronic ATS-545-M-6) were used for the temperature tests. The piano hinges were loaded into MTS grips, and DCB tests were conducted in tension according to ASTM D5528-13. The DCB test setup is shown in Figure 4. The compliance calibration method was used for data reduction in order to calculate Mode I interlaminar fracture toughness. At least nine measurements were taken during the test at different crack length values, starting from 60 mm and increasing by 5 mm. The DCB interlaminar fracture toughness is formulated as follows: , where is the crack's length, is the sample's width, is the displacement, is the load, and is the slope of a plot of compliance ( i/Pi) versus crack length ( i) on a log-log scale.

ENF Experimental Tests
ENF samples, which contained 24 layers, were cut into 21.68 mm × 170 mm × 4.1 mm dimensions. During sample preparation, the initial crack length was adjusted to 50 mm for ENF samples. Loading was applied as three-point bending following the ASTM D7905/D7905M standard. The samples were placed on roller supports, ensuring that the distance between the initial crack tip and the left roller was 30 mm. Experimental setup for the ENF test is shown in Figure 5. The compliance calibration method was used for data reduction in order to calculate Mode I interlaminar fracture toughness. At least nine measurements were taken during the test at different crack length values, starting from 60 mm and increasing by 5 mm. The DCB interlaminar fracture toughness is formulated as follows: where a is the crack's length, b is the sample's width, δ is the displacement, P is the load, and n is the slope of a plot of compliance (δ i /P i ) versus crack length (a i ) on a log-log scale.

ENF Experimental Tests
ENF samples, which contained 24 layers, were cut into 21.68 mm × 170 mm × 4.1 mm dimensions. During sample preparation, the initial crack length was adjusted to 50 mm for ENF samples. Loading was applied as three-point bending following the ASTM D7905/D7905M standard. The samples were placed on roller supports, ensuring that the distance between the initial crack tip and the left roller was 30 mm. Experimental setup for the ENF test is shown in Figure 5. Since the compliance calibration method is the only acceptable method for data reduction for the ENF test, it was used as the data reduction method. The ENF sample exhibited unstable crack growth over most of its length; therefore, Mode II interlaminar fracture toughness was obtained as a single value from maximum load points. The strain energy release rate in Mode II interlaminar fracture toughness is formulated as follows: where is the crack's length, is the sample's width, is the displacement, is the load, and is the half span length.

Fracture Examination
A Thermofisher Quanta 3D scanning electron microscope was used to examine the CDM samples' fractured surfaces. The DCB samples were split open after the completion of the tests and imaged 25 mm away from the initial crack tip; thus, we maintained consistency when comparing the samples. The DCB sample's fractured surface was measured at three different magnifications (100×, 1000×, and 10,000×) to observe micro-and nanodamages in the DCB samples. The SEM location for examining DCB samples is shown in Figure 6.

Results
Experimental compliance calibrations were performed on DCB and ENF samples at all group tests. Crack propagation in DCB tests-unlike ENF tests-was observed at a stable and almost constant speed during the experiment. The typical fracture mechanism and load-displacement curve of DCB and ENF samples that contain 1 g/m 2 CNT at 20 °C with a 1 mm/s loading rate is given in Figure 7.
Since the compliance calibration method is the only acceptable method for data reduction for the ENF test, it was used as the data reduction method. The ENF sample exhibited unstable crack growth over most of its length; therefore, Mode II interlaminar fracture toughness was obtained as a single value from maximum load points. The strain energy release rate in Mode II interlaminar fracture toughness is formulated as follows: where a is the crack's length, b is the sample's width, δ is the displacement, P is the load, and L is the half span length.

Fracture Examination
A Thermofisher Quanta 3D scanning electron microscope was used to examine the CDM samples' fractured surfaces. The DCB samples were split open after the completion of the tests and imaged 25 mm away from the initial crack tip; thus, we maintained consistency when comparing the samples. The DCB sample's fractured surface was measured at three different magnifications (100×, 1000×, and 10,000×) to observe micro-and nano-damages in the DCB samples. The SEM location for examining DCB samples is shown in Figure 6. Since the compliance calibration method is the only acceptable method for data reduction for the ENF test, it was used as the data reduction method. The ENF sample exhibited unstable crack growth over most of its length; therefore, Mode II interlaminar fracture toughness was obtained as a single value from maximum load points. The strain energy release rate in Mode II interlaminar fracture toughness is formulated as follows: where is the crack's length, is the sample's width, is the displacement, is the load, and is the half span length.

Fracture Examination
A Thermofisher Quanta 3D scanning electron microscope was used to examine the CDM samples' fractured surfaces. The DCB samples were split open after the completion of the tests and imaged 25 mm away from the initial crack tip; thus, we maintained consistency when comparing the samples. The DCB sample's fractured surface was measured at three different magnifications (100×, 1000×, and 10,000×) to observe micro-and nanodamages in the DCB samples. The SEM location for examining DCB samples is shown in Figure 6.

Results
Experimental compliance calibrations were performed on DCB and ENF samples at all group tests. Crack propagation in DCB tests-unlike ENF tests-was observed at a stable and almost constant speed during the experiment. The typical fracture mechanism and load-displacement curve of DCB and ENF samples that contain 1 g/m 2 CNT at 20 °C with a 1 mm/s loading rate is given in Figure 7.

Results
Experimental compliance calibrations were performed on DCB and ENF samples at all group tests. Crack propagation in DCB tests-unlike ENF tests-was observed at a stable and almost constant speed during the experiment. The typical fracture mechanism and load-displacement curve of DCB and ENF samples that contain 1 g/m 2 CNT at 20 • C with a 1 mm/s loading rate is given in Figure 7.

Results
Experimental compliance calibrations were performed on DCB and ENF samples at all group tests. Crack propagation in DCB tests-unlike ENF tests-was observed at a stable and almost constant speed during the experiment. The typical fracture mechanism and load-displacement curve of DCB and ENF samples that contain 1 g/m 2 CNT at 20 °C with a 1 mm/s loading rate is given in Figure 7.  Figure 8 shows the Mode I interlaminar fracture toughness of CNT-CFRP with different CNT areal densities. A non-linear relationship was observed between Mode I interlaminar fracture toughness and CNT areal density. Mode I interlaminar fracture toughness increased by increasing the CNT amount to 1 g/m 2 . Beyond this point, a decrease in Mode I interlaminar fracture toughness was observed. The optimum CNT areal density was determined to be 1 g/m 2 for the DCB test. This decrease in the fracture toughness is due to the clustering of CNTs in the fiber interface rather than the penetration of individual CNTs into the fibers, leading to a decreased resistance to crack growth [23]. This effect was further confirmed through fracture analysis presented in subsequent sections.

Loading Rate Effect
The experimental results of the DCB samples with 1 g/m 2 CNT subjected to varying loading rates at room temperature are shown in Figure 9. Mode I interlaminar fracture toughness increased with increasing loading rate. This result is due to the higher loading rates, where CFRP experiences a more rapid crack growth, leading to the formation of more damage-tolerant fracture mechanisms, such as bridging, delamination cracking, or fiber fracture. These mechanisms can increase the energy required to propagate the crack,  Figure 8 shows the Mode I interlaminar fracture toughness of CNT-CFRP with different CNT areal densities. A non-linear relationship was observed between Mode I interlaminar fracture toughness and CNT areal density. Mode I interlaminar fracture toughness increased by increasing the CNT amount to 1 g/m 2 . Beyond this point, a decrease in Mode I interlaminar fracture toughness was observed. The optimum CNT areal density was determined to be 1 g/m 2 for the DCB test. This decrease in the fracture toughness is due to the clustering of CNTs in the fiber interface rather than the penetration of individual CNTs into the fibers, leading to a decreased resistance to crack growth [23]. This effect was further confirmed through fracture analysis presented in subsequent sections.  Figure 8 shows the Mode I interlaminar fracture toughness of CNT-CFRP with different CNT areal densities. A non-linear relationship was observed between Mode I interlaminar fracture toughness and CNT areal density. Mode I interlaminar fracture toughness increased by increasing the CNT amount to 1 g/m 2 . Beyond this point, a decrease in Mode I interlaminar fracture toughness was observed. The optimum CNT areal density was determined to be 1 g/m 2 for the DCB test. This decrease in the fracture toughness is due to the clustering of CNTs in the fiber interface rather than the penetration of individual CNTs into the fibers, leading to a decreased resistance to crack growth [23]. This effect was further confirmed through fracture analysis presented in subsequent sections.

Loading Rate Effect
The experimental results of the DCB samples with 1 g/m 2 CNT subjected to varying loading rates at room temperature are shown in Figure 9. Mode I interlaminar fracture toughness increased with increasing loading rate. This result is due to the higher loading rates, where CFRP experiences a more rapid crack growth, leading to the formation of more damage-tolerant fracture mechanisms, such as bridging, delamination cracking, or fiber fracture. These mechanisms can increase the energy required to propagate the crack,

Loading Rate Effect
The experimental results of the DCB samples with 1 g/m 2 CNT subjected to varying loading rates at room temperature are shown in Figure 9. Mode I interlaminar fracture toughness increased with increasing loading rate. This result is due to the higher loading rates, where CFRP experiences a more rapid crack growth, leading to the formation of more damage-tolerant fracture mechanisms, such as bridging, delamination cracking, or fiber fracture. These mechanisms can increase the energy required to propagate the crack, resulting in increasing interlaminar fracture toughness.  Figure 10 shows the interlaminar fracture toughness of DCB samples with 1 g/m 2 CNT at different temperatures for 1 mm/s loading rate. It is observed that Mode I interlaminar fracture toughness increased with increasing for all temperatures. With the effect of temperature, Mode I interlaminar fracture toughness significantly decreased up to −60 °C. Moreover, with increasing temperature, softness was observed in the epoxy, which might have caused the fibers to be pulled out from the epoxy matrix. This phenomenon is called fiber bridging and increases fracture toughness [36,37].  Figure 11 shows CNT-CFRP's Mode II interlaminar fracture toughness with different CNT areal densities. Enhancement of Mode II fracture toughness in CFRP due to the areal loading of CNTs exhibited a trend similar to that observed in the Mode I experiments. Mode II interlaminar fracture toughness increased with increasing amount of CNTs up to 1 g/m 2 ; further increase in amount of CNTs in CFRP resulted in decreasing Mode II frac-2 Figure 9. Mode I fracture toughness at varying loading rates for DCB test. Figure 10 shows the interlaminar fracture toughness of DCB samples with 1 g/m 2 CNT at different temperatures for 1 mm/s loading rate. It is observed that Mode I interlaminar fracture toughness increased with increasing for all temperatures. With the effect of temperature, Mode I interlaminar fracture toughness significantly decreased up to −60 • C. Moreover, with increasing temperature, softness was observed in the epoxy, which might have caused the fibers to be pulled out from the epoxy matrix. This phenomenon is called fiber bridging and increases fracture toughness [36,37].  Figure 10 shows the interlaminar fracture toughness of DCB samples with 1 g/m 2 CNT at different temperatures for 1 mm/s loading rate. It is observed that Mode I interlaminar fracture toughness increased with increasing for all temperatures. With the effect of temperature, Mode I interlaminar fracture toughness significantly decreased up to −60 °C. Moreover, with increasing temperature, softness was observed in the epoxy, which might have caused the fibers to be pulled out from the epoxy matrix. This phenomenon is called fiber bridging and increases fracture toughness [36,37].  Figure 11 shows CNT-CFRP's Mode II interlaminar fracture toughness with different CNT areal densities. Enhancement of Mode II fracture toughness in CFRP due to the areal loading of CNTs exhibited a trend similar to that observed in the Mode I experiments. Mode II interlaminar fracture toughness increased with increasing amount of CNTs up to 1 g/m 2 ; further increase in amount of CNTs in CFRP resulted in decreasing Mode II frac-2 Figure 10. Mode I fracture toughness at varying temperatures for DCB test.  Figure 11 shows CNT-CFRP's Mode II interlaminar fracture toughness with different CNT areal densities. Enhancement of Mode II fracture toughness in CFRP due to the areal loading of CNTs exhibited a trend similar to that observed in the Mode I experiments. Mode II interlaminar fracture toughness increased with increasing amount of CNTs up to 1 g/m 2 ; further increase in amount of CNTs in CFRP resulted in decreasing Mode II fracture toughness. The optimal areal density of CNTs was observed to be 1 g/m 2 for the ENF test.

CNT Areal Density Effect
Nanomaterials 2023, 13, x FOR PEER REVIEW 9 of 15 Figure 11. Mode II fracture toughness at varying CNT areal density for the ENF test.

Loading Rate Effect
The experimental results of applying varying loading rates on the ENF samples with 1 g/m 2 CNT at room temperature are shown in Figure 12. It is obvious that Mode II interlaminar fracture toughness increased with increasing loading rate. However, this increase was slightly less in elevated loading rates where the ENF tests were performed.

Temperature Effect
Interlaminar fracture toughness at varying temperatures for the ENF test is shown in Figure 13. The Mode II interlaminar fracture toughness increased with increasing temperature up to 20 °C, then decreased at higher temperatures. Since delamination is smaller in the ENF test, less fiber bridging occurs compared to the corresponding DCB tests. However, the fact that fibers and the matrix have different thermal expansion coefficients introduced microcracks at the matrix-fiber interface, thus resulting in decreased Mode II interlaminar fracture toughness. Figure 11. Mode II fracture toughness at varying CNT areal density for the ENF test.

Loading Rate Effect
The experimental results of applying varying loading rates on the ENF samples with 1 g/m 2 CNT at room temperature are shown in Figure 12. It is obvious that Mode II interlaminar fracture toughness increased with increasing loading rate. However, this increase was slightly less in elevated loading rates where the ENF tests were performed. Nanomaterials 2023, 13, x FOR PEER REVIEW 9 of 15 Figure 11. Mode II fracture toughness at varying CNT areal density for the ENF test.

Loading Rate Effect
The experimental results of applying varying loading rates on the ENF samples with 1 g/m 2 CNT at room temperature are shown in Figure 12. It is obvious that Mode II interlaminar fracture toughness increased with increasing loading rate. However, this increase was slightly less in elevated loading rates where the ENF tests were performed.

Temperature Effect
Interlaminar fracture toughness at varying temperatures for the ENF test is shown in Figure 13. The Mode II interlaminar fracture toughness increased with increasing temperature up to 20 °C, then decreased at higher temperatures. Since delamination is smaller in the ENF test, less fiber bridging occurs compared to the corresponding DCB tests. However, the fact that fibers and the matrix have different thermal expansion coefficients introduced microcracks at the matrix-fiber interface, thus resulting in decreased Mode II interlaminar fracture toughness.

Temperature Effect
Interlaminar fracture toughness at varying temperatures for the ENF test is shown in Figure 13. The Mode II interlaminar fracture toughness increased with increasing temperature up to 20 • C, then decreased at higher temperatures. Since delamination is smaller in the ENF test, less fiber bridging occurs compared to the corresponding DCB tests. However, the fact that fibers and the matrix have different thermal expansion coefficients introduced microcracks at the matrix-fiber interface, thus resulting in decreased Mode II interlaminar fracture toughness. To compare the results of all the parameters' effect on the CNT samples, we present Table 2, showing the change (in percentage) of the resulting values with respect to the reference value for each parameter (temperature, speed, and CNT areal density). It must be noted that, with the minus sign, the value decreased from the reference conditions (20 °C, 1 mm/s, 1 g/m 2 ), which is denoted as "REF". As observed in Table 2, the values of each parameter decreased with decreasing temperature. Those results also proved that 1 g/m 2 is the ideal quantity of CNTs, since either increasing or decreasing this value corresponded to lower fracture toughness. This follows other studies' outcomes found in the bibliography [23].

Microstructural Analysis
The fracture surface of DCB test samples was investigated using SEM. The DCB sample's fractured surface was measured at 100×, 1000×, and 10,000× magnifications to observe micro-and nano-damages in the DCB samples. The fracture surface of the DCB test samples with 1 g/m 2 CNT-CFRP is shown in Figure 14. The loading rate was 1 mm/s performed at room temperature. To compare the results of all the parameters' effect on the CNT samples, we present Table 2, showing the change (in percentage) of the resulting values with respect to the reference value for each parameter (temperature, speed, and CNT areal density). It must be noted that, with the minus sign, the value decreased from the reference conditions (20 • C, 1 mm/s, 1 g/m 2 ), which is denoted as "REF". As observed in Table 2, the values of each parameter decreased with decreasing temperature. Those results also proved that 1 g/m 2 is the ideal quantity of CNTs, since either increasing or decreasing this value corresponded to lower fracture toughness. This follows other studies' outcomes found in the bibliography [23].

Microstructural Analysis
The fracture surface of DCB test samples was investigated using SEM. The DCB sample's fractured surface was measured at 100×, 1000×, and 10,000× magnifications to observe micro-and nano-damages in the DCB samples. The fracture surface of the DCB test samples with 1 g/m 2 CNT-CFRP is shown in Figure 14. The loading rate was 1 mm/s performed at room temperature.
3.3.1. Fracture Analysis of DCB Samples with Varying CNT Areal Density Figure 15 shows an SEM (100×) of DCB test samples with varying CNT areal density. It can be seen that fiber breakage and fiber imprints decreased with increasing amount of CNT up to 1 g/m 2 ; after this limit, a decrease was observed. In addition, fiber breakages were observed where DCB contained no CNT. Figure 16 shows the fracture surface of the DCB sample with varying CNT areal density at 1000× magnification. Fiber and matrix damage is clearly observed in detail. Specifically, damage in the DCB sample without CNT occurs at the fiber-matrix interface, while the damage in DCB samples to CNT occurs on the matrix surface. The high amounts of CNT in the fiber layers resulted in the fibers banding together at the fiber interface instead of penetrating deep into the fibers. Hence, they demonstrated no resistance to crack growth, resulting in less fracture toughness at a higher than optimum level of CNT areal density.
The fracture surface of the DCB sample with varying CNT areal density at 10,000× magnification is shown in Figure 17.  Figure 15 shows an SEM (100×) of DCB test samples with varying CNT areal density. It can be seen that fiber breakage and fiber imprints decreased with increasing amount of CNT up to 1 g/m 2 ; after this limit, a decrease was observed. In addition, fiber breakages were observed where DCB contained no CNT.   Figure 15 shows an SEM (100×) of DCB test samples with varying CNT areal density. It can be seen that fiber breakage and fiber imprints decreased with increasing amount of CNT up to 1 g/m 2 ; after this limit, a decrease was observed. In addition, fiber breakages were observed where DCB contained no CNT.  Figure 16 shows the fracture surface of the DCB sample with varying CNT areal density at 1000× magnification. Fiber and matrix damage is clearly observed in detail. Specifically, damage in the DCB sample without CNT occurs at the fiber-matrix interface, while the damage in DCB samples to CNT occurs on the matrix surface. The high amounts of CNT in the fiber layers resulted in the fibers banding together at the fiber interface instead of penetrating deep into the fibers. Hence, they demonstrated no resistance to crack growth, resulting in less fracture toughness at a higher than optimum level of CNT areal density. The fracture surface of the DCB sample with varying CNT areal density at 10,000× magnification is shown in Figure 17. while the damage in DCB samples to CNT occurs on the matrix surface. The high amounts of CNT in the fiber layers resulted in the fibers banding together at the fiber interface instead of penetrating deep into the fibers. Hence, they demonstrated no resistance to crack growth, resulting in less fracture toughness at a higher than optimum level of CNT areal density. The fracture surface of the DCB sample with varying CNT areal density at 10,000× magnification is shown in Figure 17. SEM of DCB test samples that contained 1 g/m 2 with varying temperatures i in Figure 18. It can be seen that the DCB sample exhibited a brittle behavior. With ing temperature, softness can be seen in the epoxy, which might have caused fibe

Fracture Analyses of DCB Samples with Varying Temperature
SEM of DCB test samples that contained 1 g/m 2 with varying temperatures is shown in Figure 18. It can be seen that the DCB sample exhibited a brittle behavior. With increasing temperature, softness can be seen in the epoxy, which might have caused fiber bridging and fracture toughness to increase.

Fracture Analyses of DCB Samples with Varying Temperature
SEM of DCB test samples that contained 1 g/m 2 with varying temperatures is shown in Figure 18. It can be seen that the DCB sample exhibited a brittle behavior. With increasing temperature, softness can be seen in the epoxy, which might have caused fiber bridging and fracture toughness to increase.

Conclusions
The effect of CNT areal density, temperature, and loading rate on Mode I and Mode II interlaminar fracture toughness of carbon nanotubes enhanced with carbon-fiber-reinforced polymers was investigated. The relationship between CNT areal density and Mode I and Mode II fracture toughness was found to be non-linear. Moreover, the optimum CNT amount was 1 g/m 2 , for which the interlaminar strengthening effect is maximum. It

Conclusions
The effect of CNT areal density, temperature, and loading rate on Mode I and Mode II interlaminar fracture toughness of carbon nanotubes enhanced with carbon-fiber-reinforced polymers was investigated. The relationship between CNT areal density and Mode I and Mode II fracture toughness was found to be non-linear. Moreover, the optimum CNT amount was 1 g/m 2 , for which the interlaminar strengthening effect is maximum. It has also been observed that toughness in both Modes I and II improved with increasing loading rate. In Mode II deformation, this improvement is slightly less pronounced. Furthermore, Mode I and Mode II interlaminar fracture toughness was observed to be affected differently by varying temperatures. Mode I fracture toughness increased with increasing temperature. Mode II fracture toughness increased by increasing temperature up to room temperature, while at elevated temperatures it decreased. Since CNTs, resins, and fibers shrink at different rates, stress will be generated within the fiber, matrix, and fiber-matrix interface. This stress has the potential to inflict damage upon the composite laminates. Additionally, prolonged exposure to low temperatures can cause epoxy resins to become brittle, deteriorating interlaminar fracture toughness.