Moisture Sensitivity of Asphalt Pavements (2003)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

37 TOPIC 2 Chemical and Mechanical Processes of Moisture Damage in Hot-Mix Asphalt Pavements DALLAS N. LITTLE Texas A&M University DAVID R. JONES IV Owens Corning Historically, six contributing mechanisms to moisture damage have been identified: detachment, displacement, spontaneous emulsification, pore pressure–induced damage, hydraulic scour, and the effects of the environment on the aggregate–asphalt system. However, it is apparent that moisture damage is usually not limited to one mechanism but is the result of a combination of processes. It has become necessary to seek a more fundamental understanding of the moisture damage process by carefully considering the micromechanisms that influence the adhesive interface between aggregate and asphalt and the cohesive strength and durability of the mastic. Factors that influence the adhesive bonds in asphalt mixtures and the cohesive strength of the mastic in the presence of water are presented and discussed as a fundamental approach to calculating adhesive bond strength in asphalt mixtures in the presence of water on the basis of surface free energy measurements. The adhesive bond that determines the durability of asphalt mixtures in the presence of water is described in this paper to be based on a nonuniform distribution of charges in the asphalt and on the aggregate surface. The polar compounds in the asphalt that react with the aggregate polar sites determine the strength and durability of the adhesive bond. Several processes are presented that affect this bond. The effect of aggregate mineralogy, surface properties, and the pH at the water–aggregate interface is discussed. BACKGROUND Description of Moisture Damage Moisture damage can be defined as the loss of strength and durability in asphalt mixtures due to the effects of moisture. Moisture damage can occur because of a loss of bond between the asphalt cement or the mastic (asphalt cement plus the mineral filler—74 µm and smaller aggregate) and the fine and coarse aggregate. Moisture damage also occurs because moisture permeates and weakens the mastic, making it more susceptible to moisture during cyclic loading. The literature (Taylor and Khosla 1983; Kiggundu and Roberts 1988; Terrel and Al- Swailmi 1994) refers to at least five different mechanics of stripping: detachment, displacement, spontaneous emulsification, pore pressure, and hydraulic scour. Kiggundu and Roberts (1988) suggest additional mechanisms that may well play a part in moisture damage. These include pH instability and the effects of the environment or climate on asphalt–aggregate material systems.

38 Moisture Sensitivity of Asphalt Pavements: A National Seminar Detachment Detachment is the separation of an asphalt film from an aggregate surface by a thin film of water without an obvious break in the film (Majidzadeh and Brovold 1968). Theories that explain adhesive bond energy provide the rationale to understand detachment. Several factors are involved in detachment. First of all, it is necessary to develop a good bond between the asphalt and the aggregate. Such a bond is initially dependent on the ability of the asphalt to wet the aggregate. Wettability of aggregate increases as surface tension or free surface energy of adhesion decreases (Majidzadeh and Brovold 1968). According to Majidzadeh and Brovold (1968), if a three-phase interface consisting of aggregate, asphalt, and water exists, water reduces the free energy of the system more than asphalt to form a thermodynamically stable condition of minimum surface energy. Surface energy measurements at Texas A&M University have established that when the free energy at the asphalt–aggregate interface is calculated in the presence of water, energy is released, meaning that the aggregate surface has a strong preference for water over asphalt. The negative values of free energy in Column 5 of Table 1 establish that this is true for each of four asphalt–aggregate combinations listed in Table 1. The more negative the value, the stronger is the preference for detachment of asphalt from aggregate in the presence of water. The thermodynamic basis for these calculations is presented by Cheng et al. (2002). Work at the Road Research Laboratory in 1962 suggests that most asphalts have relatively low polar activity and that the bond that develops between the aggregate and asphalt is chiefly due to relatively weak dispersion forces. Water molecules are, on the other hand, highly polar and can replace the asphalt at the asphalt–aggregate interface. Recent work at Texas A&M University by Cheng et al. (2002) has established this to be the case and will be discussed later. Texas A&M researchers have developed a methodology to measure component surface energies and to calculate adhesive bond strengths from these measurements (Cheng et al. 2002). Displacement Displacement differs from detachment because it involves displacement of asphalt at the aggregate surface through a break in the asphalt film (Tarrer and Wagh 1991; Fromm 1974). The source of the break or disruption may be incomplete coating of the aggregate surface, film rupture at sharp aggregate corners or edges, pinholes originating in the asphalt film because of aggregate coatings, and so forth. Scott (1978) states that chemical reaction theory can be used to TABLE 1 Comparison of Free Energy of Adhesion (ergs/gm) and Rate of Damage Under Repeated Load Triaxial Testing (After Cheng et al. 2002) Mix Cycles to Accelerated Damage Loss of Contact Area (Debonding), % Free Energy of Adhesion (Dry) Free Energy of Adhesion (Wet) AAD + Texas Limestone 275 34 141 -67 AAM + Texas Limestone 550 27 205 -31 AAD + Georgia Granite 250 35 150 -48 AAM + Georgia Granite 455 26 199 -30

Little and Jones 39 explain stripping as a detachment mechanism. Some authors describe film rupture as a separate mechanism of moisture damage, but it can be incorporated as part of the displacement mechanism. The process of displacement can proceed through changes in the pH of the water at the aggregate surface that enters through the point of disruption. These changes alter the type of polar groups adsorbed, leading to the buildup of opposing, negatively charged, electrical double layers on the aggregate and asphalt surfaces. The drive to reach equilibrium attracts more water and leads to physical separation of the asphalt from the aggregate (Scott 1978; Tarrer and Wagh 1991). Spontaneous Emulsification Spontaneous emulsification is an inverted emulsion of water droplets in asphalt cement. Fromm (1974) demonstrated how an emulsion forms and that once the emulsion formation penetrates the substrata, the adhesive bond is broken. Some research indicates that the formation of such emulsions is further aggravated by the presence of emulsifiers such as clays and asphalt additives (Asphalt Institute 1981; Fromm 1974; Scott 1978). Fromm (1974) observed that spontaneous emulsification occurs when asphalt films are immersed in water and that the rate of emulsification depends on the nature of the asphalt and the presence of additives. However, commercial amine-based asphalt additives, which are organic amine compounds, are chemically different from cationic asphalt emulsifiers, and they cannot function as an emulsifier in their amine form to make normal oil in water–asphalt emulsions. The cationic emulsifier solutions are obtained by reacting amines such as fatty amines with dilute hydrochloric or acetic acid to produce an amine salt (Morgan and Mulder 1995). Furthermore, organic amines, which are basic nitrogen compounds, bond strongly to aggregates in the presence of water (Robertson 2000). Kiggundu (1986) demonstrated how the rate of emulsification is dependent on the nature and viscosity of asphalt, with an AC-5 emulsifying in distilled water much faster than an AC-10. He also demonstrated that the process is reversible upon drying. Pore Pressure Pore pressure development in water that is entrapped can lead to distress. Stresses imparted to the entrapped water from repeated traffic load applications will worsen the damage as the continued buildup in pore pressure disrupts the asphalt film from the aggregate surface or can cause the growth of microcracks in the asphalt mastic. Bhairampally et al. (2000) used a tertiary damage model developed by Tseng and Lytton (1987) to demonstrate that well-designed asphalt mixtures tend to “strain harden” on repeated loading. This “strain hardening” is of course not classical strain hardening that occurs when metals are cold-worked to develop interactive dislocations to prevent slip but is the “locking” of the aggregate matrix caused by densification during repeated loading. On the other hand, some mixtures exhibit microcracking in the mastic under heavy repeated loading. This results in progressive cohesive or adhesive failure, or both, and is evident in a plot of accumulated permanent strain versus number of load cycles as the rate of damage dramatically increases as the microcracking progresses. The rate of this accelerated or tertiary damage is exacerbated in the presence of water as the pore pressure developed in the microcrack voids increases the rate of crack growth and damage through the development of higher pressures at the crack tip and through a weakening of the mastic and of the adhesive bond between the mastic and the aggregate. Terrel and Al-Swailmi (1994) described the concept of pessimum air voids, which is the range of air void contents within which most asphalt mixtures are typically compacted (between

40 Moisture Sensitivity of Asphalt Pavements: A National Seminar about 8% and 10%). Above this level the air voids become interconnected and moisture can flow out under a stress gradient developed by traffic loading. Below this value the air voids are disconnected and are relatively impermeable and thus do not become saturated with water. In the pessimum range, water can enter the voids but cannot escape freely and is, thus, subjected to pore pressure buildup upon repeated loading. Hydraulic Scour Hydraulic scour occurs at the pavement surface. Here stripping results from the action of tires on a saturated surface. Water is sucked under the tire into the pavement by the tire action. Osmosis and pullback have been suggested as possible mechanisms of scour (Fromm 1974). Osmosis occurs in the presence of salts or salt solutions in aggregate pores and creates an osmotic pressure gradient that actually sucks water through the asphalt film. Researchers are mixed on support of this process. Mack (1964) supports it, while Thelen (1958) feels it is too slow to be valid. However, several factors support the potential occurrence of this mechanism, including the fact that some asphalts are treated with caustics during manufacture, some aggregates possess salts (compositionally), and asphalt films are permeable. In fact, Cheng et al. (2002) have demonstrated that the diffusion of water vapor through asphalt cement itself is considerable and that asphalt mastics can hold a rather surprisingly large amount of water. Table 2 compares the water vapor diffusion rates and the amount of water that can be accommodated by two compositionally very different asphalts (AAD-1 and AAM-1). They also showed that the amount of water held by these asphalts is related to the level of moisture damage that occurs in the mixtures using these asphalts. pH Instability Scott (1978) and Yoon (1987) demonstrated that asphalt–aggregate adhesion is strongly influenced by the pH of the contact water. Kennedy et al. (1984) investigated the effect of various sources of water on the level of damage that occurred in a boiling test. Fehsendfeld and Kriech (undated) observed that the pH of contact water affects the value of the contact angle and the wetting characteristics at the aggregate–asphalt interface region. Scott (1978) observed that the value of interfacial tension between asphalt and glass peaked at intermediate pH values, up to about 9, and then dropped as pH increased. Yoon (1987) found that the pH of contact water increased with duration of contact and was aggregate-specific and that the values stabilized after about 5 to 10 minutes of boiling. Yoon determined that the coating retention in boiling tests TABLE 2 Effect of Moisture-Holding Potential of Asphalt on Moisture Damage in Triaxial Testing (After Cheng et al. 2002) Binder Parameter AAD-1 AAM-1 Ratio, AAD-1/AAM-1 Diffusivity, m2/s 0.0008 0.0029 Water-holding potential, W100, parts per 100,000 153 114 1.34 Percent debonding of binder from aggregate 34 (AAD/limestone) 35 (AAD/granite) 27 (AAM/limestone) 26 (AAM/granite) 1.26 1.35

Little and Jones 41 decreased as pH increased. Kiggundu and Roberts (1988) point out that these results indicate that stabilization of the pH sensitivity at the asphalt–aggregate interface can minimize the potential for bond breakage; provide strong, durable bonds; and reduce stripping. Tarrer (1996) concluded that (a) the bond between asphalt and aggregate depends on surface chemical activity, (b) water at the aggregate surface (in the field) is at a high pH, (c) some liquids used as antistrips require a long curing period (in excess of about 3 hours) to achieve resistance to loss of bond at higher pH levels, and (d) it is possible to achieve a strong chemical bond between aggregate and asphalt cement that is resistant to pH shifts and a high pH environment. This strong chemical bond can be achieved by the formation of insoluble organic salts (such as calcium-based salts), which form rapidly and are not affected by high pH levels or pH shifts. Although pH shifts affect chemical bonds, it is important to keep the magnitude of the pH shifts in proper perspective. Normally pHs as high as 9 or 10 will not dislodge amines from the acidic surfaces of aggregates, nor will they affect hydrated lime. Values of pH greater than 10 are not normally developed in asphalt mixtures unless a caustic such as lime is added. However, pHs below about 4 can dislodge amines from an aggregate surface and can dissolve lime depending on the type of acid used; these low pHs are not found in hot-mix asphalt. Environmental Effects on the Aggregate–Asphalt System Terrel and Shute (1989) report that factors such as temperature, air, and water have a profound effect on the durability of asphalt concrete mixtures. In mild climates where good-quality aggregates and good-quality asphalt cements are available, the major contribution to deterioration is traffic loading and the resulting distress manifestations. Premature failure may result when poor materials and traffic are coupled with severe weather. Terrel and Al-Swailmi (1994) identify a number of environmental factors of concern: water from precipitation of groundwater sources, temperature fluctuations (including freeze–thaw conditions), and aging of the asphalt. They identify traffic and construction techniques, which are external to the environment, as important factors. Factors considered by Terrel and Shute to influence water sensitivity in asphalt mixtures are given in Table 3. Adhesion Theories Terrel and Shute (1989) describe four theories that are often used to explain the adhesion between asphalt and aggregate: (a) chemical reaction, (b) surface energy, (c) molecular orientation, and (d) mechanical adhesion. Most likely a combination of mechanisms occurs synergistically to produce adhesion, and no one theory describes adhesion. Terrel and Shute explain that the four theories are affected by the following factors: surface tension of the asphalt cement and aggregate, chemical composition of the asphalt and aggregate, asphalt viscosity, surface texture of the aggregate, aggregate porosity, aggregate cleanliness, and aggregate moisture content and temperature at the time of mixing with asphalt cement. Chemical Reaction Chemical reaction is based on the premise that acidic and basic components of both asphalt and aggregate react to form water-insoluble compounds that resist stripping. Rice (1958) suggests the possibility of selective chemical reactions between aggregate and asphalt species. Jeon et al. (1988) described chemisorption of asphalt functionalities on aggregate surfaces and quantified the amount of coverage using a Langmuir model. Thelen (1958) had previously proposed that a

42 Moisture Sensitivity of Asphalt Pavements: A National Seminar TABLE 3 Factors Influencing Response of Mixtures to Water Sensitivity (After Terrel and Shute 1989) Variable Factor Compaction method Voids Permeability Environment Time Existing condition Water content Asphalt Aggregate Materials Modifiers or additives Curing Dry versus wet Soaking Vacuum saturation Freeze–thaw Repeated loading Conditioning Drying Traffic Environmental Other Age bond formed by chemical sorption might be necessary in order to minimize stripping potential in asphalt–aggregate mixtures. Robertson (2000) states that overall polarity or separation of charge within the organic molecules promotes attraction of polar asphalt components to the also polar aggregates. He explains that while neither asphalt nor aggregate has a net charge, components of both form nonuniform charge distributions and behave as if they have charges that attract the opposite charge of the other material. As established by Curtis et al. (1992), this is confounded by the fact that aggregates vary substantially in charge distribution and this charge distribution is affected by the environment. Robertson (2000) goes on to explain the types of reactions that might occur between the polar aggregate surface and asphalt cement. He states that, at a molecular level, basic nitrogen compounds (pyridines) adhere tenaciously to aggregate surfaces. He also describes carboxylic acids in asphalt cement. While they are quite polar and adhere strongly to dry aggregate, they tend to be removed easily from aggregate in the presence of water; but this varies with the type of acid. Plancher et al. (1977) explain that monovalent cation salts, such as sodium and potassium salts of carboxylic acids in asphalt, can be easily removed from the aggregate surface because they are essentially surfactants or soaps, which debond under the “scrubbing” action of traffic in the presence of water. On the other hand, Robertson (2000) indicates that divalent or doubly charged salts of acids (such as calcium salts from hydrated lime) are much more resistant to the action of water. This is also discussed by Scott (1978), Plancher et al. (1977), and Petersen et al. (1987). Robertson (2000) describes very recent observations at Western Research Institute (Williams et al. 1998) that indicate that aged asphalts may be much more prone to moisture damage than unaged asphalts. In some but not all asphalts, a very strongly acidic material appears with oxidation. Robertson (2000) indicates that if asphalt acids are converted to sodium salts (as can happen with some aggregates), a detergent will be formed.

Little and Jones 43 However, calcium salts of detergents are far less moisture sensitive or are deactivated with lime. Robertson (2000) concludes by warning the user to ensure that the acids in asphalts are neither free nor in the form of monovalent salts. Surface Energy and Molecular Orientation From a simplistic viewpoint, surface energy may be described in terms of the relative wettability of aggregate surfaces by water or asphalt. Water is a better wetting agent than asphalt due to lower viscosity and a lower surface tension. However, the concept of using surface energy to calculate the cohesive strength of the asphalt mastic and the adhesive bond energy between aggregate and the asphalt cement or between aggregate and the mastic is a much more complex subject that is worthy of a more detailed discussion. This is presented in the section entitled “Nature of Asphalt–Aggregate Interaction.” Molecular orientation is coupled with surface energy because both are a part of a theory that considers structuring of asphalt molecules at the asphalt–aggregate interface and assumes that adhesion between asphalt and aggregate is facilitated by a surface energy reduction at the aggregate surface as asphalt is adsorbed to the surface (Hubbard 1938; Rice 1958; Sanderson 1952). Kiggundu and Roberts (1988) describe molecular orientation and surface energy as synergistic processes. They also mention the observations of researchers regarding surface phenomena. For example, Yoon (1987) and Tarrer (1996) described how aggregates that impart a relatively high pH to contact water or that have a relatively high zeta potential have a high stripping or debonding potential. Scott (1978) stated, “If water penetrates the asphalt film to the mineral aggregate surface under conditions where micro droplets form, the pH reached may be sufficient to ionize and dissociate adsorbed water molecules.” Mechanical Adhesion This physical form of adhesion relies on physical aggregate properties, including surface texture, porosity or absorption, surface coatings, surface area, and particle size (Terrel and Al-Swailmi 1994). The philosophy is rather simple—to produce an aggregate surface capable of maximizing the surface area and texture to facilitate a strong physical bond that can synergistically improve the nature of the chemical bond between the asphalt and aggregate even in the presence of water. Aggregate properties that affect adhesion will be discussed in more detail later. Cohesion Theories Cohesion develops in the mastic and is influenced by the rheology of the filled binder. As will be discussed in more detail subsequently, Kim et al. (2002) describe how the resistance of a mastic to microcrack development is strongly influenced by the dispersion of mineral filler. Thus, the cohesive strength of the mastic is controlled not by the asphalt cement alone, but by the combination and interaction of the asphalt cement and the mineral filler. Terrel and Al-Swailmi indicate that water can affect cohesion in several ways, including weakening of the mastic due to moisture saturation and void swelling or expansion. Cohesion properties would logically influence the properties in the mastic beyond the region where interfacial properties dominate. A classic reminder of this is the work of Schmidt and Graf (1972), who show that an asphalt mixture will lose about 50 percent of its modulus upon saturation. The loss may continue with time, but upon drying, the modulus can be completely recovered. This is shown graphically in Figure 1.

44 Moisture Sensitivity of Asphalt Pavements: A National Seminar Cheng et al. (2002) describe the severe weakening of asphalt mixtures when they are subjected to moisture conditioning. Figure 2 illustrates the strength loss in a repeated load triaxial test when subjected to 85% saturation. Cheng et al. (2002) indicate that this strength loss is predictable when one compares the wet adhesive bond strength between the asphalt and the aggregate with the much higher dry adhesive bond strength. But Cheng et al. (2002) go on to demonstrate that the rate of damage in various mixtures is also related to the diffusion of water into the asphalt mastic, and that the asphalts that hold the greatest amount of water accumulate damage at a faster rate. Combining Theories Kiggundu and Roberts (1988) attempted to combine some of the theories discussed above. They realized that no single theory properly explains moisture damage. They tabulated the primary and secondary contribution relationships in Table 4. This table attempts to relate theories that explain loss of adhesion to stripping mechanisms. For example, the mechanism of pH instability is, according to Kiggundu and Roberts, explained by both chemical reaction theory and physical and chemical components of interfacial energy theory. Detachment, as a second example, is believed by Kiggundu and Roberts to be explained by physical and chemical aspects of interfacial energy theory as well as physical aspects of mechanical interlock theory. The physical aspects are manifested, according to Kiggundu and Roberts, by surface energy, while the chemical aspects are contributed by the effect of polarity of the molecules present at the common boundary. Even with this attempt to simplify the interaction of different theories and mechanisms, the interactive complexity of the processes becomes clearly evident. For example, surface bond is not solely a physical process because surface bond is dictated by the chemical nature of bonding at the asphalt and aggregate surface as well as by the presence of broken bonds or incomplete coordination of atoms due to broken bonds resulting in an increase in free energy. FIGURE 1 Effect of moisture on resilient modulus is reversible. (After Schmidt and Graf 1972.)

Little and Jones 45 LS-AAD and LS-AAM Dry and Wet 0 500 1000 1500 2000 2500 3000 0 5000 10000 15000 20000 25000 Number of Cycles Pe rm an en t M ic ro st ra in LSAADWet1 LSAADDry1 LSAAMDry1 LSAAMWet1 FIGURE 2 Repeated load permanent deformation experiments for AAD-limestone and AAM-limestone in the dry and wet conditions. (After Cheng et al. 2002.) TABLE 4 Proposed Relationships Between Theories of Adhesive Bond Loss and Stripping Mechanisms (After Kiggundu and Roberts 1988) THEORY Mechanical Interlock Chemical Reaction Interfacial Energy Proposed Operating Mode P C P-C P C P-C P C P-C Detachment S S W Displacement S S Spontaneous Emulsification S W Film Rupture S Pore Pressure S Hydraulic Scouring S St rip pi n g M ec ha n ism pH Instability S S P = Physical C = Chemical P-C = Physical – Chemical S = Primary Contributor W = Secondary Contributor

46 Moisture Sensitivity of Asphalt Pavements: A National Seminar NATURE OF ASPHALT–AGGREGATE INTERACTION Adhesive Failure Versus Cohesive Failure Damage in asphalt mixtures can occur within the mastic (cohesive fracture) or at the aggregate– mastic interface (adhesive fracture or failure). Whether or not a cohesive or adhesive failure occurs depends on the nature of the mastic and the relative thickness of the mastic around the coarse and fine aggregate. Lytton (personal communication, 2002) used micromechanics to assess the “thickness” of the asphalt film at which adhesive failure gives way to cohesive failure. Figure 3 is a plot of the cohesive and adhesive bond strength determined from cohesive and adhesive surface energies versus thickness of the asphalt binder or mastic. The theory essentially states that asphalt mixtures with thin asphalt films fail in tension by adhesive bond rupture, while those with thicker asphalt films (or mastic films) fail because of damage within the mastic (cohesive failure) as opposed to interfacial debonding. The thickness that differentiates these two types of failure is dependent on the rheology of the asphalt (or mastic), the amount of damage the asphalt or mastic can withstand prior to failure, the rate of loading, and the temperature at the time of testing. Consider an example. According to Figure 3, when asphalt or mastic coatings are thin, adhesive strength controls performance. In this stage, the adhesive bond strength in the presence of water determines mixture strength and is the critical condition. On the other hand, when asphalt or mastic coatings are relatively thick, thicker than the transition point of Figure 3, cohesive properties limit or control damage resistance. Therefore, in this situation, the impact of moisture intrusion into the mixture may be the key to assessing moisture damage of the mixture. In this case it may be more important to consider the impact of how much moisture the mastic holds and the impact on rheology of this infused water (Table 1) than to consider adhesive bond strength in the presence of water. 0 100 200 300 400 500 600 700 0 0.005 0.01 0.015 0.02 Asphalt Film Thickness, in T en si le S tr en g th , p si Cohesive Tensile Strength Adhesive Tensile Strength FIGURE 3 Adhesive versus cohesive bond failure based on asphalt film thickness.

Little and Jones 47 Realistically, it is necessary to consider both adhesive bond and cohesive strength properties of the mixtures as mastic coatings range from relatively thin to relatively thick throughout the mixture. Mixture strength then becomes a question of the statistical distribution of conditions where adhesion or cohesion strengths control. Thus the effects of moisture on adhesive bond strength and cohesive mastic strength are perhaps equally important. Kim et al. (2002) used dynamic mechanical analysis (DMA) testing to evaluate the rate of damage in asphalt binders and mastics. DMA subjects a cylindrical asphalt mastic to cyclic, torsional loading until failure occurs. Failure is due to the development and propagation of microcracks, which begin at the outer circumference of the cylindrical sample. Kim et al. (2002) demonstrated that the rate of damage and the amount of damage various mastics can accumulate before failure depend on the nature of the mastic. This is critically important because it essentially states that a mastic that is well designed will tolerate more damage prior to failure than one that is not. Kim et al. showed that mastics with the proper amount of and type of filler can accommodate more damage prior to failure than unfilled systems and that polymer-modified systems can accommodate more damage prior to failure than nonmodified systems. This indicates that the nature of the mastic (and the impact of the filler or modifier) strongly affects moisture damage because it helps control whether a cohesive or an adhesive bond failure occurs. Figure 4 is a plot of accumulated dissipated pseudostrain energy (DPSE) versus number of torsional DMA load cycles to failure of representative mastics. The filled asphalts or mastics allow a higher accumulation of DPSE prior to failure than neat asphalt. The type of filler and its physicochemical interaction with the asphalt have a strong impact (Lesueur and Little 1999). The proposed mechanism of fatigue life extension is that properly designed mastics or modified asphalts affect the rate of microdamage accumulation and resist rapid, catastrophic failure via microcrack coalescence. The process may be a redistribution, redirection, or “pinning” of crack energy. 1.E+03 1.E+04 1.E+05 1.E+06 1.E-01 1.E+00 Strain (%) C u m u la ti ve D P S E t o F ai lu re (* 10 ^1 0) AAD-1 AAM-1 AAD+LS AAD+HL AAM+HL 0.28% Strain FIGURE 4 Relationship between cycles to failure (DMA) and DPSE for various mastics. (After Kim et al. 2002.)

48 Moisture Sensitivity of Asphalt Pavements: A National Seminar Little et al. (1999) have shown that microcrack damage rates are related to cohesive mastic surface energies based on Schapery’s viscoelastic fracture theory. Cohesive bond strength can be calculated on the basis of cohesive surface energy measurements of the dry or water-saturated mastic (Cheng et al. 2002). Moisture intrusion weakens the cohesive bond and makes the resulting mixture more susceptible to damage (Table 2). Effect of Aggregate Characteristics A general hypothesis has been that acidic aggregates are hydrophobic while basic aggregates are hydrophilic. However, there are notable exceptions (e.g., Majidzadeh and Brovold 1968; Maupin 1982). The general conclusion is that few if any aggregates can completely resist the stripping action of water (Tarrer and Wagh 1991). Tarrer and Wagh (1991) list a number of factors that influence the asphalt–aggregate bond: surface texture, penetration of pores and cracks with asphalt, aggregate angularity, aging of the aggregate surface through environmental effects, adsorbed coatings on the surface of the aggregate, and the nature of dry aggregates versus wet aggregates. Surface texture of the aggregate affects its ability to be properly coated, and a good initial coating is necessary to prevent stripping (Maupin 1982). Cheng et al. (2002) have demonstrated that the adhesive bond, calculated from basic surface energy measurements of the asphalt and aggregate, between certain granites and asphalt was higher than between limestone aggregate and asphalt when the bond was quantified as energy per unit of surface area. However, when the bond was quantified as energy per unit of aggregate mass, the bond energy was far greater for the calcareous aggregates than for the siliceous. These results agreed well with mechanical mixture testing. They point out clearly the importance of the interaction of the physical and the chemical bond. Besides the importance of a good mechanical bond promoted by an amenable surface texture, stripping has been determined to be more severe in more angular aggregates (Gzemski et al. 1968) because the angularity may promote bond rupture of the binder or mastic, leaving a point of intrusion for the water. Cheng et al. (2002) substantiate this as they have shown that, regardless of the strength of the bond between the asphalt and aggregate, the bond between water and aggregate is considerably stronger. Table 5 shows adhesive bond strengths calculated in ergs/cm2 for five different liquids or semisolids (four binders and water) and three different aggregates. Note that the bond between water and either of the aggregates is at least 30% stronger than for any of the asphalts. TABLE 5 Adhesive Bond Energy per Unit Area of Sample (ergs/cm2) (After Cheng et al. 2002) Aggregate Binder Georgia Granite Texas Limestone Colorado Limestone AAD-1 153 141 124 AAM-1 198 205 179 Rubber asphalt 212 189 166 Aged rubber asphalt 171 164 145 Water 256 264 231

Little and Jones 49 The effects of crushing of the aggregate are very interesting. One might expect that a freshly crushed aggregate surface would have a greater free energy than an uncrushed aggregate surface. This is because broken bonds due to fracture should substantially increase the internal energy even though having something of a countereffect on randomness (entropy increase). However, there is another side to consider. Tarrer and Wagh (1991) point out that sometimes newly crushed faces tend to strip faster than stockpiled aggregates. They state that it is characteristic of many aggregates that one or more layers of water molecules become strongly adsorbed on the aggregate surface as a result of electrochemical attractions. Thelen (1958) states that upon aging, the outermost adsorbed water molecules may become partially replaced or covered by organic contaminants present in air (e.g., fatty acids and oils) that reduce stripping potential. However, this seems unlikely because these fatty acids are relatively heavy and are not likely to volatilize. A general oxidation process reduces free radicals at the oxidation sites and may make weathered aggregates more resistant to stripping than freshly crushed aggregates. On the other hand, if the freshly crushed aggregate can be effectively coated with asphalt and the adsorption of the water layer can be prevented from the outset, the asphalt–aggregate bond developed may be the most effective. Certainly there is much room for advancement in the state of knowledge here. Tarrer and Wagh (1991) and Hindermann (1968) discuss the effect of crushed aggregate surface on bond strength in light of the ways the aggregate surface may react to broken bonds created by crushing or cleavage. They discuss two potential reactions. In one, new coordination bonds may be formed by redirection inward to the atomic lattice. If this were the case, the aggregate would have no affinity for asphalt or water. This is a very unlikely process. In the second and more likely process, water, oil, or other contaminants in the air are attracted to the fresh surface to satisfy broken bonds. Since water is normally available, the driving force for the adsorption of water on the freshly crushed aggregate faces is that it reduces the free energy of the system. Although asphalts and other organics may also spread over the crushed faces of the aggregate, the rate at which they spread depends largely on their viscosity. Water is more prevalent and spreads much more quickly (Tarrer and Wagh 1991). Apparently, asphalt and organic materials spread over water films on an aggregate surface and tend to be stripped from water films by water (Tarrer and Wagh 1991), further complicating the process. Clearly, Tarrer and Wagh (1991) make the case that heating aggregates that contain free water and adsorbed water films may remove free water and the outermost adsorbed water molecules, causing the interfacial tension between the aggregate and the asphalt to decrease (Thelen 1958; Majidzadeh and Brovold 1968), resulting in a reduction in stripping potential. The heating effect probably also reduces asphalt viscosity and allows better penetration into the aggregate surface, promoting a more effective physical bond. According to Tarrer and Wagh (1991), the asphalt–aggregate bond is enhanced by three processes: (a) preheating the aggregate, (b) weathering the aggregate, and (c) removing aggregate coatings. When the aggregate surface is heated, the outermost adsorbed water layer is released, improving the state of interfacial tension between the asphalt and aggregate and, in turn, improving the bond between asphalt and aggregate. The weathering process results in a replacement of the adsorbed water layer with organic fatty acids from the air. This results in an improved asphalt–aggregate bond (Fromm 1974). A dust coating on the aggregate surface promotes stripping by preventing intimate contact between the asphalt and aggregate and by creating channels through which water can penetrate (Castan 1968).

50 Moisture Sensitivity of Asphalt Pavements: A National Seminar Calculation of Asphalt–Aggregate Bond Strength Fundamental Mechanisms In 1984 Schapery proposed a basic viscoelastic fracture theory, which was derived from first principles of materials science and based on an energy balance. This theory states that the load- induced energy that causes fracture damage is balanced by the energy stored on newly formed crack faces. The energy imparted to the system can be quantified as the product of two properties of the materials in question: tensile creep compliance over the time of loading and the strain energy per unit of crack area produced from one tensile load to the next. The energy stored on fracture faces can be quantified by surface energy measurements of the material. Fortunately, the material properties required to assess this energy balance can be effectively measured. Si et al. (2002) and Kim et al. (2002) demonstrate how to measure tensile creep compliance and the strain energy that causes damage (pseudostrain energy) during cyclic fatigue testing of asphalt mixtures. This concept of pseudostrain energy is not mysterious; it is merely a mathematical calculation that allows one to separate the dissipated energy that actually causes damage from the energy that is recovered over time and does not cause damage. Surface energies can also be measured. Cheng et al. (2001; 2002) have demonstrated how surface energy measurements on the aggregate and asphalt cement can be used to calculate surface energies of cohesion (related to fracture within the mastic—asphalt and filler) and adhesion (related to fracture at the asphalt–aggregate interface). Using this fundamental look at fracture damage, it is easy to relate surface energy to pavement distress and to understand the wide-ranging importance of surface energy as an indicator of distress in asphalt pavements. Obviously, surface energy can be used to directly assess fracture potential: both cohesive and adhesive. But surface energy is also related to permanent deformation distress, the fatigue failure process, and cohesive strength reduction and adhesive failure (stripping) in the presence of moisture. It is important to briefly develop this connection. As previously described, Bhairampally et al. (2000) used a tertiary damage model developed by Tseng and Lytton (1987) to demonstrate that well-designed asphalt mixtures tend to “strain harden” upon repeated loading. As previously discussed, this microcracking or tertiary damage leads to a departure from the typical “strain hardening” stress–strain curve representing an accelerated rate of damage due to the development of microcracking and the ultimate acceleration of microcrack growth. Cheng et al. (2002) have shown that the acceleration in damage, or tertiary damage, is related to cohesive and adhesive bond strengths of the mastic and asphalt mixtures in question. Table 1 presents the strong relationship between the number of cycles to failure in repeated load permanent deformation testing and cohesive and adhesive bond energies (which were calculated from bitumen and aggregate surface energy measurements). In Table 1 the free energy of adhesion in the presence of water is calculated. The negative sign indicates a preference of the aggregate for water over asphalt, and a less negative value represents a lower driving force to replace the asphalt in question with water. Thus it is consistent that asphalt AAM bonds more strongly with either the limestone or granite aggregate than asphalt AAD and that it is less likely to strip. Two back-to-back studies for the Federal Highway Administration performed at Texas A&M University’s Texas Transportation Institute have established the importance of the healing phenomenon in the fatigue damage process. Field validation of healing that occurs during rest periods was presented by Williams et al. (1998). Here the researchers measured a substantial

Little and Jones 51 recovery in modulus via surface wave measurement following rest periods. Little et al. (1999), Kim et al. (1997), Kim et al. (2002), and Si et al. (2002) measured the healing effect during repeated load tensile and torsional shear fatigue testing. They quantified the effect of healing in terms of recovery of dissipated energy during the rest period and in terms of extended fatigue life due to the cumulative effect of a number of rest periods. Little et al. (1999) further established that the healing process is composed of a short-term effect governed by the Lifshitz–Van der Waals component of surface energy and a long-term effect governed by the acid–base component of surface energy. All of these studies are consistent in their findings that a higher acid–base component of surface energy and a lower Lifschitz–Van der Waals component of surface energy produce a superior healing asphalt. Each of the studies referenced provide consistent and convincing experimental data substantiating this assertion. The fact that surface energy of dewetting is fundamentally related to fracture and that surface energy of wetting is fundamentally related to healing is discussed by Little et al. (1998) and Little et al. (1999). In fact, Schapery presented a corollary to his viscoelastic fracture theory for healing in which he related long-term healing to surface energy and found that an increase in surface energy resulted in better healing. After studying the results of a large experimental matrix comparing surface energy with healing rate plots, Lytton discovered that healing has two components: short term and long term. He determined that the short-term healing rate (and magnitude) is inversely correlated with the Liftshitz–Van der Waals component of surface energy while the long-term healing rate (and magnitude) is directly related to the acid–base component of surface energy. When this is coupled with Schapery’s theory of viscoelastic fracture, a much more complete understanding of the entire fracture fatigue process is achieved, because the fatigue process consists of fracture during loading and healing during rest periods between load applications. Lytton et al. (1993) showed that the healing process is responsible for the major component of the laboratory-to-field fatigue shift factor. Since this shift factor historically ranges between about 3 and more than 100, healing is indeed a significant part of the fatigue damage process. A logical extension can be made from understanding adhesive fracture based on surface energy to understanding the debonding process between bitumen and aggregate in the presence of moisture (stripping). Cheng et al. (2002) present a detailed methodology by which to measure the surface energies (all components) of asphalt using the Wilhelmy plate method and the surface energies of aggregates (all components) using the universal sorption device (USD). They then show how to compute the adhesive bonding energy between the bitumen and the aggregate both in a dry state and in a wet state (in the presence of a third medium—water). Table 5 (Cheng et al. 2002) demonstrates that the adhesive bond calculated per unit area of aggregate is highly dependent on the aggregate and asphalt surface energies and that the values of the adhesive bond vary over a significant range. They further point out that the affinity of the aggregate for water is far greater than it is for asphalt, so that if water has access to the aggregate interface it is likely to replace the asphalt (strip), and the rate of replacement is a function of the aggregate–asphalt bond strength. In Table 6 the same results are presented in terms of energy per unit mass instead of energy per unit area. Energy per unit mass takes the surface area into account. This is shown to be very important as the rank order of adhesive bond energy changes when this conversion is made. The far greater surface area of the limestone ranks it ahead of the granite in terms of bonding energy per unit mass even though this particular granite actually has a higher bonding energy per unit area.

52 Moisture Sensitivity of Asphalt Pavements: A National Seminar TABLE 6 Gibbs Free Energy per Unit Mass (ergs/gm × 103) (After Cheng et al. 2002) Aggregate Binder Georgia Granite Texas Limestone Colorado Limestone AAD-1 158 614 375 AAM-1 206 889 536 Rubber asphalt 219 819 497 Aged rubber asphalt 178 714 435 Cheng et al. (2002) also measured the diffusivity and moisture-holding potential of various bitumens using the USD. Lytton developed a method by which to solve Fick’s second law to differentiate adsorption from absorption in the sorption process so that diffusivity and moisture-holding potential could be determined. Using this approach, Cheng et al. (2002) found that diffusivity and water-holding potential vary significantly among bitumens and that the ultimate water-holding potential is related closely to damage. For example, asphalt AAD-1 has a lower diffusivity than asphalt AAM-1, but it has much greater water-holding potential (Table 2). This leads to a much higher level of damage in mixtures with AAD-1 than in mixtures with AAM-1. This damage may be due to two factors: the weakening of the mastic due to diffusion of water into the bitumen and the migration of water through the mastic to the mastic–aggregate interface causing stripping. Fundamental Principles Shared by Material Processes The preceding section points out that surface energy can be fundamentally related to material processes and failure mechanisms. From this background a set of principles can be established that can be used to measure material properties required to assess the basic distress processes. These principles are as follows: 1. All materials have surface energies, both asphalts and aggregates. 2. All surface energies have three components, all of which can be measured. 3. The theory of adhesive and cohesive bonding has been developed in industrial surface chemistry and chemical engineering and is used reliably and with confidence. 4. Fracture and healing involve two types of material properties: chemical and physical. Neither fracture nor healing can be properly described without the use of both properties: chemical—surface energies; physical—modulus and tensile strength and the way they change with age. 5. The presence of moisture at the asphalt–aggregate interface interrupts the bond and accelerates the rate of fracture damage. The presence of moisture in the mastic reduces cohesive strength and fracture resistance and, therefore, reduces the potential for microcracks in the mastic to heal. 6. On the basis of their surface energy characteristics, some combinations of aggregates and asphalts can be determined by calculations to bond well and heal well, whereas other combinations will not. In selecting materials for an asphalt pavement mixture from among

Little and Jones 53 several available alternatives, it will always be possible to select the best combination of all of the available aggregates and asphalts to resist fracture, heal, bond well, and resist moisture damage. Predicting their performance will require the measurement of physical properties as well. Effect of Asphalt Composition on Adhesion Asphalt Composition The chemistry of asphalt is complex. This brief overview is certainly a simplification of the complex nature of asphalt and is meant to provide (a) definitions of basic terms and (b) descriptions of basic asphalt components, which are used in discussion throughout this paper. Elemental Composition Asphalt molecules are comprised primarily of carbon and hydrogen (between 90% and 95%) by weight. However, the remaining atoms, called heteroatoms, are very important to the interaction of asphalt molecules and hence the performance of asphalt. They include oxygen, nitrogen, sulfur, nickel, vanadium, and iron. Molecular Structure Asphalt atoms are linked together to form molecules. Perhaps the simplest is the aliphatic carbon–carbon chain saturated with hydrogen bonds. The carbon–carbon bonds can also form rings saturated with hydrogen. These saturates are essentially nonpolar and interact primarily through relatively weak Van der Waals forces. A second class of asphalt molecules involves aromatics. This molecule has six carbon atoms in the form of a hexagonal ring. This ring possesses a unique bond with alternating single and double bonds between carbon atoms. Figure 5 shows representative examples of saturates (aliphaltic and cyclic) and aromatic structures. FIGURE 5 Types of asphalt molecules. (From Jones 1992.)

54 Moisture Sensitivity of Asphalt Pavements: A National Seminar Bonds Among Asphalt Molecules Strong covalent bonds link atoms together to form asphalt molecules. These molecules interact with one another through much weaker bonds (Jones 1992): pi–pi bonds, hydrogen or polar bonds, and Van der Waals bonds. These are represented in Figure 6. Pi–pi bonds are unique to aromatic molecules. They provide polarity and the ability of aromatic molecules to link together in unique configurations, including a stacked arrangement as shown in Figure 6. Heteroatoms among asphalt molecules develop polarity and link together by forming hydrogen bonds. Figure 6 shows a hydrogen bond between two very important asphalt functional groups: a sulfoxide and a carboxylic acid. Van der Waals bonding is the weakest of the secondary bonds. They form when molecules cool or stress is removed. Van der Waals bonding is responsible for the free-flowing nature of asphalt at high temperatures versus the semisolid nature at lower temperatures (Jones 1992). As a point of reference, it is important to understand that covalent primary bonds within the molecule are from 10 to 100 times stronger than secondary bonds. Polar Versus Nonpolar Molecules Polar molecules form “networks” through hydrogen and pi–pi bonds that give asphalt its elastic properties. Nonpolar materials form the body of the material in which the network is formed and contribute to the viscous properties of asphalt (Jones 1992). Degree of polarity is the most important property of polar molecules, while degree of aromaticity is the second most important. Highly polar and highly aromatic molecules form the most interactive and strongest molecular networks. FIGURE 6 Types of intermolecular asphalt bonds. (After Jones 1992.)

Little and Jones 55 Nonpolar molecules do not interact strongly enough to form networks, but they do substantially influence asphalt performance. The molecular weight of nonpolar molecules is related to low temperature performance (Jones 1992). A preponderance of high-molecular- weight nonpolar molecules will lead to asphalts that stiffen and perform poorly at low service temperatures. If nonpolars are waxy, they will crystallize at low temperatures and become crack susceptible. Nonpolar and polar molecules must interact in an acceptable manner or be “compatible.” If polar and nonpolar molecules are relatively similar in chemistry, they will be compatible; however, if they are very different, the polar network will not stay in solution, and phase separation can be a substantial problem. Asphalt Model Jones (1992) explains the history of development of an asphalt model. He describes how analytical techniques including size exclusion chromatography and ion exchange chromatography have led to viewing asphalt as a two-phase system. The polar molecules interact with each other through polar–polar or hydrogen bonding. These bonds form associations that create a network within the nonpolar solvent molecules. However, as explained by Jones, both phases make a significant contribution to asphalt performance. Figure 7 illustrates the model described by Jones (1992)—the Strategic Highway Research Program (SHRP) model. The associations of polar molecules are due to polar sites on the asphalt through hydrogen bonding. Other interactions take place through pi–pi bonding and Van der Waals bonding. These interactions provide the major contribution to viscoelastic properties of the asphalt. Actually, the term phase is not accurate in the description of polar versus nonpolar components because the mixture is homogeneous and the bonds between the polar molecules are rather weak and form and break constantly. FIGURE 7 SHRP asphalt model. (After Jones 1992.)

56 Moisture Sensitivity of Asphalt Pavements: A National Seminar Multifunctional Organic Molecules Data show that having two or more functional groups on the same molecule makes it possible to form chains of weak polar–polar interactions. According to Jones (1992), these chains are the foundation of the polar networks. An example of a multifunctional organic molecule is one with both a carboxylic acid (R-COOH) and a sulfoxide (S=O) on the same molecule. Figure 8 is an example of two multifunctional organic molecules. The first one contains three heteroatoms in its structure: a phenol group (O-H), a sulfoxide (S=O), and a ketone (C=O). The second example is a linear chain molecule that contains a carboxylic acid (R-COOH) and a mercaptan (SH) group. Multifunctional organic molecules have a major impact on aging. This is because for polar molecules to generate significant physical changes, it is necessary for them to interact in chainlike structures or form networks. Figures 9 and 10 illustrate how oxidation of molecules with a single active site results in a “quenching” of the effect of oxidation, while the oxidation of molecules with multiple active sites develops a continuous network. FIGURE 8 Types of multifunctional organic molecules. (After Jones 1992.) FIGURE 9 Asphalt with simple active sites. (After Jones 1992.)

Little and Jones 57 FIGURE 10 Asphalt with multiple active sites. (After Jones 1992.) Asphalt Chemistry and Adhesion Polarity or separation of charge within the organic molecules promotes attraction of polar asphalt components to the polar surfaces of aggregates. Although neither asphalt nor aggregate has a net charge, components of both have nonuniform charge distributions, and both behave as if they have charges that attract the opposite charge of the other material. Curtis et al. (1992) have shown that aggregates vary widely in terms of surface charge and are influenced by environmental changes. Robertson (2000) points out that adhesion between asphalt and aggregate arises between the polars of the asphalt and the polar surface of the aggregate. He also points out that polarity alone in asphalt is not sufficient to achieve good adhesion in pavements because asphalt is affected by the environment. Robertson (2000) further states that asphalt has the capability of incorporating and transporting water. Absorption of water varies with asphalt composition and changes further as asphalt is oxidized. Cheng et al. (2002), as discussed previously, have shown that a substantial quantity of water can diffuse through and be retained in a film of asphalt cement or an asphalt mastic, substantially changing the rheology of the binder. Robertson (2000) states that at the molecular level in asphalt, basic nitrogen compounds (pyridines) tend to adhere to aggregate surfaces tenaciously. Carboxylic acids are easily removed from aggregate in the presence of water if the acids form a monovalent salt by interaction at the aggregate surface, but divalent (calcium) salts of acids are much more resistant to the action of water. Curtis (1992) ranked the affinity of various asphalt functional groups to bond to aggregate surfaces by using adsorption isotherms (UV adsorption spectroscopy). In general she found acidic groups, carboxylic acids, and sulfoxides to have the highest adsorptions, while ketone and nonbasic nitrogen groups had the least. However, the sulfoxide and carboxylic acids were more susceptible to desorption in the presence of water. According to Curtis (1992), the general trend of desorption potential of polar groups from aggregate surfaces is sulfoxide > carboxylic acid > nonbasic nitrogen ≥ ketone > basic nitrogen > phenol.

58 Moisture Sensitivity of Asphalt Pavements: A National Seminar Effect of Aggregate Properties on Adhesion A number of aggregate properties affect the adhesive bond between asphalt and aggregate: size and shape of aggregate, pore volume and size, surface area, chemical constituents at the surface, acidity and alkalinity, adsorption size surface density, and surface charge or polarity. Pore Volume and Surface Area Yoon and Tarrer (1988) investigated five aggregates (granite, dolomite, chert gravel, quartz gravel, and limestone). They measured pore volume, surface area, average pore size, and percentage coating after boiling. Their study showed that stripping resistance is defined by the level of physical bond that is achieved between the asphalt and aggregate, and this is, in turn, defined by surface area, pore volume, and pore size. The optimal resistance to stripping was developed in aggregates that provide a large surface area for bonding as well as a favorable pore size for adequate (deep) asphalt penetration. This is probably because when asphalt cement coats a rough surface with fine pores, air is trapped and the asphalt has difficulty penetrating the fine pores (Yoon and Tarrer 1988). However, the penetration of asphalt cement into pores is synergistically dependent not only on the pore structure but also on the viscosity of the asphalt cement at mixing temperatures. Yoon and Tarrer (1988) also determined that aggregates with approximately equal physical properties (e.g., pore volume and structure and surface area) can have very different properties depending on their basic chemistry and mineralogy, which define surface activity. Yoon and Tarrer found substantially higher bonding power for limestone than for quartz gravel even though both had similar physical surface structures. The results of Cheng et al. (2002) were very similar; they found that a certain granite aggregate has a much higher surface energy per unit area than a certain limestone, but when bonding energy was computed in terms of unit mass instead of unit surface area (incorporating effects of surface area), the limestone was predicted to have a much greater potential to resist damage in repeated loading tests of asphalt samples at 85% saturation. pH of Contacting Water Hughes et al. (1960) and Scott (1978) reported that adhesion between asphalt cement and aggregate in the presence of water became weakened when the pH of the buffer solution was increased from 7.0 to 9.0 (Scott 1978). Yoon and Tarrer (1988) showed that if different aggregate powders (chert gravel, quartz sand, quartz gravel, granite, limestone, and dolomite) were added to water and allowed to react with water for up to about 30 minutes, the pH of the blend will increase to some asymptotic value (Figure 11). Even granite, known to be acidic, showed an increase in pH over time to about 8.8. The granite reaction in water, which leads to this gradual pH increase, is, according to Yoon and Tarrer, due to the silicate lattice reaction with the water to impart excess hydroxyl ions as follows: | | | | − S − O − Na + H2O → −SiOH + Na + + OH − (1) This is a typical hydrolytic reaction of the salt of a weak acid.

Little and Jones 59 FIGURE 11 Changes in pH of water in which aggregates were immersed. (After Yoon and Tarrer 1988.) Yoon and Tarrer (1988) assessed the sensitivity of stripping to changes in pH of water in contact with the aggregate surface. They performed boiling stripping tests to verify the sensitivity. The pH of the water was modified by using a solution of HCl or NaOH. The stripping became more severe as the pH increased. Yoon and Tarrer explain that when an aggregate is being coated with asphalt, the aggregate selectively adsorbs some components of the asphalt— the more polar compounds and hydrogen bonds or salt links are formed. Obviously, the type and quantities of the adsorbed components affect the degree of adhesion. Yoon and Tarrer state that the presence of ketones and phenolics is thought to improve stripping resistance, whereas carboxylic acids, anhydrides, and 2-quinolenes are thought to increase stripping sensitivity because of the substantial water susceptibility of the associated bonds. According to Yoon and Tarrer (1988), the water susceptibility of the hydrogen bonds and salt links between the adsorbed asphalt components and the aggregate surface would increase as the pH of the water at the aggregate surface increases. Therefore, it seems reasonable to conclude that stripping sensitivity will increase as the pH of the water increases. Experimental results of Yoon and Tarrer (1988) substantiate this hypothesis. However, they warn that other surface aggregate properties also play a role. Different types of metal ions affect stripping potential. For example, alkaline earth metals in limestone associate strongly with the asphalt components in carboxylic acids to form alkaline earth salts, and the bonds formed are not dissociated easily in water even at a high pH. In other words, the adsorption is strong because of the insolubility of the alkaline earth salts formed between the limestone and the bitumen acids. The addition of hydrated lime offers a mechanism to tie up carboxylic acids and 2- quinolenes so they cannot interact with hydrogen bonding functionalities on the aggregate surface to produce moisture-sensitive bonds. Thomas (2002) points out that the interaction of lime with components in the asphalt not only prevents the formation of moisture-sensitive bonds but also subsequently allows more resistant bonds (e.g., with nitrogen compounds from the asphalt) to proliferate. He points out that an additional benefit of the use of lime is to react with or adsorb compounds that can be further oxidized and enhance the increase in viscosity as a

60 Moisture Sensitivity of Asphalt Pavements: A National Seminar result of oxidation. In fact, experiments at Western Research Institute (2002) show a substantial improvement in moisture resistance after severe freeze–thaw experiments when lime is added directly to the bitumen and before the bitumen is coated on the aggregate. Western Research Institute is currently studying the effect of bonds between the aggregate surface and bitumen components including sulfonic acids, ketones, and 2-hydroquinolines on moisture damage. In a manner similar to the reaction between acidic compounds such as carboxylic acids in asphalt and alkaline aggregate or with lime, an amine compound either if present in asphalt or added in the form of an antistripping additive will react with acidic surfaces as in the case of siliceous aggregates to form a surface compound. Evidence of the formation of such a surface compound between siliceous surfaces and amine compounds was demonstrated by Titova et al. (1987). 2 3 2 3 2 3 2 3 2 3 2 3 0 \ / \ / / \ / \ S i lic a ,T r ie th y la m in e c o m p o u n d S ta b le > 9 0 0 F in V a c u u m C H C H C H C H S i O H N C H C H Si O H N C H C H C H C H C H C H − − + − → − − − + − Surface Potential Interfacial activity between asphalt cement and the aggregate surface is fundamentally important in assessing stripping potential. Yoon and Tarrer state that the functional groups of asphalt that are adsorbed on the aggregate surface come mainly from the acid fraction of the asphalt. Yoon and Tarrer offer the example of carboxylic acid (R-COOH), which in the presence of water separates into the carboxylate anion (R-COO-) and the proton (H+). This causes the asphalt surface to have a negative polarity at the interface. Aggregates with water present are negatively charged, and as a result, a repulsive force develops between the negatively charged aggregate surface and the negatively charged asphalt surface at the interface. Payatakis (1975) states that solid surfaces in contact with water usually acquire changes through chemical reactions at the solid surface and adsorption of complex ions from the solution. For example, metal oxide surfaces in water hydrolyze to form hydroxyl groups: / \ | | | | H H O O O −M − O − M − + H2O → − M − O − M − (2) which subsequently dissociate as

Little and Jones 61 OHMOHM −+ +↔ −−− (3) A high pH value of the water in contact with the mineral surface will cause the surface to be more negatively charged. Yoon and Tarrer (1988) report that the intensity of the repulsion developed between the asphalt and aggregate depends on the surface charge of both the asphalt and aggregate. They used zeta potential as a method to measure aggregate surface charge and found a general trend that aggregates that have a relatively high surface potential in water are more susceptible to stripping (see Figure 12). SHRP Research on Aggregate Surface Chemistry General Labib (1991) confirmed the existence of a range of acid–base types among various SHRP aggregates using zeta potential measurements and electrophoretic mobility. He reported that it is significant that the initial pH of aggregates was greater than 9.0, irrespective of aggregate type. This would neutralize the bitumen carboxylic acids at the interface and cause hydrolysis of bitumen–aggregate bonds. The high pH was attributed to basic soluble salts even in acidic aggregates. Labib (1991) documented the sensitivity of the bitumen–aggregate bond to pH. He identified three pH regions. At pHs above about 8.5 (Region 3), dissolution of the surface silica occurred in quartz or silica aggregates. In carbonate-based aggregates at pHs between about 1 and 6 (Region 1), calcium ion dissolution occurred, and the presence of carboxylic acids enhanced stripping in this region through cohesive failure in the aggregate. Podoll et al. (1991) FIGURE 12 Comparison of aggregate surface potential and stripping propensity as determined by the boiling water tests. (After Yoon and Tarrer 1988.)

62 Moisture Sensitivity of Asphalt Pavements: A National Seminar used surface analysis by laser ionization to confirm that bitumen–aggregate bond disruption occurs within the aggregate and not at the interface. They found notably less sodium, potassium, and calcium in the top monolayer of aggregate in stripped areas than in unstripped areas. This indicates that dissolution of the cations was greater where bitumen had been stripped away. Scott’s (1978) work on bitumen-coated glass slides supports Podoll et al. He found that debonding occurred in the more water-soluble glasses and not in the more stable opal glasses. Jamieson et al. (1995) conclude that net adsorption of bitumen on aggregate is a function of five aggregate variables: potassium oxide, surface area, calcium oxide, zeta potential, and sodium oxide. Alkali earth metals (sodium and potassium) are detrimental to adhesion. Higher surface area provides more active sites per unit mass for interaction. Calcium forms water- resistant bonds, and aggregates with a more negative surface charge may provide more potential for adsorption. SHRP Adhesion Model The SHRP adhesion model concludes that aggregate properties have a greater impact on adhesion than do various binder properties. Adhesion is achieved mainly by polar constituents in the bitumen bonding with active aggregate sites through hydrogen bonding, Van der Waals interaction, or electrostatic forces. The general trend is that sulfoxides and carboxylic acids have the greatest affinity for aggregates. However, in the presence of water, sulfoxides and carboxylic acid groups are more susceptible to debonding, whereas phenolic groups and nitrogen bases are more effective in providing a durable bond (Jamieson et al. 1995). It is also apparent that aromatic hydrocarbons have much less affinity for aggregate surfaces than the polar groups. SHRP Stripping Model The SHRP view is that stripping is controlled by cohesive failure within the aggregate rather than at the bitumen–aggregate interface (Jamieson et al. 1995). Surfaces rich in alkali metals are more susceptible to debonding than surfaces rich in alkaline earth metals because the latter form water-insoluble salts with acid and other groups with the bitumen. Podoll et al. (1991) state that stripping of siliceous aggregate may be associated with the presence of water-soluble cations and aluminosilicates. The mechanism is probably dissolution of salts, dissociation of silica due to the high pH environment generated by solubilization of alkaline earth cations, electrostatic repulsion between negatively charged aggregates and ionic components of the bitumen at the surface, and dissolution of soaps formed between acid anions on the bitumen surface and alkali metal cations on the aggregate surface. The superior stripping resistance of some limestones is due to the formation of water- insoluble (covalent) bonds between calcium sites on the aggregate and bitumen constituents, but stripping of calcareous aggregate can occur where their water solubility is high. WAYS TO IMPROVE ADHESION Interaction of Acidic Aggregates and Asphalt with Alkaline Amine Compounds Amines have a long hydrocarbon chain. The chain is compatible in asphalt cement, and, in the presence of water, the amine is ionized to form an amine ion, R-NH3, which has a positive charge (cationic). The physical properties of fatty amines can be altered by changing the nature of the hydrocarbon chain while the chemical nature can be altered by changing the number of amine groups and their positions in the molecule (Porubszky et al. 1969). Taken together, the

Little and Jones 63 chain length and number of amine groups greatly influence the adhesion of the asphalt. Optimum performance is typically achieved with 14 to 18 carbon chain amines with one or two amine groups (Porubszky et al. 1969; Tarrer and Wagh 1991). Fatty amines enable asphalt to wet aggregate surfaces. The amine group reacts with the aggregate surface while the hydrophobic hydrocarbon chain of the fatty amine is anchored in the asphalt. The net effect (Tarrer and Wagh) is that the long hydrocarbon chain acts as a bridge between the hydrophilic aggregate and the hydrophobic bitumen surface, encouraging a strong bond. Effect of Hydrated Lime on Adhesive Bond Plancher et al. (1977) hypothesized that hydrated lime improved binder–aggregate adhesion by interacting with carboxylic acids in the asphalt and forming insoluble salts that are readily adsorbed at the aggregate surface. This is an important reaction because hydroxyl (OH) groups are found on the surfaces of siliceous aggregates. These SiOH groups form hydrogen bonds with carboxylic acid groups from asphalt and strongly affect the adhesion between the asphalt and aggregate (Hicks 1991). But this hydrogen bond is quickly broken in the presence of water, and the two groups dissociate and reassociate with water molecules through hydrogen bonding. This means that the hydrogen bonding between the water molecules and the SiOH group and between the water molecule and the COOH group is preferred over the bond between SiOH and COOH. When lime is added, some dissociation of the Ca(OH)2 molecule occurs, resulting in calcium ions (Ca++). These ions interact with the carboxylic acids (COOH) and 2-quinolenes (Petersen et al. 1987) to form rather insoluble calcium organic salts. This leaves the SiOH molecule free to bond with nitrogen groups in the asphalt (Petersen et al. 1987). These bonds are strong and contribute to adhesion. Figure 13 illustrates some of the important functional groups in asphalt. FIGURE 13 Chemical functionalities of importance in asphalt polarity and that contribute in the reaction with calcium. (After Petersen et al. 1987.)

64 Moisture Sensitivity of Asphalt Pavements: A National Seminar Schmidt and Graf (1972) state that the effectiveness of hydrated lime as an antistripping agent cannot be completely explained by the reaction between calcium from lime and the acids in the asphalt. They state that lime provides calcium ions that migrate to the aggregate surface and replace hydrogen, sodium, potassium, or other cations. In 1997 the Western Research Institute provided an excellent explanation of the hypothesized aggregate–asphalt interaction. Susceptibility to water stripping depends, at least in part, on the water solubility of organic salts formed from the reaction of carboxylic acids in the bitumen with carbonates in the aggregates. High molecular weight magnesium and calcium salts are relatively hydrophobic and not very soluble in water. Sodium salts, being more soluble, lead to stripping. Further, it was found in SHRP research that carboxylic acids in bitumen hydrogen bond very strongly with hydroxyl groups on siliceous aggregates, these being highly concentrated on the aggregate surface. However, this hydrogen bond is highly sensitive to disruption by water, thus accounting, at least in part, for the high moisture sensitivity to moisture damage of pavement mixtures containing siliceous aggregates. Conversion of carboxylic acids to insoluble salts (e.g., calcium salts) prior to use in pavement mixtures could prevent adsorption of water-sensitive free acids on the aggregate in the first place. When pavement containing surface active materials is wet and is subjected to mechanical action of traffic, it is predictable that the surface activity of the sodium carboxylates (soaps) in the bitumen will help scrub the oil (bitumen) away from the rock. . . . The practical, perhaps conservative, solution to the historical problem of stripping is to convert all acidic materials in asphalt to water- insensitive (non-surface active) calcium salts at the time of production. This would require lime treatment at the refinery. Some refineries do this today (SHRP bitumen AAG). The recommendation here is that conversion of acids to calcium salts be made a universal requirement. The process recommended here reduces moisture susceptibility of the whole asphalt rather than just at the interface. Lime treatment of the aggregate will be desirable. Yoon and Tarrer (1988) discuss the effect of water pH on stripping potential in asphalt mixtures with respect to antistripping additives. Their analysis showed that as the pH of the water increases, the adsorptive bonds between amine-type additives and aggregate surfaces are weakened. As a result, water can more easily displace asphalt from the aggregate surface. They point out that this is not the case with hydrated lime, where the resistance to stripping is independent of the pH of the contacting water. However, other research has shown that normally pHs as high as 10 will not dislodge amines from the aggregate, and pHs greater than 10 are very unusual. The effectiveness of the polyamine additives increases with curing time in studies by Yoon and Tarrer (1988). They found that by storing asphalt–aggregate mixtures for a few hours at 300°F, the effectiveness of some additives improved considerably even at a high pH value of contacting water. Yoon and Tarrer (1988) hypothesize that the reason for the improved performance with curing might be the development of a film of polymerized asphalt.

Little and Jones 65 Other Chemical Treatments Jamieson et al. (1995) describe three possible treatments to improve adhesion: addition of cations to the aggregate surface, addition of antistripping agents to the bitumen, and aggregate pretreatment with organosilanes. Jamieson et al. (1995) point to research that shows that enhanced bonding is associated with relatively large concentrations of iron, calcium, magnesium, and aluminum at the aggregate surface. Jamieson et al. describe that the principal role of antistripping agents is to trigger the dissociation of aggregations of bitumen components, thereby increasing the availability of bitumen functional groups for active sites on the aggregate surface. Bonding energy measurements indicate that the effectiveness of aggregate pretreatment with modifiers is dependent on aggregate type, probably because antistrip agents are usually amines with relatively similar properties, whereas aggregates vary widely (Jamieson et al. 1995). Organosilane pretreatment of aggregate increases the number of polar adsorption sites on the aggregate surface (DiVito and Morris 1982; Graf 1986). Research during SHRP ranked the overall performance of organosilane treatments as a function of hydrophobic bonding enhancements and determined the order of ranking to be amino > hydrocarbon > thiol. DUSTY AND DIRTY AGGREGATES General Mechanisms of Bond Disruption with Dirty or Dusty Aggregates Dusty aggregates may generally be referred to as aggregates coated with materials smaller than 75 µm. This may cause a problem in developing an acceptable bond between fine and coarse aggregate because the asphalt binder tends to coat the dust and not the aggregate, leading to a greater probability for bond interruption and hence displacement. Dirty aggregates normally refer to aggregates coated with clay mineral fines. While clay- sized materials are soil particles smaller than 2 µm, true clays are not only very small particles but also have a unique mineralogy and morphology. Clay minerals are made up of alternating layers of silca and alumina, which comprise particles that have a great affinity to adsorb water. This is why clay fines are plastic in nature and have a large plasticity index [range of moisture content between the plastic limit (where the soil acts as a plastic semisolid) and the liquid limit (where the soil acts as a liquid)]. The presence of clay particles on the aggregate surface is similar to that of dust. The asphalt bond with the fine and coarse aggregate is disrupted by the presence of the dust of clay. In fact, the situation is worse with clay fines because these particles have a tendency to swell when they take on water, and this swelling mechanism can break or disrupt an existing bond with asphalt. Furthermore, clay is more active than other soil particles. This can lead to other complex reactions between asphalt, water, and the clays, including emulsification. Clay particles adsorb cations because of their strong negative surface charge and their enormous specific surface area. The amount and nature of the cations adsorbed can affect bond interactions and emulsification potential. In summary, aggregates coated with dust or clay disrupt the asphalt–aggregate bond and can also lead to more complex reactions among water, asphalt, and aggregate, such as emulsification. Kandhal et al. (1998) evaluated aggregate tests to assess the potential for aggregate fines to cause stripping in asphalt mixtures. They considered the sand equivalent test, the plasticity index test, and the methylene blue test. They evaluated a set of 10 asphalt mixtures using a common coarse limestone aggregate but with different fine aggregates. They used two validation tests to assess moisture damage: American Association of State Highway and Transportation

66 Moisture Sensitivity of Asphalt Pavements: A National Seminar Officials T283 and the Hamburg wheel-tracking test. After a careful statistical analysis of results, they found that the methylene blue test did the best job of identifying moisture sensitivity of the mixtures. Modification of Dusty and Dirty Aggregates to Improve Asphalt–Aggregate Interaction Hydrated lime has been used to treat dusty and dirty aggregates. The mechanism is partially because hydrated lime reacts with clay to change its properties. Two basic mechanisms are involved: cation exchange or molecular crowding of calcium hydroxide molecules at the surface of the clay and pozzolanic reaction. Cation exchange or calcium hydroxide crowding provides an abundance of divalent calcium ions, which, because of their high concentration and divalent nature, replace the normally available cations in the clay environment. This leads to a substantial reduction in clay plasticity (Little et al. 1995) and causes clay colloids to flocculate into larger aggregates (Little et al. 1995). However, the most important reaction is the pozzolanic reaction, where caustic calcium hydroxide raises the pH of the lime-water-clay system to more than 12. At this high pH, clay minerals are attacked and the silica and alumina solubilize. Soluble silica and alumina then combine with free calcium cations to form calcium silicate and calcium aluminate hydrates, further reducing plasticity, stabilizing the clay, and forming more well-cemented agglomerates (Little et al. 1995). One might expect that the lime-modified clay coatings will “peal” from the aggregates and no longer remain as coatings but as “cemented” small aggregates of flocculated clay that can be separately coated with asphalt. CONCLUSIONS Although several separate mechanisms have been identified to explain the process of moisture damage in asphalt pavements, it is more likely that most asphalt pavements suffer moisture damage as a result of a synergy of several processes. From a chemical standpoint, the literature is clear that neither asphalt nor aggregate has a net charge, but components of both have nonuniform charge distributions, and both behave as if they have charges that attract the opposite charge of the other material. Researchers point out that certain polar asphalt compounds develop more tenacious and moisture-resistant bonds with the aggregate surface than others and that the development of the more tenacious and long-lasting bonds can be promoted by treatment of the asphalt mixtures with additives. The most durable bonds appear to be formed by interaction of phenolic groups and nitrogen bases from the bitumen. These form insoluble salts. While sulfoxides and carboxylic acids have a greater affinity for the aggregate surfaces, they are most susceptible to dissolution on water. The asphalt–aggregate bond is affected by aggregate mineralogy, adsorbed cations on the aggregate surface, and the surface texture and porosity. Favorable chemical bonding between asphalt and aggregate alone will not optimize the adhesive bond and minimize moisture damage. The bond is part physical, and, therefore, the asphalt must be able to wet and permeate the aggregate surface. This process is dependent on asphalt rheology at mixing temperatures and the nature of the aggregate surface, pore size, pore shape, and aggregate mineralogy. To complicate matters somewhat, the ability to bond asphalt to aggregate is dynamic and changes with time. This is largely affected by the shift in pH at the aggregate–water interface, which can be triggered by dissociation of aggregate minerals near the surface or by the nature of the pore water (cation type and concentration).

Little and Jones 67 Moisture damage is certainly not limited to adhesive failure, but weakening of the cohesive strength of the mastic due to moisture infiltration is equally important. Recent research has shown that water can diffuse into asphalt of mastics and that each can hold an appreciable amount of water. Research over many years has clearly shown that this water can weaken the asphalt mixture, making it more susceptible to damage. Thus the logical view is that the deleterious effects of moisture on the adhesive and cohesive properties, both of which influence asphalt mixture performance, must be considered. In fact, recent work at Texas A&M University points out that the propensity for either adhesive or cohesive failure in an asphalt mixture is dependent on the thickness of mastic cover. Since the distribution varies considerably within the mixture, the statistical distribution will determine which mechanism controls. ACKNOWLEDGMENTS The authors thank Sundaram Logaraj for his careful review and comments, which have helped maintain technical accuracy of this document. The authors also thank Claine Petersen, Ray Robertson, and Ken Thomas for their meaningful technical explanations and input. RESOURCES Asphalt Institute. 1981. Cause and Prevention of Stripping in Asphalt Pavements. Educational Series No. 10, College Park, Md. Bhairampally, R. K., R. L. Lytton, and D. N. Little. 2000. Numerical and Graphical Method to Assess Permanent Deformation Potential for Repeated Compressive Loading of Asphalt Mixtures. In Transportation Research Record: Journal of the Transportation Research Board, No. 1723, TRB, National Research Council, Washington, D.C., pp. 150–158. Castan, M. 1968. Rising of Binder to the Surface of an Open-Graded Bituminous Mix. Bulletin de liaison des laboratoires routiers, No. 33, pp. 77–84. Cheng, D. Z., D. N. Little, R. L. Lytton, and J. C. Holste. 2001. Surface Free Energy Measurement of Aggregates and Its Application on Adhesion and Moisture Damage of Asphalt–Aggregate System. Proc., 9th International Center for Aggregate Research Symposium, Austin, Tex. Cheng, D., D. N. Little, R. L. Lytton, and J. C. Holste. 2002. Surface Energy Measurement of Asphalt and Its Application to Predicting Fatigue and Healing in Asphalt Mixtures. In Transportation Research Record: Journal of the Transportation Research Board, No. 1810, TRB, National Research Council, Washington, D.C., pp. 44–53. Curtis, C. W. 1992. Fundamental Properties on Asphalt Aggregate Interactions Adhesion and Adsorption. Final Report on Contract A-003B. Strategic Highway Research Program, National Research Council, Washington, D.C. Curtis, C. W., R. L. Lytton, and C. J. Brannan. 1992. Influence of Aggregate Chemistry on the Adsorption and Desorption of Asphalt. In Transportation Research Record 1362, TRB, National Research Council, Washington, D.C., pp. 1–9. DiVito, J. A., and G. R. Morris. 1982. Silane Pretreatment of Mineral Aggregate to Prevent Stripping in Flexible Pavements. In Transportation Research Record 843, TRB, National Research Council, Washington, D.C., pp. 104–111. Fehsendfeld, F. M., and A. J. Kriech. Undated. The Effect of Plant Design Changes on Hot Mix Asphalt. Heritage Research Group. Fromm, H. J. 1974. The Mechanisms of Asphalt Stripping from Aggregate Surfaces. Proc., Association of Asphalt Paving Technologists, Vol. 43, pp. 191–223.

68 Moisture Sensitivity of Asphalt Pavements: A National Seminar Graf, P. E. 1986. Factors Affecting Moisture Susceptibility of Asphalt Concrete Mixes. Proc., Association of Asphalt Paving Technologists, Vol. 55, p. 175. Gzemski, F. C., D. W. McGlashan, and W. L. Dolch. 1968. Highway Research Circular 78: Thermodynamic Aspects of the Stripping Problem. HRB, National Research Council, Washington, D.C. Hicks, R. G. 1991. NCHRP Synthesis of Highway Practice 175: Moisture Damage in Asphalt Concrete. TRB, National Research Council, Washington, D.C. Hindermann, W. L. 1968. The Swing to Full-Depth: The Case for Laying Asphalt on the Raw Subgrade. Information Series No. 146, Asphalt Institute, College Park, Md. Hubbard, P. 1938. Adhesion of Asphalt to Aggregates in Presence of Water. Highway Research Board Proceedings, Vol. 18, Part 1, pp. 238—249. Hughes, R. I., et al. 1960. Adhesion in Bitumen Macadam. Journal of Applied Chemistry, Vol. 10. Jamieson, I. L., J. S. Moulthrop, and D. R. Jones. 1995. SHRP Results on Binder—Aggregate Adhesion and Resistance to Stripping. Asphalt Yearbook 1995, Institute of Asphalt Technology, United Kingdom. Jeon, W. Y., C. W. Curtis, and B. M. Kiggundu. 1988. Adsorption Behavior of Asphalt Functionalities on Dry and Moist Silica. Submitted to TRB. Jones, D. R. 1992. An Asphalt Primer: Understand How the Origin and Composition of Paving- Grade Asphalt Cements Affect Their Performance. SHRP Technical Memorandum No. 4. Strategic Highway Research Program, National Research Council, Washington, D.C. Kandhal, P. S., C. Y. Lynn, and F. Parker. 1998. Test for Plastic Fines in Aggregates Related to Stripping in Asphalt Paving Mixtures. Report 98-3. National Center for Asphalt Technology. Kennedy, T. W., F. L. Roberts, and K. W. Lee. 1984. Evaluating Moisture Susceptibility of Asphalt Mixtures Using the Texas Boiling Test. In Transportation Research Record 968, TRB, National Research Council, Washington, D.C., pp. 45—54. Kiggundu, B. M. 1986. Effects of Submergence in Distilled Water on the Surface Coloration of Asphalt. Unpublished data, NMERI. Kiggundu, B. M., and F. L. Roberts. 1988. The Success/Failure of Methods Used to Predict the Stripping Potential in the Performance of Bituminous Pavement Mixtures. Submitted to TRB. Kim, Y. R., H. J. Lee, and D. N. Little. 1997. Fatigue Characterization of Asphalt Concrete Using Viscoelasticity and Continuum Damage Mechanics. Journal of Association of Asphalt Paving Technologists, Vol. 66, pp. 520—569. Kim, Y.-R., D. N. Little, and R. L. Lytton. 2002. Fatigue and Healing Characterization of Asphalt Mixtures. Journal of Materials in Civil Engineering, American Society of Civil Engineers Labib, M. 1991. End of Phase II Report: Evaluation of Donor-Acceptor Properties of Asphalt and Aggregate Materials and Relationship to Asphalt Composite Performance. Strategic Highway Research Program, National Research Council, Washington, D.C. Lesueur, D., and D. N. Little. 1999. Effect of Hydrated Lime on Rheology, Fracture, and Aging of Bitumen. In Transportation Research Record: Journal of the Transportation Research Board, No. 1661, TRB, National Research Council, Washington, D.C., pp. 93—105. Subbases with Lime. Kendall-Hunt Publishing Co., New York. Little, D. N., J. W. Button, and C. Estakhri. 1995. Handbook for Stabilization of Bases and

Little and Jones 69 Little, D. N., R. L. Lytton, and D. Williams. 1998. Propagation and Healing of Microcracks in Asphalt Concrete and Their Contributions to Fatigue. In Asphalt Science and Technology (A. Usmani, ed.), Marcel Dekker, Inc., New York, pp. 149–195. Little, D. N., R. L. Lytton, D. Williams, and R. Y. Kim. 1999. Analysis of the Mechanism of Microdamage Healing Based on the Application of Micromechanics First Principles of Fracture and Healing. Journal of Association of Asphalt Paving Technologists, Vol. 68, pp. 501–542. Lytton, R. L., J. Uzan, E. G. Fernando, R. Roque, D. Hiltmen, and S. Stoffels. 1993. Development and Validation of Performance Prediction Models and Specifications for Asphalt Binders and Paving Mixes. SHRP Report A-357. Strategic Highway Research Program, National Research Council, Washington, D.C. Mack, C. 1964. Bituminous Materials. Vol. 1 (A. Holberg, ed.), Interscience Publishers, New York. Majidzadeh, K., and F. N. Brovold. 1968. Special Report 98: State of the Art: Effect of Water on Bitumen–Aggregate Mixtures. HRB, National Research Council, Washington, D.C. Maupin, G. W. 1982. The Use of Antistripping Additives in Virginia. Submitted at the 51st Association of Asphalt Paving Technologists, Kansas City, Mo. Morgan, P., and P. Mulder. 1995. The Shell Bitumen Industrial Handbook, p. 120. Payatakis, A. C. 1975. Surface Chemistry Applied to Solid–Liquid Separations. In Theory and Practice of Solid–Liquid Separation. Petersen, J. C. 1988. Lime-Treated Pavements Offer Increased Durability. Roads & Bridges Magazine, Jan. Petersen, J. C., H. Plancher, and P. M. Harnsberger. 1987. Lime Treatment of Asphalt—Final Report. National Lime Association. Plancher, H., S. Dorrence, and J. C. Petersen. 1977. Identification of Chemical Types in Asphalts Strongly Absorbed at the Asphalt–Aggregate Interface and Their Relative Displacement by Water. Proc., Association of Asphalt Paving Technologists, Vol. 46, pp. 151–175. Podoll, R. T., C. H. Becker, and K. C. Irwin. 1991. Phase II Progress Report: Surface Analysis by Laser Ionization of the Asphalt–Aggregate Bond. Strategic Highway Research Program, National Research Council, Washington, D.C. Porubszky, I., M. Csizmadia, E. Szebenyi, O. K. Dobozy, and M. Simon. 1969. Bitumen Adhesion to Stones. Chimie, physique et applications pratiques des agents de surface: compte-rendus du 5ème Congrès International de la Détergence, Barcelona, Spain, Sept. 9–13, Vol. 2, Part 2, pp. 713–725. Rice, J. M. 1958. Relationship of Aggregate Characteristics to the Effect of Water on Bituminous Paving Mixtures. Symposium on Effect of Water on Bituminous Paving Mixtures, ASTM STP 240, pp. 17–34. Robertson, R. E. 2000. Transportation Research Circular 499: Chemical Properties of Asphalts and Their Effects on Pavement Performance. TRB, National Research Council, Washington, D.C. Sanderson, F. C. 1952. Methylchlorosilanes as Antistripping Agents. Highway Research Board Proceedings, Vol. 31, pp. 288–300. Schmidt, R. J., and P. E. Graf. 1972. The Effect of Water on the Resilient Modulus of Asphalt Treated Mixes. Proc., Association of Asphalt Paving Technologists, Vol. 41, pp. 118– 162.

70 Moisture Sensitivity of Asphalt Pavements: A National Seminar Scott, J. A. N. 1978. Adhesion and Disbonding Mechanisms of Asphalt Used in Highway Construction and Maintenance. Proc., Association of Asphalt Paving Technologists, Vol. 47, pp. 19–48. Si, Z., D. N. Little, and R. L. Lytton. 2002. Evaluation of Fatigue Healing Effect of Asphalt Concrete by Pseudostiffness. In Transportation Research Record: Journal of the Transportation Research Board, No. 1789, TRB, National Research Council, Washington, D.C., pp. 73–79. Tarrer, A. R. 1996. Use of Hydrated Lime to Reduce Hardening and Stripping in Asphalt Mixtures. Presented at the 4th Annual International Center for Aggregate Research Symposium, Atlanta, Ga. Tarrer, A. R., and V. Wagh. 1991. The Effect of the Physical and Chemical Characteristics of the Aggregate on Bonding. Strategic Highway Research Program, National Research Council, Washington, D.C. Taylor, M. A., and N. P. Khosla. 1983. Stripping of Asphalt Pavements: State of the Art. In Transportation Research Record 911, TRB, National Research Council, Washington, D.C., pp. 150–158. Terrel, R. L., and S. Al-Swailmi. 1994. Water Sensitivity of Asphalt–Aggregate Mixes: Test Selection. SHRP Report A-403. Strategic Highway Research Program, National Research Council, Washington, D.C. Terrel, R. L., and J. W. Shute. 1989. Summary Report on Water Sensitivity. SHRP-A/IR-89-003. Strategic Highway Research Program, National Research Council, Washington, D.C. Thelen, E. 1958. Surface Energy and Adhesion Properties in Asphalt–Aggregate Systems. Bulletin 192, HRB, National Research Council, Washington, D.C., pp. 63–74. Thomas, K. 2002. Quarterly Technical Report, Western Research Institute. FHWA Contract No. DTFH61-99C-00022. Laramie, Wyo., pp. 25–62. Titova, T. I., L. S. Kosheleva, and S. P. Zhdanov. 1987. IR Study of Hydroxylated Silica. Langmuir, Vol. 3, No. 6, pp. 960–967. Tseng, K. H., and R. L. Lytton. 1987. Prediction of Permanent Deformation in Flexible Pavement Materials. Implication of Aggregates in Design, Construction, and Performance of Flexible Pavements, ASTM STP 1016 (H. G. Schreuders and C. R. Marek, eds.), American Society for Testing and Materials, Philadelphia, Pa. Western Research Institute. 2002. Fundamental Properties of Asphalts and Modified Asphalts. Task 11-5, FHWA Quarterly Technical Report, Contract No. DTFH61-99C-00022, Nov. Williams, D., D. N. Little, R. L. Lytton, Y. R. Kim, and Y. Kim. 1998. Fundamental Properties of Asphalts and Modified Asphalts. Task K, FHWA Final Report, Volume 2, Contract No. DTFH61-92C-00170. Yoon, H. J. 1987. Interface Phenomenon and Surfactants in Asphalt Paving Materials. Dissertation, Auburn University. Yoon, H. H., and A. R. Tarrer. 1988. Effect of Aggregate Properties on Stripping. In Transportation Research Record 1171, TRB, National Research Council, Washington, D.C., pp. 37–43.