Upgrading of Alberta's Heavy Oils by Superacid-Catalyzed

558

Energy & Fuels 1999, 13, 558-569

Upgrading of Alberta’s Heavy Oils by Superacid-Catalyzed Hydrocracking Otto P. Strausz,* Thomas W. Mojelsky, and John D. Payzant Chemistry Department, University of Alberta, Edmonton, AB T6G 2G2, Canada

George A. Olah and G. K. Surya Prakash Hydrocarbon Research Institute, University of Southern California, Los Angeles, California 90089-1661 Received April 28, 1998. Revised Manuscript Received February 2, 1999

The HF‚BF3 superacid-catalyzed hydrocracking of tar sands bitumens and asphaltenes leads to deep-rooted chemical changes which affect both the aliphatic and aromatic constituents of the feedstocks even under mild experimental conditions, resulting in high yields of volatiles and liquid products. In contrast to conventional catalytic hydrocracking following free-radical mechanisms, the superacid-catalyzed reaction follows ionic mechanisms, yielding products very different from those following the free-radical reactions.When methylcyclohexane (MCH) is employed as a hydrogen-donor solvent, new products resulting from the oligomerization of MCH appear. MCH neat, without bitumen or asphaltene, does not oligomerize under identical conditions. These preliminary results point to the commercial potential of using the volatile HF‚BF3 superacid in the upgrading of bitumen and the oligomerization of cycloalkanes.

The world’s heavy oil and bitumen reserves far exceed those of conventional oils. These materials and the residue fractions of conventional petroleum refining cannot be used directly as feedstocks in oil refineries and must be upgraded first to synthetic crude oils. In the upgrading process, the hydrogen-to-carbon ratio is increased either by rejection of carbon, addition of hydrogen, or both. This is accomplished in the cracking or hydrocracking operations either with or without a suitable catalyst. At the same time, such operations reduce the molecular weight and remove much of the undesirable heteroatoms, NOS, metals, clays, and other minerals present in the heavy oils. Of the current upgrading technologies, the most advanced and efficient one is the catalytic hydrocracking process in which, parallel to the cracking reactions, partial hydrogenation of the feed takes place. The catalysts, Co/Mo and Ni/Mo, are applied either in fixedbed or fluidized, ebulliated, form. The efficiency of the catalysts is deleteriously affected by the deposition of solid polymeric materials and coke, mainly from the cracking of the high molecular weight components of the feed, on the surface of the catalysts, clogging the pore system and thus reducing the rate of hydrogenation. Ultimately, the deposition of coke and metals on the catalyst surface will spoil the catalytic performance.1,2 This becomes a serious operational problem when feedstocks such as bitumens, which are high in asphaltene, that is, high molecular weight components, (1) Gray, M. R. Petroleum Residues and Heavy Oils; Marcel Dekker Inc.: New York, 1994. (2) Fluid Catalytic Cracking III; Ocelli, M. L., O’Connor, P., Eds.; ACS Symposium Series 571; American Chemical Society: Washington, DC, 1994.

are hydrocracked. Recovery and replenishment of spent catalysts is a major expenditure in the process. An alternative to conventional catalytic hydrocracking of heavy feedstocks would be the use of a gaseous catalyst which would be free of the operational problems mentioned above. One such potential catalyst would be the gaseous superacid tetrafluoroboric acid, HBF4 or HF‚BF3. Superacids3 are acids whose activity function Ho (defined somewhat analogously to pHo) is equal to or less than -12. Superacids, as in the present case, may arise from the mixing of a strong protic acid with a strong Lewis acid. Olah4 and Olah et al.5-7 used tetrafluoroboric acid for the catalyzed depolymerizationstermed hydroliquefactionsof coals under mild conditions. Employing H2, HF‚BF3, and heating at 105 °C for 4 h it was possible to bring about coal depolymerization so that about 90% of the residue could be solubilized in pyridine. Upon heating to 150-170 °C, cyclohexane extractability of up to 22% was achieved. Pyridine is a more powerful solvent and is able to dissolve larger clusters of aromatic molecules, whereas cyclohexane is more specific in dissolving aliphatic hydrocarbon molecules. The HF‚BF3 system has Ho ≈ -16 and is useful in that it is nonreducible (hence will not oxidize products) and is gaseous, thus permitting the corrosive superacid com(3) Olah, G. A.; Prakash, G. K. S. Superacids; John Wiley & Sons: New York, 1985. (4) Olah, G. A. U.S. Patents 4,373,109 and 4,394,247, 1983. (5) Olah, G. A.; Bruce, M. R.; Edelson, E. H.; Husain, A. Fuel 1984, 63, 1130-1137. (6) Olah, G. A.; Husain, A. Fuel 1984, 63, 1427-1431. (7) Olah, G. A.; Bruce, M. R.; Edelson, E. H.; Husain, A. Fuel 1984, 63, 1432-1435.

10.1021/ef980098r CCC: $18.00 © 1999 American Chemical Society Published on Web 04/21/1999

Upgrading of Alberta’s Heavy Oils

ponents to be easily removed from the liquefaction reaction mixture. When a hydrogen-donor solvent such as isopentane or methylcyclohexane is used, the efficiency of the conversion, in terms of cyclohexane extractability of treated coal, is improved. In Olah et al.’s papers on the superacid treatment of coal, no product analysis data were reported. A study, however, was published6 on model compounds thought to represent the molecular bridging groups holding coal “monomers” together. Dibenzyl, diphenylmethane, and biphenyl were used to represent alkylidene and diphenyl bridges. Under the superacid conditions (BF3:HF:H2, heat), near quantitative conversions to products such as benzene, toluene, anthracene, p-xylene, and some volatile hydrocarbons were obtained. The ether bridges were mimicked by dibenzyl ether, benzyl phenyl ether, and diphenyl ethers which, under the conditions of the experiment, gave similar decomposition products and phenol in near quantitative yields. The sulfide bridge was represented by benzyl phenyl sulfide, dibenzyl sulfide, and diphenyl sulfide. Again, similar products were obtained in high conversion. Some of the sulfur was trapped in trace amounts as benzenethiol and thianthrene. The superacid-catalyzed hydrocracking of petroleum and petroleum fractions has also been reported.8 In these studies, however, the superacids were prepared by the mixing of anhydrous HF and metal halides, primarily tantalum pentafluoride which is a solid with mp 97 °C. In the search for new technologies aimed at effecting higher conversions of bitumen to low molecular weight pentane solubles while at the same time minimizing the coke yields, the chemical degradation of asphaltene has often been studied as a model system. Some years ago Strausz and co-workers9 investigated the Lewis-acidassisted degradation of Athabasca asphaltene under various conditions of temperature and hydrogen pressure. Both AlCl3 and ZnCl2 were shown to catalyze coal liquefaction and solubilization,10 and acid-catalyzed depolymerization became a useful method for structural investigations of coal. With Athabasca asphaltene, AlCl3 was found to be somewhat effective below 100 °C but above 250 °C ZnCl2 was a much better catalyst, especially in the presence of H2. The percentage conversion to pentane solubles increased from 39% to 51% between 300 and 400 °C, but this was accompanied by a more than 2-fold product loss in the form of volatiles (1636%). Relatively large amounts of coke were produced, 25% at 300 °C and 10% at 400 °C. Because of the large amounts of coke and cracked gases formed during the reaction, it was concluded that these catalysts would not be particularly effective as part of the bitumen upgrading operations. In the present investigation, the intention was to explore the feasibility of using the HF:BF3 system of superacid catalysts for the depolymerization/hydrogenation of Athabasca oil sand bitumens and asphaltenes, (8) Siskin, M.; Wristers, J. P.; Purcelli, J. J. U.S. Patent 3,901,790, August 25, 1975. (9) Ignasiak, T.; Bimer, J.; Samman, N.; Montgomery, D. S.; Strausz, O. P. In Chemistry of Asphaltenes; Bunger, J. W., Li, N. C., Eds.; Advances in Chemistry Series 195; American Chemical Society: Washington, DC, 1981; pp 183-201. (10) Butler, R.; Snelson, A. Fuel 1980, 59, 93-96.

Energy & Fuels, Vol. 13, No. 3, 1999 559 Table 1. Elemental Compositions of the Main Fractions % N

O

S

MW (g‚mol-1)

wt %a

C

H

Athabasca Cold Lake

83.2 84.0

10.3 10.5

Bitumen 0.4 1.6 0.2 1.0

4.6 4.7

540 490

Athabasca Cold Lake

79.9 80.6

8.3 7.4

Asphaltene 1.2 3.2 1.2 1.8

7.6 6.5

3600b 6000c

14.6 16.6

Athabasca Cold Lake

84.4 84.2

10.6 11.0

Maltene 0.9 0.6

3.9 3.9

435 430

85.4 83.4

100 100

a Weight percent of bitumen. b From Cyr, N.; McIntyre, D.; Toth, G.; Strausz, O. P. Fuel 1987, 66, 1709-1714. c From Selucky, M. L.; Chu, Y.; Ruo, T. C. S.; Strausz, O. P. Fuel 1978, 57, 9-16.

as developed by Olah et al.,5-7 for the depolymerization of coal. Under mild conditions (200 °C, 500 psi H2), this process appeared to be inviting for the following reasons: (a) it follows an ionic mechanism which is completely different from that involved in thermal or thermocatalytic cracking and therefore would be expected to lead to atypical, possibly more valuable, products than those obtained in a cracking step; (b) the temperature and hydrogen pressure required are much lower than in conventional hydrocracking; (c) a gaseous catalyst is free of the problems associated with solid catalysts (such as poisoning and core plugging); (d) both HF and BF3 can be readily separated from the product oil by distillation and be recycled; (e) neither HF nor BF3 would cause any oxidative degradation of the products because they are not reducible; and last, (f) it should be pointed out that practically no fluorine incorporation into the products takes place. Experimental Section All solvents used in this investigation were distilled prior to use. Column chromatography was done using silica gel 60 (Terochem Laboratories, particle size 0.063-0.200 mm, activity according to Brockman, 2.3). The oil sand samples were obtained from the Syncrude pit located 18 m below the surface at mine site 1-2-93-11-W4 and from the Cold Lake deposit. Gas chromatographic analyses were performed on a HewlettPackard HP5730A gas chromatograph with a 18850 GC terminal in the flame ionization mode. A 30 m × 0.252 mm J&W fused silica capillary column coated with DB-1 was used. The gas chromatograph was programmed to increase at a rate of 4 deg/min with the initial temperature set at 75 °C. A Varian Vista gas chromatograph with a splitless injector and helium carrier gas with a 30 m × 0.32 mm J&W fused silica column coated with 0.25 µm DB-1 was used in the gas chromatography-mass spectrometry (GC-MS) experiments. Samples were injected in the form of a toluene solution. The gas chromatograph was programmed from 75 to 300 °C at a rate of 4 deg/min. The end of the column was introduced into the ion source of a VG 7070E mass spectrometer. Typical mass spectrometer operating conditions were as follows: transfer line, 290 °C, ion source 250 °C, electron energy 45 eV. Data acquisition was done under VG 11-250 software using a PDP 11/24 computer. The mass range m/z ) 50-600 Daltons was scanned every 1.0 s. The elemental analyses of the isolated products were performed by the Microanalytical Laboratory of the Chemistry Department of the University of Alberta. Solvent Extraction of the Oil Sand Samples. The oil sand samples were subjected to solvent extraction according

560 Energy & Fuels, Vol. 13, No. 3, 1999

Strausz et al.

Table 2. Analysesa of Superacid-Catalyzedb Hydrocracked Asphaltene and Bitumen reaction no. (product wt, g) 1 (3.57) 2 (2.73) 3 (19.67) 4 (6.08) 5 (5.49) a

pentane solubles yield % saturates

reaction Athabasca asphaltene, 2 g, 24 h + some MCH Athabasca bitumen, 5.5 g, 24 h, 170 °C Athabasca bitumen, 5.5 g, 24 h, 170 °C + 30-mL MCH Cold Lake asphaltene, 2.5 g, 24 h, 184 °C + 30-mL MCH Cold Lake asphaltene, 2.5 g, 2.5 h, 188 °C + 30-mL MCH

2.26 g, 113%

CH2Cl2 solubles (asphaltene)

80

CH2Cl2 insolubles

total yield (wt %)

0.9 g, 45%

0.4 g, 20%

178

0.81 g, 14.7%

0.1 g, 1.8%

50

1.82 g, 33%

7.3

18.9 g, 344%

76.1

0.77 g, 14%

358

4.95 g, 198%

70

1.13 g, 45%

243

5.28 g, 211%

81

0.21 g, 8.4%

219

The percent yields refer to percent of product. b HF, BF3, H2 (50 mL, 500, 500 psi). Table 3. Elemental Analyses and Average Molecular Weights of Selected Superacid-Treated Samples maltene (%)

reaction no. 1 2 3 5 a

av

MWa

339 402 256 313

asphaltene (%)

C

H

N

S

O

C

H

N

S

O

79.25 87.51 87.50

9.50 11.25 10.61

0.26 0.08 0.26

6.68 0.18 0.87

3.43 1.49 0.92

59.08

5.75

1.07

5.65

5.55

74.08

6.71

1.20

4.71

4.07

Average molecular weight of the maltene.

to an established procedure.11 Basically, 300 g of the core sample was placed in a thimble, which was inserted into a Soxhlet extractor of appropriate size. About 700 mL of CH2Cl2 was added to a 1-L round-bottomed flask equipped with a magnetic stirrer and reflux condenser. When the apparatus was assembled, the solvent was heated and refluxed for 24 h, after which the solvent returning to the flask was clear. Upon cooling, the now black CH2Cl2 solution was concentrated, the solvent evaporated, and the residual bitumen weighed. The quantities obtained, along with typical elemental compositions of Athabasca and Cold Lake bitumen, asphaltene, and maltene fractions, are given in Table 1. Although these analyses were done on different oil sand samples, the elemental compositions of the bitumens, asphaltenes, and maltenes do not vary appreciably within a deposit. Separation of Bitumen into Asphaltene and Maltene. The bitumen was dissolved in 25 mL of CH2Cl2 and 50 mL of n-pentane. With rapid stirring, the solution was added to 400 mL of n-pentane and the total volume made up to 1 L with n-pentane. The solution was swirled, stoppered, and kept in the refrigerator overnight. The resulting insoluble asphaltene was separated by filtration through a 10-20 µm glass frit. The solid was washed with n-pentane until the washings were colorless, and then it was dried and weighed. The filtrate was concentrated to constant weight and constituted the maltene. The quantities of asphaltene and maltene isolated from each sample are given in Table 1. Hydroliquefaction of the Samples. The bitumen or asphaltene samples were subjected to hydroliquefaction treatment with HF, BF3, and H2. HF and BF3 are very corrosive substances and must be used in corrosion-resistant reactors made of Monel. The procedure is as follows. The sample (2-5.5 g) along with 0-30 mL of dry methylcyclohexane was placed in a 250-mL Parr autoclave under a dry nitrogen atmosphere. The autoclave was cooled to -78 °C (in a dry ice-acetone bath), and 50 mL (freshly distilled) of anhydrous HF was added. The vessel was closed tightly, warmed to room temperature, and 500 psi BF3 introduced, followed by 500 psi H2 (total pressure ≈ 1000 psi). Then the autoclave was heated to the desired reaction temperature and maintained at that temperature for 2-24 h. After the reaction, the autoclave was cooled to room

temperature and depressurized in an efficient hood. Then, the autoclave was opened and the contents quenched with 500 mL of ice water. The organic layer was separated and washed several times with a 10% sodium bicarbonate solution to remove traces of acid. The acid-free organic layer was then analyzed. Preparation of the Product for Analysis. The reaction product typically was a black solution, having been diluted with methylcyclohexane. Occasionally, small quantities of black, insoluble particles were visible, dispersed in the solution. This solution was poured into a separatory funnel to see if any water was present. If so, the water was separated and the organic phase was dried over anhydrous sodium sulfate and then filtered by gravity. In the event that any suspended particles were present, the solution was also filtered by gravity. Methylcyclohexane was removed by distillation, and final traces of solvent were eliminated by concentration on a rotary evaporator. The resulting weight of residual material was taken to be the total weight of the sample. This residual material was next extracted successively with n-pentane (4 × 25 mL). Each individual extract was progressively a lighter shade of reddish-yellow. Some black material appeared to be suspended in the pentane solution and was filtered by gravity. The black solid coated the filter paper, and a reddish-orange filtrate resulted. The pentane-insoluble material was next extracted with CH2Cl2 (3 × 25 mL). Each solvent portion was poured through the funnel containing the black residuum from the final pentane filtration. The combined CH2Cl2 extracts were concentrated on the rotary evaporator to give the CH2Cl2-soluble portion of the product. Any substance that was not soluble in CH2Cl2 was considered to consist of coke and/or coke precursors. The CH2Cl2 solubles were considered to be polar substances, unreacted asphaltene, or denatured asphaltene. The pentane-soluble portion was subjected to further chemical manipulations such as class, elemental, gas chromatographic, GC-MS and spectral analyses. Class Analysis of the Maltenes. Class analyses of the maltenes were performed following an established method.12 In this analysis, a 5 × 400 mm column was employed. Activity I alumina (4 g) was placed in the bottom of the column, and then 4 g of silica gel was added above the alumina. The column was flushed with n-pentane. About 200 g of maltene was

(11) Selucky, M. L.; Chu, Y.; Ruo, T. C. S.; Strausz, O. P. Fuel 1977, 56, 369-381.

(12) Sawatzky, H.; George, A. E.; Smiley, G. T.; Montgomery, D. S. Fuel 1976, 55, 16-20.


Energy & Fuels, Vol. 13, No. 3, 1999 561 Table 5. Superacid Treatmenta of Athabasca (Suncor Coker Feed) Asphaltene (Reaction 6) and Cold Lake Asphaltene (Reaction 7) product yields (%) maltene asphaltene

Athabasca 5.0 g 84.9 of products 200 of asphaltene 0.89 g 15.1 of products 35.6 of asphaltene

maltene asphaltene

Cold Lake 5.59 g 86.8 of products 223.6 of asphaltene 0.85 g 13.2 of products 34.0 of asphaltene class composition of maltene (%)

Figure 1. Preliminary experiment on Athabasca asphaltene. Table 4. Class Analyses of the Pentane-Soluble Portions of Superacid-Treated Asphaltenes and Bitumens

3.9 11.1 4.1 13.2 2.8

4.7 12.6 4.7 2.1 1.4

2.8 15.1 5.3 2.0 0.8

8.6 53.9 9.7 12.5 13.9

weighed and chromatographed. According to expectations, the saturates would be eluted with 25 mL of n-pentane, the monoaromatics with 25 mL of 5% toluene in pentane, the diaromatics with 25 mL of 15% toluene in pentane, the polyaromatics with 25 mL of 100% toluene and the polars with 25 mL of 15% CH3OH in toluene. Each eluant fraction was concentrated and the weight of the residual material determined. Gel Permeation Chromatography. A 2.5 × 100 cm glass column (Pharmacia SR 25/100) was packed with a tetrahydrofuran (THF) slurry of Bio-Beads SX-1, mesh 200-400 (BioRad Labs.). To make a slurry, 80 g of beads was allowed to swell overnight in 720 mL of THF, freshly distilled over LiAlH4. Prior to packing, the slurry was briefly (2 min) degassed in an ultrasonic bath. To replace THF by the solvent used for separation and to condition the Bio-Beads, 1800 mL of distilled and degassed CHCl3 was passed through the column. The maltene sample was dissolved in CHCl3 to give a 5% solution (including the solvent used for rinsing). The solution, after passing through a cotton wool plug to ensure the absence of insoluble particles, was applied to the column and eluted with CHCl3. Sample collection at a flow rate of 1 mL/min was begun after the void volume of 100 mL of pure solvent was eluted.

Results and Discussion The first step in the commercial upgrading of bitumen into usable products is mainly cokingsthe thermal treatment of bitumen up to ca. 470 °Csand to a lesser extent catalytic hydrocracking. As an alternative, the objective of the present study was to test the efficiency of ionic depolymerization and hydroliquefaction of oil sand bitumen and its high molecular weight asphaltene fraction using a gaseous superacid catalyst. The conditions used, HF, BF3, H2 (50 mL, 500 psi, 500 psi), a hydrogen-donor solvent (methylcyclohexane), and moderate temperatures (170-190 °C), were similar to those employed previously by Olah and co-workers5-7 for the depolymerization of coal. The superacid HF:BF3 is stable in the presence of excess HF but otherwise decomposes

66.3 13.2 4.4 2.4 13.6

elemental analysis Athabasca

reaction monopolyno. saturates aromatics diaromatics aromatics polars 80 7.3 76.1 70.1 81.2

Cold Lake

39.2 41.4 1.0 0.0 18.3

saturates monoaromatics diaromatics polyaromatics polars

%

1 2 3 4 5

Athabasca

C H N S O MW H/Catomic

Cold Lake

maltene

asphaltene

maltene

asphaltene

87.35 10.81 0.0 0.74 1.14 332 1.47

76.46 7.51 1.81 7.33 6.88 n.d. 1.17

87.92 10.30 0.06 0.23 1.19 329 1.53

81.66 7.20 0.87 6.50b 3.77 1.05

a

200 °C (Athabasca), 210 °C (Cold Lake), 500 psi BF3, 500 psi H2, 50 mL of HF, 30 mL of MCH, 2.5 g of asphaltene, 24 h. b By difference. Table 6. Superacid Treatmenta of Athabasca (Suncor Coker Feed) Bitumen (Reaction 8) and Cold Lake Bitumen (Reaction 9) product yields (%) maltene asphaltene

Athabasca 5.5 g 89.4 of products 0.65 g 10.6 of products

maltene asphaltene

1.87 g 0.33 g

208 of bitumen 24 of bitumen

Cold Lake 85 of products 15 of products

53.4 of bitumen 9.4 of bitumen

class composition of maltene (%) saturates monoaromatics diaromatics polyaromatics polars

Athabasca

Cold Lake

54.4 22.7 2.7 0.6 19.6

34.4 14.2 11.2 17.7 22.5

elemental analysis Athabasca C H N S O MW H/Catomic

Cold Lake

maltene

asphaltene

maltene

asphaltene

87.95 10.88 0.02 0.76 0.76 324 1.47

77.44 7.13 1.49 7.56 6.38 n.d. 1.10

84.25 11.31 0.22 3.49 0.79 447 1.6

80.18 7.83 1.05 7.84 2.27 2345 1.16

a Athabasca: 200 °C, 500 psi BF , 500 psi H , 50 mL of HF, 30 3 2 mL of MCH, 2.64 g of bitumen, 24 h. Cold Lake: 285 °C, 500 psi BF3, 500 psi H2, 50 mL of HF, 5 mL of MCH, 3.5 g of bitumen, 24 h.

to HF and BF3 gases, which can be distilled off and readily recovered for recycling in any potential largescale operation. The temperatures in most of the experi-


Strausz et al.

Table 7. Superacid Treatmenta of Athabasca (Suncor Coker Feed) Bitumenb and Asphaltene. Effect of the Quantity of Added MCH reaction no. bitumen (g) asphaltene (g) MCH (mL)

10

11

2.69

2.88

5

12

13

14

15

3.08 2.50 5

30

2.50 30

asphaltene

15

2.50 15

maltene

reaction no.

MCH (mL)

% recovered

% of starting material

% recovered

% of starting material

total recovery % of starting material

10 bitumen 14 bitumen 11 bitumen 12 asphaltene 15 asphaltene 13 asphaltene

5 15 30 5 15 30

18.5 13.1 5.8 57.1 54.6 19.1

8.1 9.4 11.7 44.6 49.2 38.4

81.5 86.9 94.2 42.9 45.3 80.9

35.7 62.3 191.6 33.5 40.8 161.6

43.8 71.7 203.3 78.1 90.0 200.0

a

500 psi H2, 500 psi BF3, 50 mL of HF, 1 h, 190 °C, reactor volume 250 mL. b The asphaltene content of the bitumen was 15.5%.

Figure 2. Product yields as a function of added methylcyclohexane (MCH). For experimental details see Table 7.

ments were kept low (e190 °C) in comparison to those needed to rupture C-H bonds (>300 °C) or those employed in coking. Our objective was to explore these ionic liquefaction reactions on samples of bitumen and for model studies on asphaltene from the vast Athabasca and Cold Lake deposits. We wished to establish the conditions required for optimum depolymerization, to measure the yields of synthetic crude-like products, and to determine their chemical nature. The model studies on asphaltene were intended to further the elucidation of the mechanism of depolymerization of large geopolymers during ionic hydroliquefaction. The elemental compositions of bitumen and asphaltene isolated from our Athabasca and Cold Lake samples are given in Table 1. In most experiments, methylcyclohexane was added as a hydrogen-donor solvent in order to facilitate the transfer of hydrogen atoms to the carbocations generated in the superacid-catalyzed depolymerization. In one experiment where bitumen was used as the substrate, no methylcyclohexane was added in order to assess the hydrogen-donating properties of the maltene components themselves. Preliminary, qualitative experiments on Athabasca asphaltene indicated complete conversion of asphaltene to liquid maltene and some volatile products upon

superacid treatment, Figure 1. The MW decreased to 466 g‚mol-1, a value between that of the feed bitumen and the maltene (540 and 435 g‚mol-1), with a good distribution (as determined by GPC fractionation and VPO measurements). Further experiments on asphaltene and bitumen revealed two interesting and related characteristics of the hydroliquefaction reactions, namely, (a) an increase in the weight of recovered products relative to the weight of the feedstock used, proving that methylcyclohexane (MCH), contrary to expectation, is not inert but undergoes decomposition to yield higher-boiling molecules, and (b) a conspicuous scatter in the reproducibility of the quantitative data. In blank experiments, neat MCH, under conditions similar to those employed in the liquefaction reactions, did not yield any distillation residue. Therefore, it must be assumed that some intermediate(s) in the hydroliquefaction reaction catalyze its decomposition Quantitative results on Athabasca asphaltene and bitumen obtained at 170-188 °C are listed in Table 2. In every case where MCH (30 mL) was added to the bitumen or asphaltene (2-5.5 g), the recovered product yield exceeded 100%. For example, Athabasca bitumen, neat (experiment 2) and with 30 mL of added MCH (experiment 3) afforded product yields of 50% and 358%, respectively. The excess material undoubtedly come from MCH. In the absence of MCH, the weight loss is due to the formation of volatiles and perhaps gaseous products. This feature of the reaction points out the efficiency of superacids in cleaving the carbon-carbon bond. For comparison, in thermal cracking at 420 °C, the volatile + gaseous product yields are only around 20-25% and at 190 °C they would be of the order 1-2%. (In hydrocracking at 425 °C, the volatile + gaseous product yield is only 10%). Another interesting facet of the hydroliquefaction reaction is evident upon considering the results of experiments 4 and 5. Here, under nearly identical conditions, Cold Lake asphaltene is seen to leave behind more unconverted asphaltene in a 24-h than in a 2.5-h reaction. This apparent anomaly can be attributed to the occurrence of secondary polymerization reactions upon prolonged treatment. Even the 2.5-h reaction may be overdone, and the product distribution may actually improve upon further shortening of the reaction time.


Energy & Fuels, Vol. 13, No. 3, 1999 563 Table 8. Class Compositionsa of Maltenes maltenes from reaction no.

a

fraction

Athabasca bitumen

10 (bitumen)

11 (bitumen)

12 (asphaltene)

13 (asphaltene)

saturates monoaromatics diaromatics polyaromatics polars

24.9 10.8 8.5 20.0 35.8

27.0 9.6 4.4 8.7 50.2

58.9 18.2 0.5 1.0 21.4

20.8 4.6 1.5 6.0 67.0

42.7 22.2 1.4 1.1 32.6

As wt % of maltene.

Figure 5. Gas chromatogram of the saturate fraction of Cold Lake superacid-treated asphaltene.

Figure 3. Infrared spectrum of the saturate fraction of Cold Lake superacid-treated asphaltene, stored under nitrogen.

Figure 6. GC-MS total ion current chromatogram of the saturate fraction of Cold Lake superacid-treated asphaltene.

Figure 4. Infrared spectrum of the saturate fraction of Cold Lake superacid-treated asphaltene, exposed to air.

As seen, in the shorter reaction over 90% of the asphaltene is converted to liquid and volatile products and the residual asphaltene, while being soluble in CH2Cl2, is insoluble in benzenesin sharp contrast to the native asphaltene. This phenomenon is again a manifestation of the altered nature of the pentane-insoluble/ CH2Cl2-soluble fraction, “asphaltene”, before and after hydroliquefaction. Tables 3 and 4 list some elemental and class compositional data. They show a significantly lower MW for the hydrocracked than for the native maltenes, along

with a lower sulfur content. In general, the sulfur content is reduced in the presence of MCH, but the oxygen contents of all the products show an increase. We will return to this observation later. The class composition of the produced maltene is dominated by the saturate fraction except from reaction 2, which was carried out without added MCH. Here we find the closest resemblance between the hydrocracked and native (Table 1) maltene compositions except that the amount of the saturate fraction in the hydrocracked maltene is reduced. The class compositional data also point to a deterioration in the maltene composition over extended reactions, the shorter reaction 5 yielding more saturates and less aromatics compared to the longer reaction 4. Comparison of Athabasca (Suncor coker feed, reaction 6) and Cold Lake (reaction 7) asphaltene, Table 5, reveals similar behavior, except that the Cold Lake asphaltene yields a higher percentage of saturates and a lower percentage of monoaromatics. This difference, however, can be an artifact of the analytical procedure


Strausz et al.

Figure 10. 13C NMR spectrum of the saturate fraction of Cold Lake superacid-treated asphaltene.

Figure 7. Mass spectra of scans 246, 261, and 280 (C16H24) in Figure 6.

Figure 8. Mass spectra of scans 286 (C16H24), 294, and 297 (C17H26) in Figure 6.

Figure 9. 1H NMR spectrum (80 MHz) of the saturate fraction of Cold Lake superacid-treated asphaltene.

because it is difficult to prevent the monoaromatics from eluting into the saturates and the aggregate amounts, saturates + monoaromatics, are nearly the same, ca. 80%, in both reactions. The other slight differences are attributable to experimental errors and to minor differences in the composition of the feed asphaltenes.

Some idea about the effect of temperature on the hydroliquefaction reaction can be obtained by contrasting the product yields and distributions from reactions 4 and 7 involving Cold Lake asphaltene. Under otherwise identical conditions, the increase in temperature from 184 to 210 °C appears to have caused a 12% increase in the maltene yield and a decrease of 15% in the residual asphaltene. Thus, increasing the temperaturesat least in this limited domainshas a beneficial effect on the hydroliquefaction of asphaltene. In contrast, with bitumen in the presence of 30 mL of MCH, reactions 3 and 8, Table 6, a rise in temperature from 170 to 200 °C causes a decrease in the maltene yield (or rather the percentage amount of MCH polymerization) and an increase in asphaltene formation. In the absence or with a low concentration (5 mL) of MCH, a rise in temperature from 170 to 285 °C, reactions 3 and 9, brought about an increase in the maltene and a modest decrease in the asphaltene yield from the hydrocracking of Athabasca and Cold Lake bitumens. Thus, the overall effect of increasing temperature depends on the feedstock and whether MCH is present. As is amply evident from the data, MCH, when present, may play a decisive role in the hydroliquefaction reaction and may be the source of the dominant products. For this reason, a series of six experiments were carried out using a short reaction time of 1 h to explore the effect of added MCH on the liquefaction reaction involving bitumen and asphaltene. The results, Table 7, show that when 5, 15, or 30 mL of MCH is added to 2.5-3.1 g of bitumen or asphaltene undergoing hydroliquefaction, the percentage amount of hydrocracked asphaltene and maltene (in terms of the initial amount of feedstocks) increases except in one instance (reaction 13). The increase in the maltene yield becomes particularly manifest around 15 mL of added MCH, and at 30 mL, the maltene produced well exceeds the original weight of the feedstock. A similar but decidedly more modest increase is apparent in the yield of hydrocracked asphaltene. In these experiments coke was absent and the combined yields of products, expressed as “percentage of starting material”, increase rapidly with increasing amounts of added MCH, Figure 2. In all cases, the relative amount of maltene in the hydrocracked bitumen is enhanced by increasing amounts of added MCH and, moreover, its class composition, Table 8, tends to improve inasmuch as the most valuable components, the saturates and monoaromatics, become more dominant. In the presence of 5 mL of MCH, 56% of the bitumen was volatilized and with asphaltene 22% was volatilized and 30% converted to


Energy & Fuels, Vol. 13, No. 3, 1999 565

Figure 11. Attached proton test of the saturate fraction of Cold Lake superacid-treated asphaltene.

Figure 12. GC-MS total ion current chromatogram of the saturate fraction of Cold Lake superacid-treated asphaltene after ionic hydrogenation.

maltene. In both cases, the curves show an initial slow rise with rapidly accelerating increases as the amount of MCH is increased. This trend, at least in part, is due to the distribution of solvent between the vapor and liquid phases. With 5 mL of MCH, most of the MCH is in the vapor phase, and therefore, this amount of solvent has practically no effect. As the quantity of added MCH is increased, more and more remains in the liquid phase and its effect intensifies. Before progressing further, we point out here again that in the early stages of the investigations considerable difficulties were experienced with the lack of

Figure 13. Mass spectra of scans 237 (C16H30) and 249 (C17H32) in Figure 12.

reproducibility. This initially was thought to be due to poor contact between the semisolid bitumen and the HF‚ BF3 catalyst and the ill-defined nature of the solventbitumen/asphaltene interaction. Moreover, it was noted that the maltene productsespecially from the asphaltene reactions with added MCHswas highly reactive and upon exposure to air it reacted with oxygen, transforming the colorless mobile liquid oil to a translucent gum. This observation suggested that conjugated alkenes may have been produced, providing a possible explanation for the lack of reproducibility in the early experiments. In subsequent developments it was found that when the maltene product from the hydroliquefaction of Cold


Strausz et al.

Scheme 1. Superacid-Catalyzed Solvent Polymerization

Figure 16. IR spectra of the cyclohenxane extracts of Illinois No. 6 coal: (A) before and (B) after treatment with HF‚BF3‚ H2.

Lake asphaltene was analyzed, it contained no oxygen but after standing in air for 2 days the oxygen content was 5.8%. Comparison of the IR spectra of air-protected and airexposed samples is revealing. As seen from Figures 3 and 4, the IR spectrum of the air-protected sample

Figure 17. Gas chromatogram of the saturate fraction of Cold Lake superacid-treated bitumen.

Figure 14. Chemical ionization GC-MS of the saturate fraction of Cold Lake superacid-treated asphaltene.

Figure 18. Gas chromatogram of the saturate fraction of Athabasca bitumen. B and T refer to bicyclic and tricyclic terpanes and H to hopanes. The subscripts refer to carbon numbers.

Figure 15. Mass spectra of scans 1082 (a), 1102 (b), and 1156 (c) of Figure 14.

shows only hydrocarbon bands with weak quaternary CdC double bonds whereas the spectrum of the airexposed sample shows intense C-O, CdO, and OH absorptions, indicating the presence of COOH (and possibly >CdO and C-OH functionalities). In agreement with the high reactivity and IR spectra of the oil it was also found that the saturate fraction of the oil reacted rapidly and exothermically with dilute bromine in CH2Cl2 solution, forming an insoluble and nonvolatile product. Oxidation with metachloroperbenzoic acid converted the saturate fraction to some polar material in >99% yield, suggesting the absence of any paraffinic hydrocarbon in the saturate fraction. The gas chromatogram of this saturate fraction, Figure 5, features two broad humps superimposed by a large number of peaks corresponding to individual compounds. The GC-MS chromatogram of the saturates



experimental conditions of the superacid-catalyzed hydrogenation employed, the MCH solvent undergoes ionic rearrangement, degradation, and oligomerization. Speculative structures and mechanisms for the formation of the bicyclic conjugated quaternary triolefins are shown in Scheme 1. Most of the peaks in the GC-MS of the saturate fraction in Figure 6 correspond to molecules with structures analogous to the C16H24 and C17H26 compounds. Thus, for example, the full MS traces for three of the late-eluting peaks (labeled by asterisks) appearing in the chemical ionization GC-MS of the saturate fraction, Figure 14, are reproduced in Figure 15. All three peaks correspond to isomeric C24H36 compounds which would fit the tricyclic conjugated tetraalkene structure Figure 19. GC-MS total ion current mass chromatogram of the saturate fraction of Athabasca superacid-treated bitumen.

is reproduced in Figure 6, and the full mass spectra of some of the selected peaks indicated in the figure are given in Figures 7 and 8. All the mass spectra show very low parent ion intensities; nevertheless, four of the six selected peaks can be assigned to isomers of the C16H24 and the remaining two to the C17H26 formulas. These structures possess five CdC double-bond equivalents. Next, to determine the number of double bonds in the above structures, the saturate fraction was subjected to catalytic hydrogenation (PtO2, pentane, H2 (1 atm), 25 °C, 16 h), which led to no chemical changes. It is known that tetrasubstituted double bonds are not affected under these conditions. The 1H NMR spectrum of the saturates, shown in Figure 9, consists almost entirely of aliphatic hydrogens with a small (1-2%) contribution from alkene hydrogens. The 13C NMR spectrum, Figure 10, shows significant signals in the 125-135 ppm range which are due to alkene carbons, and from the attached proton test, Figure 11, it appears that the carbons resonating in the 125-135 ppm range have no attached hydrogens. This suggests that in agreement with the 1H NMR spectrum, Figure 9, the resonances correspond to tetrasubstituted alkenes. The 19F NMR spectrum of the saturate fraction was taken, but no signal due to 19F could be detected. Thus, fluorine incorporation, if any, is minimal. In an attempt to obtain further structural information, the saturate fraction was subjected to ionic hydrogenation ((Et)3SiH, CF3CO2H, BF3‚Et2O, 0 °C, l h) which, after alumina chromatography, gave ∼35% yield of hydrogenated product. The GC-MS trace of the lead hump of this material is shown in Figure 12, and the full mass spectra of the hydrogenation products derivable from the C16H24 and C17H26 compounds are shown in Figure 13. As shown in the latter spectra, ionic hydrogenation of the above compounds converted them to C16H30 and C17H32 compounds, both of which contain two saturated rings. Therefore, the original C16H24 and C17H26 compounds each contained three carbon-carbon double bonds which must have been conjugated in order to account for the high reactivity of these molecules, as manifested by their spontaneous rapid reaction with atmospheric oxygen. Moreover, from the NMR spectra it is clear that the alkene carbons are quaternary. Therefore, we are forced to conclude that under the

Thus, the C24 compounds appear to be the homologues of the C16 and C17 compounds having one more MCH ring, with or without an extra carbon atom. The HF‚BF3-initiated oligomerization of MCH was an unexpected phenomenon. In studies on the HF‚BF3catalyzed hydroliquefaction of coal under conditions similar to those employed here using cyclohexane as the solvent and hydrogen transfer catalyst, cyclohexane was not expected to undergo chemical reaction. But comparing Olah et al.’s IR spectra on the cyclohexane extracts of Illinois No. 6 coal before and after treatment with HF‚BF3/H2, Figure 16, it is clear that the spectrum of the treated sample possesses the same OH, >CdO, and C-O absorptions as found here in the asphaltene and bitumen samples after treatment, which were not present before treatment. Consequently, cyclohexane undergoes similar oligomerization reactions forming conjugated tri- and higher olefins, as found in the present work for MCH. Therefore, we are forced to conclude that the HF‚BF3/H2-catalyzed oligomerization of cycloalkanes in the presence of asphaltene or bitumen is a general reaction producing structures in which cycloalkenes are bridged by a carbon atom in such a manner that the olefinic bonds are conjugated as shown above or as follows

and, after ionic hydrogenation, yielding the saturated counterparts

These novel processes suggest one of the many possible


Strausz et al.

Figure 20. Mass spectra of peaks a-f, Figure 19.

ways in which a superacid may attack the bitumen molecules, namely, by ionizing their reactive naphthenic ring systems. They are interesting in themselves as well because of their potential for commercial application. Some preliminary indications of the nature of the changes in the chemical composition of the saturate fraction taking place upon hydroliquefaction have also been obtained. As seen from comparing the gas chromatograms of the saturate fraction from the hydroliquefaction of Cold Lake bitumen, Figure 17, with a typical gas chromatogram of Alberta oil sand saturates, Figure 18, the differences are quite profound. GC-MS analysis of the former, Figures 19 and 20, reveals the presence of series of 2-alkylbenzyl, 2-methylalkanes. These compounds are evidently formed from the n-alkylbenzenes present in the bitumen by the well-known superacidcatalyzed isomerization of n-alkanes to isobutanes.3 Summary and Conclusions In the course of this investigation it has been clearly established that the gaseous superacid HF‚BF3 is a highly effective catalyst for the hydrocracking of oil sand bitumens. The reaction conditions required are exceptionally mild, 170-190 °C or lower temperature, 500 psi H2 pressure, and 1 h or shorter reaction time. The reactions involved proceed with ionic mechanisms as opposed to the free-radical mechanisms in thermal hydrocracking, resulting in deep-rooted alterations of the chemical composition of the bitumen, evidenced by the high (56%) conversion of the bitumen to volatiles, the drastic changes in the gas chromatograms of the

various compound classes of the bitumen, and the appearance of many prominent branched alkylbenzene peaks, etc. In asphaltene, parallel to depolymerization some secondary polymerization leading to the formation of asphaltene-like materials takes place upon prolonged reaction times. Therefore, establishing the optimal reaction time is essential to obtain the most favorable conversions. Methylcyclohexane exhibits high reactivity with respect to the HF‚BF3 superacid under the conditions of the bitumen or asphaltene hydrocracking processs typifying the general reactivity of cycloalkanessresulting in the oligomerization of methylcyclohexane to produce conjugated alkenes. The oligomerization of MCH also typifies the general reaction mechanism by which the saturate fraction and the naphthenic moieties of the aromatic, resin, and asphaltene fractions of the bitumen are eliminated in HF‚BF3-catalyzed hydrocracking. The saturates comprise one-to-five-ring naphthenic hydrocarbons. The oligomerization of MCH discovered here points to the possible complications arising in the application of hydrocarbons as “stabilizers” or “hydrogen donors” in the HF‚BF3-catalyzed hydrocracking of coal or hydrocracking of bitumen. The formation of oligomers consumes the hydrocarbons, and moreover, the presence of such highly reactive conjugated polyenic oligomers may lead to secondary reactions in the course of product analyses. The saturate fraction of the hydrocracked maltene comprises series of 2-alkylbenzyl, 2-methylalkanes formed


from the isomerization of n-alkylbenzenes in the bitumen in reactions reminiscent of the isomerization/ cracking reactions of n-alkanes to isobutane in the presence of superacids.3 Such reactions must be important and should be considered in the superacid- or strong protic acid-catalyzed transalkylations of crude oils or crude oil fractions, along with the MCH-type oligomerization reactions. Fluorine incorporation into the bitumen does not take place, and no evidence could be found for catalyst consumption. It should be pointed out that the ash content of bitumens, in general, is much lower than that of coal (e.g., e0.5% for Athabasca bitumen). BF3 is a gas and HF a low-boiling liquid (bp 20 °C); both are highly water soluble and, therefore, should be


readily separable from the hydrocracked bitumen and recycled for further use. Thus, taking the above results in conjunction with the ready recoverability of the catalyst, it can be concluded that HF‚BF3 superacid catalysis has potential for the development of new technologies for the bitumen upgrading industry.

Acknowledgment. We thank the Alberta Oil Sands Technology and Research Authority for financial assistance and Dr. E. M. Lown for assistance in the preparation of the manuscript. EF980098R

Upgrading of Alberta's Heavy Oils by Superacid-Catalyzed

Recommend Documents