hydrogen plasmas - Industrial

Ind. Eng. Chem. Res. , 1992, 31 (3), pp 818–827. DOI: 10.1021/ie00003a025. Publication Date: March 1992. ACS Legacy Archive. Cite this:Ind. Eng. Che...
2 downloads 0 Views 1MB Size
Ind. Eng. Chem. Res. 1992,31, 818-827

818 e70 eo = D-

rp2 The initiation rate equation is obtained from the dimensionless form of eq A57.

Ri

=

bD31'2CCo3+CArCHO(s)

ki

rP

k2'I2

(A60)

Registry No. Benzaldehyde, 100-52-7; perbenzoic acid, 93-59-4.

Literature Cited Bawn, C. E. H. Freeradical Reaction in Solution Initiated by Heavy Metal Ion. Discuss. Faraday Soc. 1953,14, 181-190. Blyumberg, E. A.; Malievshii, A. D.; Emanuel, N. M. Effect of Solvents on the Mechanism of Liquid-phase Oxidation of n-Butane. Int. Chem. Eng. 1964,4,400-404. Bykovchenko, V. G.; Berezin, L. V. A Study of the Kinetics and Mechanism of Cyclodecane by Molecular Oxygen IV Study of the Oxidation Mechanism of Cyclododecane by the Inhibition Method. Int. Chern. Eng. 1964,4,391-397. Caloyannis, A. G.; Graydon, W. F. Heterogeneous Catalysis in the Oxidation of p-xylene in the Liquid Phase. J. Catal. 1971, 22, 287-296.

Cerveny, L.; Ruzicka, V. Solvent and Structure Effects in Hydrogenation of Unsaturated Substance on Solid Catalysts. In Advances in Catalysis, Eley, D. D., Pines, H., Weisz, P. B., Eds.; Academic Press: New York, 1981; Vol. 30, pp 335-373. Chou, T. C.; Lin, F. S. Effect of Interface Mass Transfer on the Oxidation of Acetaldehyde. Can. J. Chem. 1983,61,1295-1300. Chou, T. C.; Lee, C. C. Heterogenizing Homogeneous Catalyst. 1. Oxidation of Acetaldehyde. Ind. Eng. Chem. Fundam. 1985,24, 32-39.

Chou, T. C.; Yeh, H. J. Heterogenized Homogeneous Catalyst. 5. The Theory of Solvent Effect and The Effect of Solvent on Adsorption and Diffusivity. Ind. Eng. Chem. Res. 1992, 31, 130. Chou, T. C.; Lin, J. Y.; Liang, C. H.; Do, J. S. Heterogenized Homogeneous Catalyst. 4. Catalyst with a Bias Active Site Distribution. Ind. Eng. Chem. Res. 1990, 29, 180-186. Dean, J. A. Physical Properties. In Lange's Handbook of Chemistry; McGraw-Hill: New York, 1985; Section 10, pp 103-114. Emanuel, N. M.; Zaikov, G. E.; Maizus, Z. K. The Role of Solvation in Chemical Reaction Kinetics. In Oxidation of Organic Cornpound: Medium Effect in Radical Reaction; Pergamon Press:

London, 1984; Chapter 3, pp 129-131. Graulke, E. A. Solubility Parameter Values. In Polymer Handbook; Brandrup, J., Immergut, H., Eds.;Wiley: New York, 1989; W i o n VII, pp 526-532. Hwang, B. J.; Chou, T. C. Heterogenizing Homogeneow Catalyst. 2. Effect of Particle Size and Two Phase Kinetic Model. Ind. Eng. Chem. Res. 1987,26, 1132-1140. Iwamoto, I.; Aonuma, J.; Keii, T. Solvent Effect on Heterogeneous Hydrogenation Reaction. Int. Chem. Eng. 1971, 1 1 , 573-577. Kharitonova, A. A.; Kozlova, 2.G.; Tsepalov, V. F.; Gladyshev, G. P. Kinetic Analysis of Antioxidant Properties in Complex Mixture Using Model Chain Reaction. Kinet. Catal. 1979, 20, 486-492. Kuo, M. C.; Chou, T. C. Heterogenizing Homogeneous Catalyst. 3. Oxidation of Benzaldehyde in a Semibatch Tubular Wall Reactor. Ind. Eng. Chem. Res. 1987,26, 1140-1147. Maslov, S. A.; Blyumberg, E. A. Liquid-pham Oxidation of Aldehyde. RUSS.Chem. Rev. 1976,45, 155-180. Morrison, R. T.; Boyd, R. N. Structure and Properties. In Organic Chemistry; Allyn and Bacon: New York, 1983; Chapter 1, pp 31-32.

Niki, E.; Kamiya, Y.; Ohtha, N. Solvent Effects in the Oxidation of Hydrocarbons 11. Oxidations of Tetralin in Various Solvents. Bull. Chem. SOC.Jpn. 1969,42,3224-3229. Ohhatsu, Y.; Takeda, M.; Hara, T.; Osa, T.; Misono, A. The Liquid Phase Oxidation of Acrolein 11. Solvent Effects in the Liquid Phase Oxidation of Acrolein and the Decomposition of Parrcylic Acid. Bull. Chem. SOC.Jpn. 1967,40, 1413-1419. Polyanskii, N. G. Recent Advances in Ion-exchange Catalysis. R w s . Chem. Rev. 1970,39,224-250. Reichardt, C. Solvent Effects in Organic Chemistry. Eble, H. F., Ed.; Monographs in Modern Chemistry; Verlag Chemie: Weinheim, New York, 1979; Vol. 3, pp 61-66, 270-273. Shendrik, A. N.; Dubina, V. N.; Opeida, I. A. Effect of the Medium on the Kinetics of Co-oxidationof Cumene in the Case of Addition of Small Amounts of Benzyl Alcohol. Kinet. Catal. 1984, 25, 634-636.

Szwarc, M.; Smid, J. Hydrogen Bonding in Radical Reactions Some Remarks on CF3 Radicals. J. Chem. Phys. 1957, 27, 421-422. Ushkalova, V. N.; Kadochnikov, G. D. Kinetics of Oxidation of Lipids 111. Comparison of the Efficiencies of Natural and Synthetic Inhibitors. Kinet. Catal. 1984,25, 421-425. Vardanyan, R. L.; Parsyan, G. V.; Gasparyan, R. A. Determination of Kinetic Parameters of Inhibited Oxidation of Two Component System. Kinet. Catal. 1984,25, 1048-1052. Zaikov, G. E.; Howard, J. A.; Ingold, K. U. Absolute Rate Constants for Hydrocarbon Autooxidation XII. Aldehyde: Photo-oxidation Co-oxidation and Inhibition. Can. J. Chem. 1969,47,3017-3029. Received for review May 30, 1991 Accepted November 5,1991

Treatment of Pitch in Argon/Hydrogen Plasmas Bogdan

Z.Dlugogorski,t Dimitrios Berk,* and R i c h a r d J. M u n z

Department of Chemical Engineering, McGill University, Montreal, Quebec, Canada H3A 2A7

Pitch-like vacuum-distilled residue from the CANMET coprocessing process was treated in a stationary particle reactor with argon/hydrogen plasmas. The production of light, unsaturated hydrocarbons was optimized as a function of the average p h a temperature, composition, and residence time. The unconverted residue was characterized in terms of ita elemental and mineralogical content. In pure argon plasma, the conversion to acetylene was about 14% and did not vary appreciably with temperature; other hydrocarbons were detected only in trace quantities. In argon/ hydrogen plasmas the maximum conversion to acetylene and ethylene (25%) was attained between 2900 and 3400 K. Large quantities of soot were produced, but no liquid hydrocarbons were detected. Sulfur was fixed in the unconverted residue by the reduction of calcium sulfate to calcium sulfide. Introduction Canada and the United States possess some of the world's most important deposita of coal and heavy oils. Present address: Department of Chemical Engineering, &ole Polytechnique, Montreal, Quebec, Canada.

The Canada Centre for Mineral and Energy Technology (CANMET) is currently developing a process to transform these into lighter petroleum products, such as C1-CI gases, naphtha, and light and heavy gas oil. The main concept of this process is to mix finely ground coal particles with the heavy oil and an additive (functioning as a catalyst) and then to hydrogenate the resulting mixture. A major

0888-5885/92/2631-0818$Q3.00/Q0 1992 American Chemical Society

Ind. Eng. Chem. Res., Vol. 31, No. 3, 1992 819 Table I. Elemental Compositions of the CANMET Residue and the Feedstocks to the CANMET Process Cold Lake element, CANMET Forestburg vacuum % residue Coal" bottoms" carbon 74.71b 64.04 83.34 hydrogen 6.60b 3.87 9.69 sulfur 3.43b 0.53 5.84 nitrogen 1.w 1.65 0.45 oxygen 4.10b 20.41' 0.68' ash 9.6gd 9.50 C/H 0.94 1.37 0.72 "From Rahimi et al. (1989). bCEC Model 240-XA elemental analyzer. By difference. Ashed at 775 OC to constant mass.

problem is the production of a by-product or a residue which is a pitch-like substance remaining after the vacuum distillation of the product obtained from the coprocessing of coals and bitumens or petroleum resids (Menzies et al., 1981; Rahimiet al., 1989). This pitch, which contains some potentially useful carbon and hydrogen as well as spent catalyst (Table I), must be treated in some way before it is disposed of. Combustion to raise steam is capital intensive and expensive while treatment by conventional technology does not appear to be feasible. It has previously been shown that high conversions of carbonaceous materials to acetylene and ash could be achieved in the thermal plasma system (Kubanek et al., 1986) and that more information is needed on the kinetics of the complex reacting system. Although much attention has been focused on the processing of fossil fuels in plasma (Venugopalan et al., 1980) in the literature, the treatment of pitches has not yet been fully investigated. In the single published study, Nicholson and Littlewood (1972) reported that the maximum conversion of pitch to acetylene was only 34.5% due to the condensed aromatic structure of pitches. This should be contrasted with the conversion of coal to acetylene of 74% reported by the same authors. The relatively low conversion of pitches to acetylene is likely predetermined by the abundance of alicyclic and aliphatic hydrocarbons which are volatile and are not normally present in pitches in large quantities (Mazumdar et al., 1959). During pyrolysis in the plasma gas, these hydrocarbons form condensable compounds called tars or primary volatiles which undergo further cracking resulting in the production of acetylene (Baumann et al., 1988). The information about small conversion of pitch to acetylene is also corroborated by the compilation of conversion data of coal to acetylene (Dixit et al., 1982). For simple and well-defined feedstock the product distributions and conversion to acetylene during plasma treatment may be predicted. For example, the decomposition of methane was extensively studied and the product distribution was well predicted from both kinetic and thermodynamic calculations (Gulyaev and Polak, 1965). Similarly, good predictions of product distribution were obtained for the treatment of gases or vapors of simple hydrocarbons in plasma (e.g., Il'in and Eremin, 1965). There is no general agreement on the controlling mechanism during the treatment of hydrocarbons in plasma although the more recent literature seems to favor kinetic control (Beiers et al., 1988). Whatever the case might be, still much useful information can be obtained from thermodynamic calculations. It appears that the presence of inorganic matter in pitch, which includes catalyst and minerals found in coals, may dramatidy change the product distribution of the plasma reaction as shown during processing of coals with very high ash content (Kozlova et al., 1977). Usually, the decom-

position of coals with higher ash content increases the conversion to carbon monoxide at the expense of the conversion to acetylene (Chakravartty et al., 1976,1984; Bittner et al., 1985). Moreover, if the inorganic components are vaporized during plasma treatment, they may increase the yield of acetylene or conversely may decrease the reaction temperature required for maximum yield (Meubus, 1975a,b; Moisyeyev et al., 1986). The objective of the present study was to examine the treatment of typical CANMET coprocessing residue in a hydrogen/argon thermal plasma. Specifically,the reaction products in both the gaseous and solid phases were examined as a function of plasma gas temperature and composition. Since small amounts of liquid hydrocarbons may be produced during plasma pyrolysis as for example in the Hiils process (Baumann, 1948; Gladisch, 1969),we explored the possibility of production of liquid hydrocarbons as well. Experimental Materials a n d Methods Apparatus. The experimental equipment consists of a high-frequency power supply, a control console, an induction plasma torch, a single particle reactor system, a sampling train, and a suction probe (Figure 1). High-frequency power at 4 MHz is supplied to the plasma torch at a maximum power of 30 kW by the induction power supply containing a grid control to match the oscillator circuit to the plasma gas. A more detailed description of the experimental apparatus may be found in Dlugogorski (1989). Most of the experiments were conducted between 7 and 14 kW of plate power. Above 14 kW the quartz tube devitrifies and breaks, and below 4 kW the flame extinguishes. Between 4 and 7 kW an inverted flame is formed which is characterized by a relatively cold axial region and hot flame fringes. If a crucible is placed inside such an inverted cone, the gas temperature is poorly defined. An attempt was made to inject hydrogen into the axial and in some instances into the tangential injection ports in the torch. Stable operation was attained with maximum 33.5% of hydrogen in argon while maintaining a maximum deviation of local and mean hydrogen concentration of 33.5 4.4%. The reactor was fabricated from a single piece of Monel (Biceroglu, 1978) and housed a reaction chamber with the crucible containing the residue. Optical pyrometry and spectroscopy measurements were performed through a window in the upper section of the reaction chamber. The upper part of the reactor was 5.08 cm (id.) and 11.4 cm in length. The diameter could be decreased to 2.54 cm by adding a water-cooled insert. The lower part of the reactor (2.54 cm in diameter and 5.08 cm in length) contained the reactor outlet (1.9 cm in diameter). Below the outlet there was a cool stagnant region which accommodated the crucible prior to reaction. As verified by a few "blank" experiments, the temperature in this part of the reactor reached about 480 K which was sufficiently high to melt the sample but too low to permit chemical reactions. The molybdenum crucible (1.9-cm length, 1.3-cm i.d., and 1.4-cm 0.d.) was mounted on a 0.32-cm-0.d. alumina rod which moved vertically. The alumina rod was placed slightly off the center line to permit installation of the suction probe. Gas samples were withdrawn from two locations: (1) from the exhaust gas stream leaving the reactor and (2) from the reactor chamber through the suction probe. The samples were taken at different positions but usually 5.5 cm below the top of the crucible slightly off center. The gases were cooled either inside the probe or by passing

820 Ind. Eng. Chem. Res., Vol. 31, No. 3, 1992

m2 TO POWER SUPPLY

MAT

I l l /

D(DFIMR

TO KME HOOD II

M A T EXWFWOER

_OR/ I BI II

PROBE

AM)

CRUCIBLE

\

TO COLLECTION BOTTLE

Figure 1. Experimental apparatus.

w

g M R I f f i BAG

WRTER CUT

WATER I N

R F R I T E E€=

Figure 2. Probe and sampling train.

through 12.2 m of stainless steel tubing (0.32-cm 0.d.) submerged in an ice-water mixture when samples were withdrawn at the reactor outlet. They were passed through a condenser and a filter to trap soot particles and then sent through a rotameter to verify that the flow rate was constant. Subsequently the gas was either exhausted to the fume hood or collected in a plastic bag as an integral sample (Figure 2). The water-cooled suction probe was designed to withdraw the hot gas from the reactor chamber and quench it at a rate of at least 0.4 X lo6 K/s (Sundstrom and DeMichiell, 1971) to reduce acetylene decomposition and to measure the gas inlet temperature using a W-5% Re/W26% Re thermocouple. The gas flow rates through the probe were usually around 250 cm3/min. The probe was constructed of three concentric stainless steel tubes with outside diameters of 1.6,3.2, and 6.4 mm and a length of 60 cm. Stainless steel was chosen for ita rigidity and inertness. The cooling water flow was 0.7 L/min. The tip of the probe which was made of molybdenum was not water cooled. The bare wire (0.13 mm in diameter) thermocouple was passed through alumina thermocouple insulation and placed inside the innermost tube. To minimize the conduction losses from the thermocouple junction, the top 1cm of the thermocouple wires was not insulated. Experimental Procedure. Before an experimental run,approximately 1.20 g of the residue was packed in the crucible and melted to eliminate trapped air. The mass

of the residue was determined with a precision of 0.0001 g. The crucible was mounted on the alumina support rod and placed in the lower part of the reactor below the gas exit. The suction probe was positioned in the reactor and was not moved during the experiment. The reactor was attached to the torch, and the cooling water was allowed to flow. After inspection for leaks and verification that the reactor system was electrically floating (not grounded), the reactor was purged with argon to remove any residual oxygen. The plasma was initiated, and the discharge was stabilized by adjusting the grid setting and the argon flow. The plate power was adjusted as required. Extra argon was added through the reactor window shroud to prevent the deposition of soot. The required amount of hydrogen was injected into the argon tailflame. Gas samples were taken with l-cm3gas-tight syringes every 3-15 s depending on the operating conditions. After 3-5 min the torch was turned off, the argon and hydrogen flows were stopped, and the crucible was allowed to cool. Immediately following an experimental run the composition of gas samples was determined on the gas chromatograph. The unreacted residue was removed from the reactor, separated into an upper and lower part, weighed, ground, and prepared for further analysis. Soot samples were also taken. Analytical Methods. Mixtures of permanent gases (H2, C02,Ar/02, N2,CO) and hydrocarbons (CHI, C2Hs, C2H4,C2H2)were separated isothermally using a twocolumn system which consisted of a Porapak N (0.32-cm i.d. X 1.98 m) followed by a Molecular Sieve 5A column (0.32-cmi.d. X 2.74 m). The gas separation was performed in the isothermal Fisher Gas Partitioner 1200 with the thermal conductivity detector and using 8.5% hydrogen in helium carrier gas. The gas partitioner was adapted by installing a six-port valve with a restrictor in the seriesbypass configuration (Dlugogorski, 1989). The elemental analysis of the fresh and the unconverted residue as well as of the soot was performed on a Control Equipment Corporation (CEC) elemental analyzer (Model 240-XA). The error in the determination of carbon, hydrogen, nitrogen, and sulfur was less than 0.30%. Because

Ind. Eng. Chem. Res., Vol. 31, No. 3, 1992 821

Y

1

4500

A From Spectroscopy

-

4000 c

0

E

e

3500 -

-

3000

-! 2500; & I

10% Hydrogen Added

i

2ooo I500

IO00

t1

2

20% Hydrogen Added

5

8

II

14

17

P l a t e Power. kW

Figure 3. Mean exit gas temperature (plasma parameters: nozzle diameter = 2.54 cm, coil turns = 6, grid = 20.25, argon flow = 20.81 g/min radial and 37.22 g/min swirl).

the residue contains metals which may remove oxygen from the sample, the oxygen analysis may be considered an approximation. The analysis of sulfur includes the sulfur content in both organic and inorganic parts of the residue. The solubility analyses of the fresh and the unconverted residues were performed with tetrahydrofuran (THF), toluene, and pentane according to ASTM D 2042-81. In order to determine the amount of ash in the residue, a sample (2 g) was placed in a crucible and ashed in a furnace at 1048 K to constant mass (f0.5 mg) (ASTM D 482-80). To gain insight into the mineralogical content of the residue, an X-ray powder diffraction (XRD) study was performed. The mineral content of the residue was concentrated by dissolving a significant percentage of the organic part of the residue in THF. Subsequently, a small amount (below 0.1 g) of the THF-insoluble matter was placed in a low-temperature asher (LTA) for more than 50 h at 100-W net FW power and 100 cm3/minoxygen flow. Since the removal of the carbonaceousmaterial is a surface phenomenon, the sample was stirred. The oxidized sample was deposited on a silver membrane in the special vacuum equipment since it was impractical to collect a sufficient amount of samples to use regular powder mounts. The analysis was performed on a Phillips PW 1710 diffractometer using nickel filtered Cu K a radiation generated a t 40 kV and 20 mA. Other techniques, such X-ray fluorescence (XRF), atomic absorption (AA), or scanning electron microscopy @EM) were also employed in the analysis. Temperature Measurements. The mean exit temperature of the plasma stream at the torch nozzle, as well as the joint torch and the power supply efficiency, was determined from calorimetric experiments. The mean exit temperature of the plasma was computed from heat capacities tabulated by Lesinski and Boulos (1978). Figure 3 illustrates the relation between the plate power and the mean plasma temperature. The effect of hydrogen injection is also shown. The torch efficiency is around 22% at 5-kW plate power, and it decreases monotonically to 15% at 14 kW. The average flame temperature could be adjusted with the precision of f150 K because of the fluctuations in the plasma current. The temperature of the crucible surface was approximated from the measurements performed by a two-color pyrometer using cooling curves, i.e., in the absence of the

strongly radiating plasma. The temperature of the molybdenum crucible depended on the crucible emissivity which was a function of the coating of the crucible surface by soot (Dlugogorski, 1989). This temperature reached 50% of the gas temperature. In addition, the intensity of light emitted by the red-hot soot particles in the plasma gas was so high that it prevented the measurement of the crucible temperature during the actual runs. Finally, the temperatures measured by a W-5% Re/W26% Re thermocouple placed inside the aspirated probe were judged not to be reliable due to the formation of a layer of soot a t the probe tip. Thus, although several techniques may be used to measure either local or average flame temperatures, it is most correct to correlate the results using the calorimetrically determined average flame temperature. Because the crucible was always in the core of the plasma jet, it was this average temperature that was most relevant.

Results and Discussion Analysis of the CANMET Residue. Prior to the analysis,the residue was homogenized by heating (to above 470 K) in an oil bath and by stirring (ASTM D 140-70). Because of the limited amount of the available residue, the sample size chosen was 2-3 g. The elemental abundances in the CANMET residue are dependent on the elemental composition of the Forestburg Coal and the Cold Lake vacuum bottoms. Sulfur,nitrogen, and oxygen are preferentially retained in higher boiling fractions from the CANMET process and in the remaining residue (Rahimi et d,1989). The C/H ratio of the residue (0.94) is between those of Forestburg coal (1.37) and the Cold Lake vacuum bottoms (0.72). It corresponds to the C/H ratio of hydrocarbons that distil between 478 K and 798 K, are soluble in methyl tert-butyl ether, and are characterized by high aromaticity and high nitrogen content (Rahimi et al., 1989) (Table I). Thus it appears that the organic part of the CANMET residue contains a highly condensed aromatic structure. This was corroborated further by investigating the solubility of the residue in different solvents. The solubilities of the residue in tetrahydrofuran (THF), toluene, and pentane were found to be 72.1, 63.3, and 46.3% respectively. THF is a powerful solvent for a wide spectrum of synthetic and natural materials. In addition to diesolvingaliphatic, alicyclic, and aromatic hydrocarbons (as pentane does), it dissolves heterocyclic aromatic compounds more efficiently than does toluene. The aromaticity of toluene solubles is higher than pentane solubles and smaller than THF solubles. According to Rahimi et al. (1989), the solubility of the Cold Lake vacuum bottoms in pentane was 76.2%. Therefore the structure of the CANMET vacuum residue is more aromatic than the structure of the Cold Lake vacuum bottoms, especially since there is a substantial difference in the solubility of the residue in pentane and in toluene. Moreover, 8.8% of the residue is soluble in THF but not in toluene and about 15% of the organic part of the residue is not THF soluble at all. Again, this indicates that the residue containsa high proportion of condensed and complex aromatic hydrocarbons which incorporate heteroatoms in their structure. The ashing of THF-insoluble matter, at 1048 K, yielded 34.57% ash, Le., 9.65% on the basis of the original residue. This may be compared with 9.68% from Table I, showing a small error in those results. If the residue is ashed at 1273 K for a few hours to a constant mass, all metal oxides are transformed to their highest and known levels of oxidation. Table I1 shows the analysis of the ashed CAN-

822 Ind. Eng. Chem. Res., Vol. 31, No. 3,1992 Table 11. Results from the XRF Analysis of the Ashed CANMET Residue content content compd compd 0.29% Si02 36.99% K20 0.11% 22.51% MnO Fe203 5302 ppm 18.42% V A1203 3852 ppm 15.57% Ni CaO 3512 ppm MgO 3.52% BaO 960 ppm NazO 0.68% PZO, 622 ppm TiOa 0.47% Cr203

0

100%Argon

~~

MET residue performed by X-ray fluorescence (XRF). W m i et al. (1989)reported that the Cold Lake vacuum bottoms contain V (235ppm), Ni (93ppm), and Fe (18 ppm). Although the presence of some metals is beneficial in the coproceasing of coal and vacuum bottoms since they act as catalysts (e.g., nickel, vanadium, iron sulfate, and chromium oxide), the assembly of the inorganic compounds (especially oxides) in the residue has an overall negative effect on the conversion of the residue to acetylene in the plasma since they enhance conversion to carbon monoxide rather than to acetylene. The amount of the high-temperature ash is leas than the amount of the inorganic part in the residue because of the thermal decomposition of salts during the ashing process. Thus the amount of hydrocarbons in the residue that could be decomposed in plasma is less than that derived from the carbon and hydrogen content in Table I. Moreover, the mineralogy of the original coal and the residue is much different from the mineralogy of the ash. The mineralogical analysis of the residue was performed by X-ray diffraction (XRD). Among the minerals present trigonal quartz was the easiest to identify. Gypsum, anhydride, hemihydrate (bassanite), and hematite were also seen. It is possible that hematite, anhydride, and to a certain extent bassanite are artifacts from the plasma ashing in the low-temperature asher. In general, carbonates (dolomite, calcite, ankerite), feldspar (orthoclase, albite), clay (kaolinite, illite), siderite, pyrite, and iron sulfate are probably present but they could not be definitely identified. Gas Composition and Conversion to Gaseous Products. For every series of experiments with different hydrogen concentration reported in this section, at least one point was replicated (Table 111). Five replicates were taken at 3390 K for 10% hydrogen plasma. The spread in the acetylene concentration was 80 ppm (between 0.140 and 0.148%), which corresponded to a difference in conversion of 6.6%. (The conversion was based on the total carbon present in the fresh residue.) Similarly, the concentrations of ethylene, methane, and carbon monoxide show a spread of 20,100and 100 ppm that corresponds to the maximum scatter in conversion of 16.9% (between 1.54 and 1.80%), 19.8% (between 1.67 and 2.00%), and 18.1% (between 2.60 and 3.07%), respectively. In general, acetylene, ethylene, methane, carbon monoxide, and hydrogen were the most important gases produced; however, the concentration of hydrogen was too small to be determined accurately. A small amount of ethane was deteded on a few gas chromatograms, but this gas was never observed in the integral sample. The concentration of hydrogen sulfide was below the detection threshold of the gas chromatograph (below 10 ppm). Table I11 presents the concentrations of hydrocarbons and carbon monoxide in the integral samples collected usually during the first few minutes of every experiment. In all experiments the flow of argon was maintained constant at 32.5 L/min; however, the flow of hydrogen was varied between 0.0 and 8.1 L/min.

D 10.01 Hydrogen

N

A 20.0% Hydrogen

h

B

A

0

.e

A

._ 6 '51

P

0 0

5[ 0

m

2500

2000

0

0

3500

4000

4500

G a s Temperature. K

Figure 4. Conversion of carbon in the CANMET residue to acetylene (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrogen flow = 0.32 or 0.73 g/min to the tailflame). I

5.0 4.5 N

4.0

-

0

100%Argon

0

10.0% Hydrogen

A

m' 3.5 5 23.0.

ill

2.5c

0

A A

0

4

A

s 2.0._

Fi

e 1.5-

0

>

6 u

0

0

1.0-

0

0.5

0

0.0'

m

2500

m

O--i-c---d

m

4000

4500

G a s Tenperature. K

Figure 5. Conversion of carbon in the CANMET residue to ethylene (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrcgen flow = 0.32 or 0.73 g/min to the tailflame).

Figure 4 shows the effect of gas temperature on the conversion of carbon to acetylene. The conversion is about twice as high in 10 and 20% hydrogen plasmas (23%) as in pure argon plasma (14%). There is no increase in the acetylene production once the hydrogen content in the plasma gas reaches 10%; thus similar results would be expected in pure hydrogen plasmas. Moveover, the conversion to acetylene in argon/hydrogen plasmas as a function of temperature exhibits a broad maximum at around 3200 K. This trend agrees with the results of the thermodynamic calculations; however, the maximum conversion is lower than predicted. The conversion with argon is almost independent of temperature. Conversion of carbon to ethylene (Figure 5)and methane (Figure 6)is higher in 20% than in 10% hydrogen plasmas. In general, the treatment in pure argon plasma does not lead to the production of ethylene and methane unless the temperatures are below 3000 K. The conversions to ethylene and methane at 2500 K are 3.7 and 4.7770,respectively, and decrease to 1.0 and 1.5% at 3800 K. Small production of methane and ethylene in argon/ hydrogen plasmas and no production in pure argon plasma were predicted from the quasi-equilibrium thermodynamic analysis. The term quasi-equilibrium refers to the fact that solid carbon phase was excluded from the calculations. It is clear that although the thermodynamic analysis predicts

Ind. Eng. Chem. Res., Vol. 31,No. 3,1992 823 Table 111. Hydrocarbons and Carbon Monoxide Contents in the Product Gases expt no. temp, K residue, g Hzflow, L/min time, min CzHz, % H6 2190 1.1950 0.0 4.0 0.076 3.0 0.100 0.0 1.2026 H5 2990 3.0 0.115 0.0 1.2024 3370 H7 3.0 0.115 0.0 1.2252 H2 3670 3.0 0.111 0.0 1.1945 H1 3670 3.0 0.124 0.0 1.2088 4110 H8 3.0 0.120 0.0 1.2085 H4 4440 3.0 0.106 1.1976 0.0 4450 H3 3.0 0.145 1.2093 3.6 J8 2890 3.0 0.160 1.2142 3.6 Jll 3160 3.0 0.148 1.1946 3.6 J1 3390 3.0 0.145 1.2038 3.6 52 3390 2.5 0.168 1.2013 3.6 3390 55 3.0 0.140 1.2040 3.6 J9 3390 3.0 0.143 1.1979 3.6 J10 3390 3.0 0.137 1.2088 3.6 3590 53 3.0 0.132 3.6 1.2134 54 3740 3.0 0.127 3.6 1.1944 J6 3810 3.0 0.125 3.6 1.2049 57 3810 3.0 0.106 8.1 1.2013 2500 I7 3.0 0.120 8.1 1.2176 2720 I6 3.0 0.148 8.1 1.2055 2920 I2 3.0 0.139 8.1 1.1931 2940 I5 3.0 0.138 8.1 I1 1.2096 3080 3.0 0.152 8.1 1.1936 3190 I4 3.0 0.137 8.1 1.1974 3260 I3

4.5

eo, %

CHI, % 0.020

0.016 0.039 0.049 0.060 0.058 0.068 0.074 0.075 0.037

0.000 0.000

O.OO0

0.003 O.OO0 O.OO0 O.OO0 O.OO0 0.019 0.016 0.012 0.014 0.014 0.012 0.013 0.011 0.010 0.008 0.008 0.025 0.022 0.021 0.021 0.018 0.017 0.015

0.006 0.000 0.000 0.000 0.000

0.051 0.037 0.029 0.031 0.037 0.026 0.027 0.027 0.024 0.021 0.025 0.064 0.062 0.057 0.057 0.048 0.046 0.041

0.040

0.040 0.046 0.050 0.048 0.045 0.052 0.055 0.057 0.057 0.023 0.026 0.035 0.031 0.032 0.033 0.035

1

5.0 I

i t

C2H4t % 0.016 0.009

1

HI

A 0

A

B

4.01

0 I O .O% Hydrogen

z

20.0% Hydrogen

U

A

0

100%Rrgon

E k0.01m

A A

U

A

Lu

0 i

2.5

=

a

r

0.001000

- 2.0 -

s

z1.5-

s 1.00.5

1

o.o--+-~--.d

m

0

2500

Jm

m

\

C2H2 4aw)

0.WOa)l

4500

1

1000

0

Gas Tenpemture. K

Figure 6. Conversion of carbon in the CANMET residue to methane (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrogen flow = 0.32 or 0.73 g/min to the tailflame).

the existence of carbon vapor and free radicals such as C2H and H at elevated temperatures, upon quenching these radicals combine to form molecular species. The thermodynamic calculations also anticipated a slightly higher concentration of methane than of ethylene (Figure 7). On the other hand, the production of carbon monoxide (Figure 8)depends only on the temperature of the plasma gas and not on the composition. The conversion of oxygen to CO is 40% at 2190 K, and it increases to slightly more than 100% at 4440 and 4450 K. This apparent inconsistency is explained by the fact that the elemental analysis underestimated the total amount of oxygen in the sample because some oxygen was removed by metals. Depending on the temperature, the emission of carbon monoxide may be divided into two stages. Below 2800 K organic oxygen is removed by cracking; at this stage the conversion of oxygen is approximately constant. Above the gas temperature of 2800 K, inorganic oxygen leaves the residue because of mineralogical transformations, especially the removal of oxygen from calcium sulfate (XRDanalysis showed that the extent of the reduction of calcium sulfate

2ooo

4003

3000

5000

1

6000

TEMPERRTURE. K

Figure 7. Decomposition of the CANMET residue in a hydrogen plasma, quasi-equilibrium calculations without the solid phase (pressure = 101.3 kPa, residue:hydrogen = 7.239 mass ratio).

N

4.5-

i4.0-

z ;3.5 6 -

0 100% Rrgon

E

0

10.0% Hydrogen

A

20.0% Hydrogen

0 0 0

on

0

I: 3.0

6

A

O A

._

e 1 m 1.0, > '

0.50.0' .._

2000

2500

3000 3500 Gas Temperature. K

4000

4500

Figure 8. Conversion of carbon in the CANMET residue to carbon monoxide (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrogen flow = 0.32 or 0.73 g/min to the tailflame).

824 Ind. Eng. Chem. Res., Vol. 31, No. 3, 1992

81

"Or----

3

70-

6

60-

b

50-

n

v A

OA'

'A0

goo

A A

m

3500 Gas Temperature. K

2500

4500

4w)O

Figure 9. Conversion of carbon and oxygen in the CANMET residue to gaseous products (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrogen flow = 0.32 or 0.73 g/min to the tailflame; open symbols denote conversion of oxygen to CO and closed symbols denote conversion of carbon initially present in the residue to all gaseous products excluding CO). N J

bor-

-

0 1LX)Y Rrgon

4 A

0 10.0% Hudrogon

A

A

A

A 20.0% H y d r m

A

0

o

B f 25

0

0 0

0

0

0 0

Y,

P 0'

r

2000

250

rn

3500

m

I

4500

Gas Terrperature. K

Figure 10. Ratio of total hydrogen in acetylene, ethylene, and methane to hydrogen initially present in the CANMET residue (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrogen flow = 0.32 or 0.73 g/min to the tailflame).

to calcium sulfide increases with temperature). At this stage, the extent of the carbon monoxide emission is proportional to the temperature. The conversion of oxygen to carbon monoxide is higher than the conversion of carbon to gaseous products such as acetylene, ethylene, and methane (Figure 9) even at low temperatures. This difference is not fully accounted for by the production of soot, and it suggests that oxygen is more easily removed from the residue than carbon. Figure 10 compares the amount of hydrogen present in acetylene, ethylene, and methane with the amount of hydrogen initially present in the residue. The difference in the conversion between pure argon and argon/hydrogen plasmas resulted not only from the better removal of hydrogen from the residue by argonlhydrogen plasmas due to improved heat transfer (see below) but also from the addition of hydrogen to the gaseous products of pyrolysis. In addition, there is a possibility of the production of small hydrocarbons from pyrolytic carbon and hydrogen catalyzed by oxygen or carbon monoxide (Cao and Back, 1985). This mechanism for improved conversion can also be considered to be directly associated with the amount

25

50

75

IO0

I25 Tim.

150

IE

203

225

260

8

Figure 11. Conversion to acetylene in argon plasma as a function of time (argon flow = 20.81 g/min radial and 37.22 g/min swirl).

of oxygen released from the residue that in turn depends on the efficiency of heat transfer to the crucible. The time required for the residue to react depended somewhat on the plasma gas temperature but was lesa than 110 s for experiments with pure argon plasma (Figure 11) and 85 s for runs with 20% hydrogen/argon plasma. For an experiment performed in an inverted flame at 2190 K, it takes substantially more time for the residue to react because in the case of an inverted flame the centerline temperature is considerably less than the mean temperature measured by calorimetry. The maximum production rates of acetylene were 2.1 X g-mol/s in pure argon g-mol/s in argon/hydrogen plasmas. and 4.5 X In general, it can be concluded that although the conversion of carbon to light hydrocarbons is limited by the cracking of the CANMET residue, the effect of the plasma temperature and ita hydrogen content on the product distribution may be approximately predicted from the thermodynamic considerations (Figure 7). The extent of the decomposition of acetylene to soot depends on the resisdence time of the gases in the hightemperature regions of the reactor. This residence time was calculated on the basis of the gas flow rate at the average flame temperature at the torch exit and the reactor volume below the top of the crucible. For some experiments, an insert (2.54 cm in diameter) was placed inside the upper part of the reactor allowing the residence time between the crucible and the sampling points to vary between 0.36 and 52 ms. No difference in the conversion of carbon to acetylene (within the gas chromatograph error) was observed; this suggests that most of the decomposition of acetylene occurs at the top of the crucible and soot is swept to the lower part of the reactor and deposited on the reactor walls. Thus it appears that the decomposition of acetylene results from the diffusion of the pyrolysis products through the porous layer of soot that forms a t the top of the crucible during the early stages in the experiment. Also, these results hint that the degradation of acetylene to soot may be slower than previously reported in the literature (Melamed et al., 1965). The quenching rate at the beginning of the cooling jacket inside the probe was estimated to be greater than 0.4 X lo6 K/s (Dlugogorski, 1990). This corresponds well with the average quench rate of 0.16 X lo6 K/s computed by Sundstrom and DeMichiell(1971) between 1500 and 600 K and the point quench rate of 0.31 X lo6 K/s at 1150 K. Due to the higher temperature gradients during plasma experiments, the quenching rate is likely to be higher and

1

II:

3

0

Ind. Eng. Chem. Res., Vol. 31, No. 3, 1992 825 1000000

110, AD

&

A

”’ 0 33.5

0

h

0

0

THF I n s o I b I e s

A

0

A O A cy

4-

100 X Rrgon

Unreacted Residue

/

U Y

> L

03

0

RLUMINUM

A o

t IO00

2400

d ”- 6

c

Rsh on THF Inso I ub les

2000

0

4-

c u

0

A

10: 1600

n

100000 [

Hydrogen

6l

0’ 1200

100.0 X Rrgon

X Hydrogen

A 17.3 % 0

33.5 X Hydrogen

A 17.3 X Hydrogen

2800

Temperature.

3200

3600

4000

4400

r(

1200

\ 1600

2wM

2400

2800

3200

NICKEL

3600

4000

4400

Temperature, K

Figure 12. Analysis of the unconverted residue (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrogen flow = 0.61 or 1.46 g/min to the tailflame).

Figure 13. Analysis of the ash obtained from the unconverted residue (argon flow = 20.81 g/min radial and 37.22 g/min swirl, hydrogen flow = 0.61 or 1.46 g/min to the tailflame).

hence the decomposition of acetylene inside the probe would remain insignificant. Some reports in the literature (Melamed et al., 1965; Bukham et al., 1965) supplied conservative theoretical estimates to predict the quenching rates necessary to preserve the whole amount of acetylene. For example, according to their calculations for the inlet temperature of 1800 K, the quenching rate should be 1.0 X lo6 K/s. Finally, in spite of the replacement of the icewater bath with dry ice and liquid nitrogen traps, no liquid hydrocarbons were condensed. I t is known that, during the decomposition of long-chain paraffins and high molecular weight cycloalkanes and aromatics to light unsaturated hydrocarbons in plasma, the conversion to liquid hydrocarbons is around 2.0 and 12.0%, respectively (Baronnet et al., 1987). The CANMET residue contains heterocyclic complex aromatic hydrocarbons rather than aliphatics and cycloalkanes. However, as determined by Baumann et al. (1988), it is the latter that decompose in plasma to form condensable compounds called tars or primary volatiles which undergo further cracking resulting in the production of acetylene. Thus, the hydrocarbon content of the CANMET residue appears to be the reason for no production of liquid hydrocarbons. Characterization of the Unconverted Residue. This section examines the effect of the temperature and the hydrogen concentration in the plasma gas on the organic and inorganic materials which were left in the crucible after the plasma treatment. In general, only small changes may be expected in the organic and inorganic parts after the pyrolysis of the hydrocarbons was completed. For example, the emission of carbon monoxide suggests that the minerals are still undergoing some transformations and that the organic carbon may be consumed in the process. However, these changes are so small that they cannot be accounted for by the analytical techniques presented below. We did not attempt to introduce any catalysts before the plasma treatment to enhance the production of acetylene since the effect of such catalysts is unclear, and it appears to depend on the type of the plasma used (Anderson et al., 1968). In this study, the production of H2S was negligible;hence there was no need to add calcium and potassium carbonate, which tend to retain sulfur in the unreacted residue (Bozzuto, 1984). The residue left in the crucible after the reaction accounts for between 40 and 50% of the initial mass. As

already mentioned, the results presented in Figure 12 are correlated with the average gas instead of the crucible temperature. These measurements are associated with an uncertainty of &5%. The amount of the residue decreases with temperature but the effect of the plasma gas cannot be unequivocally determined. The unconverted residue contained practically no hydrocarbons which were soluble in tetrahydrofuran. This implies that the aliphatic, alicyclic, and part of aromatic hydrocarbons have been cracked, and the organic part of the residue which is not soluble in THF contains only a condensed aromatic structure with heteroatoms. Since at 1600 K around 50% of the original residue is already removed from the crucible, most of the pyrolysis takes place at lower temperatures. However, the amount of ash left in THF insolubles increases slightly with temperature-up to 25% above 4200 K. This indicates that at higher gas temperatures inorganic elements become more concentrated in the crucible whereas carbon is preferentially removed to the gas phase. This is expected since increasing gas and crucible temperatures gives faster and more complete pyrolysis. After ashing, the unconverted residue was analyzed for iron, aluminum, vanadium, and nickel on an atomic absorption apparatus (Figure 13). These metals were selected for their potential to be used as catalysts in the CANMET process; therefore their recovery may be economically important. In addition, some researchers showed (Meubus, 1975a,b; Moisyeyev et al., 1986)that the presence of metal particles such as Fe, Al, and Ni in the reacting mixture of hydrocarbons in plasma may change the product distribution enhancing the production of acetylene. However, no significant variations in the contents of Fe, Al, V,and Ni were observed as a function of temperature and composition of the plasma gas. Thus, two conclusions may be drawn: (1)the metals have no effect on the reactions in the gaseous phase and (2) further separation is required to recover the catalyst from the unreacted residue. The hydrogen content of the unreacted residue may vary greatly depending on the position of the unreacted residue in the crucible. For the carbon and hydrogen analyses the unconverted residue was divided into two samples: one representing the top portion of the unreacted material (which was directly exposed to the plasma flame) and the bottom part (which remained at approximately crucible temperature). With the exception of experiments conducted under the inverted-flame conditions (Figure 14),

826 Ind. Eng. Chem. Res., Vol. 31, No. 3, 1992 7

r

1 6 I

o 1 0 0 . 0 ~Argon

A 20.0%

Hydrogen

0 10.0% Hydrogen

Inverted Flame

C

m

A

0'

EOOO

2500

I

,

3000 3500 Gas T e ~ ~ e r o t u r Ks .

4wo

4500

Figure 14. Hydrogen content of the unconverted residue: open symbols, sample from top of crucible; closed symbols, sample from bottom of crucible (argon flow = 20.81g/min radial and 37.22g/min swirl, hydrogen flow = 0.32 or 0.73 g/min to the tailflame). Table IV. Nitrogen, Oxygen, and Sulfur Contents of the Unconverted Residue in the Upper and Lower Locations of the Crucible" analyzed upper lower temp, hydrogen in K part, % elem plasma, % part, % 1.24,1.24 1.46 N 4190 0.0 1.39 1.50 N 3670 0.0 1.35,1.39 1.67 N 3370 0.0 1.68,1.58 1.67 N 2190 0.0 1.32,1.26 1.78,1.73 Nb 3390 10.0 1.41 1.60,1.61 Nb 3390 10.0 3260 20.0 0.75 1.08 N 1.32 1.56 N 3190 20.0 0.95 1.65 N 3080 20.0 0.98 1.56 N 2920 20.0 0.44 3.43 0 3080 20.0 4.19 S' 20.0 4.37 3590 'Errors in this table are below 0.3% absolute. The error associated with the determination of sulfur was *0.20% (absolute). *Separate experiments. Using LECO induction furnace and Metrohm titrator.

all results show that the top part of the unreacted material contains less hydrogen. There is a slight decrease in residue hydrogen content as the plasma gas temperature is increased. Also for the same average flame temperature, more hydrogen is removed from the residue in argonlhydrogen than in pure argon plasmas. This may be due to a more efficient heat transfer from plasma to residue for argon/hydrogen plasmas. The fact that the residue in the lower (and colder) part of the crucible changes its hydrogen content suggests that there is some reaction even here. The nitrogen and the sulfur content of the unconverted residue does not vary appreciably with the position inside the crucible. This is corroborated by the results shown in Table IV. The nitrogen and the sulfur content in the original CANMET residue were 1.18 and 3.4390, respectively (Table I). Hence nitrogen and sulfur become more concentrated in the unreacted residue. This again indicates that the aliphatic, alicyclic, and aromatic hydrocarbons are pyrolyzed first, and those containing heteroatoms such as N and S decompose during the later stages of pyrolysis and their decomposition may not be complete. In addition, as will be shown later, the organic and inorganic sulfur is retained in Cas. On the other hand, the oxygen content of the unreacted residue is high in the lower part of the crucible and low in the upper part. The inorganic oxygen is released from

the calcium and magnesium salts by thermal decomposition or during the fixation of sulfur and from kaolinite by dehydration. The presence of the organic oxygen may actually enhance the initial cracking since oxygen would be preferentially removed from the hydrocarbon structure. Finally, XRD analysis was performed for samples of the unconverted residue which was exposed to plasma at different temperatures and various hydrogen concentrations. Gypsum (CaS04.2H20) and bassanite (CaSO,. 1.5H20),which were the easiest to observe, converted initially to anhydride and then to CaS03and eventually to calcium sulfide (Cas). Thus the following mechanism is suggested for the fixation of sulfur during the plasma treatment: CaS04.2H20 CaS04-1.5H20 CaSO, CaS03 Cas (1) The fixation of sulfur in the residue might have been already inferred from the lack of H2S in the outlet gases. A similar conclusion can be drawn by analyzing the residue before and after the reaction by scanning electron microscopy. However, after the reaction, sulfur was evenly distributed in the sample and not associated with single crystals as it was before the reaction. Hence, it seems that the inorganic minerals initially melted (or partially melted) during the reaction and they did not appreciably recrystallize due to the rapid cooling of the crucible after the experiment was stopped. In addition the carbonates (dolomite, calcite) which easily decompose to calcium and magnesium oxides during the plasma treatment might have also participated in fixing the sulfur, as described by the overall reaction (Ibarra et al., 1989) CaO + H2S Cas + H,O (2) In general the transformations of the iron-bearing compounds could not be resolved by X-ray powder diffraction. For example, it appears that iron sulfate decomposes only a t higher temperatures. However, the mineralogical products of this decomposition could not be identified. Although the production of iron would be expected, it could not be confirmed due to the low resolution of the unashed XRD traces and the oxidizing character of the low-temperature asher (the unconverted residue was not ashed to avoid any chemical changes; the plasmas used were inert or reducing while LTA is strongly oxidizing).

- - -

-

-

Conclusions The conversion of carbon in the residue to light unsaturated hydrocarbons was shown to be limited by the chemical nature of the CANMET residue. Specifically, the low conversion to acetylene (and also to ethylene) resulted from the condensed heteroaromatic structure of the residue, its high oxygen content, the absence of aliphatic and alicyclic hydrocarbons, and the low cracking temperature. Due to this low temperature, carbon in the unconverted residue did not react with hydrogen present in the plasma gas. The crucible temperature was much lower than the plasma gas temperature, resulting in the slow pyrolysis and the incomplete cracking of the hydrocarbons in the lower part of the crucible. The crucible temperature, however, was sufficiently high to allow reduction of inorganic compounds in the residue. Thus, it may be concluded that the treatment of the CANMET residue consists of two stages: (1)heterogeneous reactions inside the crucible which include the pyrolysis of the hydrocarbons and the transformation of the mineral content of the residue and (2) homogeneous reactions between the gas products from the pyrolysis and the plasma gas.

Ind. Eng. Chem. Res., Vol. 31, No. 3, 1992 827 The presence of hydrogen in the plasma gas improves the pyrolysis of the residue due to more efficient heat transfer between the plasma gas and the residue. In addition, hydrogen takes part in the gaseous phase reactions improving, by almost 2-fold, the acetylene yield and enhancing the production of ethylene and methane. During the treatment in plasma, the inorganic minerals in the residue lost their crystallization water or decomposed. Sulfur was not released but was rather retained in the unconverted residue as calcium sulfide. The released oxygen reacted with carbon, forming carbon monoxide. Nitrogen became more concentrated in the unconverted residue, eapeciaUy in the lower part of the crucible. Finally, since no metals were removed from the crucible during the treatment, they were not present in the continuously flowing stream of gas; thus they had no effect on the distribution of gaseous products.

Acknowledgment We thank CANMET for the financial support for this project. The scholarships from the National Science and Engineering Council of Canada and from the Department of Chemical Engineering (Eugenie Lamothe Bequest), McGill University, are also gratefully acknowledged. Registry No. Ar, 7440-37-1; H2, 1333-74-0; CaS, 20548-54-3; C2H4,7485-1; CzH2,74-86-2; CO, 630-08-0; CHI, 7482-8; CaS04, 7778-18-9; gypsum, 13397-24-5; anhydride, 14798-040; b i t e , 17033-35-1; quartz, 14808-60-7; hematite, 1317-60-8.

Literature Cited Anderson, L. L.; Hill, G. R.; McDonald, E. H.; McIntmh, M. J. Flash Heating and Plasma Pyrolysis of Coal. Chem. Eng. Prog. Symp. Ser. 1968,64, 81-88. Baronnet, J. M.; Lesinski, J.; Sauvage, B.; Vandensteendam, C.; Meillot, E. Hydropyrolysis of Heavy Oils in H2/CH4Arc Plasma. Znt. Symp. Plasma Chem. 1987,8th, paper D I G . Baumann, H.; Bittner, D.; Beiers, H.-G.; Klein, J.; J b t g e n , H. Pyrolysis of Coal in Hydrogen and Helium Plasmas. Fuel 1988,67, 1120-1123. Baumann, P. Acetylene Production by an Electric Arc Process. Angew. Chem. 1948, B20, 257-259 (in German). Beiers, H.-G.; Baumann, H.; Bittner, D.; Klein, J.; J b t g e n , H. Pyrolysis of Some Gaseous and Liquid Hydrocarbons in Hydrogen Plasma. Fuel 1988,67, 1012-1016. Biceroglu, 0. Chlorination Kinetics of ZrO in an RF Plasma Tailflame. Ph.D. Thesis, McGill University, Montreal, 1978. Bittner, D.; Baumann, H.; Klein, J. Relation between Coal Properties and Acetylene Yield in Plasma Pyrolysis. Fuel 1985, 64, 1370-1374. Bozzuto, C. R. Acetylene from Coal and an Electric Arc. United States Patent 4,487,683, Dec 11, 1984. Bukhman, F. A.; Melamed, V. G.; Polak, L. C.; Khait, Yu. L. Calculation of Some Parameters for Acetylene Quenching during ita Production from Methane. In Kinetics and Thermodynamics of Chemical Reactions in Low Temperature Plasma; Polak, L. S., Ed.; Nauka: Moscow, 1965; pp 52-59 (in Russian). Cao, J. R.; Back, M. H. The Kinetics of the Reaction of Carbon with Hydrogen Containing 0.1% Oxygen. Carbon 1985,23,141-145. Chakravartty, S. C.; Dutta, D.; Lahiri, A. Reactions of Coals under Plasma Conditions: Direct Production of Acetylene from Coal. Fuel 1976,55,43-46. Chakravartty, S. C.; Dixit, L. P.; Srivastava, S. K. Hydrogen Enriched Plasma for Direct Production of Acetylene from Coal.

Indian J. Technol. 1984,22, 146-150. Dixit, L. P.; Srivastava, S. K.; Chakravartty, S. C.; Bandopadhyay, A. K. Theoretical Maximal Yield of Acetylene from Coals by Plasma Pyrolysis. Fuel Process. Technol. 1982,6, 85-91. Dlugogorski,B. 2.The Treatment of Vacuum Distilled Residue from the CANMET Process in Argon/Hydrogen Plasmas. M.Eng. Thesis, McGill University, Montreal, 1989. Gladisch, H. Acetylene Production in an Electric Arc, Chem.-Zng.Tech. 1969,41,205-208 (in German). Gulyaev, G. V.; Polak, L. C. Production of Acetylene from Methane in a Plasma Jet. In Kinetics and Thermodynamics of Chemical Reactions in Low Temperature Plasma; Polak, L. S., Ed.; Nauka: Moscow, 1965; pp 72-99 (in Russian). Ibarra, J. V.; Palacioa, J. M.; de AndrC, A. M. Analysis of Coal and Char Ashes and Their Ability for Sulphur Retention. Fuel 1989, 68,861-867. Il'in, D. T.; Eremin, E. N. Pyrolysis of Pure Hydrocarbons and Gasoline Vapoura in a Hydrogen Plasma Generated in an Electric Arc. Zh. Prikl. Khim. 1965,38, 2786-2896 (in Russian). Kozlova, S. G.; Kukhto, V. A.; Lebedev, V. V.; Krukovskii, V. K. Investigation of High Temperature Transformation of Mineral and Organic Parts of Coals. Khim. Tvor. Topl. 1977,11,84-85 (in Russian). Kubanek, G.; Munz, R. J.; Gauvin, W. H. Heavy Oil Processing in Steam and Hydrogen Plasmas. Can. J. Chem. Eng. 1986, 64, 803-807. Lesinski, J.; Boulos, M. I. Thermodynamic and Transport Properties of Argon, Nitrogen and Oxygen a t Atmospheric Pressure over the Temperature Range from 300 to 20000 K Internal Report; Department of Chemical Engineering, University of Sherbrooke; December 1978. Mazumdar, B. K.; Chakrabartty, S. K.; Lahiri, A. Pyrolysis and Ita Relation to the Structural Parameters of Coal. Symposium on the Nature of Coal; Central Fuel Research Institute: Jealgora, India; 1959; paper 33, pp 253-260. Melamed, V. G.; Mukhtarova, T. A.; Polak, L. C.; Khait, Yu. L. A Method for Calculation Kinetics of Chemical Reactions in Plasma Jets (Example of Methane Conversion into Acetylene). In Kinetics and Thermodynamics of Chemical Reactions in Low Temperature Plasma; Polak, L. S., Ed.; Nauka: Moscow, 1965; pp 12-51 (in Russian). Menzies, M. A.; Silva, A. E.; Denis, J. M. Hydrocracking without Catalysis Upgrades Heavy Oil. Chem. Eng. 1981,88,46-47. Meubus, P. Metal Vapour Effecta on Chemical Reactions in an Argon Plasma. J. Electrochem. SOC.1975a, 122,298-305. Meubus, P. High Temperature Propane Cracking in an Argon Plasma with the Presence of Aluminum Vapour and Tungsten Particles. Can. J. Chem. Eng. 1975b, 53, 653-658. Moisyeyev, Yu. A.; Gafarova, N. A.; Popov, V. T. Pyrolysis of Hydrocarbons in a Hydrogen Plasma Jet in the Presence of Nickel Particles. Zzu. Akad. Nauk. Kaz. SSR., Ser. Khim. 1986,1,69-72 (in Russian). Nicholson, R.; Littlewood, K.; Plasma Pyrolysis of Coal. Nature 1972, 236, 397-400. Rahimi, P. M.; Fouda, S. A.; Kelly, J. F.; Malhotra, R.; McMillen, D. F. Characteristics of CANMET Coprocessing Distillates at Different Coal Concentrations. Fuel 1989,68, 424-429. Sundstrom, D. W.; DeMichiell, R. L. Quenching Process for High Temperature Chemical Reactions. Znd. Eng. Chem. Process Des. Dev. 1971,10, 114-122. Venugopalan, M.; Roychowdhwy, U. K.; Chan, K.; Pool, M. L. Plasma Chemistry of Fossil Fuels. In Plasma Chemistry ZI; Veprek, S., Venugopalan, M., Eds.; Topics in Current Chemistry 90; Springer-Verlag: Berlin, 1980; pp 1-57. Received for reuiew April 25, 1991 Revised manuscript received October 24, 1991 Accepted November 10, 1991