Self-Healing Effect of Various Capsule-Core Materials on Asphalt Materials

,


Introduction
During the service life of roadway, trafc loading and temperature cycling may cause asphalt pavement cracking [1,2]. After cracks appear, moisture damage of asphalt concrete may occur under the repeated action of the load and rain [3]. Self-healing microcapsule materials could be used in the asphalt pavement to improve its self-repairing ability [4]. In recent years, functional materials with selfhealing and self-repairing capabilities have been developed [5][6][7].
At present, the research on self-healing asphalt concrete based on microcapsule has focused on the preparation of microcapsules, the evaluation of repair efciency, and the testing of microcapsule mechanical indexes. Te study of mechanical properties of microcapsules is limited to the feld of polymer materials, and the whole self-healing system is not considered. To better evaluate the healing properties of asphalt material with self-healing microcapsule, researchers have studied the action characteristics between microcapsule and asphalt materials [8][9][10]. Sun et al. mechanically stirred the prepared MUF microcapsules into asphalt of 160°C and studied the properties of the capsule materials in hightemperature asphalt [11]. Zhang et al. mixed self-healing microcapsule into asphalt mixture according to diferent dosages and studied the efects of microcapsule materials on high and low-temperature performance and fatigue performance of asphalt mixture [12]. Al Mansoori et al. found that the cracking of asphalt mixture containing capsule returned to 52.9% of the initial strength, and the cracking of asphalt mixture without capsule returned to 14.0% of the initial strength at 20°C [13]. Su et al. used circular heatingcooling process tests and found that microcapsules still kept a stable state after an extreme temperature change simulating temperature change in the natural environment. Moreover, interface debonding phenomenon did not appear. Tese results indicate that microcapsules can remain stable in the asphalt materials [14].
Some studies have focused on the healing ability of the asphalt itself after damage. Wang et al. [15] analyzed the efects of maintenance time, temperature, and mix type on their self-healing properties from a single factor using a fourpoint bending fatigue test and proposed evaluation indexes to quantify the healing efect. On top of that, Huang et al. [16] further investigated the efects of damage degree and asphalt modifer on the self-healing properties of the mix. Trough asphalt binder strength tests, Lv et al. [17] found that the type and amount of modifer had a great infuence on the healing ability of asphalt mastic, and the increase of modifer weakened the self-healing ability of asphalt and soft asphalt had better healing ability. In addition, several researchers have investigated the rupture mechanism and release process of self-healing microcapsules. Zhu et al. used multi-scale analysis and composite fracture mechanics to study the mechanical property requirements of microencapsulated materials in asphalt pavements and the rupture mechanism of microcapsules under the action of microcracks [18]. Moreno-Navarro et al. found that asphalt mixtures with macroscopic cracks are more sensitive to healing conditions, and the healing rate increases with the rise of healing temperature [19]. Moreover, the healing ability of matrix asphalt is higher than that of modifed asphalt; however, the application of heating or external forces as healing techniques has a lower recovery rate at macroscopic damage, but Zhong used dissipative energy theory to study and found that the healing ability of polymer modifed asphalt was higher than that of matrix asphalt [20]. Qiu et al. tested the healing ability of asphalt mixtures using elastic foundation beams and evaluated quantitatively based on strength recovery and crack width, and the results showed that delayed viscoelastic healing was the main reason for the narrowing of crack width, and viscous healing played a dominant role in prolonged or high-temperature conditions [21].
According to the above research, the interface of crack can be quickly healed due to adhesion of adhesives and cohesion of the materials. Te solidifed adhesive layer has certain strength [22][23][24]. Te capsule-core materials of microcapsules successfully prepared are divided into diluter and adhesive. Te light oil of rejuvenator is used as a diluter to improve fow deformation property of asphalt by adjusting the viscosity and colloid structure of asphalt and to improve its self-healing ability. Te epoxy resin mixed with curing agent as a adhesive to quickly heal the fractured interface of the asphalt material through its own cohesive force. Te solidifed adhesive layer has a certain strength, which could reliably connect two interfaces, transfer stress, and resist damage [25][26][27]. It is necessary to study and develop new experiments and new indicators to evaluate the self-healing properties of asphalt materials containing microcapsules.
In this study, asphalt-rejuvenator-asphalt (ARA) and asphalt-epoxy resin-asphalt (AEA) models were frst developed and used to study the healing behavior of asphalt and capsule-core materials using Materials Studio software.
Also, the molecular cross-linking and simulated stretching were implemented using the Perl script. Ten, the model of bi-plate DSR tests were designed and the model was used to characterize the self-healing properties of asphalt material. Finally, the healing indexes (HI 1 and HI 2 ) were evaluated to quantify the self-healing efect of rejuvenator and epoxy resin in asphalt materials. Tis study excluded the interference of capsule wall and compared the healing efect of two types of capsulate-core materials with two diferent mechanisms of action, which will provide theoretical guidance for the selection and use scenarios of microcapsule core materials.

Modeling.
To study the self-healing efect of epoxy resin and rejuvenator on asphalt from molecular scale, two molecular models of asphalt-epoxy resin (with hardener)-asphalt (AEA) and asphalt-rejuvenator-asphalt (ARA) were established in this study using Materials Studio 2020, as shown in Figure 1. Among them, the asphalt model is the four-component AAA-1 model proposed by Li and Greenfeld [28]. Te molecular models of epoxy resin and curing agent are DGEBA and IPD molecules, respectively. Te molecular model of the rejuvenator is shown in Figure 2.

Simulation Process.
First, the 12-component molecules were introduced into the cubic box using the Amorphous cell module according to the AAA-1 model number, and the initial density was set to 0.1 g/cm 3 . Ten, the initial model was geometrically optimized in 10,000 steps. Te optimized model was subjected to 500 ps of molecular difusion under NVT ensemble. Finally, NPT simulations of 500 ps were performed on the asphalt model to approximate the actual density at 298.15 K (25°C). COMPASS II was chosen for all force felds during the simulation process.
Te ratio of the number of DGEBA and IPD molecules is 1 : 2. AEA and ARA models were built using the Build layer function. Subsequently, 10,000 steps of geometry optimization were performed for both models. Ten, the models were simulated at NVT for 500, 1000, 1500, and 2000 ps at a difusion temperature of 293.15 K (20°C), respectively. In addition, the AEA model was subjected to molecular crosslinking of DGEBA and IPD using a Perl script to simulate the reaction of epoxy resin and curing agent, and the crosslinking process is shown in Figure 3.

Molecular Simulation Tensile Test.
In this study, a Perl script was used to implement a model for direct stretching experiments. Te healing efect of the AEA and ARA models was investigated by stretching the models. Te stretched models were subjected to geometric optimization to obtain kinetic parameters, and the stretching process is shown in Figure 4. Finally, the stresses generated after stretching the models were counted to evaluate the healing efect.

Materials and Experiments
3.1. Materials. SK-90# asphalt is selected as base asphalt. Te proportion of the mineral powder and binders in the asphalt mortar is 1 : 1. Te mineral powders are limestone fnes (≤0.075 mm). Te rejuvenator was produced by Zhuhai Jieda Petrochemical Trading Co. Ltd., and the composition is light oil containing aromatics and saturates. Te adhesive was epoxy resin (DGEBA), and curing agent (IPD) was produced by Jining Hongming Chemical Reagent Co. Ltd. Te composition of IPD is isophoronediamine (C 10 H 22 N 2 ). Te major properties of raw materials are shown in Table 1. Figure 5 shows the three microcapsules prepared using three core materials. Te research team studied the properties of the microcapsules and optimized the preparation process [27,29].

Experimental Methods and Parameters.
Te rheological properties of asphalt binders and mastics under diferent temperatures, stresses, and strains were tested using DSR. Te plates of both PP08 and PP25 were used under low and high-temperature conditions. Based on the DSR test, an experimental specimen was designed in this study to investigate the self-healing efect of the capsule-core materials in the asphalt materials. Tis specimen is mainly used to simulate the situation after rupture of microcapsule and release of core material. Firstly, the 0.003 g ± 0.0005 g capsule-core material (1.0% of asphalt content) is evenly applied between two asphalt specimens (10 mm in diameter and 2 mm thick) to prepare the bi-plate DSR tests, which formed a thickness of approximately 3.8 μm capsule-core material. Te illustration of bi-plate DSR test and test samples is shown in Figure 6.
Te asphalt-rejuvenator-asphalt specimens were placed in a constant temperature chamber at 20°C for 1-6 days. Te asphalt-epoxy resin (with curing agent)-asphalt specimens were placed in a 20°C thermostat to heal for 6, 12, 18, and 24 hours, respectively.
Te healing indexes (HI 1 and HI 2 ) were used to evaluate self-healing properties of asphalt materials, calculated as shown equations (1) and (2).
where G * healed is the initial dynamic shear modulus (Pa) of the bi-plate asphalt specimen and G * original is the initial dynamic shear modulus of the undamaged asphalt specimen.
where N healed is the number of loading cycles of the bi-plate asphalt specimen when dynamic shear modulus drops to 50% of the initial value and N original represents the corresponding loading cycles of undamaged asphalt specimen when dynamic shear modulus drops to 50% of the initial value. Moreover, linear and exponential functions were used to ft the healing index based on dynamic shear modulus (HI 1 ) and fatigue life (HI 2 ), respectively. Te ftting function used in this study was as shown in equations (3) and (4). Te R 2 values of ftting functions were found in the range of 0.91-0.98, indicating the strong correlation.

Cohesive Energy of Models with Various Simulation
Times. Te cohesive energy could be used to evaluate the cohesive properties of materials. A higher value of cohesive energy represents better intermolecular adhesion of the system. Terefore, the cohesive energy is used to evaluate the healing efect of rejuvenator and epoxy resin with asphalt. Figure 7 shows the ARA molecular model with diferent simulation times. Te cohesive energy of the ARA and AEA systems was calculated according to equation (5). Te calculated results are shown in Table 2 and Figure 8.

Advances in Civil Engineering 5
where E c is the cohesive energy of the model; E A is the energy of the asphalt phase; E B is the energy of the rejuvenator or epoxy phase; and E T is the total energy of the system.   larger than that of ARA under the same conditions, which indicates that the adhesion of epoxy resin and asphalt is better. With the increase of healing time, the cohesive energy of ARA model gradually increases while that of the AEA model increases frst and then becomes stable. Tis refects the slow healing process of the rejuvenator and the earlier strength of the epoxy resin. In terms of the healing mechanism, the rejuvenator heals mainly by promoting the diffusion of the asphalt itself, while the epoxy resin cures to increase the adhesion between the asphalt to achieve healing [27].

Stress of Models with Various Simulation Times.
To further investigate the mechanical properties of the healed model, a Perl script was used to apply 50% strain to the model and count the stresses generated in the model after stretching for evaluating the healing efect. Figure 9 shows the AEA model after stretching. Te statistical results of the stresses are shown in Table 3 and Figure 10. From Figure 10, it can be seen that the stress value of the ARA model is the highest at the healing time of 1500 ps. Tis indicates that the system has the highest strength at this time. Tis difers from the results for the cohesive energy.
Tere are two possible reasons for this scenario: frst, the experimental results are disturbed by the phenomenon of molecular chemical bond breakage that is not considered in the simulation of stretching [30]. Second, with the increase of difusion time, the molecules of the rejuvenator gradually difused into the asphalt interior causing the asphalt at the interface to become softer [31]. Te stress value of the AEA model increases with the increase of the healing time. Te model tensile state in Figure 9 also shows that the longer the healing time, the greater the degree of molecular cross-linking after stretching. Tis is basically the same as the actual epoxy curing process [32]. Te stress values of the ARA model for the same simulation time are greater than those of the AEA model. It indicates that the ductility recovery of the material by the rejuvenator is better than that by the epoxy resin. Increasing the curing agent content will reduce the ductility of the epoxy resin material [33].  Figure 11. As can be seen in Figure 11(a), the complex shear modulus (G * ) of both asphalt binder and mastic increased briefy or remained fat during the initial warming phase. Tis may be due to the reorientation of molecules under the shear action. Also, the specimen thickness was 2 mm, and the temperature of the specimen could not rise uniformly for a short time, thus making the modulus rise briefy. However, after a gradual increase in temperature, the efect of temperature on the specimen is much greater than that of thickness, which leads to a gradual decrease in the modulus of asphalt and asphalt mastic with the increase in temperature. In Figure 11(b), G * of asphalt binder and mastic gradually decreases with increasing temperature. In the initial stage, there is no similar situation as in the lowtemperature results. Tis is probably due to the fact that the thickness of the asphalt specimen is 1 mm, and the temperature of the specimen changes more uniformly during the heating process, and the efect of temperature is much larger than the efect of specimen thickness. Tis also fully illustrates that asphalt binder and mastic are temperature-sensitive materials.

Stress-Strain Characteristics.
Te main control parameters of the stress scan are loading frequency of 10 Hz, test temperature of 20°C, and stress scan range of 0.01-1 MPa. Te main control parameters of the strain scan are loading frequency of 10 Hz, test temperature of 20°C, and strain scan range of 0.01-100%. From Figure 12(a), it can be seen that G * of asphalt mastic has been residing above the matrix asphalt at the beginning of the loading phase. With the gradual increase of shear strain, the modulus of the base asphalt and mastic began to decrease slowly, and then there was a sharp decline in the later stage, which was especially obvious for the asphalt mastic. Te asphalt mastic specimens were the frst to break than the base asphalt, showing a characteristic similar to brittle fracture, while the base asphalt broke relatively slowly. Tis indicates that the toughness of the base asphalt is higher than that of the asphalt mastic. From Figure 12(b), it can be seen that the property change pattern of base asphalt and mastic under stress scan mode is basically the same as that of the strain sweep test.   Figure 13. From Figure 13(a), it can be seen that for every 0.1 MPa increase in the control stress, G * of the asphalt increases to some extent, but its fatigue life decreases by orders of magnitude. From Figure 13(b), it can be seen that the trend of G * change of asphalt mastic is similar to that of asphalt binder, but its fatigue life decreases at a slower rate compared to that of asphalt binder. G * of the asphalt binder specimen changes more slowly in the early stage and then decays faster when the specimen reaches a certain fatigue level. G * of asphalt mastic decayed more uniformly in the early stage and more slowly in the later stage of damage. Tis indicates that the fatigue resistance of asphalt mastic is signifcantly better than the base asphalt [34].

Efect of Rejuvenator on Self-Healing.
Te DSR bi-plate specimens were prepared with rejuvenator and asphalt binder or mastic as raw materials and maintained at a   Advances in Civil Engineering constant temperature of 20°C for 1-6 days in an environmental chamber. Te fatigue tests were performed sequentially. Figure 14 shows the fatigue test results of DSR biplate specimens at diferent stresses. Table 4 presents initial dynamic shear modulus of asphalt materials with rejuvenator at various stresses. It can be seen from Figure 14 and Table 4 that the initial modulus and fatigue life of the specimens increase with the extension of the curing time. It can be seen from Figure 15(a) that HI 1 of the specimens gradually increased with the increase of the healing time. Te healing index is infuenced by the control stress, and the higher the control stress is, the higher the healing index is. When the control stress was 0.3 MPa, the healing index of specimens with healing time of 5 and 6 days exceeded 1. A healing index greater than 1 indicates that the introduction of the rejuvenator could not only heal the asphalt material but also improve the performance of the asphalt material. As shown in Figure 15(b), the HI 2 index of the specimens increased relatively slowly in the early stage but increased substantially in the later stage. HI 2 is also infuenced by the control stress, and HI 2 of 0.3 MPa is always greater than the HI 2 value of 0.2 MPa. Te healing index of 0.1 MPa is less than 0.3 MPa during the short-term healing time, but the growth rate of the healing index is much greater than 0.3 MPa when the maintenance time exceeds 4 days. Comparing Figures 15(a) and 15(b), it can be seen that the  healing index calculated by the modulus ratio is much larger than the healing index calculated by the number of loadings. Similarly, HI 1 and HI 2 of asphalt mastic were calculated and the relationships between healing index and healing time are shown in Figure 16. Te R-square values of ftting functions were found in the range of 0.90-0.99, indicating the strong correlation. Te results indicate that self-healing properties of asphalt mastic and asphalt binder show consistent trends with healing time. It can be seen from Figure 16(a) that HI 1 of the specimens increased gradually  with the growth of the maintenance time. Te healing index HI 1 of the specimens was greater than 0.7 after 6 days. From Figure 16(b), it can be seen that HI 2 of the specimens grew exponentially, with a relatively slow growth in the early stage, but a burst growth in the later stage. HI 2 was also infuenced by the test control stress. HI 2 of 0.2 MPa was always at the minimum value, and the healing index of 0.3 MPa was greater than 0.4 MPa during the short-term maintenance time. Te healing index of 0.4 MPa increased faster and exceeded 0.3 MPa as the maintenance time was extended. Tis indicates that after the asphalt is damaged, the rejuvenator could heal it in a short period of time, but the damage location is still a weak surface of the whole, which can be easily destroyed under repeated loading, and the recovery of its fatigue performance is a long-term process.
Comparing the healing index results of the rejuvenator on the binder and the mastic, the healing index value of the slurry is greater than that of the binder under the same test conditions. Tis indicates that the self-healing efect of the rejuvenator on the binder is more obvious [31].

Efect of Epoxy Resin on Self-Healing.
Te DSR bi-plate specimens of epoxy resin were maintained in an environmental chamber at 20°C for 6, 12, 18, and 24 hours and then subjected to fatigue tests. Te fatigue test results of the asphalt binder and mastic specimens at the control stress of 0.2 MPa are shown in Figure 17. Table 5 presents initial dynamic shear modulus of asphalt materials with epoxy resin at various stresses. It can be seen from Figure 17 that the initial modulus and fatigue life of the specimens increase with the extension of the curing time under the same control stress condition. Figure 18 shows the ftted curve of healing index of asphalt binder specimens. It can be seen from Figure 18(a) that the healing index HI 1 of the bi-plate specimens gradually increased with the increase of the curing time. Te healing index HI 1 was greater than 1.0 after 24 hours. Under the conditions of control stress of 0.2 MPa and 0.3 MPa, the healing index HI 1 was greater than 1.0 after healing for 18 hours. Also, the healing index was infuenced by the control stress; the greater the control stress, the higher the measured healing index. When the healing index is greater than 1, the introduction of the adhesive can not only heal the asphalt material but also improve the performance of the original asphalt material. From Figure 18(b), it can be seen that the healing index HI 2 of the specimens grows in a linear manner. Comparing Figures 18(a) and 18(b), the healing index calculated by the initial complex shear modulus ratio is much larger than that calculated by the number of loadings, which indicates that the binder can heal the asphalt in a short period of time after the asphalt has been damaged [35]. Figure 19 shows the ftted curve of healing index of asphalt mastic specimens. It can be seen from Figure 19 Under the condition of control stress of 0.4 MPa, the damage of the specimen was observed at the contact interface between the parallel plate and the specimen during the test, which belonged to the debonding between the specimen and the parallel plate, rather than the fatigue damage of the specimen itself; the two main reasons for this phenomenon were as follows: (1) the adhesion between the asphalt mastic itself and the parallel plate was small and the control stress was large during the test; (2) with the extension of the curing time, the epoxy resin cured to a higher degree. Te degree of curing of the epoxy resin improves with longer maintenance time allowing the mechanical properties of the asphalt to recover and the specimen is easily debonded from the parallel plate under higher stress.     Advances in Civil Engineering 13 Comparing Figures 19(a) and 19(b), it can be seen that the variation pattern of the healing index HI 2 of the two-piece specimen is basically the same as that of the healing index HI 1 . Te healing index HI 1 was greater than 0.8 and the healing index HI 2 was greater than 0.4 for the 24 h specimens under the control stress of 0.2 MPa and 0.3 MPa. Te healing index calculated by the initial complex shear modulus ratio was much greater than that calculated by the number of loadings. Tis indicates that after the asphalt damage, the adhesive can make it heal in a short period of time, but fatigue damage is likely to occur under the action of repeated loading, and it takes a longer period of time to recover the fatigue performance.  Comparing the healing index of epoxy resin to asphalt binder and mastic, the healing index value of asphalt binder is greater than that of asphalt mastic under the same conditions. Tis result indicates that the epoxy resin has a better healing efect on the asphalt binder diferent from rejuvenator. When the control stress was 0.3 MPa, the healing indexes HI 1 and HI 2 of the epoxy were greater than those of the rejuvenator, which indicated better self-healing of the epoxy. Te value of HI 1 even   exceeded 1, which was due to the high strength of the cured epoxy [22]. Tis indicates that the epoxy resin material can better restore the shear strength of asphalt.

Conclusions
In this study, molecular dynamic simulations and rheological tests were used to investigate the efect of two capsulecore materials with various self-healing mechanisms such as rejuvenator and epoxy resin. Te main conclusions are as follows: (1) Te epoxy resin has a stronger bond to the asphalt than the rejuvenator. Tis is due to the fact that the epoxy resin can be used as a binder while the main component of the rejuvenator is light oil according to the molecular simulation results. Importantly, the ARA model produced the highest stress after stretching. Tis indicates that the efect of the rejuvenator in restoring the ductility of the asphalt is promising. (2) During the action of the rejuvenator and the asphalt material under the same controlled stress condition, the healing indexes HI 1 and HI 2 have a sound linear correlation with the maintenance time and the healing index HI 1 is much larger than HI 2 , which indicates that after the asphalt is damaged, the rejuvenator can heal in a period of time. However, the damage position is still a weak point of the whole, which is easily destroyed under repeated loading, and the recovery of fatigue performance is a longterm process. (3) Due to the high strength of the cured epoxy resin itself, the healed asphalt material has sound shear resistance. However, the amount of curing agent needs to be adjusted according to the actual situation. (4) From the above results, the rejuvenator is recommended as microencapsulated materials for healing cracks due to temperature stress, while epoxy resin is more suitable for cracks due to vehicle load shear.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.

Authors' Contributions
Suhua Chen was responsible for methodology, investigation, data curation, formal analysis, validation, original draft preparation, and review and editing. Qi Liu was responsible for conceptualization, software, review and editing, and supervision. Yanqiu Bi was responsible for original draft preparation, review and editing, and funding acquisition. Bin Yu was responsible for review and editing, supervision, and funding acquisition. Jiupeng Zhang was responsible for review and editing and validation.