Supercritical Fractions as Asphalt Recycling Agents and Preliminary

Jan 1, 1997 - Jay M. Chaffin, Meng Liu, Richard R. Davison,* Charles J. Glover, and. Jerry A. Bullin. Department of Chemical Engineering and The Texas...
0 downloads 0 Views 371KB Size
Download PDF
656

Ind. Eng. Chem. Res. 1997, 36, 656-666

Supercritical Fractions as Asphalt Recycling Agents and Preliminary Aging Studies on Recycled Asphalts Jay M. Chaffin, Meng Liu, Richard R. Davison,* Charles J. Glover, and Jerry A. Bullin Department of Chemical Engineering and The Texas Transportation Institute, Texas A&M University, College Station, Texas 77843-3122

Several asphalts were fractionated using supercritical pentane. These fractions were analyzed by gel permeation chromatography and high-performance liquid chromatography, and their viscosities were measured. The properties of these fractions vary not only among the fractions of a given asphalt but also for the same fraction produced from different asphalts. These widely varied fractions previously have been shown to have potential for reblending to produce superior asphalts. This study investigates the potential for using some of the fractions as asphalt recycling agents. A modified strategic highway research program (SHRP) pressure aging vessel (PAV) test and kinetics studies were conducted on nine recycled asphalts and the original asphalt. The aging indexes of eight of the recycled asphalts are superior to the aging index of the original asphalt. Two of the blends using industrial supercritical fractions and the three blends using laboratory supercritical fractions have lower aging indexes than blends using commercial recycling agents. The kinetics investigation also indicates that at road conditions the recycled asphalts will harden more slowly than the original asphalt. The degree of hardening for a given amount of oxidation in the recycled binders was found to be a strong function of the total saturate content in the recycled binder. Introduction There are two general approaches to upgrading the residues from crude oil distillation: chemical processing (such as visbreaking, catalytic cracking, and coking) and physical processing (such as propane deasphalting and supercritical extraction). Of these alternatives, supercritical fractionation is the least commonly used process. The two most widely known supercritical fractionation processes are the residuum oil supercritical extraction (ROSE) process licensed by M. W. Kellogg (formerly by Kerr McGee) and the Demex process licensed by UOP. These physical separation processes are capable of producing an asphaltene-rich fraction, or “asphaltenes” and a saturate-rich fraction known as deasphalted oil, or DAO. Some of the commercial units are also set up to produce a third, intermediate fraction known as resins. The industrial supercritical units utilize propane, n-butane, i-butane, n-pentane, i-pentane, or a mixture of these as the solvent. The products from the ROSE units include lube oils, FCC feedstock, fuel oils, solid fuel, and asphalt. The estimated total capacity of the ROSE units, as of July 1994, was over 200 000 barrels/stream day. Because the ROSE and Demex processes are patented and licensed trade secrets, little if any data from industrial operation are available in the literature. Fortunately, some data are available from the pilot-scale unit at Texas A&M University (TAMU). Stegeman et al. (1992) utilized supercritical n-pentane and combinations of n-pentane and cyclohexane at room temperature to fractionate asphalts. Jemison et al. (1995) used cyclohexane to perform an initial fractionation of the asphalt in the supercritical unit. The use of cyclohexane resulted in reduced selectivity and necessitated higher operating pressures and temperatures, leading to some thermal degradation of the asphaltic * Corresponding author. Telephone: (409) 845-3361. Fax: (409) 845-6446. S0888-5885(96)00443-5 CCC: $14.00

material. Jemison (1992) also fractionated reduced crudes using this methodology but found that the presence of the vacuum gas oils created cosolvency problems. While Stegeman et al. (1992) were mainly interested in characterizing supercritical fractions, Jemison et al. (1995) conducted aging studies to determine if asphalts with superior aging characteristics could be obtained by blending supercritical fractions. These aging studies were carried out in a pressure oxygen vessel (POV) described by Lau et al. (1992). Poor temperature control prevented determining accurate kinetics, but Jemison et al. (1995) were able to compare performance of the blended and original asphalts based on hardening susceptibility, HS (Lau et al., 1992), and oxidation rate at a given aging temperature. Briefly, the HS is a measure of the susceptibility of an asphalt to harden as a result of oxidation and is a good indication of an asphalt’s physicochemical properties. The HS is defined as the slope of the relationship between the logarithm of the limiting viscosity (η* 0) and the carbonyl area (CA) of the infrared spectrum for aged asphalts as shown in eq 1. The linear relationship between the two properties

(

)( )

d ln η* d ln η* 0 0 dCA ) ) HSRCA dt dCA dt

(1)

was first identified by Martin et al. (1990) when examining asphalt binders extracted from field-aged pavement samples. This relationship has subsequently been confirmed on laboratory-aged samples by other researchers and has been determined to be independent of aging temperature below 113 °C (Lau et al., 1992; Petersen et al., 1993). Obviously, a low value for the HS will help reduce the rate of increase of log η*0 (d ln η* 0/dt). Jemison et al. (1995) found that the HS values were greatly improved for the blended asphalts compared to the original asphalts but the rates of carbonyl formation (RCA) were not appreciably changed. However, the rates © 1997 American Chemical Society

Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 657

for blends made from one asphalt usually were different from the rates for blends made from another asphalt. The implications of this are that the supercritical fractions from an asphalt with a low HS could be blended with fractions from an asphalt with low aging rates (the two are not usually inclusive) to produce a superior asphalt. Although this is highly impractical, it does suggest another, more realistic possibility. A supercritical fraction could be chosen to improve the aging characteristics of a pavement currently in service, but needing repairs. At this point, a brief digression into the composition of asphalts is necessary. The well-known methods (Corbett, 1969; Rostler and Sternberg, 1949) for characterizing the composition of asphalts entail performing a binary fractionation of the material based on solubility in some arbitrary solvent, typically a saturated hydrocarbon (n-hexane, n-heptane, i-octane, etc.). The insoluble fraction, if present, is separated by filtration and is referred to as the asphaltene fraction, or simply asphaltenes. The asphaltenes have been shown to be responsible for the highly viscous nature of asphalts (Lin et al., 1995a,b; 1996). The methods differ in their treatment of the hydrocarbon-soluble fraction (maltenes), but both define a group of materials called saturates (Corbett, 1969) or paraffins (Rostler and Sternberg, 1949) which are highly unreactive and incompatible with the asphaltenes. As an asphalt ages, the saturates content remains constant while the asphaltenes content increases. This change in composition results in a decrease in compatibility. It is recommended that recycling agents contain no more than 30% saturates in order to produce good quality recycled asphalt mixtures (Epps et al., 1980). In addition, because an aged asphalt already contains too many asphaltenes, a rejuvenating agent should have few, if any, asphaltenes of its own to restore compatibility. The data reported by Stegeman et al. (1992) and Jemison et al. (1995) indicate that materials meeting these criteria can be produced through supercritical fractionation of petroleum residues. In previous experiments (Stegeman et al., 1992; Jemison et al., 1995; Jemison, 1992), several asphalts and reduced crudes were supercritically fractionated in the laboratory by using a combination of supercritical and room-temperature separations as well as by using a combination of supercritical solvents. These efforts primarily focused on the analyses of these fractions although some aging data were reported. This paper describes the products of fractionation of new asphalts and focuses on preliminary experiments investigating the potential use of supercritical fractions as recycling agents for aged asphalts. Materials Asphalts. Four different asphalts were fractionated in the supercritical pilot plant at Texas A&M University (TAMU) and analyzed in detail for this study. The first asphalt, an AC-20, was acquired from a local hot-mix contractor. This asphalt is identified as YBF. Because refineries adjust operating conditions frequently, as well as switch crude sources, and contractors switch suppliers, it was decided that asphalts whose crude source and processing could not change should be studied. The only asphalts that meet these criteria are the asphalts acquired in conjunction with the strategic highway research program (SHRP). These asphalts were sampled at one instant in time and can be considered the only

Table 1. Tank Asphalt Data asphalt

grade

YBF ABM-1 AAF-1 AAA-1

AC-20 AR-4000 AC-20 150/200 Pen

viscositya saturate asphaltene aromatic (wt %)d (wt %)e (dPa‚s)b (wt %)c 3100 2200 1950 915

9.0 6.8 12.1 10.8

17.6 6.7 15.5 19.0

74.4 86.5 72.4 70.2

a 60 °C low-frequency limiting dynamic viscosity η* b 0,60 °C. 1 dPa·s ) 1 P. c From HPLC calibration. d n-Hexane asphaltenes. e Determined by difference.

Table 2. Industrial Supercritical Fraction and Commercial Rejuvenating Agent Properties sample

viscositya (dPa‚s)b

saturate (wt %)c

asphaltene (wt %)d

aromatic (wt %)e

ISCF A ISCF B ISCF C CRA A CRA B CRA C

17.6 58.0 434.0 2.4 1.2 1.0

20.4 30.8 11.4 8.7 12.4 28.0

0.3 0.7 3.4 0.7 0.9 0.5

79.3 68.5 85.2 90.6 86.7 71.5

a 60 °C low-frequency limiting dynamic viscosity η* b 0,60 °C. 1 dPa‚s ) 1 P. c From HPLC calibration. d n-Hexane asphaltenes. e Determined by difference.

“standard” asphalts in the world. Furthermore, many of these asphalts have been studied extensively not only by our research group but throughout the world as well. As such, a large database exists for these asphalts. SHRP AAA-1, AAF-1, and ABM-1 were acquired from the SHRP material reference library. The properties of the “tank” (as received from the refiner) asphalts fractionated in this study are listed in Table 1. The asphalts chosen for fractionation in this study were picked to represent the widest variation in hardening susceptibility (HS). At the time these asphalts were chosen, the HS had been determined for more than 20 asphalts at aging pressures of 20.7 bar (300 psia) pure O2, with values ranging from 1.0 to 5.3. The 20.7 bar O2 HS for the three SHRP asphalts were determined to be 5.2, 3.8, and 1.0 for AAA-1, AAF-1, and ABM-1, respectively. SHRP AAF-1 was also aged in an airbubbling apparatus described by Chaffin et al. (1995) to produce the aged asphalt for the comparison of rejuvenating agents. Industrial Supercritical Fractions. Several refiners currently utilize supercritical operations to process various heavy-end feedstocks. Supercritical fractions (SCFs) were obtained from a few of these companies. These industrial SCFs (ISCFs) were analyzed, and three were chosen for a preliminary study to determine the potential for ISCFs to be used as recycling agents for aged asphalt binders. The three ISCFs investigated in detail have been given the arbitrary designations ISCF A, B, and C to protect the identities of the companies supplying material. The properties of these fractions are listed in Table 2. Two of the ISCFs are resin fractions (B and C), and one is a DAO (A). Even though ISCF B was identified by the manufacturer as a resin, its saturate content is significantly higher than would be expected of a resin fraction and its viscosity is a little lower than would be expected of a resin fraction. These observations are somewhat explained by the fact that the feedstock for this manufacturer is atmospheric tower bottoms, which contains many lighter, more saturated compounds than are usually present in asphaltic materials. Commercial Recycling Agents. Many commercial recycling agents (CRAs) are currently available. Most

658 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

Figure 1. TAMU supercritical unit process diagram.

of these CRAs were not developed specifically for asphalt recycling but are actually aromatic extract byproducts of lube oil manufacture and, thus, are high in saturate content. Several of these commercial recycling agents were obtained and analyzed. Three were chosen for comparison in this study. These three were obtained from different companies and have been given arbitrary designations to protect the identities of the manufacturers. The properties of CRA A, B, and C are listed in Table 2. Experimental Methods Supercritical Fractionation. The supercritical fractionation pilot plant at TAMU and its operation have been described previously (Stegeman et al., 1992; Jemison et al., 1995). However, a description of the operation of the unit at TAMU will be given here. Figures 1 and 2 illustrate schematically the SC unit. The solvent is pumped to the operating pressure in separators 1-3 (S1-S3) by the solvent metering pump (MP2). Once steady-state conditions are achieved, the asphalt metering pump (MP1) is activated, introducing feed material into the circulating solvent stream. The temperature is increased between each successive separator for S1-S3, reducing the density of the solvent and, thus, the solubility of the asphaltic components. Components of the feed precipitate when no longer soluble in the solvent. The insoluble material is transferred from the separator to its corresponding collector periodically to avoid potential plugging problems, and the soluble components are carried to the next separator. Finally, the overhead mixture from S3 passes through the control valve, where the pressure is reduced to a significantly subcritical value. At the gaseous conditions in S4, none of the asphaltic material is soluble in

Figure 2. Legend for TAMU supercritical extraction unit process diagram.

the solvent and complete solvent recovery is achieved. (Industrially this solvent purification is usually achieved by additional heating.) The solvent is then recycled back into the solvent reservoir (S0). Asphalt is fed in a semibatch manner for several hours. After the feed has been stopped, the experiment is complete when no additional material is transferred from the separators to the collectors in subsequent blowdowns. The fractionation conducted in this study differs from the previous experiments (Stegeman et al., 1992; Jemison et al., 1995) in that all of the fractionation was performed in the supercritical extraction unit using only n-pentane as the solvent. The four asphalts fractionated in this work were fractionated in two passes through the unit generally according to the fractionation scheme illustrated in Figure 3. The lightest fraction from the first pass was fed through the unit a second

Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 659

Figure 3. Typical fractionation scheme.

time, yielding eight fractions that may be analyzed. Fraction numbers were assigned according to distillation terminology where the lightest (lowest boiling) components have the lowest fraction numbers rather than according to the order in which the fractions are separated. Specifically, the lightest fraction from the second pass is designated as fraction F1 and the heaviest fraction from the first pass is designated as fraction F8 (usually fraction F5 is the feed material for the second pass through the unit). The operating conditions were chosen based on the results reported by Jemison et al. (1995) and are given in Table 3. Fraction yields for the four asphalts fractionated in this study are also listed in Table 3. YBF was the first asphalt fractionated. This asphalt was fractionated four times with the standard operating conditions shown in Table 3. One trial run was made using a lower temperature for the separation of F8, but the difference between cuts was not significant, so no further experimentation was performed. F5 and F6 from the four runs with the same operating conditions were combined and fed through the unit for a second pass. The operating conditions were controlled such that the temperature for separation of F4 was the same as the temperature for separation of F6. This way, F4 should contain all of the material from F6. In fact, second cut fractionation of the combined F5 and F6 material produced a much larger F4 than the sum of the F6 materials. This can be attributed to three possible causes: slightly different solvent:asphalt ratios between the F6 separation and the F4 separation which alter the equilibrium between phases, incomplete dissolution prior to the first separator, or increased holdup in the slightly larger separator 1 which allows for more time to achieve equilibrium in the separator. The separations with YBF suggested that only the F5 material need be refractionated. As such, the fractionation scheme shown in Figure 3 was utilized for the other three asphalts. The second asphalt fractionated was SHRP ABM-1. This asphalt was fractionated four times using the same operating conditions as those used for the initial fractionation of YBF. Most of the ABM-1 F5 was then fed through the unit in a single run for refractionation into fractions F4-F1. AAA-1 was fractionated four times using the standard operating conditions. However, two of these runs were aborted at various stages due to plugging problems. Although some F5 was obtained from these experiments, no mass balance was possible on these fractionation experiments and these data were omitted from the yield calculations reported in Table 3. The F5 materials from three of the runs were combined and refractionated separately from the F5 material from the fourth run. As such, the properties of the fractions from these second cuts vary quite drastically, as is evident from the properties tabulated in Table 4. The

final asphalt fractionated was SHRP AAF-1. By the time this asphalt was fractionated, most of the operating problems had been worked out, so that only two runs were needed to produce a sufficient quantity of F5 for refractionation. The F5 materials from these two runs were combined and refractionated in a single run. Aged Asphalt Production. To investigate the feasibility of using supercritical fractions as recycling agents, it was necessary to obtain an aged asphalt. SHRP AAF-1 was aged in an air bubbling apparatus as described by Chaffin et al. (1995). Asphalt is charged to a 1 gal paint can, air is introduced through a sparging ring, and the mixture is stirred continuously while being maintained at a temperature above 93.3 °C but below 113 °C. The aged asphalt used in this study had a final viscosity of 55 000 dPa‚s (P). This sample is referred to as AAF-AB1. Recycled Blend Composition. Nine recycled blends were produced using AAF-AB1 as the aged asphalt. Three different CRAs, three ISCFs, and three TAMU SCFs were utilized as the rejuvenating agents. The target viscosity for the recycled blends was chosen as the viscosity of unaged SHRP AAF-1 (20% so that the recycled blends could be compared not only against each other but also against the original parent asphalt. The initial amount of softening agent required for blending to achieve the target viscosity was determined using the dimensionless log viscosity (DLV) mixing rule developed by Chaffin et al. (1995). The DLV is given by eq 2. In

DLV )

ln(η* 0,m/η* 0,ra) ln(η* 0,as/η* 0,ra)

(2)

eq 2, η*0,m is the blend viscosity, η* 0,ra is the viscosity of the recycling agent, and η* 0,as is the viscosity of the aged asphalt. Blending was performed in a manner similar to that suggested in specification ASTM D4887. Accelerated Aging of the Recycled Blends. The recycled blends and the unaged SHRP AAF-1 were subjected to a slightly modified pressure aging vessel (PAV) accelerated aging test. In the standard PAV test the asphalt is pretreated according to the thin film oven test, or TFOT (ASTM D1754), and the pretreated asphalt is placed directly in the PAV vessel. The test was modified for this study in that the TFOT pretreated material was stirred and a sample was taken to measure the viscosity after the TFOT pretreatment. This may significantly alter the absolute value of the data obtained after the complete PAV test but should not inhibit comparison of the samples investigated in this study. The operating conditions for the standard PAV test are 100 °C and 300 psia air for a set time period of 20 h. Many authors have noted that asphalt aging profiles are hyperbolic in nature in that the properties change rapidly with time over a short period of time and then change in a constant manner nearly “indefinitely” (Petersen et al., 1993; Liu et al., 1996). Recent data indicate that the degree of the initial, rapid change in the carbonyl content, or “initial jump”, is a strong function of oxidation pressure (Liu et al., 1996) and composition (Jemison et al., 1995; Liu et al., 1996). Recent data also have shown that the aging kinetics in the constant rate region are dependent on both temperature (Lau et al., 1992; Liu et al., 1996; Huh and Robertson, 1996) and pressure (Liu et al., 1996). Liu et al. (1996) suggested that the oxidation kinetics, as measured by an increase in the carbonyl content, could

660 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 Table 3. Supercritical Operating Conditions and Fraction Yields fraction percentage fraction

temp (°C)

pressure (bar)

YBF

ABM-1

AAF-1

AAA-1

F8 F7 F6 F5 F4 F3 F2 F1

191 204 213 149 213 221 230 149

49.3 49.3 49.3 3.3 49.3 49.3 49.3 3.3

23.8 ( 3.9 9.6 ( 2.3 13.6 ( 6.3 53.6 ( 4.0 10.7 18.8 18.2 6.4

16.4 ( 1.5 8.2 ( 1.2 8.6 ( 2.2 66.9 (3.2 2.7 ( 1.8 25.2 ( 2.6 22.8 ( 0.5 16.3 ( 1.2

28.1 ( 6.9 6.4 ( 0.5 9.0 ( 0.9 56.6 ( 5.4 2.0 30.3 16.5 7.8

35.1 ( 4.2 4.0 ( 0.2 6.8 ( 1.5 54.1 ( 2.5 0.9 ( 0.0 17.0 ( 3.7 19.0 ( 9.2 17.0 ( 12.5

Table 4. Representative Supercritical Fraction Properties fraction YBF F1 YBF F2 YBF F3 YBF F4 YBF F5 YBF F6 YBF F7 YBF F8 ABM F1 ABM F2 ABM F3 ABM F4 ABM F5 ABM F6 ABM F7 ABM F8 AAA F1 AAA F2 AAA F3 AAA F4 AAA F5 AAA F6 AAA F7 AAA F8 AAF F1 AAF F2 AAF F3 AAF F4 AAF F5 AAF F6 AAF F7 AAF F8

viscositya (dPa‚×d7s)b 6 29.2 ( 7.3 137.4 ( 9.0 1040 43.2 ( 5.2 1813 ( 835 (5.95 ( 3.91) × 105 11.1 ( 1.5 91.1 ( 8.0 627.3 ( 18.7 1610 115.8 ( 17.1 (8.21 ( 1.32) × 104 (1.44 ( 1.10) × 107 2.0 ( 0.1 14.5 ( 3.2 99.2 ( 28.5 76.1 ( 81.8 10.7 ( 0.5 250 30 000 2.1 11.7 ( 1.6 67.1 ( 8.0 315.6 15.9 ( 0.2 3000 50 000

Mw

saturate (wt %)c

asphaltene (wt %)d

aromatic (wt %)e

649 895 1152 1388 986 (19 1460 ( 51 1912 ( 182 2668 ( 96 495 ( 9 748 ( 28 918 1085 758 ( 4 1266 ( 36 1531 ( 44 2232 ( 63 572 ( 27 875 ( 6 1237 ( 5 1178 ( 285 902 ( 5 1701 ( 304 2257 ( 233 4867 ( 83 661 928 1111 1334 966 ( 15 1417 ( 27 1696 ( 17 2495 ( 1

26.9 18.2 11.0 6.6 15.4 ( 0.3 6.0 ( 1.3 1.7 ( 0.7 0.8 ( 0.2 17.5 ( 0.6 12.9 ( 0.6 7.7 ( 0.7 9.5 ( 2.2 11.5 ( 0.5 3.4 ( 0.2 1.7 ( 0.1 1.0 ( 0.4 31.3 ( 3.4 20.0 ( 1.5 11.0 ( 1.3 16.2 ( 6.5 22.7 ( 0.3 7.6 ( 4.6 6.2 ( 0.6 1.8 ( 0.8 32.6 23.5 ( 0.4 14.3 ( 0.1 10.0 24.6 ( 0.4 8.1 ( 0.1 6.2 ( 1.7 1.2 ( 0.6

0.2 0.2 0.1 1.2 0.9 ( 0.3 1.4 ( 0.4 7.4 ( 4.5 59.5 ( 8.3 0.0 ( 0.0 0.0 ( 0.0 0.0 ( 0.1 0.0 ( 0.0 0.0 ( 0.0 0.3 ( 0.5 6.7 ( 2.0 47.7 ( 1.7 0.5 ( 0.6 0.2 ( 0.3 0.4 ( 0.5 0.6 ( 0.6 0.2 ( 0.0 0.6 ( 0.2 6.5 ( 1.6 62.0 0.8 0.4 ( 0.1 0.4 ( 0.3 1.0 0.1 ( 0.1 1.1 ( 0.1 7.4 ( 0.4 51.5 ( 2.1

73.0 81.7 88.9 92.2 83.7 ( 0.5 92.9 ( 1.4 90.7 ( 4.0 39.7 ( 8.5 82.6 ( 0.6 87.1 ( 0.6 92.3 ( 0.7 90.6 ( 2.2 88.5 ( 0.7 96.4 ( 0.5 91.6 ( 2.0 51.4 ( 1.9 68.3 ( 2.8 79.8 ( 1.4 88.6 ( 0.8 83.2 ( 5.9 77.2 ( 0.2 92.0 ( 4.3 87.4 ( 1.0 36.8 66.6 76.2 ( 0.4 85.4 ( 0.4 89.0 75.4 ( 0.4 90.9 ( 0.1 86.5 ( 1.3 47.3 ( 2.7

a 60 °C low-frequency limiting dynamic viscosity η* b c 0,60 °C. 1 dPa·s ) 1 P. From HPLC calibration. by difference.

be represented by an Arrhenius relation with respect to oxidation temperature and a power relation with respect to oxidation pressure as shown in eq 3. Unfor-

dCA ) RCA ) APRe-EA/RT dt

(3)

tunately, the single data point obtained from the PAV test is likely obtained within the initial jump region, is conducted only at a single high temperature, and is subject to the complications associated with variable pressure dependence. Thus, the relative rankings of various asphalts may be much different at road conditions than the rankings at the PAV conditions. To investigate the effects that aging temperature may have on the ranking of the asphalts, the residues from the PAV test were then further aged in the pressure oxygen vessel, POV (Lau et al., 1992; Liu et al., 1996), using atmospheric air at temperatures of 80, 90, and 100 °C. The use of atmospheric air greatly increases the duration of aging but eliminates the complications of pressure dependence and thus makes extrapolation to road conditions somewhat more reliable.

d

n-Hexane asphaltenes. e Determined

Properties Testing. The extent of oxidation was measured using Fourier transform infrared spectroscopy (FTIR). A Mattson Galaxy Series 5000 spectrometer was used with the attenuated total reflectance (ATR) method described by Jemison et al. (1992). Specifically, the extent of oxidation was quantified in terms of the carbonyl area (CA), which is the integrated area under the FTIR absorbance curve between 1820 and 1650 cm-1. It has been shown that the carbonyl area is an excellent surrogate for the total oxidation occurring as an asphalt ages (Liu et al., 1996). Low-frequency limiting dynamic viscosities were measured using a Carri-Med CSL-500 controlled stress rheometer with a 2.5 cm composite parallel plate and a 500 µm gap at a measurement temperature of 60 °C. The low frequency limiting dynamic viscosity (η*0,60 °C), further referred to as “the viscosity”, is obtained when the viscosity does not change with oscillation frequency in controlled stress measurements. To obtain the viscosity for some materials, it was necessary to conduct rheological measurements at higher temperatures and use the time-temperature superposition principle to obtain η* 0,60 °C (Ferry, 1984).

Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 661

Molecular weights were determined by gel permeation chromatography. Helium-sparged HPLC-grade tetrahydrofuran (THF) was used as the mobile phase at a flow rate of 1 mL/min. Three columns of decreasing pore size of 1000, 500, and 50 Å were utilized in series to accomplish separation. Samples were prepared by dissolving 0.2 ( 0.01 g of material in 10 mL of THF. The samples were then filtered through 0.45 µm PTFE membrane syringe filters, and 100 µL aliquots were injected onto the columns. Molecular weight distributions were determined from calibration with polystyrene standards. The columns and detector were operated isothermally at 313.2 K. Compositional analyses were performed through a combination of gravimetric and high-performance liquid chromatographic (HPLC) methods. Helium-sparged HPLC-grade n-hexane was utilized as the mobile phase at a flow rate of 2 mL/min. A single column 7.8 mm i.d. × 300 mm long packed with 10 µm µ-Bondapak (aminopropyl)methylsilyl-bonded amorphous silica was used for the analyses. The column was backflushed 15 min after injection to speed the elution of the polar aromatics and shorten the analysis time. Samples were prepared by dissolving 0.20 ( 0.01 g of material in 10 mL of n-hexane. The samples were then filtered through preweighed 0.45 µm PTFE membrane syringe filters, and the n-hexane asphaltene contents were determined gravimetrically, as described by Pearson et al. (1986). Twenty microliter aliquots of the maltene fraction were injected onto the HPLC column. The column and the detector were operated isothermally at 308.2 K. The saturate content was determined from the refractive index response factor of petroleum jelly as described by Chaffin et al. (1996). The total aromatic content, the sum of the naphthene and polar aromatic contents (Corbett, 1969), was determined by difference.

Figure 4. Fraction viscosity versus fraction number.

Figure 5. Asphaltene content versus fraction number.

Results and Discussion Supercritical Fraction Properties. Table 4 shows the viscosities (η* 0,60 °C), molecular weights (Mw), and Corbett-type composition data for the supercritical fractions. The trends reported in the previous studies (Stegeman et al., 1992; Jemison et al., 1995) were observed once again. The viscosities, Mw values, and asphaltene contents all increase nearly monotonically for F1-F4 and for F5-F8, while the content of saturates decreases nearly monotonically for F1-F4 and for F5F8. The aromatic content increases for F1-F4 and for F5-F7. Deviations from a strictly monotonic trend are attributed to possible contamination in the recovery process and random errors in GPC or HPLC sample preparation. The discontinuity in properties at F5 is due to refractionation of this material. Figure 4 shows the trends in viscosity for all asphalts, Figures 5-7 show the trend in composition for all asphalts, and Figure 8 shows fraction molecular weights. It is clear from these figures that there are significant differences between asphalts. Of particular importance are the differences in viscosities of F3 and F2 between the asphalts. The relatively low asphaltene content of asphalt ABM-1 necessitates that the aromatics are of higher viscosity. This results in fractions F3 and F2 having viscosities almost an order of magnitude above those from the other asphalts. The data in Table 4 are reported in terms of the mean ( 1 standard deviation (1σ). Those data points with only a mean were measured once. The data points with standard deviation (1σ) values were measured multiple

Figure 6. Saturate content versus fraction number.

times. Those values that have large deviations typically fall into one of two categories. The first category includes fractions where the residual pentane is difficult to remove. This will have a large effect on the viscosity and compositional analysis. F7 and F8 are highly subject to this problem. The second category is fractions of the sample asphalt from multiple fractionation experiments, such as YBF F5-F8. The data for these fractions represent at least four data points from the four different fractionation runs. Errors of this second type result from minor differences in operating conditions between runs and from performance of the different refractionation runs with potentially widely different materials (AAA-1). Many of these problems can be

662 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 Table 5. Composition and Viscosity Data for Asphalts Subjected to PAV Test sample SHRP AAF-1 AAF-AB1/ISCF A AAF-AB1/ISCF B AAF-AB1/ISCF C AAF-AB1/CRA A AAF-AB1/CRA B AAF-AB1/CRA C AAF-AB1/YBF F3 AAF-AB1/AAF F3 AAF-AB1/ABM F3

Figure 7. Aromatic content versus fraction number.

Figure 8. Molecular weight versus fraction number.

Figure 9. DLV data for recycled blends.

eliminated by operating a supercritical fractionation unit continuously, as occurs in industry. Recycled Blend Properties. Figure 9 shows the location of blend DLV data points in relation to the data points collected in the previous viscosity mixing rules study conducted by Chaffin et al. (1995). The diagonal line represents the mixing rule suggested in ASTM D4887. Chaffin et al. (1995) showed that the ASTM mixing rule is adequate for blends using low-viscosity asphalts as the softening agent but is completely unsuitable for blends using commercial recycling agents or supercritical fractions produced in the laboratory. The data presented in Figure 9 clearly show that the industrial supercritical fractions (ISCFs) behave like commercial recycling agents (CRAs) and supercritical

viscosityb composition (asphalt/agent)a (dPa‚s)c TFOT AId PAV AId N/A 72/28e 61/39 43/57 81/19 83/17 83/17 61/39e 67/33e 44/56e

1890 1900 2140 2080 1840 1850 1900 2000 2090 1670

2.80 1.68 N/A 1.67 1.85 1.70 1.96 1.50 1.67 1.59

12.42 4.21 N/A 3.89 4.30 4.46 5.53 3.00 3.85 2.93

a Approximate composition (unless denoted by footnote e). b 60 c °C low-frequency limiting dynamic viscosity η* 0,60 °C. 1 dPa·s ) 1 P. d AI ) viscosity after aging/unaged viscosity. e Target viscosity achieved after first attempt.

fractions produced at TAMU in terms of viscosity reduction. The target viscosity was obtained for four of the nine recycled blends on the first blending attempt as is indicated by footnote e in Table 5. The target viscosity was obtained after the second blending attempt for four of the remaining five recycled blends. This is a marked improvement over the results that would have been obtained using the mixing rule suggested in ASTM D4887. TFOT and PAV Test Comparison. The aging indexes (AIs), which are calculated as the viscosity after aging divided by the viscosity before aging, for all materials after TFOT pretreatment and after subsequent PAV aging are shown in Table 5. The recycled blend using ISCF B was very problematic from the initial blending through the end of the PAV test. In fact, blending with this agent required multiple tries to obtain the correct initial viscosity, and a lowfrequency limiting dynamic viscosity (η*0,60 °C) could not be obtained for the TFOT treated and PAV aged materials. Thus, the aging indexes could not be determined for this recycled blend. This is likely the result of phase separation related to the extremely high saturate content of ISCF B. The data in Table 5 also show that all of the recycled blends have lower aging indexes than the parent asphalt SHRP AAF-1. This indicates that an asphalt with superior hardening characteristics can be produced through recycling. Furthermore, the data show that the aging indexes of the recycled blends using CRAs are higher than the aging indexes of the blends which used supercritical fraction rejuvenating agents, either those obtained from industrial sources or those produced in the laboratory at Texas A&M University. In addition, the aging indexes for the blends using the laboratory SCFs are the lowest of all the aging indexes. The lower aging indexes for the recycled blends using supercritical fraction rejuvenating agents likely result, at least in part, from the relatively larger proportion of supercritical fractions (more than 50% SCF in some blends examined in this study) in the recycled blends compared to the proportion of commercial recycling agents (approximately 18% CRA in all blends examined in this study) in the recycled blends. In other words, dilution of the aged asphalt seems to play an important role in improving the behavior of recycled asphalts. This has also been proposed by Lin et al. (1995b). The lower aging indexes for the recycled blends using TAMU SCFs undoubtedly result from the narrower cuts that are obtained by fractionating into seven distinct fractions (eight total) rather than the two or three cuts that are

Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 663

Figure 10. Carbonyl formation for AAF-AB1/ISCF A.

Figure 11. Arrhenius plot for AAF-AB1/ISCF A.

produced industrially. The laboratory F3 cuts are narrower in that they are more uniform than the ISCFs in terms of composition and they also have fewer heavier aromatic materials than the ISCFs. These heavier aromatic materials present in the ISCFs oxidize, form asphaltenes, and harden more rapidly than the lighter aromatics prevalent in the TAMU F3 fractions. POV Test Comparison. The residues from the PAV aging test were then distributed into POV trays, aged at 80, 90, and 100 °C using atmospheric air, and analyzed in a manner similar to that described by Liu et al. (1996). After aging, carbonyl area measurements were performed on all of the samples and viscosity measurements were performed on enough samples (at least half of the samples) to determine HS values for each mixture. The one exception to this was the rejuvenated blend using ISCF B as the softening agent. Because low-frequency limiting viscosities could not be obtained after TFOT and PAV aging, viscosity measurements were not performed on the POV-aged samples of this blend. Figure 10 shows a plot of CA versus time (time ) 0 arbitrarily set to be when the POV aging started) for the AAF-AB1/ISCF A blend. The slopes of each of the three lines are equal to the oxidation rate as measured by CA (RCA) at the various temperatures. The use of only atmospheric pressure simplifies the kinetic expression in eq 3. This simplified expression, rewritten in linearized form, is given in eq 4 where A′ ) A × PR.

ln

EA 1 dCA ) ln RCA ) ln A′ dt R T

(4)

Figure 11 shows a semilog plot of RCA versus the reciprocal of the absolute temperature (1/T). This plot should be linear, as suggested by eq 4, with a slope proportional to the activation energy, EA. The limited data presented in Figure 11 are well described by eq 4, within experimental scatter. Figure 12 is a semilog plot of viscosity versus carbonyl area for the PAV preaged AAF-AB1/ISCF A blend. The slope of the line produced from the data in Figure 12 is the HS. Plots similar to those shown in Figures 10 and 11 were produced for all 10 AAF-1 materials examined in this experiment. Plots similar to Figure 12 were produced for all of the materials except the blend using ISCF B as the softening agent. The activation energies and HS values for these samples are reported in Table 6, as are extrapolated values of RCA and the rate of increase in ln η*0 at 50 °C (a reasonable value for road temperature).

Figure 12. Hardening susceptibility for AAF-AB1/ISCF A. Table 6. POV Aging Results material

E (kJ/mol)

HS

SHRP AAF-1 AAF-AB1/ISCF A AAF-AB1/ISCF B AAF-AB1/ISCF C AAF-AB1/CRA A AAF-AB1/CRA B AAF-AB1/CRA C AAF-AB1/YBF F3 AAF-AB1/AAF F3 AAF-AB1/ABM F3

72.1 85.0 72.5 80.5 88.9 93.4 83.3 98.3 74.4 83.1

4.48 4.91 N/A 3.41 3.49 4.70 5.11 3.20 3.53 2.22

a

50 °C CA rate 50 °C ln η*0 (10-3 CA/day)a rate (10-3)a 1.91 1.03 0.85 1.23 0.89 0.82 1.03 0.54 1.55 1.11

8.54 5.07 N/A 4.19 3.11 3.86 5.25 1.73 5.46 2.45

Subject to errors associated with extrapolation.

The data in Table 6 show that the activation energies vary widely for the AAF-1 materials investigated in this study. Of particular importance is the fact that the activation energy of the original asphalt is lower than the activation energies of all of the recycled blends. This results in carbonyl formation rates (extrapolated) at 50 °C higher than the recycled blends as shown in Table 6 and in Figure 13. The HS values are more variable, with some recycled materials having higher (worse) HS values and some having lower (better) HS values than the original asphalt. Using the data in Tables 1, 3, and 4, the approximate saturate contents in the materials were calculated. The variability in HS is roughly correlated with the saturate content in the material as shown in Figure 14. The correlation seen in Figure 14 agrees with the results reported by Peterson et al. (1994) and statements by Epps et al. (1980) that the saturate

664 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997

Figure 13. Arrhenius plot and extrapolated rates for all recycled blends.

Figure 15. Arrhenius plot for SHRP AAF-1 atmospheric data. Table 7. Comparison of Relative Rankings of Recycled Asphalts

material SHRP AAF-1 AAF-AB1/ISCF A AAF-AB1/ISCF B AAF-AB1/ISCF C AAF-AB1/CRA A AAF-AB1/CRA B AAF-AB1/CRA C AAF-AB1/YBF F3 AAF-AB1/AAF F3 AAF-AB1/ABM F3

Figure 14. HS versus recycled blend saturate content.

content should be minimized to produce good quality recycled blends. The combination of HS and RCA (extrapolated) at 50 °C yields the predicted rate of increase in ln η*0 at 50 °C. As was the case with the aging indexes determined from the TFOT and PAV tests, the data in Table 6 indicate that all of the recycled materials will harden more slowly than the original asphalt. This results mostly from the improvements in activation energy. The data in Table 6 also indicate that the predicted hardening rates of the recycled asphalts which were softened with supercritical fractions are similar to the hardening rates of the recycled asphalts softened with commercial recycling agents. In fact, the two best hardening rates were obtained for recycled asphalts using TAMU SCFs. Once again, this can likely be attributed to the narrow cuts that are obtained by fractionation into seven fractions as opposed to two or three. These results are presented with some caution due to the extreme variability that may result both from estimation of activation energy from three temperatures and from significant extrapolation. The relative rankings based on the TFOT and PAV aging indexes and the road condition hardening rates are given in Table 7. Although there are some differences between the relative rankings from the various hardening indicators, the agreement between the tests is rather good. Most importantly, however, is the fact that a highway department choosing a rejuvenating agent for this asphalt based on the relative rankings given in Table 7 would choose either the YBF F3 or the

TFOT aging index (1 ) lowest)

PAV aging index (1 ) lowest)

50 °C ln η*0 rate (1 ) lowest)

9 5

9 5

9 6

4 7 6 8 1 3 2

4 6 7 8 2 3 1

5 3 4 7 1 8 2

ABM F3, the best two agents by all three measures, if they were available. Importance of Neat AAF-1 Data. The data collected on the original asphalt AAF-1 were compared to the kinetic data reported for rolling thin film oven test (ASTM D2872) pretreated and neat AAF-1 by Liu et al. (1996). Notably, the activation energy determined from the PAV preaged AAF-1 studied here is nearly the same as the reported activation energy for AAF-1. The AAF-1 carbonyl rates determined in this study and the rates obtained from atmospheric aging conditions reported by Liu et al. are plotted together in Figure 15. Figure 15 clearly shows that the rates obtained using PAV preaged material, RFTOT pretreated material, and neat asphalt are highly consistent. In addition, the rate determined in this study at 80 °C effectively extends the atmospheric data 8 °C beyond the data reported in the literature. The activation energy determined using the data in Figure 15 is 69.4 kJ/mol, which is highly consistent with the activation energy reported in Table 6 which was estimated from the three oxidation rates measured in this study. Because Liu et al. (1996) described only the oxidation kinetics of asphalts, no HS values were reported. However, subsequent viscosity analyses performed on the samples they aged at atmospheric pressure yielded an HS value for the AAF-1 nearly identical to the HS value for the PAV preaged AAF-1 investigated in this study. The consistency of the activation energy and the HS prove conclusively for the first time that the aging history of the material has little, if any, influence on the kinetic parameters. (Obviously, if the samples have ever been subjected to extreme oxidation conditions such as 500 °F, this may alter this conclusion.) This means that it may be possible to extract an asphalt from a pavement sample and subject it to the POV test to

Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 665

determine kinetic parameters. From these kinetic parameters and the HS, it may be possible to accurately predict the future condition of the pavement. This is a pleasing, but unintended, result of this experiment. Conclusions It has been shown that the properties of supercritical fractions produced from fractionation of asphalt vary not only among the fractions of a given asphalt but also for the same fraction produced from different asphalts. Several of the supercritical fractions produced in this study have low enough saturate and asphaltene contents to restore an aged asphalt to a like-new compositional state. Furthermore, several of the fractions produced in this study also have viscosities suitable for blending with an aged asphalt to restore it to a likenew consistency state. One aged asphalt was blended with three of the laboratory supercritical fractions, three industrial supercritical fractions, and three commercial recycling agents. The aging indexes after thin film oven test (TFOT) treatment, the aging indexes after pressure aging vessel (PAV) aging, and the extrapolated road conditions hardening rates all indicate that supercritical fractions can be used to produce recycled blends with properties superior to those of the original asphalt. Furthermore, the TFOT and PAV aging indexes from the blends using supercritical fractions are superior to those of the blends produced by using commercial recycling agents. The hardening susceptibilities of the supercritical fraction blends are also generally superior to those of the commercial recycling agent blends. The lower aging indexes for the recycled blends using supercritical fraction rejuvenating agents indicate that using a higher viscosity rejuvenating agent is desirable and that asphaltene dilution is a key factor in asphalt recycling. The hardening susceptibilities for the materials examined in this study are correlated strongly with the saturate content, indicating that minimized saturate content is also important. The generally superior performance of the laboratory supercritical fraction recycled blends can thus be attributed to the narrower cuts produced in the laboratory compared to industrial operation. This limited study indicates that a supercritical fraction constituting as much as 30% of the asphalt from which it was produced may be useful as a recycling agent. Through proper selection of operating conditions and feedstocks, this percentage could be dramatically increased. Obviously, this could have tremendous impact on asphalt processing and economics. Acknowledgment Support for this work by the Texas Department of Transportation (TxDOT), in cooperation with the U.S. Department of Transportation, and the Federal Highway Administration (FHWA). This work was also supported by the U.S. Department of Energy DOE, Assistant Secretary for Energy Efficiency and Renewable Energy under DOE Albuquerque Operations Office Cooperative Agreement DE-FC04-93AL94460. Fellowship support provided by the Dwight David Eisenhower Transportation Fellowship Program is gratefully acknowledged. This report was prepared as an account of work sponsored by an agency of the U.S. Government. Neither the U.S. Government nor any agency thereof, nor any of their employees, makes any warranty,

expressed or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the U.S. Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the U.S. Government or any agency thereof. List of Abbreviations and Nomenclature AI ) aging index (η* 0,aged/η* 0,unaged), dimensionless CA ) carbon area, dimensionless CRA ) commercial recycling agent DAO ) deasphalted oil DLV ) dimensionless log viscosity ({ln(η* 0,m/η* 0,ra)}/ {ln(η* 0,as/η* 0,ra)}, dimensionless GPC ) gel permeation chromatography HPLC ) high-performance liquid chromatography HS ) hardening susceptibility, dimensionless ISCF ) industrial supercritical fraction PAV ) pressure aging vessel POV ) pressure oxygen vessel ROSE ) residuum oil supercritical extraction SCF ) supercritical fraction SHRP ) strategic highway research program TAMU ) Texas A&M University TFOT ) thin film oven test (ASTM D1754) A ) preexponential factor, CA/day‚atmR A′ ) preexponential factor for the reaction in air, CA/day EA ) reaction activation energy, kJ/mol R ) universal gas constant, kJ/mol‚K T ) absolute temperature, K Greek Letters R ) order of reaction relative to oxygen, dimensionless η* 0 ) low-frequency limiting viscosity, dPa‚s ) P

Literature Cited Chaffin, J. M.; Davison, R. R.; Glover, C. J.; Bullin, J. A. Viscosity Mixing Rules for Asphalt Recycling. Transp. Res. Rec. 1995, 1507, 78. Chaffin, J. M.; Lin, M. S.; Liu, M.; Davison, R. R.; Glover, C. J.; Bullin, J. A. The Use of HPLC to Determine the Saturate Content of Heavy Petroleum Products. J. Liq. Chromatogr. 1996, 19, 1669. Corbett, L. W. Composition of Asphalt Based on Generic Fractionation Using Solvent Deasphaltening, Elution-Adsorption Chromatography, and Densimetric Characterization. Anal. Chem. 1969, 41, 576. Epps, J. A.; Little, D. N.; Holmgreen, R. J.; Terrel, R. L. Guidelines for Recycling Pavement Materials. NCHRP Report No. 224; Transportation Research Board, National Research Council, Washington, DC, 1980. Ferry, J. Viscoelastic Properties of Polymers, 4th ed.; John Wiley and Sons: New York, 1985. Huh, J. D.; Robertson, R. E. Modeling of Oxidative Aging Behavior of Asphalt from Short Term, High Temperature Data as a Step Toward Prediction of Pavement Aging. 75th Annual Meeting of the Transportation Research Board, Washington, DC, Jan 1996; Preprint 961112. Jemison, H. B. Supercritical Refining of Asphalts. Ph.D. Dissertation, Texas A&M University, College Station, TX, 1992. Jemison, H. B.; Burr, B. L.; Davison, R. R.; Bullin, J. A.; Glover, C. J. Application an Use of the ATR, FT-IR Method to Asphalt Aging Studies. Fuel Sci. Technol. Int. 1992, 10, 795. Jemison, H. B.; Lunsford, K. M.; Davison, R. R.; Glover, C. J.; Bullin, J. A. Fractionation of Asphalt Materials by Using

666 Ind. Eng. Chem. Res., Vol. 36, No. 3, 1997 Supercritical Cyclohexane and Pentane. Fuel Sci. Technol. Int. 1995, 13, 605. Lau, C. K.; Lunsford, K. M.; Glover, C. J.; Davison, R. R.; Bullin, J. A. Reaction Rates and Hardening Susceptibilities as Determined from POV Aging of Asphalts. Transp. Res. Rec. 1992, 1342, 50. Lin, M. S.; Glover, C. J.; Davison, R. R.; Bullin, J. A. The Effects of Asphaltenes on Asphalt Recycling and Aging. Transp. Res. Rec. 1995a, 1507, 86. Lin, M. S.; Lunsford, K. M.; Glover, C. J.; Davison, R. R.; Bullin, J. A. The Effects of Asphaltenes on the Chemical and Physical Characteristics of Asphalts. In Asphaltenes: Fundamentals and Applications; Sheu, E. Y., Mullins, O. C., Eds.; Plenum Press: New York, 1995b. Lin, M. S.; Chaffin, J. M.; Liu, M.; Glover, C. J.; Davison, R. R.; Bullin, J. A. The Effect of Asphalt Composition on the Formation of Asphaltenes and Their Contribution to Asphalt Viscosity. Fuel Sci. Technol. Int. 1996, 14, 139. Liu, M.; Lunsford, K. M.; Glover, C. J.; Davison, R. R.; Bullin, J. A. The Kinetics of Carbonyl Formation in Asphalt. AIChE J. 1996, 42, 1069. Martin, K. L.; Davison, R. R.; Glover, C. J.; Bullin, J. A. Asphalt Aging in Texas Roads and Test Sections. Transp. Res. Rec. 1990, 1269, 9.

Pearson, C. D.; Huff, G. S.; Gharfeh, S. G. Technique for the Determination of Asphaltenes in Crude Oil Residues. Anal. Chem. 1986, 58, 3266. Petersen, J. C.; Branthaver, J. F.; Robertson, R. E.; Harnsberger, P. M.; Duvall, J. J.; Ensley, E. E. Effects of Physicochemical Factors on Asphalt Oxidation Kinetics. Transp. Res. Rec. 1993, 1391, 1. Peterson, G. D.; Davison, R. R.; Glover, C. J.; Bullin, J. A. The Effect of Composition on Asphalt Recycling Agent Performance. Transp. Res. Rec. 1994, 1436, 38. Rostler, F. S.; Sternberg, H. W. Compounding Rubber with Petroleum Products. Ind. Eng. Chem. 1949, 41, 598. Stegeman, J. R.; Kyle, A. L.; Burr, B. L.; Jemison, H. B.; Davison, R. R; Glover, C. J.; Bullin, J. A. Compositional and Physical Properties of Asphalt Fractions Obtained by Supercritical Extraction. Fuel Sci. Technol. Int. 1992, 10, 767.

Received for review July 25, 1996 Revised manuscript received November 13, 1996 Accepted November 18, 1996X IE9604435 X Abstract published in Advance ACS Abstracts, January 1, 1997.