1438

Energy & Fuels 2002, 16, 1438-1443

Catalyst Deactivation during Hydroprocessing of Maya Heavy Crude Oil. 1. Evaluation at Constant Operating Conditions
J. Ancheyta,*,, G. Betancourt, G. Centeno, G. Marroqu n, F. Alonso, and E. Garciafigueroa
Instituto Mexicano del Petro leo, Eje Central La zaro Ca rdenas 152, Me xico 07730 D.F., Mexico, and Instituto Polite cnico Nacional, ESIQIE, Me xico 07738 D.F., Mexico Received February 25, 2002

A deactivation study during hydroprocessing of Maya heavy crude oil is reported. Experiments were carried out in fixed-bed pilot plant at the following constant operating conditions: pressure of 70 Kg/cm2, Hydrogen-to-oil ratio of 5000 ft3/bbl, space velocity of 1.0 h-1, and reaction temperature of 400 C. Deactivation of catalyst was followed by monitoring the changes in asphaltenes, sulfur, metals (V and Ni), Rambottom carbon, and nitrogen contents in the products. The catalyst was characterized before and after reaction. Results indicate that catalyst deactivation is very fast during the first hours of the run. Coke and metals depositions were around 18 and 10%, respectively. Accumulation of vanadium was higher in the lobes of the catalyst.

1. Introduction Deactivation of catalysts is one of the main concerns during hydroprocessing of petroleum fractions. The profitability of this process depends very often on the lifetime of the catalyst. For light fractions (naphtha and middle distillates), catalyst deactivation is minimal, but for heavy fractions (residua and heavy crude oil), catalysts deactivation can be severe, resulting in short periods of operation. Loss of active sites is the main reason for catalyst deactivation, which is mainly due to active site poisoning by strongly absorbed species, active site coverage by deposits (coke and metals), pore mouth constriction/ blockage, and sintering of the active phase. Each of these causes have been subject of many investigations, which were recently summarized and deeply discussed by Furimsky and Massoth.1 Three deactivation periods are identified during catalytic hydroprocessing of oil fractions: (1) deactivation by coke, which is formed from virtually all feeds employed, and in the case of heavy oils is mainly caused from asphaltenes; (2) deactivation by metals, which is irreversible and whose rate depends on the metals level in the feed; and (3) pore constriction and blockage, characterized by a very strong loss in activity, after which the plant has to stop operation. The first two periods of deactivation are obviously the most important ones to study, since they define the life of the catalyst. In fact, the primary catalyst deactivation causes in hydroprocessing of petroleum distillates have
* To whom correspondence should be addressed. Fax: (+52-55) 3003-8429. E-mail: jancheyt@imp.mx. Instituto Mexicano del Petro leo. Instituto Polite cnico Nacional. (1) Furimsky, E.; Massoth, F. E. Catal. Today 1999, 52, 381.

been considered to be the accumulation of carbonaceous and metallic deposition.2,3 It is well-documented in the literature that coke is formed very rapidly during the first hours of time-onstream, and catalyst deactivation by this carbonaceous material appears to rapidly reach a pseudo steady-state level, while deactivation by the metals takes a longer time period. Both deactivation mechanisms during hydroprocessing, which occur simultaneously if metals are present in the feed, have been studied by other researchers, who have employed atmospheric or vacuum residua for conducting their experiments. Most of these studies have been reported in the Catalyst Deactivation Symposia issues. However, deactivation of catalyst during hydroprocessing of heavy crude oils has not received too much attention. The nature of the metals (mainly V and Ni), which are usually distributed between porphyrin and nonporphyrin type of structures, and coke deposited on the catalyst depends on the origin of the feed. Then, each feed will exhibit different deactivation patterns. For this reason we conducted an experimental evaluation about catalyst deactivation in the hydroprocessing of Maya heavy crude oil. The main objective of the study is to evaluate changes in hydroprocessed product quality during time-on-stream as well as modifications in catalyst properties. 2. Experimental Section
Deactivation studies were carried out in a high-pressure pilot plant. The reactor operates in isothermal mode and
(2) Absi-Halabi, M.; Stanislaus, A.; Trimm, D. F. Appl. Catal. A 1991, 72, 193. (3) Callejas, M. A.; Mart nez, M. T.; Blasco, T.; Sastre, E. Appl. Catal. A 2001, 218, 181.

10.1021/ef020045g CCC: $22.00 2002 American Chemical Society Published on Web 08/24/2002

Hydroprocessing of Maya Heavy Crude Oil

Table 1. Properties of Maya Crude Oil properties API gravity sulfur, wt % nitrogen, wppm Ramsbottom carbon, wt % asphaltenes in nC7, wt % V, wppm Ni, wppm Maya crude oil 20.9 3.44 3700 10.5 12.4 299 55

Energy & Fuels, Vol. 16, No. 6, 2002 1439

Pressure is then reduced from operating value (70 kg/cm2) to atmospheric and temperature is decreased from 40 C to ambient. When these conditions are reached, catalyst is unloaded and immediately transferred to a clean and dry vessel. The vessel has to be with nitrogen bubbling before and after catalyst unloading. Finally, the vessel is hermetically closed. The spent catalyst was washed with toluene and dried at 80 C before characterization. Catalysts were analyzed before and after hydroprocessing reaction by using the following methods: Metal content (nickel and vanadium) was determined by atomic absorption spectrometer Specific surface area, pore volume and pore size distribution were measured by nitrogen adsorption at 77 K. Carbon content was measured by combustion with an infrared detector. Scanning electron microscopy of spent catalyst was performed in a JSM-35 CF JEOL model electron microscope working at 20 kV. The microscope is equipped with a Sigma 2 Kevex model disperse energy system. The samples were deposited on a carbon holder and covered with a thin carbon film.

consists of a tube with an inside diameter of 2.54 cm and a total length of 143 cm. Experiments were conducted in oncethrough hydrogen in down-flow mode of operation. More details about this plant, catalyst loading, experimental procedure, and catalyst sulfiding were reported elsewhere.4,5 Maya crude oil was used for deactivation experiments. The main properties of this heavy crude oil are shown in Table 1. The catalyst employed during the tests was a Ni-Mo tetralobe commercial sample (2.54 mm diameter). Its properties will be shown and discussed in next sections together to properties of used catalyst. The catalyst loaded to the reactor (75 mL) was in situ activated by sulfiding with a hydrodesulfurized naphtha containing 0.8 wt % CS2 at the following operating conditions: pressure of 54 kg/cm2, H2-to-oil ratio of 2000 ft3/bbl, temperature of 230 C, and a liquid-hourly space velocity (LHSV) of 3.2 h-1. The sulfiding time was 18 h. Deactivation study was carried out at the following constant operating conditions: 400 C of reaction temperature, 70 kg/ cm2 of total pressure, 5000 ft3/bbl of hydrogen-to-oil ratio, and 1.0 h-1 of LHSV. Product quality was monitored during the 490 h time-onstream. Product samples were recovered every 50 h for determining physical and chemical properties. In some cases, products were also recovered 8 h after the 50 h sample (i.e., 58 h, 108 h, etc.) in order to check repeatability of analyses. In addition, to determine the initial activity of the catalyst, a liquid product sample was taken at 4 h of time-on-stream. Physical and chemical properties of the feed and products were determined with the following methods: total sulfur, ASTM D-4294 total nitrogen, ASTM D-4629 Ramsbottom carbon, ASTM D-524 asphaltenes in nC7, ASTM D-3279 metals (Ni and V), ASTM D-5863 For our study it was considered to be very important to determine changes in properties of a catalyst when it finishes its operation. However, sometimes not too much care is taken during catalyst unloading procedure, which is very crucial since catalyst properties may change if it is not unloaded correctly. The following procedure was employed in this work for catalyst unloading: Once the last experiment has finished, hydroprocessing feedstock was changed to another lighter stream without stopping hydrogen flowrate. In this case we changed Maya heavy crude oil to straight-run gas oil. It is very important that the catalyst never operates without liquid flow. Reactor temperature is reduced from operating value (400 C) to 200 C at a rate of 30 C/h. When 200 C temperature is achieved, the feed (SRGO) is again changed to another lighter feed without stopping hydrogen flow. Desulfirized naphtha was used for this purpose. Reactor cooling is continued from 200 to 40 C at a rate of 20 C/h. When temperature is of 40 C, naphtha flow is stopped and at the same time hydrogen is changed to nitrogen to have an inert atmosphere.
(4) Ancheyta, J.; Betancourt, G.; Marroqu n, G.; Pe rez, A.; Maity, S. K.; Cortez, M. T.; del R o, R. Energy Fuels 2001, 15, 120. (5) Ancheyta, J.; Maity, S. K.; Betancourt, G.; Centeno, G.; Rayo, P.; Go mez, M. T. Appl. Catal. A 2001, 216, 195.

3. Results and Discussion 3.1. Results of Product Quality. Variations of sulfur, vanadium, and nickel contents, and Ramsbottom carbon (RBC), asphaltenes, and nitrogen contents in hydroprocessed products against time-on-stream are shown in Figures 1 and 2, respectively. Samples analyzed after 8 h of the every 50 h sample indicate a very good repeatability of experimental results. It is clearly observed that deactivation of catalyst is very rapid during first hours of the run, which is attributed to losses of specific area and pore volume due to coke formation. Initial percentages of removal of RBC, asphaltenes, S, N, V, and Ni were 38.8, 54.6, 71.8, 45.9, 56.5, and 35.9%, respectively. In the first 50 h of time-of-stream, RBC, asphaltenes, S, and N contents increased very sharply and hence conversions decreased up to 17.7, 38.9, 50, and 34.5%, respectively. During this period, V and Ni exhibited a lower reduction in conversion. From 50 to 100 h of run, all contents continued increasing with more or less constant rates. After 100 h, RBC, S, N, and V contents maintain this increase but at a lower rate. However, asphaltenes and Ni contents reached a constant value. This is not surprising since asphaltenes, which are very large polyaromatic compounds, are the major coke precursors. This confirms that deposition of coke is the main source of catalyst deactivation during first hours of run, and metals deposition is lower than coke deposition during this period. It is also seen from Figures 1 and 2 that after 100 h of run, asphaltene conversion and consequently coke deposition are drastically reduced while metals, especially V, continue being deposited at a low but constant rate. Nickel and vanadium removals showed different behavior during time-on-stream. At about 200 h, Ni content reached a constant value while V continues increasing. It indicates that the rate of metal deposition varies from metal to metal. This can be attributed to the differences in Ni- and V-porphyrin type of structure, since in the case of vanadium a perpendicular oxygen

1440

Energy & Fuels, Vol. 16, No. 6, 2002

Ancheyta et al.

Figure 1. Vanadium, nickel, and sulfur contents in product as function of time-on-stream.

Figure 2. Ramsbottom carbon, asphaltene, and nitrogen contents in product as function of time-on-stream.

atom is linked to it, which forms strong bond with the catalyst surface, and nickel does not have this oxygen links. Another explanation can be that the rate of Ni deposit formation is strongly influenced by formation of V deposits. It should also be mentioned that initial V content is more than five times higher than Ni content (299 and 55 wppm, respectively), and this difference also affects metals deposition on the catalyst. From the above observations and results showed in Figures 1 and 2 both coke and metal depositions are clearly distinguished. Carbon deposition is the main cause of catalyst deactivation during first hours of the run, and losses of specific area and pore volume are observed. After this period of deactivation, metal deposition is more important than coke deposition.

3.2. Textural Analyses of Fresh and Spent Catalysts. After the catalyst was unloaded, various analyses were done which are shown in Table 2. Properties of fresh catalyst are also included in this table for comparison. Specific area of catalyst decreased from 175 to 88 m2/g while loss of pore volume was from 0.56 to 0.33 cm3/g, which means around 50% and 40% of reduction, respectively. Deposition of coke in early period of run and consequently blockage of pores is the main responsible of these reductions. This decrease in properties agrees with those reported in the literature.6,7 Bulk density increased due to metals and coke deposits.
(6) Thakur, D. S.; Thomas, M. G. Ind. Eng. Chem. Prod. Res. Dev. 1984, 23, 349.

Hydroprocessing of Maya Heavy Crude Oil

Table 2. Textural Properties of Fresh and Spent Catalysts fresh catalyst physical properties specific surface area, m2/g pore volume, cm3/g mean pore diameter, bulk density, g/cm3 pore size distribution, vol % <100 100-250 250-500 > 500 175 0.56 127 0.72 9.8 69.1 15.0 6.1 spent catalyst 88 0.33 152 1.041 31.1 63.7 4.5 0.7

Energy & Fuels, Vol. 16, No. 6, 2002 1441

Figure 4. Possible blockage mechanism of pores during hydroprocessing. Void circles represent unblocked pores, and black crowns indicate covered pores.
Table 3. Carbon and Metals Deposition on Spent Catalyst physical properties molybdenum, wt % nickel, wt % sodium, wt % vanadium, wt % iron, wt % carbon, wt % sulfur, wt % fresh catalyst 10.66 2.88 0.041 0 0 0 0 spent catalyst 5.56 2.64 0.13 4.41 0.082 18.3 8.56

Figure 3. Pore size distributions of fresh (s) and spent (- -) catalysts.

Pore size distributions of fresh and spent catalysts are also shown in Figure 3. The differences between catalyst before and after hydroprocessing reactions are clearly observed. Percentage of pore volume having pore diameter higher than 500 exhibited a very important reduction (88%) and these pores almost disappeared completely. Similar behavior but with slightly less reduction (70%) was observed with pores in the range of 250-500 . Percentages of pore volume having pore diameter 100250 remained almost unchanged since they only decreased from 69.1 to 63.6% (8% of reduction). Smaller pores (<100 ) exhibited a notable increase in their percentages from 9.8 to 31.1% (more than 3 times), which is mainly due to the reduction of the percentage of pore volumes having pore diameters higher than 100 , mainly those higher than 250 . This behavior in catalyst pore size distribution suggests that during pore blockage large pores may be totally or partially covered. The partially covered pores then became smaller pores. For instance, most of the pores having pore diameter higher than 500 are totally covered as can be seen in Table 2. Part of them may be blocked partially, and they contribute to increase the percentage of pores having 250-500 pore diameter, which at the same time suffer the same blockage behavior. In the case of 100-250 pores, some of them are formed from large pores and others are covered to form smaller pores. The smallest
(7) Song, C.; Nihonmatsu, T.; Nomura, M. Eng. Chem. Res. 1991, 30, 1726.

pores (<100 ) are not blocked relatively, and they are only formed from large pores. This is the reason for showing an increase in their percentage. We can imagine these changes in pore size distribution of catalyst as a typical consecutive-parallel reaction scheme as shown in Figure 4. Hence, metals and coke deposits not only block the pore mouth but also help to increase pores with smaller size. As consequence of all these modifications in pore size distribution mean pore diameter of catalyst showed an increase from 127 to 152 . This increase in MPD has also been reported in the literature.8-10 3.3. Metals and Coke Deposition. Carbon content was used as a measure of the amount of coke on the catalyst. Then, coke deposited on the catalyst was of 18.3 wt % as can be observed in Table 3. This coke deposition is mainly formed from asphaltenes. It has been reported in the literature that coke is deposited up to 25% of the weight of the original catalyst during hydrotreating of residual fractions.11 Integration of the amounts of metals removed from the feed as function of time-on-stream can be calculated and are shown in Figure 5. The total amount of metals deposited on the catalyst during 490 h time-on-stream on the basis of fresh catalyst was around 10%.
(8) Kobayashi, S.; Kushiyama, S.; Aizawa, R.; Koinuma, Y.; Keiichi, I.; Shimizu, Y.; Egi, K. Ind. Eng. Chem. Res. 1987, 26, 2245. (9) Beuther, H.; Larson, O. A.; Perrotta, A. J. In Catalysis Deactivation; Delmon, B., Froment, G. F., Eds.; Elsevier: Amsterdam, 1980; p 271. (10) Nakamura, N.; Togari, O.; Ono, T. In Proceedings of the 45th API Midyear Meeting, Houston, TX, May 1980; Vol. 201, p 22. (11) Wiwel, P.; Zeuthen, P.; Jacobsen, A. C. In Catalysis Deactivation Bartholomew, C. H., Butt, J. B., Eds.; Elsevier: Amsterdam, 1991; p 257.

1442

Energy & Fuels, Vol. 16, No. 6, 2002

Ancheyta et al.

Figure 5. Metals deposition on the catalyst during hydroprocessing of Maya crude oil.

Figure 7. Intraparticle V profiles of spent catalyst.

Figure 6. Zones for determination of V profiles in tetra-lobe spent commercial catalyst.

With metals content in fresh catalyst and those deposited on spent catalyst reported in Table 3 and with density values of both samples reported in Table 2, a metals balance can be done and then the percentage of metals on fresh-basis catalyst can also be determined, which was very close to 10%. This confirms that for the

490 h time-on-stream the metal deposit in catalyst is around this value. It should be mentioned that asphaltenes are condensed polynuclear aromatic systems containing different heteroatoms, among them metals are located inside the molecule. During hydroprocessing asphaltenes are converted to smaller molecules and metals (nonporphyrin and porphyrin types) are also transformed and deposited into catalyst causing catalyst deactivation. Because of their large size, these molecules do not penetrate deeply into the catalyst. They are accumulated as metal sulfides into the pore mouth of the catalyst and block the way to enter the reactants. During hydrodemetallization two extreme cases of metal deposition are presented: (1) for large pore catalysts, molecules can penetrate deep into the particle interior, and (2) for small pore catalysts, metal deposition occurs predominantly on the external surfaces. The first case ensures that more metals can be accommodated because pore-mouth plugging is minimized, but such catalysts are less active because less active surface is available. The second case leaves a large portion of the active surface available for reaction, but this is not easy to achieve because of significant pore-mouth plugging.12 To determine how metals are being deposited on the catalyst used in our deactivation experiment we conducted a scanning electron microscopy analysis. It should be remembered that the catalyst is a tetralobe commercial sample. Then, two profiles were examined considering the longest (profile A) and shortest (profile B) extremes of the particle as can be seen in Figure 6.
(12) Furimsky, E. Appl. Catal. A 1998, 171, 177.

Hydroprocessing of Maya Heavy Crude Oil

Energy & Fuels, Vol. 16, No. 6, 2002 1443

Figure 7 shows the results of vanadium profiles for the two selected zones. It is observed that these profiles are between the two extreme cases of metals deposition mentioned earlier, and that they are different between them. Profile A has higher V content on the external surface compared to that of profile B. This means that there is more V accumulation in the lobes of the catalysts. At the interior of the particle, V tends to deposit almost in similar way in both zones. It can be finally mentioned that this commercial catalyst shows a more or less uniform profile of vanadium which is highly desirable during hydroprocessing of heavy crude oils, because this pattern results in maximum metals storage capacity without considerable pore mouth plugging. Conclusions The following conclusions can be raised from our present study: Deactivation of catalyst during hydroprocessing of Maya crude oil is very rapid during first hours of timeon-stream due mainly to coke deposition.

Removals of nickel and vanadium showed different behavior during time-on-stream, which was attributed to an oxygen atom linked to vanadium and to the influence of V on rate of Ni deposit. Similar losses of specific area and pore volume of catalyst (40-50%) were found with respect to literature reports. Analyses of pore size distribution suggested that during pore blockage large pores may be totally or partially covered contributing to increase smaller pores percentage. Coke and metals depositions were around 18 and 10%, respectively, which agree with reports in the literature. Vanadium accumulation was higher in the lobes of the commercial catalyst.

Acknowledgment. The authors thank Instituto Mexicano del Petro leo for its financial support.
EF020045G