Transportation Research Record 999 13

Chemical Composition of Asphalt as Related to Asphalt Durability: State of the Art

J. CLAINE PETERSEN

ABSTRACT For the purposes of this review, a durable as­ phalt is defined as one that (a) possesses the phys­ ical properties necessary to produce the desired The literature on asphalt chemical composi­ initial product performance properties and (b) is tion and asphalt durability has been re­ resistant to change in physical properties during viewed and interpreted relative to the cur­ long-term in-service environmental aging. Although rent state of the art. Two major chemical design and construction variables are major factors factors affecting asphalt durability are the in pavement durability, more durable asphalts will compatibility of the interacting components produce more durable pavements. of asphalt and the resistance of the asphalt The importance of chemical composition to asphalt to change from oxidative aging. Histori­ durability, although not well understood, cannot be cally, studies of the chemical components of disputed. Durability is determined by the physical asphalt have been facilitated by separation properties of the asphalt, which in turn are deter­ of asphalt into component fractions, some­ mined directly by chemical composition. An under­ times called generic fractions~ however, standing of the chemical factors affecting physical these fractions are still complex mixtures properties is thus fundamental to an understanding the composition of which can vary signifi­ of the factors that control asphalt durability. cantly among asphalts of different sources. The purpose of this paper is to examine the The reaction of asphalt with atmospheric literature dealing with the chemical composition of oxygen is a major factor leading to the asphalt and changes in composition during environ­ hardening and embrittlement of asphalt. The mental aging that affect durability-related prop­ hardening phenomenon is primarily a result erties. Both past and recent research important to of the formation in asphalt of polar oxygen­ the state of the art will be considered. Because of containing functional groups that increase the extreme breadth of the subject and the volumi­ asphalt consistency through strong molecular nous literature related to durability, a complete interaction forces. The identification and bibliography will not be attempted, but sufficient characterization of the chemical functional references will be cited to allow the serious re­ types normally present in asphalt or formed searcher to find additional literature. The author's on oxidative aging that influence molecular approach to the fundamental chemical factors affect­ interactions afford a fundamental approach ing asphalt properties and durability will also be to relating asphalt composition with asphalt presented. properties and thus the performance of both asphalts and asphalt-aggregate mixtures. In addition to the polar chemical functional PRELIMINARY CONSIDERATION OF FACTORS AFFECTING groups formed on oxidation, asphalt prop­ ASPHALT DURABILITY erties can also be significantly altered by molecular structuring, sometimes called To provide a background for the documented discus­ steric hardening. This potentially revers­ sions that follow, the major composition-related ible phenomenon, although highly elusive and factors affecting durability are briefly outlined. difficult to quantify in asphalt pavement The most important aspect of a durable asphalt, as­ mixtures, may also be a major factor con­ suming that it meets initial performance require­ tributing to pavement embrittlement. ments, is its resistance to change while in service. The dominant physical change leading to reduced as­ phalt durability is a change in flow properties re­ lated to excessive stiffening or hardening of the asphalt. Three fundamental composition-related fac­ Differences in the quality of asphalts from differ­ tors govern the changes that could cause hardening ent sources (different composition) and relation­ of asphalts in pavements: ships between composition and performance properties have long been recognized, as shown by the many pub­ lications on the subject, a few of which are cited 1. Loss of the oily components of asphalt by (l-13). Asphalts meeting the same specifications volatility or absorption by porous aggregates, often produce pavements with widely differing per­ 2. Changes in the chemical composition of as­ formance and serviceability. Admittedly such factors phalt molecules from reaction with atmospheric oxy­ as aggregate characteristics, design, construction gen, and variables, and environment play major roles in de­ 3. Molecular structuring that produces thixo­ termining pavement performance and often overshadow tropic effects (steric hardening). the contribution to performance made by var iabil­ ities in asphalt cement quality. However, such stud­ With current specifications and construction prac­ ies as the well-known Zaca-Wigmore Experimental Road tices, volatility loss is probably not a significant Test (4,9), in which construction variables were contributor to pavement hardening. Reaction with carefully- controlled and asphalt source was inten­ atmospheric oxygen is probably the major and best tionally varied, clearly demonstrate the importance understood cause. Molecular structuring, although of asphalt chemical composition in pavement dura­ elusive and difficult to quantify, may also be a bility. significant contributor. 14 Transportation Research Record 999

Irreversible adsorption of polar asphalt compo­ omatic ring systems has been estimated by correla­ nents by mineral aggregate surfaces, although not a tions based on carbon-hydrogen analyses and den­ factor that might be expected to harden asphalt, sities (~,17) and directly by nuclear magnetic will produce compositional changes in the asphalt resonance (NMR) (18) • Concentrations of aromatic that may also significantly affect asphalt prop­ carbon determined by NMR typically run from 25 to 35 erties and aging characteristics. Finally, it is percent for petroleum asphalts. The aromatic carbon recognized that environmental factors affecting the is incorporated in condensed aromatic ring systems properties of the asphalt-aggregate bond, partic­ containing from 1 to possibly 10 rings per aromatic ularly water, can seriously affect the performance moiety (18). These ring systems may be associated and durability of asphalt pavements i however, even with saturated naphthenic (cycloalkyl) ring systems, though moisture-induced damage may be related to as­ and both the aromatic and naphthenic ring systems phalt composition and adsorption of asphalt compo­ may have attachments composed of a variety of types nents on aggregate surfaces, it is primarily an in­ of normal or branched hydrocarbon side chains. By terfacial phenomenon and beyond the scope of this NMR, carbon associated with naphthenic ring systems paper. typically ranges from 15 to 30 percent (18). Normal and branched chain hydrocarbons are present either as individual molecules or as the previously men­ CHEMICAL COMPOSITION OF ASPHALT tioned moieties associated with naphthenic or aro­ matic rings. The nonaromatic and nonnaphthenic car­ Elemental and Molecular Composition bon content of asphalt would typically range from 35 to 60 percent. It should be emphasized that examples Before an attempt is made to discuss relationships outside these ranges may be found and the variety of between chemical composition and asphalt properties possible combinations of molecular structures in an affecting durability, the chemical composition of asphalt is astronomically large and may vary widely asphalt will be reviewed. Asphalt is not composed of from one crude source to another. The hydrocarbon a single chemical species but is rather a complex molecular structures are further complicated by the mixture of organic molecules that vary widely in heteroatoms sulfur, nitrogen, and oxygen, which are composition from nonpolar saturated hydrocarbons to often present in sufficient combined amounts so highly polar, highly condensed aromatic ring sys­ that, on the average, one or more heteroatom(s) per tems. Elemental analyses of several representative molecule may be present. These may be incorporated petroleum asphalts are presented in Table 1. Al­ within the ring or nonring components or in more though asphalt molecules are composed predominantly discrete chemical functional groups attached to of carbon and hydrogen, most molecules contain one these components. or more of the so-called heteroatoms nitrogen, sul­ The heteroatoms, particularly nitrogen and oxy­ fur, and oxygen together with trace amounts of gen, and the aromatic ring systems contribute con­ metals, principally vanadium and nickel. As seen in siderable polarity or polarizability to the mole- Table 1, the heteroatoms, although a minor component cules that roduce the fl!!jor association for~e~s'------ompareu~ro-t.he-h ydrocarbon-mo±-e:ty , can vary---m---con­ affecting physical properties. This will be dis- centration over a wide range depending on the source cussed in more detail in a later section. of the asphalt. Because the heteroatoms often impart Because the number of molecules in asphalt with functionality and polarity to the molecules, their presence may make a disproportionately large contri­ different chemical structures and reactivities is bution to the differences in physical properties extremely large, chemists have not seriously con­ sidered attempts to separate and identify them. Con­ among asphalts from different sources. siderable progress, however, has been made in the study of asphalt composition by separation or char­ acterization of asphalt based on the reactivity or TABLE 1 Elemental Analyses of Representative Petroleum polarity or both of the various molecular types Asphalts (14) present. The molecules in asphalt can be con­ veniently separated or grouped into classes of molec­ Asphalt• ular types or fractions based on their chemical B-2959 B-3036 B-305 l B-3602 functionality. This separation and classification of Element (Mexican) (Ark.-La.) (Boscan) (Calif.) molecular types has been useful to provide simpler Carbon(%) 83.77 85.78 82.90 86.77 component fractions that permit further characteriza­ Hydrogen(%) 9.91 10.19 10.45 10.94 Nitrogen(%) 0.28 0.26 0.78 1.10 tion and has aided in determining how different Sulfui (%) 5.2 5 3.41 5.43 0.99 molecular types affect the physical and chemical Oxygenb (%) 0.77 0.36 0.29 0.20 properties of the whole asphalt. Vanadium (ppm) 180 7 l,380 4 Nickel (ppm) 22 0.4 109 6

a From study by Welborn et al. (J 5). Asphalt Composition as Defined by Fractionation bBy difference, A variety of procedures has been employed in at­ tempts to fractionate asphalt into less complex and Elemental analyses are average values and reveal more homogeneous fractions. Some are simple (19) and 1 i ttle information about how the atoms are incor­ others are more complex (20, 21) • Many are special­ porated in the molecules or what type of molecular ized and unique to a given-research endeavor in structures are present. Molecular type and structure which they were used to prepare fractions for fur­ information is necessary for a fundamental under­ ther characterization. Several, however, have found standing of how composition affects physical prop­ more general use to characterize and classify as­ erties and chemical reactivity. The molecular struc­ phalts. These separation schemes can be classified tures in asphalt will be discussed in more detail in into three general types based on the procedure used: subsequent sections: however, an overview is impor­ (a) partitioning with partial solvents carbons, saturated cyclic hydrocarbons (sometimes Figure 1. Partitioning with selective solvents called naphthenic hydrocarbons) , and possibly a (scheme 1-A) has not been widely used. Although it small amount of mono-ring aromatic hydrocarbons; avoids contact with reactive adsorbants and chemi­ however, those molecules containing ring systems are cals that might irreversibly adsorb or alter asphalt dominated by attached saturated hydrocarbon side components, the fractions obtained are usually not chains. Sulfur is often found incorporated in mole­ as distinctively different as with the other separa­ cules of the saturate fraction. Addition to the tion types. In general, one sequentially treats the chromatographic column of a more polar aromatic asphalt with increasingly polar solvents precipitat­ solvent such as benzene (now usually replaced by ing a series of fractions with decreasing polarity. toluene) , which competes for the polar sites on the Selective adsorption-desorption chromatography adsorbants, displaces the more weakly adsorbed as­ has probably found the widest use as a research phalt molecules. These molecules usually contain separation technique. This technique is typified by condensed nonaromatic and aromatic ring systems, and scheme 1-B in Figure 1. The asphaltenes are sepa­ in addition to sulfur, the heteoratoms oxygen and rated first based on their insolubility in a non­ nitrogen may also be part of the molecule. In scheme polar paraffinic solvent. This removes the most 1-B, this fraction has been called the naphthene­ polar and least soluble asphalt components and gen­ aromatic fraction. erally facilitates further separation. The remaining Finally, a highly polar solvent such as an alco­ petrolene (maltene) fraction, which is dissolved in hol is added to the benzene to displace the most the paraffinic solvent, is then adsorbed on a chro­ strongly adsorbed and most polar components of the matographic column and sequentially desorbed with petrolene fraction. The alcohol debonds these com­ solvents of increasing polarity. By proper selection ponents, which are held strongly to the adsorbant by of the adsorbant and desorption solvents, a series highly polar functional groups, and the benzene pro­ of fractions with increasing polarity is obtained. vides solubility for these components as they elute The fractions obtained will be described in some de­ from the column. This fraction, called polar aroma­ tail to provide background for discussion in later tics, contains more highly condensed aromatic ring sections of this paper. systems and functional groups containing hetero-

SCHEME 1-A. Partitioning with Partial Solvents SCHEME 1-B. Selective Adsorption-Desorption (Schweyer and Traxler, Rel. 23) (Corbett, Rel. 10)

ASPHALT ASPHALT I I n-Butanol n-Heptone

Precipitate l. Evaporate butonol Precipitate PETROLENES I 2. Dissolve in acetone, (MALTENESI chill to -23°C ASPHAL TICS ASP HAL TEN ES A dso rptio n-E lutio n Precipitate Solution Chromatography I on alumina n-Heptane PARAFFIN I CS CYCLICS SATURATES {elute)

NAPHTHENE Benzene AROMATICS (eluta) SCHEME 1-C. Chemical Precipitation (Rostler and Sternberg, Ref. 32) POLAR 1. Methanol-be nio no AROMATICS 2 . Trichloroelhylone ASPHALT (elute) 1 n-Pentone

P recipitote Solution

NITROGEN ASPHAL TEN ES BASES (85%)

H1so. 1st ACIDAFFINS 1-- --''-'acv,"""',"'"i --I 19

2 2nd ACIDAFFINS l-- H--'so,'-S_0.;;. • --1

PARAFFINS

FIGURE 1 Schematic diagrams illustrating three types of fractionation schemes used to separate asphalt into component fractions. 16 Transportation Research Record 999 atoms. It should be noted here, and will be impor­ Although any of the fractionation schemes dis­ tant l ater when durabili t y and oxidation s uscepti­ cussed separate asphalt into less complex and more bility are discussed, that all fractions contain, to homogeneous fractions, the generic fractions ob­ a greater or lesser degree, cyclic and noncyclic tained are in themselves still complex mixtures and saturated hydrocarbon fragments either as individual not well-defined chemical species. In all the sepa­ molecules in the saturate fraction or as moieties ration methods, equilibria are involved. This may be attached to the aromatic ring fragments in the more solid-liquid equilibria in the chromatographic sepa­ polar fractions. Sulfur is usually found in a large ration or liquid-liquid partitioning in the chemical percentage of the individual asphalt molecules, but precipitation method. The same generic fraction can in unoxidized asphalt, it is rather nonpolar and vary considerably in composition and properties from thus is found distributed among all the component one asphal t source t o a not her ; howeve r, the separa­ fractions. The so-called asphaltene fraction is tions a re o f t e n s uffic i entl y defin i tive to p r ovide chemically similar to the more resinous or polar us e f u l i nforma t ion in studies rela t ing c hemical components of the petrolene fraction (31). Although composition with physical properties. the. asphaltene fraction may contain small amounts of occluded or insoluble saturate-type material, the significant differentiating feature of this fraction is the pre1;>0nder a nce of molecules with highly con­ INTERACTIONS AMONG ASPHALT COMPONENT FRACTIONS AND densed planar a nd polar izable aromatic ring systems RELATIONSHIPS WITH DURABILITY together with a high concentration of polar, hetero­ atom-containing functional groups. Because of these As is shown from the results of component fractiona­ features, molecules of this fraction are strongly tion, a wide spectrum of molecular types is present attracted to each other, associate strongly, and are in asphalt. The most nonpolar or oily fracti on, in difficult to disperse even in polar solvents. the absence of the resinous components, is so unlike The last fractionation scheme to be considered is the asphaltene fraction that the two fractions are that of chemical precipitation (scheme 1-C in Figure not mutually solublei yet, these extremes in molec­ 1). This scheme and its modifications are baccd on n ular types must ~OP.xis~ in neat asphalt as a micro­ fractionation procedure developed by Rostler and s copically homogeneous mixture. This is made pos­ Sternberg ( 32) and later applied to asphalts ( 5, 6) • s i ble by the interaction of various components of It may notbe exactly co r r ect to call this fr ac­ asphalt with each other to form a balanced or com­ t ionation procedure a separation scheme because it patible system. I t i s t h is bala nce of component s is really a method of analysis. Some of the steps that g i ves a s phal t its uni que viscoel astic prop­ are destructive and the method does not require the erties that are so impor tan t to i ts a pplic ation as a recovery of the altered fractions for further analy ­ pavement binder . Lack of compatibility or balance, sis. After separation of the asphaltenes, the re­ as so111etimes man ifested by compone nt phase separa­ maining components are sequentially separated into tion, l eads t o undesirable properties . The r o l e o f fractions based on their reactivity with sulfuric the various component fractions in contributing to acid of increasing acid strengths (decreasing degree asphalt compone nt compatibility, and thus durabil­ of hydration). In practice, the sulfuric acid phase ity, will be considered next. is added to a nonpolar hydrocarbon solution of the It has long been recognized that asphalts exhibit components to be separated, thus forming a second properties that deviate from those of a true solu­ polar acid phase containing those components reac­ tion. The colloidal nature of asphalt was first tive with the sulfuric acid together with other com­ recognized by Nellenstyn (33,34), who considered ponents sufficiently polar or polar izable to parti­ asphalt a dispersion of micelle;-in an oily medium. tion toward the sulfuric acid phase. In the first The asphaltene fraction was early associated with step, an 85 percent sulfuric acid solution is used the dispersed or micelle phase (35). It was also to remove the most polar components, including most recogni ?.ed that the i nability of the resinous com­ of the basic and nonbasic nitrogen compounds and ponents to keep these highly associated asphaltene many of the oxygenated molecules. This fraction is components dispersed in the oily phase largely de­ called nitrogen bases. Because the nitrogen in as­ termined the gel or non-Newtonian flow characteris­ phalt is usually associated with condensed aromatic tics of the asphalt (36,37). Rostler (5) described ring systems, the so-called nitrogen base fraction the asphaltene fractioii"""" as the component of asphalt is quite aromatic. Concentrated sulfuric acid (98 primarily responsible for asphalt viscosity and percent) is used next to precipitate the first acid­ colloidal behavior because of its limited solubility affin fraction, which has been reported as contain­ in the remaining components. He concluded that the ing unsaturated hydrocarbons (5). The use of the asphaltenes are kept dispersed by the peptizing term •unsaturated" is unfortunate because the term ability of the nitrogen bases. The peptized asphal­ is usually reserved for the designation of double tenes are in turn solvated by the resinous acidaff in and triple carbon-carbon bonds in nonaromatic ring fractions and gelled by the paraf fins. Co rbett (38) structures (olefinic and acetylenic types) , which described the effects on physi cal properti es of the have not been found in significant amounts, if at four fractions separated by his procedure: the as­ all, in petroleum residues. The first acidaffin phaltenes function as solution thickeners; fluidity fraction is quite aromatic and low in nitrogen con­ is imparted by the saturate and naphthene aromatic tent. The most likely reaction leading to separation fractions, which plasticize the solid polar aromatic of this fraction is sulfonation or complex formation and asphaltene fractionsi the polar aromatic frac­ involving the aromatic ring systems. Sulfuric acid tion imparts ductility to the asphalti and the containing 30 percent SO], which is a powerful saturates and naphthene-aromatics in combination sulfonating and complexing agent, is used to precip­ with asphaltenes produce complex flow properties in itate those components with less reactive or polar­ the asphalt. In summary, he concluded that "each izable aromatic ring systems. Analysis of this frac­ fraction or combination of fractions perform sepa­ tion, called second acidaffins, indicates that it is rate functions in respect to physical properties, considerably less aromatic than the nitrogen bases and it is logical to assume that the overall phys­ or first acidaffins (5). The final remaining frac­ ical properties of one asphalt are thus dependent tion is called paraff:Gis. This fraction is the oily upon the combined effect of these fractions and the component of asphalt and is composed primarily of proportions in which they are present.• molecules embodying straight-chain, branched, and For purposes of comparison, the polar aromatic cyclic alkanes. fraction from the Corbett separation (10, scheme 1-B Petersen 17

in Figure 1) might be considered to contain many of hardens the asphalt, has been recognized for more the components found in the Rostler nitrogen base than 50 years, and the literature in this area is fraction plus possibly part of the first acidaffin voluminous. Atmospheric oxidation is the major fac­ fraction. The Corbett naphthene aromatic fraction tor responsible for the irreversible hardening of may be roughly compared with the Rostler second asphalts (~) and is the reason why pavement void acidaffin fraction plus some components found also content (which allows access to atmospheric oxygen) in the Rostler first acidaffin fraction. The satu­ correlates so strongly w1th asphalt pavement hard­ rate and paraffin fractions from the two schemes ening (_!!!,~). Hardening from loss of volatile might also be somewhat comparable. components, the physical factor that might affect It has been widely recognized (6,7,12,39-43) that the correlation of hardening and void content, is a proper balance of chemical componentS-iS-tleC;essary not considered a significant factor when asphalts in a durable asphalt. Not only may too much or too meeting current specifications are used. The poten­ little of a given generic type as defined by the tially volatile components would be part of the fractionation schemes be detrimental to the compati­ saturate fraction; Corbett and Merz showed that the bility of the system, but so also may variations in amount of this fraction remained virtually constant chemical composition within the same generic type during 18 years of service in the well-known Michi­ classification be detrimental. For example, the gan Road Test (50). Thus, in dealing with asphalt presence of waxes in the oily fraction, which tend durability a major factor that must be addressed is to crystallize and cause phase separation, can be change that takes place in asphalt composition from detrimental (39). Asphaltenes that are not properly oxidative aging. dispersed, either because of their inherent solu­ The change in the amounts of fractional compo­ bility properties or because of the solvent nents of asphalt generally seen on oxidative aging properties or dispersing power of the resinous is a movement of components from the more nonpolar components of the petrolenes, will have reduced to the more polar fractions. The saturates in the compatibility with the oily fraction and thus reduce Corbett analysis (2.2_) and the paraffins in the asphalt durability (5,42). Exudation of oil may oc­ Rostler analysis (6) show the least change on oxida­ cur and undesirable gel characteristics thus be tion. There is uiually some loss of the Rostler imparted. second acidaffins and a greater loss of the more Rostler and coworkers (6,43,44) showed that the reactive first acidaffins and these losses are off­ balance of the component fractions, as indicated by set by a significant increase in the asphaltene the ratio of the most reactive fractions (nitrogen fraction. Similarly, the Corbett naphthene aromatics bases plus first acidaffins) to the least reactive and the polar aromatics decrease as asphaltenes fractions (paraffins plus second acidaff ins) , was increase. important to the resistance of pellets of asphalt King and Corbett (2!_) using thin films at 150°C and Ottawa sand to abrasion loss in laboratory test­ and Knotnerus (11) using dilute toluene solutions ing. Although the Rostler fractionation scheme has showed that th;- saturate fraction is relatively been used by many materials laboratories and cor­ inert to reaction with oxygen as measured by oxygen relations with field performance have been attempted uptake. The naphthene aromatic (King and Corbett) (45,46), it has generally not been accepted as an and aromatic (Knotnerus) fractions showed slight and accurate predictor of field performance. Lack of no reactivity, respectively. However, the Corbett consideration of the asphaltene fraction, which polar aromatic fraction and the Knotnerus resin and contributes so significantly to flow properties, in asphaltene fractions were highly reactive with oxy­ the Rostler durability parameter may be unfortunate gen. Corbett's asphaltene fraction showed inter­ and a serious omission. In field tests conducted in mediate reactivity. Direct measurement of the forma­ California (46), the Heithaus parameter (P) (state tion of oxygen-containing functional groups by of peptization) , which is an attempt to measure the Petersen et al. (52) ranks the relative reactivity internal compatibility of an asphalt by evaluating with atmospheric oxygen of the saturate, aromatic, the peptizability of the asphaltenes and the dis­ polar aromatic, and asphaltene fractions as persing power of the petrolenes C1ll , was found to 1:7:32:40, respectively, for a Wilmington (Califor­ correlate better with pavement field hardening than nia) asphalt when the fractions were oxidized sepa­ the Rostler durability parameter. Using six asphalts rately at 130°C. However, evidence was found that in of widely differing composition, Traxler (i2_) found neat asphalt, components of the more polar fractions a correlation between his coefficient of dispersion may promote more oxidation of components of the less (resins plus cyclics divided by asphaltenes plus polar fractions than when they are oxidized sepa­ saturates) and the rate of hardening during labora­ rately. tory oxidative aging. Better-dispersed asphalts The asphaltene fraction has been considered by hardened more slowly. He further suggested that the some (5) to be chemically almost inert; however, the degree of dispersion of the asphalt components is foregoing data indicate that asphaltenes are quite inversely related to the complex (non-Newtonian) reactive with oxygen. This supports the author's flow properties of the asphalt and is indicative of criticism made earlier that it may not have been the asphalt's colloidal characteristics. justified to eliminate the asphaltene fraction from the Rostler durability ratio. The apparent con­ tradiction regarding the chemical reactivity of CHANGES IN CHEMICAL COMPOSITION ON AGING asphaltenes might be explained by the following observations made by the author. Isolated asphal­ The discussion thus far has addressed chemical com­ tenes are brittle solids and in this state at positional factors that determine the physical prop­ ambient temperatures are indeed quite unreactive erties of asphalts. Without question, a durable with atmospheric oxygen, probably because their asphalt must possess acceptable physical properties solid, highly structured state reduces molecular to produce a pavement with initially acceptable mobility, which in turn reduces reactivity with performance properties. The companion requirement of oxygen. However, when the asphaltenes are melted (as a durable asphalt is that these initial properties in the 130 and l50°C oxidations) or in solution in be resistant to change during environmental aging in solvents, their mobility is increased and thus so is field service. However, asphalt composition changes their apparent reactivity. One might assume that if with time when the asphalt is exposed as a thin film asphaltene components are well dispersed in neat to atmospheric oxygen in the pavement. That asphalt asphalt, they might also be chemically quite active. reacts with atmospheric oxygen, which stiffens or Their chemical structure, highly condensed ring 18 Transportation Research Record 999 systems with alkyl attachments, also suggests a TABLE 4 Carbonyl Functional Groups Formed in system sensitive to hydrocarbon-type oxidation. Wilmington Asphalt Fractions During Column More recent studies (14, 52-57) have yielded con­ Oxidation (52) siderable information ~the specific chemical 1 changes that take place in asphalt on oxidative Concentration (mol · liter- ) aging by reaction with atmospheric oxygen. The major oxygen-containing functional groups formed on aging Carboxylic are listed in Table 2 for four asphalts of different Fraction Ketone Anhydride Acid crude sources and aged under identical conditions Saturate 0.045 0.010 Trace (air, 130°C, thin film). The data (14) represent Aromatic 0.32 0.017 a averages for the four asphalts aged on~our differ­ Polar aromatic 1.48 0.088 ent aggregates and are for the same asphalts shown Asphaltene 1.8 2 0.080 NDb in Table 1. The level of oxidation has been judged Whole asphalt 1.02 0.052 0.007 to be equivalent to that typically found in asphalts 3 Some held.I lost on alumina column during component fractionation. after 5 years or more of pavement service (9) • That bNot dNcrmined. the chemical functionality developed during- labora­ tory oxidation at l30°C is similar to that developed during normal pavement aging at ambient temperatures is supported by data given in Table 3 (59). Reasons hydrocarbon types in asphalt and the general chemis­ for the lower levels of oxidation in some pavement try of hydrocarbon oxidation. The major reaction samples compared with laboratory oxidation are dis­ pathway of hydrocarbon air oxidation is the forma­ cussed elsewhere (59), but they relate to the inac­ t ion of carbonyl compounds via the hydroperoxide cessibility of some of the asphalt in the pavement intermediate (53,54). The most sensitive hydrocarbon to atmospheric oxygen. moiety expected to be present in asphalt is that associated with the carbon atom adjacent to an aro­ matic ring system, commonly called a benzylic car­ TABLE 2 Chemical Functional Groups Formed in Aaphalts bon. The hydrogen attached to the carbon in this During Oxidative Aging (14) position is relatively easy to displace, forming a free radical on the carbon. Branching in the at­ Concentration (mol · liter-I) Average tached hydrocarbon chains also increases the sensi­ Carboxylic Hardening tivity of the asphalt to oxidation. Asphalt Ketone Anhydride Acid' Sulfoxide Indexb A simplified, generalized scheme proposed by the B-2959 0.50 0.014 0.008 0.30 38.0 author for the oxidation of the hydrocarbon moieties B-3036 0.55 0.015 0.005 0.29 27.0 in asphalt is proposed in Figure 2, which shows B-3051 0.58 0.020 0.009 0.29 132.0 ketones as the major functional group formed, con­ 30.0 B-3602 0.77 0.043 0.005 0.18 sistent with the data in Tables 2 and 3. In the Note: Column oxidation (58), 130°C, 24 hr, lS·micron film. reaction scheme presented, the symbol R may repre­ aNaturally occurring acids have been subtracted from reported value. bRatio of viscosity after oxidative aging lo viscosity before oxidative aging. sent either a hydrogen atom or an alkyl group. The reaction is initiated by the abstraction of a hydro­ gen atom attached to a benzylic carbon of an asphalt molecule (I) to form a free radical (II). The free TABLE 3 Comparison of Oxidation Products in Column-Oxidized radical reacts with atmospheric oxygen to form a and Pavement-Aged Samples peroxy radical (III). This in turn rapidly decom­ poses to form a ketone (IV) or, more likely, ab­ Concentration (mol · liter-I) stracts a hydrogen atom from the benzylic carbon of Column Oxidized' Pavement Agedb another asphalt molecule (V) to form a hydroperoxide Asphalt Ketone Anhydride Ketone Anhydride (VII) • The asphalt-free radical formed (VI) can 60 0.76 0.024 0.53 0.018 react with atmospheric oxygen to repeat the process. 25 0.70 0.025 0.53 0.022 The hydroperoxide is rather unstable and may decom­ 30 0.64 0.027 0.64 0.038 pose to form either a ketone (VIII) or an alkoxy 61 0.67 0.022 0.44 0.020 radical (IX). The alkoxy radical may rapidly decom­ 67 0.43 0.013 0.32 0.010 71 0.76 0.026 0.51 0.022 pose to form a ketone (X) • This mechanism is ad­ 72 0.82 0.033 0.68 0.029 mittedly oversimplified and minor amounts of other 73 0.49 0.013 0.35 0.011 oxidation products not shown are undoubtedly formed. 74 0.72 0.027 0.43 0.017 However, it adequately accounts for the major hydro­ aThin film oxidation, I 30°C, 24 hr (58). carbon oxidation product, ketones. Ke tones as the bRecovered From 11- to 13-year-old pavements (JS). major oxidation product in oxidized asphalt have been positively verified (~. Smaller amounts of anhydrides are formed (54) through what is believed Data in Table 2 show that ketones and sulfoxides by the author to be analternate hydroperoxide de­ are the major oxidation products formed during oxi­ composition route in certain asphalt molecules dative aging; anhydrides and carboxylic acids are having stereospecific ring systems associated with formed in smaller amounts. In some asphalts, the the oxidizable alkyl moieties. Under certain condi­ summed concentrations approach 1 mole per liter. If tions, the alkyl moiety may oxidize to the carbox­ a molecular weight of 1, 000 is assumed for an as­ ylic acid; however, only small amounts of carboxylic phalt molecule, on the average one functional group acids [and no measurable esters (54)] have been is formed for each asphalt molecule. Of course, not found in laboratory- or pavement-aged asphalts. It all molecules of asphalt have the same reactivity. appears that the oxidation reaction almost always Data in Table 4 show that ketones are formed in the stops at the ketone stage. highest concentrations in the asphaltene and polar Because the polar aromatics (or nitrogen bases aromatic fractions; lesser amounts are formed in the and first acidaffins) and asphaltene fractions are aromatic fraction, and considerably less in the known to contain the highest concentrations of aro­ saturate fraction. These data are consistent with matic ring systems, and thus benzylic carbons vi a the earlier-cited oxygen uptake experiments (11,..?..!l. the alkyl moieties attached, they have the highest The oxidation products formed are consistent and content of hydrocarbon types sensitive to air oxida­ in good agreement with what is known about the tion. It is then not surprising that the polar aro- Petersen 19

H I Initiation ~ rR - +RO· I H I ~C -R ~~

/OH 0 0 I 11 -i:-R + ~-C-R --~C-R + ROH I R /7H~~ ~ 0 0 I II · OH + ~C -R ~ C-R + R• ~~ :x: FIGURE 2 Suggested mechanism for the free·radical air oxidation of asphalt.

ma tics and asphaltenes (Table 4) showed the highest reactive. Thus, it is understandable that a correla­ ketone formation on oxidation. tion was found between the Rostler durability param­ The formation of sulfoxides, the second most eter (ratio of most reactive to least reactive com­ dominant oxidation product, has been shown to result ponents as recognized by Rostler) and abrasion loss from oxidation of organic sulfides that are part of in the pellet abrasion test ( 5, 6), which test is complex asphalt molecules ( 57) • These sulfides are sensitive to asphalt hardening- on oxidation. Al­ highly reactive. Sulfoxides are formed in asphalt at though the ratio may indicate the amounts of compo­ a much higher rate than ketones and their formation nents most reactive to oxygen (excluding the asphal­ often precedes significant ketone formation. This is tenes) relative to the amounts least reactive, it is probably because sulfides are hydroperoxide scaven­ not a precise measure of the compatibility of the gers and are converted to sulfoxides by the scaven­ sample, because the relative amounts of saturates ger reaction. and second acidaffins and the relative amounts of The significance of the polar oxygen-containing first acidaffins and nitrogen bases, whose sums make functional groups to physical properties will be up the denominator and numerator of the ratio, are discussed in detail in a later section of this not specified. paper. However, their influence on the hardening of Also, as previously mentioned, the asphaltenes, the asphalt is apparent from the aging indexes which have such a profound effect on asphalt com­ (ratio of viscosity after oxidation to viscosity patibility and viscosity, are not considered in the before oxidation) of the asphalts in Table 2, which ratio. These problems are compounded by differences range from 27 to 132. It is apparent that the rela­ in composition that may occur among similar generic­ tive amount of hardening is not directly related to type fractions from different asphalts. Consider the amount of oxidation when one asphalt source is again the large differences in the effects of simi­ compared with another. Note that the asphalts with lar amounts of oxidation products (from chemical the smallest and greatest aging indexes (Table 2) reactivity) on the hardening rate of two different both showed about the same chemical reactivity with asphalts, B 3036 and B 3051, shown in Table 2. atmospheric oxygen. This is because all asphalts do Initial compatibility, rate of formation of oxida­ not show the same sensitivity to the oxidation pro­ tion products, and response of the system to the ducts formed. Asphalts from different sources have oxidation products produced are all interdependent differing composition and thus their components variables and cannot be sufficiently defined by a interact differently with the oxidation products single numerical value. The asphalt system is much formed to increase viscosity. This varying sensitiv­ too complex for this. It is the considered opinion ity to oxidation products will be discussed later in of the author that although the Rostler parameter more detail. should show a general correlation with pavement per­ At this point it is instructional to consider the formance, it is not sufficiently precise to be used Rostler durability parameter with regard to the as an accurate predictor and if used must be con­ chemical information just presented. As previously sidered together with additional physiochemical data stated, this parameter is the ratio of the quantity further defining the composition of the asphalt. of nitrogen bases plus first acidaffins divided by the quantity of second acidaffins plus paraffins. The polar aromatic fraction, which should include MOLECULAR INTERACTIONS--A FUNDAMENTAL APPROACH TO the Rostler nitrogen bases and a good part of the CHEMICAL FACTORS AFFECTING ASPHALT DURABILITY first acidaffins, was shown to be the fraction most chemically reactive with oxygen after asphaltene In this section of the paper a fundamental approach separation. Also, the saturates and aromatics, which to asphalt chemical composition--physical property should account for much of the Rostler paraffin and relationships--will be addressed. The approach draws second acidaffin fractions, were shown to be least heavily on past chemical data and information from 20 Transportation Research Record 999 component analyses but is based on more recent re­ and breakdown of molecular structure via association search that recognizes the fundamental chemical forces are extremely rapid. In liquids the average factors that affect asphalt properties. Parts of lifetime of a given arrangement of molecule s, that this section have been summarized previously (~) . is, a s tructural unit, may be as short a s lo-13 Because the physical properties of asphalt are con­ sec a nd for water is estimated at lo-11 sec . trolled by the interactions of the molecules from To further illustrate the great influence of the which it is composed, an understanding of these electronegative oxygen atom on physical properties, interactions should provide the basis for under­ consider replacing the oxygen in water with sulfur standing its physical behavior and thus its dura­ to yield hydrogen sulfide, H2S. Because sulfur is bility. It is not necessary (and would be virtually less electronegative than oxygen, the strength of impossible) to know the exact structure of each the hydrogen bond is greatly reduced. As a result, molecule for a workable understanding. It should be hydrogen sulfide, even though a heavier molecule sufficient to identify or characterize th various than water, is not a liquid but rather a gas at room types of chemical or structural features of the temperature. Its boiling point is 162°C lower than asphalt molecules and how these interact with each that of water. The effects of intermolecular forces, other and their environment. as illustrated by the hydrogen bond, are basic to Many molecules of different composition will have understanding the effects of chemical composition on similar features, or functionality, that will pro­ asphalt properties because the intermolecular forces duce similar effects on physical properties. Func­ are the primary determinants of physical properties. tionality analysis has the advantage over component It should not be implied from the foregoing dis­ fractionation in that it can take into account the cussion that the hydrogen bond is the most important multiple functionality of asphalt molecules. Many interaction force in asphalt. Many other reversible asphalt molecules have several types of chemical interaction forces are important in a material so functionality on the same molecule, often of diverse chemically complex. These include a variety of types, which frustrate chemical fractionation pro­ dipole and induced dipole interactions. For simplic­ cedures because whole molecules must be moved into a ity in discussing molecular interactions in this given fraction. The molecular interaction approach paper, all molecular structural types in asphalt taken here is primarily that of the author; however, that exhibit these forces are considered as chemical many researchers, both past and present, have rec­ functionality. These functionalities include, but ognized the importance of chemical functionality on are not limited to, the more classical chemical asphalt properties. functional groups. Nonpolar hydrocarbon components in asphalt such as those dominant in the saturate fraction exhibit Fundamentals of Molecular Interactions and Effects of only weak interaction forces, which accounts for the Molecular Interactions on Flow Properties rather fluid character of this fraction. On the other hand, asphalt components containing highly Because the chemical and physical properties of an condensed ring systems and chemical functional asphalt are the sum total of the composition and groups containing oxygen and nitrogen atoms may be interactions of its individual molecules, it is highly polar or polarizable and thus interact instructive at this point to briefly review some strongly with each other. These strong interaction fundamentals of molecular interactions. Molecules forces largely account for the fact that asphal­ attract one another and interact through a variety tenes, even though they may not be significantly of secondary bonds or association forces. These different in molecular weight than the saturates association forces are generally one to two orders (61), are high melting solids. of magnitude weaker than the covalent chemical bond­ ~To illustrate the potential applicability of ing forces that hold the atoms together in the mole­ molecular interaction theory and molecular structur­ cule. The association forces are significantly dif­ ing to the physical properties of asphalt, it is ferent from covalent bonding forces in that they helpful to examine the effects of the chemical func­ form bonds that are generally reversible and are tionality of a series of model compounds on the usually in dynamic equilibrium. That is, they "make" physical properties of these compounds. A list of and "break" under forces induced by such factors as selected model compounds together with their struc­ temperature and external stress, and thus they ture, boiling point, and melting point are shown in largely determine the physical properties of the Table 5. These chemical functionalities also repre­ composite material. sent important types found in asphalt and thus the To illustrate the principles involved, the reasoning developed may have direct application. classic example of the hydrogen bonding of water First, consider the series of compounds n-heptane molecules is considered. A simplified schematic of through benzoic acid. All were chosen from the same this bonding is as follows: molecular weight range to minimize the effects of this variable and are listed in order of increasing (',-

TABLE 5 Effects of Molecular Interact.ions and Molecular explain why asphalts that have quite different flow Structuring on the Physical Properties of Model Compounds properties at low temperatures look more alike at higher temperatures. At lower temperatures, however, as the kinetic Boiling Melting energy of the system is lowered, the asphalt mole­ Compound Structure Point, °C Point, °C cules tend to associate or agglomerate into immobi- 1 ized entities with a more or less ordered or struc­ n-Heptane ~ 98 .4 -90.5 tured spatial arrangement. Although this ordered arrangement is influenced by polar functionality, it is also greatly influenced (as was the melting point Methycyclohexane @-cH 100.3 -126.4 of pure compounds) by the geometry of the molecules. 3 Thus, at low temperatures the effects of differences in chemical composition of asphalt play a more sig­ Toluene 110 .6 -95 nificant role in determining the complex low-temper­ OcH3 ature flow properties, for example, viscosity shear and temperature susceptibility. Data obtained on asphalt-based systems will now Cyclohexanone @= o 156 .7 -45 (lrz.) be considered. To illustrate the effect of different types of molecular interactions on viscosity, con­ sider data taken from an early paper by Griffen et Phenol OoH 182 41 al. (12) and abstracted as follows (to convert to visco~y in poise, multiply by 10):

Benzoic Acid Oc" o 249 122 Apparent ' OH Molecular Viscosity Fraction Weight (Pa•sec), 25°C Benzene 80 5.5 Saturate 500 10 0 Aromatic 500 1,000 Resin 500 100,000 Hex a hydro n op hthole ne 205 liquid 0iJ Griffen separated asphalt into component fractions and then determined molecular weight versus viscos­ Te!rahyd ronaphtha le ne ()i) 207 .2 -30 ity profiles on the fractions. The foregoing data are taken for components of each fraction having the same molecular weight of 500. Naphthalene 217.9 80 .2 The saturates had a viscosity of only 10 Pa•sec. 00"" They do not contain polar chemical functionality and CH3 molecular interactions are weak. Molecular interac­ 1- Methylnaphthalene 240 -22 t ions of increased intensity are exhibited by the co aromatic fraction the viscosity of which was 1,000 Pa•sec. Finally, the resins that contain polarizable roCH 3 condensed-ring systems and heteroatom functionality, 2 - Methylnaphthalene 245 35 .l and thus exhibit even more intense molecular interac­ tions, had a viscosity of 100,000 Pa•sec. Because the molecular weights of each fraction were the same, the differences in viscosity resulted primarily from group that forms strong hydrogen-bonded dimers. Note differences in the type and strength of molecular in­ its extremely high boiling point. teractions. The association forces among asphalt Next consider the properties of the series hexa­ molecules give asphalt many of the properties of hydronaphthalene through 2-methylnaphthalene in Table 5. As expected, the boiling point increases in high-molecular-weight polymers. a regular fashion with increasing aromaticity and I the introduction of the methyl group; however, the Major Chemi c al Fu nctional Group Types Affecting melting point shows no such correspondence. Note Asphalt Proper ties also in Table 5 how introducing a methyl group on benzene to form the higher-boiling toluene greatly As previously stat ed , a sphalt molecules contain reduced the melting point. The melting point, which hydrocarbon structural cons t ituents var ying from reflects the interaction forces between molecules saturated, paraffinlike chains ta highly condensed when in an ordered or structured configuration, is and polarizabl e aromatic ring systems. From the greatly influenced by the geometry of the molecule. previous discussion of model compounds , it is ap­ Interfering appendages on, or structural arrangements parent that the relative amou nts of the different of, the molecules that do not allow them to fit to­ structural or func tional moieties of each molecule gether easily in the necessary geometric pattern for determine how molecules interact with each other. effective interaction greatly reduce the melting Those molecules that are most alike are most com­ point. Thus, molecular shape dominates the low­ patibl e a nd vice versa. In fact , as stated earlier, temperature structuring properties. s ome asphalt components such as t he saturates and It is instructive to consider the molecular in­ asphaltenes are not mutually soluble when separated teraction effects just described with regard to the f rom the whole asphalt. It is the wide spectrum of flow properties of asphalts. The rather predictable molecular types in asphalt all i nteracti ng toget her effect of polar functionality on boiling point can that gives asphalt its unique properties and makes be related to the effects of polar functionality in it appear as a homogeneous material. However, on a asphalt on its flow properties at higher tempera­ molecular level, asphalt is undoubtedly heteroge­ tures in the Newtonian flow region. The polar inter­ neous, and a delicate balance exists among strongly actions between molecules dominate in influencing associated or agglomerated components and dispersing the flow behavior, and the effects of molecular or solubilizing components. It is this delicate shape on geometry are minimized. This reasoning may balance, or the lack of it, that affects the perfor- 22 Transportation Research Record 999 mance properties of asphalts, Incompatibility and Effec ts o n Aspha l t Performance of Pol ar Functional poor performance generally follow when one component Groupe Formed on Oxidation type unduly dominates at the expense of others. As discussed previously, although asphalt mole­ Asphalts vary considerably in their susceptibility cules are composed primarily of hydrocarbon constit­ to the effects of deteriorating oxidation. If during uents, heteroatoms such as nitrogen, sulfur, and aging the concentration of polar functional groups oxygen may also be present as part of the molecule, becomes sufficiently high to cause molecular im­ Heteroatom concentrations vary widely among as­ mobilization through increased intermolecular inter­ phalts. In many asphalts heteroatom concentrations action forces, that is, the asphalt molecules or are sufficient to average one or more heteroatoms micelles are not sufficiently mobile to flow past for every asphalt molecule, Oxygen, nitrogen, and one another under the stress applied, fracturing or some forms of sulfur may introduce high polarity cracking of the asphalt will result. into the molecule, and although only a minor compo­ Data in Figure 4 show a relationship between the nent, they can exhibit a controlling influence on amount of strongly interacting polar groups formed the molecular interactions that control asphalt flow in a series of asphalts during controlled laboratory properties. Thus chemical functionality containing oxidative aging and the resistance to failure from these atoms becomes a major consideration in under­ cracking in roads in which these asphalts were used. standing asphalt properties. The asphalts were from the California Zaca-Wigmore Much work has been done by the author and co­ Experimental Road Test mentioned previouely ( 4, 9) , workers in identifying polar, strongly associating in which construction variables were kept as -con­ functional groups in asphalts, either naturally stant as possible to evaluate the effect of asphalt present or formed on oxidation, and in characteriz­ composition (or source) on durability, In the ing their association forces (9,14,52-56,59,62-69). laboratory studies (9), these asphalts were coated Structural formulas of importairt'° chemical-func­ as thin films on inert fluorocarbon particles and tionality in asphalts are shown in Figure 3. Nitro­ were aged in a gas chromatographic (GC) column at gen, which occurs in concentrations of 0.2 to l 130°C for 24 hr by passing air through the column percent, is present in several forms from the (59). ('T'hiR pror.P.ilnre oxidizes the asphalt an amount equal to approximately 5 years or more in a road.) Following oxidative aging, a polar test compound, phenol, was passed through the GC column by using an 0 0)0 inert gas carrier, and its interaction with the O, H I'~ polar groups formed in the asphalt during oxidation Polynuclear Phenolic (1) 2-Quinolone was determined from the phenol retention time. AS aromatic (1) type (1) phenol passes through the column it is in equilibrium with the asphalt: those asphalts having a greater concentration of polar groups interact more strongly 0• with the phenol functionality, giving a larger phenol - s- - s- interaction coefficient. nN 0N As seen from Figure 4, an excellent correlation H' was found between the asphalt polarity as measured (1) Pyridinic (1) Sulfide (1) Sulloxide (2) Pyrrolic by the phenol interaction coefficient and the ser­ vice performance rating after 51 months of road service. Those asphalts that developed greater ,, o 0. amounts of strongly interacting polar groups during -c -c- aging failed sooner in the road. Figure 5 shows a ' O-H similar correlation developed on 20 roofing asphalts Anhydride (2) Carboxylic Ketone (2) the weatherability of which was determined by re­ acid (1, 2) sistance to cracking when aged in a carbon arc Weather-:-Ometer. Thus, strong evidence exists relat­ ing the development of polar functional groups in

llJ Nal uroll y o c c urr ing asphalts on aging with asphalt failure from embrit­ (2} Formed o n oiddative ag ing tlement and cracking. FIGURE 3 Chemical functionality in asphalt molecules normally present or formed on Effec ts o f Polar Asp halt Componen ts o n Viscosity a nd oxidative aging. Aging C h arac te ~istlcs

Data plotted in Figure 6 show the important influence slightly acidic pyrrole types to the more basic, of polar constituents (in this case, asphaltenes) on strongly interacting pyridine types (69). The nitro­ asphalt viscosity. In the study cited <.!.!>, four gen types naturally occurring in asphalt are not asphalts (all meeting the same specifications) were known to be significantly altered by oxidation. oxidatively aged in the laboratory both before and Sulfur, in concentrations from about l to 5 percent, after a hydrated lime treatment (the lime with ad­ is present primarily as sulfides. Many of these sorbed asphalt components--about 4 percent of the sulfides are readily oxidized to polar sulfoxides by asphalt--was separated from the asphalt during the atmospheric oxygen during normal aging (57). Pheno­ treatment) • The asphalts were aged by different meth­ lics are also usually present. Naturally occurring ods to achieve different levels of oxidation. Fol.low­ carboxylic acids and 2-quinolone-type functionality, ing aging, both asphaltene content and viscosity although occurring in relatively small amounts in were determined at 25°C. Each data point in the asphalts, are highly polar and associate strongly figure represents a separate level of oxidation. (64). As discussed previously, asphalts are sus­ Al though each asphalt had its own unique relation­ ceptible to oxidative aging by reaction with atmos­ ship, a good correlation existed for each asphalt pheric oxygen, which is a major factor contributing between asphalt viscosity and asphaltene content. to age hardening and embrittlement. The major oxy­ Several important points can be inferred from the gen-containing functional groups formed on oxidative data in Figure 6. First, the asphaltene fractions of aging are also included in Figure 3, the different asphalts are quite different from each Petersen 23

c ~ a :'!! 210 "0

..~ :> 0 ..c.. 190 N

~ .! ~ 170 c .!!- ~ 150 ~ 0 u c 130 ·~u ~ .! c 110 0 c ..c.. a.. 90 Foiled 7 6 5 4 Surface Performance Roting Alter 51 Months FIGURE4 Relationship between phenol interaction coefficient and pavement surface performance rating.

200 c ·~ 0 :'!! 0 " 180

..~ :> 0 ..c

N.. ..~ -=~ c .-!! :£.. 0 u c ·~ u ~ 16 .! 120 • .: I 0 c ..c.. a..

Accelerated Weathering Durability, days to failure FIGURE 5 Relationship between phenol interaction coefficient and durability of 20 roofing asphalts.

other in composition or the petrolene fractions have of four different aggregates. Note that except for widely different solubility power for the asphal­ asphalt B-3602, lime treatment reduced the hardening tenes or both. This is a necessary condition in index of the asphalts by more than 50 percent. As­ order for asphalts with widely varying asphaltene phaltene formation on aging was also reduced by contents to have the same viscosity. Second, asphal­ about 50 percent (!!) • Functional group analyses, tenes formed in asphalt on oxidative aging appear to however, showed that the oxidation reaction, as have a predictable effect on viscosity increase. measured by the formation of ketones, was reduced by The data in Table 6 give the effect of the lime only ab0ut 10 percent by lime treatment. What was treatment in reducing the hardening rate of the concluded was that lime removed carboxylic acids and asphalts when subjected to a laboratory GC column other highly polar functionality that would have oxidation procedure (14,59) during which the as­ otherwise interacted with oxidation products to phalts were supported as thin films on the surface increase asphalt viscosity. Separate studies showed 24 nnn Transportation R~s~arch Record """

2

107 - ~ 8 0 Q_ 6 >­ I­ (/j 8 "'> 6 10 8 UNTREATED ASPHALTS o fl o 0 6 LIME - TREATED ASPHALTS •••• ASPHALT 8-2959 0. 4 ASPHALT 8-3036 fl ' ASPHALT 8-3051 0 • 2 ASPHALT 8-3602 0 •

105 L-~L-~'--~"---~-'--~.'-~.J.-~--'---~-'-~--'---~-'-~-'-~--'-~-'-~-,'-~-'-~-' 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 ASPHALT ENE CONTENT , wt pcl FIGURE 6 Relationship between viscosity at 25°C and asphaltene content of untreated and lime-treated asphalts aged by different laboratory methods.

TABLE 6 Reduction of Hardening Rate o[ Asphalt by of molecular interactions through chemical modif ica­ Treatment with Hydrated Lime (14) tion or the addition of modifiers that can interact with polar chemical functionality in the asphalt and Hardening Index• alter its activity as suggested by the data on lime addition. The ability of surface-active materials Lime Reduction such as antistripping agents (often amines) to alter Sample Untreated Treated (%) asphalt viscosity, often to an extent not expected Asphaltb from simple additive effects, is familiar to many, B-2959 37 17 54 Because antistripping agents have polar chemical B-3036 27 10 63 functionality, they might be expected to affect the B-3051 132 35 73 dispersibility of asphaltenelike components in as­ B-3602 29 18 39 phalt by associating with polar functionality and Aggregate0 thus altering association and micelle structure Quartzite 57 22 61 within the asphalt. The ability of high-molecular­ Hol limestone 58 22 61 weight amines to interfere with molecular structure Riverton limestone 36 13 63 Granite 75 22 70 buildup and subsequent viscosity increase in asphalt cutbacks during storage was reported as early as n ll"tdo11 ing index ==viscosity after oxidative aging divided by viscosity 1951 by Hoiberg (71). b befo re oxidative aging. The effects of antistripping additives on alter­ Averaged for aggregates. e Averaged for asphalts. ing the flow properties and oxidative hardening rate of asphalts have recently been reported (~) • It has been suggested by J.N. Dybalski of Akzo Chemie America that cationic asphalt additives augment the that the introduction of carboxylic acid functional peptization of the asphal tene constituent in many groups into asphalt molecules greatly increased asphalts and thus reduce the asphalt hardening rate asphalt viscosity. as reported in a trade newsletter

solubility of a variety of components with widely chemical composition of the asphalt molecules. Sev­ differing solubility and solvent powers. Earlier eral factors account for its elusive nature and workers in the field considered compatibility in resistance to quantification. First, in asphalts at terms of the state of peptization of asphaltenes. ambient temperatures, structuring is a slow process Heithaus (£) stated that asphaltenes are the most that may go on for days and even yearsi second, it polar and aromatic components in asphalt and are is promoted by mineral aggregate surfaces (as in present in asphalts as rather concentrated solutions pavement mixtures) in an environment where it is ( 10 to 40 percent). To obtain maximum solvency for difficult to measure its effects; and third, most the polar, strongly associating species in asphalts, normal asphalt recovery methods using solvents or the polarity of the solvent (maltenes, now prefer­ heat or both destroy or reverse the structuring ably called petrolenes) must be matched to the mate­ process. Yet molecular structuring is largely re­ rials to be dispersed (asphaltenes). For example, sponsible for asphalt's unique physical properties. more polar asphaltenes require petrolenes with Without it, pavement mixtures would not set to pro­ greater solvent power to effectively dissolve or duce a nontender pavement with the desired struc­ disperse them. tural integrity, and too much structuring can pro­ A number of studies have addressed the dispersi­ duce pavements with poor low-temperature properties bility of asphaltenes and methods to measure dis­ and high shear susceptibility. persibility and compatibility (7,42,75-77). Altgelt Although the phenomenon of structuring in asphalt and Harle (76) clearly demonstrated-that selected is as yet little understood, it was recognized and asphaltenes ""from different sources had different received considerable study by early investigators effects on blend viscosity (different thickening (71,78-84) and is often called steric hardening power) when added to the same petrolene fraction. (82) ~it has been related to the thixotropic flow They further showed that petrolene fractions derived behavior commonly encountered in colloidal materials from asphalts of different sources had different (79). The early literature is extremely revealing solubility power for asphaltenes. To illustrate the and might well be read carefully by present-day principles they developed, consider what happens technologists, because little work has been done to when a highly polar asphaltene fraction having a build on or take advantage of the excellent past strong tendency to self-associate is added to a studies. petrolene fraction having relatively poor solvent Traxler and coworkers characterized molecular power for the asphaltenes. Intermolecular agglomera­ structuring (which they also called age hardening, tion will result, producing large, interacting, not to be confused with age hardening caused by viscosity-building networks. Conversely, when an oxidation) in a number of asphalts from a variety of asphaltene fraction is added to a petrolene fraction sources and related it to the degree of complex flow having relatively high solvent power for the asphal­ (sensitivity of measured viscosity to shear rate) tenes, molecular agglomerates are broken up or dis­ (78-!!!). Their experiments were carried out on bulk persed to form smaller associated species with less asphalt in the absence of significant oxidation. interassociation; thus the viscosity-building effect Selected data in Figure 7 and in Table 7 document of the asphaltenes is reduced. It must be concluded this nonoxidative age hardening. that chemical composition becomes an important fac­ In Figure 7, selected data on three paving as­ tor to be considered when asphalts (or crudes) are phalts show different hardening rates as a function blended from different sources or when asphalt com­ of asphalt source. Air-blown asphalts showed a ponents are blended, The principles described help greater rate of structure hardening than unblown explain why viscosities of asphalt mixtures or asphalts as illustrated by the air-blown Venezuelan blends often show irregularities when compared with asphalt in Figure 7. Note also that for the air­ the viscosities of the original asphalts. blown asphalt, the log viscosity versus log time It follows that asphalt compatibility can be plot was not linear as with the unblown asphalts but improved or worsened by blending and that the ini­ the rate of change increased with time. Traxler and tial properties of blended asphalts are not neces­ coworkers (78) devised an expression called the sarily additive in determining blend properties. asphalt aging index to quantitatively measure re­ Compatibility considerations are also important with versible hardening from molecular structuring. This regard to the aging characteristics of asphalt be­ index is the slope of the log viscosity versus log cause, as already described in detail, oxidative time plot. Asphalt aging indexes for eight unblown aging greatly alters the polarity of asphalt mole­ asphalts from different sources are shown as follows cules and therefore their interactions. The discus­ (~): sion thus far implies that a better knowledge of the chemical functionality responsible for the strong Asphalt Aging interaction forces in asphalt, and the ability to Asphalt Index manipulate these forces, has great practical value Californian 0.012 in altering the flow and thus the performance prop­ Californian 0.018 erties of asphalt. Application of compatibility Trinidad 0.026 principles to pavement recycling in which a recy­ Venezuelan 0.037 cling agent is added to restore useful properties Midcontinent 0.039 should also greatly benefit this rapidly emerging Mexican 0.051 practice. Venezuelan 0.063 Venezuelan 0.071

Importance of Molecular Structuring to the The structuring phenomenon was found reversible and Flow Properties of Asphalt structured asphalt could be brought back to near its initial viscosity by heating to a temperature above Molecular structuring in asphalts is probably one of its softening point or by continued mechanical work­ the least understood physiochemical phenomena af­ ing (80,84). fecting physical properties. Unlike oxidative aging, Theseearly researchers found a correlation be­ which produces irreversible changes in asphalt com­ tween the complex flow of asphalt (non-Newtonian position, moiecular structuring is a reversible behavior or viscosity lowering with increasing shear phenomenon that can produce large changes in the rates) and its tendency to exhibit molecular struc­ flow properties of asphalt without altering the turing. This is illustrated by data in Table 7, 26 Transportation Research Record 999

6.1

6.0

S.9

s.a Californian I Ref. 78) ~ 0 A. 5 .7

·;;;~ 0 5 .6 u.. > Trinidad (Ref. 78) DI 5 .5 ....0 ..

5."4

5.3

5 .2 1 10 100 1000 Time, hrs. (log scale) FIGURE 7 Hardening of asphalts from reversible molecular structuring at 25°C.

TABLE 7 Correlation Between Reversible A problem in the determination and quantification Molecular Structuring and Degree of Complex of structural hardening is that the structuring is Flow (81) destroyed during the solvent recovery of asphalts from aged pavements and therefore escapes detection Asphalt during measurement of recovered asphalt properties. Aging Com~lex Thus, the tendency for an asphalt to harden from Asphalt Index• Flow structuring, which may be a major contributor to loss of durability and pavement failure, is being Pressure· still 0.017 l.O Air-blown Californian 0.023 0.95 virtually ignored in pavement performance considera­ Air-blown Midcontinent 0.073 0.85 tions by present-day pavement technologists. Brown Air-blown Northeast Texas 0.183 0.50 ( 82, ~ , who studied reversible molecular structur­ ing (called by him "steric hardening"), in 1957, a Slope of log-log plot of viscosity versus time in hours (nonoxi­ dative hardanlnt). approximately two decades after the studies of bSlope or lof.· l o~ plot of shearing stress versus shear rate (mea­ Traxler and coworkers, noted that "this is not a new sure of non-Newtonian flow). discovery, but has had relatively 1i ttle emphasis.• In the last two decades, not much has changed in this regard but the need is still there. More recently, studies of molecular interactions which compare the complex flow (a value of 1 indi­ of asphalts and of asphalt-aggregate interactions cates Newtonian flow) for a series of asphalts with have been conducted at higher temperatures (above the asphalt aging index resulting from structural 130°C) by microcalorimetry (66,85,86). Reversible hardening. Those asphalts that showed more rapid molecular interactions were showntobe present at structuring possessed the greatest degree of non­ temperatures as high as 250°C. Interactions at these Newtonian flow behavior. high temperatures undoubtedly reflect the disasso­ The relationship between complex flow and struc­ ciation-association of the more stable micellar tural hardening has considerable potential signifi­ bodies in asphalt (87). Aggregate surfaces have been cance with regard to hardening of pavement mixtures. shown to promote molecular structuring and immobili­ It is known that as asphalts undergo oxidative zation of asphalt molecules in the vicinity of the aging , their flow behavior becomes more complex or aggregate surface, which should have considerable non-Newtonian. Recently, the author and coworkers, effect on bond properties and thus properties of while studying oxidation (SB), found that two as­ pavement mixtures (66,86,BB). A relationship between phalts that had been previously oxidatively aged molecular structuring and the setting characteris­ increased s ignificantly in measured viscosity be­ tics of pavements was proposed by Hveem and cowork­ cause of reversible molecular structuring after ers (~) and a cohesiograph test was proposed to standing for 2 years; the increase was from 2. S x measure the setting property. A good correlation was 10' to 1.2 x 10 5 and 1.1 x 10 5 to 3.0 x 10 6 Pa•sec. found between the tendency for asphalts to form The structural hardening was reversed on heating of "tender-mix" pavements and their lack of molecular the samples. The original unoxidized asphalt showed structuring when in contact with the surface of a almost no increase from structural hardening (3.B x standard Ottawa sand aggregate (_!!.§.). Thus the phenom­ 10 2 to 4.2 x 10 2 Pa"sec) on standing for the same enon of molecular structuring is important not only length of time. This strongly suggests that the oxi­ to the bulk properties of asphalt but also to the dation that takes place during the natural aging of asphalt-aggregate interaction. Both types of struc­ asphalts in pavements significantly increases the turing will affect the performance and durability of rate of structural hardening and that oxidative and asphalts in pavements. The common practice of eval­ structural hardening may be synergistic. uating asphalt performance in the absence of the Petersen 27

aggregate with which it is to be used leaves uneval­ nations of asphalt components with varying chemical uated important criteria for pavement performance. structures in individual asphalts may produce as­ phalts that will provide satisfactory service, it seems unlikely to the author that satisfactory SUMMARY AND CONCLUDING COMMENTS chemical composition specifications can be devised for asphalts. Such specifications would likely ex­ Chemical composition is important in determining the clude from use many asphalts that would otherwise physical properties and performance characteristics perform satisfactorily. Ideally, specifications should define the performance properties desired. of asphalts. The interactions of polar or polariz­ able chemical functionality, either naturally pres­ Chemical information would be important in producing asphalts that meet the performance criteria. ent or formed on oxidative aging, play a major role Composition information is useful in helping to in determining asphalt viscosity and related complex flow properties. understand asphalt--what makes it behave as it does and what makes one asphalt behave differently from Two major factors affecting asphalt durability another. With given asphalt sources available, com­ are (a) the compatibility of the interacting compo­ position information can be used to improve the nents of asphalt and (b) the resistance to changes product through modification with additives, by resulting from oxidative aging. Both factors are a blending, and so on, or to alter use design proce­ function of chemical composition, which can vary dures to accommodate specific properties. Composi­ widely from one asphalt source to another because of tion information can be used to match asphalt and inherent differences in crude sources or from pro­ aggregate, provide clues as to what modifications cessing and blending. are necessary to make an asphalt-aggregate system Historically, the study of asphalt chemical com­ more serviceable under a given environment, diagnose position has been facilitated by the separation of failures, and provide information needed in taking asphalt into component fractions based on the polar­ corrective measures. ity or adsorption character is tics or both of the ~s asphalts from new sources are utilized, and as molecular components present. The component frac­ processing and design procedures change, the use of tions, sometimes called generic fractions, although more fundamental information, both chemical and useful in classifying and characterizing asphalts and to provide simplified mixtures for further physical, and particularly as related to long-term study, are still complex mixtures the composition of performance and durability, should provide for prod­ uct improvement and reduce the number of early or which is a function of asphalt source. The component fractions are, however, sufficiently unique to unexpected failures of asphalt products. It is hoped that this review will bring about a identify their particular contribution to the com­ plex flow properties of asphalt. A proper balance of better understanding of the chemical compositional component types is necessary for a durable asphalt. factors that control the properties of asphalt and Because asphalt occurs as a film exposed to at­ will assist in providing direction to both research mospheric oxygen in pavements, it rapidly oxidizes and application leading to improved asphalt products in service with the formation of polar, strongly with better performance and durability. interacting, oxygen-containing chemical functional groups that greatly increase viscosity and alter complex flow properties. This hardening often leads RESEARCH NEEDS to asphalt embrittlement and ultimately asphalt pavement failure. Not only does the susceptibility In the opinion of the author, future research effort to oxidation vary from one asphalt to the other, but on chemical factors that affect asphalt durability the effect of the oxidation products formed on should include, as some of the most important, the physical properties also varies widely with asphalt following interrelated areas. source. The sensitivity of the asphalt to the chemi­ cal functionality produced on oxidation can be sig­ 1. Development of techniques to measure and nificantly reduced by removing or altering the polar evaluate the effects of molecular structuring (in chemical functionality initially present that would both neat asphalt and on the presence of aggregates) otherwise interact with the oxidation products to on the physical properties of pavements; increase viscosity. 2. Determination of the relationships between The identification and characterization of the molecular structuring and complex flow properties; chemical functional types normally present in as­ 3. Identification and characterization of the phalt or formed on oxidative aging that influence chemical factors that cause and control molecular molecular interactions afford a fundamental approach structuring, including the effects of oxidation; to the chemical compositional factors that determine 4. Development and validation of methods that physical properties, which in turn governs the per­ will predict changes in physical properties of as­ formance properties of both asphalts and asphalt-ag­ phalts in aged pavements because the properties of gregate mixtures. aged asphalts, not initial properties, determine the Asphalt physical properties are significantly properties affecting durability; altered not only by the oxidative formation of polar 5. Relation of the complex flow properties of chemical functional groups but by reversible molec­ asphalt (especially at low temperatures) with ular structuring (also called steric hardening). changes in chemical functionality resulting from This latter phenomenon is a slow process that ap­ oxidative aging and quantification of relationships pears to proceed concurrently and synergistically between chemical functionality and changes in physi­ with oxidative aging during the lifetime of the cal and performance-related properties; pavement and may be a major factor contributing to 6. Development of relationships between asphalt asphalt pavement embrittlement in the later stages compatibility, composition, and changes in physical of pavement service life. Limited data indicate that properties on oxidative aging; and the complex flow properties of asphalt and the ten­ 7. Examination of ways to improve asphalt compo­ dency of asphalt to structure are directly related. nent compatibility and durability by altering molec­ This suggests possibilities for ways to evaluate ular interactions by such means as chemical modifi­ this elusive property. cation, additives, component blending, and so on, Because asphalt chemical composition can vary This area should be highly rewarding in upgrading widely with asphalt source, and a variety of combi- the performance of otherwise less durable asphalts. 28 Transportation Research Record 999

ACKNOWLEDGMENT 15. J.Y. Welborn, F.R. Oglio, and J.A. Zenewitz. A Study of Viscosity-Graded Asphalt Cements. The author is grateful to the FHWA, u.s. Department Proc., Association of Asphalt Paving Technol­ of Transportation, for partial financial support ogists, Vol. 35, 1966, pp. 19-60. during the compilation of this review. 16. H.E. Schweyer and E.L. Chipley. Composition Studies on Asphalt Cements: A Progress Report. In Highway Research Record 178, HRB, National Research Council, Washington, D.C., 1967, pp. REFERENCES 30-59. 17. L.W. Corbett. Densimetric Method for Charac­ 1. F. N. Hveem. Quality Tests for Asphalts--A Pro­ terizing Asphalt. Analytical Chemistry, Vol. gress Report. Proc., Association of Asphalt 36, 1967, pp. 1967-1971. Paving Technologists, Vol. 15, 1943, pp. 111- 18. J.w. Ramsey, F.R. McDonald, and J.C. Petersen. 152. Structural Study of Asphalts by Nuclear Mag­ 2. W.P. Van Oort. Durability of Asphalt. Indus­ netic Resonance. Industrial and Engineering trial and Engineering Chemistry, Vol. 48, 1956, Chemistry Product Research and Development, pp. 1196-1201. Vol. 6, 1967, pp. 231-236. 3. P.C. Blokker and H. van Hoorn. Durability of 19. R.L. Hubbard and K.E. Stanfield. Determination Bitumen in Theory and Practice. Proc., Fifth of Asphaltenes, Oils, and Resins in Asphalt. World Petroleum Congress, Section VI, 1959, pp. Analytical Chemistry, vol. 20, 1948, pp. 460- 417-432. 465. 4. F.N. Hveem, E. Zube, and J. Skog. Progress 20. G. O'Donnell. Separating Asphalt into Its Report on the Zaca-Wigmore Experimental Asphalt Chemical Constituents. Analytical Chemistry, Test Project. In Special Technical Publication Vol. 23, 1951, pp. 11-20. 277, American """Society for Testing and Mate­ 21. M. Bestougeff. Isomerism of Petroleum Constit­ rials, Philadelphia, Pa., 1959, pp. 1-45. uents. Proc., Third World Petroleum Congress, 5. F.S. Rostler and R.M. White. Influence of Section VI, The Ha~ue, Netherlands, 1951, pp. Chemical Composition of Asphalts on Perfor­ 11-20. mance, Particularly Durability. In Special 22. A.J. Hoiberg and W.E. Garris, Jr. Analytical Technical Publication 277, American Society for Fractionation of Asphalts. Analytical Edition, Testing and Materials, Philadelphia, Pa., 1959, Industrial Engineering Chemistry, vol. 16, pp. 64-08. 1944, pp. 294-302. 6. F.S. Rostler and R.M. White. composition and 23. H.E. Schweyer and R.N. Traxler. Separating Changes in Composition of Highway Asphalts, Asphalt Materials--Butanol-Acetone Method. Oil 85-100 Penetration Grade. Proc., Association of Gas Journal, Vol. 52, 1953 1 p. 133. Asphalt Paving Technologists, Vol. 31, 1962, 24. M. Bestougeff. Constitution des Composies pp. 35-89. soutres Cycliques du Petrole. Proc., Fifth 7. S.H. Greenfeld and J.R. Wright. Four Methods World Petroleum Congress, Section V, New York, for Predicting the Durability of Roofing As­ N.Y., 1959, pp. 143-164. 25. L.R. Kleinschmidt. Chromatographic Method for phalts. Materials and Research Standards, Vol. the Fractionation of Asphalt into Distinctive 2, 1962, pp. 738-745. 8. C.D. Smith, c.c. Scheutz, and R.S. Hodgson. Groups of Components. Journal of Research, Relationship Between Chemical Structures and National Bureau of Standards, Vol. 54, 1955, Weatherability of Coating Asphalts as Shown by pp. 163-166. Infrared Adsorption Spectroscopy. Industrial 26. K.H. Altgelt and E. Hirsh. GPC Separation and and Engineering Chemistry Production Research Integrated Structural Analysis of Petroleum and Development, Vol. 5, 1966, pp. 153-161. Heavy Ends. Separation Science, Vol. 5, 1970, 9. T.C. Davis and J.C. Petersen. An Inverse GLC pp. 855-862. Study of Asphalts used in the zaca-Wigmore 27. D. Bynum, Jr., and R.N. Traxler. Gel Permeation Experimental Test Road. Proc., Association of Chromatography Data on Asphalts Before and Asphalt Paving Technologists, Vol. 36, 1967, After Service in Pavements. Proc., Association pp. 1-15. of Asphalt Paving Technologists, Vol. 39, 1970, 10. L.W. Corbett. Composition of Asphalt Based on pp. 683-702. Generic Fractionation Using Solvent Deasphalt­ 20. H. Reerink and J. Lijzenga. Gel-Permeation ening, Elution-Adsorption Chromatography, and Chromatography Calibration Curve for Asphal­ Densimetric Characterization. Analytical Chem­ tenes and Bituminous Resins. Analytical Chem­ istry, Vol. 41, 1969, pp. 576-579. istry, Vol. 47, 1975, pp. 2160-2167. 11. J. Knotnerus. Bitumen Durability--Measurement 29. c. Such and B. Brule. Characterization of a by Oxygen Adsorption. Industrial and Engineer­ Road Asphalt by Chromatographic Techniques (GPC ing Chemistry Product Research and Development, and HPLC). Journal of Liquid Chromatography, Vol. 11, 1972, pp. 411-422. Vol. 2, 1979, pp. 437-453. 12. R.L. Griffin, w.c. Simpson, and T.K. Miles. 30. P.W. Jennings, J.A.S. Pribanic, K.R. Dawson, Influence of Composition of Paving Asphalts on and C.E. Bricca. use of HPLC and NMR Spectros­ Viscosity, Viscosity-Temperature Susceptibil­ ity, and Durability. Journal of Chemical and copy to Characterize Asphaltic Materials. Pre­ Engineering Data, Vol. 4, 1959, pp. 349-354. prints, American Chemical Society, Division of Petroleum Chemistry, Vol. 26, No. 4, 1981, pp. 13. R.J. Peters. Compositional Considerations of 915-922. Asphalt for Durability Improvement. In Trans­ portation Research Record 544, TRB,--i.iational 31. T.C. Davis and J .c. Petersen. Inverse Gas­ Research Council, Washington, D.C., 1975, pp. Liquid Chromatographic Studies of Asphalt--Com­ 46-55. parison with Analyses by Fractionation. Analyt­ 14. H. Plancher, E.L. Green, and J.C. Petersen. ical Chemistry, Vol. 39, 1967, pp. 1852-1857. Reduction of Oxidative Hardening of Asphalts by 32. F.S. Rostler and H.W. Sternberg. compounding Treatment with Hydrated Lime--A Mechanistic Rubber with Petroleum Products. Industrial and Study. Proc., Association of Asphalt Paving Engineering Chemistry, vol. 41, 1949, pp. 598- Technologists, Vol. 45, 1976, pp. 1-24. 608. Petersen 29

33. F.J. Nellenstyn. The Constitution of Asphalt. tal Microbalance. Analytical Chemistry, Vol. Journal of the Institution of Petroleum Tech­ 41, 1969, pp. 580-583. nologists, Vol. 10, 1924, pp. 311-325. 52. J .c. Petersen, F.A. Barbour, and s.M. Dorrence. 34. F.J. Nellenstyn. Relation of the Micelle to the Catalysis of Asphalt Oxidation by Mineral Ag­ Medium in Asphalt. Journal of the Institution gregate Surfaces and Asphalt Components. Proc., of Petroleum Technologists, Vol. 14, 1928, pp. Association of Asphalt Paving Technologists, 134-138. Vol. 43, 1974, pp. 162-177. 35. C.J. Mack. Colloid Chemistry of Asphalts. Jour­ 53. s.M. Dorrence, F.A. Barbour, and J.C. Petersen. nal of Physical Chemistry, Vol. 36, 1932, pp. Direct Evidence of Ketones in Oxidized As­ 2901-2914. phalts. Analytical Chemistry, Vol. 46, 1974, 36. J.Ph. Pfeiffer and R.N.J. Saal. Asphaltic Bitu­ pp. 2242-2244. men as Colloid System. Journal of Physical 54. J.C. Petersen, F.A. Barbour, and s.M. Dorrence. Chemistry, Vol. 44, 1940, pp. 139-149. Identification of Dicarboxylic Anhydrides in 37. R.N.J. Saal, P.W. Bass, and w. Heukelom. The Oxidized Asphalts. Analytical Chemistry, Vol. Colloidal Structure of Asphaltic Bitumens. 47, 1975, pp. 107-111. Journal de Chimie Physique, Vol. 43, 1946, pp. 55. J .c. Petersen. Quantitative Method Using Dif­ 235-261. ferential Infrared Spectrometry for the De­ 38. L.C. Corbett. Relationship Between Composition termination of Compound Types Absorbing in the and Physical Properties of Asphalt. Proc., Carbonyl Region in Asphalts. Analytical Chemis­ Association of Asphalt Paving Technologists, try, Vol. 47, 1975, pp. 112-117. Vol. 39, 1970, pp. 481-491. 56. J.C. Petersen and H. Plancher. Quantitative 39. J. Manheimer. The Effects of Paraffins on As­ Determination of Carboxylic Acids and Their phalt. Proc., First World Petroleum Congress, Salts and Anhydr ides in Asphalts by Selective Vol. 2, 1933, pp. 553-556. Chemical Reactions and Differential Infrared 40. G.L. Oliensis. Compatibility Between Bitumens-­ Spectrometry. Analytical Chemistry, Vol. 53, Exudition Versus Insudation. Materials Research 1981, pp. 786-789. and Standards, Vol. 1, 1961, pp. 723-727. 57. J.C. Petersen. Oxidation of Sulfur Compounds in 41. w.c. Simpson, R.L. Griffin, and T.K. Miles. Petroleum Residues: Reactivity-Structural Rela­ Relationships of Asphalt Properties to Chemical tionships. Preprints, Division of Petroleum Composition. Preprints, Division of Petroleum Chemistry, American Chemical Society, Vol. 26, Chemistry, American Chemical Society, Vol. 5, No. 4, 1981, pp. 898-906. No. 4, 1960, pp. A79-A87. 58. T.C. Davis and J .c. Petersen. An Adaptation of 42. J.J. Heithaus. Measurement of Significance of Inverse Gas-Liquid Chromatography to Asphalt Asphaltene Peptization. Preprints, Division of Oxidation Studies. Analytical Chemistry, Vol. Petroleum Chemistry, American Chemical Society, 38, 1966, pp. 1938-1940. Vol. 5, No. 4, 1960, pp. A23-A37. 59. J.C. Petersen, H. Plancher, and G. Miyake. 43. R.M. White, W.R. Mitten, and J.B. Skog. Frac­ Chemical Reactivity and Flow Properties of tional Components of Asphalt--Compatibility and Asphalts Modified by Metal Complex-Induced Interchangeability of Fractions Produced from Reaction with Atmospheric Oxygen. Proc., As­ Different Asphalts. Proc., Association of As­ sociation of Asphalt Paving Technologists, Vol. phalt Paving Technologists, vol. 39, 1970, pp. 52, 1983, in press. 498-531. 60. J.C. Petersen. Relationships Between Asphalt 44. W.J. Halstead, F.S. Rostler, and R.M. White. Chemical Composition and Performance Related Properties of Highway Asphalts--Part III, In­ Properties. Proc., Ninth Annual Meeting of the fluence of Chemical Composition. Proc., Asso­ Asphalt Emulsion Manufacturers Association, Las ciation of Asphalt Paving Technologists, Vol. Vegas, Nevada, March 1982, and meeting of the 35, 1966, pp. 91-138. International Slurry Seal Association, Phoenix, 45. J.A. Zenewitz and J.Y. Welborn. Predicting Arizona, Jan. 1983. Asphalt Performance. Report FHWA-RD-76-133. 61. M.M. Boduszynski, J.F. McKay, and D.R. Latham. FHWA, u.s. Department of Transportation, 1975. Asphal tenes, Where Are You? Proc. , Association 46. G.R. Kemp. Asphalt Durability Tests and Their of Asphalt Paving Technologists, Vol. 49, 1980, Relationships to Field Hardening. In Special pp. 123-143. Technical Publication 532, American Society for 62. J .c. Petersen. An Infrared Study of Hydrogen Testing and Materials, Philadelphia, Pa., 1973, Bonding in Asphalt. Fuel, Vol. 46, 1967, pp. pp. 100-122. 295-305. 47. R.N. Traxler. Relation Between Hardening and 63. T.C. Davis, J .c. Petersen, and W.E. Haines. Composition of Asphalt. Preprints, Division of Inverse Gas-Liquid Chromatography. A New Ap­ Petroleum Chemistry, American Chemical Society, proach for Studying Petroleum Asphalts. Analyt­ Vol. 5, No. 4, 1960, pp. A71-A77. ical Chemistry, Vol. 38, 1966, pp. 241-243. 48. J.F. Goode and E.P. Owings. A Laboratory-Field 64. J.C. Petersen. R.V. Barbour, S.M. Dorrence, study of Hot Asphalt Concrete wearing Course F.A. Barbour, and R.V. Helm. Molecular Interac­ Mixtures. Public Roads, Vol. 31, 1961, pp. tions of Asphalt: Tentative Identification of 221-228. 2-Quinolones in Asphalt and Their Interaction 49. B.A. Vallegra, R.M. White, and K.S. Rostler. with Carboxylic Acids Present. Analytical Chem­ Changes in Fundamental Properties of Asphalts istry, Vol. 43, 1971, pp. 1491-1496. During Service in Pavements. Office of Research 65. J.C. Petersen, E.K. Ensley, and F.A. Barbour. and Development, U.S. Bureau of Public Roads, Molecular Interactions of Asphalt in the As­ Jan. 1970. phalt-Aggregate Interface Region. In Transpor­ 50. L.W. Corbett and R.E. Merz. Asphalt Binder tation Research Record 515, TRB, National Re­ Hardening in the Michigan Test Road After 18 search Council, Washington, D.C., 1974, pp. Years of Service. In Transportation Research 68-78. Record 544, TRB, National Research Council, 66. E.K. Ensley. A Study of Asphalt-Aggregate In­ Washington, D.C., 1975, pp. 27-34. teractions and Asphalt Molecular Interactions 51. W.H. King and L.W. Corbett. Relative Oxygen by Microcalorimetric Methods.: Postulated In­ Absorption and Volatility Properties of sub­ teraction Mechanism. Journal of the Institute micron Films of Asphalt Using the Quartz Crys- of Petroleum, Vol. 59, 1973, pp. 279-289. 30 Transportation Research Record 999

67. R.V. Barbour and J.C. Petersen. Molecular In­ 79. R.N. Traxler and C.E. Coombs. The Colloidal teractions of Asphalt: An Infrared Study of the Nature of Asphalt as Shown by Its Flow Prop­ Hydrogen-Bonding Basicity of Asphalt. Analyti­ erties. Journal of Physical Chemistry, Vol. 40, cal Chemistry, Vol. 46, 1974, pp. 273-277. 1936, pp. 1133-1147. 68. J.C. Petersen, E.K. Ensley, H. Plancher, and 80. R.N. Traxler and C.E. Coombs. Development of W.E. Haines. Paving Asphalts: Asphalt-Aggregate Internal Structure in Asphalts with Time. Interactions and Asphalt Intermolecular Inter­ Proc., American Society of Testing Materials, actions. Report FHWA-RD-77-25. FHWA, u.s. De­ Vol. 37 (II), 1937, pp. 549-557. partment of Transportation, Aug. 1976. 81. R.N. Traxler, H.E. Schweyer, and J.W. Romberg. 69. J.C. Petersen, H. Plancher, E.K. Ensley, and Rheological Properties of Asphalts. Industrial G. Miyake. Chemistry of the Asphalt-Aggregate and Engineering Chemistry, Vol. 36, 1944, pp. Interaction: Relationship with Pavement Mois­ 823-829. ture Damage Prediction Tests. In Transportation 82. A.B. Brown, J.W. Sparks, and F.M. Smith. Steric Research Record 483, TRB, National Research Hardening of Asphalts. Proc., Association of council, Washington, D.C., 1982, pp. 95-104. Asphalt Paving Technologists, Vol. 26, 1957, 70. H. Plancher,C.-K. Chow, and J.C. Petersen. Vis­ pp. 486-494. coelastic Measurements and Their Relationship 83. A.B. Brown and J .w. Sparks. Viscoelastic Prop­ to Bitumen Properties. In Proc., International erties of a Penetration Grade Asphalt at Winter Symposium on Progrossi Nella Technologia Dei Temperature. Proc., Association of Asphalt Bitumi, Stazione Sperimentale per i Combus­ Paving Technologists, Vol. 27, 1958, pp. 35-51. tibili, San Donato Milanese, Italy, pp. 248-262. 84. C.E. Coombs and R.N. Traxler. Rheological Prop­ 71. A.J. Hoiberg. Asphalt-Solvent Blends. Indus­ erties of Asphalts IV. Observations Concerning trial and Engineering Chemistry, Vol. 43, 1951, the Anomalous Flow Characteristics of Airblown pp. 1419-1423. Asphalts. Journal of Applied Physics, Vol. 8, 72. D.A. Anderson, E.L. Dukatz and J.C. Petersen. 1937, pp. 291-296. The Effect of Antistripping Additives on the 85. E.K. Ensley and H.A. Schulz. A Study of As­ Properties of Asphalt Cements. Proc., Associa­ phalt-Aggregate Interactions hy HP~~s of Tmmer­ tion of Asphalt Paving Technologists, Vol. 51, sion. Journal of the Institute of Petroleum, 1982, pp. 298-317. Vol. 58, 1972, pp. 95-101. 7 3. Highway Chemicals Newsletter. Akzo Chimie 86. R.E. Robertson, E.K. Ensley, and J.C. Petersen. America, (formerly Armak Company), McCook, Physiochemical Studies of Tender Mix and Non­ Ill., Fall 1979. tender Mix Asphalts. Report FHWA/RD-80/130. 74. N.P. Khosla. Effect of Emulsified Modifiers on FHWA, U.S. Department of Transportation, Dec. the Characteristics of Recycled Mixtures. 1980. Proc., Association of Asphalt Paving Technol­ 87. E.K. Ensley. A Kinetic Investigation of Asso­ ogists, Vol. 51, 1982, pp . 522-539. ciation in Asphalt. Journal of Colloid and 75. E.W. Mertens. Predicting weatherability of Interface Science, Vol. 53, 1975, pp. 452-460. coating-Grade Asphalts from Asphaltene Charac­ 88. E.K. Ensley. Multilayer Adsorption with Molec­ teristics. Bull. 250. ASTM, Philadelphia, Pa., ular Orientation of Asphalt on Mineral Aggre­ 1960, pp. 40-44. gate and Other Substrates. Journal of Applied 76. K.H. Altgelt and O.L. Harle. The Effect of Chemistry and Biotechnology, Vol. 25, 1975, pp. Asphaltenes on Asphalt Viscosity. Industrial 671-682. and Engineering Chemistry, Product Research and 89. F.N. Hveem, E. Zube, and J. Skog. Proposed New Development, Vol. 14, 1975, pp. 240-256. Tests and Specifications for Paving Crude As­ 77. H. Plancher, A.J. Hoiberg, s.c. Suhaka, and phalts. Proc., Association of Asphalt Paving J.C. Petersen. A Settling Test to Evaluate the Technologists, Vol. 32, 1963, pp. 271-327. Relative Degree of Dispersion of Asphaltenes. Proc., Association of Asphalt Paving Technol­ ogists, Vol. 48, 1979, pp. 351-374. 78. R.N. Traxler and H.W. Schweyer. Increase in Viscosity of Asphalts with Time. Proc., Ameri­ can Society of Testing Materials, Vol. 36 (II), Publication of this paper sponsored by Committee on 1936, pp. 544-551. Characteristics of Bituminous Materials.