Performance and Fracture Analysis of Composite Interfaces for Semi-Flexible Pavement

: Semi-ﬂexible pavement is widely used in pavement engineering due to its excellent rutting resistance; however, it mainly fails due to cracking. Therefore, it is important to understand the properties of the aggregate–mortar–asphalt interfacial transition zone, to better understand the cracking mechanism of the semi-ﬂexible pavement. In this work, we used pull-off tests and digital image analysis technology to compare and analyze the interfacial tensile strength and granite– bitumen–mortar interactions in three types of asphalt (70# matrix asphalt, PG76-22 modiﬁed asphalt and S-HV modiﬁed asphalt) at different curing ages. The analysis results showed that, for the three different bitumen materials, with settled mortar, the peak interfacial tensile strength values all occurred at approximately 14 d of curing. In addition, the order of the tensile strength followed the order of asphalt penetration degree; the order of the interfacial water damage resistance from weak to strong was 70# asphalt cementation specimen, PG76-22 modiﬁed asphalt cementation specimen, and S-HV modiﬁed asphalt cementation specimen. The results of this analysis highlight the original contributions of the optimum curing time for the composite interface of semi-ﬂexible pavement materials prepared with different asphalts to reach optimum crack resistance.


Introduction
Semi-flexible pavement (SFP) refers to composite pavements that have characteristics of both asphalt and cement concrete, and are formed by filling a special slurry with cement as the main component in the open-graded asphalt mixture with a large void matrix [1][2][3][4]. Semi-flexible pavements are also known as semi-rigid pavements, grouted macadam composite materials, cement grout asphalt composites, asphalt-Portland cement concrete composites (APCCC), and composite cement-asphalt mixtures [5]. The SFP is constructed without the use of expansion, contraction, and construction joints and demonstrates prominent rutting and shoving-corrugation resistance. These new pavement materials possess both the flexible characteristics of asphalt pavement and the high strength (hardness) of concrete pavement [5]. As a result, they have broad application prospects, such as overloaded road surfaces, intersections, and bus stations [6]. However, cracking is the main problem that hinders the acceptance of semi-flexible pavements [7]. In addition, the interfacial bonding strength of the composite material will affect its overall performance [8], and the semi-flexible pavement mortar-asphalt interface is considered the weak link.
In 1960, SFP was first successfully used as airport pavement in France [2,9]. Ever since, there have been numerous studies on the structural design and engineering properties of SFP materials, and the corresponding grouting cement mortar [6,[10][11][12][13][14]. Cihackova [15] used wheel tracking, stiffness modules, skid resistance, and low-temperature cracking tests and found that the resistance of SFP against rutting is considerably higher than asphalt mixtures. However, the resistance of SFP against low-temperature cracking is lower than asphalt mixtures. Corradini [16] et al. studied the fatigue properties of SFP by indirect Table 1. Properties of the 70# petroleum bitumen.

Test Aggregates
Granite was selected as the test aggregate. As a type of magmatic rock, granite is mainly composed of feldspar, quartz, biotite, and other minerals and, owing to its high density, it has a denser texture and good thermal stability. When heated in an oven, some of the water was removed before reacting, as the material exhibits poor physical permeability and poor water absorption. When placed in a curing box at greater than 80% humidity, the material will not absorb a considerable amount of water. Thus, the moisture will not penetrate the asphalt interface from the aggregate; otherwise, it could affect the test results. The basic physical properties of granite are shown in Table 4, and the granite used in this work were pitted granite, which was purchased from the Guangzhou Stone Market. The granite specimens were formed by a double-sided cutting machine in the laboratory, and the cutting machine is shown in Figure 1. The molded granite specimen was 4 cm × 3 cm × 3 cm in size, and the specimen is shown in Figure 2. used in this work were pitted granite, which was purchased from the Guangzhou Stone Market. The granite specimens were formed by a double-sided cutting machine in the laboratory, and the cutting machine is shown in Figure 1. The molded granite specimen was 4 cm × 3 cm × 3 cm in size, and the specimen is shown in Figure 2.   The mortar was prepared using water, cement, and special additives (including sand), which was provided by Guangdong Longhu Technology Co., Ltd. (Shantou, China) [26], using a ratio of 0.23 of water to powder (cement and additives), and a cement/additive ratio of 45:55 to produce the cement mortar. An image of the prepared mortar is shown in Figure 3.
This work utilized white Portland cement, and the initial setting time of the test piece was no longer than 45 min, with a final setting time of no longer than 390 min, and a standard strength grade of 42.5R. The chemical composition and physical properties are shown in Tables 5-7. used in this work were pitted granite, which was purchased from the Guangzhou Stone Market. The granite specimens were formed by a double-sided cutting machine in the laboratory, and the cutting machine is shown in Figure 1. The molded granite specimen was 4 cm × 3 cm × 3 cm in size, and the specimen is shown in Figure 2.   The mortar was prepared using water, cement, and special additives (including sand), which was provided by Guangdong Longhu Technology Co., Ltd. (Shantou, China) [26], using a ratio of 0.23 of water to powder (cement and additives), and a cement/additive ratio of 45:55 to produce the cement mortar. An image of the prepared mortar is shown in Figure 3.
This work utilized white Portland cement, and the initial setting time of the test piece was no longer than 45 min, with a final setting time of no longer than 390 min, and a standard strength grade of 42.5R. The chemical composition and physical properties are shown in Tables 5-7.  The mortar was prepared using water, cement, and special additives (including sand), which was provided by Guangdong Longhu Technology Co., Ltd. (Shantou, China) [26], using a ratio of 0.23 of water to powder (cement and additives), and a cement/additive ratio of 45:55 to produce the cement mortar. An image of the prepared mortar is shown in Figure 3.
This work utilized white Portland cement, and the initial setting time of the test piece was no longer than 45 min, with a final setting time of no longer than 390 min, and a standard strength grade of 42.5R. The chemical composition and physical properties are shown in Tables 5-7.   After using a cutting machine (TNS Infratest Testing Systems CO., LTD., Munich, Germany) to shape the specimen, we used a high-pressure water gun to rinse the surface. We then placed the granite specimen in an oven at 135 °C and dehydrated it for 6 h. The specimen was then marked and weighed as granite specimen . Then, the temperature was adjusted to 175 °C and the specimen was heated for 2 h.
The 70# petroleum asphalt, PG76-22-modified asphalt, and S-HV-modified asphalt were prepared and placed in the oven for the test. The upper surface of the heated granite specimen was placed on the asphalt for 3-5 s. Then, the upper surface of the granite specimen was attached to a layer of asphalt.
We then placed the granite specimen with the accompanying asphalt layer in the oven, and placed the 70# petroleum asphalt specimen in the oven at 135 °C. In addition, the S-HV modified asphalt and PG76-22 modified asphalt specimens attached were placed in the oven at 165 °C and heated for 20 min, causing the asphalt to flow freely and evenly on the upper surface of the granite specimen, forming a relatively uniform asphalt film.
Finally, the asphalt-coated granite specimen was cooled for one hour. Then, we weighed the asphalt-coated granite specimen ( ), using Equation (1) to calculate the asphalt film thickness (d)   After using a cutting machine (TNS Infratest Testing Systems CO., LTD., Munich, Germany) to shape the specimen, we used a high-pressure water gun to rinse the surface. We then placed the granite specimen in an oven at 135 • C and dehydrated it for 6 h. The specimen was then marked and weighed as granite specimen m 0 . Then, the temperature was adjusted to 175 • C and the specimen was heated for 2 h.
The 70# petroleum asphalt, PG76-22-modified asphalt, and S-HV-modified asphalt were prepared and placed in the oven for the test. The upper surface of the heated granite specimen was placed on the asphalt for 3-5 s. Then, the upper surface of the granite specimen was attached to a layer of asphalt.
We then placed the granite specimen with the accompanying asphalt layer in the oven, and placed the 70# petroleum asphalt specimen in the oven at 135 • C. In addition, the S-HV modified asphalt and PG76-22 modified asphalt specimens attached were placed in the oven at 165 • C and heated for 20 min, causing the asphalt to flow freely and evenly on the upper surface of the granite specimen, forming a relatively uniform asphalt film. Finally, the asphalt-coated granite specimen was cooled for one hour. Then, we weighed the asphalt-coated granite specimen (m 1 ), using Equation (1) to calculate the asphalt film thickness (d) where d is the asphalt film thickness (µm), m 0 is the weight of the granite specimen (g), m 1 is the weight of the granite specimen with the asphalt film (g), A is the area of granite covered with asphalt (cm 2 ), and ρ is the density of the asphalt (g/cm 3 ). The film thickness is shown in Table 8, and the molded test piece is shown in Figure 4. where is the asphalt film thickness (μm), is the weight of the granite specimen (g), is the weight of the granite specimen with the asphalt film (g), is the area of granite covered with asphalt (cm ), and is the density of the asphalt (g/cm ). The film thickness is shown in Table 8, and the molded test piece is shown in Figure 4.

Preparing Mortar and Test Specimen
Cement mortar samples were prepared according to the JGJ/T 70-2009 standard[27]. First, we formed the mold before pouring the mixture, which was cut into a square piece with a side length of 3 cm using a sandwich wood board, as shown in Figure 5a. We then used an oil-based pen and a steel ruler to measure 1.1 cm on the small wooden block and marked the scale. Then, a 0.5 cm wide tape was used to fix the small wooden block and form the mold according to the scale line. Note that the small wooden blocks were sealed without leaving any gaps during splicing; otherwise, the mortar would leak out of the gap when pouring the mortar. The test piece is shown in Figure 5b.
Then, we poured the prepared cement mortar into the mold and allowed it to cure at room temperature for 24 h. Afterward, it was placed in a constant temperature curing box with greater than 95% humidity, and at a temperature of 20 °C for 24 h. The mold was then removed, and the test piece was placed in the curing box. After aging for a prescribed time, the test piece was removed, and a layer of mortar was scraped off the surface of the test piece with a utility knife. Then, we used sandpaper to reduce the mortar to a thickness of 1 cm ± 0.1 cm, as shown in Figure 5b. Afterward, we used AB glue to glue the mold at both ends of the test piece, as shown in Figure 5c. The final test specimen is shown in Figure 5d.

Preparing Mortar and Test Specimen
Cement mortar samples were prepared according to the JGJ/T 70-2009 standard [27]. First, we formed the mold before pouring the mixture, which was cut into a square piece with a side length of 3 cm using a sandwich wood board, as shown in Figure 5a. We then used an oil-based pen and a steel ruler to measure 1.1 cm on the small wooden block and marked the scale. Then, a 0.5 cm wide tape was used to fix the small wooden block and form the mold according to the scale line. Note that the small wooden blocks were sealed without leaving any gaps during splicing; otherwise, the mortar would leak out of the gap when pouring the mortar. The test piece is shown in Figure 5b.
Then, we poured the prepared cement mortar into the mold and allowed it to cure at room temperature for 24 h. Afterward, it was placed in a constant temperature curing box with greater than 95% humidity, and at a temperature of 20 • C for 24 h. The mold was then removed, and the test piece was placed in the curing box. After aging for a prescribed time, the test piece was removed, and a layer of mortar was scraped off the surface of the test piece with a utility knife. Then, we used sandpaper to reduce the mortar to a thickness of 1 cm ± 0.1 cm, as shown in Figure 5b. Afterward, we used AB glue to glue the mold at both ends of the test piece, as shown in Figure 5c. The final test specimen is shown in Figure 5d.

Pull-Off Test
As shown in Figure 6, we used a universal testing machine (MTS SYSTEMS (CHINA) CO., LTD. Shenzhen, China) to perform the uniaxial tensile test with a pull rate of 1 mm/min, and we measured the peak force.

Pull-Off Test
As shown in Figure 6, we used a universal testing machine (MTS SYSTEMS (CHINA) CO., LTD. Shenzhen, China) to perform the uniaxial tensile test with a pull rate of 1 mm/min, and we measured the peak force.
To control for errors, three sets of tests were conducted for each test condition in this work, and we calculated the average of three test results. If the ratio of the average to the maximum and minimum was greater than 15%, the test value was considered invalid and the test was rerun, especially if one of the ratios was greater than 10%, and the other was less than 10%, whichever was smaller. The average of each value was used as the test result, and when the ratio was between 15% and 10%, the median of the three values was taken as the test result. If the ratio was less than 10%, the average of the three values was taken as the test result. To control for errors, three sets of tests were conducted for each test condition in this work, and we calculated the average of three test results. If the ratio of the average to the maximum and minimum was greater than 15%, the test value was considered invalid and the test was rerun, especially if one of the ratios was greater than 10%, and the other was less than 10%, whichever was smaller. The average of each value was used as the test result, and when the ratio was between 15% and 10%, the median of the three values was taken as the test result. If the ratio was less than 10%, the average of the three values was taken as the test result.
Before starting the test, we placed the test specimen in a 25 °C constant-temperature curing box for 12 h to control the test variables to reduce error. The specimen interface stress, Rm, was calculated by Equation (2): where is the stress at the interface (MPa), is the peak force at the interface (N), and is the interface area (cm ).

Digital Image Processing Technology
To evaluate the adhesion behavior of the asphalt-mortar on the fractal surface more accurately, we used digital image-processing technology to count the area ratios of the mortar and asphalt on the fracture surfaces.
Digital imaging uses numerical values to represent the surface of an object, and pixel points divide the object surface into discrete units, where the quantized gray numerical values represent the gray level. Digital image-processing consists of processing a series of values in the digital imager to obtain the desired result. Numerical image-processing technology can be used to perform image preprocessing, image segmentation, image expression, image pattern recognition, and feature extraction to analyze digital images. This analysis technique uses two main methods: spatial processing and frequency domain processing. In spatial processing, the operation is conducted on the pixels in the image, while in frequency domain processing, the numerical image is converted into a two-dimensional function to create other orthogonal coordinate bases for corresponding processing. In this study, we utilized airspace processing technology [28]. Before starting the test, we placed the test specimen in a 25 • C constant-temperature curing box for 12 h to control the test variables to reduce error. The specimen interface stress, Rm, was calculated by Equation (2): where R m is the stress at the interface (MPa), F p is the peak force at the interface (N), and A is the interface area (cm 2 ).

Digital Image Processing Technology
To evaluate the adhesion behavior of the asphalt-mortar on the fractal surface more accurately, we used digital image-processing technology to count the area ratios of the mortar and asphalt on the fracture surfaces.
Digital imaging uses numerical values to represent the surface of an object, and pixel points divide the object surface into discrete units, where the quantized gray numerical values represent the gray level. Digital image-processing consists of processing a series of values in the digital imager to obtain the desired result. Numerical image-processing technology can be used to perform image preprocessing, image segmentation, image expression, image pattern recognition, and feature extraction to analyze digital images. This analysis technique uses two main methods: spatial processing and frequency domain processing. In spatial processing, the operation is conducted on the pixels in the image, while in frequency domain processing, the numerical image is converted into a two-dimensional function to create other orthogonal coordinate bases for corresponding processing. In this study, we utilized airspace processing technology [28].

Pull-Off Test Results
In this work, 45 groups of pull-off tests were conducted, and the test results were as follows (Figure 7).

Pull-Off Test Results
In this work, 45 groups of pull-off tests were conducted, and the test results were as follows (Figure 7). When the water/powder ratio of the cement mortar was 0.23, three types of asphalt were used as the cement agents for curing times of 3 and 7 d. The order of interfacial tensile strength followed: PG76-22 modified asphalt > 70# petroleum asphalt > S-HV modified asphalt. At curing times of 21-28 d, the interfacial tensile strength of the specimens with 70# petroleum asphalt as the cementing agent decreased rapidly, and the strength was the lowest among the three asphalt types. This was attributed to the 70# petroleum asphalt, which was not as dense and durable as the other two asphalts. Moisture can also penetrate the asphalt faster, causing asphalt emulsification and interfacial bonding failure. When the S-HV modified asphalt was used as an interfacial binder, the tensile strength of the interface increased rapidly during the early stages of curing, and reached an average standard after 14 d, which was the peak tensile strength.
In summary, we found that during the early stages of curing, moisture ingress into the mortar accelerated the hydration of the mortar and enhanced its strength. As a result, the tensile strength of the interface increased with increased aging time and, during the later stages of curing, moisture ingress caused the asphalt to emulsify. In addition to the failure of the asphalt-granite interfacial bond, both conditions can cause a decrease in the tensile strength of the interface.

Image Processing and Analysis of Fracture Surface
Three materials with different colors appeared on the fracture surface, namely, asphalt, the interface, and the mortar, as shown in Figure 8, where the gray-black dots in the image are asphalt, the interface is dark brown, and the gray-white material is the mortar. When the water/powder ratio of the cement mortar was 0.23, three types of asphalt were used as the cement agents for curing times of 3 and 7 d. The order of interfacial tensile strength followed: PG76-22 modified asphalt > 70# petroleum asphalt > S-HV modified asphalt. At curing times of 21-28 d, the interfacial tensile strength of the specimens with 70# petroleum asphalt as the cementing agent decreased rapidly, and the strength was the lowest among the three asphalt types. This was attributed to the 70# petroleum asphalt, which was not as dense and durable as the other two asphalts. Moisture can also penetrate the asphalt faster, causing asphalt emulsification and interfacial bonding failure. When the S-HV modified asphalt was used as an interfacial binder, the tensile strength of the interface increased rapidly during the early stages of curing, and reached an average standard after 14 d, which was the peak tensile strength.
In summary, we found that during the early stages of curing, moisture ingress into the mortar accelerated the hydration of the mortar and enhanced its strength. As a result, the tensile strength of the interface increased with increased aging time and, during the later stages of curing, moisture ingress caused the asphalt to emulsify. In addition to the failure of the asphalt-granite interfacial bond, both conditions can cause a decrease in the tensile strength of the interface.

Image Processing and Analysis of Fracture Surface
Three materials with different colors appeared on the fracture surface, namely, asphalt, the interface, and the mortar, as shown in Figure 8, where the gray-black dots in the image are asphalt, the interface is dark brown, and the gray-white material is the mortar. We first conducted image preprocessing. Due to the noise in the spatial domain while acquiring the digital images, the images had distorted details, which caused large errors in image analysis. To reduce the effect of noise on the digital images, we preprocessed the We first conducted image preprocessing. Due to the noise in the spatial domain while acquiring the digital images, the images had distorted details, which caused large errors in image analysis. To reduce the effect of noise on the digital images, we preprocessed the digital images before conducting the analysis. In this work, we used denoising, image enhancement, and other imaging technologies to process the digital fracture surface image. Some studies have indicated that the use of wiener filtering and median filtering to denoise digital images may also provide a good denoising effect [29]. The image-processing results are shown in Figure 9. After image noise reduction, the images were blurred at the junction of the two substances, causing large errors in image segmentation. Therefore, after the denoising process, the images were enhanced, to remove irrelevant information and enhance regional contrast. We first conducted image preprocessing. Due to the noise in the spatial domain acquiring the digital images, the images had distorted details, which caused large e in image analysis. To reduce the effect of noise on the digital images, we preprocesse digital images before conducting the analysis. In this work, we used denoising, i enhancement, and other imaging technologies to process the digital fracture surfac age. Some studies have indicated that the use of wiener filtering and median filteri denoise digital images may also provide a good denoising effect [29]. The image cessing results are shown in Figure 9. After image noise reduction, the images blurred at the junction of the two substances, causing large errors in image segment Therefore, after the denoising process, the images were enhanced, to remove irrel information and enhance regional contrast. Next, we used image grayscale and threshold selection, using MATLAB 2018 ware to analyze and process the digital images to show the fracture surface. The se developed program, as well as the original software, were used to analyze and pr the pixels in the images. A grayscale image was regarded as a two-dimensional fun of f (x, y), where (x, y) represented the spatial plane coordinates, and the function v represented image brightness, which was termed "gray", to express the brightness o image. Each (x, y) coordinate point in the image corresponded to a gray value, and i information and gray values for all the coordinate points in the image were used to a digital image. This was denoted as a digitization of the coordinate values by sam and the digitization of the gray values was denoted by quantization.
The gray values for each coordinate point were divided into 266 small divi ranging from pure black to pure white, and were represented by values ranging from 255, where 0 indicated pure black and 255 indicated pure white. When taking a phot gray value displayed on the picture was different because of varying luminous flux, w Next, we used image grayscale and threshold selection, using MATLAB 2018 software to analyze and process the digital images to show the fracture surface. The seconddeveloped program, as well as the original software, were used to analyze and process the pixels in the images. A grayscale image was regarded as a two-dimensional function of f (x, y), where (x, y) represented the spatial plane coordinates, and the function value f represented image brightness, which was termed "gray", to express the brightness of each image. Each (x, y) coordinate point in the image corresponded to a gray value, and image information and gray values for all the coordinate points in the image were used to form a digital image. This was denoted as a digitization of the coordinate values by sampling, and the digitization of the gray values was denoted by quantization.
The gray values for each coordinate point were divided into 266 small divisions, ranging from pure black to pure white, and were represented by values ranging from 0 to 255, where 0 indicated pure black and 255 indicated pure white. When taking a photo, the gray value displayed on the picture was different because of varying luminous flux, which was reflected by the different materials. On this basis, we analyzed the gray values on the coordinate point to determine the composition and area ratio of the materials in the photos. Non-linear, linear, and piecewise linear transformations were the most-used gray-scale transformation methods.
This work utilized the three-stage piecewise linear transformation method for piecewise linear transformation to divide the asphalt, interface, and mortar on the fracture surface into different gray levels. We obtained 10 pictures of the mortar, interface, and asphalt sections in the different fracture surfaces, as shown in Figure 10. The statistical analysis results showed that the mortar on the fracture surface was gray-white, with a gray range of~160-255, and the interface was gray, with a gray range of 100-160, and the asphalt was gray-black, with a gray range of 0-100. This work utilized the three-stage piecewise linear transformation method for piecewise linear transformation to divide the asphalt, interface, and mortar on the fracture surface into different gray levels. We obtained 10 pictures of the mortar, interface, and asphalt sections in the different fracture surfaces, as shown in Figure 10. The statistical analysis results showed that the mortar on the fracture surface was gray-white, with a gray range of ~160-255, and the interface was gray, with a gray range of 100-160, and the asphalt was gray-black, with a gray range of 0-100.

Asphalt
Interface Mortar (a)

Asphalt
Interface Interface Mortar (c) This work also calculated the proportion of the asphalt and mortar area, showing that the three types of asphalt binder on the granite fracture surfaces varied with curing age, This work also calculated the proportion of the asphalt and mortar area, showing that the three types of asphalt binder on the granite fracture surfaces varied with curing age, as shown in Figure 11. After three days, the specimens with 70# petroleum asphalt and mortar fractured during the test, because of the insufficient tensile strength of the mortar.
As shown in Figure 11, for the specimens cured for three days, the mortar accounted for the largest area. Except when using PG76-22 as a binder, the asphalt area accounted for the smallest proportion. The mortar area proportion continued to decline with increasing curing age [30,31], possibly for the following reasons. As the curing age increased, the hydration reactions in the cement mortar continued, and the tensile strength of the cement mortar increased, resulting in a decrease in the potential mortar fracture performance of the specimen when subjected to tensile testing. As the curing age increased, the interactions between the asphalt and mortar increased, which increased the stress range of the mortar, resulting in cracks at the interface. Finally, with prolonged curing time, the continuous intrusion of moisture caused the asphalt to emulsify, which caused the failure in interfacial bonding and caused the asphalt to crack at the interface.
Coatings 2021, 11,1231 12 of 16 as shown in Figure 11. After three days, the specimens with 70# petroleum asphalt and mortar fractured during the test, because of the insufficient tensile strength of the mortar. As shown in Figure 11, for the specimens cured for three days, the mortar accounted for the largest area. Except when using PG76-22 as a binder, the asphalt area accounted for the smallest proportion. The mortar area proportion continued to decline with increasing curing age [30,31], possibly for the following reasons. As the curing age increased, the hydration reactions in the cement mortar continued, and the tensile strength of the cement mortar increased, resulting in a decrease in the potential mortar fracture performance of the specimen when subjected to tensile testing. As the curing age increased, the interactions between the asphalt and mortar increased, which increased the stress range of the mortar, resulting in cracks at the interface. Finally, with prolonged curing time, the continuous intrusion of moisture caused the asphalt to emulsify, which caused the failure in interfacial bonding and caused the asphalt to crack at the interface.
The asphalt area accounted for the overall upward trend with prolonged aging time. The area ratio of asphalt dropped suddenly after 14 d, when PG76-22 was used as the interfacial binder. Quartz was present on the granite surface, and the adhesion of quartz and asphalt is poor; thus, the asphalt peeled off when the quartz was pulled, and the proportion of asphalt area suddenly decreased. Then, owing to the emulsification of the asphalt, the cohesive forces of the asphalt itself decreased, causing the asphalt to eventually fracture.
As noted in Figure 12, when we used 70# petroleum asphalt, many small folds were observed in the asphalt on the fracture surface of the specimen in the early stages, and the materials started to pull apart. This indicated that when the test piece was cured for three days, the cohesive forces of the mortar and 70# petroleum asphalt were already greater than the cohesive forces of the 70# petroleum asphalt. Therefore, for the specimens containing 70# petroleum asphalt, and at curing times of 3-14 d, the mortar strength controlled the tensile strength of the specimen interface. With a prolonged curing time, the strength of the mortar increased, and the tensile strength of the interface increased. After 14 d of curing, the interfacial tensile strength was dominated by the cohesion of the asphalt. With an increased curing time, moisture continued to invade the asphalt, causing the asphalt to emulsify and reducing its cohesive properties, causing a decrease in the tensile strength of the specimen interface. Therefore, when using 70# asphalt as a binder for semi-flexible pavement construction, it should be not applied to areas which receive an excessive amount of rain. As a result, we recommend that road access be restricted for at least three days, before the semi-flexible pavement is made accessible to traffic. The asphalt area accounted for the overall upward trend with prolonged aging time. The area ratio of asphalt dropped suddenly after 14 d, when PG76-22 was used as the interfacial binder. Quartz was present on the granite surface, and the adhesion of quartz and asphalt is poor; thus, the asphalt peeled off when the quartz was pulled, and the proportion of asphalt area suddenly decreased. Then, owing to the emulsification of the asphalt, the cohesive forces of the asphalt itself decreased, causing the asphalt to eventually fracture.
As noted in Figure 12, when we used 70# petroleum asphalt, many small folds were observed in the asphalt on the fracture surface of the specimen in the early stages, and the materials started to pull apart. This indicated that when the test piece was cured for three days, the cohesive forces of the mortar and 70# petroleum asphalt were already greater than the cohesive forces of the 70# petroleum asphalt. Therefore, for the specimens containing 70# petroleum asphalt, and at curing times of 3-14 d, the mortar strength controlled the tensile strength of the specimen interface. With a prolonged curing time, the strength of the mortar increased, and the tensile strength of the interface increased. After 14 d of curing, the interfacial tensile strength was dominated by the cohesion of the asphalt. With an increased curing time, moisture continued to invade the asphalt, causing the asphalt to emulsify and reducing its cohesive properties, causing a decrease in the tensile strength of the specimen interface. Therefore, when using 70# asphalt as a binder for semi-flexible pavement construction, it should be not applied to areas which receive an excessive amount of rain. As a result, we recommend that road access be restricted for at least three days, before the semi-flexible pavement is made accessible to traffic. As shown in Figure 13, when we used PG76-22 modified asphalt, which was cured for three days, the asphalt surface on the fractured surface was still relatively smooth. This indicated that when the interface was cracked, the bonding and interactions of the asphalt and mortar were poor, and the interfacial tensile strength of the specimen was mainly dominated by the strength of the mortar. After curing for seven days, bonding failure of the asphalt and aggregate interface occurred, with wrinkles in the asphalt when the as- As shown in Figure 13, when we used PG76-22 modified asphalt, which was cured for three days, the asphalt surface on the fractured surface was still relatively smooth.
This indicated that when the interface was cracked, the bonding and interactions of the asphalt and mortar were poor, and the interfacial tensile strength of the specimen was mainly dominated by the strength of the mortar. After curing for seven days, bonding failure of the asphalt and aggregate interface occurred, with wrinkles in the asphalt when the asphalt broke. In addition, the tensile strength of the specimen interface was mainly dominated by the adhesion of the asphalt and the aggregate interface. After curing for 14 d, interfacial failure between the asphalt and aggregate became more noticeable. At this point, the tensile strength of the interface was mainly controlled by the adhesion capacity of the asphalt and aggregate. As the cohesive forces of asphalt and mortar and the cohesive forces of the asphalt itself are greater than the cohesive forces between the asphalt and the granite, the proportion of asphalt area was reduced when the interface fractured. After 21 d of curing, the bonding failure of the asphalt and aggregate interface was reduced and, after 28 d, the bonding failure between the asphalt and the aggregate interface completely disappeared. The wrinkles were caused by the self-cohesive forces of the asphalt surface, which appeared when the asphalt cracked. In short, starting at 21 d of curing, the tensile strength of the specimen interface was mainly controlled by the cohesive forces of the asphalt. This result was attributed to two reasons. First, with the continuous ingress of moisture into the asphalt, the cohesion of the asphalt emulsification decreased. Second, with increased aging time, the interactions between mortar and asphalt increased. Therefore, when we used PG76-22-modified asphalt as a semi-flexible pavement interfacial binder, water damage resistance was stronger compared to the 70# petroleum asphalt, but was greatly affected by the adhesion of aggregate and asphalt. Therefore, the adhesion capacity between the aggregate and asphalt should be carefully considered when building roads, and after road construction, we recommend a curing time of more than seven days.
As shown in Figure 13, when we used PG76-22 modified asphalt, which was cured for three days, the asphalt surface on the fractured surface was still relatively smooth. This indicated that when the interface was cracked, the bonding and interactions of the asphalt and mortar were poor, and the interfacial tensile strength of the specimen was mainly dominated by the strength of the mortar. After curing for seven days, bonding failure of the asphalt and aggregate interface occurred, with wrinkles in the asphalt when the asphalt broke. In addition, the tensile strength of the specimen interface was mainly dominated by the adhesion of the asphalt and the aggregate interface. After curing for 14 d, interfacial failure between the asphalt and aggregate became more noticeable. At this point, the tensile strength of the interface was mainly controlled by the adhesion capacity of the asphalt and aggregate. As the cohesive forces of asphalt and mortar and the cohesive forces of the asphalt itself are greater than the cohesive forces between the asphalt and the granite, the proportion of asphalt area was reduced when the interface fractured. After 21 d of curing, the bonding failure of the asphalt and aggregate interface was reduced and, after 28 d, the bonding failure between the asphalt and the aggregate interface completely disappeared. The wrinkles were caused by the self-cohesive forces of the asphalt surface, which appeared when the asphalt cracked. In short, starting at 21 d of curing, the tensile strength of the specimen interface was mainly controlled by the cohesive forces of the asphalt. This result was attributed to two reasons. First, with the continuous ingress of moisture into the asphalt, the cohesion of the asphalt emulsification decreased. Second, with increased aging time, the interactions between mortar and asphalt increased. Therefore, when we used PG76-22-modified asphalt as a semi-flexible pavement interfacial binder, water damage resistance was stronger compared to the 70# petroleum asphalt, but was greatly affected by the adhesion of aggregate and asphalt. Therefore, the adhesion capacity between the aggregate and asphalt should be carefully considered when building roads, and after road construction, we recommend a curing time of more than seven days. As shown in Figure 14, when the S-HV modified asphalt was used as a binder, the asphalt to mortar ratio curve was similar to curing age over time. Before seven days of curing, the interfacial tensile strength was mainly controlled by the strength of the mortar and the bonding forces between the asphalt and the mortar. After curing for 14 d, the asphalt portion in the fracture surface started to wrinkle, indicating that the interactions As shown in Figure 14, when the S-HV modified asphalt was used as a binder, the asphalt to mortar ratio curve was similar to curing age over time. Before seven days of curing, the interfacial tensile strength was mainly controlled by the strength of the mortar and the bonding forces between the asphalt and the mortar. After curing for 14 d, the asphalt portion in the fracture surface started to wrinkle, indicating that the interactions and adhesion between asphalt and mortar weakened. The wrinkles produced during the time were larger, but did not cover all the asphalt, indicating that the adhesion between the asphalt and the interface was less than the cohesive forces of the asphalt. Between 7 and 14 d of curing, the interfacial tensile strength was mainly dominated by the adhesion between the asphalt and the mortar, due to insufficient interfacial cohesion between the asphalt and the mortar. Starting at 14 d of curing, the interfacial tensile strength was mainly controlled by the cohesion between the asphalt. This was due to the increased interactions between the asphalt and the mortar and the ingress of water into the asphalt, which caused the asphalt to emulsify and reduced the cohesion of the asphalt. Therefore, when S-HV modified asphalt was used as a semi-flexible pavement interfacial binder, its ability to resist water damage was better than the 70# petroleum and PG76-22-modified asphalt. However, if the road is to be opened to traffic afterward, we recommend a curing time of longer than 14 d. mainly controlled by the cohesion between the asphalt. This was due to the increased interactions between the asphalt and the mortar and the ingress of water into the asphalt, which caused the asphalt to emulsify and reduced the cohesion of the asphalt. Therefore, when S-HV modified asphalt was used as a semi-flexible pavement interfacial binder, its ability to resist water damage was better than the 70# petroleum and PG76-22-modified asphalt. However, if the road is to be opened to traffic afterward, we recommend a curing time of longer than 14 d. In summary, the curing times required for the three asphalt and mortar types to achieve better interactions and adhesion in the SFP were as follows. Before three days of curing, 70# petroleum asphalt and mortar achieved better interactions and adhesion. Between three and seven days of curing, PG76-22-modified asphalt and mortar achieved better interactions and adhesion. However, between 14 and 21 d of curing, the S-HV-modified asphalt and mortar exhibited better interactions and adhesion. Therefore, the order of interfacial water damage resistance of the specimens with different asphalt materials as binder was as follows: S-HV-modified asphalt > PG76-22-modified asphalt > 70# petroleum asphalt.

Conclusions
In this work, different types of asphalt were used as interfacial binders according to pull-off, tensile strength measurements, and fracture surface image analysis, and the following conclusions were drawn.
When using PG76-22-modified asphalt as the interface binder, the maximum tensile strength of the interface was larger than specimens with 70# asphalt as the binder. However, the interfacial tensile strength was lower with less curing time. When the S-HV modified asphalt was used as an interfacial binder, more curing time was required for the specimen interface to reach peak tensile strength. The order of peak tensile strength values for the interfacial materials was as follows: PG76-22-modified asphalt >70# petroleum asphalt >S-HV-modified asphalt.
On the fracture surface, the mortar was gray-white, the gray-scale range was 160-255, the interface was gray, with a gray-scale range of 100-160, and the asphalt was grayblack, with a gray-scale range of 0-100.
The curing times required for the three asphalts and mortar types to achieve improved interactions and adhesion were as follows: the 70# petroleum asphalt required 3 d of curing, PG76-22-modified asphalt required 7 d of curing, and S-HV-modified asphalt In summary, the curing times required for the three asphalt and mortar types to achieve better interactions and adhesion in the SFP were as follows. Before three days of curing, 70# petroleum asphalt and mortar achieved better interactions and adhesion. Between three and seven days of curing, PG76-22-modified asphalt and mortar achieved better interactions and adhesion. However, between 14 and 21 d of curing, the S-HV-modified asphalt and mortar exhibited better interactions and adhesion. Therefore, the order of interfacial water damage resistance of the specimens with different asphalt materials as binder was as follows: S-HV-modified asphalt > PG76-22-modified asphalt > 70# petroleum asphalt.

Conclusions
In this work, different types of asphalt were used as interfacial binders according to pull-off, tensile strength measurements, and fracture surface image analysis, and the following conclusions were drawn.
When using PG76-22-modified asphalt as the interface binder, the maximum tensile strength of the interface was larger than specimens with 70# asphalt as the binder. However, the interfacial tensile strength was lower with less curing time. When the S-HV modified asphalt was used as an interfacial binder, more curing time was required for the specimen interface to reach peak tensile strength. The order of peak tensile strength values for the interfacial materials was as follows: PG76-22-modified asphalt >70# petroleum asphalt >S-HV-modified asphalt.
On the fracture surface, the mortar was gray-white, the gray-scale range was 160-255, the interface was gray, with a gray-scale range of 100-160, and the asphalt was gray-black, with a gray-scale range of 0-100.
The curing times required for the three asphalts and mortar types to achieve improved interactions and adhesion were as follows: the 70# petroleum asphalt required 3 d of curing, PG76-22-modified asphalt required 7 d of curing, and S-HV-modified asphalt required 21 d of curing. In addition, the order of interfacial specimen water damage resistance, with different asphalt types was as follows: S-HV-modified asphalt >PG76-22-modified asphalt >70# petroleum asphalt.