Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
19 CHAPTER THREE TECHNIQUES FOR CHEMICAL CHARACTERIZATION OF ASPHALT ASPHALT FRACTIONATION Historically, the study of asphalt chemical composition has been facilitated by the separation of asphalt into component fractions based on the polarity or adsorption characteristics, or both, of the molecular components present. The component fractions, sometimes called generic fractions, are useful in classifying and characterizing asphalts and in providing simpli- fied mixtures for further study. They are still complex mixtures whose composition is a function of asphalt source. Many techniques using different properties of the molecules for separations have been used. The fractions are generally very hetero- geneous and are defined only by the method of separation. A common misunderstanding is to consider the fractions as âthe components in bitumenâ and claim that bitumen consists of a mixture of three, four, or five types of compounds instead of a continuum of molecules. The continuum consists of relatively large hydrocarbons with different sizes, polarity, and aromatic- ity. The component fractions are, however, sufficiently unique to identify their particular contribution to the complex flow properties of asphalt. A proper balance of component types is necessary for a durable asphalt. Many techniques, using different properties of the molecules for separations, have been used as shown in Table 3. These fractions have then been correlated to physical properties, sometimes rather successfully but more commonly without find- ing any general correlation. Historically, the two most common methods to fractionate asphalt are the chemical precipitation method (Rostler and White 1959) and the selective adsorptionâdesorption (chromatographic) method (Corbett 1970). In both methods, the asphaltenes are identified as the fraction insoluble in pentane or heptane. The pentane or heptane soluble frac- tion is classified as maltenes, which can be further separated into mixtures with differing solubility properties. The chemical precipitation method uses various sulfuric acid concentrations to identify and characterize four classes of molecular mixes comprising the maltenes: nitrogen bases, first acidiffines, second acidiffins, and parafins. The fractional components of four representative asphalts have been summarized by White et al. (1970). In the absence of asphaltene, the fractional maltene TABLE 3 METHODS FOR FRACTIONATING ASPHALTS AND CRUDE OILS Principle Products Precipitant References Solvent extraction Parafinnics, cyclics, asphaltics n-butanol, acetone Traxler and Schweyer (1953) Acid precipitation Paraffins, nitrogenbases, acidaffins no. 1, acidaffins no. 2, asphaltenes n-pentane Rostler and White (1970); White at al. (1970) Adsorption on alumina Saturates, naphtene aromatics, polar aromatics, asphaltenes n-heptane Corbett and Swarbrick (1966) Adsorption on clay Saturates, aromatics, polars, asphaltenes n-pentane Corbett and Swarbrick (1966) Gel permeation chromatography Molecules of different sizes N/A Davison et al. (1995) Gas chromatography Volatile hydrocarbons N/A Tang and Isacsson (2005); Fernandes et al. (2009) High-pressure liquid chromatography Molecules of different polarities N/A Such et al. (1979) Thin-layer chromatography (Iatroscan) Saturates, aromatics, resins A, resins B (SARA) N/A Bharati et al. (1994) Ion exchange chromatography Neutrals, bases, acids N/A Branthaver et al. (1992) Flash chromatography Saturates, aromatics, resins A, resins B n-heptane Raki et al. (2000) Asphaltene determinator Aromatics, saturates, resins, asphaltenes N/A Boysen and Schabron (2013) Automatic saturates, aromatics, resins and asphaltene-determinator Aromatics, saturates, resins, asphaltenes N/A Schabron and Rovani (2013) Source: Adapted from Masson et al. (2001). N/A = not available.
20 components from different asphalts are compatible. Variations in the asphalts are the result of changes in the asphaltene com- ponents. The research demonstrates that the durability of the asphalts depends on the proportions of the components and that the asphalt quality could be improved by blending to change the proportions of maltene components to reach a desirable com- position ratio. Rostler and White (1970) demonstrated that the primary differences in asphaltene fractions was their molecular weights. Asphaltene molecular weight variations among the asphalts studied was not very significant (2,000â6,000), but the addition of compatible rubbers enhanced the properties of the corresponding HMAs. TABLE 4 FRACTIONS OBTAINED USING CORBETT ANALYSIS Rings/Mole Group Weight Percent Range Average Molecular Weight Fraction Aromatic Naphthene Aromatic Description Saturates 5â15 650 0 3 2.6 Pure paraffins + pure naphthenes mixed paraffin naphthenes Naphthene aromatics 30â45 725 0.25 3.5 7.4 Mixed paraffin â naphthene aromatics + sulfur-containing compounds Polar aromatics 30â45 1,150 0.42 3.6 Not determined Mixed paraffin naphthene aromatics in multiring structures and sulfur, oxygen. nitrogen-containing compounds Asphaltenes 5â20 3,500 0.5 Not determined Not determined Mixed paraffin â naphthenc aromatics in polycyclic structures + sulfur, oxygen, nitrogen-containing compounds Source: Corbett (1969). Corbett Fractionation This Rostler method has been essentially superseded by the chromatographic method as developed by Corbett (Corbett and Swarbrick 1966; ASTM 2008). Corbett used a densometric procedure coupled with molecular weight determination by VPO at 37oC to determine the structure of his fractions (Corbett 1964). Asphaltenes could not be characterized completely because of the difficulties in molecular weight determination as a result of asphaltene molecular association. However, the asphaltenes precipitated by heptane can be further fractionated by the relative solubility of the asphaltenes in toluene and carbon disulfide (Figure 7) (Speight 2004). After precipitating the asphaltene fraction by addition of n-heptane, the maltenes in the heptane solution are coated unto an alumina column. Sequential elution of the column with heptane and benzene yields the least polar saturated hydrocarbons followed by the naphthenes and aromatic hydrocarbons. Increasing the polarity of the eluting solvents by adding methanol to the benzene and changing to a trichloroethylene eluent yields the polar aromatics and resins. Alternatively the deasphaltened oil can be fractionated by sequential extraction with heptane, toluene, and pyridine. The fractions are denoted as saturates, aromatics, resins, and asphaltenes (SARA). The properties of the fractions are compiled in Table 4. Bulk flash chro- matography of the maltenes can be used to obtain gram quantities of saturates, aromatics, and resins fractions (Masson 2000). FIGURE 7 Separation of bitumen into its various fractions, highlighting the SARA (saturates, aromatics, resins, and asphaltenes) fractions. (Source: Speight 2004.)
21 Table 5 shows additional structural data estimated for the fractions (Corbett 1970). These results are all dependent on the composition of the crude oil source, particularly heteroatom content and metals. Both nickel and vanadium are found primarily in the heptane-precipitated asphaltenes and are evenly distributed in the resins and asphaltenes. They appear to be interchange- able in structureâin fractions of a given asphalt the ratio of vanadium to nickel is constant over wide ranges of composition. TABLE 5 ELEMENTAL CHARACTERIZATION OF CORBETT FRACTIONS Average Number of Atoms per Molecule in: Element Saturates Naphthene Aromatics Polar Aromatics Asphaltenes Carbon Paraffin chain 31 21 24 85 Naphthene ring 14 17 18 29 Aromatic ring 0 13 25 115 Hydrogen 85 94 105 350 Sulfur 0 0.5 1 4 Nitrogen 0 0 1 3 Oxygen 0 0 1 2.5 Avg. molecular weight 625 730 970 3,400 Source: Corbett (1969, 1970). Heteroatoms are important because of their inordinate contribution to resin properties. The presence of heteroatoms enhances the activity of asphalt molecules toward oxidation. Large increases in asphalt hardening occur with the uptake of only 1 weight percent oxygen. Petersen has carried out extensive work on functional group analysis. A typical analysis is shown in Table 6 (Petersen 1986). When asphalt oxidizes, the principal increases in oxygenated species are in ketones and sulfoxides. Carboxylic acids and anhydrides tend to concentrate at the aggregate surface in asphalt concrete and may produce sensitivity to water damage. TABLE 6 DISTRIBUTION OF FUNCTIONAL GROUPS IN FRACTIONS FROM CORBETT SEPARATION Concentration in Fractiona (mole/liter) Group Whole Asphalt Saturates Naphthene Aromatics Polar Aromatics Asphaltenes Ketone 0 0 0 0.11 Trace 0.027 0 0 0 0.034 Anhydrides 0 0 0 Trace Trace 2-Quinoline types 0.021 0 0 0.023 0.046 Sulfoxides 0.019 0 0 0.12 0.09 Pyrrolics 0.17 0 0 0.21 0.23 Phenolics 0.035 0 0 0.055 0.075 aYield of fractions based on whole asphalt were saturates, 9.9%; naphthene aromatics, 25.3%; polar aromatics, 38.1%; asphaltenes, 21.6%; loss on the column (which should be added to polar aromatics) 5.1%. Source: Petersen (1986). Studies have shown that increases in asphalt viscosity with oxidation can be correlated with increases in carbonyl forma- tion, which has been shown to be proportional to oxygen uptake (Liu et al. 1998). Almost certainly this hardening results from hydrogen bonding between heteroatom groups in asphaltene molecules and also between polar aromatics, which then may become asphaltenes (Barbour and Petersen 1974; Herrington and Wu 1999). This association strongly impacts attempts to measure unimolecular size by GPC or colligative properties. Iatroscan Analysis The Corbett analysis became more routine when a procedure was developed that allowed rapid sample analysis. Thin layer chromatography (TLC) on a thin layer of silica powder that is fused onto a quartz rod (a Chromarod) coupled with
22 a flame ionization detector has been incorporated into a commercial instrument, an Iatroscan. The instrument requires milligram quantities of a sample, and the asphalt is separated into four fractions with better resolution and more rapidly than the column chromatography procedure. The four fractions are pyrolyzed on the developed Chromarod and the pyrolysates pass through the FID. The signal output of the FID detector is plotted versus the Chromarod peak position to produce an Iatrogramme (Masson 2001). The signals are attributed to SARA; the procedure is called a SARA analysis and it is used extensively in the petroleum industry to analyze crude oils as well as HMAs (Bharati et al. 1994). Cor- relation of the asphaltene fraction with alternate separation methods is improved by precipitating the asphaltenes with heptane and only depositing the maltenes on the Chromarod. It is important to realize that the SARA analysis provides general information about the composition, depending on the solubility, the adsorption, and the partition coefficient of the compounds in each solvent, all of them affected by mutual interactions between the sample components. Therefore, the SARA fractionation has limitations for providing an accurate description of the particular structural and chemical composition of the sample. Automated SAR-AD A new on-column precipitation and redissolution separation technique uses a continuous flow system to precipitate and redissolve various chemical species from oil strictly on their relative solubility (Boysen and Schabron 2013). This saturates, aromatics, resins, and asphaltene-determinator method (SAR-AD) involves precipitation of asphaltene components from residua on a ground PTFE or activated silica column using a heptane mobile phase. The precipitated material is redissolved and eluted at 30Â°C in three steps using solvents of increasing solubility parameter: cyclohexane, toluene, and methylene chlo- ride/methanol (98:2, v/v). A series of automated switching valves is used to direct solvent flow into the various columns in forward and reverse directions in a complex series of steps. The eluted fractions are detected using both an evaporative light scattering detector (ELSD) and a 500 nm detector. An illustrative chromatogram is shown in Figure 8. FIGURE 8 Automated SAR-AD separation profile for 2 mg Lloydminster vacuum residuum (Source: Boysen and Schabron 2013). Note: SAR-AD = saturates, aromatics, resins, and asphaltene-determinator method; ELSD = evaporative light scattering detector.
23 The SAR-AD results in Table 7 are different than those from the manual SARA separations performed during the original SHRP program (Jones 1993) and other SARA separations performed on these binders using different methods. In particular, the quantity of saturates fractions is greater in the automated SAR-AD method. Differing asphaltene separation protocols and separation mediums between these two studies is likely responsible for these discrepancies. The many different SARA methods utilize procedures with different solvents and filters for isolating asphaltenes and different solvents and sorbents for the maltenes separations. Results from these different methods are not necessarily comparable. What is important is that the new method can be applied to separate and quantify the asphaltene components. Further applications include the ability to study how asphalt binders change with oxidation, how oils change with processing, and also the differences between samples in a repeatable manner. TABLE 7 AUTOMATED SAR-AD RESULTS FOR SHRP CORE ASPHALT BINDERS Maltenes Asphaltenes Sample Detector Saturates Aromatics Resins CyC6 Toluene CH2Cl2/MeOH Total ELSD AAA-1 ELSD 19.0 19.9 47.4 5.2 8.3 0.2 13.6 500 nm 0.6 31.9 27.6 37.7 2.3 AAE-1 ELSD 16.9 10.4 50.4 4.7 17.4 0.2 22.3 500 nm 0.4 28.4 19.2 49.4 2.6 AAB-1 ELSD 18.0 15.5 54.6 3.3 8.5 0.1 11.9 500 nm 0.5 36.6 18.1 41.5 3.3 AAC-1 ELSD 29.2 13.1 51.6 2.2 3.8 0.1 6.1 500 nm 0.9 48.7 17.0 30.5 2.9 AAD-1 ELSD 11.8 15.5 54.7 5.8 12.2 0.1 18.1 500 nm 0.6 26.6 23.8 46.3 2.7 AAF-1 ELSD 17.4 13.2 62.1 2.5 4.7 0.0 7.2 500 nm 0.8 46.5 18.8 31.9 2.1 AAG-1 ELSD 18.0 12.4 67.3 0.5 1.8 0.0 2.3 500 nm 1.5 65.8 9.4 20.9 2.5 AAK-1 ELSD 11.3 18.6 54.9 4.3 10.9 0.1 15.3 500 nm 0.5 30.7 23.5 42.4 3.0 AAM-1 ELSD 18.3 13.7 65.9 0.8 1.2 0.1 2.1 500 nm 1.1 76.6 8.8 11.4 2.2 Note: SAR-AD = saturates, aromatics, and resins asphaltene-determinator method; ELSD = evaporative light scattering detector. Source: Boysen and Schabron (2013). The results presented in Table 7 show real differences between binders. They also show a change resulting from oxidation; for example, binder AAE-1 is air-blown AAA-1. The ELSD toluene-soluble asphaltenes-to-aromatics ratio for the original AAA-1 is 0.4, and this ratio increases to 1.7 with air-blowing oxidation treatment. These ratios are less than 1 for all of the other original asphalt binders. The 500 nm aging index ratio is the ratio of the toluene soluble asphaltenes area to the area of the resins at 500 nm. This ratio is 1.2 for the original AAA-1 binder, and it increases to 1.7 with the air-blowing treatment. All of the other binders except AAD-1 have values less than 1.7. The automated SAR-AD separation is highly repeatable and gives real content differences between asphalt binders and residua, which allows for an understanding of the relationship between chemical composition and physical properties. The total pericondensed aromaticity, ratio of 500 nm toluene soluble asphaltenes to resins, and ratio of ELSD toluene soluble asphaltenes to aromatics are being explored for possible use as aging indices and for asphalt performance correlations (Boysen and Schabron 2013). Chemical composition is important in determining the physical properties and performance characteristics of asphalts. The interactions of polar or polarizable chemical functionality, either naturally present or formed on oxidative aging, play a major role in determining asphalt viscosity and related complex flow properties. Two major factors affecting asphalt durability are (1) the compatibility of the interacting components of asphalt and (2) the resistance to changes resulting from oxidative aging. Both factors are a function of chemical composition, which can vary widely from one asphalt source to another because of inherent differences in crude sources or from processing and blending.
24 ANALYTICAL INSTRUMENTATION FOR ASPHALT ANALYSIS Thermal Analysis Thermogravimetric analysis and differential scanning calorimetry are used to characterize petroleum bitumens and their chromatographic fractions. The influence of the different constituents on the thermal stability of bitumen was studied by TGA. In this analysis, the change in mass of a material is measured as a function of temperature or time. TGA facilitates acquisition of information on properties of a material and its composition. When a sample is heated it often loses mass. Loss of mass may be caused by vaporization or chemical reactions that evolve volatile products from the sample. A decomposition as a result of chemical reaction leads to changes in the mass of the sample in a stepwise manner as the onset temperature for each phase of the decomposition is reached. The onset temperature at which decomposition occurs provides information on the stability of the material in that atmosphere. For example, if a reactive gas atmosphere is used, reaction of the material with the gas can result in mass change. Typically, this mass change is exhibited in the form of mass loss; however, in cases such as oxidation there may be a gain in mass. Composition of a material can be determined by analyzing the temperatures and the extent of the individual weight losses. The derivative of the weight loss curve, DTA, facilitates identification of the thermal transitions. TGA can be used to mea- sure the thermal stability of a polymer and the thermal degradation of polymer blends owing to the simplicity of the weight loss method. The potential of TGA for quantitative analysis of vulcanizates based on binary elastomer blends of natural rub- ber (NR) and styrene-butadiene rubber (SBR) has been previously reported (Lee et al. 2007). A typical thermogram from an analysis of a natural rubber vulcanizate is shown in Figure 9. FIGURE 9 Thermogravimetric analysis/differential thermal analysis thermogram of natural rubber vulcanizate (Source: Baumgardner et al. 2014). The sample is heated under a nitrogen atmosphere according to a set protocol of isothermal and temperature ramping sequences. Weight loss steps on the TGA output appear as downward slopes of the solid thermogravimetric curve and peak in the dashed DTG curve. The slope of the thermogravimetric curve corresponds to the rate of change of sample mass. In region 1, volatile compounds such as water, residual solvents, and oils are evolved from 0 to 25 minutes at relatively low temperatures from 40Â°C to 250Â°C (104Â°F to 482Â°F). In region 2, pyrolytic decomposition occurs in an inert atmosphere (nitrogen) from 25 to 50 minutes at 250Â°C to 550Â°C (482Â°F to 1022Â°F) (Juma et al. 2006), allowing for analysis of the content (step height) and material type of the rubber hydrocarbon component. In region 3, the carbon black is combusted upon switching to an oxidative (air) atmosphere from 50 to 70 minutes at 550Â°C to 750Â°C (1022Â°F to 1382Â°F). Region 4 is residual ash remaining from the
25 entire TGA process. When materials of binary NR/SBR compounds were decomposed in nitrogen between 250Â°C and 550Â°C (482Â°F to 1022Â°F), two distinct regions of decomposition were observed. Calibration of TGA data with known samples led to a protocol that allowed the ratio of NR to SBR in crumb rubber samples to estimated (Baumgardner et al. 2014). TGA measurements also provide a simple means to determine the thermal stability of bitumen (Jimenez-Mateos et al. 1996; Mothe et al. 2008). TGA is an excellent procedure for determining the volatile asphalt components and residual solvents in binder extracts. Thermal analysis can be considered as one method for measuring variations in asphalt microstructures. If the energy required to maintain a constant temperature between a reference sample and the analyte is plotted, one obtains a differential scanning chromatogram. The calorimeter is sensitive to thermal transitions such as glass transitions and melting points of crystalline fractions in asphalt binders, which impact their physical and rheological properties. Noel and Corbett used differential scanning calorimetry to estimate both the crystalline fraction and glass transition temperature, Tg, of binders and recognized that it is critical to establish a consistent thermal history prior to comparing different binder samples (Noel and Corbett 1970; Claudy et al. 1992b). Crystalline Components of Asphalt The crystallization process of asphalt fractions depends on thermal history. Slow heating or annealing before analysis to experimentally realize a near-equilibrium state is necessary to study the system with more thermodynamic rigor. Using an annealing protocol, a number of low-temperature transitions characteristic of a given crude and/or refinery source can be identified; thus, comparison of thermograms from an unknown source with reference thermograms from documented sources allows the unknown source to be identified. The crystallization process of a selected AC-20 asphalt, ACB, was systematically studied by doping with the following pure crystalline hydrocarbons: octadecene-1, eicosane, and octacosane. The impact of doping on the crystalline fractions in asphalt varies from asphalt to asphalt. The crystalline components in asphalt exhibit distinct endothermic patterns that depend on their chemical structure, interactions with the amorphous phase, and interac- tions among themselves. The most significant endothermic effect is produced by cocrystallization of components with similar crystalline chain lengths. For example, the crystalline components of asphalt ACB do not interact with octadecene-1, but do cocrystallize with eicosane and octacosane (Daly et al. 1996). Chambrion observed two glass transitions in bitumen after cooling at constant rate. The magnitude and temperature of these transitions depended on the cooling rate. At low cooling rates (<1ÂºK/min), the glass transition at the higher temperature vanishes and an endothermic peak is obtained. From these observations, a segregation mechanism is proposed to explain the behavior of bitumen during cooling (Chambrion et al. 1996). Glass transition temperatures (Tg) of an asphalt and its fractions obtained by preparative GPC decreased drastically from 17.5Â°C to â63.36Â°C as the apparent molecular weight of the fractions increased from 500 to 800 Daltons, then increased regularly to â53Â°C as the molecular weight increased further to 3,000 Daltons. Composition rather than molecular weights of the fractions is responsible for the control of the Tg values (Hon et al. 1978). Bitumen was analyzed by modulated DSC (Masson et al. 2002). This method allows for the deconvolution of overlapping reversing and nonreversing thermal events and it allows for the observation of transitions not visible on the thermal curve obtained by standard DSC. The reversing thermal curve revealed two Tgs in bitumen, which had an 85/100 penetration grade and respective saturates, aromatics, resins, and asphaltenes contents of 9%, 27%, 43%, and 20%, as measured with the Iat- roscan by successive elution in heptane, toluene, and tetrahydrofuran (THF). One transition, at â20Â°C, was assigned to the maltenes, the other at 70Â°C to the asphaltenes. The heat capacity of these transitions was found to depend on thermal history. After cooling from the melt and annealing at 22Â°C, bitumen microstructure was found to develop in four stages. Most rapid is an ordering process that occurs when bitumen is quenched from the melt. It is postulated that this first stage arises from the partial ordering of simple aromatic structures into micro- and nano-phases; a second stage when low-MW saturated seg- ments crystallize. In the third stage, high-MW saturated segments crystallize. In the fourth stage, resins and asphaltenes order into a mesophase. The third and fourth stages are responsible for the room-temperature (steric) hardening of bitumen. The development of bitumen microstructure and the calculations of the entropy and enthalpy of transitions suggest that bitumen is a structured amorphous phase with a small crystalline phase (Masson and Polomark 2001; Masson et al. 2002). From the enthalpy of the endotherms, ÎH, it is estimated that more than 50% of the total endotherm arises from isotropization of these aromatics, the rest arising from the melting of low-MW saturated segments (Masson and Polomark 2004). Wax crystallization and melting in bitumen is usually considered detrimental to bitumen quality and asphalt performance. DSC is used to study the wax morphology in bitumen with respect to time, temperature, and thermal cycling. Eight waxy
26 bitumens from different sources and three lab blends prepared by adding a slack wax (a mixture of oil and wax, obtained from lubricating oil) and two isolated bitumen waxes to a nonwaxy bitumen were characterized using DSC, polarized light microscopy, confocal laser scanning microscopy, and freeze etching (fracture) in combination with transmission electron microscopy. The DSC results indicated that the selected bitumen samples differ widely in wax content and initial wax crystal- lizing and melting-out temperatures. Nonwaxy bitumen displayed no structure or crystals in the other three methods, while waxy bitumens from different crude origins showed a large variation of structures. The morphology of wax crystals was highly dependent on crystallization temperature as well as thermal history. The wax that has been isolated from waxy bitu- men and mixed into nonwaxy bitumen displayed similar morphology as the wax in the original bitumen. It was also found that bitumen wax usually melted at temperatures lower than 60Â°C although in one case a temperature of 80Â°C was required to complete melting of the wax (Lu et al. 2005). The nature and origin of bee-like microstructures (bees) in asphalt binders and their impact on asphalt oxidation have been the subject of extensive discussions in recent years. Although several studies refer to the bees as solely surface features, some others consider them to be bulk microcrystalline components that are formed as a result of the co-precipitation of wax and asphaltene molecules. Pahlavan et al. (2016) use a rigorous theoretical and experimental approach to study the interplay of asphalt components (mainly asphaltene and wax) and their impact on bee formation In the theoretical section of their study, quantum- mechanical calculations using density functional theory are used to evaluate the strength of interactions between asphaltene unit sheets in the presence and absence of a wax component, as well as the mutual interactions between asphaltene molecules (mono- mers and dimers) and paraffin wax. The results reveal that paraffin waxes not only do not reinforce the interaction between the asphaltene unit sheets, they destabilize asphaltene assembly and dimerization. This destabilization among interacting systems (asphalteneâasphaltene and waxâasphaltene) does not support the hypothesis that interaction between paraffin waxes and nonwax components, such as asphaltene, is responsible for their co-precipitation and âbee formation.â To further examine the effect of wax component on asphalt microstructure experimentally, the authors used AFM to study the surface morphology of an asphalt sample doped with 1% to 25% paraffin wax. The experiments indicate that paraffin wax tends to crystallize separately and form lamellar paraffin wax crystal inclusions with 10 nm thickness. Also, the addition of 3% wax into asphalt results in a significant increase in surface roughness from 0.5 nm to 4.1 nm and an increase in bee wavelength from 651 nm to 1038 nm (Pahlavan et al. 2016). FOURIER TRANSFORM INFRARED SPECTROSCOPY FTIR is one of the more important methods for fingerprinting asphalt materials (Lamontagne et al. 2001; Durrieu et al. 2003), based on its sensitivity and exploiting ease. It is able to quickly offer reliable information regarding aliphaticity, aromatic- ity, and oxygenation rate. This technique can give more accurate data such as the average distribution length of aliphatic chains, oxygenation, and substitution mode of aromatics (Lu and Isacsson 1998; Lamontagne et al. 2001). Combining infrared spectrometry with the specialized use of selective chemical reactions and differential spectra allowed quantification of the analytical absorption bands of interest including the naturally occurring functional groups; carboxylic acids and their salts, 2-quinolone types, phenolics, and pyrrollitcs; and those formed on oxidation; ketones, anhydrides, small amounts of acids, and sulfoxides (Petersen 1986). Relatively low-cost ($20,000 to $40,000) portable devices have become available for FTIR. These can be employed in the field to test the chemical composition of the delivered materials. These are point-and-shoot applications that could potentially be used by field technicians with accuracy similar to that obtained by using traditional stationary laboratory equipment (Zolfka et al. 2013). Characteristic Absorption Bands An infrared spectrum is a plot of the energy absorbed at specific wavelengths by the chemical functional groups in the asphalt. Each type of chemical bond can be identified by characteristic bands in the absorption spectrum of the transmitted IR radia- tion. The concentration of the functional groups associated with a given absorption can be deduced from the intensity of these bands using the BeerâLambert Law. The absorption maxima for typical groups found in AC are compiled in Table 8. Although transmission IR spectroscopy is the most efficient technique for making quantitative measurements, an FTIR spectrometer in the attenuated total reflectance (ATR) mode is the routine tool for studying asphalt samples (Jemison et al. 1990; Yut and Zofka 2011). In an ATR mode, the infrared beam is directed onto an optically dense crystal with a high refrac- tive index at a certain angle (Figure 10) (Perkin Elmer 2015). The internal reflection creates an evanescent wave that extends beyond the surface of the crystal, into the sample held in contact with the crystal. This evanescent wave extends only a few
27 microns (0.5Î¼â5Î¼) beyond the crystal surface and into the sample. If the sample is placed in good contact with the surface of the crystal, the sample absorbs energy and the evanescent wave will be attenuated or altered. The attenuated energy from each evanescent wave is passed back to the IR beam and then exits the opposite end of the crystal and is passed to the detector in the IR spectrometer. The detector then generates the IR spectrum. A graphical representation of ATR phenomena (a) and an image of the real crystal (b) are shown in Figure 10. TABLE 8 FTIR ABSORPTION MAXIMA OF ASPHALT FUNCTIONAL GROUPS Functional Group Wavenumber, cm-1 Species Source -OH 3,300 Alcohol/phenol Natural -C-H stretch 3,000 Aromatic Natural -CH2 stretch 2,920, 2850 Aliphatic Natural -C(=O)-O-C(=O) 1780 Anhydride Oxidative aging product -C=O 1,700 Ketone Oxidative aging product -C=O 1,700 Esters Oxidative aging product -C=O(OH} 1,650 Carboxylic acid Oxidative aging product -C=C- aromatic rings 1,600 Natural -CH2 bending 1,460 Aliphatic Natural -CH3 bending 1,375 Aliphatic Natural -O- Ether Natural -S- 690 Sulfide Natural -S=O 1,030 Sulfoxide Oxidative aging product -(CH2)n 745 Aliphatic chain Natural Source: Petersen (1986). FIGURE 10 (a) Graphical representation of attenuated total reflectance (ATR) phenomena; (b) an image of the real crystal (Source: Perkins Elmer 2015). An FTIR spectrum illustrating the characteristic absorption bands for bitumen is shown in Figure 11. The functional and structural indices are calculated from the band areas measured from valley to valley. The use of the areas rather than the band heights allows incorporation of several vibrations of the same type (for example, the C=O ester, acid, and ketone vibrations between 1,753 and 1,635 cm-1). One approach for estimating the concentration of several important functional groups using band area ratios is defined here (Lamontagne et al. 2001). aromatic structures (aromaticity index) A1600 /Î£Î aliphatic structures (aliphatic index) Î£(A1460 +A1376)/Î£A branched index A1376/Î£(A1460 +A1376)
28 long chains index A724/Î£(A1460 +A1376) oxygenated functions (carbonyl index) A1700 /Î£Î sulfoxide A1030 /Î£Î The sum of the area represents: Î£Î = A1700 + A1600 + A1460 +A1376 + A1030 + A864 + A814 + A743 + A724 + A (2953,2923,2862) Using band area ratios ensures spectrum normalization so that the bitumen film does not have a constant thickness. FIGURE 11 Fourier transform infrared spectra showing valley to valley area integration (Source: Lamontagne et al. 2001). Asphalt Aging The extent of aging of a given asphalt binder sample can be followed using the area ratio technique to monitor the oxygenated functions, carbonyl index, and sulfoxide index. Extensive studies have confirmed that the aging process is dependent on the type of binder (crude oil origin, refining process) and the type of aggregate used in the mix. Direct FTIR index comparisons should be limited to the same asphalt/aggregate combination. The potential benefits and limits of the FTIR methods for reclaimed asphalt characterization based on an international round robin test are summarized by Marsac et al. (2014). The identification and characterization of the chemical functional types normally present in asphalt or formed on oxidative aging that influence molecular interactions afford a fundamental approach to the chemical compositional factors that determine physical properties, which in turn governs the performance properties of both asphalts and asphalt-aggregate mixtures. Elemen- tal analysis revealed differences in carbon and sulfur contents and an increase in oxygen content on aging. FTIR confirmed that an increase of oxygen was caused mainly by carbonyl and sulfoxide groups. FTIR also showed a higher content of arylalkylk- etones and a higher content of oxidizable sulfur compounds in asphalts derived from sour crudes (Michalica et al. 2008). In addition to the oxidative formation of polar chemical functional groups, aged asphalt physical properties are signifi- cantly altered by reversible molecular structuring (also called steric hardening). This latter phenomenon is a slow process that appears to proceed concurrently and synergistically with oxidative aging during the lifetime of the pavement and may be a major factor contributing to asphalt pavement embrittlement in the later stages of pavement service life. Limited data indicate that the complex flow properties of asphalt and the tendency of asphalt to form microstructures are directly related. Additive Identification The FTIR method is an efficient technique for identifying additives in a binder. Quantitative determination of polymer content is essential to quality control and quality assurance during the processing and application of PMAs. Research on the quantita-
29 tive determination of polymer content in modified asphalt has been conducted (Sun and Zhang 2013). A general description of four methods for additive analysisânamely, performance-based methods, dissolving-separation methods, gel permeation chromatography methods, and IR spectroscopy methodsâis given. Of these methods, IR spectroscopy is probably the most appropriate test method to determine several types of polymer [SBR, styrene-butadiene-styrene copolymer (SBS), and EVA content in modified asphalt]. A typical ATR FTIR spectrum of EVA exhibits several distinctive absorption peaks (Table 9). The polyethylene chain is characterized by sharp peaks at 2,920, 2,850, and 724 cm-1. The carbonyl peak at 1,735 cm-1 coupled with a double peak at 1,240 to 1,270 cm-1 represents the acetate component in EVA. Finally, two peaks at ~1,640 and ~1,560 cm-1 are associated with vinyl vibrations. When added to asphalt binder, EVA has two distinctive peaks at ~1,240 and 1,735 cm-1, which can be used for positive identification of this additive. The most distinctive chemical bonds in a typical styrene-butadiene-based polymer are aromatic CâH bonds in polystyrene and trans-alkene (vinyl) CâH bonds in polybutadiene. The out-of-plane vibrations yield prominent IR peaks at ~700 and ~965 cm-1 for polystyrene and polybutadiene, respectively, and confirm the presence of SBS. It is important to note that polybutadiene peak at 967 cm-1 is likely to be obscured by a strong and wide band centered on 1,000 cm-1 that is associated with the silicate component of the aggregates (Zofka et al. 2013). TABLE 9 FTIR ABSORPTION MAXIMA OF DISTINCTIVE POLYMER FUNCTIONAL GROUPS Polymer Additive Functional Group Wavenumber, cmâ1 Species Elvaloy EVA Methylene 2,920, 2,285, 724 Polyethylene block Elvaloy EVA Carbonyl 1,735 Vinyl acetate Elvaloy EVA Double bond to oxygen 1,240, 1,270 Vinyl acetate Elvaloy EVA Vinyl 1,640, 1,560 Vinyl substituents Styrene butadiene copolymers Aromatic C-Hl 700 Polystyrene block Styrene butadiene copolymers Trans-alkene C-H 965 to 970 Butadiene block Source: Zofka et al. (2013). Aging of Polymer-Modified Binders One of the most important issues in asphalt chemistry is to identify the processes occurring during PMA aging. These processes involve bitumen aging, polymer aging, or both at the same time (Mouillet et al. 2008). Moreover, most PMAs feature a two- phase structure made of polymer-rich areas along with polymer-poor regions, depending on the bitumen chemistry, the polymer nature, and content. It is therefore important to take this binary phase behavior into consideration when trying to sort out the respective effect on aging of the polymer and the bitumen. IR microscopy, which can focus on separate phases in the blend, facilitates the characterization of different phases in heterogeneous products. PMAs can be studied in their original state and after conventional aging tests that simulate the aging during the mixing process and several years of road service (RTFO + PAV tests). IR microscopy was used to determine for each phase the polymer aging rate and functional indices characterizing the bitumen, such as aromaticity, aliphaticity, and condensation, and also to map the polymer distribution in the PMA. Characterization of PMAs in their original state shows that species of bitumen are involved in the polymer swelling and the effect of the polymer interaction with the bitumen phase. Characterization of the same PMAs after RTFO + PAV aging shows how the bitumen species responsible for the swelling evolve during aging. In addition, kinetic studies can be performed using a heating cell fitted to the IR microscope. The chemical composition of the bitumen part of the bitumen phase and the polymer phase before and after aging could be assessed by subtracting infrared spectra. These studies demonstrate an inter- dependence of the aging of the different constitutive phases in a PMA and of chemical exchanges between them, which leads to modification of the micro-morphological structure during aging in a PMA. The general trend for SBS-modified bitumen is that the binder becomes more homogeneous upon aging. This is the result of both some polymer degradation (chain scission) and a better compatibility of the smaller polymer chains with the oxidized bitumen molecules. GEL PERMEATION CHROMATOGRAPHY Differences in the molecular structures of maltenes and asphaltenes have prompted efforts to separate these components using size exclusion chromatography, more commonly known as gel permeation chromatography. In theory GPC provides a
30 simple separation of molecules in a sample according to their sizes or, more specifically, their hydrodynamic volumes. GPC is a method of separating molecules based on their size and shape in solution. The column used for separating the molecules (stationary phase) is packed with a porous bead-like crosslinked polymer network of styrene-divinylbenzene copolymer with closely controlled pores of variable sizes that can separate molecules in a particular molecular weight range. Depending on the size and shape, solute molecules may be able to enter the pores of the stationary-phase particles. Molecules larger than the pores will be totally excluded and will elute first. Very small molecules can enter every pore and permeate well into the stationary-phase particles. These are retained most and hence appear last in the chromatogram. Intermediate-size molecules elute at times depending on their comparative size. The size-separated molecules are detectedâtypically by either a differen- tial refractive index (ÎRI) detector or an ultraviolet detectorâand recorded according to their concentration. Through calibra- tion with molecules of known molecular weight, hydrodynamic volumes are converted to MWs and various MW parameters for the sample are calculated from the MW concentration data. GPCâs ability to separate mixtures by molecular size rather than by some complex property such as solubility or absorptivity is one of the great advantages of the technique. This feature has made GPC a useful alternate technique for fractionating complicated mixtures, such as crude oil residua, asphalts, and asphaltenes, for nearly 50 years (Altgelt 1965; Dickie and Yen 1967; Synder 1969; Yapp et al. 1991; Jennings et al. 1993). GPC ascertains the quantitative distribution of all species present in a binder, including maltenes, asphaltenes, and poly- mers. The detector signalâthat is, the difference between the refractive indices of the eluting solution containing the asphalt and that of the solvent (ÎRI)âis plotted versus the eluting volume (milliliters). The process allows the differentiation of asphalt species over a molecular weight range of 106â102 Daltons. A set of gel permeation columns is selected to optimize the separation of asphalt samples. Choice of the solvent system is of great importance, particularly with a complex material such as asphalt. The fractionation parameters in a given solvent include the sample concentration, temperature, injection volume, and flow rate. All these parameters affect the column performance and determine the separation efficiency of a given column set. The solvents most widely used in asphalt analysis are THF and toluene; the associated asphalt components remain unal- tered in toluene but THF breaks up some association (Jennings et al. 1992b). The key factors impeding an efficient separation are the tendency of polar materials in asphalt to associate or to be adsorbed on the column. To a lesser extent, but still important, is the fact that selective solvent interactions can affect the apparent hydrodynamic volume. For instance, associating substances, such as asphaltenes, show much higher molecular size in a poor solvent owing to the presence of large complexes. Adsorption of polar species leads to slow desorption from the column (tail- ing) and distorts the column calibration. Asphalt from a given source of crude oil has its own characteristic chromatogram that changes only slightly with grade. For this reason, GPC is a very effective tool for detecting changes in asphalt as a result of processing changes, crude source, or con- tamination. Each asphalt chromatogram exhibits its characteristic shape, but some of the asphalts show considerable seasonal change, probably reflecting asphalt processing changes (Glover et al. 1987). It must be emphasized that characterizations of this kind require that all GPC parameters be held constant. This is a major disadvantage, making comparisons difficult between laboratories and even over time. An asphalt standard should be run periodically to confirm constant operating parameters. Fractionation by Molecular Size (Hydrodynamic Volume) A correlation of the eluting volume with the relative molecular weight of the eluting fraction is difficult to achieve. Because GPC responds directly to apparent molecular size, it appears to be a simple method for obtaining the molecular weight dis- tribution of asphalt. However, it turns out not to be a straightforward determination for a number of reasons. The first is the tendency of some asphalt fractions to associate in solution. These same fractions also may tend to be adsorbed in the column. A final factor is the chemical complexity of asphalt. It is well known that the order of elution of polar and nonpolar compounds can be considerably altered by changing solvents, so it is difficult to choose calibrating compounds for such a complex mix- ture. A detailed evaluation of this calibration problem has been summarized (Davison et al. 1995). The chief utility of GPC in molecular size distribution measurements is not to obtain absolute values of molecular weight but to determine apparent molecular sizes that measure the degree of association in asphalts of different properties and composition, particularly to note the changes that occur during aging. It is likely that the effect of solvent power on the change in apparent molecular size car- ries information about the internal stability of the asphalt. Thus, a practical approach is to assign apparent molecular weights relative to commercially available polystyrene standards. In most of the reported GPC studies, the chromatogram is divided into three or more equal slices defined as large molecular size (LMS), medium molecular size (MMS), and small molecular size (SMS) (Brule et al. 1986; Price and Burati 1989; Yapp et al. 1991; Lee et al. 2008). It is also reported that the LMS region can be correlated with the binderâs physical properties and
31 field performance (Glover et al. 1987). Many of these reports have not presented the range of apparent MW of these fractions relative to polystyrene or other standards, which can reflect the separation efficiency of the column set. A detailed compilation of conditions reported for GPC determination of asphalts has been reported (Davison et al. 1995), and a standard practice for determining MW averages and molecule weight distribution of hydrocarbons and terpene resins by size exclusion chroma- tography (ASTM D 6579) can be applied to asphalts. A correlation of the eluting volume with the apparent molecular weight of the eluting fraction can be achieved using narrow-MW polystyrene standards (Daly et al. 2013). The apparent molecular weights of the eluting fractions, the GPC chromatograms, have been divided into three regions: LMS, MMS, and SMS based on elution volume. Researchers have stated that the LMS and SMS regions are significant with respect to predicting pavement performance (Garrick and Wood 1986; Garrick 1994; Wahhab et al. 1996, 1999). The impact of asphalt aging on the GPC chromatograms has been reported extensively (Kim and Burati 1993; Churchill et al. 1995; Siddiqui and Ali 1999; Negulescu et al. 2006; Shen et al. 2006; Lee et al. 2008; Xiao et al. 2009). Although the division of the chromatograms into arbitrary regions based on elution volume is used routinely, it is preferable to calibrate the GPC chromatograms and identify the maltenes, asphaltenes, and polymer com- ponents on the basis of their apparent MW ranges (Daly et al. 2013). By using MW regions, it is possible to divide the LMS fraction into ranges that change when the asphalt ages or is modified. To define LMS, MMS, and SMS fractions, the chromatogram was divided into three slices based on the apparent MW of the eluting species using a calibration curve from polystyrene standards (Daly et al. 2013). The three fractions are polymers (MW greater than 19,000), asphaltenes (MW from 19,000 to 3,000), and maltenes (MW less than 3,000), as shown in Figure 12. Quantitative data can be obtained by determining the area under the curve. As asphalt ages, the asphaltenes aggregate and begin to contribute to the polymer fraction, so the increase in the concentration of materials in the polymer fraction reflects the degree of aging. Therefore, the polymer region was subdivided into three fractions: very high molecular weight, with MW greater than 300,000 Daltons; high molecular weight, with MW between 45,000 and 300,000 Daltons; and medium molecular weight, with MW between 19,000 and 45,000 Daltons. The polymers added to modify the asphalt binders generally exhibit MWs above 45,000 Daltons so the medium molecular weight subfraction reflects aggregates formed during binder aging. Deconvolution of the chromatogram facilitates quantitative analysis by integration of the area under the curves shown in Figure 13. FIGURE 12 Molecular weight zones assigned in PMAC gel permeation chromatography (GPC) chromatogram (Source: Daly et al. 2013). Earlier determinations by osmometry indicated that the average MW of maltenes (as heptane soluble binder fraction) is 700â900 Daltons and that of asphaltenes (as heptane insoluble binder fraction) ranges between 2,000 and 10,000 Daltons (Dickie and Yen 1967). Morgan et al. (2010) used laser desorption mass spectrometry along with size exclusion chroma- tography and planar chromatography to study the MW of maltenes and asphaltenes of Mayan crude oil. The fractions were separated using extraction with pentane. The results revealed a small portion of asphaltenes extending to 10,000 Daltons, and
32 maltenes extending to 2,000 Daltons, but continuing to fall within the polymer ranges defining the maltene and asphaltene regions of the chromatograph. These MW data have been confirmed by the GPC method (Daly et al. 2013). The polymer and asphalt components of polymer-modified asphalt cements could be separated completely with accurate determination of the molecular weight of species by calibration with standard narrow-MW polystyrenes (Figure 14). FIGURE 13 Determination of maltenes and asphaltenes content of PG 64-22 binder by deconvolution of the GPC curve (Source: Daly et al. 2013). FIGURE 14 GPC elution curve of PG 70-22M containing 1% polymer extracted from a mixture aged for 5 days at 85ÂºC (Source: Daly et al. 2013). Correlating Physical Properties with GPC Results Attempts to correlate asphalt physical properties with chemical properties have not been particularly successful. This no doubt is primarily the result of the lack of uniqueness in the chemical properties. For instance, a Corbett fraction from one asphalt may have very different physical properties from those of the same fractions from another asphalt. Also, two asphalts with similar physical properties can have radically different GPC chromatograms (Davison et al. 1995).
33 Elseifi et al. (2010) studied the relationship between asphalt binder deformation properties at intermediate and low tem- peratures, its molecular composition, and mix performance. Nine straight-run binders obtained from two asphalt suppliers were tested using ductility and direct tensile tests. To assess the results of these tests, selected asphalt binders were evaluated using GPC, DSC, and dynamic mechanical analysis. Measurements showed that an inverse correlation exists between binder ductility at intermediate temperatures and failure strain at low temperatures. In other words, a binder that provides high duc- tility at intermediate temperatures would be characterized by poor elongation properties at low temperatures. This trend was related to the binder molecular composition, as characterized by GPC. An increase in the binder content of low MW results in an increase in binder ductility at intermediate temperatures. If the aggregate is heated higher than the specified level, the asphalt binder in the mixture will be aged at a much higher level during the short-term oven aging (STOA) period. The asphalt binder in that mixture will be oxidized (aged) more than expected during STOA time because of the highly elevated aggregate temperature. If the binder in the mixture is severely oxidized, the asphalt pavement will have a diminished service life. A gel permeation chromatography technique was used on the mixture particles without binder extraction to estimate the significance of aging for each case of STOA (Kim et al. 2016). Churchill et al. (1995) evaluated three different slice profiles (three, four, and 10 slices) to correlate aging times with the GPC LMS region. Partitioning of chromatograms into 10 segments provided a better resolution for predicting aging viscosity, aging index, viscosity number, and penetration from GPC data. It was also found that the aging level of binders in the warm-mix asphalt mixture was significantly lower than that of binders in hot-mix asphalt (Kim et al. 2016). GPC chromatograms were divided into 13 slices to evaluate relative impact of an RTFO test versus STOA on the properties of nine binders and six field-aged samples (Lee et al. 2008). The LMS (slices 1â5) ratio of aged to unaged samples consistently coincided with the aging times. The RTFO method was found to induce less aging than the STOA method of aging asphalt mixtures in the laboratory. Estimation of binder viscosity of recycled asphalt pavement (RAP) could be accomplished using GPC by direct sampling. The results exhibited a positive correlation with binders from corresponding samples recovered by the Abson method (Kim et al. 2006). A correlation of multiple stress creep recovery (MSCR) parameters with elastic recovery and molecular weight of differ- ent polymer-modified binders was demonstrated (Batten et al. 2011). The MSCR test that was recently developed by FHWA enables the correct grading of the field performance of polymer-modified asphalts. The polymer modification of asphalts is one of the solutions to overcome factors leading to asphaltâs demise; that is, the corresponding asphalt concrete defects such as rutting, thermal cracking, fatigue, and stripping. However, the different polymer-modified asphalts can behave differently even though they have the same performance grade. The MSCR test measures the high temperature binder specification parameter, called nonrecoverable creep compliance or Jnr, and percent recovery. In this study eight different polymer-modi- fied binders are measured. The molecular weights were determined using GPC. The correlation of the MSCR results with the molecular weight provides an insight into how modification affects mechanical response and the rutting potential. Mathematic models have been generated to predict rheological properties of asphalt cement based on the molecular size of its components. A set of polymer-modified asphalt samples with 5%, 10%, and 15% crumb rubber and 3%, 6%, and 9% styrene-buta- diene-styrene was prepared to evaluate the effect of polymer modification on the molecular size distribution of the asphalt samples. In general, the distribution of increasing molecular size could be correlated with changes in the asphalt binder physical properties. Predictive models could be generated for the different rheological properties from GPC chromatograms (Wahhab et al. 1999). NUCLEAR MAGNETIC RESONANCE ANALYSIS Proton nuclear magnetic resonance (1H NMR) spectroscopy has emerged as a very powerful and versatile tool for bitumen characterization (Jennings et al. 1992b; Borrego et al. 1996; Jain et al. 1998). 1H NMR allows investigations in solids as well as in solution. As an alternative to the conventional analysis, the 1H NMR spectrometry method does not require sample pre- treatment and thus considerably reduces manipulation time. The 1H NMR method is also capable of simultaneously detecting and quantifying a number of constituents in a single spectrum. The direct 1H NMR spectrometry quantitative method presents advantages over some routine methods: simplicity, rapidity, selective recognition, and quantitative determination of aliphatic hydrogens and aromatic hydrogens in bitumen. Asphalt Molecular Structure Ramsey et al. (1967) published the first structural characterization of asphalts by 1H NMR. Hasan et al. (1983) published a more structured analytical approach to characterize a petroleum vacuum distillation residue (boiling point >454Â°C); later,
34 this methodology was slightly modified by Siddiqui and Ali (1999) and Siddiqui (2010). Since then other authors have used the 1H NMR spectrum, sometimes the carbon-13 nuclear magnetic resonance (13C NMR) spectrum, and elemental analysis for the structural characterization of asphalt fractions or similar materials to define indices that intend to correlate with the properties and future behaviors of asphalts in roadworks (Huang 2010; Ma et al. 2011). Betancourt Cardoza et al. (2016) improved the characterization of some heavy fractions of petroleum based on elemental analysis, 1H NMR, and 13C NMR spectra of some heavy fractions of petroleum. They proposed a methodology useful to characterize oil, coal, or its fractions using a method of data processing and spectra interpretation that benefited from recent advances in NMR (in hardware and software). Three unfractionated fresh commercial asphalts named P1, P2, and P3 produced in Colombia were examined. The viscos- ity, the penetration (80/100), the softening point, and the colloidal instability index (Table 10) indicate that P1 and P2 have a softer consistency than P3. The three asphalts show high thermal susceptibility, with P3 being the most sensitive (penetra- tion index â1.4). This result agrees well with the common origin. The SARA fractionation data (Table 10) suggest that their higher content of resins and their smaller fraction of aromatics justify the softer consistency of P1 and P2 compared with P3 (Betancourt Cardozo et al. 2016). TABLE 10 PHYSICAL-CHEMICAL PROPERTIES OF FRESH AND UNFRACTIONATED ASPHALTS Samples Assay Units P1 P2 P3 Conventional assays viscosity at 60Â°C Poises 1403 1263 2193 Viscosity at 135Â°C Poises 3.10 2.89 4.23 Viscosity at 150Â°C Poises 1.27 1.17 1.46 Ductility at 25Â°C cm >150 >150 >150 Penetration at 25Â°C 0.1 mm 84 93 64 Specific gravity 1.004 1.006 1.008 Softening point Â°C 45.8 44.4 47.0 Colloidal instability index 0.36 0.36 0.38 SARA separation Saturatesa % 14.18 13.86 13.54 Aromaticsb % 39.59 39.59 43.43 Resinsc % 33.69 33.06 29.03 Asphaltenesd % 12.53 13.48 13.99 Elemental analysis carbon % 85.69 85.73 85.84 Hydrogen % 10.50 10.45 10.42 Nitrogen % 0.94 0.99 1.02 Sulfur % 1.64 1.65 1.65 Oxygene % 1.24 1.19 1.07 Source: Betancourt Cardozo et al. (2016). Uncertainty of a Â± 0.50; b Â± 0.83; c Â± 0.54; d Â± 0.67. e Measured values, not deducted as the difference from 100%. The elemental composition (Table 10) and the 1H NMR spectra (Table 11) show very similar quantities of hydrogen in the different structural fragments (aliphatic, aromatic, olefinic, or phenolic) of P1, P2, and P3. Although all these data do not discriminate clearly between the three asphalts, they evidence a predominance of alkyl substituents with a length of four or more carbon atoms in their average structure. Most of the hydrogen in its aliphatic fraction belongs to CHn moieties located at three or more bonds from the aromatic fragment. In total, more hydrogen was found bonded to carbons in positions Î², Î³, Î´, or farther (HÎ², HÎ³, HÎ´, ...) than bonded to carbons in positions Î± to aromatic fragments, HÎ±. 13C NMR spectra (Table 12), viscosity, penetration, and the softening point (Table 10) confirm some degree of similitude between P1 and P2, and show differences between them regarding P3. Table 10 data show that the content of carbon bonded to hydrogen present as methyl, methylene, or methine in the structure of P1, P2, or P3 is quite similar. The quaternary carbon content in alkyl or aryl fragments is practically equal in the structure of P1 or P2. P3 has the lowest content of quaternary
35 carbon in its aliphatic fragments and the highest content of quaternary carbon in polycondensed aromatic fragments (most of these are catacondensed carbons bonded to a heteroatom or an alkyl group distinct to methyl). TABLE 11 HYDROGEN PERCENTAGE DETERMINED BY 1H NMR SELECTED ASPHALTS Samples Type (assignments) Interval (ppm) P1 P2 P3 General aliphatic or saturatesa 0.5â4.5 94.48 94.33 94.35 Olefins 4.6â6.2 0.26 0.21 0.21 Aromatic 6.3â9.3 5.23 5.44 5.43 Phenolic OH 5.0â9.0 0.02 0.02 0.02 Particular undefined CHn Î³, Î´ or more to aromatic 0.5â2.0 46.04 46.08 45.92 CHn Î², Î³, Î´ or more to aromatic 0.5â2.0 81.37 81.26 80.95 CHn Î² to aromatic 0.5â2.0 35.32 35.19 35.04 CHn Î± to aromatic 2.0â4.5 13.10 13.05 13.38 CHn Î± to Câ«C (it is CHnâCâ«C) 1.9â2.1 0.01 0.01 0.01 CHn in a C sp 3 joined to oxygenb 3.1â3.3 0.01 0.01 0.01 CH3, CH2 or CH Î³, Î´ or more c 0.5â1.0 16.08 16.34 16.32 CH3-Î², CH2-Î² or Î³, CHâÎ² or Î³ d 1.0â2.0 59.93 59.48 59.19 CH in monocyclic aromatice 6.3â7.3 1.85 2.07 1.94 CH in polycyclic aromatic 7.2â9.3 3.38 3.37 3.49 a Paraffins and naphthalenes b CH n-C-O cTo aromatic d To aromatic and CH2 or CH e Estimated by excess (6.3â7.3). Source: Betancourt Cardozo et al. (2016). TABLE 12 CARBON PERCENTAGE DETERMINED BY NMR IN SELECTED ASPHALTS Type (assignment) Interval (ppm) Samples P1 P2 P3 General aliphatic or saturates 10â60 70.5 71.35 68.11 Olefinic 105â153 1.38 1.05 1.14 Aromatic 102â165 28.12 27.6 30.74 Carbon types Total CH3 24â28 2.69 2.66 2.66 Total CH2 23â60 8.27 8.17 8.13 Quaternary sp3 10â60 56.09 57.10 53.91 Total aromatic CH 102â131 1.55 1.6 1.59 Quaternary sp2 123â165 26.57 26 29.15 Source: Betancourt Cardozo et al. (2016). Using 1H and 13C NMR can yield information on average structural parameters of asphalt and asphaltenes, such as percent- ages of aromatic carbons, aliphatic carbons, bridged carbons, methyl carbons, ring carbons, naphthenic carbons, paraffinic chain lengths, and other parameters. The specific environment of the different types of hydrogens and carbons can be defined (Betancourt Cardozo et al. 2016). Thus, NMR spectroscopy is a powerful tool for predicting the structure of complex organic molecules. Mechanisms of Asphalt Aging Combining NMR and GPC information, Saddiqui suggested possible structures for asphalt and mechanisms of aging (Sid- diqui and Ali 1999; Siddiqui 2010). This generic model incorporates aliphatic branch structures elaborating pericondensed rings with low ratios of hydrogen to carbon. The presence of heteroatoms in the rings is shown in Figure 15.
36 FIGURE 15 Possible representation of reaction types in a hypothetical asphalt structure on laboratory aging (Source: Siddiqui and Ali 1999). In this work (Siddiqui and Ali 1999), the chemical properties of a commercial grade Saudi Arabian asphalt procured from Ras Tanura refinery were evaluated. The RTFO short-term aging and PAV long-term aging tests were used to simulate the laboratory aging of this asphalt. PAV has more severe effects on the chemical properties of asphalt than does the RTFO method. The Corbett fractionation procedure was used to separate fresh and aged asphalts into four generic fractions; namely asphaltenes, polar aromatics, naphthene aromatics, and saturates. Various analytical techniques were applied to evaluate the chemical changes that occurred during the aging processes. High pressure-gel permeation chromatography molecular weight and size distributions suggested that molecular rearrangements occur predominantly on aging. Based on the molecular size distribution of isolated asphaltenes, it is suggested that dissociation, isomerization, and fragmentation were predominant reac- tion types that occurred in asphaltenes during extensive PAV laboratory aging. Further chemical reactions, such as molecular association, polymerization, or polycyclic condensation within the complex asphaltenes, were inferred from NMR studies of extensively aged asphaltene fractions (Siddiqui 2010). X-RAY FLUORESCENCE SPECTROSCOPY X-ray fluorescence spectroscopy (XRF) is a simple technique for quantitative analysis of elements ranging from sodium to uranium in the periodic table. The XRF method is widely used to determine the elemental composition of materials. Because this method is fast and nondestructive to the sample, it is often used in field and industrial applications for quality control. Portable XRF devices facilitate measurements in the field. The heavy metal content of an asphalt sample can be used to iden- tify some asphalts. XRF can detect other characteristic elements in binder additives such as phosphorus (P) in polyphosphoric acid (PPA); calcium, zinc, and molybdum in recycled engine oil bottoms (REOB); and crumb rubber modifier (CRM). The
37 use of XRF for forensic analysis of HMAs is facilitated by the ability to determine the concentration of multiple elements simultaneously. XRF is the emission of characteristic âsecondaryâ (or fluorescent) X-rays from a material that has been excited by a bom- bardment of high-energy X-rays or gamma rays. When materials are exposed to short-wavelength X-rays or gamma rays, their component atoms may ionize. As a result, the material emits radiation at the energy characteristic of the atoms present. The term âfluorescenceâ is applied to phenomena in which the absorption of high-energy radiation results in the re-emission of lower-energy radiation (Jenkins 2012). The wavelength of this fluorescent radiation can be calculated and analyzed either by sorting the energies of the photons (energy-dispersive analysis) or by separating the wavelengths of the radiation (wavelength- dispersive analysis). Once sorted, the intensity of each characteristic radiation is directly related to the amount of each element in the material. Asphalt contains high levels of sulfur and varying amounts of iron, vanadium, and nickel. Zinc compounds are added to control hydrogen sulfide emissions in the refinery. The distribution of the heavy metal elements in unmodified binders is a fingerprint that can be used to identify their corresponding crude oil sources. AC does not naturally contain phosphorus. If it is assumed that all the phosphorus comes from PPA, measurement of the phosphorus content using XRF can estimate the amount of PPA used to produce the binder. However, other phosphorus-con- taining additives such as REOB can distort the results (Reinke and Glidden 2010). Waste engine oil contains copper and iron (wear metals) as well as zinc and molybdenum from anti-wear and corrosion-inhibiting oil additives. The phosphorus source in lube oil is zinc thiophosphate. Zinc and molybdenum are not found in natural asphalt, therefore XRF analysis of these ele- ments can help identify the presence of REOB. Quantifying REOB content can be achieved by determining the intensity of those elements and correlating that intensity to a known dosage of REOB with a calibration standard (Clifton et al. 2016). Ground tire rubber (GTR) also contains some of the metals (zinc and iron) found in REOB, so analysis of REOB in crumb rubber-modified binders is quite complicated since the levels of zinc in GTR are much higher than those in REOB (Arnold and Shastry 2014). If no REOB is present there is a linear correlation between the GTR content and zinc (Reinke and Glidden 2010). FIGURE 16 Schematic of an AFM scanning an asphalt surface (not to scale). ATOMIC FORCE MICROSCOPY In simple terms, the atomic force microscope has a nanometer-sized tip at the end of a cantilever that probes the surface of the material (typically over an area of 10â50 micrometers x 10â50 micrometers) recording the topography of the surface and changes in the surfaceâtip interaction. Figure 16 shows a schematic of an atomic force microscope (not to scale) scanning the surface of an asphalt binder. For imaging, the reaction of the probe to the forces that the sample imposes on it can be used to form an image of the three-dimensional shape (topography) of a sample surface at a high resolution. This is achieved by raster scanning the sample surface and recording the height of the probe that corresponds to a constant tipâsample interaction. The surface topography is commonly displayed as a pseudocolor plot. An AFM image is a simulated image based on the height of
38 each point of the surface, and in fact each point (x, y) of the surface has a height h (x, y). The color variations reflect the rela- tive heights of the surface topography. The impact of asphalt chemical composition on the microstructure and performance characteristics of asphalt binders was studied. The methods implemented included adsorptionâdesorption chromatography analysis and a range of AFM and chemical force microscopy techniques. AFM can be a powerful complementary tool to rhe- ology and spectroscopy for characterizing asphalts. It is possible to compare the mechanical properties of pure and modified bitumen using appropriate techniques. Asphalt Microstructure An intriguing AFM image of an asphalt surface is a peculiar âbeeâ structure that was initially found only for a âgelâ bitumen. The phase images, 25 Î¼m by 25 Î¼m, of one of the asphalts are shown in Figure 17, where the continuous phase is represented with white color contrast. The impact that each SARA chemical fraction has on asphalt phase structuring, most notably the well-known asphalt bee structures, is shown. Certain asphalt chemical parameters have a consistent and measurable effect on the asphalt microstructure that is observed with AFM. Particular microstructures that emerged through chemical doping were then discovered to have unique chemical polarity, which explicitly impacts the durability and performance of asphalt. A surprising correlation was found between the saturates chemical parameter and the effects of oxidative aging on asphalt behavior (Allen et al. 2014). FIGURE 17 AFM phase images of 25 Î¼m by 25 Î¼m of Asphalt ARC BI0002 derivatives taken at room temperature (25Â°C): (a) control; (b) asphaltene-doped blend; (c) naphthenic aromatic-doped blend; (d) polar (resin)-doped blend; and (e) saturate- doped blend (enlarged locations are 10 Î¼m by 10 Î¼m for each asphalt blend) (Source: Allen et al. 2014).
39 The same bee structure (also called catana phase) was repeatedly observed in other works with an average height between 22 and 85 nm and a typical distance between strips on an order of 150 nm (Masson et al. 2006). A link between the extent of the bee phase and the asphaltenes has been confirmed in one study (Masson et al. 2007). The scanning electron microscopy observation of the same gel bitumen that gave the bee structure in AFM showed connecting aggregates of what was believed to be asphaltene particles with a diameter of around 100 nm (Loeber et al. 1996). Several studies using the AFM and mathematical modeling at a similar scale have revealed that the binder has a distinct microstructure that is related to its chemical composition and thermal history (Masson et al. 2006; Schmets et al. 2010; Das et al. 2013; Allen et al. 2014; Rebelo et al. 2014; Bhasin and Ganesan 2015; Zhao et al. 2015; Menapace et al. 2015). Studies have also shown that such a structure can influence the nucleation of damage within the binder (Jahanigir et al. 2015). Although most of the aforementioned studies have been limited to the surface of the binder, some studies have also demonstrated that structures observed on the surface are correlated with similar, albeit smaller-sized, structures (referred to as ant-like struc- tures in some literature) below the surface in the sample mass (Fischer and Dillingh 2014; Ramm et al. 2016). For example, Ramm et al. (2016) used optical techniques to demonstrate that (1) these structures exist in the bulk of the binder; (2) for a given chemical composition and temperature, the thermal history of the binder can be manipulated to alter the size and dis- tribution of the structures; and (3) the structures influence the rheological properties of the binder, such as complex modulus (e.g., in some cases the G* of the binder varied by 20% at the same temperature, composition, and rate of loading, simply by manipulating the thermal history and concomitant internal structure). Property Changes at a Microscopic Level AFM results were presented for a base binder and the mastics prepared with granite, portland cement, and hydrated lime. A clear identification of three micro domainsâparaphase, periphase, and beesâwas displayed. It was apparent that the rela- tive proportion and size of the features changes with the addition of filler and appears to be filler-specific, providing evidence of physico-chemical interactionâinduced changes to the effective binder matrix. Image analysis was used to determine the relative surface area each micro domain occupies for the materials studied. The size decreased and relative number of bee structures increased with the addition of filler. The relative change in the size, relative quantity, and special distribution of bee features is greatest for the mastic prepared with lime, indicating a higher degree of physico-chemical interaction intensity. This was expected because the lime filler has a much higher specific surface area than the other fillers and hence, given the same volume concentration, provided more opportunity for adsorption. A greater stiffening effect was observed for fillers that induce the greatest microstructural changes as a result of physico-chemical interaction. Physico-chemical interaction is anticipated to lead to a softening of the effective asphalt binder matrix since polar components adsorb to the surface of the filler. Thus, results suggest the formation of an adsorbed interphase layer plays a more critical role in determining the macro- scopic rheology of mastics than does softening of the binder matrix (Davis and Castorena 2015). Conventional rheological and chemical tests provide a global view of asphalt property and composition changes upon aging, but offer little details on changes at the microscopic level. Using AFM, Yuhong Wang et al. (2015) analyzed the micro- mechanical properties of five asphalts with different aging conditions. Aging was found to significantly increase the spatial variations of the sample properties. The asphaltene content and the size of microstructures both appear to affect the microme- chanical properties of the binders. Using two different antioxidant additives, PPA and cashew nut oil, at concentrations of 1% and 2%, respectively, it is possible to use the AFM technique to distinguish the effects of both additives in the morphological and micromechanical properties of bitumen films (Rebelo et al. 2014). Moisture damage in polymer-modified asphalts has been studied for decades, yet the effects of chemical functional groups on moisture sensitivity are not known. A nanoscale experiment was conducted to measure these effects in terms of adhesive/ cohesive forces using AFM. A base asphalt binder and asphalts modified by two polymers (styrene-butadiene and styrene- butadiene-styrene) were tested in dry and wet conditions. Using the AFM, these samples were probed by silicon nitrite (Si3N4), carboxyl (-COOH), methyl (-CH3), and hydroxyl (-OH) functionalized AFM tips, and nanoscale pull-off or adhe- sion/cohesion forces between asphalt and tip molecules were measured. Based on the ratio of wet to dry adhesion/cohesion forces, it was shown that the polymer modification makes binders less susceptible to moisture damage. The moisture perme- ability is impacted by the polymer molecular weight and compatibility with the binder. Among the four tips, the -COOH tip showed almost no difference in adhesion forces between wet and dry samples. Using -OH tips shows that the cohesion in SBS- modified wet asphalt samples is significantly higher than the cohesion in SB-modified wet asphalt samples. The Si3N4 tip showed higher adhesion in SB-modified wet samples than in the SBS polymer-modified wet samples. Based on the adhesion/ cohesion force, 3% polymer is found to be optimum for minimizing moisture susceptibility with both SB and SBS polymers. This study illustrates the potential for nanoscale AFM testing on asphalt binders (Tarefder and Zaman 2010).
40 Worldwide, several research groups are reporting AFM results on bitumen, so it is becoming important to improve the understanding of the reproducibility and objectivity of the technique for studying bituminous samples. When reproducibility and stability are proven, AFM can be a tool for asphalt professionals to rapidly screen bituminous binders. In this context, two independent laboratories have developed a standard method for preparing and conditioning bitumen for AFM imaging. By means of an interlaboratory comparison of independently imaged specimens, the reproducibility of microstructure mea- surements was investigated. A quantitative comparison on different microstructures was developed, and the consistency of independently obtained results was confirmed. The results from both labs were comparable: the microstructural properties were found to be randomly distributed within a 5% interval. Also, the influence of temperature on the microstructure was demonstrated to be reproducible and consistent. With the increase of temperature, the microstructure gradually disappeared; however, traces of the microstructure remained visible up to the highest measurement temperature of 60Â°C. The conclusion is that given well-defined sample preparation and measurement procedures, the microstructure of bitumen can be reproducibly imaged by AFM from room temperature up to temperatures where bitumen becomes liquid (Nahar et al. 2013). The current understanding of bitumenâs surface microstructures as characterized by AFM is evolving. Microstructures of bitumen develop in different forms depending on crude oil source, thermal history, and sample preparation method. Although some bitumens display surface microstructures with fine domains, flake-like domains, and dendrite structuring, bee struc- tures with wavy patterns several micrometers in diameter and tens of nanometers in height are commonly seen in other bind- ers. Controversy exists regarding the chemical origin of the bee structures, which has been related to the asphaltene fraction, or the crystallizing waxes in bitumen. The rich chemistry of bitumen can result in complicated intermolecular associations, such as co-precipitation of wax and metalloporphyrins in asphaltenes. Therefore, it is the molecular interactions among the different chemical components in bitumen, rather than a single chemical fraction, that are responsible for the evolution of bitu- menâs diverse microstructures, including the bee structures. Mechanisms such as curvature elasticity and surface wrinkling that explain the rippled structures observed in polymer crystals might be responsible for the formation of bee structures in bitumen. Despite progress on morphological characterization of bitumen using AFM, the fundamental question of whether the microstructures observed on bitumen surfaces represent its bulk structure remains to be addressed. Combining AFM with other chemical analytical tools that can generate high resolution comparable to AFM would provide an avenue to linking bitumenâs chemistry to its microscopic morphology and mechanical properties and consequently benefit the efforts focusing on developing structure-related models for bituminous materials across the different-length scales (Yu et al. 2015).