Comparative Study on the Impact Wedge-Peel Performance of Epoxy-Based Structural Adhesives Modified with Different Toughening Agents

Epoxy adhesives are widely used in various industries because of their high heat and chemical resistance, high cohesion, and minimal shrinkage. Recently, epoxy adhesives have been applied in the automotive industry as structural adhesives for lightweight vehicles. However, the brittleness of the epoxy is an obstacle for this application, since the automotive industry requires epoxy-based structural adhesives to have a high level of high-speed impact resistance. Hence, we used phenol-terminated polyurethane (PTPU) as a toughening agent for epoxy adhesives and compared the results with those that were obtained with carboxyl-terminated butadiene acrylonitrile copolymer (CTBN). The high-energy impact resistance of the epoxy adhesives was measured by the impact wedge-peel (IWP) test, and the shear strength was measured by the single lap joint test. As a result, the 20 wt % PTPU-modified epoxy adhesive showed remarkably higher total absorbed energy (25.8 J) during the IWP test and shear strength (32.3 MPa) as compared with the control epoxy adhesive (4.1 J and 20.6 MPa, respectively). In particular, the total absorbed energy of the PTPU-modified epoxy adhesive was much larger than that of the CTBN-modified epoxy adhesive (5.8 J). When more than 10 wt % PTPU was added, the modified epoxy adhesives showed stable crack growth and effectively transferred external stress to the substrate. These results were explained by changes in the glass transition temperature, crosslinking density, and morphology due to the toughening agents.


Introduction
Structural adhesives are used for structural bonding that requires high adhesion strength, elasticity, and environmental resistance. Particularly, structural adhesives are emerging as a fastening method for replacing bolts and nuts in the field of transportation vehicles in order to reduce carbon emissions and energy consumption. To be used as a structural adhesive for automobiles, high shear strength as well as high peel, bending, impact, and fatigue strengths are required [1]. Therefore, structural adhesives for automobiles are commonly prepared with epoxy and polyurethane [2,3]. Epoxy-based structural adhesives are the most suitable structural adhesives for fastening parts that require high-speed impact resistance, because they have high cohesive strength, low shrinkage, and high heat resistance and chemical resistance. In addition, epoxy-based structural adhesives can provide various formulations that are suitable for bonding processes and required properties by adjusting the curing agent and Polymers 2020, 12, 1549 3 of 15 the fracture behaviour of adhesive joints in response to high-speed impacts and was adopted as an international standard (ISO 11343) test method by the automotive industry in 1993 [38]. There have been several studies on epoxy-based structural adhesives while using the IWP test method [39][40][41]. Blackman et al. were the first to report the measurement of the resistance to cleavage fracture of a structural adhesive by the IWP test; these authors measured the fracture energy through a T-peel test and comparing it with the IWP test result [39]. Later, Taylor et al. evaluated the performance of commercially available epoxy-based structural adhesives using the IWP test method [40]. Although Back et al. recently used the IWP test to measure the impact strength of form-type epoxy adhesives containing CSR and analysed the effect of the degree of foaming, there are few studies on the effect of thermoplastic polyurethanes and reactive liquid rubbers on the impact resistance of epoxy-based structural adhesives using the IWP test [41].
Therefore, in this study, phenol terminated polyurethane (PTPU), a thermoplastic polyurethane, and carboxyl-terminated butadiene acrylonitrile copolymer (CTBN), a reactive liquid rubber, were used as toughening agents for structural adhesives prepared with diglycidyl ether of bisphenol-A (DGEBA) epoxy as a main component, and their effects on the physical properties of the structural adhesives were examined according to their content. The resistance to cleavage fracture, an index of impact resistance, was determined by the IWP test using CR340, which is used for automobile pillars [42], and the adhesive strength was measured by the single lap shear test. To analyse the results of the IWP and lap shear test, changes in the T g , crosslinking density and viscosity of the epoxy adhesives according to the addition of PTPU and CTBN were measured, and fracture surface and surface morphology analyses were performed in this study.

Materials
The bisphenol-A type of epoxy resin (DGEBA, diglycidyl epoxy of bisphenol A, YD-128), the carboxyl-terminated butadiene copolymer with 26% acrylonitrile (Hypro 1300X13 CTBN), and the dimer acid-modified epoxy resin used for diluent (YD-171) were purchased from Kukdo Chemical Company (Seoul, South Korea). Phenol-terminated polyurethane (DY-965) was purchased from Huntsman Corporation (USA). A 40 wt % core-shell rubber (CSR) dispersed in DGEBA epoxy resin (MX-154) was purchased from Kaneka Corporation (Tokyo, Japan). Dicyandiamide hardener (Dyhard 100) and N,N-dimethyl-N-phenylurea accelerator (Omicure U-405) were purchased from AlzChem Group (Trostberg, Germany) and DyStar Group (Singapore, Singapore), respectively. For the fillers, calcium carbonate (Omyacarb 10) was used and supplied by OMYA Company (Oftringen, Switzerland). Additionally, glass beads with a diameter of 0.2 mm to control the thickness of the adhesive were supplied by UNITECH (Ansan, South Korea). Figure 1 shows the chemical structures of PTPU and CTBN, and Table 1 shows the details of the materials used for the epoxy adhesive.
Polymers 2020, 12, x FOR PEER REVIEW 3 of 16 the fracture behaviour of adhesive joints in response to high-speed impacts and was adopted as an international standard (ISO 11343) test method by the automotive industry in 1993 [38]. There have been several studies on epoxy-based structural adhesives while using the IWP test method [39][40][41]. Blackman et al. were the first to report the measurement of the resistance to cleavage fracture of a structural adhesive by the IWP test; these authors measured the fracture energy through a T-peel test and comparing it with the IWP test result [39]. Later, Taylor et al. evaluated the performance of commercially available epoxy-based structural adhesives using the IWP test method [40]. Although Back et al. recently used the IWP test to measure the impact strength of form-type epoxy adhesives containing CSR and analysed the effect of the degree of foaming, there are few studies on the effect of thermoplastic polyurethanes and reactive liquid rubbers on the impact resistance of epoxy-based structural adhesives using the IWP test [41]. Therefore, in this study, phenol terminated polyurethane (PTPU), a thermoplastic polyurethane, and carboxyl-terminated butadiene acrylonitrile copolymer (CTBN), a reactive liquid rubber, were used as toughening agents for structural adhesives prepared with diglycidyl ether of bisphenol-A (DGEBA) epoxy as a main component, and their effects on the physical properties of the structural adhesives were examined according to their content. The resistance to cleavage fracture, an index of impact resistance, was determined by the IWP test using CR340, which is used for automobile pillars [42], and the adhesive strength was measured by the single lap shear test. To analyse the results of the IWP and lap shear test, changes in the Tg, crosslinking density and viscosity of the epoxy adhesives according to the addition of PTPU and CTBN were measured, and fracture surface and surface morphology analyses were performed in this study.

Materials
The bisphenol-A type of epoxy resin (DGEBA, diglycidyl epoxy of bisphenol A, YD-128), the carboxyl-terminated butadiene copolymer with 26% acrylonitrile (Hypro 1300X13 CTBN), and the dimer acid-modified epoxy resin used for diluent (YD-171) were purchased from Kukdo Chemical Company (Seoul, South Korea). Phenol-terminated polyurethane (DY-965) was purchased from Huntsman Corporation (USA). A 40 wt % core-shell rubber (CSR) dispersed in DGEBA epoxy resin (MX-154) was purchased from Kaneka Corporation (Tokyo, Japan). Dicyandiamide hardener (Dyhard 100) and N,N-dimethyl-N-phenylurea accelerator (Omicure U-405) were purchased from AlzChem Group (Trostberg, Germany) and DyStar Group (Singapore, Singapore), respectively. For the fillers, calcium carbonate (Omyacarb 10) was used and supplied by OMYA Company (Oftringen, Switzerland). Additionally, glass beads with a diameter of 0.2 mm to control the thickness of the adhesive were supplied by UNITECH (Ansan, South Korea). Figure 1 shows the chemical structures of PTPU and CTBN, and Table 1 shows the details of the materials used for the epoxy adhesive.    Table 2 shows the epoxy adhesive composition according to the PTPU or CTBN content. First, PTPU or CTBN was poured in the prepared container at 0, 5, 10, 15, 20, and 30 wt % according to the composition and, then, the CSR mixture was added. Additionally, DGEBA was added to adjust the epoxy ratio to the epoxy resin contained in the CSR mixture. Before adding the curing agent, the materials were blended and defoamed in a paste mixer (ARV-310, Thinky, Tokyo, Japan) at 2000 rpm under 1.0 kPa vacuum condition for three minutes. When the mixing was complete, DICY, accelerator, and CaCO 3 were added and mixed in the paste mixer at 2000 rpm for three minutes. Finally, 1 wt % DAME by weight was added, mixed in the paste mixer, and defoamed at 2000 rpm under 1.0 kPa vacuum condition for three minutes.

Preparation of the Impact Wedge-Peel (IWP) Test and Single Lap Joint (Lap Shear) Test Specimens
The ISO 11343 Standard does not specify the dimensions to be used for the IWP specimens. Hence, we adopted the widely used Ford specimen design (length: 90 mm, width: 20 mm, thickness: 1.6 mm, material: CR340). Two specimens were prepared and the surface of each specimen was wiped with acetone to remove dust. The mixture was applied to an area of 30 mm × 20 mm on the specimen surface, and a small amount of glass beads was spread onto the well-coated mixture. Each specimen was nested, as shown in Figure 2a, and cured for 28 min. at 180 • C in a conventional oven.

Preparation of Dynamic Mechanical Analyser (DMA) Specimens
DMA specimens were prepared by casting the epoxy adhesives in silicone mould. The rectangular specimens (length: 32 mm, width: 2.0 mm, thickness: 1.0 mm) were employed for DMA measurement. The epoxy adhesive mixtures were poured into the silicone mould and degassed for an hour in an 80 °C vacuum oven and cured 28 min. in a 180 °C conventional oven. After curing, the  We used a specific lap shear specimen to prepare for the single lap joint (lap shear) test, as described in the ISO 4587 Standard (length: 100 mm, width: 25 mm, thickness: 1.6 mm, material: CR340) [43]. In the same manner as the IWP test, two specimens were prepared, and the surface of each specimen was wiped with acetone to remove dust. The epoxy mixture was applied to an area of 25 mm × 12.5 mm on the specimen surface, and a small amount of glass beads, which controlled the thickness of applied adhesive, was spread onto the well-coated mixture. Each specimen was overlaid, as shown in Figure 2b, and then cured for 28 min. at 180 • C in a conventional oven.

Preparation of Dynamic Mechanical Analyser (DMA) Specimens
DMA specimens were prepared by casting the epoxy adhesives in silicone mould. The rectangular specimens (length: 32 mm, width: 2.0 mm, thickness: 1.0 mm) were employed for DMA measurement. The epoxy adhesive mixtures were poured into the silicone mould and degassed for an hour in an 80 • C vacuum oven and cured 28 min. in a 180 • C conventional oven. After curing, the specimens were cooled down to room temperature and separated from the silicone mould.

Characterizations
The viscosity of the epoxy mixture was measured by a Brookfield DV1 viscometer (AMETEK Brookfield, Middleborough, MA, USA). The viscosity measurements were conducted at room temperature while using a HB-07 spindle at 2.0 rpm.
The IWP tests of the epoxy adhesive test specimens were performed using an impact drop tower (Model 7520, Instron, Norwood, MA, USA) at room temperature. Two bent steel plates were bonded using the epoxy adhesives (area: 30 mm × 20 mm, thickness: 0.2 mm), and the force was measured when the adhesive layer was cleaved by the wedge at a velocity (v) of 2.0 m/s, as per ISO 11343. The cross-section morphologies of the epoxy adhesives after the IWP tests were observed by FE-SEM (JSM 6701F, JEOL, Tokyo, Japan).
The single lap shear tests of the epoxy adhesive test specimens were conducted using a dual-column universal testing machine (Model 5969, Instron, Norwood, MA, USA). The crossed specimens were bonded using the epoxy adhesives (area: 25 mm × 12.5 mm, thickness: 0.2 mm), and the shear stress measurements were performed at room temperature at 5 mm/min. crosshead displacement, as per the ISO 4587 standard.
The crosslink density (ρ) and glass transition temperature (T g ) of the epoxy adhesives were measured by a dynamic mechanical analyser (DMA 8000, Perkin Elmer, Waltham, MA, USA). The rectangular specimens (length: 32 mm, width: 2.0 mm, thickness: 1.0 mm) were prepared and tested in tension mode from −100 to 250 • C with 5 • C/min. ramp rate at a frequency of 1 Hz.
The molecular weights of PTPU and CTBN were measured by gel-permeation chromatography (GPC) (P-4000, FUTECS, Daejeon, South Korea). As a control, polystyrene with a PDI of approximately 1 was prepared. Subsequently, the sample was dissolved in THF for HPLC at 1 wt %. After that, the solution was placed in a cylinder and filtered by a syringe filter with 0.2 µm pore size. Finally, each sample was injected into GPC and then measured at 40 • C.

Thermomechanical Properties of the Epoxy Adhesives
The glass transition temperature (T g ) and crosslink density (ρ) of the cured epoxy adhesives were measured by a dynamic mechanical analyser (DMA) in order to determine the effect of the toughening agents. Figure 3 shows the storage modulus, loss tan δ, loss modulus, T g s and crosslink density of the epoxy adhesives as a function of the PTPU content. The glass-transition temperature was determined by the peak value of the tan δ curves, and the crosslink density was calculated based on the rubber elasticity theory [44,45]. As expected, the T g and storage modulus of the epoxy adhesives decreased as the PTPU content increased. Figure 3d shows that the T g decreased by approximately 8 • C for every Polymers 2020, 12, 1549 6 of 15 5 wt % increase in PTPU. The crosslink density remarkably decreased after the initial addition of PTPU, but the extent of decrease after the further addition of PTPU was diminished.
adhesives decreased as the PTPU content increased. Figure 3d shows that the Tg decreased by approximately 8 °C for every 5 wt % increase in PTPU. The crosslink density remarkably decreased after the initial addition of PTPU, but the extent of decrease after the further addition of PTPU was diminished.
FT-IR analysis of PTPU heat-treated at 180 °C was conducted to observe whether the urethane groups in PTPU reverted to isocyanate and alcohol to form allophanates during curing reaction. The reverse reaction scarcely occurred under our curing condition, and allophanate formation was too trivial, as shown in Figure S1. 30   CTBN led to results that were similar to those of PTPU, as shown in Figure 4. However, the Tg decreased by 4 °C for every 5 wt % because of the relatively high Tg of CTBN (−12 °C vs. −70 °C for PTPU). The crosslink density was also changed in a manner that differed from that of PTPU; it was constantly decreased with the CTBN content. When CTBN exceeded 15 wt %, the crosslink density became lower than that of the PTPU-modified sample. This result seemed to be due to the differences in the size, miscibility, and functional groups of CTBN and PTPU. FT-IR analysis of PTPU heat-treated at 180 • C was conducted to observe whether the urethane groups in PTPU reverted to isocyanate and alcohol to form allophanates during curing reaction. The reverse reaction scarcely occurred under our curing condition, and allophanate formation was too trivial, as shown in Figure S1.
CTBN led to results that were similar to those of PTPU, as shown in Figure 4. However, the T g decreased by 4 • C for every 5 wt % because of the relatively high T g of CTBN (−12 • C vs. −70 • C for PTPU). The crosslink density was also changed in a manner that differed from that of PTPU; it was constantly decreased with the CTBN content. When CTBN exceeded 15 wt %, the crosslink density became lower than that of the PTPU-modified sample. This result seemed to be due to the differences in the size, miscibility, and functional groups of CTBN and PTPU.
The GPC results show that CTBN has a much lower molecular weight (M n : 6800) than PTPU (M n : 18,700). CTBN may be more uniformly distributed in the epoxy network, causing a reduced crosslink density as compared with that caused by PTPU, owing to its small size and high miscibility. This observation is supported by the CTBN domain size, which is relatively smaller than that of PTPU. On the other hand, the reduced crosslink density of the PTPU-toughened epoxy adhesives could be explained, as follows. Based on the crosslink density and morphology of the PTPU-toughened epoxy adhesives, the sharp decrease in the crosslink density of the PTPU-toughened epoxy (at 5 wt % of PTPU) was believed to result from the hindrance of the curing reaction by the PTPU domains. However, as the PTPU content increased, more PTPU-epoxy adducts were formed and included in the network structure, alleviating the decrease in the crosslink density. In conclusion, it is believed that the crosslink density is determined by the degree of phase separation and the ease of toughening agent-epoxy adduct formation.   The GPC results show that CTBN has a much lower molecular weight (Mn: 6800) than PTPU (Mn: 18,700). CTBN may be more uniformly distributed in the epoxy network, causing a reduced crosslink density as compared with that caused by PTPU, owing to its small size and high miscibility. This observation is supported by the CTBN domain size, which is relatively smaller than that of PTPU. On the other hand, the reduced crosslink density of the PTPU-toughened epoxy adhesives could be explained, as follows. Based on the crosslink density and morphology of the PTPU-toughened epoxy adhesives, the sharp decrease in the crosslink density of the PTPU-toughened epoxy (at 5 wt % of PTPU) was believed to result from the hindrance of the curing reaction by the PTPU domains. However, as the PTPU content increased, more PTPU-epoxy adducts were formed and included in the network structure, alleviating the decrease in the crosslink density. In conclusion, it is believed that the crosslink density is determined by the degree of phase separation and the ease of toughening agent-epoxy adduct formation.
Viscosity is also an important physical property of epoxy adhesives. Normally viscosity increases with the addition of PTPU or CTBN [46]. As shown in Figure 5, the viscosity of the PTPUmodified epoxy adhesive was twice as high as that of the CTBN-modified epoxy adhesive, because PTPU has higher viscosity than CTBN. Because too-high viscosity causes a wetting problem, more than 30 wt % of PTPU could not be used. Viscosity is also an important physical property of epoxy adhesives. Normally viscosity increases with the addition of PTPU or CTBN [46]. As shown in Figure 5, the viscosity of the PTPU-modified epoxy adhesive was twice as high as that of the CTBN-modified epoxy adhesive, because PTPU has higher viscosity than CTBN. Because too-high viscosity causes a wetting problem, more than 30 wt % of PTPU could not be used.

Impact Wedge-Peel (IWP) Properties
The dynamic mechanical properties of the epoxy adhesives were measured by the IWP test, which is a particularly easy method for measuring the resistance to cleavage fracture. Figure 6 shows

Impact Wedge-Peel (IWP) Properties
The dynamic mechanical properties of the epoxy adhesives were measured by the IWP test, which is a particularly easy method for measuring the resistance to cleavage fracture. Figure 6 shows the time-to-force curves obtained from the IWP tests of the PTPU-and CTBN-toughened epoxy adhesives. According to Blackman et al., the case in which the force value is not immediately broken after the impact and has a plateau region between 25% and 90% of the measurement time is called "stable crack growth." In contrast, the case in which the adhesive force cannot be sustained and is broken after the impact is called "unstable crack growth" [38].

Impact Wedge-Peel (IWP) Properties
The dynamic mechanical properties of the epoxy adhesives were measured by the IWP test, which is a particularly easy method for measuring the resistance to cleavage fracture. Figure 6 shows the time-to-force curves obtained from the IWP tests of the PTPU-and CTBN-toughened epoxy adhesives. According to Blackman et al., the case in which the force value is not immediately broken after the impact and has a plateau region between 25% and 90% of the measurement time is called "stable crack growth." In contrast, the case in which the adhesive force cannot be sustained and is broken after the impact is called "unstable crack growth" [38]. In Figure 6, the PTPU-toughened epoxy showed a plateau region when 10 wt % or more PTPU content was added. However, the CTBN-toughened epoxy adhesives showed no plateau region, regardless of the CTBN content, meaning that all of the CTBN-toughened epoxies were broken by unstable crack growth. Based on the IWP test results, the dynamic resistance to cleavage, cleavageforce, displacement for cleavage, and total absorbed energy were obtained, as follows. The dynamic resistance to cleavage is the average value of force in the plateau region, and the cleavage-force is the average value of force measured between 25% and 90% of the measurement time. The displacement for cleavage is obtained by multiplying the measurement time and test speed (2 m/s), and the total absorbed energy is calculated by multiplying the area of force-to-time curve and test speed. In Figure 6, the PTPU-toughened epoxy showed a plateau region when 10 wt % or more PTPU content was added. However, the CTBN-toughened epoxy adhesives showed no plateau region, regardless of the CTBN content, meaning that all of the CTBN-toughened epoxies were broken by unstable crack growth. Based on the IWP test results, the dynamic resistance to cleavage, cleavage-force, displacement for cleavage, and total absorbed energy were obtained, as follows. The dynamic resistance to cleavage is the average value of force in the plateau region, and the cleavage-force is the average value of force measured between 25% and 90% of the measurement time. The displacement for cleavage is obtained by multiplying the measurement time and test speed (2 m/s), and the total absorbed energy is calculated by multiplying the area of force-to-time curve and test speed. Figure 7 shows the cleavage-force and displacement for the cleavage of the PTPU-and CTBN-toughened epoxy adhesives, and Figure 8 shows the total absorbed energy of the PTPUand CTBN-toughened epoxy adhesives. Figure 7 shows that the cleavage-force values of 0 and 5 wt % PTPU contents were greatly decreased. The higher the PTPU content, the higher the cleavage-force value in the stable crack growth region. However, the stable and unstable crack growth coexisted in the 30 wt % PTPU samples. The displacement for cleavage increased with the PTPU content, especially in the stable crack growth region, and then decreased in the 30 wt % PTPU samples, like the cleavage-force value. Moreover, the cleavage-force of the CTBN-toughened epoxy adhesives decreased monotonically with the CTBN content, and the displacement for cleavage seemed to increase and then decrease with the CTBN content, but the difference was trivial.
In the case of the total absorbed energy, the PTPU-toughened epoxy adhesives exhibited much higher values than the CTBN-toughened epoxy adhesives for all of the compositions. In addition, the PTPU-toughened epoxy adhesives showed noticeable changes in the total absorbed energy with the PTPU content. However, the CTBN-toughened samples showed almost constant values, regardless of the content. Thus, the CTBN-toughened epoxy adhesives are not tough enough to endure the high-speed impact energy delivered by the IWP test. When considering Figure 7, this difference seems to originate mainly from the displacement for cleavage. To explain this, the deformed IWP test specimens and their fracture surfaces were examined. toughened epoxy adhesives. Figure 7 shows that the cleavage-force values of 0 and 5 wt % PTPU contents were greatly decreased. The higher the PTPU content, the higher the cleavage-force value in the stable crack growth region. However, the stable and unstable crack growth coexisted in the 30 wt % PTPU samples. The displacement for cleavage increased with the PTPU content, especially in the stable crack growth region, and then decreased in the 30 wt % PTPU samples, like the cleavage-force value. Moreover, the cleavage-force of the CTBN-toughened epoxy adhesives decreased monotonically with the CTBN content, and the displacement for cleavage seemed to increase and then decrease with the CTBN content, but the difference was trivial. In the case of the total absorbed energy, the PTPU-toughened epoxy adhesives exhibited much higher values than the CTBN-toughened epoxy adhesives for all of the compositions. In addition, the PTPU-toughened epoxy adhesives showed noticeable changes in the total absorbed energy with the PTPU content. However, the CTBN-toughened samples showed almost constant values, regardless of the content. Thus, the CTBN-toughened epoxy adhesives are not tough enough to endure the highspeed impact energy delivered by the IWP test. When considering Figure 7, this difference seems to originate mainly from the displacement for cleavage. To explain this, the deformed IWP test specimens and their fracture surfaces were examined.   In the case of the total absorbed energy, the PTPU-toughened epoxy adhesives exhibited much higher values than the CTBN-toughened epoxy adhesives for all of the compositions. In addition, the PTPU-toughened epoxy adhesives showed noticeable changes in the total absorbed energy with the PTPU content. However, the CTBN-toughened samples showed almost constant values, regardless of the content. Thus, the CTBN-toughened epoxy adhesives are not tough enough to endure the highspeed impact energy delivered by the IWP test. When considering Figure 7, this difference seems to originate mainly from the displacement for cleavage. To explain this, the deformed IWP test specimens and their fracture surfaces were examined.   Figure 9 shows the deformed specimens after the IWP test. The specimens showing large total absorbed energies were severely deformed in the adhesive layers (Figure 9c-f)) and had a plateau region of stable crack growth. Thus, the more external impact energy that was transferred to the specimen, the higher the absorbed energy. Therefore, not only the toughness of the adhesive but also the interfacial adhesion is very important for achieving large absorbed energy.
It is necessary to introduce a soft material into the crosslinking structure or to complicate the crack propagation pathway through the formation of separate domains in order to improve the toughness of the epoxy. Due to this soft material, the T g and crosslink density of the epoxy were reduced. It is also known that excessive reduction in the crosslink density leads to a decrease in toughness [47]. When considering Figures 3d and 4d, at 10 wt % of toughening agent, the PTPU-toughened epoxy adhesive seems to be softer than the CTBN-toughened epoxy adhesives based on their T g and crosslink densities. As a result, the PTPU-toughened epoxy has higher total absorbed energy than the CTBN-toughened epoxy. As mentioned in 3.1, the PTPU-toughened epoxy adhesives showed a trivial reduction in crosslink density, while their T g significantly decreased with further increases in PTPU. However, the CTBN-toughened epoxy adhesives showed a serious decrease in crosslink density, which became lower than that of the PTPU-toughened epoxy adhesives after 15 wt % CTBN was added. This result implies that a high content of CTBN reduced the fracture toughness of the epoxy adhesives by an excessive reduction in crosslink density.
toughness [47]. When considering Figures 3d and 4d, at 10 wt % of toughening agent, the PTPUtoughened epoxy adhesive seems to be softer than the CTBN-toughened epoxy adhesives based on their Tg and crosslink densities. As a result, the PTPU-toughened epoxy has higher total absorbed energy than the CTBN-toughened epoxy. As mentioned in 3.1, the PTPU-toughened epoxy adhesives showed a trivial reduction in crosslink density, while their Tg significantly decreased with further increases in PTPU. However, the CTBN-toughened epoxy adhesives showed a serious decrease in crosslink density, which became lower than that of the PTPU-toughened epoxy adhesives after 15 wt % CTBN was added. This result implies that a high content of CTBN reduced the fracture toughness of the epoxy adhesives by an excessive reduction in crosslink density. Based on the fracture surface analysis, most of the samples showed cohesive failure of the epoxy adhesives, except for the samples in Figure 10j,k, which showed interfacial failure. It is difficult to distinguish between interfacial and cohesive failures because for some specimens, the adhesive at the impact site was broken off from the surface due to the high-impact energy. However, when considering the fracture surfaces after the lap shear test, we know that epoxy adhesives that contain more than 15 wt % CTBN have weak interfacial adhesion. This weak interfacial adhesion is because the amount of epoxy in the adhesives decreases as the content of CTBN increases. This interfacial failure is more easily observed in the CTBN-toughened epoxy adhesives, because CTBN is smaller than PTPU (Mn = 6800 vs. 18,700). Therefore, epoxy adhesives have three times more CTBN molecules than PTPU molecules, even at the same wt %. Based on the fracture surface analysis, most of the samples showed cohesive failure of the epoxy adhesives, except for the samples in Figure 10j,k, which showed interfacial failure. It is difficult to distinguish between interfacial and cohesive failures because for some specimens, the adhesive at the impact site was broken off from the surface due to the high-impact energy. However, when considering the fracture surfaces after the lap shear test, we know that epoxy adhesives that contain more than 15 wt % CTBN have weak interfacial adhesion. This weak interfacial adhesion is because the amount of epoxy in the adhesives decreases as the content of CTBN increases. This interfacial failure is more easily observed in the CTBN-toughened epoxy adhesives, because CTBN is smaller than PTPU (M n = 6800 vs. 18,700). Therefore, epoxy adhesives have three times more CTBN molecules than PTPU molecules, even at the same wt %.

Fracture Surface Morphology of IWP Specimens
It is well known that the separate domain formed by the toughening agent during the curing reaction plays an important role in improving the toughness of the epoxy. The number and size of the domain depend on the solubility parameter and chemical interaction between the toughening agent and epoxy [48][49][50]. Figure 11 shows the fractured surface morphology of the epoxy adhesives that were modified with different amounts of PTPU or CTBN. In the control sample, 200-300-nm spherical domains and

Fracture Surface Morphology of IWP Specimens
It is well known that the separate domain formed by the toughening agent during the curing reaction plays an important role in improving the toughness of the epoxy. The number and size of the domain depend on the solubility parameter and chemical interaction between the toughening agent and epoxy [48][49][50]. Figure 11 shows the fractured surface morphology of the epoxy adhesives that were modified with different amounts of PTPU or CTBN. In the control sample, 200-300-nm spherical domains and 12-µm irregular-shaped particles are observed due to the CSR and CaCO3 particles (Figure 11a,b). These domains and particles are observed in all of the samples in Figure 11. After adding the PTPU, 1-µm spherical domains appeared, and their size increased with the addition of up to 10 wt % PTPU (see Figure 11c,d). However, as shown in Figure 11e,f, the number and size of the spherical domains decreased rapidly, and they are not observed in Figure 11g. From Figure 11h-l, the CTBN-toughened epoxy adhesives show similar morphology to the PTPU-toughened epoxy adhesives. However, CTBN developed slightly smaller domains than PTPU.

Fracture Surface Morphology of IWP Specimens
It is well known that the separate domain formed by the toughening agent during the curing reaction plays an important role in improving the toughness of the epoxy. The number and size of the domain depend on the solubility parameter and chemical interaction between the toughening agent and epoxy [48][49][50]. Figure 11 shows the fractured surface morphology of the epoxy adhesives that were modified with different amounts of PTPU or CTBN. In the control sample, 200-300-nm spherical domains and 12-µm irregular-shaped particles are observed due to the CSR and CaCO3 particles (Figure 11a,b). These domains and particles are observed in all of the samples in Figure 11. After adding the PTPU, 1-µm spherical domains appeared, and their size increased with the addition of up to 10 wt % PTPU (see Figure 11c,d). However, as shown in Figure 11e,f, the number and size of the spherical domains decreased rapidly, and they are not observed in Figure 11g. From Figure 11h-l, the CTBN-toughened epoxy adhesives show similar morphology to the PTPU-toughened epoxy adhesives. However, CTBN developed slightly smaller domains than PTPU.

Single Lap Shear Properties
The shear strength of the epoxy adhesives was measured and is shown in Figure 12. When PTPU was used, the shear strength and displacement of the epoxy adhesives increased with the PTPU content due to the increased toughness. In particular, the failure mode changed from brittle to ductile at 30 wt % PTPU, where plastic deformation was clearly observed. However, all of the CTBN-toughened epoxy adhesives were fractured in brittle mode and their shear strength and displacement decreased when the CTBN content was over 15 wt %. In particular, clear substrate was observed in the fracture surfaces taken from the 15 wt % or more CTBN-toughened epoxy adhesives (see Figure 13). After examining the fracture surface and T g and ρ values, the drastic decrease in the shear strength of the CTBN-toughened epoxy adhesives is attributed to a decrease in the interfacial adhesion and the crosslink density.
toughened epoxy adhesives were fractured in brittle mode and their shear strength and displacement decreased when the CTBN content was over 15 wt %. In particular, clear substrate was observed in the fracture surfaces taken from the 15 wt % or more CTBN-toughened epoxy adhesives (see Figure  13). After examining the fracture surface and Tg and ρ values, the drastic decrease in the shear strength of the CTBN-toughened epoxy adhesives is attributed to a decrease in the interfacial adhesion and the crosslink density.

Conclusions
Polyurethane (PTPU)-and butadiene-acrylonitrile rubber (CTBN)-based toughening agents were used to improve the high-speed impact resistance of epoxy-based adhesives, and the following conclusions were obtained.
The glass-transition temperatures and crosslink densities of the epoxy adhesives changed differently, depending on the toughening agent (PTPU or CTBN). The degree of Tg reduction was highly dependent on the Tg of the toughening agent itself, and the crosslink density depends on the phase separation and the ease of formation of toughening agent-epoxy adducts.
From the results of the IWP test, stable crack growth is observed if the epoxy adhesive effectively transfers external stress to the substrate, which means that the adhesive strength with the substrate

Conclusions
Polyurethane (PTPU)-and butadiene-acrylonitrile rubber (CTBN)-based toughening agents were used to improve the high-speed impact resistance of epoxy-based adhesives, and the following conclusions were obtained.