Id.Eng. Chem. Process Des. Dev. lQ81,20, 91-94
Coal Hydrogenation and Hydrocracking Using a Metal Chloride-Gaseous HCI Catalyst System F. P. MCCandiess,’ John Jay Watermhn, and David Lee Sire Department of Chemical Engineering, Montana State Unhwshy, Bozeman, Montana 59717
A novel catalyst system consisting of a metal chloride impregnated on the s o l i coal in a 5% HCi-H2 atmosphere was investigated for the hydroiiquefaction of a Montana subbituminous coal. A preliminary screening of various metal chlorides was prried out in a semi-batch reactor. The best catalysts were then tested in a continuous short residence time tubular reactor. When 5% HCI was included in the reacting gas yields of oils increased from about two to over five times the oil yields when no HCI was present, depending on the specific metal chloride catalyst used. Of the catalysts tested in the continuous system, activlty was in the order CoCI, > SnCi, > CuCi, > NiCl, > ZnCI, > FeCI,. Preliminary data indicate that essentially 100% of nonvolatile chloride catalysts such as NICip remains in the unreacted residue in the chlorkle form and hence wid be easily recovered by simple water leaching.
Introduction and Background Previous research in this laboratory showed that a catalyst system of NiC12 impregnated on alumina or silica-alumina in a hydrogen atmosphere containing gaseous HC1 is an extremely active hydiocracking/hydrotreating catalyst for nitrogen removal from high nitrogen content gas oils. Under identical operating conditions the acid catalyst system was more active than the most active conventional catalysts such as nickel tungsten sulfide or cobalt molybdate (McCandless and Berg, 1970). Thus, it seemed appropriate to investigate this catalyst system for the hydrogenation/hydrocracking of coal. Experimental Section Coal used in the study was Montana subbituminous from the Rosebud bed obtained from the Western Energy Company from their Colstrip Mine. This coal was pulverized in a ball mill and screened to -60 mesh. The total -60 mesh product was used as feed rather than any fraction of it. Coal samples were pre-dried for about 48 h at 100 OC in a forced convection oven. Catalytic coal was prepared by adding coal to an aqueous solution of the metal chloride followed by drying. The reactor used for the preliminary screening of metal chloride catalysts is shown in Figure 1. The reactor tube was a 0.6m length of schedule 80,2.5cm 0.d. Inconel Alloy 600 pipe. The inlet section was fitted with a high-pressure type 316 stainless steel tee and appropriate Swagelock tubing fittings. The inside upper portion of the reactor was machined slightly to accommodate a porous alundum extraction thimble, which contained the coal charge. Since the remaining ridge supports the thimble, its height in the reactor was determined by the length of pipe machined. A thermowell made of 6.35mm Inconel Alloy 600 tubing extended from the reactor outlet to the bottom of the thimble. Temperature was detected by a chromel-alumel thermocouple and was recorded on a chart recorder. The reactor temperature was regulated by a 110-Vpowerstat controlling a 5-A Hoskins furnace. Pressure within the reactor was maintained by a Grove “Mighty Mite” back pressure regulator. To make a test, a 5-g sample of coal which had previously been impregnated with catalyst was mixed with 15 g of 40 mesh Ottawa sand. The sand prevented coal caking, which would cause diffusional resistance. The mixture was blended and then poured into a 25 mm 0.d. X 70 mm high coarse alundum extractable thimble. After the thimble was weighed, it was inserted into the reactor 0198-4305/81/1120-0091$01 .OO/O
and the run was carried out, passing the H2-HC1 gas mixture past the coal charge for 0.5 h. All runs were at 6.89 MPa and 450 “C. For this part of the study conversion to toluene-soluble material (on a moisture and ash free basis) was used as a basis of catalyst activity comparison. A diagram of the continuous system is shown in Figure 2. Pulverized coal impregnated with a metal chloride catalyst was charged batchwise to the hopper of the star feeder where it was continuously fed by a rotating star wheel to the coiled tube reactor. In the feeder, the coal is entrained in the hydrogen stream which propels the coal through the reactor. The solid-liquidproduct was collected in a high-pressure bomb modified for this use while the gas flow is depressurized, scrubbed to remove acid gasses, metered, and vented to the atmosphere. Since HCl was mixed with the hydrogen for some of the runs, all internal parts were made of Inconel whenever possible for resistance to corrosion. The details of the star feeder are also shown in Figure 2. Coal from the hopper falls into one of the holes in the rotating shaft, and as it turns BO”,the coal is carried into the reactor along with the hydrogen which flows through the shaft. The feeder is constructed on Inconel and rated at 34.5 MPa. The diagram also shows the details of the tubular reactor and heater. The reactor itself is a length of 6.35.” Incone1 617 tubing. It is coiled around a 82.5 mm X 12.7 mm wall stainless steel pipe. The heavy walled pipe was used to provide an approximately constant temperature heat source for the reactor. Process heat was supplied to the pipe by strip heaters that fit inside it. Power was supplied from variable transformers. The products were collected in a Parr Series 4000 pressure reaction apparatus modified to accept the 6.35mm tubing fittings. This vessel was rated to 68.9 MPa. Pressure was controlled using a Grove “Mighty Mite” back-pressure regulator. Experimental Results. Preliminary Screening The results of a preliminary screening of various metal chlorides are shown in Table I. As can be seen, the most active catalysts tested were the two tin chlorides and nickelous chloride. Significantly,zinc chloride is considerably less active than these and it is also less active than the copper chlorides, cobalt chloride, and iron chloride. It is also interesting to note that the alkaline earth chlorides CaC12,MgC12,SrC12,and BaC12,exhibited some catalytic 0 1980 American Chemical Society
92
Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 1, 1981 -vlTp2GE* PUOM -~~ -~ *-_-&__-
'
k€
-ALq--;
Of4 r -p
TO
PRDWCT CLILLSCTION
VENT
ciusric
SCRUEEER
( a ) Schematic Diogrom of Continuous System.
Figure 1. Schematic diagram of semi-batch reactor for preliminary catalyst screening.
Table I. Coal Conversion with Metal Chloride Catalyst Preliminary Screening in Semi-Batch Reactora av conv, %, dry ash free basis
catalyst SnC1, NiCI, SnC1, CUCl,
coc1,
CUCl FeCl, ZnC1, CrCl, FeCl, CaCl, MgC1, MnCl, SrCl, BaC1, noncatalytic ( 5 %HC1-95% H, gas) LiCl NaCl noncatalytic (H, only) KC1 noncatalytic (N, only)
-
--cONT.INER INSIJl ATION
67.5 67.3 62.5 57.4 55.6 54.4 54.2 51.7 48.3 47.7 44.4 44.3 41.7 41.6 38.6 38.4
-STRIP
( b ) Coal Stor Feeder Defoils.
HEAR-RS
( C ) Reocfor ond Heater Defoii.
Figure 2. Schematic diagram of continuous system.
38.1 37.7 34.0 33.9 33.8
(I Conditions for all tests: 6.89 MPa, 450 "C using 5% HCl in H, except for one run each passing N, or H, only through bed; catalyst concentration on coal was 3 wt 5% metal in all cases; run time was 3 0 min.
activity while the alkali chlorides LiC1, NaC1, and KC1, appeared to act as poisons. Since the economic use of expensive metal chlorides as coal hydrogenation catalysts would require essentially 100% recovery of the metal, a brief investigation was made of catalyst loss using the most active catalysts. For these tests an amount of catalysts equivalent to that normally impregnated on the coal was mixed with Ottawa sand and the mixture was placed in the reactor and subjected to normal reaction conditions. Since the mixture was nonreactive, any loss in weight was catalyst loss. This was an approximation of actual catalyst loss due to the volatility of the metal chloride since the catalyst would be adsorbed on the coal. The results of these tests are shown in Figure 3. As can be seen, essentially all of the SnC14is lost at
U
\n
Table 11. Material Balance o n Nickel during Reaction wt ash and unreacted coal remaining ash and unreacted coal after reaction 32.3 14.4 11.1 24.4 39.4 13.8 14.7 48.0 37.2 13.5 % Ni o n
run no,
% Ni on coal
g of Ni on 100 g of coal
1 2 3
4.64 2.70 5.45 7.09 5.03
4.64 2.70 5.45 7.09 5.03
4 5
g of Ni remaining ash and unreacted Ni consumed coal in reaction 4.65 0 2.71 0 5.44 0 7.06 0 5.02 0
Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 1 , 1981 93 Table 111. The Effect of Metal Chloride Catalysts on Individual Coal Conversion conversion catalvst
oil
ZnC1, SnC1, CUCl, NiCl, coc1, FeC1, noncatalytic unreacted
16.9 14.3 13.0 14.0 9.9 5.7 3.8 0.4
asphal- asphaltene to1 6.1 2.9 2.8 1.5 0.9 1.9 0.3 0.2
1.0 0.0 0.7 0.0 0.0 0.7 0.1 0.1
total 24.0 17.2 16.5 15.5 10.8 8.3 4.2 0.7
Table IV. The Effect of Hydrogen Chloride on Coal Conversion total conv. total conv. catalyst H, H,-HC1 COCl, SnC1, CUCl, NiC1, ZnC1, FeCl, noncatalytic
10.8 17.2 16.5 15.5 24.2 8.3 4.2
44.2 38.3 37.8 37.6 34.3 34.2 25.0
Table V. The Effect of Hydrogen Chloride on Individual Coal Conversion asphal- asphalcatalyst oil tene to1 total ~~
COCl, SnC1, CUCl, NiC1, ZnC1, FeCl, noncatalytic
29.7 33.3 33.2 34.2 31.9 29.9 20.6
12.6 4.1 3.7 2.9 2.4 4.3 3.5
1.9 0.9 0.9 0.5 0.0 0.0 0.9
44.2 38.3 37.8 37.6 34.3 34.2 25.0
manner, and analyzing the unreacted coal residue for nickel by dimethylglyoxime precipitation methods. The results of these tests are shown in Table 11. As can be seen, essentially 100% of the nickel remained on the unreacted coal. Analysis for chloride by silver chloride precipitation indicated that is was all in the chloride form. Continuous Wactor Studies Standard reactor conditions, using the continuous reactor, were 6.89 MPa and 425 "C with an H2rate of about 850 L/h (STP) and a coal feed rate of about 70 g/h. The reactor was 1.8 m long. Under these conditions coal residence time in the reactor was about 4 s. Three percent metal chloride was impregnated on the coal for all tests. Conversion was determined (on a moisture and ash free basis) by extracting the product that accumulated in the product collection vessel with cyclohexane, toluene, and pyridine to yield oils, asphaltenes, and asphaltols, respectively. To date conversion to gaseous producta has not been determined nor have the oil, asphaltene, and asphaltol products been further characterized. The results of these tests are shown in Tables 111, IV, and V. Table I11 shows the effect of using different metal chloride catalysts on the conversion when using H2 only. As can be seen, conversion varied from 4.2% for noncatalytic coal to 24.0% when ZnC12 was used. SnC12,CuCl,, NiCl,, CoCl2, and FeC12 gave intermediate conversions. Table IV shows the effect of adding 5% HC1 to the H2 feed gas. As can be seen, total conversion varied from 25.0% for noncatalytic coal to 44.2% for coal impregnated with 3% CoC1,. In this case, activity was in the order CoC4 > SnC1, > CuC1, > NiCl, > ZnC12> FeC12with little difference in the activity for CoCl,, SnCl,, CuC12, and NiC1,.
Table VI. Effect of HCl on Conversion to Oils (conv. to oil-HCl)/ (conv. to oil without HCl) catalyst ZnC1, SnCl, CUCl, NiCl, COCl, FeC1, noncatalytic
1.89 2.33 2.55 2.44 3.00 5.25 5.42
Table V shows the conversion to oils, asphaltenes and aspha,ltols when HC1 is included in the feed hydrogen. Significantly, most of the increased conversion appears in the oil fraction. Discussion The conditions for the semi-batch preliminary testa were arbitrarily set at 6.89 MPa and 450 "C with a run time of 0.5 h. Under these conditions conversion on a moisture and ash free basis varied from 34% for noncatalytic coal (H, only) to 67.5% when 3% SnCll was impregnated on the coal and the feed hydrogen contained 5% HC1. In these tests conversion was determined by the difference in weight of the extraction thimble before reaction and after extraction with toluene. Hence, this conversion also included conversion to gaseous products. The continuous runs were carried out at 6.89 MPa and 425 "C with a residence time in the tubular reactor of about 4 s. Under these conditions conversion varied from 4.2% with noncatalytic coal (H, only) to 44.2% with C0C12 impregnated on the coal and with 5% HC1 in the feed hydrogen. In these tests, the conversions were based on the extraction of the entire product with cyclohexane, toluene, and pyridine and conversion to gaseous products was not determined. hence, the reported overall conversion is probably low for the continuous runs. In addition, in the continuous system, the feed hydrogen was not preheated and calculations indicate that the coal entrained in the hydrogen did not reach the desired reaction temperature until it had traveled approximately one-fourth of the length of the reactor tube. As a result, average reactor temperature was somewhat less than reported. The increased conversion to oils is rather dramatic when HC1 is included in the feed as shown by comparing Tables I11 and V, and as shown in Table VI. Over five times as much oil is produced from the noncatalytic coal and the coal impregnated with FeC12when HC1 is included in the gas stream compared with hydrogenation when no HCl is present. From two to three times as much oil is produced when the coal is impregnated with ZnC12, SnC12, CuC12,and CoCl2. At least two large coal liquefaction research projects utilize ZnCl, as the catalyst (Zielke et al., 1980; Wood and Wiser, 1976; Lantz, 1977). However, this research shows that in the presence of HC1, CoC12,SnC12,CuC12,and NiCl, are all more active than ZnCl, for conversion of Montana subbituminous coal, and that FeCl, is almost as active. The impregnated metal chloride-gaseous HC1 catalyst system appears to offer several important advantages when compared with conventional hydrocracking catalysts and the ZnC12 processes under development. Conventional hydrotreating catalysts such as sulfided cobalt-nickeltungsten-molybdenum impregnated on alumina or silica alumina lose activity very rapidly and the spent catalyst are very difficult to regenerate because of interaction with coal ash constituents. Sulfur and nitrogen in the coal are liberated as H,S or NH, and this can react with both the impregnated ZnCl, or the molten ZnC12catalysts and this
94
IM.Eng. Chem. Process Des. Dev. 1981, 20, 94-104
complicates recovery and regeneration. However, with the excess HC1 in the system the catalyst is maintained in the chloride form and this would greatly simplify recovery of the catalyst from the unreacted coal. Indeed, batch tests using coal impregnated with NiC12 indicate that, in the presence of HC1, all of the nickel remains on the spent coal residue as the chloride and that virtually all of its can be recovered from the residue by simple leaching. Tests have not yet been made on the recovery of the metal halide catalysts from the products of the continuous runs, however. Recovery would have to be virtually 100% if the process is to be economical. The mechanism for increased conversion to oils is not known, but based on the data obtained to date it is interesting to conjecture. From the preliminary data it appears that the thermal splitting of the coal molecules is catalyzed by HCl with the split fragments being stabilized by the short residence time and by the addition of hydrogen which is catalyzed by the metal. The rate-limiting step appears to be the terminal cracking of the coal molecules. It is also possible that HC1 contributes to the stabilization of the split fragments. Based on the acidity of the metal halides, HC1 promotes liquefaction using the better hydrogenation,less acidic chloride while liquefaction is poorer with the more acidic ZnC1,. It is possible that this system is a Friedel-Crafts catalyst which is prevented from deactivating as anhydrous HC1 replaces adsorbed H 2 0 (from hydrodeoxygenation of the subbituminous coal). This study was primarily a screening study and much more work needs to be done to show that such a process is feasible and before an understanding of the effects of different catalysts and the synergistic effects of the gaseous HC1 and impregnated metal chlorides can be obtained.
Conversion to and analysis of gaseous products, a better characterization and analysis of the oils, asphaltenes, and residual unreacted coal all need to be determined and correlated with the catalyst type and amount of HC1 included in the feed gas. The amount of sulfur, nitrogen, and chloride remaining in the entire product and each fraction would also be important. The system is quite corrosive and the reactor tubes must be replaced frequently due to both corrosion from the HCl and erosion of the metal by the abrasive coal. It may be possible to minimize this by reducing the amount of HC1 in the reacting gas and by use of other alloy tubing. These studies will be made in the future and will be the basis for future publications. Acknowledgment Montana State University Chemical Engineering students David Alzheimer, Michael Gerondale, and Michael Biegalke contributed greatly to the experimental investigation. In addition, the invaluable assistance of the late Sila Huso and Jim Tillery who constructed much of the laboratory equipment is gratefully acknowledged. The work was carried out under NSF Grant No. ENG 74-23009 A01. Literature Cited Lank, P. M., “The Unlverslty of Utah’s Continuous Coel Hydrogenation Process”, Oak Ridge Natlonal Laboratory, Oak Ridge, Tenn., Report No. ORNL/TM 1537 (June 1977). McCandless, F. P.; Berg. L. I d . Eng. Chem. Process Des. Dev. 1970, 9 , 110-1 15. Wood, R. E.; Wiser, W. H. Ind. Eng. Chem. Process Des. Dev. 1978. 15, 144- 149. Zielke. C. W.; Klunder, E. B.; Maskew, J. T.; Struck, R. T. Ind. Eng. Chem. Process Des. Dev. 1980, 19, 85-91.
Received for review April 28, 1980 Accepted September 19,1980
VolatCHty of Coal Liquids at High Temperatures and Pressures Grant
M. Wllson
Wllco Research Company, Provo, Utah 8460 1
Robert H. Johnston, Shuen-Cheng Hwang, and Constantlne Tsonopoulos’ Exxon Research and Engineering Company, Fbrham Park, New Jersey 07932
The volatility of coal liquids has been experimentally determined at 700-880 O F and about 2000 psia. These measurements were made in a flow apparatus to minimize thermal decomposition effects at high temperatures. Three coal iiqulds in mixtures with H2,methane, and H2S were investigated. Measurements were also made up to 900 OF on the vapor pressure of pure compounds found In coal liqulds and on the equilibrium pressure of narrow coal liquid cuts. These data were used to develop a new method for the prediction of the critical point and the superatmospheric vapor pressures of aromatic fractions that is superbr to the Maxwell-Bonnell correlation. The VLE data on coal liqulds and some recent high-temperature VLE data on binaries of aromatics with H2or methane were analyzed with a modified Chao-Seader correlation and a modified Redllch-Kwong equation of state. Both VLE correlations are shown to be equivalent in the predlctkm of the volatillty of coal liqulds-when the new vapor pressure procedure is used.
Introduction Coal liquefactionhas drawn a lot of attention in the past few years. Several processes are currently being investigated, and in every case the process development has been hampered by the unavailability of phase-equilibrium data 0196-4305/81/1120-0094$01.00/0
at the conditions of interest. One of the principal areas of data uncertainties that would affect the Exxon Donor Solvent coal liquefaction process (Furlong et al., 1976) is the volatility of coal liquids at liquefaction reactor conditions. 0 1980 American Chemical Society
