Impact of Multiphase Behavior on Coke Deposition in a Commercial

Diverse coke deposition mechanisms and models, all supported by ... of coke deposited on and within catalyst pellets and the distribution of coke with...
1 downloads 0 Views 317KB Size
Energy & Fuels 2005, 19, 1405-1411

1405

Impact of Multiphase Behavior on Coke Deposition in a Commercial Hydrotreating Catalyst under Sedimentation Conditions† Xiaohui Zhang, Martin Chodakowski, and John M. Shaw* Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Canada Received August 5, 2004. Revised Manuscript Received February 10, 2005

Diverse coke deposition mechanisms and models, all supported by experimental data, have been proposed for catalytic hydrogenation processes related to heavy oil and bitumen refining. The influence of multiphase behavior on the observed deposition mode is an important but unresolved question in this literature. The model mixture Athabasca vacuum bottoms (ABVB) + decane, which is shown to exhibit L1L2V phase behavior at elevated temperatures and a commercial heavy oil hydrotreating catalyst (NiMo/γ-Al2O3) were employed in this preliminary investigation. Under sedimentation conditions, the impact of phase behavior, per se, on the amount of coke deposited on and within catalyst pellets and the distribution of coke within catalyst pellets was found to be a secondary one despite the differences in the physical properties of the solventrich L1 phase and asphaltene-rich L2 phase. In all cases, the exterior surfaces of pellets were coated with a thick nanoporous coke layer. The ratio of pore surface area to pore volume also increased with the extent of reaction in all cases, indicating that larger pores are filled in part or plugged in preference to smaller ones. Observed differences in the properties of coked catalyst exposed to the L1 and L2 phases and to multiphase environments are attributed to differences in the asphaltene aggregate size distribution in the two phases and to multiphase hydrodynamic effects.

Introduction Catalyst deactivation may generally be attributed to poison adsorption, sintering, and mineral and/or coke deposition1 and may occur singly or in combination. At moderate operating conditions and short reaction times, coke deposition is the most significant one among them. Modes of coke deposition within a catalyst pore may affect macroscopic catalytic behavior. The performance of a catalyst with uniform coke deposition, for example, may differ significantly from ones in which coke deposition takes place at fixed sites.2 Coke deposition mechanisms on catalyst can be classified into three simplified modes: uniform surface deposition,3 pore-mouth plugging,4 and bulk-phase coke formation.3 The uniform deposition mode assumes that coke forms uniformly on catalyst inner surfaces. The pore-mouth plugging mode includes uniform coke deposition on inner surfaces of catalysts but also allows for coke deposition at small pore mouths within catalysts, leading to local pore blockages. The bulk-phase coke formation mode asserts that coke forms in the bulk liquid phase and deposits on all surfaces within reactors and includes both the uniform coke deposition and pore† Presented at the 5th International Conference on Petroleum Phase Behavior and Fouling. * Corresponding author: e-mail [email protected]. (1) Butt, J. B.; Petersen, E. E. Activation, Deactivation, and Poisoning of Catalysts; Academic Press: San Diego, CA, 1988; pp 3-26. (2) Arbabi, S.; Sahimi, M. Chem. Eng. Sci. 1991, 46 (7), 1739-1747. (3) Richardson, S. M.; Nagaishi, H.; Gray, M. R. Ind. Eng. Chem. Res. 1996, 35, 3940-3950. (4) Muegge, B. D.; Massoth, F. E. Fuel Process. Technol. 1991, 29, 19-30.

mouth plugging modes. Uniform surface deposition and pore-mouth plugging modes are two hotly debated modes because conflicting results were reported on the probable location and thickness of coke deposits on catalyst.3 Observation of bulk-phase coke deposition makes the understanding of coke deposition mechanisms more complex. All modes are supported by experimental findings. A complete description of coke deposition modes has to account properly for several important factors such as catalyst pore structure, reactant molecular size, reaction time, etc., which were investigated widely in the past. Phase behavior may be another important contributing factor that influences coke formation and deposition. For example, if multiple phases are present, which phase is continuous and which phase wets a catalyst may have an important bearing on outcomes. The role solvents play in suppressing coke formation in hydroprocessing of heavy oils is unresolved. Explanations range from improving hydrogen solubility, which promotes hydrogenation vs condensation reactions at fixed temperature and pressure,5,6 to the high radical scavenging ability of additives,5 to the improved solubility of heavy reaction products, which also suppresses condensation reactions and deposition.7 Supercritical conditions induced by adding a solvent to the feed also reduce coke formation in heavy oil upgrading.8 Diffusion rates of supercritical fluids into catalyst pores are much (5) Kubo, J.; Higashi, H.; Ohmoto, Y.; Arao, H. Prepr.sAm. Chem. Soc., Div. Pet. Chem. 1994, 39 (3), 416-421. (6) Kubo, J.; Higashi, H.; Ohmoto, Y.; Arao, H. Energy Fuels 1996, 10, 474-481.

10.1021/ef049802w CCC: $30.25 © 2005 American Chemical Society Published on Web 03/18/2005

1406

Energy & Fuels, Vol. 19, No. 4, 2005

higher than those for liquids,9,10 and an optimum fluid density that minimizes coke deposition and mass transfer limitations can be expected to exist at supercritical or near-supercritical conditions.11 The relative importance of these effects and their link with phase behavior and coke deposition mechanisms have yet to be determined. The phase behavior within heavy oil upgrading processes can be changed during thermal reaction and/ or hydroprocessing reactions. It has been concluded that heavy oils such as crude oils, residues, bitumen, etc., exhibit colloidal behavior in the presence of asphaltenes and resin fractions.12-14 The asphaltenes are believed to exist in heavy oils partly dissolved and partly in steric-colloidal and/or micellar forms depending on the polarity of their oil medium and presence of other compounds.15 But asphaltenes and heavier species can precipitate from the oil matrix if the colloid stability of the feed cannot be maintained. Storm et al.16,17 observed that asphaltenes could flocculate and precipitate to form another asphaltenic phase at elevated temperature even well below the temperature at which chemical reactions occur. They suggested that the asphaltenic phase is the precursor to the coke-producing phase in the reacting residue. Under catalytic hydroprocessing conditions, this colloidal structure is more likely disrupted and asphaltenes and heavier species can precipitate on the catalyst surface because of instabilities caused by thermal and hydrogenation reactions, which may enhance coke formation.18 The instability of hydrotreated heavy oil products was recently proved by use of a flocculation onset titration method.19,20 It is widely accepted that coke formation is triggered by liquid-liquid phase separation in the thermal upgrading of heavy oils.21-23 One of the two liquid phases is the heavy aromatic polar liquid phase that is the cokeproducing phase but different from the asphaltenic phase discussed by Storm et al.16,17 The phase separation here is caused by asphaltenes exceeding the solubility limit of the oil as reactions proceed, while the (7) Gray, M. R. Upgrading Petroleum Residues and Heavy Oils; Marcel Dekker: New York, 1994. (8) Scotta, D. S.; Radleinb, D.; Piskorzb, J.; Majerskib, P.; deBruijn, Th. J. W. Fuel 2001, 80 (8), 1087-1099. (9) Lee, S. Y.; Seader, J. D.; Tsai, C. H.; Massoth, F. E. Ind. Eng. Chem. Res. 1991, 30, 29-38. (10) Lee, S. Y.; Seader, J. D.; Tsai, C. H.; Massoth, F. E. Ind. Eng. Chem. Res. 1991, 30, 607-613. (11) Baptist-Nguyen, S.; Subramanjam, B. AIChE J. 1992, 38 (7), 1027-1037. (12) Li, S. H.; Liu, C. G.; Que, G. H.; Liang, W. J.; Zhu, Y. J. Fuel 1996, 75, 1025-1029. (13) Bardon, Ch.; Barre, L.; Espinat, D.; Guille, V.; Li, M. H.; Lambard, J.; Ravey, J. C.; Rosenberg, E.; Zemb, T. Fuel Sci. Technol. Int. 1996, 14 (1/2), 203-242. (14) Branco, V. A. M.; Mansoori, G. A.; Xavier, L. C. D. A.; Park, S. J.; Manafi, H. J. Pet. Sci. Eng. 2001, 32, 217-230. (15) Priyanto, S.; Mansoori, G. A.; Suwono, A. Chem. Eng. Sci. 2001, 56, 6933-6939. (16) Storm, D. A.; Barresi, R. J.; Sheu, E. Y. Energy Fuels 1995, 9, 168-176. (17) Storm, D. A.; Barresi, R. J.; Sheu, E. Y. Fuel Sci. Technol. Int. 1996, 14 (1/2), 243-260. (18) Furimsky, E.; Massoth, F. E. Catal. Today 1999, 52, 381-495. (19) Bartholdy, J.; Andersen, S. I. Energy Fuels 2000, 14, 52-55. (20) Bartholdy, J.; Lauridsen, R.; Mejlholm, M.; Andersen, S. I. Energy Fuels 2001, 15, 1059-1062. (21) Wiehe, I. A. Ind. Eng. Chem. Res. 1993, 32 (11), 2447-2454. (22) Li, S.; Liu, C.; Que, G.; Liang, W.; Zhu, Y. Pet. Sci. Technol. 1999, 17 (7/8), 693-709. (23) Rahmani, S.; McCaffrey, W.; Elliott, J. A. W.; Gray, M. R. Ind. Eng. Chem. Res. 2003, 42, 4101-4108.

Zhang et al.

formation of asphaltene-rich phases discussed by Storm et al.16,17 is purely physicalsflocculation and precipitation. Regardless of the reason causing another phase to form during upgrading, under delayed coking conditions at least, phase behavior has a dramatic effect on coke yield.24 However, while multiphase behavior and phase-behavior-dependent deposition phenomena have been observed under hydroprocessing conditions25 and there has been speculation that phase behavior affects coke formation under these conditions,7,26 no direct experimental proof has yet been presented. Minicucci et al.27 provide cogent arguments for how phase behavior could account for the amount of coke deposited and deposition mechanisms. It is also widely known that asphaltenes, which are concentrated in heavy crude oil, are the most troublesome fractions for upgrading and processing technologies and asphaltenes are responsible for the formation of coke precursors and for catalyst deactivation. In catalytic hydroprocessing of heavy oils, only the molecules/particles that can penetrate inside the catalyst can be upgraded. Early work showed that the size of asphaltene particles varies between nanometer and micrometer length scales.28 Asphaltene size may be another key factor influencing coke deposition in hydroprocessing, an effect yet to be explored. In this preliminary paper we focus on the impact of phase behavior on coke deposition in porous catalyst pellets under sedimentation conditions. The phase to which the catalyst is exposed from the inception of the experiments is controlled. Envisioned applications include heavy oil hydroprocessing, although the results may be applicable to other processes where sedimentation occurs even inadvertently through poor selection of operating conditions. Experimental Section Materials. Athabasca vacuum bottoms, ABVB, is the 525 °C+ boiling fraction of Athabasca Bitumen comprising approximately 32 wt % asphaltenes. A detailed composition analysis for ABVB is available elsewhere.29 Decane and dodecane were supplied by Aldrich with 99.5+% purity. The catalyst is a commercial 1 mm diameter cylindrical extrudate NiMo/γ-Al2O3 catalyst with 10-15 wt % MoO3 and 2-4 wt % NiO. The catalyst has a surface area of 220 m2/g and a pore volume of 0.59 cm3/g. Reactor. The coking experiments were performed in a variable-volume X-ray view cell30,31 with which one can observe the phase behavior of mixtures opaque to visible and infrared light. The advantage of using this view cell as a batch (24) Ali, V. M.Sc. Thesis, University of Toronto, Ontario, Canada, 2002. (25) Abedi, S. J.; Seyfaie, S.; Shaw, J. M. Pet. Sci. Technol. 1998, 16 (3/4), 209-226. (26) Ternan, M.; Rahimi, P. M.; Clugston, D. M.; Dettman, H. D. Energy Fuels 1994, 8 (3), 518-530. (27) Minicucci, D.; Zou, X.-Y.; Shaw, J. M. Fluid Phase Equilib. 2002, 194-197, 353-360. (28) Mullins, O. C.; Sheu, E. Y. Structures and Dynamics of Asphaltenes; Plenum Press: New York and London, 1998; pp 1-20. (29) Zou, X. Y.; Dukhedin-Lalla, L.; Zhang, X. H.; Shaw, J. M. Selective Rejection of Inorganic Fine Solids, Heavy Metals and Sulfur from Heavy Oils/Bitumen Using Alkane Solvents. Ind. Eng. Chem. Res. 2004, 43 (22), 7103-7112. (30) Abedi, S. J.; Cai, H. Y.; Seyfaie, S.; Shaw, J. M. Fluid Phase Equilib. 1999, 158-160, 775-781. (31) Zou, X. Y.; Zhang, X. H.; Shaw, J. M. The Phase Behavior of Athabasca Vacuum Bottoms (ABVB) + Alkane Mixtures. Manuscript in preparation.

Impact of Multiphase Behavior on Coke Deposition

Energy & Fuels, Vol. 19, No. 4, 2005 1407

Figure 2. X-ray transmission video image and schematic representation showing (a) catalyst held in the L1 and L2 phases and (b) catalyst held in the L1 phase with the L2 phase dispersed.

Figure 1. Phase behavior of ABVB + decane mixtures at 380 °C: (a) pressure-composition diagram showing critical phenomena and curves indicating phase behavior boundaries as described in the text; (b) video stills showing the relative amounts of phases present at the ABVB compositions (weight percent) and pressures (kilopascals) noted beneath the stills. The black horizontal line appearing beneath the stirrer (black circle) in the images is the stirrer support plate. hydroprocessing reactor is that one can monitor and control the phase to which catalyst pellets are exposed. Individual phase volumes and densities were also monitored concurrently. To ensure that the catalyst remained in the targeted phase, preliminary uncatalyzed phase behavior experiments were performed. A variable-position catalyst holder was designed and installed, and operating protocols were adopted. Method. First the phase behavior for the mixture ABVB + decane was evaluated by the synthetic method for the temperature range 20-380 °C. A pressure-composition diagram at 380 °C is shown in Figure 1a, and a series of video stills that provide a guide to its construction are shown in Figure 1b. As 380 °C is above the critical temperature of decane, the mixture exhibits vapor phase behavior at low ABVB mass fractions. As the mass fraction of ABVB is raised, a dense liquid (L2) appears at low pressures. As the pressure is raised, this liquid becomes miscible with the gas phase (column 1 of Figure 1b). In the center of the diagram one of three phenomena is observed, depending on the mass fraction of ABVB: (1) A low-density liquid (L1) appears above the L2 phase as pressure is raised and then disappears as pressure is raised further; (2) A low-density liquid (L1) appears above the L2 phase as pressure is raised and then the vapor phase (V) disappears as pressure is raised further (column 4 of Figure 1b), revealing the presence of a K-point in the diagram, that is, where a low-

density liquid and a vapor become critically identical in the presence of another liquid phase; (3) A high-density liquid (L2) appears beneath a low-density phase (L1) as pressure is raised, revealing the presence of an L-point in the diagram, that is, where two liquids become critically identical in the presence of a vapor phase. At high ABVB mass fractions, L1 and L1V phase behavior is observed. The composition and operating conditions selected for the coking experiments (30 wt % ABVB + decane at 380 °C and ∼3.0 MPa) were chosen to ensure that the amount of the targeted L1 (low-density liquid) and L2 (high-density liquid) phases present were large compared to the total catalyst charge (0.3 g) given a total liquid charge of ∼60 g, and the prospect of adequate reaction rates, that is, conditions where significant coke formation arises over the time scale of the experiment, ∼5 h. The catalyst charge placed in both L1 and L2 phases is 0.15 g, while the catalyst charge in the L1 + L2 dispersed experiment is 0.3 g. Care was taken to ensure that catalyst was exposed to only one bulk phase. The bulk phase to which catalyst pellets were exposed was monitored and controlled by manipulating cell volume and stirring rates. For L2 placement, catalyst and ABVB were placed in the view cell under vacuum and heated to 150 °C. The view cell was then cooled to room temperature, below the glass transition temperature of the ABVB, before the decane was added. For L1 placement, ABVB and decane were added to the view cell, and then the catalyst pellets were suspended in the view cell at an elevation above the anticipated L1/L2 interface. Once the catalyst and fluids were in place, the reactor was degassed under mild vacuum (∼30 kPa) for 30 min prior to heating. The reaction times at 380 °C were 2 and 5 h. As reaction rates are negligible at temperatures less than ∼300 °C,32,33 there is sufficient time, during the 3 h heating period for diffusion processes to approach completion prior to the occurrence of chemical reaction. At the end of an experiment, the reactor was cooled to room temperature over a 3-h period. Catalysts and fluids were then removed sequentially. The coked catalyst was extracted for 24 h in toluene and then dried for 2 h at room temperature under vacuum. The pellets were then analyzed. To exemplify the general arrangements for experiments, two X-ray transmission video images taken during experiments, one for catalyst placed in both L1 and L2 phases and another for catalyst placed in L1 + L2 dispersed, are shown in Figure 2. For clarification, the schematic representations are also shown in Figure 2. In Figure 2a, catalyst pellets were placed in both the L1 and L2 phases. To avoid catalyst placed in one phase having contact with the other phases, slow agitation was applied. In Figure 2b, catalyst pellets were (32) Cai, H.-Y.; Shaw, J. M.; Chung, K. H. Fuel 2001, 80, 10551063. (33) Cai, H.-Y.; Shaw, J. M.; Chung, K. H. Fuel 2001, 80, 10651077.

1408

Energy & Fuels, Vol. 19, No. 4, 2005

Zhang et al.

Table 1. Coked Catalyst Analytical Dataa Coking for 5 h

catalyst form during analysis surface area (m2/g) pore volume (cm3/g) pore surface area-to-volume ratio (m2/cm3) coke (wt %) C (wt %) H (wt %) S (wt %) N (wt %)

catalyst exposed to L1

catalyst exposed to L2

powder 112 0.14 800 33 28 2.1 5.1 0.6

powder 113 0.17 665 32 26 2.0 5.2 0.6

Coking for 2 h

catalyst exposed to L1 + L2 (dispersed) powder 127 0.17 747 32 26 2.1 4.7 0.6

pellet 126 0.17 741

catalyst exposed to L1

catalyst exposed to L2

powder 157 0.24 654

powder 165 0.29 569

23 1.9 2.9 0.5

21 1.8 3.6 0.5

a Coking time was 5 or 2 h. Catalysts were exposed to L1, L2, or L1 + L2 (dispersed) phases. The catalysts were in either powder or pellet form during analysis.

placed in the L1 phase and the view cell was agitated such that the L2 phase was dispersed but a vortex was not created at the L1/V interface. Catalyst Characterization. Bulk elemental analyses for carbon, hydrogen, and nitrogen were obtained from an elemental analyzer (Carlo Erba model Strumentazione 1108). Sulfur content was measured by the Schoniger method. To ensure representative sampling, 150 mg of coked catalyst was ground to a fine powder and then 5 mg samples were used for elemental analysis. Cross sections of whole coked catalyst pellets were also examined by use of a JEOL 8900 electron microprobe. Pellets were sectioned radially at the midpoint and then mounted in a mold with epoxy resin. Exposed surfaces were dry-polished with 3-µm diamond as the final abrasive. Finally, the polished surfaces were evaporatively coated with a thin layer of carbon. Local composition measurements were obtained at 35 µm intervals from the center to the exterior surface of the particles. Also, the cross sections of catalyst pellets were observed by visible light microscopy. Fresh catalyst pellets were used as a control for these experiments. The surface area, pore volume, and pore size distribution of the catalysts were determined with an Omnisorb 360. Nitrogen adsorption measurements were used to calculate surface areas, by use of the BET equation, and pore size distributions and pore volume were determined from nitrogen desorption data and the Kelvin equation.34 Asphaltene Aggregate Size and Concentration Measurement. Small-angle X-ray scattering experiments, which provide size distributions and concentrations for nanodispersed materials in liquids, were conducted at the Argonne National Laboratories. Scatterers in unreacted ABVB and 5 wt % ABVB + dodecane samples were assessed by use of an apparatus and method described elsewhere in detail.35

Results and Discussion Phase Behavior of Decane + ABVB Mixtures. Key phase boundaries (Figure 1a) place the experimental composition and operating conditions in context. Dashed curves in the diagram connote approximate placement of phase behavior boundaries that are difficult to observe experimentally. The L1L2V phase behavior has a lower pressure boundary that includes a L1 ) L2 + V critical point. At the experimental conditions (30 wt % ABVB + decane at 380 °C and ∼3.0 MPa), L1L2V phase behavior is exhibited. At higher pressures there is a transition to L1L2 phase behavior, and at lower pressures there is a transition to L2V phase behavior. The densities for the L1 and L2 phases are 470 and 870 kg/m3 and the volumes for the L1 and (34) Gregg, S. J.; Sing, K. S. Adsorption, Surface Area and Porosity; Academic Press: London and New York, 1982. (35) Zhang, X. H.; Chodakowski, M. G.; Shaw, J. M. J. Dispersion Sci. Technol. 2004, 25, 1-9.

Figure 3. Pore size distribution of fresh and coked catalysts in powder form for (a) 2 h and (b) 5 h of coking time.

L2 phases for a total charge of 60 g are 93.7 and 8.8 cm3. The mass in the gas phase is ∼8 g. As the L1L2V zone extends to room temperature it is possible to maintain the catalyst in the target phase by adjusting the total volume of the view cell. Bulk Properties of the Coked Catalyst Pellets. Catalyst pellets, extracted with toluene as described above, were subject to bulk analyses, whole and in powdered form as reported in Table 1. The carbon content of the catalyst increased with coking time as expected.37-39 From the sharp drop in the pore volume with coke mass fraction and the increase in the pore area to pore volume ratio with reaction from 372 m2/ cm3 to the values indicated, it is evident that larger pores are partially filled or blocked prior to the smaller pores. One sees this more directly in Figure 3, where the pore size distributions for the various cases are reported. The fresh catalyst has only one broad peak (36) Gray, M. R.; Zhao, Y.; McKnight, C. M.; Komar, D. A.; Carruthers, J. D. Energy Fuels 1999, 13, 1037-1045. (37) Gualda, G.; Kasztelan, S. J. Catal. 1996, 161, 319-337. (38) Marafi, M.; Stanislaus, A. Pet. Sci. Technol. 2001, 19 (5/6), 697710. (39) Matsushita, K.; Hauser, A.; Marafi, A.; Koide, R.; Stanislaus, A. Fuel 2004, 83, 1031-1038.

Impact of Multiphase Behavior on Coke Deposition

Energy & Fuels, Vol. 19, No. 4, 2005 1409

Figure 4. Element distributions within catalyst pellets (excluding the external coke deposit) for (a) carbon (including carbon coating), (b) vanadium, (c) sulfur for 5 h of coked catalyst, and (d) carbon coating on fresh catalyst.

with a median pore radius at about 40 Å. After coking, the spent catalysts possess a bimodal pore size distribution.3 The median pore size for the principal peak decreases with increasing coke content and a secondary peak arises at 18 Å. Coke deposition is clearly not uniform. The fraction of pores blocked vs filled is difficult to estimate as the coke fraction within vs without the particle was not determined. The carbon contents and pore volume losses for catalyst pellets exposed to L1 are greater than for the corresponding L2 or L1 + L2 dispersed cases, and the pore area loss is substantially less for the L1 + L2 dispersed case than for the others. As the mass of the catalyst is small compared to the mass of the bulk phases in which they reside and the mass of coke precursors present in each phase, and the residence times are relatively long, one may examine transport processes for an explanation of these results. The viscosity of L1 (solvent-rich phase) is less than the viscosity of L2 (bitumen-rich phase), and one is tempted to assert that therefore L1 circulates more readily within the macropores of the pellet, facilitating asphaltene transport and hence coke deposition. High-temperature viscosity data for bitumen combined with a variety of alkane solvents does not support this assertion, as the viscosities of both phases at 380 °C are low, less than

2 mPa s.40 An alternative explanation must be sought for this effect. However, for the L1 + L2 (dispersed) case, one can readily envision catalyst macropores being blocked by L2 drops, which inhibit asphaltene penetration within the catalyst pellets. Thus transport issues account only in part for the observations. On the basis of this result, one would anticipate a similar impact for the L2 + L1 (dispersed) case, and such an experiment is planned. Characteristics of Cross Sections of Coked Catalyst Pellets. It is difficult to section and analyze interior surfaces of catalyst pellets without introducing artifacts, and care must be taken to avoid overinterpretation of data obtained. For example, some smearing of all elements arises on exposed surfaces during polishing. This is unavoidable. Fortunately, the microprobe sample volume extends well below the surface, and the smearing represents only a small fraction of the volume sampled. Similarly, it is difficult to analyze for carbon in gold-coated samples, due to the relative weakness of the signal from carbon, and one must accept that the carbon content of carbon-coated samples is relative. Element distributions within mounted and polished catalyst pellets following 5 h of coking, as well as for a control experiment (mounted and carbon-coated fresh catalyst) are reported in Figure 4. For the coked catalyst

1410

Energy & Fuels, Vol. 19, No. 4, 2005

Zhang et al.

Figure 5. Photomicrographs of catalyst cross sections for (a) fresh catalyst, (b) 2 h coked catalyst in L1, (c) 2 h coked catalyst in L2, (d) 5 h coked catalyst in L1 + L2 dispersed, (e) 5 h coked catalyst in L1, and (f) 5 h coked catalyst in L2.

pellets, the average value for carbon is ∼20 wt %, for sulfur ∼6.5 wt %, and for vanadium ∼0.03 wt % (6 times the detection limit). The average value for carbon in the control is ∼10 wt %. The basis for the local composition measurements includes the carbon coating, the organic deposit, and catalyst. Clearly, the coating represents only a small fraction of the volume sampled. The carbon content of the control would otherwise be higher. The mass fraction for carbon in deposits is overstated, whereas the mass fraction of vanadium in deposits is understated due to the carbon coating. The local mass fraction for sulfur includes sulfur in the sulfidized catalyst and the organic deposit. These cannot be discriminated on the basis of the measurements. Despite these artifacts, it is clear that carbon, comprising ∼27 wt % for the spent catalyst pellets as a whole (obtained from bulk measurements and reported in Table 1) is deposited preferentially on exterior as opposed to interior surfaces of the catalyst irrespective of which of the two phases, L1 or L2, wets the catalyst. The carbon content within the pellets is significantly lower in all cases, ∼10-20 wt %, than the bulk measurement. The results do confirm that carbon, vanadium, and sulfur are well distributed throughout the pellets, as one would expect for catalysts possessing a network of macropores. The coke layer thicknesses for the cases investigated are presented in Figure 5 and range from 10 to 20 µm. A fresh catalyst pellet, Figure 5a, acts as a control. It is evident that coke layer thickness is variable at fixed reaction time regardless of the phase to which the catalyst was exposed. Coke also deposited on vessel surfaces. Such deposits are thought to arise from asphaltene precipitation during thermal reaction.36 If so, asphaltenes are clearly precipitating from both L1 and L2 phases and the catalyst pellets act as traps for them, as the liquid circulates among the pellets. The mechanism appears to be the same in both L1 and L2s a surprising result given the composition differences between the two phases. Impact of Asphaltene “Size” on Coked Catalyst Properties. A fraction of asphaltenes present in heavy oils are small enough to be observed during SAXS measurements. Figure 6 shows scatterer size distribution data for ABVB and ABVB (5 wt %) + dodecane, measured by small-angle X-ray scattering. The method is limited to scatterers with leading dimensions in the

Figure 6. Asphaltene aggregate size distribution for (a) ABVB (32 wt % asphaltenes) and (b) 5% ABVB in dodecane (1.6% asphaltenes). The maximum dimension for each bin is shown.

1-100 nm size range. While the conditions are not identical to the ones employed in the coking experiments, it is clear that a greater fraction of smaller scatterers are present in ABVB diluted in dodecane over a broad range of temperatures than in the parent oil under similar conditions. Further, the leading dimensions of these small scatterers fall well within the size range of the catalyst pores. Thus, the average scatterers present in L1 (the dilute ABVB phase) can penetrate the catalyst more readily than the average ones present in L2 (the more concentrated ABVB phase). This difference readily accounts for the greater pore area, pore volume loss, and coke content within pellets for the L1 vs L2 cases. Origin of Bimodal Pore Size Distribution. There is no significant difference between the reported surface areas and pore volumes when spent catalyst powder or

Impact of Multiphase Behavior on Coke Deposition

Energy & Fuels, Vol. 19, No. 4, 2005 1411

Figure 7. Comparison of (a) pore size distribution and (b) sorption isotherms for powdered and pellet coked catalyst samples from the 5 h, L1 + L2 dispersed case.

pellets from the 5 h L1 + L2 (dispersed) experiment are analyzed (Table 1 and Figure 7a). However, nitrogen uptake in catalyst pellets at low pressure (P/P0 < 0.3) is significantly less than for powdered catalyst (Figure 7b), and the sorption isotherms are functionally different. According to the classification of isotherm types,34 the isotherm for the powder is a type II isotherm, while the isotherm for the pellets shows type III isotherm behavior. Clearly, the external coke layer possesses a small mean pore size that hinders but does not prevent nitrogen diffusion into the catalyst. The secondary peak at 18 Å, only present in the coked pellets, is attributed to the external coke layer and not the catalyst per se. Conclusions The influence of multiphase behavior on coke deposition on and within commercial hydrotreating catalyst pellets under sedimentation conditions was explored by use of the model mixture Athabasca vacuum bottoms (ABVB) + decane. Compositions and operating conditions were defined such that individual catalyst pellets were exposed to targeted phases from the inception to the completion of experiments and that the ratio of the mass of liquid to the mass of catalyst in each phase exceeded 50:1. Under sedimentation conditions, the impact of phase behavior, per se, on the amount of coke deposited on and within catalyst pellets and the distribution of coke within catalyst pellets was found to be a secondary one despite the differences in the physical (40) Seyer, F. A.; Gyte, C. W. Viscosity. In AOSTRA Oil Sands Handbook; Hepler, L. G., Hsi, C., Eds.; Alberta Oil Sands Technology and Research Authority: Edmonton, Alberta, Canada, 1989; Chapter 7.

properties of the solvent-rich L1 phase and asphaltenerich L2 phase. In all cases, the exterior surfaces of pellets were coated with a thick nanoporous coke layer. The ratio of pore surface area to pore volume also increased with the extent of reaction in all cases, indicating that larger pores were filled in part or plugged in preference to smaller ones. Observed differences in mean pore size, pore surface area, and pore volume of coked catalyst exposed to the L1 and L2 phases and to multiphase environments are consistent with observed differences in the asphaltene aggregate size distribution in the two phases and to multiphase hydrodynamic effects. Acknowledgment. We thank Dr. Murray R. Gray of the Chemical and Materials Engineering Department, University of Alberta, for providing the catalyst and Dr. Xiangyang Zou for his help with phase behavior experiments. We gratefully acknowledge the sponsors of the NSERC Industrial Research Chair in Petroleum Thermodynamics (Alberta Energy Research Institute; Albian Sands Energy, Inc.; Computer Modeling Group, Ltd.; ConocoPhillips, Inc.; Imperial Oil Resources; NEXEN, Inc.; Natural Resources Canada; Petroleum Society of the CIMM; Oilphase-DBR, Oilphasesa Schlumberger Company, Schlumberger; Syncrude Canada, Ltd.; NSERC). Use of the BESSRC-CAT 12-ID-C SAXS beam line at the Advanced Photon Source (Argonne National Laboratories) was supported by the Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy (under Contract No. W-31-109-Eng-38). EF049802W