Estimating moisture susceptibility of asphalt modified with alumina trihydrate

The moisture susceptibility of asphalt containing alumina trihydrate (ATH) was assessed through the surface free energy. Physical properties and flame retardancy of ATH samples were also investigated. For research purposes SBS modified asphalt samples with different dosages (0-14 %) of ATH were prepared. Increased viscosity, softening point limiting oxygen index and lower penetration ductility, were investigated for the ATH modified samples. ATH has a significant negative effect on the moisture-induced damage potential of asphalt mixture from the view of micromechanisms.


Introduction
Many road tunnels have been built in mountainous areas in the scope of rapid development of highway construction incentives in China. Thus, over 15,000 highway tunnels, with a total length of 14,039.7 km, have been built in China by the end of 2016 [1]. Out of this total tunnel length, long tunnels (L ≥ 1000 m) account for as many as 740 km. Although tunnel construction brings convenience to transport sector, a problem that cannot be ignored is the semi-closed space, which is highly detrimental with regard to rescue efforts in case of fire [2]. In fact, many high-profile road tunnel fires have occurred in tunnels over the past decades, and the consequences of such fires have been quite serious [3]. To counter this problem, flame-retardant materials should be used in tunnels. Asphalt pavements have been used widely in road and tunnel construction owing to their superior performance and long service life. However, such pavements can easily be ignited in case of traffic accidents involving outbreak of fire. In such cases, asphalt can catch on fire and, while burning, it releases large amounts of smoke and toxic gases, impeding the traffic and hampering rescue operations [4]. A growing attention is currently paid to possible improvements of flame retardancy of asphalt pavements in tunnels, and adding flame retardants into asphalt has proven to be a successful popular measure to inhibit asphalt combustion and/or pyrolysis [5]. In recent years, many flame retardants such as magnesium hydroxide, decabromodiphenyl oxide, decabromodiphenyl ethane, alumina trihydrate, antimonous oxide, ammonium polyphosphate, and pentaerythritol, have been used as additives to asphalt [6]. Due to its cost effectiveness, safe handling, and environmental friendliness, alumina trihydrate (ATH) has become a popular flame retardant for reducing or preventing asphalt combustion. The issues such as flame retardant properties, ageing characteristics of flame retardant asphalts, asphalt mixture design methods, and the effect of flame retardant on pavement performance, are currently studied, and the thermal degradation behaviour and modification mechanism of flame retardant (FR) asphalt materials are investigated by thermogravimetric and infrared analyses [7,8]. However, some researchers have placed emphasis on the moisture-induced damage potential of asphalts containing FRs, especially from the point of view of micro-mechanisms. Actually, this is very necessary because some tunnel environments are very humid. At present, many theories and methods have been developed to estimate and explain damage that could result from moisture. Surface free energy (SFE) has proven to be an effective and successful method for evaluating moisture susceptibility of asphalt mixtures, as it can characterize the cohesive strength and adhesion strength of asphalt and asphalt-aggregate from the point of view of micro-mechanisms [9,10]. In general, the surface free energy of a material could be considered as a capacity needed to create a unit area of new surface of such material. The SFE has been successfully used to evaluate moisture susceptibility of asphalt materials. The objective of this study is to examine the moisture-induced damage potential of the ATH modified asphalt using the SFE theory. The penetration, ductility and softening point, rutting resistance performance, and flame retardancy of the ATH modified asphalt, are investigated. The SFE components of asphalt are determined by static contact angle measurement. The cohesion energy, debonding work, adhesion work, and energy ratio, are calculated to explore the moisture susceptibility of the ATH modified asphalt.

Asphalt and flame retardant
SBS (I-C) modified asphalt, widely used in China, was chosen as the matrix asphalt in the test. Its basic physical properties are presented in Table 1. Aluminium trihydrate (ATH) consisting of micro AH-2 was used as the flame retardant in this study. Its basic technical index is presented in Table 2.

Aggregate and probe liquids
Three types of aggregate provided from Jiansu Province, China, were used in this study. Their basic physical properties and surface free energy components are listed in Table 3. The formamide, distilled water, and glycerol, were selected as probe liquids to measure the contact angle with the tested binders. The SFE parameters of the above liquids are listed in Table 4.

Preparation of ATH modified bitumen
The ATH modified bitumens were prepared using a high shear mixer at a rotational speed of 4,000 rpm at 165 °C. First, the matrix binder was heated in an iron container to make it fluid. Then the 6-14 wt % of ATH was added into the binder and sheared for 30 min to produce the ATH modified asphalt. Finally, the asphalt samples were used for the follow-up tests.

Measuring performance of ATH modified asphalts
Physical and mechanical properties of the ATH-modified asphalts were determined using a series of tests. The penetration at 25 °C, ductility at 5 °C, softening point, and viscosity at 135 °C of the asphalt samples, were determined on the basis of the Chinese standard JTG E20-2011 [11]. The antirutting performance of ATH modified asphalt was evaluated by the rutting factor (G*/sind), measured using DSR at the 10 rad/s frequency in the temperature range from 58 to 82 °C according to ASTM D 7175 [12].

Measuring flame retardancy of ATH modified asphalts
The flame retardancy of the ATH modified asphalt was characterized by the limiting oxygen index (LOI), which is the minimal oxygen concentration allowing asphalt to burn as per ASTM D-2863 [13]. The sample size of the ATH modified asphalt was 110 × 6.5 × 3 mm 3 .

Measuring contact angles
Surface energies of asphalt were measured by contact angle, which is the simplest and the most common approach for measuring the surface energy. The Drop Shape Analyser 10 was used for measuring contact angles of the ATH modified asphalt. The distilled water, glycerol, and formamide liquids were used in this analysis.

SFE parameters and fundamentals
Based on earlier literature surveys [14][15][16], the moisture susceptibility of the ATH modified asphalt was evaluated by the SFE components, as well as the cohesion energy of asphalt, the work of adhesion (under dry conditions), work of debonding (under wet conditions), wettability, and energy ratio. The relationships between the contact angle (q), SFE of liquid or solid (g l or g s ), and interaction energy of liquid-solid (g ls ) was determined using the Young equation expressed by Eq. (1) [17]: where g is the surface free energy of the material, while l and s denote the liquid and solid, respectively. g ls is the interaction energy between liquid and solid materials. It can be calculated from Eq. (2) based on the Owens and Wendt method [18]: where g LW is the Lifshitz-van der Waals component and g AB is the acid base component. It should be noted that SFE is composed of two components: Lifshitz-van der Waals (nonpolar component) and acid-base component [19]. The total surface free energy was obtained by combining these components as follows, when the material is liquid and solid. (3) According to the research of Akvarez et al. [20], the wettability can be represented by the Spreading Coefficient as shown by Eq. (9): Howson et al. [21], and Little et al. [22] combined cohesive and adhesive energies into a single term as shown by Eq. (10).

Physical properties
The results of penetration, ductility, softening point, and viscosity of ATH modified asphalt are shown in Figure 1. It can be seen in that figure that the penetration 25 °C and ductility 5 °C decrease with an increase in ATH dosage, while the softening point and viscosity 135 °C significantly improve in comparison with the neat (base) asphalt. For instance, when the 6 % ATH was mixed into asphalt, the penetration and ductility was reduced by 6 % and 8 %, respectively. This result indicates that ATH additives increase the stiffness of asphalt. Furthermore, the softening point and viscosity increased by 11.18 % and 36.51 %, respectively, for asphalt samples with 6 % of ATH. This also proves that stiffness increases with the addition The above analysis was used to establish the relationship between the contact angle and SFE components as shown by Eq. (5): Furthermore, the cohesive energy (W c ll ) providing the energy of materials to attract each other due to the nature of mutual attraction was calculated from Eq. (6): The work of adhesion is considered as the energy for asphalt stripping from the aggregate, and this could characterize the bond strength of the asphaltaggregate system. The work of adhesion (W a dry ) under dry conditions, and the work of debonding (W wet ) under wet conditions, can be calculated by means of Eqs. (7) and (8), respectively: GRAĐEVINAR 70 (2018) 10, 863-871 Estimating moisture susceptibility of asphalt modified with alumina trihydrate of ATH to asphalt. As a result, the ATH modified asphalt is more stable at high temperatures and may also be more resistant to permanent deformation (rutting) compared to matrix asphalt.

Rutting resistance
In order to better characterize the hightemperature rutting resistance, the G*/ sind of asphalt with or without ATH was analysed, as shown in Figure 2. It can be seen that the G*/sind increases when ATH is added to the matrix asphalt at different temperatures. For instance, the G*/sind of asphalt at 58°C, with 6 % ATH dosage, increased by 19.15 %. In general, the greater the G*/sind, the better the rutting resistance performance of asphalt. The results indicate that the high-temperature performance of the ATH modified asphalt improves with an increase in ATH content. Furthermore, the change of amplitude of asphalt at different temperatures reveals a great difference when the ATH content is increased. For matrix asphalt, the G*/sind decreased by 41.4 % when the temperature increased from 58 °C to 9 °C, while for the modified asphalt containing 6 % ATH, the G*/sind decreased by 43 % when the temperature increased from 58 °C to 64 °C. The difference in the changed G*/sind also shows that the hightemperature performance of asphalt alters with the addition of ATH.

Flame retardancy
The limiting oxygen index (LOI) was selected as the indicator to evaluate the flame-retardant capability of the ATH modified asphalt. Asphalt with different dosages of ATH was investigated to illustrate the effect of ATH on flame retardancy in the similar test environment. Figure 3 presents the LOI test results of asphalt samples. The results illustrate that the LOI of the ATH modified asphalt was greater than that of the matrix asphalt. Meanwhile, the LOI shows an increasing trend with an increase in ATH dosage. It was identified that the addition of ATH enhanced the flame retardant performance of the modified asphalt. In addition, as shown in Figure 4, the incremental of LOI was similar when the LOI increased in the same range. To be specific, the LOI increased by 2.93 %, 3.41 %, 3.11 %, and 4.27 %, when ATH dosage increased from 6 % to 8 %, 8 % to 10 %, 10 % to 12 %, and 12 % to 14 %, respectively.

Contact angle, SFE, and cohesive energy
Measured contact angles of the unmodified and ATH-modified bitumen with formamide, distilled water, and glycerol under different ATH dosages are presented in Figure 4.a. The van der Waals component of (g LW ) of SFE and Lewis acid-base component (g AB ) of the surface free energy of each asphalt were calculated using the contact angle based on Eq. (5). The corresponding results are shown in Figure 4.b. The cohesive energy results were calculated based on Eq. (6). The corresponding results are presented in Figure 4.c. Figure 4.a shows that the contact angle of asphalt increases first and then decreases at 6 % ATH. In the case of water, for 6 % ATH, the contact angle of asphalt increased from 91.28° to 93.47°. A contrary decreasing trend was observed when the ATH dosage increased from 6 % to 14 %, and the contact angle of asphalt decreased to 91.42°. The contact angle could directly indicate the wetting ability of the probe liquid with a solid. Generally, if the contact angle is close to zero, the solvent spreads completely on the surface; conversely, the contact angle > 90° indicates that the solvent is not wetting, or is poorly wetting the surface; and if it is in the range from 0° to 90°, it nicely wets the surface of the material. The contact angles were > 90° when the water was taken as the probe liquid for all ATH dosages, indicating that the water cannot wet them well. Furthermore, the use of higher amounts of ATH was found to possibly reduce the contact angle. In particular, the asphalt with 14 % ATH has the contact angle similar to that of the unmodified asphalt. Figure 4.b shows that the van der Waals component (g LW ) of asphalt increases first and then decreases at the turning point of 10 % ATH. A reverse trend was observed, showing the decrease of the acid-base component (g AB ), followed by an increase in the ATH content up to 10 %. However, the trend of total SFE was found to be similar to g LW . This indicates that the van der Waals component is the most significant contributing factor to the total SFE, and that the total SFE slightly improves with an increase in ATH content. The van der Waals component (g LW ), which corresponds to nonpolar molecules, is usually considered as solvent for polar molecules in asphalt. Therefore, an increase in the total SFE of asphalt could be deemed to provide a higher potential for bonding with aggregates in the presence of moisture. In particular, the acid-base component (g AB ), or polar molecules within the asphalt binder, play a vital role in the adhesion to aggregate surface, because of high polarity of aggregate surface. Therefore, if the content of polar molecules in asphalt is lower, water can easily displace the asphalt binder from the aggregate surface in the absence of a strong bond, as the polarity of water is higher than that of the asphalt binder. In this study, the acid-base component of asphalt decreased first and then increased at 10 % ATH. This may be explained based on the fact that the ATH has both acidic and alkaline characteristics. When ATH (< 10 %) is added into the asphalt (almost acidic), it may lead to neutral reaction, reducing the acid-base component. Furthermore, the acidity of ATH increases with an increase in ATH dosage, causing the acid-base component to increase. Therefore, the change trend of the adhesion of asphaltaggregate system may be similar to the acid base component. However, the potential moisture damage to asphalt should still be analysed in depth as the moisture damage causes the loss of the strength and stiffness at the binder aggregate interface. Estimated cohesive energies of the tested binder samples are shown in Figure 4.c. As the cohesive energy is twice of the total SFE, it can be found that the cohesive energy exhibits an increase similar to the SFE. Furthermore, the cohesive energy of asphalt at all ATH dosages is higher than that of the unmodified asphalt, although it increases first and then decreases at the turning point of 10 % ATH. Because a higher cohesive energy will lead to greater external energy for crack propagation, it suggests higher resistance to the moisture damage for the ATH modified binder.

Work of adhesion and work of debonding
The indexes of work of adhesion and work of debonding for the asphalt samples are shown in Figure 5. There it can be seen that the work of adhesion decreases at 6 % to 10 % of ATH content, while it increases at higher ATH contents (12 % and 14 %) compared to the control asphalt, as shown in Figure 5.a. This is similar to the changing trend of acid components. Especially, all the studies of adhesion of the ATH modified binders are lower compared to control asphalt. These results illustrate that the addition of ATH reduces both indexes for the three aggregates. The decrease of adhesion will cause an exponential increase in fracture energy [23]. Therefore, this finding points to a higher moisture damage resistance of the ATH modified asphalt mixture. Furthermore, the limestone-asphalt system has been shown to have the highest adhesion energy, while the basaltasphalt system has the lowest adhesion energy at different ATH contents. Thus, it is much harder to induce adhesive fracture at the limestone-asphalt interface under the same wheel load and environmental stresses. Figure 5.b also shows that the work of adhesion was slightly reduced by the addition of ATH. This reduction increased for the granite aggregate compared to other aggregates. For example, when ATH dosage increased from 0 % to 10 % for the asphaltlimestone, asphalt-basalt and asphalt-granite systems, the   Estimating moisture susceptibility of asphalt modified with alumina trihydrate The energy ratio (ER) calculated based on Eq. (10) is presented in Figure 6.b. This figure shows that the ER decreases at the 6 % to 10 % ATH content, while it increases at higher ATH contents (12 % and 14 %). The ER values for the ATH modified asphalt are lower than those for the matrix asphalt, although they increase at ATH dosage >10 %. A higher ER value represents better moisture resistance for the asphalt-aggregate system. This result shows that ATH has a negative effect on the resistance of the moisture-induced damage potential. Furthermore, the ER values for granite aggregate combinations are lower than those for the limestone aggregate and basalt aggregate combinations, while the ER values for the limestone aggregate are the highest. This indicates that the resistance to moisture damage is the lowest for granite combinations, and the one for limestone combinations is the highest. The sensitivity was also measured to analyse if ATH and aggregate have significant effects on the SC and ER. The independent variables were also designed as factor A: ATH content (0 %, 6 %, 8 %, 10 %, 12 %, and 14 %), and factor B: aggregate type (limestone, basalt and granite), while dependent variables were designed as SC and ER. The sensitivity results are listed in

Spreading coefficients and energy ratio
The spreadability coefficient (SC) of asphalt samples over aggregate surface is calculated using Eq. (9). The results calculated and determined for bitumen blends and aggregates are presented in Figure 6.a. It can be seen in that figure that the spreadability of matrix asphalt reduced after addition of ATH, although it slightly improved at the ATH dosage >10 %. In general, a higher SC value indicates better wettability of asphalt with aggregate. If the wettability is higher, the aggregate surface can easily be coated by asphalt. This result indicates that ATH negatively affects the aggregate coating with asphalt, especially as physical and chemical properties of asphalt and aggregates are originally different. Furthermore, as to various asphalt-aggregate systems, the limestone clearly showed the highest SC, while the granite exhibited the lowest SC. Therefore, the asphalt-limestone system has the best wettability and moisture resistance damage potential.

Summary and conclusions
In this study, SFE components, work of cohesion of the matrix asphalt and ATH modified asphalt, the work of adhesion, work of debonding, wettability, and energy ratio of various asphalt and aggregate systems, were investigated by means of the SFE approach. The static contact angles were measured to calculate the SFE parameter. Physical properties and flame retardancy of ATH binder samples were also measured to study their performance. The experimental results of this study leads to the following conclusions: -Increased viscosity, softening point, G*/sind, LOI, and lower penetration and ductility, were observed because of addition of the ATH modifier. -The addition of ATH increased the contact angle, van der Waals component, total SFE, and cohesion energy, but decreased the acid base component for all binder samples.
-The work of adhesion of all asphalt aggregates increased due to addition of ATH additive, but the work of debonding decreased. All the adhesion and debonding values of ATH binders were lower than those of the matrix asphalt for various aggregates.
-The addition of ATH decreased the spreading coefficients and energy ratio, which also shows that the ATH reduced moisture resistance of asphalt mixes.
-The limestone-asphalt system has the highest adhesion work and the lowest adhesion work among all the tested aggregatesasphalt systems. Therefore, the asphalt-limestone system has the best moisture resistance damage potential.
-10 % ATH dosage is a turning point for the cohesion energy, work of adhesion, work of debonding, wettability, and energy ratio. At this dosage, the moisture resistance damage potential of asphalt and asphalt-aggregates system reached the lowest value.