Extraction of oil shales under supercritical conditions

Extraction of oil shales under supercritical conditions

Extraction conditions Riri Kramer of oil shales and Moshe under supercritical Levy Department of Materials Research, Weizmann Institute (Receive...

767KB Sizes 11 Downloads 145 Views

Extraction conditions Riri Kramer

of oil shales

and Moshe

under

supercritical

Levy

Department of Materials Research, Weizmann Institute (Received 79 April 7988; revised 17 August 7988)

of Science,

Rehovot,

Israel

Supercritical extraction of oil shales was carried out in a number of solvents and solvent mixtures at 400°C. Shales from Mishor Rotem in Israel and from Kentucky and Green River in the USA were studied and

compared. Extraction was carried out both batchwise, in autoclaves, and in a continuous flow system. The best solvents seem to be isopropanol/water and water/CO mixtures. After extraction, an additional kerogen fraction can be recovered by heating the shales to 600°C in an inert atmosphere. A simple and efftcient method was developed for the determination of volatile and non-volatile fractions of kerogen by thermogravimetric analysis. The average MWs of the extracted fractions were determined by g.p.c.. (Keywords: oil shale; supercritical; thermogravimetric analysis)

The organic matter present in oil shales is a complex mixture of high MW organic molecules, finely distributed in an inorganic matrix. To convert the organic matter into useful liquids or gases it is common practice to use retorting processes under an inert atmosphere. Retorting is usually carried out at 500_6OO”C, a temperature at which considerable degradation takes place, resulting in a volatile fraction which is recovered and a char fraction which remains with the inorganic residue. Supercritical extraction can be carried out under milder conditions than retorting, resulting in less degradation and charring. This technique has been widely used for the study of bituminous coal. It was reported’ that CO/H,0 at 400°C and pH > 12.6 converted bituminous coal into a pyridine-soluble product; similar results were also obtained by reaction with alcoholic alkali’. It was assumed that hydrolytic reactions take place, as well as some bond degradation by pyrolysis. Supercritical extraction has also been widely applied to oil shales to obtain soluble extracts3-h. In a previous publication7, a study of supercritical extraction of Israeli shales in toluene was presented. This has now been extended to a number of other solvents and applied to oil shales from Israel and the USA. The aim of this work was to compare different solvent systems, some of them chemically active and some not, under supercritical conditions, to evaluate the feasibility of the method as a potential practical process, and as a way of obtaining more structural information on oil shales from different sources. EXPERIMENTAL The samples of oil shales used in this work were: Israeli shales from Mishor Rotem; eastern American shales from Kentucky; and western American shales from the Green River oil shale deposit. The analysis of the oil shales is presented in Table 1. The calculated amount of kerogen, given in the last line, was derived from the organic carbon 0016-2361,‘89/060702~8$3.00 I(’ 1989 Butterworth & Co. (Publishers) Ltd.

702

FUEL, 1989, Vol 68, June

analysis according to the method of McKay et u!.~, where Kerogen

in shales (%) =

wt% organic carbon wt% carbon

in shales

x 100

in kerogen

The wt% of organic carbon was determined as the difference between total carbon and mineral carbon. Total carbon was determined by heating the shales in air to 950°C followed by coulometric titration of the released carbon dioxide. Mineral carbon was determined by perchloric acid digestion of the shales followed by titration. The wt% of carbon in the kerogen was determined from the elemental analysis of the liquid organic material recovered from the shale. Mineral-free kerogens, for i.r. analysis, were prepared by dissolving the mineral matrix, while minimizing changes in the chemical composition of the organic matter. The method used, described previously’, is based on the decomposition of the carbonates with HCl, followed by dissolution of the silicates in HF at 70°C. The treatment with LiAlH, applied previously was not carried out, to avoid changes in the structure of the kerogen. Mineral-free chars were prepared in the same way using the residues from the extractions. Table 1 Analysis of the oil shales Green

Carbon Hydrogen Nitrogen Sulphur Moisture Carbonate (as CO,) Organic matter” ‘See experimental section

Rotem (%I

Kentucky (%)

River (“/J

9.7 1.1 0.2 1.8 1.5 21.0 14.8

11.8

39.1 5.4 1.5 1.9 I .6 5.6 45.1

1.4 0.5 4.4 1.5 0.7 15.6

Extraction

of oil shales under supercritical

For the supercritical batch extraction experiments, a well-mixed sample of raw shale was ground to < 177 pm and charged into a sintered glass crucible that was suspended from the cover of a 50ml autoclave. Various solvents were used under supercritical conditions: toluene, tetralin, isopropanol, and isopropanol/water at pH 14, water at pH 12.3, and water/CO at pH 12.3. The solvent/shale ratio was 10 in all of the experiments. The vapour density was 0.5 g ml-’ for toluene and tetralin, 0.2 g ml-’ for isopropanol and isopropanol/water, and 0.12 gml-’ for the water and water/CO systems. The autoclave was sealed and checked for leaks with argon at 50atm pressure, then the pressure was adjusted to 1 atm argon for the experiments with toluene, isopropanol, isopropanol/water and water, 30 atm argon for tetralin and 20atm CO for water/CO. The reaction temperature was 400°C in all cases and was maintained for 60min; the time taken to reach the working temperature was 15 min. The experiment was terminated by immersing the autoclave in cold water, which cooled the mixture to room tempeature in less than 10min. In cases where an increase in pressure could be detected, the gaseous products were collected, weighed and analysed with a gas chromatograph fitted with a 6m long x 3.2mm i.d. Porapak Q column. The autoclave was opened and the spent shale in the sintered glass crucible was extracted (Soxhlet) with toluene and dried in an oven under an argon atmosphere at 60°C. The amount of organic matter (OM) remaining in the residue was determined thermogravimetrically. The yield obtained from supercritical extraction was calculated as follows: Kerogen extracted (%) = (wt shales)(%OM in shales)- (wt residue)(%OM in residue) (wt shales)(%OM in shales) Part of the solution obtained was used directly for the determination of the molecular weights of the organic materials recovered. The measurements were carried out immediately after extraction, to avoid possible secondary reactions in the extract. The average MWs of the organic materials recovered were determined using a Tracer 985 Model h.p.1.c. system equipped with three g.p.c. columns in series (Li Chrogel PS 400; PS 20 and PS 4). The solvent used during g.p.c. was tetrahydrofuran and the U.V. detector was operated at 300nm. A standard graph of MWs versus retention times, which was obtained using polystyrene standards (MWs 600 and 2100) and anthracene (MW = 178) was used in the molecular weight determinations. The additional Soxhlet extracts were also analysed by g.p.c. and usually gave molecular weights similar to or somewhat larger than those of the direct extracts. Supercritical extraction was also carried out in a continuous flow system. The reactor used, shown schematically in Figure 1, was made from 12.5mm stainless steel tube, and the internal pressure was maintained above the critical value of the solvent using a back pressure regulator (82atm for experiments with toluene and 30atm for those with tetralin). The thermocouple placed inside the reactor was used to control the surrounding heating element; the reaction temperature was kept constant at 400°C. The shale sample (1 g) was supported in the reactor on a stainless

conditions:

R. Kramer and M. Levy

steel grid. The solvent was pumped into the reactor with a metering pump (Eldex, Model A-30s) at a constant flow of 3mlmin-‘. Samples of the extract were collected at intervals of 8 min and analysed by g.p.c.. The total absorption of the extracts at 300nm gave an estimate of the total amount of kerogen extracted. This was not a quantitative measure because different compositions have different absorption coefficients; however, measurements at higher and at lower wavelengths, and measurements with a refractive index detector, all gave similar results, and it was therefore decided that these values would suffice until further work could be done with total scanning of every sample. The residue remaining after the extraction process was terminated, was extracted with toluene (Soxhlet) and then analysed by TGA to determine the amount of unextracted kerogen. The reproducibility of the method was checked and found to be accurate within 10%. The shale samples were from well mixed, ground and sieved stock and uniformity could therefore be assumed even in 1 g samples; to obtain reproducible results it was important to keep the flow of solvent at a very constant rate and the temperature and pressure constant. The amount of organic matter initially present in the shales was determined thermogravimetrically. A weighed amount of shale was oxidized completely in a Mettler TA 3000 thermal analysis system equipped with a TG 50 thermobalance by heating from 25 to 980°C at 50”Cmin’ under a flowing oxygen atmosphere. The derivative thermogravimetric (DTG) oxidation profiles and quantitative weight loss data produced by the Mettler TClOTA processor measured the content of organic material in the shale. To obtain a more detailed analysis of the organic material present in the initial shales, and to differentiate between volatile fractions and char, a sample of shale was heated under a flowing nitrogen atmosphere from

Thermocouple Metermg

I

pump

Argon Back pressure regulator

Extract solution

Figure 1 Schematic supercritical extraction

diagram of the of oil shales

FUEL,

continuous

1989,

Vol

flow

system

68, June

for

703

Extraction

of oil shales

under

supercritical

conditions:

R. Kramer

a

25 to 600°C at 50”Cmin-‘, and then reheated under flowing oxygen from 25 to 900°C at the same rate. The measurements of heat of combustion (AH,,,) of the shales and of the non-volatile char fractions, were performed with the Mettler TA 3000 system equipped with a differential scanning calorimetry (d.s.c.) attachment and the TClOTA processor via a Mettler 03 data interface. The temperature was measured with a Pt 100 temperature sensor, in contact with the crucible holding the sample, which also controlled the rate of heating.

-0.006

RESULTS AND DISCUSSION

-0.004

TGA

-0.002

r

/j\

\!

‘\

i*

2 .z $ .g al z : % al z LL

and d.s.c. of initial

-o.ooa

-0.006

-0.004

-0.002

-0.06

Table 2

TGA of initial shales

Shale

Volatiles in shale WI

Char in shale WJ

Total kerogen in shale (%I

Volatiles in kerogen WI

Char in kerogen WI

Rotem Green River Kentucky

11.0 40.4 8.5

3.2 6.4 6.6

14.2 46.8 15.1

71.5 86.3 56.3

22.5 13.7 43.1

T(OC)

Figure 2 Differential thermogravimetric analysis of oil shales in: -, oxygen; ---, nitrogen; -. ~. , oxygen after preheating in nitrogen; Rotem; b, Kentucky; c, Green River

704

FUEL, 1989, Vol 68, June

shales

TGA of shales in an oxygen atmosphere leads to complete combustion of the kerogen, so the weight loss is an accurate measure of the amount of kerogen in the shales. TGA in a nitrogen atmosphere gives only the fraction that degrades thermally and yields volatile products. The residue contains char products that cannot be volatilized, but can be burned in oxygen. The results of the TGA experiments with the three shales studied, in atmospheres of oxygen, nitrogen and nitrogen followed by oxygen, can be seen in Figure2. A summary of the respective weight losses is shown in Table2. The Rotem shale showed two distinct peaks on combustion in oxygen. The first (at 360°C) is probably due mainly to the aliphatic fraction of the kerogen, and the second (at SOO’C)to the aromatic fraction. When the heating is done in a nitrogen atmosphere, the peak starts at a higher temperature than in oxygen, indicating that combustion in oxygen starts in the condensed phase before volatilization. Combustion of the char in oxygen takes place over about the same temperature range as devolatilization in nitrogen. The sum of the areas of the latter two peaks equals the peak area in oxygen for the initial sample; this confirms the present contention that the area is indeed a measure of the total kerogen content of the shale. The Kentucky shale showed similar behaviour. The first peak in oxygen is small and merges into the second, which is the main aromatic peak. The peaks for the volatile and the char fractions are almost equal in area. The Green River shale showed a very sharp peak on oxidation, and some solid particles may have been lost during the intense exothermic reaction because of the high rate of heating. When TGA in oxygen was carried out at a lower heating rate, the first peak, at 300°C increased and the second, at 450°C decreased. The kerogen content calculated from the direct oxidation thermogram was 56%, which is much higher than the value found by analysis (45.1%, see Table I). This problem however, did not exist when the sample was first pyrolysed in nitrogen and the char subsequently burned in oxygen; the total weight loss was then 46.8%, much closer to the real value.

\ i

. j

:

and M. Levy

a,

Extraction

of oil shales under supercritical

SE;” --

L

----

3000

2000

1500

1000

500

Wavenumbers Figure 3 Kentucky;

FT-ix. spectra c, Green River

of demineralized

kerogens;

R. Kramer and M. Levy

aromaticity. This was verified by FT-i.r. measurements of the mineral-free kerogens and of the mineral-free chars from the three sources. The intensities of the different peaks in the kerogen spectra (Figure3) relative to the intensity of the CH, stretching frequency at 2925 cm ’ are given in Table3. It is evident that the Kentucky kerogen has higher contents of aromatic and phenolic groups than the other two. The spectra seem to show higher aromaticity in the Green River shale; however, its higher volatility is probably determined by other factors, such as low phenolic content. The FT-i.r. spectra of the respective chars (Figure 4), show complete disappearance of the aliphatic bands, while the aromatic and phenolic bands remain, although as expected they are much broader than in the original spectra. It was concluded therefore that the TGA method described above can be used as a simple and accurate way of determining the total kerogen in shales as well as of distinguishing between the volatile and the char fractions. The same method was used for determining the residual kerogen in the spent shale after supercritical extraction. The two oxidation reactions, of the initial kerogen and of the char, are also evident in the d.s.c. thermograms (Figure 5). The respective heats of combustion are given in Table#. The temperatures of the peaks are generally lower than those obtained by TGA, probably because of the lower heating rates. The shapes are not identical, because in d.s.c. the heat of combustion is a function of the hydrogen content of the molecules. It was also noticed that the heats of combustion derived from the d.s.c. curves, for all three shales, were lower than those measured by ordinary calorimetric methods. This may be due to the fact that part of the combustion occurs in the gas phase after vaporization, so that the detector cannot register all the heat evolved. Nevertheless, the values for the percentage of char in kerogen calculated from the d.s.c. results are very close to those calculated from the TGA measurements (Table 4). The d.s.c. method was tried for analysis of the residues remaining after supercritical extractions, but because of the limited accuracy of the instrument reproducible results could not be obtained, and the attempt was not pursued further.

f

i

4000

conditions:

a, Rotem;

b,

The percentages of char in the shales are shown in Table 2. From the known kerogen contents, one can then calculate the percentages of char in the kerogens as shown in the last column. The Green River shale showed the lowest char content, 13.7%, and the Rotem shale a higher value, 22.5%, while the Kentucky shale had the very high value of 43.7%, which can be attributed to its high

Batch supercritical extraction of shales in an autoclave A previous paper described a series of supercritical extractions of Rotem shales, in toluene, at 340°C. It was shown that the amount of extract increased with time and reached an assymptotic value after about 60min. Prolonged extraction beyond 60 min led to a considerable decrease in the MW of the extract. The maximum yield obtained was 60%. The present extractions were carried out at 400°C which is above the critical temperature of all the solvent systems used. The results are summarized in Table5. To distinguish between the thermal and the solvolytic reactions, the shales were also heated in argon at 400°C for the same period, and the products were then extracted with toluene in the same way as the other samples. It is seen (Table5) that pyrolysis alone for a prolonged period results in considerable degradation of the kerogen into molecules small enough to be soluble in hot toluene. Furthermore, there is an additional fraction that is not soluble but is volatile, which can be estimated from the weight loss shown by TGA under nitrogen. The two fractions combined coincide almost

FUEL,

1989,

Vol 68, June

705

Extraction

of oil shales under supercritical

Table 3 Relative absorption of demineralized kerogens

intensities

of major

conditions:

functional

groups

R. Kramer and M. Levy relative

to the intensity

Relative

Kerogen Rotem Green

River

Kentucky

intensity

(frequency,

in the FT-i.r.

Ar C-C (1620)

0.24

0.29

0.26

0.42

0.28

0.09

0.54

0.43

0.50

0.32

0.66

0.33

0.40

0.80

0.50

0.36

0.98

2000

1500

1000

Wavenumbers of demineralized

FUEL, 1989, Vol 68, June

chars;

a,

Rotem:

b,

spectra

cm- ‘)

co (1710)

Symmetric bend CH, (1380)

Plane deformation CH aromatic (740) 0.05

D.s.c. of initial shales

Rotem Green River Kentucky

706

at 2925cm-’

OH (3410)

Shale

FT-i.r. spectra c, Green River

CH,

Asymmetric bend CH,, CH, (1460)

Table 4

Figure 4 Kentucky;

of asymmetric

H of kerogen kJg-’ shale

H of char kJg-’ shale

Char in kerogen by d.s.c. (%)

Char in kerogen by TGA (%)

1.9 10.3 2.9 _________.

0.46 1.40 1.10 __~

24.2 13.6 37.9

22.5 13.7 43.7

exactly with the volatile fraction of the initial shales from Rotem and Kentucky. This indicated that heating in argon for an hour did not lead to any additional charring. However, with Green River shale there was a definite increase in the amount of char after prolonged heating in argon. The extraction results obtained with the different solvents show considerable variation. Isopropanol/water at pH 14 is a very efficient solvent for Rotem shale, leaving only 1% of non-extractable or non-volatile kerogen in the residue. Under similar conditions, Kentucky shale yielded 13.9% char. The MWs of the products were quite low in both cases, indicating that considerable degradation took place due to alkaline hydrolysis. With this solvent system, a considerable amount of hydrogen was detected in the gas phase. As is well known, alkaline isopropanol is a very active solvent and will react with kerogen mainly by hydride transfer. Even so it can be seen that, unlike Rotem shale, Kentucky shale was not completely solubilized. Green River shale was not tested as it is dissolved even by non-alkaline isopropanol. Isopropanol alone is also a good solvent for Rotem shale but not for Kentucky shale; this may be explained by the aromatic nature of the latter. Tetralin is known as a hydrogen donor and therefore it is expected to be a good solvent. Note that the molecular weights of the resulting extracts are much higher than in any other solvent, indicating that hydrogen transfer rather than extensive degradation was the main reaction. Water and water/CO mixtures give good results with Green River and Rotem shales, but poor results with Kentucky shale. This again is consistent with the highly aromatic structure of the Kentucky shale. Here again, the solvent is hydrolytically active and gases are produced during the extraction. Hydrogen formed 18% of the gases produced during the extraction of Rotem shale, compared with 6% for the other two shales as well as in blank experiments. However, detailed analysis of all the gaseous products requires further work.

Extraction Table 5

Batch

supercritical

Shale

Solvent

extraction

of oil shales under supercritical

conditions:

R. Kramer and M. Levy

of shales at 400°C in an autoclave

Pressure

SCE + volatiles

(atm)

(%)

SCE (%)

Volatiles

(%)

Char

MW

(%)

R

Argon

80

78.1

64.1

14.0

21.9

240

GR

Argon

80

79.5

78.3

1.8

20.5

270

K

Argon

80

56.5

47.1

9.0

43.5

280

R

Isopropanol/water pH 14

170

99.0

93.2

5.8

K

pH 14

170

86.1

79.5

6.6

13.9 5.4

I .o

245 280

R

Isopropanol

70

94.6

85.4

9.2

CR

Isopropanol

70

98.7

97.2

1.5

K

Isopropanol

70

63.1

49.5

13.6

36.9

270 440

1.3

370

R

Tetralin

70

96.7

86.1

13.1

3.3

GR

Tetralin

70

98.3

96.1

2.2

1.7

810

K

Tetrahn

70

95.2

83.6

9.1

4.8

660 420

R

Toluene

120

91.9

79.1

12.8

8.1

GR

Toluene

120

97.9

95.5

1.6

2.9

510

K

Toluene

120

75.9

58.6

17.3

24.1

430

R

Water pH 12.3

180

87.8

83.0

4.8

12.3

280

GR

pH 12.3

180

91.6

89.8

1.8

8.4

290

K

pH 12.3

180

63.3

51.6

11.7

36.7

240

R

Water/CO pH 12.3

210

96.6

90.0

6.6

3.4

360

GR

pH 12.3

210

98.7

96.2

2.5

1.3

380

K

pH 12.3

210

66.4

53.3

13.1

33.6

310

R = Rotem GR = Green River K = Kentucky

Table 6

Comparison

Mode” time (min)

Shale Solvent-toluene R

GR

K

Solvent-tetralin R

GR

K

‘b=

of supercritical

batch

in autoclave;

extraction

of shales at 400°C in flow and batch

SCE +volatiles

(%)

systems

SCE (%)

Volatiles

(%)

Char

(%I

MW

b/60

87.2

69.7

17.5

12.8

f/48

91.1

74.2

16.9

8.9

f/32

92.2

73.0

19.6

7.4

50&280

f/16

94.1

72.7

21.4

5.9

470-380

340 [email protected]

b/60

97.1

95.5

1.6

2.9

510

f/48

94.0

90.8

3.2

6.0

53tX640

f/32

92.0

87.9

4.1

8.0

500-620

b/60

75.9

58.6

17.3

24.1

430

f/48

71.6

55.1

16.5

28.4

540-520

f/16

76.0

45.4

21.6

33.0

520-460

b/60

96.7

83.6

13.1

3.3

440

f/60

98.8

86.0

12.8

1.2

6w20

b/60

98.3

96.1

2.2

1.7

810

f/180

99.9

98.3

1.6

0.1

50&600

b/60

97.0

82.1

14.9

3.0

720

f/48

98.3

83.8

14.5

1.7

500

f=flow

system

FUEL,

1989,

Vol 68, June

707

Extraction

of oil shales

under

supercritical

conditions:

R. Kramer

and M. Levy

Figure 6 compares the results of g.p.c. of the materials obtained from the three shales by supercritical extraction in toluene. From these curves the average MWs were calculated, and one can also derive the MW distribution. Furthermore it is possible to stop the flow of eluate at any point during chromatography and to scan the whole spectrum for a specific fraction. Supercritical

of shales in a jlow system

extraction

Supercritical extraction in a flow system allowed the extraction process to be followed as a function of time; the solution was collected every few minuts and subjected to g.p.c. analysis for total absorption at 300nm and for MW determination. Because the solvent is under constant flow through the heated zone, the soluble products are removed from the heated zone as soon as they are dissolved, to avoid any further degradation. From Figure 7 it is seen that the maximum rate of extraction occurred after z l&20 min. Table 6 compares the amounts extracted in the batch and in the flow systems. It can be seen that the char formed in extraction of Rotem shale in toluene is less in the flow system than in the batch system. When the time of extraction was reduced from 48 to 32 and 16 min, the char was reduced

1

,

;

1\ 1

/Ii \,\ I

I

f

I

\I 1I \’

I

\’

‘.

1 IO

I 12

I

I

I

I

I

I

14

16

18

20

22

24

Elution volume (ml)

I 10000

I 1000

I 300

I

I 100

MW

Figure 5 preheating

708

D.s.c. of oil shales in: -, oxygen; -. -. , oxygen in nitrogen; a, Rotem; b, Kentucky; c, Green River

FUEL, 1989, Vol 68, June

after

Figure 6 extraction River

G.p.c. of oil shale extracts in toluene: -, Rotem;

obtained

by batch

-. -. , Kentucky;

supercritical ---, Green

Extraction

I! 3

"0 -

of oil shales under supercritical

conditions:

R. Kramer and M. Levy

The MW of the initial extract was high (600), but the value was reduced to 420, the value of the batch extract, as extraction progressed. Kentucky shale was not easily extracted by toluene in either the batch or the flow system. The flow results were worse than in batch, and the amount of char at 16min was greater than after 48 min flow or 60min batch extraction. With tetralin, on the other hand, the flow system gave less char. Green River shale gave better results in batch than in flow. This was probably due to the high content of kerogen in this shale, which necessitated a higher volume of fluid. With tetralin, a long experiment (3 h) was carried out, giving practically total extraction of the kerogen.

1

<

b-4 0. 5 I

ACKNOWLEDGEMENT

0

I

I

I

I

I

I

I

8

16

24

32

40

48

56

The authors thank Professor Yuda Yurum from the Chemistry Department, Hacettepe University, Ankara, Turkey, who initiated this work during his tenure as Visiting Scientist at the Weizmann Institute of Science. This work was supported by a grant from the US Department of Energy, in the framework of the cooperation agreement in Energy Research and Development with the Israeli Ministry of Energy and Infrastructure.

f (mln)

Figure 7 Supercritical extraction in flow system-total extracts in toluene as a function of time of flow: -, Kentucky: p-p) Green River (x 2)

absorption Rotem; .~.

of

~. ,

from 8.9 to 7.4 and 5.9% respectively. The amount of kerogen extracted by the solvent, was73374% in each case. However, the volatile fraction increased, so that the total extracted kerogen also increased. The MW decreased from 460 to 270 during the 48 min experiment, while at 16min it decreased from 470 to only 380. This is another indication that less degradation took place over the shorter time. Tetralin, which was a very good solvent for batch extraction, was also applied in flow experiments, and gave a very low char yield, only 1.2%.

REFERENCES Ross, D. S., Blessing, J. E., Nguyen. Q. C. and Hum, G. P. Furl 1984, 63, 1206 MaKabe, M., Fuse, S. and Ouchi, K. Fuel 1978, 57, 801 Vandergrift, G. F., Winmans. R. E. and Horwitz, F. P. Fuel 1980, 59, 634 Compton, L. E. Am. Chem. Sot., Dio. Fuel Chem. Prepr. 1983, 28(41), 205 McKay, J. F., Chong, S.-L. and Gardner, G. W. Liquid Fuels Technol. 1983, 1, 259 Chong, S.-L. and McKay, J. F. Fuel Science nnd Tech&. Int. 1987. 5,513 Yurum, Y., Kramer, R. and Levy, M. Fuel Science and Technol. Inc. 1986, 4, 501 Yurum, Y., Kramer, R. and Levy, M. Thermochim. Acta 1985.94. 285

FUEL,I~~~,VOI

68, June

709