Energy & Fuels 1991,5, 333-340
333
Fractionation and Characterization of Utah Tar Sand Bitumens: Influence of Chemical Composition on Bitumen Viscosity K. Bukka,*>tJ. D. Miller,+ and A. G. Obladt Department of Metallurgical Engineering and Department of Fuels Engineering, University of Utah, Salt Lake City, Utah 84112 Received August 13, 1990. Revised Manuscript Received December 1, 1990
The viscosity of tar sand bitumen, among other properties such as thermal conductivity and specific heat, represents an important property that controls various recovery processes. Tar sand bitumens, particularly from Utah,have been shown to vary significantly in their physical and chemical properties from one location to another. The highly viscous nature of the bitumen was understood, based on a model developed by Nellensteyn and further revised by Pfeiffer et al. and Marvillet, to arise from a dispersion of asphaltenes in viscous oils with constituent aromatics and resins acting as suspension stabilizers. Based on this colloidal nature of bitumens, the differences in viscosities were accounted for by different degrees of peptization of the asphaltenes. In a recent related study by Christensen et al., it was shown that for an Asphalt Ridge (Utah)bitumen the viscosity increased with an increase in asphaltenes content. Variation in the asphaltene content of the bitumen was accomplished by a prior fractional separation of the bitumen into maltenes and asphaltenes and mixing these fractions together in various proportions. From the colloidal suspension model for bitumen and the results of Christensen et al., it may appear that asphaltenes alone are responsible for the observed high viscosity of bitumen. However, a comparative analysis of Athabasca (Alberta, Canada) and Asphalt Ridge bitumens, particularly in terms of their viscosities and asphaltene contents, reveals otherwise. Asphalt Ridge bitumen, though containing only 7% asphaltenes, has a viscosity value of 80 P e s at 50 "C whereas Athabasca bitumen, which has 17% asphaltenes, has a viscosity of only 5 P a s at 50 "C. These results indicate that in addition to asphaltenes there are other compositional properties of the bitumen which influence its viscosity. In this paper a detailed analysis consisting of fractionation and characterization of Asphalt Ridge and Sunnyside bitumens was undertaken in order to determine what compositional properties other than asphaltene content may influence bitumen viscosity. The results suggest that a discrete group of compounds present in the polar fractions of the bitumen may have a significant influence on bitumen viscosity and may be of equal, if not greater, importance than the asphaltene content itself. The interpretation of the observed results not only lends support to the model of Nellensteyn, but also identifies another compositional factor which determines bitumen viscosity.
Introduction Asphaltenes are considered to enhance the viscous behavior of bitumens.14 This inference is based on an earlier studyS of bitumen derived from Asphalt Ridge tar sands of Utah. In this study the bitumen was fractionated into maltenes and asphaltenes and these two components were mixed in different proportions to evaluate the effect of asphaltenes on the viscosity of the synthetically prepared bitumen. The finding that asphaltenes increase the viscosity of such a bitumen is an important step in understanding bitumen properties. However, it appears that this result cannot be generalized and bitumen viscosities cannot be estimated from their asphaltenes contents alone. This is perhaps best illustrated by considering two samples of bitumen extracted from Athabasca and Asphalt Ridge tar sands. Asphalt Ridge bitumen, though containing only 7% asphaltenes, has a viscosity value of 80 Pas at 50 OC whereas Athabasca bitumen, which has 17% asphaltenes, has a viscosity value of only 5 Pas at the same temperature. This obviously indicates that in addition to asphaltenes content there may be other compositional
* To whom correspondence should be directed. 'Department of Metallurgical Engineering. t Department of Fuels Engineering.
properties which influence bitumen viscosity. Bitumen is a complex mixture consisting of compounds ranging from nonpolar aliphatic and naphthenic hydrocarbons to highly polar aromatic molecules containing heteroatoms such as oxygen, nitrogen, and sulfur. Bitumen also contains several transitional metal ions complexed with various polar molecules such as porphyrins. A number of attempts have been made to fractionate and characterize the bitumen into broadly defined compound The method which is commonly adopted for this separation is based on the polarity of the constituent compounds comprising the bitumen. The separated groups (1)Nellensteyn, F.J. Manufacture and constitution of asphaltic bitumen. Dissertation Technische Hoogeschool, Delft, Holland, 1923. (2)Pfeiffer, J. PH.; Van Doormad, P. M.J . Inst. Pet. Technol. 1936,
22,414. (3)Pfeiffer, J. PH.; Sad, R. N. J. J. Phys. Chem. 1940, 4, 139. (4)Marvillet, J. h o c . Assoc. Asphalt Technol. 1975,44, 416. (5) Christensen, R.J.;Lindberg, W. R.; Dorrence, S. M. Fuel 1984,63, 1312. (6)Jewell, D. M.; Weber, J. H.; Bunger, J. W.; Plancher, H.; Latham, D. R. Anal. Chem. 1972,44, 1391. (7) Syncrude Anulytical Methods for Oil Sand and Bitumen Processing; Bulmer, J., Starr., Eds.; Alberta Oil Sands Technology and Research Authority: Edmonton, Alberta, 1979. (8)Sadeghi, K.M.; Sadeghi, M. A.; Wu, W. H.; Yen, T. F. Fuel 1989, 68, 782.
0S87-0624/91/2505-0333$02.50/00 1991 American Chemical Society
Bukka et al.
334 Energy & Fuels, Vol. 5, No. 2, 1991 Bitumen
I
(Solubles)
1
Pentane (Insolubles)
1
Asphaltenes
Maltenes
1
Methanol
Pentane
Saturates 6 Aromatics-I 6 I1
Resins-I
1-
Res ins I I
Neutral Alumina
1
Pentane
Saturates
THF
1
Aromat ics-I
Methanol
1
Aromatics-I1
Figure 1. Bitumen fractionation scheme.
of compounds which progressively increase in their polarities are saturates, aromatics, resins, and asphaltenes. As an elegant separation into well-defined groups of compounds from the complex bitumen is nearly impossible, these four groups are considered at best to be broad and are found to be useful for comparison purposes. In order to account for some of the physical properties, particularly viscosity, a conceptual model for bitumen was developed by Nellensteyn.’ It assumes bitumen to be a colloidal system in which the solid asphaltenes are suspended in viscous oils and the suspension is considered to be stabilized by polar molecules belonging to the aromatics and resins group. Though this model is seemingly appropriate, it is difficult to provide experimental evidence to substantiate its features. In this paper, a study involving two different Utah tar sand bitumens, viz., Asphalt Ridge (AR) and Sunnyside (SS),which differ extensively in their viscosities (at 50 “C the values of AR and SS bitumens are 80 and 1500 Pa.s, respectively) is presented. These bitumens were fractionated by a slightly modified method reported elsewhere.’ Characterization studies were extended to the various separated fractions. From these analyses an attempt was made to identify the compositional features which influence bitumen viscosity.
Experimental Section Materials. The two bitumens used in this study were obtained by conventional toluene Dean-Stark extraction of the tar sand samples of AR and SS deposits (Utah). The solvents such as pentane, toluene, tetrahdyrofuran, and methanol were of spectroscopic grade and were obtained from EM Science Inc. Fuller’s earth was of 3W30 mesh and obtained from Sigma Chemical Co. It was washed with pentane prior to its use in the fractionation step. Neutral alumina (type 507C) was a chromatographicgrade material and obtained from Fluke Chemicals. Before use,alumina was dried in the oven at 110 OC for 24 h. Potassium bromide was spectroscopic grade, which was obtained from EM Science Inc. Fractionation. The fractionation scheme followed in this study is illustrated in Figure 1. For the initial separation of bitumen into maltenes and asphaltenes, the commonly used solvents are C5-C7 normal alkanes. The solvating ability of the alkane with respect to the bitumen increases gradually with the carbon chain length.*” This would mean that pentane yields
quantitatively more aspMtenes than heptane for a given bitumen. Since pentane is relatively easy to remove, it has gained acceptance and has become a standard solvent for maltene and asphaltene separation from bitumen. The procedure for fractionation consisted of the followingsteps. Bitumen was weighed in a tared Erlenmeyer flask (250 mL) to which toluene was added (1mL of toluene per gram of bitwen) and heated on a water bath to obtan a homogeneous suspension. The suspension was cooled, and followed by gradual addition of pentane (40 times the volume of toluene) with constant mixing. The resulting mixture was thoroughly shaken and stored in the dark for 16-20 h. The insoluble solids (asphaltenes)from soluble maltenes were separated by filtration using a fine sintered glass funnel. The solids were washed thoroughly with pentane till the washings were colorless, dried in the oven at 110 OC for 2 h, and weighed. Since the asphaltenes fraction was separated by virtue of ita insolubility in pentane, and considering the possibility of fine minerals association with the bitumen, a separation between the two was effected by dissolution in dichloromethane and filtration. The asphaltene fraction was finally isolated from its solution in dichloromethaneby distillation on a Buchi btavapour. In a similar manner, the maltene fraction was also obtained by the removal of pentane from its solution. The maltene fraction was further divided into two components, one containing a mixture of saturates and aromatics and the other resins. This separation was carried out in a Soxhlet extraction unit. The Fuller’s earth was taken in a fritted glass thimble (the ratio of maltenes to Fuller’s earth was maintained at 1:lO) and washed with pentane prior to use. The maltenes were then loaded onto Fuller’s earth and capped with some glass wool, and the thimble was transferred to the Soxhlet unit. The extraction of the maltene fraction was carried out successively with solvents of increasing polarity. The solvents used were in the order pentane, tetrahydrofuran, and methanol. Extraction with each solvent was carried out for a minimum of 24 h. The extracts thus collected were labeled as a mixture of saturates and aromatics, resin-I, and resin-11, respectively. In each case, the solvent was removed as mentioned in the previous paragraph. In the finalstep, the mixture containing saturates and aromatics was further separated into ita components by column chromatography. Neutral alumina, dried at 110 OC for 24 h, was used as support material. A ratio of 151for the support material to the mixture was maintained. The saturates and aromatics mixture was dissolved in a small volume of pentane (- 20 mL) the solution was loaded onto the column. The column was eluted with the same solvent sequence as described earlier. After the removal of the solvent from the extracted solutions, they were labeled as saturates, aromatics-I,and aromatics-11. All fractions were, after removal of the solvent on the Buchi Rotavapour, placed in a vacuum oven at 60 “C (10 mmHg) for 16 h to remove the residual traces of solvent. The mass balances obtained in the fractionation were 99.7% and 97.8% for AR and SS bitumens, respectively. The greater loss observed in the case of SS bitumen was mainly confined to the step involving the separation of saturates and aromatics. Characterization studies were carried out on major fractions of the bitumens viz., saturates, aromatics-I, resins-I, and asphaltenes. The remaining fractions such as aromatics-I1 and resins-I1 were too small and together amounted to 1%. The characterization of the bitumen fractions consisted of elemental analyses, molecular weight determination by vapor-phase osmometry, and FTIR and 13C NMR spectroscopic examination. Elemental Analysis and Molecular Weight Determinations. Elemental analysis of bitumens and their major fractions consisted of carbon, hydrogen, nitrogen, and sulfur. These were performed on a Leco-600 and Leco-SCl32analyzers. Molecular weights were determined at a commercial laboratory (Alltech Associates Inc., San Jose, CA). These measurements were carried out using Wescan Model 233 vapor-phase osmometer with toluene as the solvent and at an operating temperature of 47 OC. A
-
(9)Mitchell, D.L.; Speight, J. G. Fuel 1973,52, 149. (10) Long, R. B. Chemistry of Asphaltenes, Advances in Chemistry Series; Bunger J. W., Li, N. C., Eds.; American Chemical Society: Washington, DC, 1981;Vol. 195, p 17. (11) Andersen, S. I.; Birdi, K. S. Fuel Sci Technol. 1990,8, 593.
Energy & Fuels, Vol. 5, NO.2, 1991 335
Utah Tar Sand Bitumens
Table I. Fractional Composition of Asphalt Ridge and Sunnyeide Bitumens bitumen fraction saturates aromatics-I aromatics-I1 resins-I resins-I1 asphaltenes
weight percent Asphalt Ridge Sunnyside 32.4 24.9 22.4 18.1 n.d. 0.5 37.6 30.0
n.d.a 7.3
0.6 23.7
an.d. = not determined.
standard solution of sucrose octaacetate (MW = 678.6) was used to obtain the instrument calibration factor. The concentrations of bitumen fractions in toluene were varied over an order of magnitude (0.5-5.0 g/L). The results have shown that the calculated molecular weights were independent of solute concentration. 1wNMR Spectroscopy. 13C NMR spectra of bitumen fractions were obtained on a Bruker AC-F 200 NMR spectrometer. Spectra of saturate fractions were obtained in the liquid state. In all other fractions deuterated chloroform was used as a solvent. The operating conditions for spectral analysis were as follows: quadrature detection and phasing, spin temperature alternation, 5 90"pulse width, 2K data points, 15-kHzspectral and 19-kHz filter width, 0.024 s acquisition time, 3-kHz magic angle spinning in 9-mm ceramic rotors, 12-bit digitizer resolution, and 50-H~line broadening at an ambient temperature of 20 "C. All chemical shifts were referred to TMS via adamantane as a secondary reference. The magic angle was set by using the 79Brresonance in KBr. FTIR Spectroscopy. FTIR spectra of the bitumen fractions were obtained on a Digilab FTS-40 spectrometer equipped with a Model 3240-SPC data-processing system and liquid nitrogen cooled MCT detector. The software capability of the instrument included base-line correction, deconvolution, and subtraction of the primary spectra. Transmission spectra of asphaltenes and resin-I samples were obtained in a KBr matrix. The sample was mixed with KBr (-3%) in a Wig-L-Bug for 5 min, and a portion of the mixture (-150 mg) was pressed into a disk (13 mm) under 20000 psi. Transmission spectra of the saturates and aromatics-I fractions were obtained by directly smearing the samples on KBr windows. Although the thickness of sample film on the windows was not determined, the amount taken was adjusted till major peaks in the spectrum were sharp. Primary spectra of the bitumen fractions were obtained with a nominal resolution of 4 cm-' in the range 4000-700 cm-'.
Results Fractionation. Bitumen components separated by the fractionation scheme, shown in Figure 1, are quantitatively described in Table I. Although the distinctions were made within the aromatic and resin fractions of SS bitumen as aromatics-I and -11and resins-I and -11,the subfractions designated as aromatics-I1 and resins-I1 were too small to afford a detailed analysis. Consequently, the terms aromatics and resins will be used in the remainder of the text to identify aromatics-I and resins-I, respectively. As seen from Table I, differences in the composition of the two bitumens are quite significant in every major fraction. AR bitumen contains a relatively high proportion of saturates, aromatics, and resins whereas SS bitumen contains a very large proportion of asphaltenes. Although the asphaltenes content of the AR bitumen agrees well with the earlier reported value of 6.3% by Bunger et al.,12the SS asphaltenes value is significantly different from 17.8%, as reported by Kotlyar et al.13 (12) Bunger, J. W.; Thomas,K.P.; Dorrence, S. M. Fuel 1979,58,183. (13) Kotlyar, L.S.;Ripmeester, J. A.; Sparks, B. D.; Woods,J. A. Fuel 1988,67, 15%.
140 120
100
80
60
40
20
0
PP M
Figure 2. 13C NMR spectra of (a) SS and (b) AR saturates fractions.
The saturate fractions of both AR and SS bitumens were clear, viscous liquids. The former had a faintly yellow color and the latter was colorless. Other than possible viscosity differences, both aromatics and resin fractions looked alike. When the resins were freshly separated, they were black crystalline solids but, on long standing, became viscous semisolids,possibly due to their low softening points. Both asphaltene fractions were glossy black crystalline solids but lost their luster when they were finely ground. Elemental Analysis. Elemental analysis of AR and SS bitumens and their major fractions with their molecular weights and representative molecular formulae are presented in Table 11. The results indicate that, for a given fraction, the difference between the two bitumens is rather small. Considering the probable uncertainties in the determination of some of the elements such as nitrogen and sulfur, particularly when their content is very small, the saturate fractions were observed to be practically identical. However, some differences in the contents of the heteroatoms among other fractions were noticeable. Nitrogen is consistently higher in the fractions of AR bitumen while differences in the sulfur contents are noticeable only in the asphaltene fractions, with a higher value observed in the case of SS bitumen. Since oxygen contents were determined by difference, interpretation of its trends in all fractions of AR and SS bitumens warrants some caution. The results presented in Table I1 also lead to some general conclusions for both bitumen samples. For a given bitumen fraction, both carbon and hydrogen contents slowly decrease from the relatively nonpolar saturates to the highly polar asphaltenes and the ratio H/C is also observed to decrease gradually from nearly 1.8 to 1.2. As expected, the contents of heteroatoms also increase from saturates to asphaltenes, though the increase in sulfur is not as pronounced as nitrogen. Molecular Weights. Molecular weights of the bitumen fractions listed in Table I1 show some mixed trends. It is observed that, for a given bitumen, the molecular weight values gradually increase from saturates to asphaltenes with the exception of AR resins, which was observed to have a higher values than asphaltenes. A fractionwise comparison of both AR and SS bitumens indicates that for saturates and aromatics, the molecular weight values are marginally higher for SS bitumen. A similar comparison of resin and asphaltenes fractions reveals reversed trends. '3c NMR Spectroscopy. The 13CNMR spectra of the saturate fractions of both AR and SS bitumens are pres-
Bukka et al.
336 Energy & Fuels, Vol. 5, No. 2, 1991
Table 11. Elemental Compositions of the Bitumens and Their Fractions Derived from Asphalt Ridge (AR) and Sunnyside (99) Tar Sands weight percent bitumen t-*mc eC H N S 0“ H / C ratio aromaticit? mol wt represent. mol formula‘ -. - __- ._ n.d.d 490 3.03 1.56 11.00 0.44 84.47 1.06 AR bitumen 588 3.40 1.45 n.d. 10.28 0.50 84.92 0.90 ss 388 0.50 1.78 0.04 0.07 12.83 0.09 86.51 AR saturates
aromatics resins asphaltenes
ss ss AR ss AR ss AR
86.57 85.83 85.85 81.45 83.38 84.26 83.49
12.89 10.51 10.51 10.81 10.37 9.19 8.68
0.10 0.65 0.53 1.57 1.13 2.23 1.56
0.04 0.41 0.43 0.42 0.40 0.51 0.64
0.40 2.60 2.68 5.75 4.80 3.71 5.63
403 465 480 1510 670 1100 2240
0.04 0.31 0.27 0.18 0.17 0.26 0.27
1.79 1.47 1.47 1.59 1.49 1.31 1.23
a Values were obtained by difference. No provision was made for metals present in all these fractions which were considered to be very small in comparison. *Determined from the 13C NMR spectroscopic analysis (see text for the definition). cFormula based on the elemental analysis and molecular weight listed in the table. dn.d. not determined.
ented in Figure 2. The intense bands in these spectra were confined to the region 10-70 ppm, which is characteristic of the presence of saturated carbon atoms. The unsaturated or what is referred to as aromatic carbon content was deduced to be very small from its relatively weak bands observed in the region 90-170 ppm. The proportion of the aromatic carbon to the sum of the saturated and aromatic carbon was calculated from the integration of band intensities. These values are often referred to as aromaticitiesI4and are presented in Table I1 for all fractions. The spectra of saturate fractions exhibited identical features with the same relative intensities for all major bands and also have shown the similar aromaticities. Since these are complex mixtures of both paraffinic and naphthenic hydrocarbons, band assignments are a t best a qualified guesswork. Consequently, a precise analysis is considered to be extremely difficult. However, based on the study of model compounds like phytane and farnesane, some attempts have been made to identify the observed bands in I3C NMR spectra of Athabasca bitumen fraction^'^ and coal extracts16 to carbon atoms with certain structural characteristics. Although these band assignments are useful for identification purposes, the overall information that can be derived from the ‘3c Nh4R spectra still appears to be of limited value. The spectra of aromatic and resin fractions of both AR and SS bitumens have also shown remarkably similar features. The only difference observed was in the relative intensities of some bands in the resin fractions. For both aromatic and resin fractions, the calculated aromaticities were found to be slightly higher for AR bitumen (Table 11). In the comparison of asphaltene fractions, some distinct difference were observed between AR and SS bitumens. The spectra of these fractions are presented in Figure 3. Differences are noted both in the number of bands and their relative intensities. However, the aromaticity values for asphaltene fractions of both the bitumens are similar (Table 11). FTIR Spectroscopy. Infrared spectra of the saturate fractions of both the bitumens were very simple, consisting of only absorption bands arising from C-H stretching and bending vibrations. The other bands were observed to be extremely weak and appear almost to be of no significance. This observation indicated an effective separation of non polar fractions from the bitumens. As expected, both fractions have identical features. A typical spectrum of (14) Axelson, D.E.Fuel Process. Technol. 1987, 16, 257. (16) Thiel, J.; Grey, M. AOSTRA J. Res. 1988,4,63. (16) Thiel, J.; Wachowska, H.Fuel 1989,68, 758.
80 6 0 40 20 0 PPM Figure 3. 13C NMR spectra of (a) SS and (b) AR asphaltenes fractions. 140
120 100
A 0.3 b
:“ n
0.0 4000 3500
Figure
3000
2500 2000 W8venufnber.s
I500
I000
4. FTIR spectrum of SS saturates fraction.
a saturate fraction, representative of both the bitumens, is presented in Figure 4. Aromatic fractions of AR and SS bitumens show marked differences. Spectra of both aromatic fractions are presented in Figure 5. The majority of the bands observed in the spectra of all bitumen fractions and their probable assignments are listed in Table 111. A comparison of the spectra (Figure 5) reveals that the difference between AR and SS aromatic fractions lie in three different regions viz., 3648, 1704, and 1330-1100 cm-l, which are characteristic regions involving mainly oxygen functional groups. From the elemental analysis of the aromatic fractions, as listed in Table 11, it is clear that both of the aromatic fractions have similar amounts of oxygen. However, a close exam-
Utah Tar Sand Bitumens
Energy & Fuels, Vol. 5, No. 2, 1991 337
n
1
2.04
0.5 A b 1.5-
'
0
r
&5O.0.00 O
b 1.0-
1300
a
1200
1100
P 3500 3000 2500 2000 1500 1000
4000
Wavenumbers
0.64
2.0A b
S
I I
I
s 1.50
r b 1.0-
n C
e 0.5-
0.07
4000
I
I
I
3500
3000
2500
4000
I
2000 Mavenumbers
1500
Figure 5. FTIR spectra of (a) AR and (b) SS aromatic fractions
(expanded portion in the inset).
Table 111. Absorption Bands Observed in the FTIR Spectra of Bitumen Fractions and Some of Their Probable
Designations ~
3500
1000
~
~
wavenumber, cm-l band designation 3648 0-H stretching in non-H-bonded alcohols 3500-3200 0-H and N-H stretching in multiple H-bonded compounds 2956 C-H stretching (asymmetrical) in -CH3 2929 C-H stretching (asymmetrical) in -CH2 2852 C-H stretching (svmmetrical) in C H , C-H stretching (sy"etrica1) in -CH; 2829 2728 C-H stretching at the bridge head position in naphthenes 1700 C=O stretching in carboxylic acids 1600 C = C stretching (conjugated) and skeletal ring vibrations 1460 C-H deformation in -CH2 C-H deformation in -CH3 1370 1314 0-H deformation in phenols 1267 C-0 stretching in aromatic and cyclic ethers 1231 C-O-H deformation 1196 C-0 stretching in tertiary alcohols 1158 C-0 stretching in alkyl ethers 1121 aliphatic C-0 and/or S=O stretching 1071 C-O-C stretching in alkyl ethers and C-N stretching 1033 C-0-H deformation C-H out of plane bending vibration from a 879 penta-substituted benzene derivative 810 two adjacent C-H groups on the aromatic ring 747 four adjacent C-H groups on the aromatic ring -CH2 rocking 727
ination of the spectra indicates that the oxygen content in the aromatic fractions of both the bitumens is distributed in a very different manner. In the aromatic fraction of the AR bitumen (Figure 5a), it is noted that a good portion of the oxygen is distributed in a functional group which absorbs at 3648 cm-'. This band indicates the presence of non-hydrogen-bonded free -OH groups. A second portion of the oxygen is seen tied up in bands appearing in the region 1330-1100 cm (Figure 5a, inset), which is characteristic of CH20-Hand CO-H deformation modes of vibrations and antisymmetrical bridge stretching for cyclic ethers." Finally, the remaining portion of the
3000
25bO 2000 Wavenumbers
1500
1000
Figure 6. FTIR spectra of (a) AR and (b) SS resin fractions.
oxygen is considered to be bound in the carboxylic groups, as indicated by a weak absorption band observed at 1704 cm-l 18
In the case of the aromatic fraction of SS bitumen (Figure 5b), distribution of oxygen content is seen to be opposite to that of AR bitumen. Here it appears that most of the oxygen is distributed in carboxylic groups as indicated by the relatively more intense band at 1704 cm-'. The amount of oxygen engaged in free -OH groups was observed to be relatively small, and in the compounds which give rise to absorption bands in the region 1330-1100 cm-' is insignificant, as can be seen from the intensities of respective bands (Figure 5b, inset). These subtle differences in composition, as indicated by FTIR spectra, are implied to contribute significantly to the physical properties of bitumen, as will be discussed later. Finally, there are also other small differneces noticed in the region 900-700 cm-', a region which is characteristic of unsaturated carbon and substitution in six-membered ring aromatic compounds. FTIR spectra of the resin fractions also show some interesting differences between AR and SS bitumens. The spectra of these fractions are presented in Figure 6. From the design of the fractionation scheme, it is expected that resins would accumulate polar compounds. Indeed the spectral features do indicate the presence of extensive polar functional groups in these fractions. Other than differences in the relative intensities of some bands, the spectra of AR and SS resin fractions (Figure 6)appear identical. Distinct features of the resin fractions include bands at 3648 and -3400 cm- due to hydrogen-bonded 4 H , NH2,and N-H groups, and at 1700 cm-' due to carboxylic group containing compounds. Occurrence of a broad band at -3400 cm-' coupled with the band at -1700 cm-' strongly indicates the participation of the carboxylic groups in hydrogen-bonded interactions. In comparing the resin fractions of AR and SS bitumens, the trend observed in the aromatic fraction, viz., the relatively higher carboxylic N
(17) Rhoads, C.A.; Painter,P. C.; Given, P. H. Int. J. Coal Geol. 1987, 8, 69. (18) Speight, J. G.; Moachopedis,S. E. Prepr.-Am. Chem. Soc., Diu. Pet. Chem. 1981,26, 907.
Bukka et al.
338 Energy & Fuels, Vol. 5, No. 2, 1991
o,221
;0.20
4000
3500
2500
3000
2000
1500
1000
Usvenunbers
1
0,304 A
I
1I ,
r
I
I I
t
4000
I
I
3500
3000
I
1
I
I
2500
2000
I500
1000
Uavenumbers
Figure 7. FTIR spectra of (a) AR and (b) SS asphaltene fractions.
acid content of SS fraction, was not only maintained but more pronounced. Asphaltenes are known to possess the highest molecular weight and be the most polar of the bitumen fractions. FTIR spectra of these fractions are presented in Figure 7. From their spectra it becomes apparent that the polar groups, particularly those involving oxygen, nitrogen, and sulfur, are noticeably lower in content than those present in the corresponding resin fractions. Perhaps the high polarity of the asphaltene fractions is derived from the presence of porphyrins and their complexes with transitional-metal ions. In general, the spectra of these fractions appear to be simpler than that of the corresponding resin fractions. Once again a comparison between these two asphaltene fractions indicates that the carboxylic groups in the SS asphaltenes, as inferred by the intensity of the band observed at -1700 cm-', were more abundant than in the corresponding fraction of AR bitumen. Discussion Fractionation and characterization of AR and SS bitumens have been done with the purpose of relating the bitumen viscosity to chemical composition. It has been reported6 earlier that the asphaltene content of the bitumen plays an important role in determining the rheological properties of bitumen. As pointed out in the Introduction, a comparison of the asphaltene content of two bitumens (viz., AR and Athabasca) indicates that there may be other factors which influence viscosity. From the results of this study it is demonstrated that, in addition to asphaltenes, other features of chemical composition, particularly the presence of polar functional groups in some bitumen fractions, could play an important role in the determination of bitumen viscosity. The idea that bitumen viscosity can be related to its chemical composition was a consequence to an earlier conceptual model developed for bitumen by Nellensteyn.' It was further revised by Pfeiffer213and applied recently to the study of asphalts by Marvillet.' The essential feature of this model is that bitumen is viewed as a colloidal system in which solid asphaltenes are suspended as micelles in viscous oils and the suspension is stabilized by other constituents of bitumen such as aromatics and resins.
Based on this theory, differences in the rheological properties of various bitumens were accounted for by differences in the degree of peptization of the asphaltene micelles, thus resulting in differences in the tendency to form gels.3 In turn, the degree of peptization or the tendency of the bitumen to gel depends on the nature of the colloidal interactions and the type of functional groups present in the components which stabilize the colloidal asphaltenes. Although it is difficult to find experimental evidence for this model, it nonetheless helps to define variables which may have influence on bitumen viscosity. Among these variables are the nature and amount of asphaltenes, interaction between the asphaltenes and their suspensionstabilizing components, and finally the oil medium itself in which the asphaltenes are suspended. From the fractional compositions of AR and SS bitumens, as shown in Table I, it is clear that there are significant differences between the two bitumens in every major fraction. Perhaps it is superfluous to conclude that differences in every fraction have an impact on bitumen viscosity. In an ideal comparison of the viscosities of two different bitumens, with other properties being similar, a bitumen with a larger proportion of asphaltenes and smaller proportion of saturates would be expected, based on the studies of Christensen et al.,5 to have higher viscosity. However, the presence of merely a higher proportion of the asphaltenes in a bitumen cannot be assumed to render higher viscosity, as illustrated by the comparison of Athabasca and Asphalt Ridge bitumens earlier in the Introduction. In view of the conflicting results from earlier studies of Christensen et aL5 and the previously cited comparison of Athabasca and Asphalte Ridge bitumens, it becomes clear that the higher asphaltenes content of SS bitumen need not necessarily contribute to its higher viscosity. Differences in the elemental analysis of AR and SS bitumen fractions, as presented in Table 11, provide no consistent clues to explain the viscosity differences. Elemental analysis and molecular weights of both the bitumens are also included in Table 11. A comparison of the elemental analysis of both the bitumens show some differences in all elements except carbon. Perhaps the notable differences observed are in the values of atomic ratio H/C and molecular weights. The higher value of H/C ratio, which indicates higher proportion of saturated hydrocarbons, and lower molecular weight for AR bitumen may possibly explain its lower viscosity. The elemental analysis of bitumen fractions also show similar trends as observed for the composite bitumens. While nitrogen content was higher in AR fractions, it was found that sulfur was marginally higher in SS fractions. Comparisons of aromaticities and even molecular weights were of no practical utility in differentiating the two bitumens. The H/C ratio values for AR resin and asphaltene fractions were larger than those of SS, which is understandable in view of larger proportion of saturates in AR bitumen. Even the 13CNMR analyses of the fractions of AR and SS bitumens seemed inadequate to provide a meaningful explanation of the viscosity difference between the two. In three of the four fractions, viz., saturates, aromatics, and resins, both bitumens seem to have identical spectral features. This may be, in part, due to the relatively low sensitivity of the instrument and also the complexity of the mixture being analyzed. Asphaltenes have shown some small differences but the available band characterization techniques do not enable a better identification. In summary, a simple '% NMR analysis of the bitumen fractions
Utah Tar Sand Bitumens
was not of much help in the development of a relationship between viscosity and chemical composition. However, with the use of NMR technique and on the basis of theoretical considerations of molecular motion and intermolecular attractions, Netzel et al.19 have recently developed a relationship between the viscosity of bitumen and spinspin relaxation rates for three bitumens of widely differing viscosities. Among the characterization methods employed in this study, FTIR analysis was found to be the most useful in terms of identification of distinct functional groups and their variations in different bitumen fractions. This may be attributed to its superior sensitivity. By use of the data from FTIR analysis, distinct compositional differences between the bitumen fractions have been identified. In the following paragraphs these differences are further examined and a tentative relationship is suggested between chemical composition and viscosity, in agreement with the bitumen model of Nellensteyn.' The FTIR profiles in both saturate fractions are identical. However, the similarities between the two bitumens diminish as the comparison was extended to other fractions. Beginning with the aromatic fractions, distinct differences emerge (Figure 5 ) . Since all the absorption bands in the spectra cannot be satisfactorily accounted for, due to the complex nature of the sample, attention was focused only on the regions where differences were observed. These regions are 3648, -1700, and 1330-1100 cm-', which are characteristic to the polar functional groups such as free 0-H, N-H, C-OH, C-O-C,and C=O. Of these, it is necessary to establish, in a qualitative manner, which group plays an important role in the enhancement of the gel structure as described earlier in the discussion. The band at 3648 cm- was attributed to either free non-hydrogen-bonded 0-H group, be it from an alcohol, phenol and a carboxylic group, or to free N-H of amine. Though this indicates that the -OH or the amine group is free in the separated aromatic fraction, it may be expected to take part in hydrogen-bond formation provided there are no structural or steric constraints exerted on these groups. However, if the band at 3648 cm-' is due to the -OH stretching of a carboxylic group, its corresponding carbonyl group will influence the aforementioned interactions. The compounds which give rise to the bands observed in the region 1300-1100 cm-', particularly for the AR aromatics fraction (Figure 5 ) , were not considered to be effective to increase intermolecular attractions. The band observed at -1700 cm-1 can be unequivocally assigned to a carbonyl stretching vibration of the carboxylic acid group.18 The only mechanism for carboxylic group participation in the stabilization of asphaltene suspension or the gel structure enhancement is through hydrogenbond formationz0with alcohols, carboxylic acids, amines, amides, and imines. Presumably all these compounds are present in the bitumen, although some of them are in low concentrations and defy available detection methods. Thus the relative intensities of this band in fractions from two different bitumens, in addition to the differences in the asphaltene contents, would enable a qualitative comparison of their viscosities. It must be pointed out that, in addition to carboxylic acids, there may be other polar compounds containing heteroatoms such as nitrogen and sulfur which may potentially contribute to the stabilization of asphaltene suspension in the bitumen. But their contribution is expected to be considerably smaller for two reasons, viz., hydrogen bonds formed with molecules con(19)Netzel, D.A,; Turner, T. F.&el Sci. Technol. Znt. 1990,8,379. (20) Moschopedis, S. E.; Speight, J. G. Fuel 1976,55, 187.
Energy & Fuels, Vol. 5, No. 2, 1991 339 taining such heteroatoms are energetically weaker and their abundance in the bitumen is much smaller than oxygen. A comparative examination of the FTIR profiles of the aromatic, resin, and asphaltene fractions of AR and SS bitumens indicates consistently that the SS fractions show relatively intense bands at 1700 cm-', which are attributed to the presence of carboxylic acid groups. These acidic groups are responsible for the stabilization of the suspended asphaltenes in bitumen through interactive hydrogen-bond formation. If the resin fraction is assumed to stabilize the asphaltnes suspension, as inferred by the model, then one would expect a larger proportion of the carboxylic acids in both the asphaltenes and resins would give rise to enhanced interactions, thus registering a high viscosity as a result. From the results obtained in this study it also leads to the conclusion that resin-asphaltene interactions offer more stability to the suspension than the aromatics-asphaltene interactions, since the carboxylic groups are more abundant in resins than in aromatics fraction. The difference between the asphaltenes of AR and SS bitumens (Figure 7), though small in terms of their carboxylic acid contents, indicates that their chemical nature could also play an important role in the determination of the bitumen viscosity. The material balances obtained in the fractionation of both AR and SS bitumens were not similar. The losses incurred were 0.3% and 2.2% for AR and SS bitumens, respectively. These losses were mainly due to the polar compounds which may have strongly adsorbed on both Fuller's earth and neutral alumina used in the fractionation scheme. These compounds could not be eluted with methanol, which is considered to be the most polar solvent in this types of separations. Although these polar compounds were not separated and characterized, they are also expected to influence the viscosity in the composite bitumen. The results of this study indicate that the viscosity of bitumen may be dependent on several variables. Among these are the proportion of saturates and asphaltenes, and the chemical composition of the resin and aromatic fractions present in the bitumen. Particularly, the presence of carboxylic acids in the polar constituenb of the bitumen is inferred to positively influence the property of viscosity. The higher viscosity of SS bitumen can be explained not only by its relatively higher amounts of asphaltenes, but also by its higher carboxylic acid content in the overall composition of the SS bitumen. These carboxylic acids, which are more abundantly present in the resin fraction, effectively engage the suspended solid asphaltenes in the saturate fraction through hydrogen-bond formation. These enhanced interactions between resins and asphaltenes lead to greater stabilization of the suspension, thereby increasing the viscosity. Although the results obtained from the study of these two bitumens point to the outlined conclusions, further work involving a number of bitumens which vary significantly in their viscosities and asphaltene contents, including Athabasca bitumen, is in progress.
-
Conclusions In this study two Utah tar sand bitumens, AR (Asphalt Ridge) and SS (Sunnyside), which differ vastly in their viscosities, were fractionated in a preparative manner and the recovered fractions were subjected to a comprehensive characterization. From the analysis it is noted that, while other methods of comparison have shown relatively small differences between the two bitumen fractions, FTIR spectroscopy was shown to offer a superior diagnostic method to distinguish compositional differences between
Energy & Fuels 1991,5, 340-346
340
bitumens' fractions. The two bitumens were found to differ not only in the relative amount of a given fraction, but also in that fraction's chemical composition. The most notable variation was found in the presence of carboxylic acids in most of the fractions. For a given bitumen, these carboxylic acid compounds were found to reach a maximum level in the resin fraction. As a result, a comparison of the two bitumens in terms of their resin fractions provided the greatest difference. Differences in the viscosities of these two bitumens were explained to arise not only from the differences in their asphaltene contents, but also due to the differences in the chemical compositions of aromatic, resin, and asphaltene fractions. The SS bitumen with the higher viscosity was observed to contain a relatively larger proportion of compounds containing carboxylic groups for a particular fraction. The presence of larger amounts of these compounds in the SS bitumen appears
to lead to more extensive hydrogen-bonding interaction between aromatics and resins with the suspended asphaltenes in the composite bitumen. According to Nellensteyn's model' these enhanced interactions result in higher viscosity for the bitumen with higher carboxylic acid content. The results from this study show that bitumen viscosity can be correlated not only with asphaltene content, but also with the polar functionalities of the resin and aromatic fractions of a particular bitumen. Research in this area is being continued in order to provide a more comprehensive analysis in the future. Acknowledgment. We thank members of the tar sand research team at the University of Utah for encouragement and review of the work. Also it should be noted that this research has been supported by U.S.Department of Energy, Grant No. DE-FC21-89MC 26268.
A New Coal Flash Pyrolysis Method Utilizing Effective Radical Transfer from Solvent to Coal Kouichi Miura,* Kazuhiro Mae, Seiji Asaoka, Tomonori Yoshimura, and Kenji Hashimotot Research Laboratory of Carbonaceous Resources Conversion Technology, and Department of Chemical Engineering, Kyoto University, Kyoto, 606 Japan Received August 16, 1990. Revised Manuscript Received January 4, 1991 A new coal flash pyrolysis method was developed for drastically increasing the total volatile matter and the liquid products under mild conditions. The original idea of this method lies in the realization of effective hydrogen transfer from the hydrogen donor solvent to the coal fragments during the flash pyrolysis. To realize it, we found the contact at the molecular level between the solvent and the coal functional groups was essential in addition to the matching of the dehydrogenation rate of the solvent and the primary decomposition rate of coal. Contact at the molecular level was achieved by swelling the coal with solvent at 100-250 "C under a pressurized atmosphere (1MPa). When two coals were swollen by tetralin, tetralin penetrated into the micropores of its molecular size by enlarging the pores. Pyrolyzing the swollen coal in a Curie-point pyrolyzer at 670, 764, and 920 "C in a helium stream at atmospheric pressure resulted in production of 56% of volatile matter for Taiheiyo (TC, Japanese subbituminous coal) and surprisingly 67% for Morwell (MW, Australian brown coal). A maximum liquid yield from MW reached more than 4270, which was larger by 24% than that from the raw coal. It was clarified that this significant increase was brought about by a physical effect as well as the effective hydrogen transfer from/via tetralin.
Introduction Since pyrolysis of coal is the first step of all coal conversion processes, its understanding is essential for the effective use of coal. The pyrolysis of coal consists of two series of reactions: the first is the primary decomposition, which consists of the formation of radical fragments and their stabilization, the other is the secondary gas-phase reaction of the gaseous components produced by the primary decomposition. The product yield during the pyrolysis depends strongly on coal type and operating conditions of pyrolysis. In several coal pyrolysis methods, flash pyrolysis is a promising process for producing chemicals such as benzene, toluene, and xylene (BTX),but the yield of liquid products including BTX (tar) is limited because of low hydrogen to carbon ratio in coal. It is necessary to supply effectively Department of Chemical Engineering.
hydrogen from other sources to coal for increasing the tar yield. Several methods have been proposed to supply hydrogen to coal during the pyrolysis. Hydropyrolysis is a representative one of such methods. This generally requires severe reaction conditions: high hydrogen pressure, high temperature, and long residence time. Even if the total volatile matter was increased by this method, the increase of liquid yield could not be expected as Wanzl indicated.' Flash pyrolysis in methane,2s3toluene? or methanol6 atmosphere was proposed. These methods intended to supply hydrogen or methyl radicals produced by the de(1)Wanzl, w.Fuel Process. Technol. 1988, 20,317-336. (2) Calkins, W.H.; Bonifaz, C . Fuel 1984,63, 1716-1719. (3)Steinberg, M.;Fallon, P. T. Hydrocarbon Process. 1982, Nou. 92-96. (4) Doolan, K. R.; Mackie, J. C. Fuel 1986,64, 400-405. (5)Run-Ling, R.; Itoh, H.; Makabe,M.;Ouchi, K. Fuel 1987, 66, 643-653.
OSS7-Q624/91/25Q5-Q34Q~Q2.5Q/Q 0 1991 American Chemical Society