Direct Liquefaction of Coal Using Aerosol-Generated Ferric Sulfide

Department of Chemical Engineering, P.O. Box 6102, West Virginia University, ... The activity of aerosol-generated ferric sulfide based mixed-metal ca...
0 downloads 0 Views 124KB Size
312

Energy & Fuels 1998, 12, 312-319

Direct Liquefaction of Coal Using Aerosol-Generated Ferric Sulfide Based Mixed-Metal Catalysts R. K. Sharma, J. S. MacFadden, A. H. Stiller, and D. B. Dadyburjor* Department of Chemical Engineering, P.O. Box 6102, West Virginia University, Morgantown, West Virginia 26506-6102 Received July 25, 1997

The activity of aerosol-generated ferric sulfide based mixed-metal catalysts for direct coal liquefaction was studied at 400 °C and nominally 2000 psi hydrogen pressure. Aluminum, cobalt, copper, lead, silver, and tin were used in turn as the second metal. The typical fraction of the second metal was 10 atom % of total metal, although the concentration was varied in some cases. The catalysts were prepared in an aerosol reactor at 250 °C and 70 psi and were characterized in terms of their skeletal density, surface area, pyrrhotite/pyrite ratio, and X-ray diffraction. Of the catalysts tested, only those in which Al (and perhaps Pb) was used as the second metal cause an increase in conversion compared to the iron-alone catalyst. Selectivity to oil-range products is higher for catalysts containing Ag, Co, Cu, or Pb than for the iron-alone catalyst and is highest for the Fe-Pb-S catalyst. Hence, the Fe-Pb-S catalyst appears to be the one most suitable. The relative size of the ions of the second metal may be important for the performance of the catalyst. These aerosol-generated catalysts are slightly less active (in overall conversion) than the corresponding catalysts impregnated in situ in coal but are slightly more selective (to oilrange products).

Introduction Catalysts based on Al, Bi, Cd, Co, Cu, Fe, Mo, Ni, Pb, Sn, Zn, and other metals have been used for direct coal liquefaction (DCL).1-8 Among these catalysts, those based on iron are particularly desirable, since they are cheap and environmentally benign, and have relatively good activity for coal liquefaction.4-8 In particular, catalysts using ferric sulfide as a precursor have been shown to be especially advantageous.9 Ferric sulfide (Fe2S3) disproportionates into FeS2 (pyrite, PY), a non* To whom inquiries should be addressed. Voice: (304) 293-2111, ext 411. Fax: (304) 293-4139. E-mail: [email protected]. (1) Anderson, L. L.; Miin, T. C. Catalysis of coal conversion at mild temperatures. Fuel Process. Technol. 1986, 12, 165-174. (2) Besson, M.; Becaud, R.; Charcosset, H.; Burillo, V. C.; Oberson, M. Catalytic hydroliquefaction of coal: about the methodology in batch experiments. Fuel Process. Technol. 1986, 12, 91. (3) Weller, S. W. Catalysis and catalyst dispersion in coal liquefaction. Energy Fuels 1994, 8, 415-420. (4) Hager, G. T.; Compton, A. L.; Givens, E, N.; Derbyshire, F. J. The effect of promoter metal concentration on the catalytic activity of sulfated hematite for the liquefaction of a subbituminous coal. Prepr. Pap.sAm. Chem. Soc., Div. Fuel Chem. 1994, 39 (4), 1083-1087. (5) Pradhan, V. R.; Tierney, J. W.; Wender, I.; Huffmann, G. P. Catalysis in direct coal liquefaction by sulfated metal oxides. Energy Fuels 1991, 5, 497-507. (6) Tian, D.; Sharma, R. K.; Stiller, A. H.; Stinespring, C. D.; Dadyburjor, D. B. Direct liquefaction of coal using ferric-sulfide-based, mixed-metal catalysts containing magnesium and molybdenum. Fuel 1995, 75, 751-758. (7) Zhao, J.; Feng, Z.; Huggins, F. E.; Huffman, G. P. Binary iron oxide catalysts for direct coal liquefaction. Energy Fuels 1994, 8, 3845. (8) Sharma, R. K.; Stiller, A. H.; Dadyburjor, D. B. Effect of preparation conditions on the characterization and activity of aerosolgenerated ferric-sulfide-based catalysts for direct coal liquefaction. Energy Fuels 1996, 10, 757-65. (9) Dadyburjor, D. B.; Stewart, W. R.; Stiller, A. H.; Stinespring, C. D.; Wann, J.-P.; Zondlo, J. W. Disproportionated Ferric-Sulfide Catalysts for Coal Liquefaction. Energy Fuels 1994, 8, 19-24.

stoichiometric sulfide FeSx with x ≈ 1 (pyrrhotite, PH), and elemental sulfur. The juxtaposition of the PY and PH sublattices, and the resulting lattice strain, are thought to provide the improved catalytic performance of this material. The relative amounts of PY and PH and the composition of the nonstoichiometric PH (i.e., the value of x) have been shown9 to depend on the conditions of the disproportionation, viz., the temperature, time, and gas-phase composition. Currently, much of the research is focused on obtaining catalysts with small particle size that can be dispersed uniformly in the coal and thereby increase the activity. In one method of achieving small particle sizes, ferric sulfide based catalysts have been impregnated on coal.6 However, the disadvantage of the impregnation method is that it may not be cost efficient, especially for making large quantities of the catalyst.3 We have also used an aerosol technique8 to obtain fine particles of catalyst in the size range 3-20 nm. The aerosol technique also has an added advantage that a large amount of catalyst of known composition can be prepared under precisely controlled conditions.10,11 To alter further the activity and selectivity of the ferric sulfide based aerosol catalysts, we are investigating the effect of incorporating a second metal in the Fe-S lattice. According to the Hume-Rothery rules,12 materials with similar ionic sizes ((15%) may be (10) Pratsinis, S. E. An overview of material synthesis by aerosol processes. AIChE Symp. Ser. 1987, 85 (270), 57-68. (11) Gadalla, A. M.; Hsuan, F. Y. Preparation of fine hollow spherical NiFe2O4 powders. J. Mater. Res. 1990, 5, 12. (12) See, for example, the following. Askeland, D. R. The Science and Engineering of Materials, 2nd ed.; PWS-Kent Publishing: Boston, 1989; p 253.

S0887-0624(97)00121-7 CCC: $15.00 © 1998 American Chemical Society Published on Web 01/24/1998

Aerosol Ferric Sulfide Mixed-Metal Catalysts

substituted into one another’s lattice framework relatively easily. Owing to the different ionic sizes of various metals, the magnitude of the strain exerted by their substitution in the Fe-S lattice may be different, which, in turn, may affect the activity of the catalyst. The ionic radi of metals such as Co, Cu, Mg, Mo, Ni, or Sn are close to that of iron. Hence, it is reasonable to expect that these sulfides may be particularly suitable as additives. Some of these metals have DCL activities of their own, which could also be helpful. However, for multimetal catalysts to be economically viable, a low concentration of the second metal is desired.4-7 Hager et al.4 studied the effect of adding Ni, Co, W, and Mo on the activity of an iron-based catalyst for the liquefaction of Black Thunder coal and observed an increase in activity with Mo. Pradhan et al.5 also noted a beneficial effect on the activity of a sulfated hematite catalyst upon adding Mo or W. In both these cases, up to 10 wt % of the second metal was used, a relatively large fraction, considering the cost of these metals. Tian et al.,6 using smaller amounts of the second metals, observed an optimum concentration of Mo, above which the activity of the Fe-Mo-S catalyst decreased. Hence, the beneficial effect of Mo probably was not merely due to the good activity of Mo for coal liquefaction. The distribution of Mo on the catalyst may also be important; i.e., there may be an interaction between Fe and Mo. The addition of the second metal may also suppress the agglomeration and growth of catalyst particles; Zhao et al.7 observed a reduction in the agglomeration of ferrihydrite catalyst when Si or Al was introduced in the catalyst. The mixed-metal catalysts in all these studies were prepared by nonaerosol techniques. In this work is reported the characterization and DCL activity of aerosol-generated ferric sulfide based mixedmetal catalysts. Ag, Al, Co, Cu, Pb, or Sn were used in turn as the second metal. Some of the metals have ionic radii that are (15% of that of Fe, others do not. Most of these metals are relatively inexpensive, so we are not restricted to small fractions of the second metal.The catalysts were prepared at 250 °C and 70 psi pressure. The catalytic activity for coal liquefaction was measured at 400 °C and nominally 2000 psi H2 pressure. The products were analyzed in terms of asphaltene, oil, and gas fractions. The results are compared to those from iron-alone aerosol-generated catalysts and with those from multimetal nonaerosol catalysts. Experimental Method Coal. The coal used in this study was DECS-6, which is a high-volatile A bituminous coal from the Blind Canyon seam in Utah and was received from the Pennsylvania State University Coal Bank. The properties of the coal are given in Table 1. Not explicitly noted in Table 1 is the fact that the coal has a low iron content compared to other bituminous coals (0.3% compared with 2.8% for Illinois No. 6 coal, for example). This makes the DECS-6 coal particularly suitable for testing the activity of iron-based catalysts. Solutions. A series of solutions were made, containing ferric acetate and acetates of the other anions used as second metals. To prepare ferric acetate, 0.1 mole of hydrated ferric chloride was dissolved in 500 mL of water and approximately 100 mL of ammonium hydroxide was added until all the ferric ions were precipitated. The precipitate was repeatedly centrifuged and washed with water. The remaining solid was

Energy & Fuels, Vol. 12, No. 2, 1998 313 Table 1. Proximate and Ultimate Analysis of DECS-6 Coala Proximate Analysis water, % ash, % dry basis volatile matter, % dafb basis fixed carbon, % daf basis

1.8 6.3 49 51

Ultimate Analysis C, % H, % N, % S, % O, %c

81.9 6.3 1.5 0.9 9.4

a From Galbraith Laboratories. b daf ) dry, ash-free. c Obtained by difference.

then dissolved in 250 mL of acetic acid, with the resulting solution heated to 100 °C. Approximately 20 mL of ammonium hydroxide was added until the pH was 3.5. The solution was diluted to 0.1 M. The same procedure was followed for zinc acetate and silver acetate. Lead nitrate was used as the parent salt for lead acetate. Magnesium acetate was purchased. For tin, 0.1 mole of stannous chloride was dissolved in 500 mL of water. To this solution was added a solution containing 0.2 mole of ferric chloride. Ammonium hydroxide (200 mL) was added to form the precipitate, this time of stannic ions. The precipitate was then centrifuged and washed and subjected to the rest of the preparation protocol. Aerosol System. The mixed-metal catalysts were prepared by reacting appropriate mixtures of the precursor solutions batchwise with H2S in an aerosol system. This system for the preparation of the catalysts has been somewhat modified since the time it was last described;13 hence, a brief description is given below. The overall system consists of four parts: a pump, a spray nozzle, the reactor itself, and the vapor disposal system (see Figure 1). The pump is used to transfer the liquid reactants (the ferric salt and a salt of the second metal) to the spray nozzle in the reactor under high pressure. A Haskel CP-101 positivedisplacement pump is used, with a maximum displacement pressure of 10 000 lbs/cycle. The maximum flow rate is 1 mL/ cycle; we use 0.3 mL/cycle. The pump frequency can vary from 1 to 90 cycles/min (cpm); we find that around 80 cpm is optimum. The pump is made from 316 stainless steel (SS), and gasket seals for the plunger and check valves are of Teflon. The power is supplied by 1500 psi of nitrogen. Nitrogen is used in place of air to minimize contact between catalyst materials and oxygen. The nitrogen is fed to the pump through a 0.25-in. (inside diameter) steel-braided cable, while a polyethylene hose is used to connect the solution of the metal salts to the pump. A flexible 10 000-psi hydraulic hose connects the pump to the inlet of the spray nozzle. The nozzle is a Hago injector nozzle, also made from 316 SS. Spraying tips are purchased separately from Hago Spraying Systems. The pin lift pressure is 400 lbs. A needle pressure controller is attached to the lift pin, creating a back pressure. This prevents large drops from forming in the nozzle throat and clogging the line. The spray of the salt solution is an even conical shape, with a radius at the reactor bottom corresponding to the reactor radius, thus allowing a more uniform residence time and minimizing particle size. The nominal particle size is cited as 5 µm, but we expect that the actual droplet size is smaller because of the higher pressures used in our work. The nozzle is attached to the aerosol reactor through a bar of carbon steel clamped to a carbon-steel flange. Between the flange and the top of the reactor is a Graphoil (13) Dadyburjor, D. B.; Stiller, A. H.; Stinespring, C. D.; Chadha, A.; Tian, D.; Martin, S. B., Jr.; Agarwal, S. Use of an Aerosol Technique to Prepare Iron Sulfide Based Catalysts for Direct Coal Liquefaction. In Advanced Catalysts and Nanostructured Materials; Moser, W. R., Ed.; Academic: San Diego, 1996; p 563.

314 Energy & Fuels, Vol. 12, No. 2, 1998

Sharma et al.

Figure 1. Schematic of aerosol system for preparation of ferric sulfide based mixed-metal catalysts: BH, band heater; PG, 316SS 400-psi pressure gauge; V1, 1/4-in. 316-SS 6000-psi bellows valve; V2, 1/4-in. 316-SS relief valve; V3, 3/4-in., 316-SS hightemperature ball valve; V4, 3/4-in. 316-SS ball valve; V5, 1/4-in. 316-SS 6000-psi plug valve.

Figure 2. Detail of aerosol reactor. Symbols are as in Figure 1. gasket. Twelve bolts connect the flange to the reactor top. Before operation of the system, the bolts are tightened to 45 ft-lbs using a torque wrench. As shown in Figure 2, the reactor is a stainless steel tube 11 in. in diameter and 36 in. long, and is supported on a stand 9 in. above the floor. Five “Wrap-It-Heat” band heaters from Acra Heating, 1500 W each, are located on the curved external wall of the reactor. The temperature in the reactor is measured by a thermocouple extending radially to the center of the reactor. Inside the reactor bottom is a vertical channel

that serves as a collection trough when the solution is drained at the end of the batch treatment. The top flange of the reactor contains the injector nozzle as well as two vapor ports. One port is for the entry of H2S and/or N2 to the reactor. The other port, connected to a high-temperature/high-pressure ball valve, allows gases (acetic acid and/or its decomposition products, as well as N2, H2O, and H2S) to exit the reactor. The vapor disposal system consists of a collector/condenser, a scrubber, an indicator flask, and a backup scrubber in series, leading to a vent. The gases from the reactor first pass through a collection vessel and then to a condenser filled with stainless-steel wool. Aqueous acetic acid condenses and is drained to the collection vessel. The condenser is not heated. The remaining gases pass next through a sparger to a scrubbing vessel. This is made from 316 SS and contains a 50 wt % aqueous solution of sodium hydroxide to remove the H2S. The top of the scrubber is flanged and sealed with seven 1/ -in. bolts and a rubber gasket.The bottom of the scrubber 4 contains a 1/4-in. spigot to drain the contents. Gas leaves the scrubber through a 1/4-in. diameter stainless-steel tube to a 2-L flask containing 0.1 M ferric acetate solution as an indicator. The solution turns from cherry-red to black if H2S is present in the gas. A backup scrubber is located after the indicator flask. Also filled with the aqueous NaOH solution, this scrubber ensures that no H2S is released to the atmosphere even if the indicator solution changes color. The H2Sfree gas is vented to the atmosphere. Catalyst Preparation. The procedure was designed to prevent (or at least to minimize) the contact between the catalyst and oxygen. Before each batch, the aerosol reactor was purged by pressurizing to 50 psi with N2, evacuating, and repeating the procedure five times. The reactor was then heated to 250 °C using the heating bands, then pressurized to 70 psi using a N2/H2S ratio of 6. The feed solution to the pump was a mixture of ferric acetate and the acetate of the second metal, the relative proportion of each depending on the desired composition of the catalyst. The typical atom fraction (fM, M/(Fe + M)) of the second metal was 0.1, although 0.5 was also used. The precursor solution was then sprayed

Aerosol Ferric Sulfide Mixed-Metal Catalysts through the nozzle into the reactor at high pressure. The H2S reacts with acetates at the surface of the aerosol droplets to form acetic acid and solid metal sulfides. Simultaneously, water is evaporated from the droplets. These two competing processes govern the size and thickness of the resulting solid material, as has been discussed elsewhere.8,11,13 The temperature and pressure in the reactor were controlled by adjusting the backpressure valve and by varying the liquid flow rate. Owing to a continuous loss of H2S to the exit gases, additional H2S was added to the reactor intermittently. The combination of inlet flow rate, water condensation, and effluent flow rate maintained the pressure reasonably constant within the reactor. During the batch run, the catalyst particles accumulated at the bottom of the reactor. The ferric sulfide disproportionated into pyrite and pyrrhotite. At the end of the run, the reactor was cooled and purged with N2. Approximately 2 L of deionized, deoxygenated water was sprayed into the reactor to slurry the product. The catalyst-water slurry was flushed from the reactor bottom into a collection vessel, using N2. At this point, the collection vessel was sealed off, and the reactor was opened and washed thoroughly with water. The collection vessel was placed in a nitrogen-filled glovebox, and the catalyst in the slurry was allowed to settle. At this point, most of the supernatant water was suctioned off. The last 150 mL of water was removed by mild heating under vacuum to recover the catalyst. The dry catalyst was doublesealed under vacuum in polyethylene bags while inside the nitrogen glovebox. Catalyst Characterization. The catalysts were characterized in terms of their density, surface area, pyrrhotite/pyrite ratio (PH/PY), and X-ray diffraction (XRD). During these measurements, catalyst exposure to air could not be prevented, but this is not expected to alter the bulk measurements significantly. The density and surface area were measured using a He pycnometer and a BET surface-area analyzer, both from Micromeritics. The amounts of PH and PY were measured in terms of hydrochloric acid solubility and nitric acid solubility, respectively, in that order. The solutions from the acid dissolutions were analyzed by atomic-absorption spectroscopy (AA) to obtain the amounts of iron present as PH and as PY. These assignments of acid solubilities have been confirmed by Auger electron spectroscopy. The XRD patterns were obtained without any background subtraction, with peak identification using d-spacing values. Liquefaction Procedure. A stainless-steel tubing-bomb reactor with a volume of 27 mL was used for the liquefaction. The reactor was charged with 3 g of coal, 4.7 g of phenanthrene, and 0.1 mL of CS2. Phenanthrene was used as solvent because of its poor ability to donate hydrogen; this property is useful in determining the intrinsic activity of the catalyst.6 In the catalytic runs, 0.05 g of catalyst was added to the reaction mixture. This corresponds to a loading of 1.67% based on dry, ash-free (daf) coal. The reactor was purged and pressurized with hydrogen to 1000 psi at room temperature, i.e., cold. The reactor was heated in a fluidized sand bath that was preheated to 400 °C before the run. Calculations indicate that the pressure in the reactor under liquefaction conditions is approximately 2250 psi. At the end of the run, which lasted 30 min, the reactor was quenched in water. The gaseous products were collected in an evacuated sampling flask of known volume and were analyzed by gas chromatography. The amount of gas was evaluated on a hydrogen-free, ethaneequivalent basis; i.e., the detector-response factors for all the components were considered to be the same as that for ethane, since ethane was a median component of the gaseous product. The solid and liquid products in the reactor were washed and extracted with tetrahydrofuran (THF) for 24 h. The THFinsoluble (TI) material was separated by filtration. The conversion of coal was calculated from the amount of TI. After

Energy & Fuels, Vol. 12, No. 2, 1998 315 Table 2. Characterization of Aerosol-Generated Mixed-Metal Catalysts second metal

fM M/ (M+Fe)

surface area,a m2/g

skeletal density,b g/mL

total Fe,c mg/g

PH/PY

none Ag Al

0.0 0.1 0.01 0.1 0.5 0.1 0.26 0.1 0.4 0.1 0.1 0.5 0.1 0.5

3.1 6.2 2.5 2.2 5.3 2.8 3.3 13.7 23.1 3.8 2.1 4.3 15.5 23.0

4.0 4.4 3.0 3.7 2.8 2.9 2.8 3.9 3.9 3.5 3.5 6.0 4.3 4.8

367.5 462.0 319.4 233.6 201.6 248.6 205.3 d d 326.5 215.0 214.0 504.3 366.3

1.0 0.1 1.3 0.3 0.1 1.1 1.8 d d 0.4 1.0 0.1 0.1 0.1

Co Cu Mg Pb Sn a

N2 adsorption BET.

b

He pycnometry. c AA.

d

Not available.

the removal of THF by rotary evaporation, the THF solubles were extracted with hexane for 2 h. The resulting mixture was separated into hexane-insoluble (HI) and hexane-soluble fractions by filtration. The HI fraction represents asphaltenes and preasphaltenes. The conversion (X) and the yield of asphaltenes and preasphaltenes (A) were calculated as follows:

X ) (Fm - TI)/Fdaf

(1)

A ) HI/Fdaf

(2)

Here, Fm and Fdaf represent the amount of feed on a moisturefree basis and a daf basis, respectively. With the gas yield (G) determined independently from the gas analysis, the oil yield (O) was obtained by difference:

O)X-A-G

(3)

In repeated experiments carried out in our laboratory over an extended period of time, we have found that results are consistent within 2 percentage points in conversions and asphaltene yields. In almost all the experiments performed in this work, runs were carried out in duplicate simultaneously, and the results of the two runs agree to within 1 percentage point.

Results and Discussion Catalyst Characterization. Table 2 shows the surface area, density, total amount of iron, and the PH/ PY ratio for the fresh catalysts used in the present work.The surface areas of the catalysts are measured using the BET equation. The values shown in Table 2 are low, between 2 and 23 m2/g. However, transmission electron micrographs presented earlier8 show that typically the catalyst particles are 3-20 nm in size, corresponding to a much higher surface area. Hence, it is reasonable to believe that the surface areas, and other properties of Table 2, are essentially those of clusters or aggregates of fine particles. (It is tempting to speculate that these clusters disintegrate in the liquefaction reactor, resulting in an increase in the surface area under the liquefaction conditions. However, this cannot be verified; the spent catalyst could not be analyzed, since it was mixed with the unreacted coal and coal residue.) Among the multimetal catalysts, those containing Cu or Sn have surface areas that are

316 Energy & Fuels, Vol. 12, No. 2, 1998

Sharma et al.

Figure 3. X-ray diffraction results of aerosol-generated ferric sulfide based mixed-metal catalysts. Peaks of pyrite (PY), elemental sulfur (S), greigite (G), and monoclinic pyrrhotite (PH) are indicated.

much higher than the corresponding values for the other catalysts. Interestingly, for all the catalysts, the surface area increases as the concentration of the second metal increases. The addition of a second metal may decrease the rate of agglomeration of the catalyst particles during catalyst preparation, analogous to the agglomerationrate decrease observed by Zhao et al.7 during the liquefaction run. The density measured in Table 2 is the so-called skeletal density of the individual particle, not the bulk density of the powder. The values vary between 2.8 and 6 g/mL.The variations in skeletal density are consistent with differences in the densities of various metals. For example, metals such as Ag, Pb, and Sn are relatively dense so that the catalysts containing these metals have higher densities compared with those containing the other metals. Interestingly, the density of the ironalone catalyst is lower than that of ferric sulfide (4.3 g/mL) or pyrite (5 g/mL). This suggests that the catalyst probably contains impurities such as elemental sulfur. The low densities could also be due to the presence of large, hollow catalyst particles, as postulated by Gadalla and Hsuan.11

The mixed-metal catalysts contain between 202 and 504 mg iron/g catalyst. The iron content decreases as fM, the fraction of the second metal in the catalyst, increases. This is expected, since the amount of iron replaced by the second metal should increase with an increase in the concentration of the second metal. The iron content of Table 2 is below the stoichiometric amounts of iron in ferric sulfide (538 mg/g), pyrrhotite (636 mg/g, for x ) 1), and, in most cases, even pyrite (467 mg/g). This indicates that the catalysts probably contain significant amounts of pyrite and sulfur as well as other materials. This is confirmed by the XRD results presented later. The pyrrhotite/pyrite (PH/PY) ratios of the catalysts are between 0.1 and 1.8, consistent with the disproportionation of ferric sulfide. The XRD results are presented in Figure 3. The results are shown without any background subtraction. A comparison of the results for mixed-metal catalysts with those for iron-alone catalyst (also shown in the figure) indicates that the XRD peaks mostly correspond to pyrite (FeS2), with a few S peaks and even a few greigite (Fe3S4) peaks. The existence of this typically unstable compound in the products from an aerosol

Aerosol Ferric Sulfide Mixed-Metal Catalysts

Energy & Fuels, Vol. 12, No. 2, 1998 317 Table 3. Ionic Radii of Different Metals19,a metal

R

metal

R

Ag Al Co Cu

1.2 0.69 0.97 0.97

Mg Ni Pb Sn

0.89 0.93 1.62 1.1

a R is the ratio of ionic radius of metal to that of Fe. All ions considered have +2 charge except Al with +3.

Figure 4. Comparison of the performance of mixed-metal catalysts. Liquefaction conditions are the following: 400 °C, 30 min, 1000 psi H2 (cold), phenanthrene solvent, 1.67% catalyst loading, air-isolated catalyst, fM ) 0.1.

reactor has been noted earlier.13 The peak area corresponding to monoclinic pyrrhotite (Fe7S8) is small, even though the PH/PY ratios for the catalysts are high. This suggests that the pyrrhotite structures are probably too small to show a clear crystal pattern, as noted previously.8 The presence of sulfur in these catalysts is consistent with their low skeletal densities. As noted earlier, the sulfur is formed in the aerosol reactor during the disproportionation of ferric sulfide. Finally, the mixed-metal catalysts contain additional peaks, i.e., peaks not present in the Fe-S catalyst. This indicates the presence of additional phases in these catalysts. The additional phases do not correspond to any of the pure sulfides of the second metals and, hence, may consist of some complex alloy structures. In the case of the FeCu-S aerosol catalyst, the additional phase has been identified8 as chalcopyrite (CuFeS2). Catalytic Activity. A. Conversion and Oil Yield. As mentioned before, the liquefaction runs were performed at 400 °C using phenanthrene as the solvent, and the catalyst loading is 1.67%, based on daf coal. Figure 4 shows the activity of mixed-metal catalysts in which the fraction of the second metal (fM) is kept constant at 0.1. The results from the thermal-base case and from the iron-alone catalyst are also shown for comparison. The asphaltene yields are not explicitly shown in Figure 4 but can be determined by using eq 3. Each error bar of Figure 4 indicates the difference between the mean value and the extreme value of the corresponding parameter. In most cases, the error bars are below 1 percentage point, as mentioned earlier. From Figure 4, the conversion is low in the absence of catalyst. This is as expected, since phenanthrene was used as the solvent. According to McMillen et al.,14 a solvent stabilizes the coal free-radicals by donating hydrogen, and promotes the cracking of coal molecules. Although phenanthrene may enhance the cracking reactions of coal or act as a hydrogen-shuttler, the hydrogen-donor properties of phenanthrene are weak.15 Thus, unlike tetralin, which can supply almost all the (14) McMillen, D. F.; Malhotra, R.; Tse, D. S. Cleavage of benzylaromatics and their relevance to coal conversion. Prep. Pap.sAm. Chem. Soc., Div. Fuel Chem. 1991, 36, 498-503. (15) Whitehurst, D. D.; Mitchell, T. D.; Farcasiu, M. Coal Liquefaction; Academic Press: New York, 1980.

hydrogen necessary for the liquefaction of coal, the contribution of phenanthrene to hydrogen donation is small. As noted earlier, the poor hydrogen-donor properties of phenanthrene make it particularly suitable for measuring the intrinsic activity of the catalyst. In the presence of the Fe-S catalyst, the conversion increases. Adding a second metal changes the conversion further, but not always positively. The highest increase is approximately 3 percentage points, small but significant. This value is observed with Al as the second metal. A smaller increase, but probably still significant, is observed for the case of the Fe-Pb-S catalyst. Clearly, the addition of any one of these two second metals is beneficial to the activity of the Fe-S catalyst. With other metals, the conversions are up to 8 percentage points lower than with the iron-alone catalyst, though still significantly higher than that in the absence of a catalyst. This indicates that the addition of the second metal alters the catalyst activity. Following the Hume-Rothery formulation, the beneficial effect of Al or Pb may be related to the ionic size of these metals. Table 3 shows the ratios of ionic radii of the various metals with respect to iron. Except for Al and Pb, the ionic radii of the metals are within 20% (to 2 significant figures) of that of iron. In the case of Al and Pb, the ratios of ionic radii are far from unity. This would seem to imply that lattice substitution would be difficult. But recall that the Fe-S lattice is probably strained anyway, owing to the juxtaposition of PY and PH. Hence, perhaps those metals are most beneficial for which the differences in ionic radii are relatively large. The yields of oil and gas are also presented in Figure 4. With most catalysts, the gas yields are not statistically different and are around 6%. Oil yields are significantly different (and higher) for the mixed-metal catalysts relative to the Fe-S catalyst, with the exception of the Fe-Sn-S catalyst, discussed later. The highest yield is obtained using the Fe-Pb-S catalyst, 7 percentage points higher than with the iron-alone catalyst. Since the Fe-Pb-S catalyst has only a slightly higher activity than Fe-S, the improved yield arises not only from the increased activity but also from the net increased conversion of asphaltenes. The use of Co increases the oil yield but by a smaller amount, about 5 percentage points. However, there is only a small improvement in the yield when Al or Cu is used as the second metal. Since the addition of Al increases the overall conversion, this additive enhances mainly the coal-to-asphaltene reaction. It is interesting to note that the addition of Sn does not improve the activity of the Fe-S catalyst. Besson et al.2 used a sulfided tin oxide catalyst for the liquefaction of a high-volatile bituminous coal and reported a similarly low activity in terms of THF solubility. However, Weller3 used a tin chloride catalyst to obtain a high

318 Energy & Fuels, Vol. 12, No. 2, 1998

Sharma et al.

Figure 6. Comparison of the selectivities of various mixedmetal catalysts. Liquefaction conditions are as in Figure 4.

Figure 5. Effect of fM on conversion and oil, and gas yields. Liquefaction conditions are as in Figure 4 except that fM is as shown.

conversion (in terms of benzene solubility) of Rock Springs coal at 460 °C. The conversion was attributed to the presence of chloride during the liquefaction. To investigate whether the low activity of the Fe-Sn-S catalyst in the present study was due to the absence of chloride during liquefaction, additional runs were made with the Fe-Sn-S catalyst. Here ammonium chloride was added to the reaction mixture, with a catalyst/ ammonium-chloride ratio of 1. The conversion increased by 2 percentage points and the oil yield increased by 5 percentage points, relative to the values in the absence of chloride. Hence, the presence of chloride may have some effect on the activity of the tin chloride catalyst. It should be noted that the reaction conditions in this study are different from those used by Weller, in particular, the type of coal and the liquefaction temperature. Further, in this study, the concentration of tin in the catalyst is low. B. Effect of Concentration of Second Metal. Figure 5 shows the effect of varying the concentration of the second metal (fM) on conversion and oil yield. The error bars are too small to be visible. As the value of fM increases, the conversion and yield decrease, except in the case of Al and Pb. As noted earlier, the iron content of the catalysts also decreases with an increase in the fM value. Accordingly, the decrease in activity at high fM values may be due to the decrease in the iron content of the catalysts. In the case of Al as the second metal, there is a slight maximum in conversion at fAl ) 0.1. With the Fe-Pb-S catalyst, maxima are observed in both conversion and oil yield at fPb ) 0.1. Thus, there may be an optimum distribution of Al or Pb in the catalyst at fM ) 0.1, after which the activity decreases. A similar optimum was observed in the case of a FeMo-S catalyst, not made in the aerosol reactor but by an in situ impregnation procedure.6 Similarly, in their studies on pyrene hydrogenation over a NiMo catalyst, Gardner et al.16 reported an optimum in the concentration of Mo at fM ) 0.1. The presence of a maximum in conversion with loading of the second metal catalyst (16) Gardner, T. J.; McLaughlin, L. I.; Lott, S. E.; Oelfke, J. B. Prep. Pap.sAm. Chem. Soc., Div. Fuel Chem. 1994, 39, 1078.

indicates that the effect of the second metal is not merely due to the activity of Al, Pb, or Mo for coal liquefaction but that there exists an interaction of Fe with these metals. C. Selectivity to Oil. Here, we define selectivity to oil as

So ) O/X

(4)

This is a useful measure of the performance of the catalyst. Selectivities of various catalysts to the oil product are reported in Figure 6 as a function of the corresponding conversion. In most cases, the selectivity increases with conversion, consistent with oil being a secondary product of the reaction. There seem to be two trends in the selectivity/conversion relationship, as evidenced by the two straight lines. With Fe-Cu-S, Fe-Co-S, Fe-Pb-S, and Fe-Ag-S, the increase in selectivity with conversion is relatively large. With FeAl-S and Fe-Sn-S, the increase is small and may be of the order of the experimental errors for Fe-Sn-S. The single point for the Fe(-alone)-S catalyst can also be considered to lie on the line of lower slope. However, in all the cases, the selectivity decreases as the concentration of the second metal in the catalyst, fM, is increased from 0.1 to 0.5. The magnitudes of the selectivity changes do not correlate with whether the ionic ratio is greater than or less than unity or with whether the ratio lies within the range 0.8-1.2. D. Effect of Exposure to Air. Previous work6 with iron-molybdenum catalysts, not made in an aerosol reactor but by an in situ impregnation procedure, indicates that the conversion and oil yield deteriorate, at least to some extent, when the catalyst is exposed to air over a period of time. To study the effect of exposure to air on the activity of the aerosol catalyst, runs were made with an earlier sample of Fe-Al-S catalyst (with fAl ) 0.01), which was purposely exposed to air over an extended period of time, approximately 6 months. This was done by allowing air into the sample vial before storage. It was found that the conversion of coal decreases from 78% (for the air-isolated Fe-Al-S aerosol catalyst) to 73% (for the air-exposed catalyst). Further, the oil yield was approximately 2 percentage points higher with the exposed catalyst compared with that with the air-isolated catalyst. This indicates, first, that the surface of the catalyst may be slowly oxidizing in the presence of air and, second, that the loss in activity due to the exposure of the catalyst to air manifests itself mainly in the yield of asphaltenes, with

Aerosol Ferric Sulfide Mixed-Metal Catalysts

Energy & Fuels, Vol. 12, No. 2, 1998 319

liquefaction of Blind Canyon coal (DECS-6) using an impregnated FeOOH catalyst. The value is comparable to the conversions observed in this study. An exact comparison of the present results with others in the literature is difficult because of differences in the types of coal used and in the liquefaction conditions, including the nature of the solvent. However, the conversions in this study are comparable to those reported in the literature using tetralin as solvent. Thus, aerosol-generated catalysts may make it economically advantageous to use a cheaper poor-hydrogendonor solvent. Conclusions

Figure 7. Comparison of the performances of aerosol mixedmetal catalysts (A) and in situ impregnated mixed-metal catalysts (I). Liquefaction conditions are as in Figure 4.

little or no effect on the oil yield. In other words, the exposure to air affects mainly the catalytic activity for the coal-to-asphaltenes reaction. E. Comparison with Other Ferric Sulfide Based Catalysts. It is interesting to compare the performance of the aerosol catalysts with those of other iron-based catalysts in the literature. Figure 7 shows the activities of Fe(-alone)-S and Fe-Mo-S catalysts impregnated in situ on the coal. For simplicity, the only aerosol catalysts shown in Figure 7 are Fe(-alone)-S and the two shown earlier to have the greatest activity enhancement, viz., Fe-Al-S and Fe-Pb-S. The impregnated Fe-S catalyst has been shown to be a superior ironbased DCL catalyst.17 The oil yields with aerosol catalysts are generally higher compared with those with impregnated catalysts, although the aerosol catalysts are slightly less active. This indicates that the aerosol catalysts are more selective to the oil product. The aerosol Fe-Pb-S catalyst seems to be particularly selective, since the oil yield is highest with this catalyst. Our previous work8 shows that the selectivity differences between the aerosol and impregnated catalysts increase even further as the liquefaction temperature is increased; i.e., the aerosol catalysts are even more selective at high temperatures than the impregnated catalysts. Among the other studies reported in the literature, Cugini et al.18 reported a conversion of 85% for the (17) Stohl, F. V.; Diegert, K. V.; Goodnow, D. C. Evaluation of West Virginia University’s iron catalyst impregnated on coal. Proceedings, Coal Liquefaction and Gas Conversion Contractors Review Conference; U.S. Department of Energy, Pittsburgh Energy Technology Center: Pittsburgh, 1995; pp 679-686.

The aerosol technique can be used to generate ferric sulfide based mixed-metal catalysts. The second metals used here are, in general, relatively cheap. The aerosol materials generally have lower densities compared with the typical iron sulfide. This is due to the presence of elemental sulfur from disproportionation, to the presence of the second metal, and to the hollow nature of the particles. The presence of the second metal affects the activity and yields of the catalysts. The effect is different for different metals and also varies with the fraction of the second metal (fM). The optimum value of fM appears to be around 0.1 for the catalysts tested. The use of Al (and probably Pb) as the second metal enhances the activity of the catalyst, whereas the use of Pb or Co improves the yield of the oil to a large extent.The selectivity to oil-range products, i.e., yield/conversion, is highest with the Fe-Pb-S catalyst. This appears to be the best catalyst among the aerosol-generated materials tested. Prolonged exposure of these catalysts to air is somewhat detrimental to their activity. The aerosol Fe-S catalysts are more selective to the oil-range products compared with catalysts impregnated in situ on coal, although the overall activity of the latter catalyst is larger. The selectivity increases further by incorporating Pb or Co in the aerosol-generated catalyst. Acknowledgment. This work was conducted under U.S. Department of Energy Contract No. DE-FC2290PC90029 under the Cooperative Agreement to the Consortium for Fossil Fuel Liquefaction Science. The authors are grateful to Professor J. Renton for the XRD results. EF970121T (18) Cugini, A. V.; Krastman, D.; Lett, R. G.; Balsone, V. D. Development of a dispersed iron catalyst for first-stage liquefaction. Catal. Today 1994, 19, 395-407. (19) Handbook of Chemistry and Physics, 60th ed.; CRC Press: Boca Raton, FL; 1980.