MATRIX EFFECTS IN CATALYTIC CRACKING
Lance D. Silverman,
Steven Winkler, Technical Service Engineer
Jack A. Tiethof, Group Leader
Anatol Witoshkin, Manager, Catalyst Technical Service
Edison, New Jersey
Presented at the
March 23-25, 1986
Westin Bonaventure Hotel
Los Angeles, California
The FCC catalyst matrix plays a significant role in determining catalytic performance as well as affecting heat transfer in the FCC unit and imparting important physical properties to the catalyst. This paper focuses on the effects of the matrix on catalytic performance. Because large feed molecules cannot readily diffuse into zeolite pores, primary cracking reactions on the catalyst matrix make a major contribution to upgrading bottoms to light cycle oil. This is illustrated by the strong correlation of improved bottoms upgrading with increasing matrix surface area in both laboratory and commercial data. Gasoline octane and middle distillate cetane are also significantly increased by an active matrix, due to low hydrogen transfer of matrix cracking and to the cracking of saturates into light cycle oil from bottoms. The matrix also plays an important role of trapping feed contaminants. Results are discussed which show the effect that matrix has on contaminant metals and on cracking feeds containing basic nitrogen.
The matrix of an FCC catalyst serves both physical and catalytic functions. Physical functions include providing particle integrity and attrition resistance, acting as a heat transfer medium, and providing a porous structure to allow diffusion of hydrocarbons into and out of the catalyst microspheres. The matrix can also affect catalytic selectivity, product qualities, and resistance to poisons. The nature of these catalytic functions will be the focus of this paper.
The paper will present and discuss laboratory data obtained on a large number of commercial FCC catalysts in order to show catalytic effects of the matrix. In addition, commercial data are presented which support the conclusions of the laboratory investigations. Because the commercial catalysts studied vary widely in their method of manufacture and in numerous properties, the effects of the matrix will only stand out where the matrix has a very strong influence on overall catalytic properties. This tends to be true for reactions which directly involve large molecules, where the influence of zeolitic cracking is limited by diffusion, such as primary cracking of large molecules or the depositing of metals from asphaltenes. This is of increasing importance to refiners because of the need to upgrade low value bottoms and to process resid.
The laboratory evaluations of FCC catalysts presented were usually performed using a fixed bed microactivity test (MAT) (Ref. 1, 2). In some instances, a circulating pilot unit (Ref. 3) was used to collect sufficient quantities of products for product quality studies. Details of these tests are described in the references cited. Prior to MAT testing, catalysts were steam deactivated at temperatures between 1350 and 1550°F for four hours in a fluid bed of 100% steam. MAT activity, i.e., conversion/(100-conversion), was varied by changing steaming temperature. Regression analysis of each selectivity versus activity allowed determination of product yield at a constant conversion of 70 weight percent. Cut points of 421 and 602°F were used for gasoline and light cycle oil (LCO), respectively. Catalyst activities are compared after steam deactivation at 1450°F. Methods for treatment of catalysts with contaminant metals have been described in detail elsewhere (Ref. 1).
A wide boiling range mid-continent gas oil was used for all laboratory studies, except where noted. In addition, a high nitrogen feed containing coker gas oil was used for feed contaminant studies. Pertinent properties are presented in Table I.
The circulating pilot unit has the major features of a commercial unit, with short contact time riser cracking and continous regeneration. Catalysts were steam deactivated at 1450°F prior to testing, with steaming time adjusted to yield 70% MAT conversion at constant severity of operation. Products were separated by conventional distillation for product quality analyses.
PONA analyses were determined on the gasoline and LCO fractions. Also, Carbon-13 nuclear magnetic resonance (NMR) spectra were taken on these fractions as well as the bottoms fractions. This complements PONA data (which could not be measured on the bottoms fractions) and measures the percent carbon atoms which are contained in aromatic rings (Ref. 4). NMR spectra were taken on a Varian XL-400 instrument.
Matrix surface areas of commercial catalysts were determined by the T-plot method(Ref. 5, 6).
Results and Discussion
1. Matrix Effects on Product Selectivities
The catalyst matrix is capable of cracking large molecules which cannot readily diffuse into zeolite pores. The resulting fragments are small enough to enter the zeolite pores. Thus, the matrix can play an important role in upgrading bottoms. Both the amount of matrix surface area in a catalyst and the composition of the matrix contribute to its activity for cracking. The amount and strength of acid sites on a silica-alumina matrix, which are responsible for its cracking activity, are associated with its alumina content. The matrix surface area and alumina content of those commercial catalysts which we have investigated generally vary in the same direction, as shown in Figure 1. Because of this, it would be very difficult to separate the effects of surface area from surface acidity from these results.
The effect of an active matrix on upgrading bottoms to LCO is seen from both laboratory and commercial data. Figure 2 shows bottoms and LCO yields at 70% conversion as a function of matrix surface area from MAT tests on commercial FCC catalysts. The effectiveness of the matrix in reducing bottoms yield and increasing LCO is clearly evident and has been found to occur on other feeds as well. Another example will be discussed in the section on feed contaminants.
The upgrading properties of catalysts with active matrices have been observed in commercial operation, as illustrated in Figures 3 a and b. At Refinery A, a high matrix catalyst (fresh matrix area > 150 m2/g) replaced a low matrix surface area product (fresh surface area ca. 50 m2/g). As observed, the bottoms yield decreased (bottoms selectivity improved) in direct proportion to the increase of the high matrix surface area catalyst in the circulating catalyst inventory. The reduction in bottoms yield approached 40% and was evident over a wide range conversions (Fig 3b). The decrease in the bottoms resulted in equal volumetric increase in the light cycle oil yield. Additional observations illustrating the bottoms upgrading by active matrix catalysts are presented in Table II.
Cracking sites have been attributed to the external surfaces of zeolite crystals, and these also can crack molecules which are too large to readily enter zeolite pores (Ref. 7). This is an important path with pure zeolite materials for cracking molecules significantly larger than zeolite pores. Increasing the amount of external zeolite surface by reducing zeolite crystal size has been suggested as a means of improving bottoms upgrading on a resid feed, using a catalyst with a very inactive matrix(Ref. 8). The relative benefits of matrix and external zeolite surface cracking of large molecules cannot be separated from available data on commercial catalysts because the relative rates of cracking on these surfaces are not known. However, calculations show that external zeolite surfaces contribute less than 10 m2/g of surface to an equilibrium commercial catalyst (i.e., for 10% zeolite content with crystals larger than 0.1 microns), while matrix surface areas can be over 100 m. The relative benefits of matrix and external zeolite surface cracking of large molecules cannot be separated from available data on commercial catalysts because the relative rates of cracking on these surfaces are not known. However, calculations show that external zeolite surfaces contribute less than 10 m2/g for moderate and high matrix surface area catalysts. Thus, cracking on zeolite external surfaces is more important with catalysts having a very inactive matrix.
Although matrix cracking is a major contributor to bottoms cracking, it has a much smaller effect on other selectivities. This is because with smaller molecules, which can readily enter zeolite pores, zeolite cracking is the dominant route and masks the smaller effect of the matrix. It has been shown, for example, that the relative rates of cracking on zeolite, compared to amorphous silica-alumina, increase as molecular size of the feed decreases(Ref. 9). Figures 4 a-c presents laboratory MAT data which shows that there is no apparent correlation of other selectivities (coke, C4- gas, and gasoline) with matrix surface area. Although matrix cracking probably does make some contribution to coke and gas make, this is overshadowed by other factors.
2. Matrix Effects on Product Qualities
In addition to its effects on product yields, matrix cracking also affects the product qualities of both gasoline and LCO. Gasoline octane is improved by a high olefin content and a high concentration of aromatics. On the other hand, properties which contribute to high LCO cetane (high paraffin content) are opposite to those which contribute to high gasoline octane(Ref. 10).
Matrix cracking improves gasoline octane because hydrogen transfer is reduced compared to zeolitic cracking(Ref. 11). In part, this is because hydrogen transfer reactions are bimolecular, and their rate is enhanced by concentration of reactants in zeolite pores and by acid sites in close proximity to one another, The hydrogen transfer reaction of interest is depicted in Equation 1 and leads to saturation of olefins which reduces gasoline octane. Data in Table III compares product qualities from FCC pilot tests on high rare earth catalysts with low and medium activity matrices. It can be seen that the gasoline from the active matrix catalyst has both a higher gasoline octane and a higher olefin content by PONA analysis.
Napthenes + Olefins ----> Aromatics + paraffins Eq. 1
While the shift in olefin and paraffin content between the gasolines produced by these two catalysts is seen, the data does not show the corresponding shift in naphthenes and aromatics predicted by Eq. 1. This may be because olefins, which occur primarily in gasoline and lighter products, are saturated at the expense of naphthenes in heavier products. This is difficult to assess, since these products are difficult or impossible to analyze by conventional means, eg., naphthene content of bottoms or coke, and because matrix cracking has other effects on the composition of these products (see below).
The pilot data in Table III shows that the catalyst with the active matrix produced LCO with higher cetane, as well as gasoline with higher octane. Earlier work(Ref. 11, 12) has shown that an improvement in gasoline octane and LCO cetane occurs simultaneously when hydrogen transfer is reduced by substituting a hydrogen zeolite for a rare earth zeolite. In this case the two catalysts had very similar matrices, so that differences in matrix cracking did not play a role. For the non-rare earth catalyst, low hydrogen transfer is probably raising cetane by minimizing naphthene conversion to aromatics in LCO and raising octane by minimizing olefin conversion to paraffins in gasoline.
While low hydrogen transfer may explain some of the cetane benefit of increased matrix activity, it appears that a more significant effect of an active matrix is the preferential cracking of aliphatic material (paraffin and napthene molecules or fragments) from bottoms into LCO. This is evident from NMR data showing a much lower aliphatic carbon content in the bottoms from the catalyst with the more active matrix. The preferential cracking of aliphatic compounds (or fragments) reflects the fact that aromatic structures are very resistant to cracking. As the catalyst with the active matrix cuts deeper into bottoms at constant conversion, some of the aliphatic compounds cracked from the bottoms end up in the LCO fraction and raise its fuel quality. This is consistent with the higher aliphatic content of the LCO from the active matrix catalyst and the corresponding higher LCO cetane.
Increases in octane which are attributed to catalysts with increased matrix activity have been observed in commercial units. Examples are shown in Table IV. There are no significant changes in factors other than the catalyst matrix (e.g. operating severity, feed changes, zeolite unit cell size or rare earth content) in these examples.
Simultaneous improvements in the gasoline octane and light cycle oil cetane index were observed in refinery B. as illustrated in Figure 5. At this refinery a REY zeolite catalyst with moderate matrix surface area (125 m2/g) replaced one with a low matrix surface area. The reduced hydrogen transfer and selective cracking of high boiling range molecules associated with matrix cracking resulted in a gasoline octane improvement of approximately two RON and light cycle oil cetane index increase of almost three numbers. The changes in qualities were proportional to the increase in the matrix surface area in the circulating inventory. This simultaneous catalytic improvement of both octane and cetane is related to catalytic properties in this operation, since most operational changes that improve octane (e.g. increased conversion, increased feed aromaticity) will depress the cetane value.
As the cetane of the light cycle oil improves with the higher matrix surface area catalyst, the yield and properties of bottoms are also affected, as can be seen from the pilot data in Table III. Note that the higher matrix surface area catalyst produced more aromatic bottoms with lower API gravity. The change in bottoms quality is illustrated for Refinery C in Figure 6. A low matrix surface area catalyst replaced a high matrix product. The bottoms yield increased steadily as the matrix surface area declined. Simultaneously the bottoms API gravity also increased. The higher API gravity indicated that lighter, more paraffinic molecules were being left uncracked in the bottoms product from the low matrix catalyst. Additional commercial data illustrating improvements in the light cycle oil qualities seen with increases in matrix activity are summarized in Table V.
3. Matrix Effects on Feed Contaminant Tolerance
Both metals and basic nitrogen compounds poison FCC catalysts and are concentrated in the heavier end of gas oils, especially in the residuum(Ref. 13). The large molecules containing these poisons deposit on the matrix which can then bind or "trap" the poisons. Each type of poison affects FCC catalysts differently. Vanadium causes permanent deactivation of the catalyst(Ref. 1) while nitrogen causes only temporary deactivation which is reversed by regeneration. Nickel, on the other hand, causes little or no deactivation, but is deleterious to product selectivities.
Vanadium, which deposits on catalyst particles in the riser as organometallic compounds, is converted to vanadium oxides in the regenerator. In the regenerator environment, vanadium can migrate to zeolite crystals and form a low melting eutectic with the silica-alumina of the zeolite. This leads to destruction of the crystal structure and loss in activity(Ref. 14). The binding of vanadium to catalyst matrix inhibits the vanadium from reaching the zeolite and thus minimizes deactivation(Ref. 1). Laboratory data on commercial catalysts which demonstrate this effect are shown in Figure 7. It illustrates how the activity maintenance of catalysts (aged with 4500 ppm vanadium, 4 hrs., 1450°F steam) correlates with their matrix surface areas, at least up to a moderate value of 120-130 m2/g. Certainly, the surface chemistry is important as well. It is reported that alumina materials are effective at trapping vanadium(Ref. 15). However, the alumina in this case is probably not acting as an acidic material in trapping the vanadium. In fact other materials which are clearly basic, i.e., calcium compounds(Ref. 16), magnesium compounds(Ref. 17), and rare earth compounds(Ref. 18), have been shown to be effective at trapping vanadium. Although it is possible that acid-base chemistry as well as the formation of high melting species of vanadium are involved in trapping of vanadium by the matrix, the mechanism may be more complicated. In Figure 8 vanadium tolerance data is shown for three catalysts with nearly the same high alumina, matrix composition (all are REY catalysts). Clearly, the catalysts with moderate and high matrix surface areas have superior vanadium tolerance, and the loss is greatest for the low matrix surface area catalyst.
Commercial support for the vanadium tolerance effects of an active matrix is shown in Figure 9. In this resid cracking operation at Refinery D, a low matrix surface area catalyst replaced one with a moderate matrix surface area. The catalysts were similar in other respects, such as matrix alumina content and zeolite type. Over a period of weeks as the catalyst inventory changed, the equilibrium catalyst MAT activity gradually declined by 5 numbers. The activity decline closely followed the decline in catalyst matrix surface area. Unit operating conditions, fresh catalyst makeup, and equilibrium catalyst vanadium loading were essentially constant during the catalyst changeout period. The refinery then changed back to the previously used moderate surface area catalyst and the activity trend reversed itself.
When nickel deposits on FCC catalysts it acts as a dehydrogenation catalyst which contributes to coke and gas production. The FCC catalyst matrix may act as a catalyst support for nickel, so that a high matrix surface area could increase the harmful effects of nickel. However, the use of antimony passivators(Ref. 18, 19) largely offsets the effects of nickel and greatly reduces the importance of nickel effects relative to vanadium.
The heavy ends of gas oils and residua contain both basic and non-basic nitrogen compounds. The basic nitrogen compounds are mainly responsible for poisoning of FCC catalysts because poisoning occurs by the reaction of the basic nitrogen species with acid sites on the catalyst. The neutralization of catalytically active acid sites results in deactivation of the catalyst. Nitrogen poisons also affect selectivities deleteriously(Ref. 21). In addition, nitrogen in feed is associated with asphaltenes and other coke precursors, which contribute to catalyst poisoning.
The acid sites on the matrix of an FCC catalyst act to intercept nitrogen poisons and provide "sacrificial sites" which protect zeolite in the catalyst from poisoning. Thus, the matrix can reduce the deactivation effects of nitrogen. Figure 10 illustrates the effectiveness of an active catalyst matrix at reducing deactivation by basic nitrogen in feed. The ratio of activity of the steamed catalyst on a high nitrogen feed to activity on the mid-continent feed is plotted against matrix surface area. The higher matrix surface area catalysts, which have high alumina contents, are far more resistant to poisoning.
Figure 11a shows that an active matrix still plays its major selectivity role of helping to upgrade bottoms to LCO on the high nitrogen feed. Laboratory data also shows an advantage in other selectivities (gas, gasoline, and coke) for active matrix catalysts, as is shown in Figure 11b for gasoline.
The high nitrogen feed used in this investigation is more aromatic and refractory than the mid-continent feed (properties in Table 1). In order to distinguish the effects of basic nitrogen from other differences in the feeds, experiments were performed using the mid-continent feed doped with quinoline, a basic nitrogen compound having a relatively low molecular weight. Quinoline, which is known to be a potent poison(Ref. 21), was added to the mid-continent feed to yield the same basic nitrogen content as the high nitrogen feed. Table VI compares activity losses caused by quinoline to losses caused by the high nitrogen feed for three catalysts. These are two low matrix surface area catalysts with different alumina contents and a moderate matrix surface area catalyst. Quinoline caused roughly half of the deactivation which occurred with the high nitrogen feed. Thus, half of the deactivation caused by the high nitrogen feed is associated with other feed properties, such as higher aromaticity and coke forming tendencies. This data suggests that the active matrix helps protect the catalyst from deactivation due to these other factors as well. The fact that the activity loss caused by quinoline is less for the catalyst with the higher surface area matrix is consistent with protection of zeolite from basic nitrogen by the matrix. Lastly, the difference in alumina content between the low matrix catalysts shows only a small effect.
Table VII summarizes two cases in which high nitrogen feedstocks were being cracked in commercial FCC units. Both FCC units operated at high unit severity in order to compensate for catalyst deactivation due to nitrogen poisoning. In the first case, Refinery E used a gas oil containing 0.37 wt.% total nitrogen and 0.14 wt.% basic nitrogen. The conversion was 59.4 vol.% when a low matrix surface area catalyst was used. The conversion rose by 0.6 vol.% after a change was made to a high matrix surface area catalyst, even though operating conditions changed in a direction which should have resulted in lower conversion. This included a reduction in catalyst make up rate on the higher matrix surface area catalyst. The table shows unit conversion corrected to account for changes in operating conditions using a computer model. The corrected data shows that the high matrix surface area catalyst results in 4.6 vol.% higher conversion at equal operating conditions and indicates that the high matrix surface area catalyst exhibited a significant resistance to nitrogen deactivation. In the second case, Refinery F used a gas oil containing 0.23 wt.% total nitrogen with a low matrix surface area catalyst, followed by a high matrix surface area catalyst. In this FCC unit the high surface area catalyst was responsible for raising the conversion by 2.5 vol.% after correcting for differences in cracking severity. As in the previous case, the improved conversion can be attributed to the nitrogen tolerance provided by the high surface area matrix. The improvement in nitrogen tolerance is not as great in Case 2 because the nitrogen level in the feedstock is lower.
In conclusion, the matrix of an FCC catalyst significantly influences bottoms upgrading, liquid product qualities, and feed contaminant tolerance. Matrix cracking increases LCO yield while reducing bottoms by cracking large molecules. Both gasoline octane and LCO cetane are improved by matrix cracking because of low hydrogen transfer and selective cracking of paraffinic portions of the bottoms. Finally, the catalyst matrix traps vanadium and basic nitrogen contaminants in the feed to reduce catalyst deactivation.
1. B. K. Speronello and W. J. Reagan, Oil Gas J., 82(5), 139 (1984)
2. D. B. Bartholic and R. P. Haseltine, NPRA Annual Meeting,Paper No. AM81-45 (1981)
3. W. H. Humes, Chem. Eng. Prog., Feb. 1983, p. 51
4. T. Yoshida, Y. Maekawa, H. Uchino, and S. Yokoyama, Anal.Chem., 52, 817 (1980)
5. B. C. Lippens and J. H. de Boer, J. Catal., 4, 319 (1965)
6. M. F. L. Johnson, J. Catal., 52, 425 (1978)
7. C. L. Thomas and D. S. Barmby, J. Catal., 12, 341 (1968)
8. J. M. Maselli and A. W. Peters, Catal. Rev. Sci. Eng., 26,525 (1984)
9. D. M. Nace, Ind. Eng. Chem. Prod. Res. Dev., 9, 203 (1970)
10. J. M. Collins and G. H. Unzelman, Oil Gas J., 80(22), 87(1982)
11. W. J. Reagan, G. M. Wolterman, and S. M. Brown, Div. of Petr. Chem., A.C.S. Symposia, 28(4), 884 (1983)
12. Unpublished results
13. C. L. Hemmler and W. L. Vermilion, Oil Gas J., 71(45), 88 (1973)
14. L. Upson and S. Jaras, NPRA Annual Meeting, March 1982
15. M. Occelli and J. V. Kennedy, UK Patent Application GB 2,116,062A (1983)
16. B. Mitchell and R. Vogel, US Patent 4,451,355 (1984)
17. M. Occelli and J. V. Kennedy, U.S. Patent 4,465,588 (1984)
18. H. Beck, C. Lochow, and C. Nibert, U.S. Patent 4,515,683 (1985)
19. G. H. Dale and D. L. McKay, Hydrocarbon Proc., 56(9), 97 (1977)
20. D. L. McKay, U.S. Patent 4,025,458 (1977)
21. C.-M. Fu and A. M. Schaffer, Ind. Eng. Chem. Prod. Res. Dev.,24, 68 (1985)
*[Technical information and data regarding the composition, properties, or use of the products described herein is believed to be reliable. However, no representation or warranty is made with respect thereto. BASF Corporation cannot assume responsibility for any patent liability which may arise from the use of any product in a process, manner or formula not designed by BASF.]