Energy & Fuels 1987,1, 332-338
332
Catalysts Supported on Hydrous “Titanates”for Hydroprocessing Mixtures of Heavy Oil and Coal J. Monnier,*+ G . D8nBs,x J. Potter,§ and J. F. Krizt CANMET Energy Research Laboratories, Energy, Mines and Resources Canada, Ottawa, Ontario K 1 A OG1, Canada, Department of Chemistry and Laboratories for Inorganic Materials, Concordia University, MontrEal, QuEbec H3G 1M8, Canada, and Energy Research Unit, University of Regina, Regina, Saskatchewan S4S OA2, Canada Received September 2 1 , 1986. Reuised Manuscript Received March 18, 1987
Hydroprocessing catalysts were prepared by ion exchange of a sodium hydrous “titanate” support with aqueous solutions of transition metals (Ni, Co, Mo, Zn) or main-group elements (Sn, Al). These catalysts were tested in an autoclave system with a constant flow of hydrogen for the hydrocracking of mixtures of 1 part of subbituminous coal and 2 parts of residual oil from vacuum distillation of an Alberta heavy oil. The operating pressure and temperature were chosen so that the hydrocracking reactions occurred at the threshold of coking. Catalysts were compared on the basis of yields of oil, asphaltenes, and preasphaltenes, conversion of THF insolubles, and sulfur content in the hydrocarbon distillates collected from exiting gases. The presence of semicoke was determined by petrographic analysis of samples of THF insolubles. Mossbauer spectroscopy provided information on the chemical changes occurring to the tin active sites during reaction, and X-ray powder diffraction provided data on the nature of the crystalline phases present in these catalysts and on their average particle dimension. Experimental results indicate that some metal-exchanged hydrous “titanates” can help produce more pentane-soluble oil and less gas than commercial hydrotreating catalysts.
Introduction Table I. Composition of Hydroprocessing Catalysts In recent years, coal conversion research and developcomposition, wt % ment has focused on the concept of coprocessing where coal catalyst Mo Co Pd Sn Zn Ni Na Ti is hydrogenated in the presence of bitumen or heavy oil MB-582 15.8 0.1 42.0 of petroleum Coprocessing is an attractive alMB-586 12.7 1.1 43.2 ternative in Canada where large deposits of both coal and MB-588 12.5 4.1 43.0 heavy oil are found in geographical proximity. An ecoMB-592 10.1 5.0 0.1 40.5 11.2 MB-593 1.5 46.6 nomic advantage over conventional liquefaction may be MB-596 11.4 5.6 47.5 in the elimination of recycle streams of coal-derived liquids. MB-599 16.8 3.9 36.5 However the principal advantage in coprocessing is the MB-602 7.8 6.0 41.2 increase in distillate yield based on the amount of heavy MB-531 (ref) 8.9 2.6 oil used. As in any process utilizing high-pressure hydrogen, better conversions and liquid product quality can washed with water and acetone and dried under vacuum overnight. be achieved by the addition of catalysts to the feed s l u ~ ~ y . ~ J In addition to the referenced procedures, successive exchanges were also performed with solutions of molybdenum and cobalt Recently, the Sandia National Laboratories of the US. to obtain Co-Mo hydrous “titanate” catalysts. Before being tested Department of Energy developed new catalyst supports for hydroprocessing, the titanate catalysts were calcined in air for coal hydrogenation utilizing metal-exchanged hydrous a t 400 “C for 2 h. Cobalt-molybdenum supported on alumina titanate^.^^^ In our laboratory catalysts prepared in a similar manner were applied in the coprocessing of residual oil and coal. The present study discusses the results ob(1)Monnier, J. CANMET Rep. 1984,84-5E. ( 2 ) Kelly, J. F.; Fouda, S.A,; Rahimi, P.; Ikura, M. In Proceedings of tained.
the Coal Conuersion Contractors’ Review Meeting; Kelly, J. F., Ed.; CANMET, Energy, Mines and Resources Canada: Ottawa, Canada, 1985;
Experimental Section
pp 397-423.
Catalyst Preparation. Metal-exchanged hydrous “titanates”1o (MEHT) were prepared according to the method of Dosch, Stephens, and c o - w o r k e r ~ .The ~ ~ ~catalyst support is produced by reacting titanium(1V) 2-propoxide, Ti[OCH(CH3)J4, with a solution of sodium hydroxide in methanol to produce a soluble intermediate, and precipitating the sodium hydrous “titanate”, NaTi20SH,by using a mixture of 10 w t % water in acetone. The solution of sodium hydroxide was kept below 0 “C during the reaction, an aspect not brought up in the literature. The precipitate is washed with water and acetone and then dried under vacuum. The resulting material is a white, fluffy powder, which can be loaded with a metal from an aqueous solution of its salt via the exchange with the sodium ions. The catalyst is then Energy, Mines and Resources Canada.
* Concordia University.
*University of Regina.
0887-0624/87/2501-0332$01.50/0
(3)Ignasiak, B.;Kovacik, G.; Ohuchi, T.; Lewkowicz, L.; du Plessis, M. P. In Proceedings of the Coal Conuersion Contractors’ Review Meeting; Kelly, J. F., Ed.; CANMET, Energy, Mines and Resources Canada: Ottawa, Canada, 1985;pp 385-395. ( 4 ) McLean, J. B.; Duddy, J. E. P r e p . Pap-Am. Chem. SOC.Diu. Fuel Chem. 1986,31(4),169-180. (5)Chem. Eng. News 1986, 64(28), 26. (6) Curtis, C. W.; Teai, K.-J.; Guin, J. A. Prep.-Am. Chem. SOC. Diu. Pet. Chem. 1985, 30(4), 688-696. (7) Monnier, J.; Kriz, J. F. Prepr.-Am. Chem. SOC. Diu. Pet. Chem. 1985, 30(3),513-520. (8) Dosch, R.G.; Stephens, H. P.; Stohl, F. V. US.Patent 4511 455, 1985. (9)Stephens,H.P.;Dosch, R. G.; Stohl, F. V. Znd. Eng. Chem. Prod. Res. Deu. 1985, 24, 15-20. (10)We use the name hydrous “titanates”,as do Dosch, Stephens,et
d8t9 for these catalysts, although no structural data suggest that discrete titanate ions exist. The name hydrous ‘titanates” for these types of compounds is also used by Cotton and Wilkinson: Cotton, F. A.; Wilkinson, G. Advanced Inorganic Chemistry, 4th ed.; Wiley: New York, 1980; p 695.
Published 1987 by the American Chemical Society
Catalysts Supported on Hydrous "Titanates"
Energy & Fuels, Vol. 1, No. 4, 1987 333
Table 11. Characteristics of Feedstocks Forestburg Subbituminous C Coal ultimate anal., wt % proximateanal., wt % carbon hydrogen sulfur nitrogen oxygen (by diff) ash
64.04 3.87 0.53 1.65 20.41 9.50
moisture ash volatiles fixed carbon (by diff)
9.2 7.8 46.5 36.5
Cold Lake Vacuum Bottoms ultimate anal., w t % other characteristics carbon hydrogen. sulfur nitrogen oxygen
a ,
.
84.04 9.94 5.46 0.63
0.50'
IBP oil asphaltenes -525 "C +525
"C
420 "C 72.3 % 27.7 % 20% 80%
(Harshaw HT-400E) was used in a powder form (grain size 4 4 9 pm) as the reference catalyst. The compositions of the calcined MEHT catalysts, analyzed by inductively coupled Vgon plasma spectrometry and neutron activation, are presented in Table I. Their surface area,measured by (BET) nitrogen adsorptioq.varied from 60 to 150 m2/g. Pore volume data were obtained 59 mercury porosimetry using a Micromeritics Autopore 92OQ. Information on the size and shape of the catalyst particles, a t various stages of their preparation, was provided by scanning electron microscopy, by use of a Hitachi S-520 scanning electron microscope operating with the filament energized a t 15 kV. The catalysts for electron microscopy were dried a t 120 OC overnight to remove adsorbed water, then stored in a dessicator, and gold-coated. Mossbauer Spectroscopy and X-ray Diffraction of the Tin-ContainingCatalysts. Information on the oxidation state and chemical environment of the metal active sites in some catalysts can be provided by Mossbauer spectroscopy. Since among the metals used in this work only tin has a convenient Miissbauer isotope,samples of tin catalystswere analyzed by l19Sn Mijssbauer spectroscopy (MES). The 23.875-keV Miissbauer y-ray line was obtained using a 15-mCi Ca11g"Sn03 source purchased from Amersham with a 0.1 mm thick P d foil filter. All spectra were recorded with both the source and absorber (sample) a t ambient temperature. The amount of sample used was determined in order to have 10 mg of Sn/cm2. The spectra were calibrated by using a-SnF2 and CaSn03 as standards. Computer fitting was performed by using GMFP5," a revised version of the General Mossbauer Fitting Program (GMFP) of Ruebenbauer and Birchall.12 All chemical isomer shifts are given relative to CaSn03 a t room temperature [6 (CaSn03) = 0 a t room temperature]. X-ray powder diffraction tests, performed on the same tin samples to complement the Mossbauer data (X-ray diffraction giving long-range order, Mossbauer spectroscopy being a local probe) were expected to reveal information on the structure of the catalysts. All X-ray powder patterns were recorded by using Cu Ka radiation (A&& = 1.541 78 A) obtained by using a Ni foil filter. Catalytic Hydroprocessing Tests. The hydrocarbon feedstock was a mixture of 33 wt % Forestburg subbituminous C coal (as received) and 67 wt % Cold Lake vacuum bottoms (CLVB) from an Alberta heavy oil reservoir. As shown in Table 11, CLVB is about 72 wt % oil (pentane soluble) and contains about 5.5 wt % sulfur. Table I1 also gives the proximate and ultimate analyses of the coal used as fine particles smaller than 149 pm in diameter. To simplify handling, 60 g of coal and heavy oil was placed inside a reactor liner with 3.2 g of catalyst. Initially, the feed slurry of coal and residual oil contained about 2.3 wt 9% sulfur, which, depending on the catalyst, corresponded to a minimum of 12 S atoms per atom of exchanged cation. I t was thus expected that sulfiding would occur in situ under reaction (11) Monnier, J.; DbnBs, G.; Anderson, R.B. Can. J. Chem. Eng. 1984, 62,419-424. (12) Ruebenbauer, K.; Birchall, T. Hyperfine Interact. 1979, 7, 125-133.
Figure 1. SEM micrograph of sodium hydrous titanate before cationic exchange: top, 5000X magnification; bottom, lOOOOX magnification. conditions by contact with the liquid phase rich in sulfur compound~.'~Consequently, no presulfiding procedures were applied in any case. I t should also be mentioned that in a continuous reactor system, a "disposable" catalyst would probably be added without any pretreatment to the slurried feed and then discarded after primary upgrading of the heavy oil-coal mixture. The catalytic hydroprocessing tests were performed at 410 "C in a 300-mL stirred autoclave equipped to operate under continuous hydrogen flow. A back-pressure regulator maintained the pressure inside the autoclave a t 3.4 MPa. A cold trap placed before this regulator removed the distillates from the outlet gases. The hydrogen flowrate of about 275 mL (STP)/min was sufficient to replace the internal gas volume in about 15 min. The system temperature and pressure conditions were chosen so that reaction occurred a t the threshold of coking as shown previ~usly.~A reaction time of 225 min was maintained in an attempt to observe the hydroprocessing selectivity while reasonably enhancing the distillate yield. On completion of a run, the oil remaining in the autoclave (between the wall and the reactor liner) was drained. The distillates were the liquids including aqueous portions collected from the trap. Tetrahydrofuran (THF) used to remove residues from the metal liner was subsequently evaporated in a vacuum oven. The residues were analyzed by solvent extraction for the oil content (pentane solubles), asphaltenes (pentane insolublestoluene solubles; letter A in figures described later), preasphaltenes (toluene insolubles-THF solubles; letter P), and THF insolubles (shaded area). Total oil content (letter 0) was calculated by adding the weight of the distillates, excluding water (letter W), and the oil drained from the autoclave to the weight of the pentane solubles. The distillates were characterized by elemental analysis and by gas chromatography-simulated distillation. The outlet gases (letter R) were analyzed by gas chromatography. Semicoke formation was determined by petrographic analysis of samples of T H F insolubles. This analysis was carried out in plane-polarized light on the polished, pelleted samples by using a Leitz "Orthoplan-pol" incident light microscope, a t a magnification of 500X. An accessory X plate was inserted in the light path to enhance the optical properties of anisotropic solids and facilitate the identificationof anisotropicsemicokes. The relative abundances (Vol %) of unaltered coal, altered coal, granular (13) Ternan, M.; Whalley, M. J. Can. J. Chem. Eng. 1976,54,642-644.
Monnier et al.
334 Energy & Fuels, Vol. 1, No. 4, 1987
Table 111. Average Particle Size of Catalyst MB-599 As Determined by Scherrer’s Method from the Broadening of the Bragg Peaks B,b deg t,b A compd stageo 6 ( K a Cu), deg BMpb deg BSyb deg C 13.58 0.70 0.17 0.68 120 Til-,Sn,02 rutile C 17.40 1.05 0.17 1.04 80 Til,Sn,02 rutile 0.78 107 Til,Sn,02 rutile 0.80 0.17 17.60 u (63) 0.25 330 Ti02 anatase 0.30 0.17 12.64 u (76) 0.50 0.17 0.47 177 Til-,Sn,02 rutile 17.60 u (76) OStage: c = calcined catalyst and u = used catalyst. MB-599 was used in two runs (run 63 refers to Figure 4d and run 76 to Figure 4e). bTheaverage particle size t was obtained from Scherrer’s formula t = (0.9A)/(B cos 6) (A = wavelength; 6 = Bragg angle), corrected by using Warren’s method B2 = BM2 - BS2,where B is the broadening at half-height due to the small size of the particles and BM and Bs are the experimental values for the sample and for a well-crystalline standard, re~pective1y.l~
1 ________ MB-53 FIB-592
.16-
\
-
-
€ W
W
E
3
,
.12
-
-1 0 .08-
> w
-
2
.04-
w
a
10
100
PORE DIAMETER
1000
10.000
(A>
Figure 3. incremental pore volume as a function of pore diameter, (from mercury porosimetry data) for Co-Mo catalysts supported on alumina (MB-531) or on hydrous titanate.
Figure 2. SEM micrograph of calcined tin hydrous titanate, MB-599: top, lOOOX magnification; bottom, loooOX magnification.
Table IV. ‘19SnMossbauer Parameters for Tin Catalyst MB-599 at Room Temperature A t mm stage” b,b*cmm s-l s-l nd % centre assigntf C -0.03 100 Sn/O octa dist 0.10 +4 85 Sn/O octa dist u (63) -0.03 0.10 +4 +2 15 SnS 0.86 3.22 u (63) +4 90 Sn/O octa dist 0.10 -0.03 u (76) 0.86 +2 10 SnS u (76) 3.28
residue, coal/pitch-derived solids, and mineral matter, terms that are defined in the Discussion, were measured by point counting with the aid of a Swift automatic point counter and a Leitz “click-stop” sample holder.
“Stage c = calcined and u = used; runs 63 and 76: as in Table 111. b 6 = isomer shift relative to CaSn03 at room temperature. Error bar = 0.01. n = tin oxidation state. e Percent contribution to the total spectrum; error bar = 1%. fSn/O octa dish distorted octahedral environment of oxygen.
Catalyst Characterization Scanning Electron Microscopy. Figures 1 and 2 present SEM micrographs of MEHT catalysts at two stages of their preparation. In the first figure, the sodium hydrous “titanate” support is shown after overnight drying under vacuum at room temperature. It consists of 1-2pm-diameter particles that have an irregular shape and are probably composed of smaller grains since no Bragg peak was observed by X-ray diffraction. After cationic exchange and catalyst calcination, these l-pm particles agglomerate to give very porous lumps of materials varying in size from 30 to 60 pm (Figure 2). Calcination of the support without any exchange also leads to agglomeration. Mercury Porosimetry. The expected high porosity of these catalysts was confirmed by mercury porosimetry. In Figure 3, the pore distribution of a cobalt-molybdenum hydrous “titanate” (MB-592), which is plotted as an incremental pore volume vs. pore diameter, is compared with that of a commercial alumina-supported catalyst containing the same metals (MB-531). A very wide range of pore diameters is observed for MB-592, with a minimum size of about 150 A. Such a distribution of pore diameters
can be advantageous in facilitating the diffusion, to the metal active sites, of large “molecules” of asphaltenes and preasphaltenes, the size of which range in the hundreds of ang~tr0ms.l~ This contrasts remarkably with the narrow pore distribution of MB-531, centering at about 90 A. Characterization by Mossbauer Spectroscopy and X-ray Diffraction. Mossbauer spectroscopy, by probing shifts and splittings of the nuclear ground and first excited states of l19Sn,provides information on the causes of these effects, i.e., oxidation state, electronic structure, ligand changes, and coordination of tin. The Mossbauer results for MB-599 are summarized in Table IV. To complement these data, Figure 4 presents results of X-ray powder diffraction for the same catalyst. The sodium hydrous “titanate”, the catalyst support, is amorphous or microcrystalline (particle size C50 A) prior to its loading with active metals, as none of the samples characterized by X-ray diffraction give rise to Bragg peaks (14)Baltus, R. E. In Catalysis on the Energy Scene; Kaliaguine, S., Mahay, A., Eds.; Elsevier: Amsterdam and New York,1984;pp 553-560.
Catalysts Supported on Hydrous “Titanates”
Energy & Fuels, Vol. 1, No. 4, 1987 335
f
5
Y
E
b
I
I
30
25
Br-9
I 20 v l g l n Thrtr
I
I
15
10
(0)
Figure 4. X-ray powder pattern of the MB-599catalyst a t various stages: (a) support; (b) noncalcined ion-exchanged catalyst; (c) calcined catalyst; (d) used catalyst in T H F insolubles of run 63;(e) used catalyst in T H F insolubles of run 76. The interpolated base line (dashed line) is just a guide for the eye.
(Figure 4A). A similar finding was reported by Stephens et aL9 The noncalcined MB-599 catalyst, dried a t room temperature under vacuum, contains tin in mixed oxidation states. Tetravalent tin is the major component, the remaining tin being tin(I1). After calcination a t 400 “C for 2 h in air,the tin species was identified as being Sn02by Mossbauer spectrocopy, which was confirmed by X-ray diffraction. Calcination also resulted in significant growth of particles are shown in the X-ray powder patterns (Figure 4C) by the presence of several groups of overlapping Bragg peaks. The average particle size of this calcined catalyst, given in Table 111, was determined by using Schemer’s formu1a,15corrected with Warren’s method15 for instrumental broadening. In spite of inaccuracy due to the strong peak overlapping, it is clear that the samples are microcrystalline with an average particle diameter in the range 100-300 A. Therefore, the particles from 30 to 60 pm observed in electron microscopy on calcined samples (Figure 2) are, indeed, agglomerates of a large number (about loio) of much smaller size particles (diameter about lo00 times smaller), which explains the irregular shape and absence of internal structure of the aggregates. Additional information was obtained from the diffraction spectra of the titanate. The main crystalline component was identified as a rutile-type Tii,Snx02 solid solution. Its cell parameters are intermediate between (15) Cdlity, B. D. Elements of X-ray Diffraction, 2nd ed.; AddisonWesley: Don Mills, Ontario, Canada, 1978, pp 192, 284.
IO0
n -0
80
Q,
0)
k-
80
G
0
40
s & 3 v
20
0
FEED Blank MB-531 MB-588 MB-593 ME-882 ME-598 MB-599 Pd Mo Co .Zn Sn
Figure 5. Product distribution for MEHT catalysts tested at 410 O C (0 = oil, W = aqueous phase, A = asphaltenes, P = preasphaltenes, shaded area = THF insolubles, R = gas and loss).
those of Sn02 and TiOz rutile.16
Test Results and Discussion Coproceasing testa were performed with MEHT catdysta containing different metal cations. Product distributions (16) D6n& G., ‘Characterization of Tin Catdysta by Mbsbauer Spectroscopy”; final report to CANMET under Supply and Services Canada Contract Serial No. OST85-00243; Department of Chemistry, Concordia University: Montreal, 1987.
Monnier et al.
336 Energy & Fuels, Vol. 1, No. 4, 1987 MB-531 582
588
592
593
586
596
COKE FORMATION
Figure 7. Yields of gases including losses for MEHT catalysts tested at 410 "C. FEED ME-502 ME-53 1 MB-592 MB-53 I 410'C 410'C 430'C 430'C
Figure 6. Product distribution for Co-Mo catalysts supported on alumina (MB-531) or on hydrous titanate and tested at 410 and 430 "C (0 = oil, W = aqueous phase, A = asphaltenes, P = preasphaltenes, shaded area = THF insolubles, R = gas and loss).
corresponding to the metal contained within the support are depicted in Figure 5. It shows the product distribution in terms of oil (letter 01,asphaltenes (A), preasphaltenes (P), and gaseous (R) and aqueous products (W), the latter being collected in the trap following the autoclave, expressed in weight percent (ma0 of the feed mixture. Coal conversion was measured in terms of THF insolubles (shaded area) before and after each run. This determination was thus complicated by the formation of THFinsoluble semicoke, an undesirable competing reaction, occurrence of which indicates the upper limit of the usable temperature range. Figure 6 shows that the addition of a catalyst to the feed slurry enhances the conversion of THF insolubles and the oil yield. In the absence of catalysts, more of the THFinsoluble material is found in the product than in the feed slurry, which suggests the occurrence of coking at these experimental conditions that is confirmed by the petrographic analysis described later in this section. However, in the presence of catalysts, up to 65 wt % of the THFinsoluble material in the feed was converted and oil contents up to 65% were observed. The catalyst modification through exchange of sodium for molybdenum and palladium led to conversions of about 65% and 59% respectively. These conversions appear to be significantly higher than those observed for nickel and cobalt (about 47%). However, not all MEHT catalysts worked as well. For tin, the outcome was not significantly different from that observed without catalysk the negative values of conversion indicated significant semicoke formation. A zinc-exchanged hydrous "titanates" catalyst performed somewhat better in terms of oil yield and conversion. It is of interest to note that good activity of tin and zinc chlorides and oxides, as coal liquefaction catalysts, was reported in the literature.17J8 Similarly, net gains in oil fraction were obtained for all metal cations except tin and zinc. The oil content in the products varied from 48 wt % for cobalt to 65 wt % for molybdenum. This may appear as a small gain when compared to the initial content of about 47 wt % in the feed slurry. However, one must also consider the difference (17) Weller, S. W. In Catalysis ZV; Emmett, P., Ed.; Reinhold New York, 1966; Chapter 7. (18)Mizumoto, M.; Yamashita, H.; Matauda, S.Znd. Eng. Chem. Prod. Res. Deu. 1986, 24, 394-397.
in quality of the resulting oil of which 30-4090 boils below the initial boiling point for CLVB in the feed, indicating an appreciable decrease of the average molecular weight of the oil. The performance of palladium was likely impeded by the high sulfur content in the feed slurry (about 2.3 wt %) but was still comparable to that of the sulfurtolerant molybdenum. For tin and zinc, the oil contents were from 36 to 40%, similar to the blank test. Performance variations between the individual metals could be caused by the differences in initial metal dispersions within the support or dispersion stabilities under reaction conditions. To make a meaningful comparison between the hydrous "titanates" and the alumina-supported catalyst MB-531, a cobalt-molybdenum HT (MB-592) was prepared by successive cationic exchange in aqueous solutions of molybdenum and cobalt. Figure 6 presents the product distribution of both catalysts in coprocessing at 410 and 430 "C. A t both temperatures, MB-592 was superior in terms of oil content by about 20%. As the temperature increases, the fraction of asphaltenes and preasphaltenes decreases for both catalysts, whereas the fraction of THF insolubles increases sharply. As shown earlier, some of the MEHT catalysts produced up to 25% more oil than the reference catalyst (cobaltmolybdenum on alumina). Furthermore, the increase in oil yield with MEHT catalysts was accompanied by a lower yield of gaseous hydrocarbons, such as methane, ethane, and propane, as calculated by difference from the mass balance of the product distribution and shown in Figure 7. Although a negative (gas + loss) value was observed, this was evidently caused by errors in the analysis. Six tests performed at identical reaction conditions with MB-531 gave an average gas yield of about 12 wt % The MEHT catalysts, excluding tin and zinc, performed significantly better, the gas yield decreasing to an average value of about 5 wt %, based on 13 tests. Cracking reactions causing the formation of these hydrocarbons were evidently impeded, probably because of weaker surface acidity of these catalysts. With tin and zinc HT, the hydrogen transfer within the reacting hydrocarbon mixture was likely less effective, causing insufficient saturation of the large cracked intermediate molecules. Figure 8 compares the MEHT catalysts with the reference catalyst MB-531 in terms of sulfur removal from the distillate oil collected in the trap (boiling range from 60 to 350 "C). The sulfur concentrations are typically from 1.6 to 1.8 wt % and thus are lower than the 1.9 wt % observed in the blank test. Multiple ion exchanges can provide MEHT catalysts containing more than one cation. For a cobalt-molybdenum MEHT catalyst, the sulfur
.
Catalysts Supported on Hydrous “Titanutes”
Energy & Fuels, Vol. I, No. 4, 1987 337
A
ME-531 Co-Mo HEHT X no-fra. MEHT -k Blank
X
x x xxxx xxxx
X
x
A A I
I
I
1.0
x
BBBX I
I
1.2 ut X S
I
xx I
I
1.4 1.6 DISTILLATE
IN Figure 8. Performance of desulfurization.
I
I
+ I
I
1.8
MEHT catalysts for hydro-
Table V. Petrographic Analysis of THF-Insoluble Residues (vol % Mineral Matter Free) coal/
catalyst Sn
zn
blank Ni Co-Mo Pd
Mo MB-531 0
unaltered altered coal huminite 0.2 29.5 1.2 63.9 0.4 52.8 2.2 3.5 3.2 2.6 1.6
83.6 80.5 80.7 82.5 95.5
granular residues 0.0 0.0 0.4 14.1 16.1 16.1 14.9 2.9
pitchderived semicoke’ solidsb 51.4 18.8 25.2 9.6 33.4 12.9 0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.0 0.0
0.0
Anisotropic. *Isotropic.
content is further lowered to 1.3 w t %, indicating that a better sulfur removal is achieved when Co-Mo type catalysts are used.’ Although the commercial cobalt-molybdenum supported on alumina still outperformed this MEHT catalyst in terms of sulfur removal, it was encouraging to note that the difference was less significant, leading to sulfur levels in the product oil of 1.0-1.1 wt %. In the overall perspective, better refining can therefore be seen with the use of MB-592. Petrographic Analysis. Samples of THF-insoluble residues were characterized to determine the presence of anisotropic semicoke. Table V distinguishes five different solids found by petrographic analysis. The coal/pitchderived solids are an intermediate product of coal/pitch conversion. Ultimately, they yield a granular residue as a byproduct. This residue consists of a very fine grained, carbonaceous material interdispersed with inorganic minerals. Similarly, the anisotropic semicoke can also be considered as a byproduct (from coal or coal/pitch-derived solids), but from a different process involving mainly carbonization reactions that occur when the hydrogen transfer is insufficient within the reaction mixture for hydrogenation reactions to p r 0 ~ e e d . l ~The anisotropic semicoke is easily detected when the polarizers of the microscope are crossed or partially crossed. Altered coal ie unconverted coal that has been thermally or chemically affected, and finally, the fraction of unaltered coal represents reactive macerals that have not been affected by chemical or thermal treatment during coprocessing. Additional details can be found in the literature.zo The results in Table V confirm the presence of anisotropic semicoke in tests with negative conversion (corre(19)Belinko, K.;Kriz, J. F.; Nandi, B. N. Prepr. Can. Symp. Catal. 1979,6th. (20)Potter, J. “Characterizationof Solid Residues from Coal Liquefaction Processes”;final report to CANMET under Supply and Services Canada Contract Serial No. OST86-00100;Energy Research Unit, University of Regina: Regina, Canada, 1986.
sponding to the blank test and the tests with Zn H T or Sn HT). However, some coal/pitch-derived intermediates were present, but the absence of granular residues in these particular samples precludes any significant extent of liquefaction during these tests. With Mo, Ni, Co, Pd, and CO-Mo hydrous “titanates” that produced high oil yields, the fraction of granular residue was appreciable,accounting for up to 16.1 vol % of the THF insolubles. The absence of semicoke in these tests, even after the relatively long residence time, suggests that these catalysts provided effective hydrogen transfer during the fulllength of the tests. Mossbauer Spectroscopy and X-ray Diffraction Characterization of the THF Insolubles. The Mossbauer parameters of Table IV indicate that, after coprocessing, part of the tin(1V) initially contained in the catalyst was reduced to SnS. This is confirmed by X-ray diffraction (Figure 4D,E). The traces of elemental tin, P-Sn, suggested by X-ray diffraction (ASTM 4-0673), could not be corroborated by Mossbauer spectroscopy because of the strong overlapping of the p-Sn and SnS lines. Characterization of the THF insolubles did not show any evidence for the formation of tin(1V) sulfide, or titanium(IV) sulfide, or of mixed-oxidation state tin sulfides, Sn2S3 or Sn3S4. Only the presence of sulfur (2.3 wt %) in the feed slurry of Cold Lake vacuum bottoms and Forestburg coal can explain the formation of tin(I1) sulfide from tin(IV) oxide of the unused catalyst. Calculations of equilibrium constantsz1for the reduction of oxides of Sn, Mo, Co, and Ni at reaction temperatures used in these experiments show that Sn02is more difficult to reduce in Hz than NiO, COO,and Moog. Similar calculations indicate that the sulfides are relatively more stable than oxides of these metals. Thus, the presence of SnS in the THF insolubles suggests that formation of sulfides of the other metals also occurs during coprocessing. X-ray diffraction also indicated the presence of Ti02 anatase, Ti02 rutile, and Sn02 rutile in these catalystcontaining residues. The used MB-599 catalyst was found to contain much more Ti02 anatase than the unused calcined catalyst, alluding to formation of anatase during the hydroprocessing reaction. The change can be explained by using the hypothesis of a Til,Sn,Oz solid solution in the calcined catalyst. Indeed, as no tin(I1) is present in the calcined catalysts, simultaneous reduction and sflidation of some of the tin in the solid solution must occur, which releases Ti02 as anatase. Conclusions Metal-exchanged hydrous “titanates” have been investigated for the catalytic hydroprocessing of mixtures of coal and heavy oil. These MEHT catalysts consist of microcrystallites that agglomerate and provide catalyst particles with large pores, facilitating the diffusion of heavy hydrocarbon molecules to the active sites. The performance of some MEHT catalysts (molybdenum,palladium, nickel, cobalt, and cobalt-molybdenum) compared favorably with a commercial Co-Mo catalyst in terms of oil yield, conversion of THF insolubles, and gas formation. The petrographic analysis of the THF-insoluble residues confiimed the conversion data obtained by solvent extraction. Mossbauer spectroscopy, corroborated by X-ray powder diffraction, showed that catalysts are partially reduced and sulfided in the process. Acknowledgment. We wish to thank M. R. Fulton, P. Y. Dionne, P. S. Soutar, P. Zourdos, M. Skaff, and J.-F. (21) Barin, I,; Knacke. 0. Thermochemical Properties of Inorganic Substances; Springer-Verlag: West Berlin and New York, 1973.
Energy & Fuels 1987,1,338-343
338
Bilodeau for technical assistance at CANMET, N. Kapoor, J. Fitzgerald, and M. DBn6s a t Concordia University, and W. G. McDougall at the University of Regina. Financial assistance from Concordia University for the purchase of the Mossbauer spectrometer is acknowledged. G.D. thanks
the Natural Sciences and Engineering Research Council of Canada for a University Research Fellowship. Registry No. Mo, 7439-98-7; Co, 7440-48-4; Pd, 7440-05-3; Sn, 7440-31-5; Zn, 7440-66-6; Ni, 7440-02-0; TQ, 13463-67-7; NaTiz05H, 60704-88-3; SnOz, 18282-10-5; SnS, 1314-95-0.
Liquefaction of Yallourn Coal by Binary System Solvent Koji Chiba,* Hideyuki Tagaya,* and Naoki Saito Faculty of Engineering, Yamagata University, 4-3-16 Johnan, Yonezawa, Yamagata 992, Japan Received August 25, 1986. Revised Manuscript Received February 2, 1987
To clarify the interactions between coal liquefaction solvents, Yallourn coal was liquefied by using binary solvents composed of tetralin and polynuclear aromatic compounds. The hydrogen-transfer reaction from tetralin was accelerated by the addition of fluorene or phenanthrene, and coal conversions in those mixtures were larger than those observed with mixtures of 1-methylnaphthalene or naphthalene and tetralin, regardless of atmosphere and reaction time (20-60min). The physical affinity of phenanthrene or fluorene for Yallourn coal is large and shuttles hydrogen from tetralin to coal efficiently. Fluorene donates hydrogen to coal. The donor ability of binary solvent systems was evaluated quantitatively by using anthracene as the coal model. The hydrogen donor abilities of the fluorene-tetralin mixtures were larger than those of the 1-methylnaphthalene-tetralin or naphthalene-tetralin mixtures. This hydrogen donor scale reflects the results of coal conversion very well.
In coal liquefaction, the solvent plays a vital role as a hydrogen donor and a reaction medium to dissolve reactants and products. The selection of the solvent is an important factor for effective coal liquefaction, especially for solvent extraction liquefaction without a catalyst. In the operation of a continuous coal liquefaction process, synergistic effects during conversion have been observed when SRC or vacuum bottoms were added to the recycle solvent.' It has been suggested that large polynuclear aromatics, though not hydrogen donors themselves, act as hydrogen shuttlers and accelerate hydrogen transfer from the hydrogen gas or the donor solvent to the coal.2 Although the effects of such interactions between solvents during coal conversion appear to be they are not completely understood. In order to operate liquefaction processes smoothly, donor solvent evaluation has been attempted by many workers."" Model compound studies can evaluate the (1) hrbaty, M. L.;Lamen, J. w.; Wender, 1. Coal Science; Academic: New York, 1983; Vol. 2; pp 84-95. (2) Whitehuret, D. D.; Mitchell, T. 0.;Farcasiu, M. Coal Liquefaction; Academic: New York, 1980, pp 315-324. (3) Volker, E. J.; Bockrath, B. C. Fuel 1984,63, 285-287. (4) Kamiya, Y.; Yao, T.; Nagae, S. Bull. Chem. SOC. Jpn. 1982, 55, 3873-3877.
(6)I"&
753-758. . - - . - -.
M.; UiYma, T.; Matsuda, M. J . Fuel.
s O C * JPn.
1982~61,
(6) Chiba, K.; Tagaya, H.; Sato, S.;Ito, K. Fuel 1986, 64, 68-70. (7) Clarke, J. W.; Rantell, T. D.; Snape, C. E. Fuel 1982,61,701-712. (8)Curtis, C. W.; Guin, J. A.; Kwon, K. C. Fuel 1984,63, 1404-1409. Eng. (9) Cronauer,D. C.; Jewell, D. M.; Shah, Y. T.; Kueser, K. A. Id. Chem. Fundam. 1978,17, 291-297.
Table I. Yallourn Coal Analyses ultimate anal.' proximate anal.b C H N 0' O w t
69.8 4.0 0.8 25.4
% daf coal base. bwt
volatile matter fixed carbon moisture ash
32.8
51.9 13.3 2.0
5% coal base. cBy difference.
solvent donor ability quantitatively, but sufficient relationships with coal conversion were not obtained.8 In our laboratory, studies aimed at clarifying the solvent interactions and determining the solvent donor ability using a model compound have been undertaken to obtain more effective coal liquefaction solvents. In this study, compounds such as polynuclear aromatic contained in the recycle solvent were mixed with the representative donor, tetralin, and they were used as liquefaction solvents of Yallourn coal. To clearly elucidate the role of molecular hydrogen, experiments were performed in a nitrogen or a hydrogen atmosphere in the absence of catalyst. Experimental Section Coal Sample and Solvent. Yallourn coal analyses are given in Table I. Yallourn coal was ground to pass through a 100-mesh screen and dried at 107 oc in a nitrogen atm,osphere. (10) Raaen, V. F.; Roark, W. H. Fuel 1978,57,650-651. (11) Bockrath, B. C.; Bittner, E.; McGrew, J. J.Am. Chen. SOC.1984, 106, 135-138.
0887-0624/87/2501-0338$01.50/00 1987 American Chemical Society
