Enhancement of Bonding and Mechanical Performance of Epoxy Asphalt Bond Coats with Graphene Nanoplatelets

Improving bonding and mechanical strengths is important for the application of bond coats used in the construction of steel deck bridges. Graphene nanoplatelets (GNPs) are attractive nanofillers for polymer modification because of their low cost, ultra-high aspect ratio, and extraordinary thermal and mechanical performance. In this paper, GNPs were used to reinforce the epoxy asphalt bond coat (EABC). The morphology, viscosity–time behavior, contact angle, dynamic mechanical properties, and mechanical and bonding strengths of GNP-reinforced EABCs were investigated using laser confocal microscopy, a Brookfield rotational viscometer, a contact angle meter, dynamic mechanical analysis, a universal test machine, and single-lap shear and pull-off adhesion tests. GNP dispersed non-uniformly in the asphalt phase of EABC. The viscosity of the neat EABC was lowered with the inclusion of GNPs and thus the allowable construction time was extended. The existence of GNPs enhances the hydrophobicity of the neat EABC. When adding more than 0.2% GNP, the storage modulus, crosslinking density and glass transition temperatures of both asphalt and epoxy of the neat EABC increased. The mechanical and bonding properties of the neat EABC were greatly enhanced with the incorporation of GNPs. Furthermore, the mechanical and bonding strengths of the modified EABCs increased with the GNP content. GNP-reinforced EABCs can be utilized in the pavement of long-span steel bridges with long durability.

Orthotropic steel decks (OSDs) have been extensively utilized in bridge construction all over the world since 1948 thanks to their rapid erection, light weight, low maintenance costs, and high degree of prefabrication and standardization [20]. To protect the deck from Graphene nanoplate powders were bought from Suzhou Tanfeng Graphene Technology Co., Ltd. (Suzhou, China). The physical properties of GPNs are summarized in Table 1. Asphalt binder was provided by China Offshore Bitumen (Taizhou) Co., Ltd. (Taizhou, China). Table 2 presents an overview of the asphalt binder. Bisphenol A epoxy prepolymer with an epoxide number of 0.52 mol/100 g was obtained from Nantong Xingchen Synthetic Material Co., Ltd. (Nantong, China). The acid-based curing agent was self-prepared in the laboratory.

Preparation of GNP-Reinforced EABCs
Graphene nanoplatelet powders were mixed at 120 • C with the curing agent and base asphalt to prepare a masterbatch using a high-speed emulsifier at 4000 rpm for 15 min. Afterward, the GNP-reinforced bituminous curing agent was mixed with the epoxy prepolymer at the same temperature for five minutes at 200 rpm. Finally, the mixture was poured into a circular Teflon mold with a diameter of 100 mm. The cured GNP-reinforced EABC was removed from the mold after curing at 120 • C for four hours in an oven. A scheme of the preparation of GNP-reinforced EABC is shown in Figure 1. The mass ratio of the epoxy prepolymer, asphalt binder and curing agent is 1:2.5:2. In the modified EABCs, the weight percentages of GNPs are 0, 0.2%, 0.5% and 1%, respectively.

Preparation of GNP-Reinforced EABCs
Graphene nanoplatelet powders were mixed at 120 °C with the curing agent and base asphalt to prepare a masterbatch using a high-speed emulsifier at 4000 rpm for 15 min. Afterward, the GNP-reinforced bituminous curing agent was mixed with the epoxy prepolymer at the same temperature for five minutes at 200 rpm. Finally, the mixture was poured into a circular Teflon mold with a diameter of 100 mm. The cured GNP-reinforced EABC was removed from the mold after curing at 120 °C for four hours in an oven. A scheme of the preparation of GNP-reinforced EABC is shown in Figure 1. The mass ratio of the epoxy prepolymer, asphalt binder and curing agent is 1:2.5:2. In the modified EABCs, the weight percentages of GNPs are 0, 0.2%, 0.5% and 1%, respectively.

Viscosity-Time Behavior
The viscosity-time behavior during curing was tested on a Brookfield rotational viscometer (NDJ-1C, Changji, Shanghai, China) under the guideline of ASTM D4402. The uncured GNP-reinforced EABC was transferred into the chamber of the viscometer. The test was conducted at 120 °C using the No. 28 spindle at 50 rpm.

Phase-Separated Microstructures
The dispersion of GNPs in EABCs was observed on a confocal microscope (LSM710, Zeiss, Jena, Germany) with Ar + as the laser source (488 nm). To prepare the sample for confocal microscopy observation, a drop of uncured GNP-reinforced EABC liquid was the slide glass. After covering a cover glass, the glass slide sample was cured at 120 °C for four hours. The distribution of discontinuous asphalt domains in EABCs was determined by an image analysis software. The number-and weight-average diameters (dn and dw) and polydispersity index (PDI, dw/dn) were calculated using Equations (1) and (2):

Viscosity-Time Behavior
The viscosity-time behavior during curing was tested on a Brookfield rotational viscometer (NDJ-1C, Changji, Shanghai, China) under the guideline of ASTM D4402. The uncured GNP-reinforced EABC was transferred into the chamber of the viscometer. The test was conducted at 120 • C using the No. 28 spindle at 50 rpm.

Phase-Separated Microstructures
The dispersion of GNPs in EABCs was observed on a confocal microscope (LSM710, Zeiss, Jena, Germany) with Ar + as the laser source (488 nm). To prepare the sample for confocal microscopy observation, a drop of uncured GNP-reinforced EABC liquid was the slide glass. After covering a cover glass, the glass slide sample was cured at 120 • C for four hours. The distribution of discontinuous asphalt domains in EABCs was determined by an image analysis software. The number-and weight-average diameters (d n and d w ) and polydispersity index (PDI, d w /d n ) were calculated using Equations (1) and (2): where n i is the number of domains with a diameter of d i . To determine these parameters, three confocal microscopy images for each sample were used.

Hydrophobicity
The contact angles of GNP-reinforced EABCs were determined using a KSV CAM contact angle meter (Helsinki, Finland). The contact angle was recorded within 60 s after dripping 5 µL deionized water on the sample. At least five measurements were performed for each sample. Dynamic mechanical properties were evaluated on a DMA instrument (DMA + 450, 01 dB-Metravib, Limonest, France). The measurement was performed in tension mode at a heating rate of 3 • C/min and 1 Hz between −50 • C and 100 • C.

Mechanical Performance
Mechanical properties were determined using a universal testing machine (Instron 3366, Instron, Norwood, MA, USA) as per ASTM D638. The dog-bone-shaped samples were measured at room temperature and a crosshead speed of 200 mm/min. The mechanical properties were averaged from five samples.

Shear Strength
The single-lap shear measurements were also performed on a universal testing machine (Instron 3366, Instron, Norwood, MA, USA) as per ASTM D1002. The end surfaces (12.5 mm) of steel plates (100 × 25 × 2 mm 3 ) were sandblasted to remove all impurities and corrosion and cleaned with acetone. After coating with the uncured bond coat (12.5 mm), the parts of the two plates were overlapped and cured. Single-lap shear strength was determined at a crosshead speed of 50 mm/min after curing at 120 • C for 4 h and being cooled to room temperature. Five replicates were tested for each sample.

Pull-Off Adhesion Strength
Pull-off adhesion tests were conducted on a DeFelsko PosiTest AT-A portable automatic adhesion tester (Ogdensburg, NY, USA). Q345D steel plates (150 × 150 × 20 mm 3 ) were sandblasted and cleaned with acetone. The polished surfaces of steel plates were coated with a layer of uncured bond coat (600 g/m 2 ). Afterward, five dollies were secured on the steel plate surface. After curing at 120 • C for 4 h, a core drill was used to separate the testing areas around the dollies. Pull-off adhesion tests were performed using a uniform tensile force on the dolly by using the automatic adhesion tester under a speed of 0.7 MPa/s at room temperature and 60 • C.

Phase Separation
Laser confocal microscopy is a powerful tool for the investigation of the phase separation of epoxy asphalts due to the high contrast resolution between asphalt and crosslinked epoxy resin [31,44]. Figure 2 illustrates the microscopy images of GNP-reinforced EABCs in the fluorescence mode. As is known, asphalt and uncured epoxy resin, mainly composed of epoxy prepolymer and curing agent, are compatible [45]. However, the crosslinked networks formed during epoxy curing lead to the incompatibility between the crosslinked epoxy and asphalt. In this case, as shown in Figure 2, phase separation occurs, and thus results in the dispersion of spherical asphalt particles (black) in the continuous epoxy phase (yellow). When incorporating 0.2% GNP, the shape of asphalt particles is unchanged. However, as the GNP content increases, the shape of asphalt particles becomes irregular. It is worth mentioning that the dispersion of GNPs in EABCs cannot be observed in all of these fluorescence confocal microscopy images.
To distinguish GNPs from both asphalt and epoxy phases of EABCs, transmittance confocal microscopy images were used. As shown in Figure 3, the dispersion of GNPs in the EABC is still unseen. However, darker lamellar GNPs are non-uniformly dispersed in asphalt particles. It is important to note that GNPs are unseen in the epoxy phase, which does not mean that no GNPs exist in the epoxy phase. The lighter color of GNPs with fewer layers may be caused by the invisibility of GNPs in the epoxy phase. crosslinked epoxy and asphalt. In this case, as shown in Figure 2, phase separation oc and thus results in the dispersion of spherical asphalt particles (black) in the contin epoxy phase (yellow). When incorporating 0.2% GNP, the shape of asphalt particl unchanged. However, as the GNP content increases, the shape of asphalt particle comes irregular. It is worth mentioning that the dispersion of GNPs in EABCs cann observed in all of these fluorescence confocal microscopy images. To distinguish GNPs from both asphalt and epoxy phases of EABCs, transmitt confocal microscopy images were used. As shown in Figure 3, the dispersion of GN the EABC is still unseen. However, darker lamellar GNPs are non-uniformly dispers asphalt particles. It is important to note that GNPs are unseen in the epoxy phase, w does not mean that no GNPs exist in the epoxy phase. The lighter color of GNPs fewer layers may be caused by the invisibility of GNPs in the epoxy phase. Table 3 collects the dn, dw and PDIs of the asphalt particles in GNP-reinforced EA The inclusion of GNPs increases the average diameters of asphalt particles of the EABC. Besides, the average diameters of asphalt particles of the reinforced EABC crease in the GNP content. The GNP-reinforced EABCs have a higher PDI value than of the neat EABC, indicating that GNPs lead to the inhomogeneous dispersion of asp particles in the continuous epoxy phase. Furthermore, the GNP content has a mar influence on the dispersion of asphalt particles in the epoxy phase.    Table 3 collects the d n , d w and PDIs of the asphalt particles in GNP-reinforced EABCs. The inclusion of GNPs increases the average diameters of asphalt particles of the neat EABC. Besides, the average diameters of asphalt particles of the reinforced EABCs increase in the GNP content. The GNP-reinforced EABCs have a higher PDI value than that of the neat EABC, indicating that GNPs lead to the inhomogeneous dispersion of asphalt particles in the continuous epoxy phase. Furthermore, the GNP content has a marginal influence on the dispersion of asphalt particles in the epoxy phase.  Figure 4 presents rotational viscosity as a function of time of GNP-reinforced EABCs. At 0 min, the viscosities of the neat EABC and the modified EABCs containing 0.2%, 0.5% and 1.0% GNPs are 390, 360, 320 and 310 mPa·s, respectively, indicating that the viscosity of the neat EABC is declined somewhat with the inclusion of GNPs, which is attributed to the low viscosity of low-molecular-weight curing agent and epoxy prepolymer and the small size of the graphene nanoplatelet, whose thin lamellar structure plays a lubricating role [46]. As time proceeds, the viscosity of all EABCs slightly declines within 5 min, because the reaction between epoxy prepolymers and curing agents is exothermic, and thus results in a typical viscosity decrease [43]. As the reaction proceeds, the viscosity decrease caused by heat absorption is much lower than the viscosity increase generated by the molecular weight increase in epoxy resin. In this case, the viscosity increases gradually with time as the molecular weight increases until the appearance of the gel point. After that, the viscosity increases sharply. Evidently, the viscosity of the neat EABC is higher than that of GNP-reinforced ones during the whole curing process. Furthermore, the viscosity of the reinforced EABCs decreases in the GNP content. In other words, the existence of GNPs delays the curing between epoxy prepolymers and curing agents due to the lubricating effect of the lamellar structure of graphene nanoplatelet. Thus, the addition of GNPs extends the allowable construction time of the neat EABC. Additionally, with the increase in the GNP content, the allowable construction time of the reinforced EABCs is prolonged.  Similar to GNPs, other nanofillers, such as GO, MMT and ATT, also extend the allowable construction time of the neat EABC [35,47,48]. Table 4 summarizes the time to 5000 mPa·s of GNP-reinforced EABCs at 120 °C . Among all of the modified EABCs, GNPreinforced EABC with 1.0% GNPs exhibits the longest allowable construction time.  Similar to GNPs, other nanofillers, such as GO, MMT and ATT, also extend the allowable construction time of the neat EABC [35,47,48]. Table 4 summarizes the time to 5000 mPa·s of GNP-reinforced EABCs at 120 • C. Among all of the modified EABCs, GNP-reinforced EABC with 1.0% GNPs exhibits the longest allowable construction time.

Hydrophobicity
The water contact angles of GNP-reinforced EABCs are presented in Figure 5. The water contact angle of the neat EABC is 98.2 • . Clearly, GNPs increase the water contact angle of the neat EABC. The water contact angle of the reinforced EABCs slightly increases with GNP content. A maximum value (102.6 • ) appears at 0.5%, which is 4.4 • higher than that of the neat EABC. The water contact angle (102.3 • ) slightly decreases when the GNP content increases to 1.0%. It is known that a material is hydrophobic when its water contact angle is over than 90 • . Therefore, GNPs improve the hydrophobicity of the neat EABC, which is attributed to the inherently high hydrophobicity of GNPs [49].

Storage Modulus (E′)
The E′ as a function of temperature of GNP-reinforced EABCs is shown in Figure 6. As the temperature increases, the storage modulus gradually decreases at the glassy state (T < Tg, glass transition temperature), followed by a sharp decrease and a gradual decrease during the glass transition and rubbery (T > Tg) stages. When adding 0.2% GNP, the E′ of the neat EABC is improved in the glassy state, while the increase in the E′ during the glass transition and rubbery stages is imperceptible. When adding more than 0.2% GNPs, the E′ of the neat EABC is enhanced over the whole temperature range.

Storage Modulus (E )
The E as a function of temperature of GNP-reinforced EABCs is shown in Figure 6. As the temperature increases, the storage modulus gradually decreases at the glassy state (T < T g , glass transition temperature), followed by a sharp decrease and a gradual decrease during the glass transition and rubbery (T > T g ) stages. When adding 0.2% GNP, the E of the neat EABC is improved in the glassy state, while the increase in the E during the glass transition and rubbery stages is imperceptible. When adding more than 0.2% GNPs, the E of the neat EABC is enhanced over the whole temperature range. The E′ as a function of temperature of GNP-reinforced EABCs is shown in Figure 6. As the temperature increases, the storage modulus gradually decreases at the glassy state (T < Tg, glass transition temperature), followed by a sharp decrease and a gradual decrease during the glass transition and rubbery (T > Tg) stages. When adding 0.2% GNP, the E′ of the neat EABC is improved in the glassy state, while the increase in the E′ during the glass transition and rubbery stages is imperceptible. When adding more than 0.2% GNPs, the E′ of the neat EABC is enhanced over the whole temperature range.

Loss Modulus (E″)
Generally, in DMA, the Tg can either be determined as the inflection point at which a sudden drop of the storage modulus takes place in a storage modulus vs. temperature curve, or the temperature at the maximum E″ or loss factor (tan δ) in an E″ or tan δ vs. temperature curve [50]. In this case, the Tg value can be different for a given polymer when being determined using different DMA parameters, because the inflection point of the E′ takes place at a temperature lower than the temperature of the maximum E″, eventually

Loss Modulus (E )
Generally, in DMA, the T g can either be determined as the inflection point at which a sudden drop of the storage modulus takes place in a storage modulus vs. temperature curve, or the temperature at the maximum E or loss factor (tan δ) in an E or tan δ vs. temperature curve [50]. In this case, the T g value can be different for a given polymer when being determined using different DMA parameters, because the inflection point of the E takes place at a temperature lower than the temperature of the maximum E , eventually followed by the temperature of the maximum tan δ. In this work, both E" and tan δ were used to evaluate the T g s of GNP-reinforced EABCs.
E" as a function of temperature of GNP-reinforced EABCs, is depicted in Figure 7. Two peaks appear at the glassy state in all E"-versus-temperature curves, which can be attributed to the T g s of asphalt and of epoxy [51]. From a polymeric material point of view, epoxy asphalt is a polymer blend containing epoxy and asphalt, which have two individual T g s [52]. Epoxy asphalt (epoxy/asphalt blend) exhibits two distinct T g s-the T g of crosslinked epoxy resin at a higher temperature, and the T g of asphalt at a lower temperature-because of the incompatibility between crosslinked epoxy resin and asphalt. Table 5 presents the T g s of epoxy and asphalt of GNP-reinforced EABCs obtained from loss modulus-temperature curves. The T g s of asphalt and epoxy of the neat EABC are −21.3 • C and 14.5 • C, respectively. GNPs increase the T g of asphalt for the neat EABC. GNPs increase the T g of epoxy of the neat EABC apart from the 0.2% loading. Additionally, with the increase in the GNP content, the T g s of both asphalt and epoxy of the reinforced EABCs increase. linked epoxy resin at a higher temperature, and the Tg of asphalt at a lower temperaturebecause of the incompatibility between crosslinked epoxy resin and asphalt. Table 5 presents the Tgs of epoxy and asphalt of GNP-reinforced EABCs obtained from loss modulustemperature curves. The Tgs of asphalt and epoxy of the neat EABC are −21.3 °C and 14.5 °C , respectively. GNPs increase the Tg of asphalt for the neat EABC. GNPs increase the Tg of epoxy of the neat EABC apart from the 0.2% loading. Additionally, with the increase in the GNP content, the Tgs of both asphalt and epoxy of the reinforced EABCs increase.

Loss Factor
The tan δ as a function of temperature of GNP-reinforced EABCs is illustrated in Figure 8. Like loss modulus-temperature curves, two peaks are exhibited in all tan δ-temperature curves. In addition, compared to the tan δ peak of the asphalt, the peak of epoxy is more intensive and broader. The Tg of asphalt and the Tg of epoxy of the pure EABC are −12.9 °C and 28.1 °C , respectively, as shown in Table 5. GNPs increase the Tg of asphalt of the pure EABC. Nevertheless, the GNP content has little influence on the Tg of asphalt of the reinforced EABCs. When adding 0.2% GNP, the Tg of epoxy of the neat EABC is nearly unchanged. The Tgs of epoxy of the modified EABCs with more than 0.2% GNP are greater than that of the pure EABC. Furthermore, with increasing GNP content, the glass transition temperature of epoxy of the reinforced EABCs increases.

Loss Factor
The tan δ as a function of temperature of GNP-reinforced EABCs is illustrated in Figure 8. Like loss modulus-temperature curves, two peaks are exhibited in all tan δ-temperature curves. In addition, compared to the tan δ peak of the asphalt, the peak of epoxy is more intensive and broader. The T g of asphalt and the T g of epoxy of the pure EABC are −12.9 • C and 28.1 • C, respectively, as shown in Table 5. GNPs increase the T g of asphalt of the pure EABC. Nevertheless, the GNP content has little influence on the T g of asphalt of the reinforced EABCs. When adding 0.2% GNP, the T g of epoxy of the neat EABC is nearly unchanged. The T g s of epoxy of the modified EABCs with more than 0.2% GNP are greater than that of the pure EABC. Furthermore, with increasing GNP content, the glass transition temperature of epoxy of the reinforced EABCs increases. It is believed that the Tg of a thermosetting polymer depends greatly on the crosslinking density (CD). The crosslinking density can be calculated using Equation (3) based on the elasticity theory [45]: where E′r is the E′ at Tr (Tg of epoxy + 40 K) in the rubbery state. R is the gas constant. As can be observed from Table 5, except for the loading of 0.2%, GNPs increase the crosslinking density of the pure EABC. Moreover, the crosslinking density of the reinforced EABCs increases in the GNP content. Consequently, the Tgs of both asphalt and epoxy of the neat EABC are enhanced with the incorporation of GNPs apart from the 0.2% GNP content for the Tg of epoxy, as discussed previously.

Cole-Cole Plot
The E″-E′ curve is called a Cole-Cole plot, and is a useful tool for describing the effect of microfiller and nanofiller on the homogeneous or heterogeneous phase and structural changes in polymer composites [53]. A homogeneous polymeric system shows a smooth semicircular arc, whereas a heterogeneous polymeric system exhibits two or more semicircular arcs. Furthermore, the imperfect or irregular shape of semicircular arcs indicates phase heterogeneity or ununiform dispersion in the polymeric system [54]. The Cole-Cole plots of the GNP-reinforced EABCs are shown in Figure 9. For the neat EABC, there are It is believed that the T g of a thermosetting polymer depends greatly on the crosslinking density (CD). The crosslinking density can be calculated using Equation (3) based on the elasticity theory [45]: where E r is the E at T r (T g of epoxy + 40 K) in the rubbery state. R is the gas constant. As can be observed from Table 5, except for the loading of 0.2%, GNPs increase the crosslinking density of the pure EABC. Moreover, the crosslinking density of the reinforced EABCs increases in the GNP content. Consequently, the T g s of both asphalt and epoxy of the neat EABC are enhanced with the incorporation of GNPs apart from the 0.2% GNP content for the T g of epoxy, as discussed previously.

Cole-Cole Plot
The E -E curve is called a Cole-Cole plot, and is a useful tool for describing the effect of microfiller and nanofiller on the homogeneous or heterogeneous phase and structural changes in polymer composites [53]. A homogeneous polymeric system shows a smooth semicircular arc, whereas a heterogeneous polymeric system exhibits two or more semicircular arcs. Furthermore, the imperfect or irregular shape of semicircular arcs indicates phase heterogeneity or ununiform dispersion in the polymeric system [54]. The Cole-Cole plots of the GNP-reinforced EABCs are shown in Figure 9. For the neat EABC, there are two smooth semicircular arcs: the one at the higher modulus is the epoxy resin and the other one at the lower modulus is for asphalt and for the epoxy resin because of the incompatibility between the crosslinked epoxy and asphalt. In this case, phase-separated microstructures appear in all EABCs (Figures 2 and 3). With the addition of GNPs, both arcs for asphalt and epoxy shift to the higher modulus because of the reinforcement effect of graphene nanoplatelets. Meanwhile, the shape of the asphalt arcs of the GNP-reinforced EABCs becomes irregular compared to that of epoxy arcs, indicating that the dispersion of GNPs in the asphalt is inhomogeneous as shown in Figure 3.  Figure 10 presents the mechanical properties of GNP-reinforced EABCs. GNPs significantly enhance the mechanical performance of the neat EABC. Furthermore, with increasing GNP content, the mechanical performance of the modified EABCs increases. The elongation at break and tensile strength of the neat EABC increase by 82% and 68%, respectively, with the inclusion of 1.0% GNP. Klimek-McDonald and coworkers revealed that the tensile strength of the neat epoxy is improved when incorporating less than 1.5 vol.% GNP [55]. However, a contrary trend was found in another epoxy system [11]. The increase in tensile strength is due to the higher crosslinking density of the neat EABC increased with the addition of GNPs as mentioned previously. Noteworthily, the elongation at break in all these epoxy systems decreases with the presence of GNPs. As shown in Figure 3, GNPs are dispersed in the dispersed asphalt phase, and thus increase the elongation at break of the pure EABC [56].  Figure 10 presents the mechanical properties of GNP-reinforced EABCs. GNPs significantly enhance the mechanical performance of the neat EABC. Furthermore, with increasing GNP content, the mechanical performance of the modified EABCs increases. The elongation at break and tensile strength of the neat EABC increase by 82% and 68%, respectively, with the inclusion of 1.0% GNP. Klimek-McDonald and coworkers revealed that the tensile strength of the neat epoxy is improved when incorporating less than 1.5 vol.% GNP [55]. However, a contrary trend was found in another epoxy system [11]. The increase in tensile strength is due to the higher crosslinking density of the neat EABC increased with the addition of GNPs as mentioned previously. Noteworthily, the elongation at break in all these epoxy systems decreases with the presence of GNPs. As shown in Figure 3, GNPs are dispersed in the dispersed asphalt phase, and thus increase the elongation at break of the pure EABC [56].

Mechanical Performance
vol.% GNP [55]. However, a contrary trend was found in another epoxy system [11]. The increase in tensile strength is due to the higher crosslinking density of the neat EABC increased with the addition of GNPs as mentioned previously. Noteworthily, the elongation at break in all these epoxy systems decreases with the presence of GNPs. As shown in Figure 3, GNPs are dispersed in the dispersed asphalt phase, and thus increase the elongation at break of the pure EABC [56]. In a tensile test, the integral of the stress (σ) vs. strain (ε) curve indicates the toughness (τ) of the material [57,58]. Thus, where εb is the elongation at break. To determine the brittleness (B) of polymeric materials, Brostow et al. [59] proposed an empirical equation: In a tensile test, the integral of the stress (σ) vs. strain (ε) curve indicates the toughness (τ) of the material [57,58]. Thus, where ε b is the elongation at break. To determine the brittleness (B) of polymeric materials, Brostow et al. [59] proposed an empirical equation: where E RT is the E at room temperature gained from DMA. The toughness and brittleness of GNP-reinforced EABCs are illustrated in Figure 11. GNPs improve the toughness and thus decrease the brittleness of the neat EABC. Additionally, with the increase in the GNP content, the toughness of the reinforced EABCs increases (Figure 11a). Correspondingly, the brittleness of the modified EABCs exhibits an opposite trend (Figure 11b). With the incorporation of 1.0% GNP, the toughness of the pure EABC increases by 79%.
FOR PEER REVIEW 12 of 18 where E′RT is the E′ at room temperature gained from DMA. The toughness and brittleness of GNP-reinforced EABCs are illustrated in Figure 11. GNPs improve the toughness and thus decrease the brittleness of the neat EABC. Additionally, with the increase in the GNP content, the toughness of the reinforced EABCs increases (Figure 11a). Correspondingly, the brittleness of the modified EABCs exhibits an opposite trend (Figure 11b). With the incorporation of 1.0% GNP, the toughness of the pure EABC increases by 79%.

Single-Lap Shear Strength
The shear strengths of GNP-reinforced EABCs at room temperature and 60 °C are shown in Figure 12. GNPs enhance the shear strength of the neat EABC. Furthermore, with the increase in GNP content, the shear strength of the reinforced EABCs increases. When adding 1.0% GNP, the shear strengths of the neat EABC at 60 °C and room temper-

Single-Lap Shear Strength
The shear strengths of GNP-reinforced EABCs at room temperature and 60 • C are shown in Figure 12. GNPs enhance the shear strength of the neat EABC. Furthermore, with the increase in GNP content, the shear strength of the reinforced EABCs increases. When adding 1.0% GNP, the shear strengths of the neat EABC at 60 • C and room temperature increase by 26% and 36%, respectively.

Single-Lap Shear Strength
The shear strengths of GNP-reinforced EABCs at room temperature and 60 °C are shown in Figure 12. GNPs enhance the shear strength of the neat EABC. Furthermore, with the increase in GNP content, the shear strength of the reinforced EABCs increases. When adding 1.0% GNP, the shear strengths of the neat EABC at 60 °C and room temperature increase by 26% and 36%, respectively.  The failure surfaces of GNP-reinforced EABCs after single-lap shear tests are shown in Figures 13 and 14. The failure surfaces of steel plates exhibit a mixed mode of cohesive and adhesive failures. Generally, adhesive failure takes place at the interface of the bond coat and the substrate. The bonding layer at the interface weakens and leads to failure with the smooth surface left on the substrate, which exhibits weak bonding strength. Cohesive failure occurs within the bond coat with the rough surface left on the substrate, indicating that the bonding strength of the bond coat itself is lower than the bonding strength between the bond coat and the substrate. In this case, cohesive failure reveals good bonding strength [60]. For a bond coat with a mixed failure mode, the lower the area of adhesive failure is, the higher the bonding strength is.
The areas of adhesive failure on the steel plates bonded with GNP-reinforced EABCs after single-lap shear tests are illustrated in Figure 15. For the neat EABC, the area of adhesive failure and cohesive failure at room temperature are nearly the same. However, at 60 • C, the area of adhesive failure is slightly lower than that of cohesive failure. When adding GNPs, cohesive failure mode becomes dominant, and thus the presence of GNPs enhances the shear strength of the neat EABC. Furthermore, with increasing GNP content, the area of adhesive failure of the reinforced EABCs slightly decreases. Consequently, for the reinforced EABCs, the shear strength increases as the GNP content increases. with the smooth surface left on the substrate, which exhibits weak bonding strength. Cohesive failure occurs within the bond coat with the rough surface left on the substrate, indicating that the bonding strength of the bond coat itself is lower than the bonding strength between the bond coat and the substrate. In this case, cohesive failure reveals good bonding strength [60]. For a bond coat with a mixed failure mode, the lower the area of adhesive failure is, the higher the bonding strength is.    The areas of adhesive failure on the steel plates bonded with GNP-reinforced EABCs after single-lap shear tests are illustrated in Figure 15. For the neat EABC, the area of adhesive failure and cohesive failure at room temperature are nearly the same. However, at 60 °C , the area of adhesive failure is slightly lower than that of cohesive failure. When adding GNPs, cohesive failure mode becomes dominant, and thus the presence of GNPs enhances the shear strength of the neat EABC. Furthermore, with increasing GNP content, the area of adhesive failure of the reinforced EABCs slightly decreases. Consequently, for the reinforced EABCs, the shear strength increases as the GNP content increases.   The areas of adhesive failure on the steel plates bonded with GNP-reinforced EABCs after single-lap shear tests are illustrated in Figure 15. For the neat EABC, the area of adhesive failure and cohesive failure at room temperature are nearly the same. However, at 60 °C , the area of adhesive failure is slightly lower than that of cohesive failure. When adding GNPs, cohesive failure mode becomes dominant, and thus the presence of GNPs enhances the shear strength of the neat EABC. Furthermore, with increasing GNP content, the area of adhesive failure of the reinforced EABCs slightly decreases. Consequently, for the reinforced EABCs, the shear strength increases as the GNP content increases.

Pull-Off Adhesion Strength
The pull-off adhesion strengths of GNP-reinforced EABCs at room temperature and 60 • C are shown in Figure 16. Like single-lap shear strength, the pull-off adhesion strength of the neat EABC is enhanced with the incorporation of GNPs. For the modified EABCs, the pull-off adhesion strength increases with the GNP content. When adding 1.0% GNP, the pull-off adhesion strengths of the pure EABC at 60 • C and room temperature increase by 61% and 28%, respectively. The pull-off adhesion strengths of GNP-reinforced EABCs at room temperature and 60 °C are shown in Figure 16. Like single-lap shear strength, the pull-off adhesion strength of the neat EABC is enhanced with the incorporation of GNPs. For the modified EABCs, the pull-off adhesion strength increases with the GNP content. When adding 1.0% GNP, the pull-off adhesion strengths of the pure EABC at 60 °C and room temperature increase by 61% and 28%, respectively.   Figures 17 and 18 present the surfaces of steel plates and dollies bonded with GNPreinforced EABCs after pull-off adhesion tests. For the dollies, adhesive failure is overwhelming both at room temperature and 60 • C. For steel plates, the room-temperature pull-off failure mode is a mixed failure. However, adhesive failure is overwhelming at 60 • C. Figure 16. Pull-off adhesion strengths of GNP-reinforced EABCs at room temperature and 60 °C . Figures 17 and 18 present the surfaces of steel plates and dollies bonded with GNPreinforced EABCs after pull-off adhesion tests. For the dollies, adhesive failure is overwhelming both at room temperature and 60 °C . For steel plates, the room-temperature pull-off failure mode is a mixed failure. However, adhesive failure is overwhelming at 60 °C .   Figure 19 shows the areas of adhesive failure of steel plates bonded by GNP-reinforced EABCs after room-temperature pull-off adhesion tests. Similar to single-lap shear tests, the areas of adhesive failure and cohesive failure of the neat EABC are nearly equivalent. When incorporating GNPs, the area of adhesive failure of the neat EABC significantly decreases. Besides, in the case of GNP reinforced EABCs, the area of adhesive failure decreases with the GNP content. Thus, GNPs enhance the pull-off adhesion strength of the neat EABC. Moreover, with the increase in the GNP content, the pull-off strength of reinforced EABCs increases. It is known that, in addition to high mechanical strength and modulus, graphene materials have a high aspect ratio, which increases the load transfer to the polymer matrix and thus improves the bonding strength of the neat EABC [61].  Figure 19 shows the areas of adhesive failure of steel plates bonded by GNP-reinforced EABCs after room-temperature pull-off adhesion tests. Similar to single-lap shear tests, the areas of adhesive failure and cohesive failure of the neat EABC are nearly equivalent. When incorporating GNPs, the area of adhesive failure of the neat EABC significantly decreases. Besides, in the case of GNP reinforced EABCs, the area of adhesive failure decreases with the GNP content. Thus, GNPs enhance the pull-off adhesion strength of the neat EABC. Moreover, with the increase in the GNP content, the pull-off strength of reinforced EABCs increases. It is known that, in addition to high mechanical strength and modulus, graphene materials have a high aspect ratio, which increases the load transfer to the polymer matrix and thus improves the bonding strength of the neat EABC [61]. alent. When incorporating GNPs, the area of adhesive failure of the neat EABC significantly decreases. Besides, in the case of GNP reinforced EABCs, the area of adhesive failure decreases with the GNP content. Thus, GNPs enhance the pull-off adhesion strength of the neat EABC. Moreover, with the increase in the GNP content, the pull-off strength of reinforced EABCs increases. It is known that, in addition to high mechanical strength and modulus, graphene materials have a high aspect ratio, which increases the load transfer to the polymer matrix and thus improves the bonding strength of the neat EABC [61]. Figure 19. Area of adhesive failure of steel plates bonded by GNP-reinforced EABCs after roomtemperature pull-off adhesion tests.

Conclusions
In this study, GNP-reinforced EABCs were prepared and characterized. When adding more than 0.2% GNP, the shape of spherical asphalt particles in EABCs becomes irregular. GNPs increas the diameters of the neat EABC. Furthermore, with the increase in GNP content, the average diameters of the reinforced EABCs increase. Laser confocal microscopy observation revealed that GNPs are dispersed in the dispersed asphalt phase of the EABC. Due to the lubricating effect of the GNPs, the viscosity of the pure EABC is lowered, indicating that the allowable construction time of the neat EABC is extended Figure 19. Area of adhesive failure of steel plates bonded by GNP-reinforced EABCs after roomtemperature pull-off adhesion tests.

Conclusions
In this study, GNP-reinforced EABCs were prepared and characterized. When adding more than 0.2% GNP, the shape of spherical asphalt particles in EABCs becomes irregular. GNPs increas the diameters of the neat EABC. Furthermore, with the increase in GNP content, the average diameters of the reinforced EABCs increase. Laser confocal microscopy observation revealed that GNPs are dispersed in the dispersed asphalt phase of the EABC. Due to the lubricating effect of the GNPs, the viscosity of the pure EABC is lowered, indicating that the allowable construction time of the neat EABC is extended with the inclusion of GNPs. In addition, the extension effect is more evident with higher GNP content. GNPs improve the hydrophobicity of the neat EABC. When introducing over 0.2% GNP, the storage modulus during the whole temperature range, the crosslinking density and the T g of the epoxy and the T g of asphalt are increased. Cole-Cole plots revealed that GNPs result in the inhomogeneity of asphalt and thus the non-uniform dispersion of graphene nanoplatelets in the asphalt particles of the EABC. GNPs significantly improved both the mechanical and bonding properties of the neat EABC. Furthermore, with increasing GNP content, the mechanical and bonding strengths of the reinforced EABCs increased. When adding 1.0% GNP, the tensile strength, elongation at break and toughness of the neat EABC increased by 68%, 82% and 79%, respectively. The shear strengths and pull-off adhesion strengths of the neat EABC at room temperature and 60 • C improved by 36%, 28%, 26% and 61%, respectively.