- Email: [email protected]
FCC Catalyst Deactivation: A Review and Directions for further Research
B. Delmon and G.F.Froment (Eds.) Calalyst Deactivarion 1994 Studies in Surface Science and Catalysis, Vol. 88 0 1994 Elsevier Science B.V. All rights reserved.
I29
FCC Catalyst Deactivation: A Review and Directions for further Research P. O'ConnoP and A.C. Pouwelsb "Akzo Catalysts, P.O.Box 247, 3800 AE, Amersfoort, The Netherlands bAkzo Catalysts, Nieuwendammerkade 1-3, 1022 AE, Amsterdam, The Netherlands 1. SUMMARY
Various forms of catalyst deactivation in FCC and the consequences for catalyst activity and selectivity are reviewed. The complexity of the inter-relationsbetween catalyst deactivation by aging, poisoning and fouling and the effect on heat balance and catalyst circulation rate are described. For catalyst poisoning at high metal contents as well as for hydrothermal aging with low metals, cyclic deactivation of the catalyst in various age fractions is the preferred route for a realistic simulation and quantification of these phenomena. The ability of a catalyst to rapidly deactivate the deposited poisons as nickel and vanadium is an important feature. Several factors contribute to the effect of coke on deactivation. The amount of hydrocarbons ('soft'' coke) entrained to the regenerator depends on the properties of the aged catalyst. Notwithstanding the extensive work carried out in this field, there is still ample scope for improvement in the quantification and modeling of FCC catalyst deactivation. 2. INTRODUCTION: SCOPE OF THE PAPER
The scope of this paper is to review the various forms in which deactivation of FCC Catalyst takes place. The consequences of deactivation on catalyst activity and selectivity are discussed and possible relations between the various deactivation phenomena are qualitatively indicated. A few cases of FCC catalyst deactivation are highlighted, specifically addressing the question how to simulate the deactivation phenomena properly, in order to study and estimate the impact on catalyst activity and selectivity.
130
3. DYNAMICS OF FCC CATALYST BEHAVIOUR
3.1. Forms of Deactivation Deactivation of FCC Catalysts does not only yield a drop in activity, but usually also a change in selectivity. Basically, three types of phenomena should be considered when studying the changes in catalyst activity and selectivity: (1) Catalvst Aaing
How does the catalyst change its behaviour in time. (2) Catalvst Poisoninq How do external poisons affect catalyst behaviour in time. (3) Catalvst Fouling How does formation of coke and/or metal deposits affect catalyst behaviour. There are several ways of classifying the various forms of catalyst poisoning [1,2,3]. In this paper we will focus on the difference between reversible (regenerable) and irreversible catalyst poisons in the FCC unit, as one of the typical features of the FCC operation is the continuous regeneration of the catalyst being circulated. The average catalyst goes through 10.000 up to 50.000 regeneration cycles. Table 1 Dynamics of FCC Catalyst Behaviour Deactivation:
Reversible
Irreversible
Catalyst Aging Catalyst Poisoning Catalyst Fouling
Coke, N, S, 0 (Polars) Coke deposits
HydrothermaI Na, V, Ni, etc. Metal deposits
To illustrate the complex inter-relations between the deactivation phenomena, we have made use of causal loop diagrams [4]. These diagrams can be used to integrate the various relations into an overall description of the FCC deactivation phenomena. 3.2. Hydrothermal Deactivation With amorphous silica-alumina catalysts (5, 61, the primary mode of aging involves steam-induced loss of surface area by the growth of the ultimate gel particles, resulting also in loss of porosity. While amorphous catalysts deactivate thermally as well as hydrothermally, thermal deactivation is a significantly slower process. The introduction of zeolites in cracking catalysts combined with various non-zeolite matrix types (a.0. higher stability silica-aluminatypes) certainly complicates the picture of FCC hydrothermal deactivation. Letzsch et a1 [7] have shown that like amorphous catalysts the zeolite is more strongly deactivated hydrothermallythan purely thermally.
131
Figure 1 shows the effect of steam during hydrothermal deactivation for a 1990's state-of-art medium RE,O, zeolite catalyst, containing also an active matrix contribution.
0.4 I 0
% Steam
T,deg C
*.
I
I
I
I
5
10 Time (hours)
15
20
0% 788
25%
h_
788
10% .-.*--. 25% 0% .--AL--. 50% 100% +.--*--. * 788
7aa
goo
goo
goo
Figure 1. Hydrothermal Deactivation
The first 10 to 25% of steam has the greatest influence. The zeolite unit cell size reduction, which should give an indication of the zeolite activity loss by dealumination [a] is not very sensitive to steam partial pressure, with the exception that some steam is necessary for cell size shrinkage. Chester et al [6] indicate that the relative contributions of zeolite deactivation (e.9.
loss of crystallinity) and matrix deactivation (e.g. loss of porosity) in different
temperature ranges can be significantly different. They therefore conclude that increasing temperature as a means of increasing catalyst steam deactivation severity can give misleading estimates of overall catalyst stability. This has also been confirmed with "today's'' FCC catalysts [lo]. As the relative contribution of zeolite and matrix activity will have an impact on catalyst selectivity, we can conclude that the foregoing is also valid for catalyst selectivity.
132
3.3. Reversible Catalyst Poisons and Deposits Examples are basic and polar molecules as for instance nitrogen compounds which are readily adsorbed on to the catalyst acidic sites, leading to an instantaneous, but temporary deactivation [ l , 21. Also polycyclic aromatics and other organic and non-strippable molecules which lead to coke formation are considered reversible (regenerable) catalyst poisons [ 111. If we assume that the poisoning effect will increase with the concentration of poisons on the catalyst, than the poisoning effect will be inversely proportional to the catalyst-to-oil ratio (CTO). Nitrogen poisoning of FCC catalyst [12] is often roughly correlated in this way.
The effect of reversible poisons will be dependent on the catalyst-to-oil, and therefore also of the coke selectivity of the catalyst and the heat balance of the FCC operation. 3.4. Irreversible Catalyst Poisons and Deposits These catalyst poisons (or deposits) can already start to influence the catalyst during the first passage through the reactor, but are not (easily) removed during the stripping and/or regeneration stages. Examples are the heavy metals in feed as vanadium and nickel and other poisons as for instance alkali components, iron and copper.
Irreversible catalyst poisons will often continue to interact with the total catalyst inventory of the FCC unit. If we assume that the poisoning effect will increase with the concentration of poisons on the catalyst [ 13,14,15], we can model this effect by for instance assuming steady state addition and removal of catalyst, see example Leuenberger 1141. The catalyst poisoning effect will then be proportional to the ratio between catalyst replacement and feed rate. Unfortunately, the metal level on FCC catalysts is hardly ever in equilibrium and as catalyst deactivation by vanadium does not take place in isolation, but combined with and influenced by hydrothermal deactivation [14, 151, more sophisticated dynamic equations will be needed to describe this behaviour also including the effects of the catalyst age distribution 115, 16, 171. Figure 2 and 3 give examples of causal loop diagrams for reversible and irreversible catalyst deactivation: Reversible deactivation is affecting hydrothermal deactivation, via the deterioration of the coke selectivity of the catalyst and hence higher regenerator temperatures. This can continue until the regenerator reaches a new equilibrium, because of the drop in catalyst activity.
133
7
Deactivation Conditions
++
Aging
I - Coke
Selectivity
Poisons on catalyst
\\
Catalyst to Oil Ratio
f
+ \ Poisons in feed
Figure 2. Dynamics of reversible deactivation Irreversible deactivation can have a similar effect on the hydrothermal deactivation by deteriorating coke selectivity (for instance for nickel poisoning). The hydrothermal deactivation on its turn will now also have an effect on the catalyst poisons, as for instance on the mobility of vanadium and on the deactivation of vanadium and nickel as dehydrogenation catalysts [2].
L----
Aging
6 +
\
Poisoning
Fouling
L
Deactivation Conditions
- Coke
Selectivity
Deactivation
7
Poisons on catalyst
7 Actili
-
Catalyst Replacement to Oil Ratio
y + \ Poisons in feed
Figure 3. Dynamics of irreversible deactivation
134
According to Yo0 [9] some of the vanadium poisoning is regenerable. While the poisoning effect of Vanadium on the FCC catalyst can at least be partially reversed, this type of regeneration does not take place in conventional FCC operations. 3.5. Catalyst Deposits: Fouling and Pore Mouth Plugging With the deposits of catalyst poisons as coke and heavy metals, fouling and pore mouth plugging phenomena can be observed [18, 191. Fouling can result in bigger differences in selectivity of various catalysts, because of changes in pore architecture [2, 101. The catalysts which are relatively less accessible for large hydrocarbons will be more sensitive to pore mouth blocking and plugging 1181. Khouw et al[20] report that catalysts contaminated to high Vanadium levels are still capable of converting light feeds, but not heavier feeds. This is illustrated in the following table. Table 2 Vanadium contamination has higher effect on conversion of residue feed Feedstock
Activity Loss in wt% Conversion per 1000 DDm V From [20] Own Data
VGO, CCR = 1 wt% RESID, CCR = 3-4 Wt% CCR: Conradson Carbon Residue in feed.
-1 3
0.7 1.8
135
Apparently catalyst poisons can block access for the larger hydrocarbon molecules to the most accessible sites. This is illustrated by the following model [lo]:
SUPPLY
DEMAND I
I
I
I
I
I
Figure 4. Supply and demand model of cracking
If a larger fraction of the sites are more accessible, the detrimental effect of poisons on the resid cracking selectivity will be less.
4. CASE OF HYDROTHERMAL AGING WITH LOW METALS Assuming that the metals and other poisons on catalyst are low, we can expect that traditional catalyst steaming will be sufficient to simulate catalyst deactivation. Keyworth et al [I61recommend to make a composite of several steamings in order to address the age distribution of equilibrium catalyst in a commercial unit. Beyerlein et al [17, 211 critically question the possibility of improving catalyst ageing procedures, which rely only on steam treatment at constant temperature for varying times. We find [lo, 221 that the decay behaviour of zeolite catalysts by steaming differs significantly from the activity and selectivity results after Cyclic Deactivation without metals. As described by Gerritsen et al [23]in the Cyclic Deactivation method the catalyst is deactivated by several Reaction and Regeneration (coke burning) cycles. As we will discuss in the next section of this paper, this is essential for the realistic aging of the metals.
136
The following example shows that even without metals, the catalyst seems to deactivate differently by Cyclic Deactivation compared to steaming. Table 3 Ranking changes dependent of deactivation conditions
ST
Method Catalyst A Conversion , Coke I
C4-olefinicity
Catalyst 6 Conversion , Coke C,-olefinicity 9
ST CD-1 CD-2
: : :
%wt %owt
67.7 2.1
0.64
72.5 2.5 0.60
68.5 3.4 0.61
%wt
67.0 2.2 0.69
72.7 2.6 0.60
69.2 3.9 0.57
%wt
Steaming 5 hours at 788"C, 100% steam Cyclic deactivation, 50 hrs, no metals Cyclic deactivation, 50 hrs, 1000 ppm Ni, 1000 ppm V
Table 3 is an example where a clear change in ranking is obtained with respect to C,-olefinicity (total C,= olefins/total C), and hence hydrogen transfer activity of the catalyst. A possible explanation for this is that while dealumination in a commercial unit is fast, migration of non-framework alumina from the zeolite structure will be a function of temperature and steam partial pressure [24]. This is an area in which the Cyclic Deactivation method approaches the commercial conditions much closer than traditional steaming methods. Also the presence of coke and coke burning in the regenerator stage can have an effect on the mobility and aging of the non-framework alumina species. This has been proposed to be the case for Vanadium [ I , 251. The foregoing combined with the observations made in 3.2 lead us to conclude that we need to try to simulate the deactivated catalyst as close as possible, preferably using the Cyclic Deactivation method. Short-cuts, even in the case of hydrothermal deactivation, can lead to critical errors in the performance ranking of FCC catalysts.
137
5. CASE OF CATALYST POISONING AND FOULING 5.1. The Poisoning Power of Heavy Metals and Other Compounds The literature on FCC catalyst deactivation by Vanadium and Nickel is extensive [l, 2, 10,13,14,26, 271. Basically Nickel and Vanadium influence the catalyst via three main reactions:
Mechanism
Metal Potency (V > Ni) (Ni > V)
(1) Destruction or neutralization of catalyst active sites
(2) Dehydrogenation reactions leading to coke and gas formation
(3) Oxidation promotion, leading to a higher COJCO ratio in
(Ni > V) (Ni > V)
the regenerator [28]
(4) Pore mouth blockage
The following table gives a rough impression of the relative poisoning power and dehydrogenation activity of some fresh compounds based on several literature sources available [lo, 26, 27,29, 30, 31,321. Table 4 Indications for Fresh Poisoning Power and Dehydrogenation Activity Relative Activitv Loss' per ppm weight
Relative Activitv Loss'' per ppm moles
V Ni Fe cu
1 .o 0.1
0.1
Na K Mg Ca Ba
0.9 0.9 0.5 0.5
C (Coke)"? N (Nitrogen)"?
Relative H, Production") per ppm weight
1 .o
0.3
0.1 0.1
0.3 0.4 < 0.1 < 0.1
0.1
2.0 1.2 1 .o 0.6 < 0.1
0.8 > 1.2
> 4.0
0.1
0.1
*) Defined as 1.O for Vanadium **) Defined as 1.0for Nickel ***) Very rough indications, for comparison only
0.2
1.o
< 0.1 < 0.1 < 0.1 < 0.1 < 0.1
138
Unfortunately, this information is not sufficient. The method in which the poisons are deposited [2, 23, 331 and the rate with which the poisons are (de)activated (dehydrogenation activity of Ni and V, or the mobility and acid site poisoning by Vanadium) have still to be taken into account. Consequently, we fully agree with Tatterson et al [33] who conclude that the FCC catalyst ability to rapidly deactivate the deposited metals will be an important factor in resid cracking. They find that Vanadium interacts with Nickel in a manner which inhibits the deactivation behaviour of Nickel. They therefore conclude that metals resistant cracking catalysts must be evaluated in the presence of both Nickel and Vanadium. We find that also the mobility of Vanadium is reduced by the presence of Nickel. A Cyclic Deactivation procedure will be preferred in order to simulate the actual metal distribution and interactions on the catalyst and the correct metal age distribution [2, 23, 27, 341. Furthermore, presence of SOX during the regeneration stage seems to be essential [35] as the SOX in the regenerator flue gas competes with Vanadium oxide in the reaction with certain compounds to non-mobile Vanadate species. There is only a limited amount of information on the deactivation mechanisms and rates of Vanadium and Nickel. The formation of metal silicates and/or aluminates have been proposed [26, 33, 34, 361, which seem to form more easily by reduction and oxidation cycles [37]. Taterson et al[33] indicate that the sites in which nickel is easily reduced are the sites in which nickel generates the most coke and find that Ni ions in tetrahedral sites are far less active than the ions in octahedral sites. Cheng et al [37] report that coke selectivity can also be correlated with the reducibility of Vanadium. Rajagopalan et al [38]confirm that methods involving cyclic redox aging of metals in the presence of sulphur are needed for screening metals tolerant catalyst. A simplified cyclic test (cyclic propylene steam) method is proposed which addresses the redox ageing of the metal, but not the non-uniform laydown and age distribution of metals on the catalyst.
Table 5: Comparison of Various Laboratory Metals Testing Procedures [38] Properties
Mitchell Method
Cyclic Metals Impregnation
Cyclic Propylene Steam
Deactivation Environment
Inert or oxidizing
Cyclic redox
Cyclic redox
Effect on V
V remains in +5 oxidation state where it is most mobile and most acidic. Severe zeolite attack.
V cycles between +5 and +3 oxidation states. Lower V mobility and less severe zeolite attack. V in the +3 state can react to form stable vanadates
V cycles between +5 and +3 oxidation states. Lower V mobility and less severe zeolite attack. V in the +3 state can react to form stable vanadates.
Effect on Ni
Ni remains in +2 oxidation Ni cycles between Ni (+2) and Ni (0) when alloy state. formation with Sb is possible.
Ni cycles between Ni (+2) and Ni (0)when alloy formation with Sb is possible.
Sb Passivation
Not always observed
Observed
Observed
Ratio of V/Ni Dehydrogenation Activity
0.6 (too high)
Ca. 0.25 (expected)
Ca. 0.25 (expected)
Sulfur Effect on V
Minimal
S competes for V trap adsorption sites
S injection currently under development
Metals Spatial Distribution
Uniform
Shell concentrated
Uniform
Metals Age Distribution
Uniform
Non-uniform
Uniform
Ease of Implementation
Easy
Tedious and difficult to scale-up
Easy
-
W
\o
140
Under actual FCC conditions, the penetration and age profile of the metals will and influence the efficiency of the catalyst metal trapping function [23,39,40] consequently catalyst activity and selectivitiy. Therefore, our opinion is that it is essential to simulate also the metal profile over the catalyst. Figure 5 illustrates relations which need to be taken into account in order to arrive at a correct description of the effect of heavy metals poisoning.
,
t+
Aging
f +
ip:
Deactivation Conditions
\
Poisoning Fouling
1
Active Metals
+/-
t
I
- Coke
Selectivity
Deactivation
7
Metals on
Actvi!;
2Catalyst Replacement
s
to Oil Ratio
Catalyst
Metals in feed
Figure 5. Dynamics of deactivation by metals
5.2. The Poisoning Power of Nitrogen From table 5 we observe that nitrogen in feedstock can have quite a big impact on the activity of a catalyst [lo, 321.The large effect of Nitrogen may be explained by blockage of a number of sites through coke formation, related to the adsorption of the nitrogen containing hydrocarbon molecules. Ho et al [41]show that the poisoning power of a nitrogen aromatic (polar) compound is primarily determined by a balance between its heaviness or size and basicity. The former may be measured by molecular weight, the latter by proton affinity. For example [lo]at end of run conditions of FCC pretreatment, the nitrogen left in feed can have a much higher poisoning power (smaller molecules containing nitrogen) than in the case of a non-treated nitrogen containing feedstock (nitrogen still in larger, less mobile molecules).
141
Formation of Coke Deposits Coke is a typical example of a reversible catalyst poison. The deactivation influence of coke depends very much on the nature of the coke, its structure and morphology and the exact location of its deposition on the catalyst surface [42, 43, 441. Coke formation follows the adsorption of coke precursors on the catalyst surface. The adsorption depends on the strength of the interaction and the volatility of the species. Polar Sulphur, Nitrogen and Oxygen containing compounds will tend to be adsorbed more strongly than neutral hydrocarbons [3]. Catalyst age distribution is also a factor here as the coke deactivation is more severe for the ralatively fresh catalyst, because of a larger surface area available for adsorption. 5.3. The
Various mechanisms of coke poisoning: active site coverage, pore filling as well as pore blockage have been observed in FCC [18, 19, 431 and Percolation theory concepts have been proposed for the modelling here of [45, 46, 47, 481. This approach provides a framework for describing diffusion and accessibility properties of randomly disordered structures. Basically, we could consider the FCC catalyst system as a combination of a shrinking core of sites not yet deactivated by coke and a progressing shell of large hydrocarbon molecules and metal contaminants, penetrating into the catalyst particle. The relative velocities of these fronts will be of great importance and will be strongly determined by the accessibility of the various functional sites of the catalyst [40].
enetration of reactants
of non-deactivated sites
Figure 6. Shell progression and shrinking core
142
Figure 7 summarizes the main relations, which determine the effect of coke on deactivation. Note that a poor coke selectivity (or low cat-to-oil ratio) will aggravate the poisoning effect of the fraction of the Conradson Carbon Residue, which is converted to coke. The amount of "soft" coke or hydrocarbons entrained to the regenerator without being stripped 13, 101 will have a significant effect on the overall coke selectivity and will depend @nthe surface area and pore size architecture of the aged catalyst [lo, 491.
-
Coke removed by stripping
adsorbed HC
feactio;
1-
Coke
Activity in Reactor
-
CCR on cat -Catalyst
to Oil Ratio
1\
CCR in feed
Figure 7. Dynamics of deactivation by coke
6. CONCLUDING REMARKS
Notwithstandingthe extensive work performed and voluminous literature in this field, we feel that there is still a need for further improvements in the quantification of the deactivation phenomena in FCC. Considering the many relationships and the complexity of the various mechanisms involved, a system dynamics approach [4] could be beneficial.
143
REFERENCES 1. J.L. Mauleon and J.B. Sigaud; "Characterization and selection of heavy feeds for upgrading through FCC", 12th WPC, Houston 1987, John Wiley & Sons Ltd. 2. P.O'Connor, A.W. Gevers, A. Humphries, L.A. Gerritsen and P.H. Desai; "Concepts for Future Residue Catalyst Development", ACS Symposium Series 452, Chapter 20, pg. 318, M.L. Occelli, editor, ACS Washington DC 1991. 3. E. Furimsky; Erdol und Kohle, Bd. 32, Heft 8, August 1979, pg. 383. 4. "Managerial Applications of Systems Dynamics"; MIT Press, Cambridge, Mass, 1986. E.B. Roberts, editor. 5. G.S. John and R.J. Mikovsky; Chem. Eng. Sci (1961) 15,161, 172. 6. A.W. Chester and W.A. Stover; Ind. Eng. Chem., Prod. Des. Dev. (1977) vol 16, no. 4. 7. W.S. Letzsch, R.E. Ritter and E.W. Vaughn; O&GJ, 130, January 1976, pg. 130. 8. L.A. Pine, P.J. Maher and W.A. Wachter; J. Catal (1984) 85,pg. 466. 9. J.S. Yoo, E.H. Burk, J.A. Karch and A.P. Voss; Ind. Eng. Chem. Res. (199O)a, pg. 1183. 10. P. O'Connor, A.C. Pouwels and J.R. Wilcox; "Evaluation of Resid FCC Catalysts" Symposium on Catalytic Cracking of Heavy Oils, paper 242E, 1992 AIChE, Annual Meeting, Miami Beach, 1-6 November 1992. 11. A. Voorhies; Ind. Eng. Chem. (1945) 37, pg. 318-322. 12. C. Fu and A.M. Schaffer; Ind. Eng. Chem. Prod. Res. Dev. (1985) 24, pg. 68-75. 13. D.F. Tolen; O&GJ, 30 March 1981, pg. 90. 14. E.L. Leuenberger; O&GJ, 15 July 1985, pg. 125. 15. D.J. Rawlence and K. Gosling; Catalysis Today (1991) 11pg. 47. 16. D.A. Keyworth, W.J. Turner and T.A. Reid; O&GJ, 14 March 1988, pg. 65. 17. R.A. Beyerlein, G.A. Tamborski, C.L. Marshall, B.L. Meyers, J.B. Hall and B.J. Higgins; "Monitoring FCC Deactivation Profile by Equilibrium Catalyst Separation", ACS Symposium Series 452, Chapter 8, pg. 109, M.L. Occelli, editor, ACS Washington DC 1991. 18. R. Man, F.Y.A. El-Kady and R. Marzin; Chem. Eng. Sci. (1985) 40, no. 2, 249. 19. P. O'Connor and F.W. van Houtert; Paper F-8, Akzo Catalyst Symposium 1986, The Netherlands, H.Th. Rijnten and H.J. Lovink, editors. 20. F.H.H. Khouw, M.J.R.C. Nieskens, M.J.H. Borley and K.H.W. Roebschlaeger; "The Shell Residue FCC Process Commercial Experiences and Future Developments", 1990 NPRA Annual Meeting, 25-27 March '90, paper AM-90-42. 21. R.A. Beyerlein, C.C. Feng, J.B. Hall, B.J. Higging and G.J. Ray; Symposium on Advances in FCC, ACS Chicago, August 1993, Paper in press. 22. P. O'Connor and A.C. Pouwels; "Realistic FCC Commercial Catalyst Testing in the Laboratory", 8th International Symposium on Large Chemical Plants, page 241, October 1992, Antwerpen, Belgium. Proceedings by Royal Flemish Society of Engineerings, G. Froment, editor. 23. L.A. Gerritsen, H.N.J. Wijngaards, J. Verwoert and P. O'Connor; Catalysis Today (1991) 11,pg. 61. 24. B.J. Meyers, T.H. Fleisch and C.L. Marshall; Applied Surface Science (1986), 26, pg. 503.
144
25. W.P. Hettinger, H.W. Beck, E.B. Cornelius, P.K. Doolin, R.A. Kmecak and S.M. Kovack; "Hydrothermal Screening of Reduced Crude Conversion Catalysts", Symposium on Advances in FCC, ACS Washington, August 1983, American Chem. SOC.Div. Petr. Chem. 28 (4), 920 (1983). 26. M.L. Occelli; "Metal Resistant FCC: Thirty Years of Research", ACS Symposium Series 452, Chapter 21, pg. 343, M.L. Occelli, editor, ACS Washington DC 1991. 27. A.W. Chester; Ind. Eng. Chem. Res. (1987) 26. pg. 863-869. 28. P.K. Doolin, J.F. Hoffman and M. Mitchell Jr.; Applied Catalysis (1991) 71, pg. 223. 29. L.A. Pine; J. Catal (1990) 125,pg. 514. 30. R.H. Nielsen and P.K. Doolin; "Metals Passivation" in Fluid Catalytic Cracking: Sciene and Technology Studies in Surface Science and Catalysis, vol. 76, 1993 Elsevier Science Publishes B.V. J.S. Magee and M.M. Mitchell Jr., editors. 31. W.S. Letzsch and D.N. Wallace; O&GJ, 29 November 1992, pg. 58. 32. J. Scherzer and D.P. Mac Arthur;"lnd. Eng. Chem. Res. (1988) 27, pg. 1571. 33. D.F. Tatterson and R.L. Mieville; "lnd. Eng. Chem. Res. (1988) 27, pg. 1595. 34. S.A. Roth, L.E. Iton, T.H. Fleisch, B.L. Meyers, C.L. Marshall and W.N. Delgass; J. Catal (1987) 108,pg. 214. 35. R.F. Wormsbecher, A.W. Peters and J.M. Maselli; J. Catal (1986) 100,pg. 130. 36. C. Lam, P. O'Connor and C.P. Smit; "The Advance Catalyst Series", Akzo Catalyst Symposium 1988, The Netherlands, H.J. Lovink, editor. 37. W.C. Cheng, M.V. Juskelis and W. Suarez; Applied Catalysis A. General (1993) 103, pg. 87. 38. Rajagopalan, W.C. Cheng, W. Suarez and C.C. Wear; "Resid FCC Catalyst Technology: Today and Future", 1993 NPRA Annual Meeting, March 1993, AM-93-53. 39. D.M. Stockwell, G.S. Koermer and W.M. Jaglowski; United States Patent, 5082814, 21st January 1992. 40. P. O'Connor and A.P. Humphries; "Accessibility of Functional Sites in FCC", Symposium on Advances in FCC, ACS Chicago, August 1993. American Chem. SOC.Div. Petr. Chem. (1993) 38,no. 3, pg. 598. 41. T.C. Ho, A.R. Katritzkyand S.J. Cato; Ind. Eng. Chem. Res. (1992)a, pg. 1589. 42. R.G. Menon; J. of Mol Catalysis (1990) 59,pg. 207. 43. J.W. Dean and D.B. Dadyburjor; lnd. Eng. Chem. Res. (1989) 28, pg. 271. 44. T.C. Ho; Ind. Eng. Chem. Res. (1992) 31, pg. 2281. 45. M.A. Uguina, D.P. Serrano, R. van Grieken and S. Venes; Applied Catalysis A: General (1993) !39, pg. 97. 46. G.F.A. Froment; "A quantitative Approach of Catalyst Deactivation by Coke Formation",Proceedings International Symposium on Catalyst Deactivation, Elsevier, Amsterdam 1980. 47. M.A. Sahimi and T.T. Tsotsis; J. Catal (1985) 96, pg. 552. 48. S.C. Reyes and L.E. Scriven; Ind. Eng. Chem. Res. (1991) 30,pg. 71. 49. P. O'Connor; "RFCC Operating Regimes and Catalyst Selection", paper to be published at Akzo Catalyst Symposium, June 1994, The Netherlands.
E.
I29
FCC Catalyst Deactivation: A Review and Directions for further Research P. O'ConnoP and A.C. Pouwelsb "Akzo Catalysts, P.O.Box 247, 3800 AE, Amersfoort, The Netherlands bAkzo Catalysts, Nieuwendammerkade 1-3, 1022 AE, Amsterdam, The Netherlands 1. SUMMARY
Various forms of catalyst deactivation in FCC and the consequences for catalyst activity and selectivity are reviewed. The complexity of the inter-relationsbetween catalyst deactivation by aging, poisoning and fouling and the effect on heat balance and catalyst circulation rate are described. For catalyst poisoning at high metal contents as well as for hydrothermal aging with low metals, cyclic deactivation of the catalyst in various age fractions is the preferred route for a realistic simulation and quantification of these phenomena. The ability of a catalyst to rapidly deactivate the deposited poisons as nickel and vanadium is an important feature. Several factors contribute to the effect of coke on deactivation. The amount of hydrocarbons ('soft'' coke) entrained to the regenerator depends on the properties of the aged catalyst. Notwithstanding the extensive work carried out in this field, there is still ample scope for improvement in the quantification and modeling of FCC catalyst deactivation. 2. INTRODUCTION: SCOPE OF THE PAPER
The scope of this paper is to review the various forms in which deactivation of FCC Catalyst takes place. The consequences of deactivation on catalyst activity and selectivity are discussed and possible relations between the various deactivation phenomena are qualitatively indicated. A few cases of FCC catalyst deactivation are highlighted, specifically addressing the question how to simulate the deactivation phenomena properly, in order to study and estimate the impact on catalyst activity and selectivity.
130
3. DYNAMICS OF FCC CATALYST BEHAVIOUR
3.1. Forms of Deactivation Deactivation of FCC Catalysts does not only yield a drop in activity, but usually also a change in selectivity. Basically, three types of phenomena should be considered when studying the changes in catalyst activity and selectivity: (1) Catalvst Aaing
How does the catalyst change its behaviour in time. (2) Catalvst Poisoninq How do external poisons affect catalyst behaviour in time. (3) Catalvst Fouling How does formation of coke and/or metal deposits affect catalyst behaviour. There are several ways of classifying the various forms of catalyst poisoning [1,2,3]. In this paper we will focus on the difference between reversible (regenerable) and irreversible catalyst poisons in the FCC unit, as one of the typical features of the FCC operation is the continuous regeneration of the catalyst being circulated. The average catalyst goes through 10.000 up to 50.000 regeneration cycles. Table 1 Dynamics of FCC Catalyst Behaviour Deactivation:
Reversible
Irreversible
Catalyst Aging Catalyst Poisoning Catalyst Fouling
Coke, N, S, 0 (Polars) Coke deposits
HydrothermaI Na, V, Ni, etc. Metal deposits
To illustrate the complex inter-relations between the deactivation phenomena, we have made use of causal loop diagrams [4]. These diagrams can be used to integrate the various relations into an overall description of the FCC deactivation phenomena. 3.2. Hydrothermal Deactivation With amorphous silica-alumina catalysts (5, 61, the primary mode of aging involves steam-induced loss of surface area by the growth of the ultimate gel particles, resulting also in loss of porosity. While amorphous catalysts deactivate thermally as well as hydrothermally, thermal deactivation is a significantly slower process. The introduction of zeolites in cracking catalysts combined with various non-zeolite matrix types (a.0. higher stability silica-aluminatypes) certainly complicates the picture of FCC hydrothermal deactivation. Letzsch et a1 [7] have shown that like amorphous catalysts the zeolite is more strongly deactivated hydrothermallythan purely thermally.
131
Figure 1 shows the effect of steam during hydrothermal deactivation for a 1990's state-of-art medium RE,O, zeolite catalyst, containing also an active matrix contribution.
0.4 I 0
% Steam
T,deg C
*.
I
I
I
I
5
10 Time (hours)
15
20
0% 788
25%
h_
788
10% .-.*--. 25% 0% .--AL--. 50% 100% +.--*--. * 788
7aa
goo
goo
goo
Figure 1. Hydrothermal Deactivation
The first 10 to 25% of steam has the greatest influence. The zeolite unit cell size reduction, which should give an indication of the zeolite activity loss by dealumination [a] is not very sensitive to steam partial pressure, with the exception that some steam is necessary for cell size shrinkage. Chester et al [6] indicate that the relative contributions of zeolite deactivation (e.9.
loss of crystallinity) and matrix deactivation (e.g. loss of porosity) in different
temperature ranges can be significantly different. They therefore conclude that increasing temperature as a means of increasing catalyst steam deactivation severity can give misleading estimates of overall catalyst stability. This has also been confirmed with "today's'' FCC catalysts [lo]. As the relative contribution of zeolite and matrix activity will have an impact on catalyst selectivity, we can conclude that the foregoing is also valid for catalyst selectivity.
132
3.3. Reversible Catalyst Poisons and Deposits Examples are basic and polar molecules as for instance nitrogen compounds which are readily adsorbed on to the catalyst acidic sites, leading to an instantaneous, but temporary deactivation [ l , 21. Also polycyclic aromatics and other organic and non-strippable molecules which lead to coke formation are considered reversible (regenerable) catalyst poisons [ 111. If we assume that the poisoning effect will increase with the concentration of poisons on the catalyst, than the poisoning effect will be inversely proportional to the catalyst-to-oil ratio (CTO). Nitrogen poisoning of FCC catalyst [12] is often roughly correlated in this way.
The effect of reversible poisons will be dependent on the catalyst-to-oil, and therefore also of the coke selectivity of the catalyst and the heat balance of the FCC operation. 3.4. Irreversible Catalyst Poisons and Deposits These catalyst poisons (or deposits) can already start to influence the catalyst during the first passage through the reactor, but are not (easily) removed during the stripping and/or regeneration stages. Examples are the heavy metals in feed as vanadium and nickel and other poisons as for instance alkali components, iron and copper.
Irreversible catalyst poisons will often continue to interact with the total catalyst inventory of the FCC unit. If we assume that the poisoning effect will increase with the concentration of poisons on the catalyst [ 13,14,15], we can model this effect by for instance assuming steady state addition and removal of catalyst, see example Leuenberger 1141. The catalyst poisoning effect will then be proportional to the ratio between catalyst replacement and feed rate. Unfortunately, the metal level on FCC catalysts is hardly ever in equilibrium and as catalyst deactivation by vanadium does not take place in isolation, but combined with and influenced by hydrothermal deactivation [14, 151, more sophisticated dynamic equations will be needed to describe this behaviour also including the effects of the catalyst age distribution 115, 16, 171. Figure 2 and 3 give examples of causal loop diagrams for reversible and irreversible catalyst deactivation: Reversible deactivation is affecting hydrothermal deactivation, via the deterioration of the coke selectivity of the catalyst and hence higher regenerator temperatures. This can continue until the regenerator reaches a new equilibrium, because of the drop in catalyst activity.
133
7
Deactivation Conditions
++
Aging
I - Coke
Selectivity
Poisons on catalyst
\\
Catalyst to Oil Ratio
f
+ \ Poisons in feed
Figure 2. Dynamics of reversible deactivation Irreversible deactivation can have a similar effect on the hydrothermal deactivation by deteriorating coke selectivity (for instance for nickel poisoning). The hydrothermal deactivation on its turn will now also have an effect on the catalyst poisons, as for instance on the mobility of vanadium and on the deactivation of vanadium and nickel as dehydrogenation catalysts [2].
L----
Aging
6 +
\
Poisoning
Fouling
L
Deactivation Conditions
- Coke
Selectivity
Deactivation
7
Poisons on catalyst
7 Actili
-
Catalyst Replacement to Oil Ratio
y + \ Poisons in feed
Figure 3. Dynamics of irreversible deactivation
134
According to Yo0 [9] some of the vanadium poisoning is regenerable. While the poisoning effect of Vanadium on the FCC catalyst can at least be partially reversed, this type of regeneration does not take place in conventional FCC operations. 3.5. Catalyst Deposits: Fouling and Pore Mouth Plugging With the deposits of catalyst poisons as coke and heavy metals, fouling and pore mouth plugging phenomena can be observed [18, 191. Fouling can result in bigger differences in selectivity of various catalysts, because of changes in pore architecture [2, 101. The catalysts which are relatively less accessible for large hydrocarbons will be more sensitive to pore mouth blocking and plugging 1181. Khouw et al[20] report that catalysts contaminated to high Vanadium levels are still capable of converting light feeds, but not heavier feeds. This is illustrated in the following table. Table 2 Vanadium contamination has higher effect on conversion of residue feed Feedstock
Activity Loss in wt% Conversion per 1000 DDm V From [20] Own Data
VGO, CCR = 1 wt% RESID, CCR = 3-4 Wt% CCR: Conradson Carbon Residue in feed.
-1 3
0.7 1.8
135
Apparently catalyst poisons can block access for the larger hydrocarbon molecules to the most accessible sites. This is illustrated by the following model [lo]:
SUPPLY
DEMAND I
I
I
I
I
I
Figure 4. Supply and demand model of cracking
If a larger fraction of the sites are more accessible, the detrimental effect of poisons on the resid cracking selectivity will be less.
4. CASE OF HYDROTHERMAL AGING WITH LOW METALS Assuming that the metals and other poisons on catalyst are low, we can expect that traditional catalyst steaming will be sufficient to simulate catalyst deactivation. Keyworth et al [I61recommend to make a composite of several steamings in order to address the age distribution of equilibrium catalyst in a commercial unit. Beyerlein et al [17, 211 critically question the possibility of improving catalyst ageing procedures, which rely only on steam treatment at constant temperature for varying times. We find [lo, 221 that the decay behaviour of zeolite catalysts by steaming differs significantly from the activity and selectivity results after Cyclic Deactivation without metals. As described by Gerritsen et al [23]in the Cyclic Deactivation method the catalyst is deactivated by several Reaction and Regeneration (coke burning) cycles. As we will discuss in the next section of this paper, this is essential for the realistic aging of the metals.
136
The following example shows that even without metals, the catalyst seems to deactivate differently by Cyclic Deactivation compared to steaming. Table 3 Ranking changes dependent of deactivation conditions
ST
Method Catalyst A Conversion , Coke I
C4-olefinicity
Catalyst 6 Conversion , Coke C,-olefinicity 9
ST CD-1 CD-2
: : :
%wt %owt
67.7 2.1
0.64
72.5 2.5 0.60
68.5 3.4 0.61
%wt
67.0 2.2 0.69
72.7 2.6 0.60
69.2 3.9 0.57
%wt
Steaming 5 hours at 788"C, 100% steam Cyclic deactivation, 50 hrs, no metals Cyclic deactivation, 50 hrs, 1000 ppm Ni, 1000 ppm V
Table 3 is an example where a clear change in ranking is obtained with respect to C,-olefinicity (total C,= olefins/total C), and hence hydrogen transfer activity of the catalyst. A possible explanation for this is that while dealumination in a commercial unit is fast, migration of non-framework alumina from the zeolite structure will be a function of temperature and steam partial pressure [24]. This is an area in which the Cyclic Deactivation method approaches the commercial conditions much closer than traditional steaming methods. Also the presence of coke and coke burning in the regenerator stage can have an effect on the mobility and aging of the non-framework alumina species. This has been proposed to be the case for Vanadium [ I , 251. The foregoing combined with the observations made in 3.2 lead us to conclude that we need to try to simulate the deactivated catalyst as close as possible, preferably using the Cyclic Deactivation method. Short-cuts, even in the case of hydrothermal deactivation, can lead to critical errors in the performance ranking of FCC catalysts.
137
5. CASE OF CATALYST POISONING AND FOULING 5.1. The Poisoning Power of Heavy Metals and Other Compounds The literature on FCC catalyst deactivation by Vanadium and Nickel is extensive [l, 2, 10,13,14,26, 271. Basically Nickel and Vanadium influence the catalyst via three main reactions:
Mechanism
Metal Potency (V > Ni) (Ni > V)
(1) Destruction or neutralization of catalyst active sites
(2) Dehydrogenation reactions leading to coke and gas formation
(3) Oxidation promotion, leading to a higher COJCO ratio in
(Ni > V) (Ni > V)
the regenerator [28]
(4) Pore mouth blockage
The following table gives a rough impression of the relative poisoning power and dehydrogenation activity of some fresh compounds based on several literature sources available [lo, 26, 27,29, 30, 31,321. Table 4 Indications for Fresh Poisoning Power and Dehydrogenation Activity Relative Activitv Loss' per ppm weight
Relative Activitv Loss'' per ppm moles
V Ni Fe cu
1 .o 0.1
0.1
Na K Mg Ca Ba
0.9 0.9 0.5 0.5
C (Coke)"? N (Nitrogen)"?
Relative H, Production") per ppm weight
1 .o
0.3
0.1 0.1
0.3 0.4 < 0.1 < 0.1
0.1
2.0 1.2 1 .o 0.6 < 0.1
0.8 > 1.2
> 4.0
0.1
0.1
*) Defined as 1.O for Vanadium **) Defined as 1.0for Nickel ***) Very rough indications, for comparison only
0.2
1.o
< 0.1 < 0.1 < 0.1 < 0.1 < 0.1
138
Unfortunately, this information is not sufficient. The method in which the poisons are deposited [2, 23, 331 and the rate with which the poisons are (de)activated (dehydrogenation activity of Ni and V, or the mobility and acid site poisoning by Vanadium) have still to be taken into account. Consequently, we fully agree with Tatterson et al [33] who conclude that the FCC catalyst ability to rapidly deactivate the deposited metals will be an important factor in resid cracking. They find that Vanadium interacts with Nickel in a manner which inhibits the deactivation behaviour of Nickel. They therefore conclude that metals resistant cracking catalysts must be evaluated in the presence of both Nickel and Vanadium. We find that also the mobility of Vanadium is reduced by the presence of Nickel. A Cyclic Deactivation procedure will be preferred in order to simulate the actual metal distribution and interactions on the catalyst and the correct metal age distribution [2, 23, 27, 341. Furthermore, presence of SOX during the regeneration stage seems to be essential [35] as the SOX in the regenerator flue gas competes with Vanadium oxide in the reaction with certain compounds to non-mobile Vanadate species. There is only a limited amount of information on the deactivation mechanisms and rates of Vanadium and Nickel. The formation of metal silicates and/or aluminates have been proposed [26, 33, 34, 361, which seem to form more easily by reduction and oxidation cycles [37]. Taterson et al[33] indicate that the sites in which nickel is easily reduced are the sites in which nickel generates the most coke and find that Ni ions in tetrahedral sites are far less active than the ions in octahedral sites. Cheng et al [37] report that coke selectivity can also be correlated with the reducibility of Vanadium. Rajagopalan et al [38]confirm that methods involving cyclic redox aging of metals in the presence of sulphur are needed for screening metals tolerant catalyst. A simplified cyclic test (cyclic propylene steam) method is proposed which addresses the redox ageing of the metal, but not the non-uniform laydown and age distribution of metals on the catalyst.
Table 5: Comparison of Various Laboratory Metals Testing Procedures [38] Properties
Mitchell Method
Cyclic Metals Impregnation
Cyclic Propylene Steam
Deactivation Environment
Inert or oxidizing
Cyclic redox
Cyclic redox
Effect on V
V remains in +5 oxidation state where it is most mobile and most acidic. Severe zeolite attack.
V cycles between +5 and +3 oxidation states. Lower V mobility and less severe zeolite attack. V in the +3 state can react to form stable vanadates
V cycles between +5 and +3 oxidation states. Lower V mobility and less severe zeolite attack. V in the +3 state can react to form stable vanadates.
Effect on Ni
Ni remains in +2 oxidation Ni cycles between Ni (+2) and Ni (0) when alloy state. formation with Sb is possible.
Ni cycles between Ni (+2) and Ni (0)when alloy formation with Sb is possible.
Sb Passivation
Not always observed
Observed
Observed
Ratio of V/Ni Dehydrogenation Activity
0.6 (too high)
Ca. 0.25 (expected)
Ca. 0.25 (expected)
Sulfur Effect on V
Minimal
S competes for V trap adsorption sites
S injection currently under development
Metals Spatial Distribution
Uniform
Shell concentrated
Uniform
Metals Age Distribution
Uniform
Non-uniform
Uniform
Ease of Implementation
Easy
Tedious and difficult to scale-up
Easy
-
W
\o
140
Under actual FCC conditions, the penetration and age profile of the metals will and influence the efficiency of the catalyst metal trapping function [23,39,40] consequently catalyst activity and selectivitiy. Therefore, our opinion is that it is essential to simulate also the metal profile over the catalyst. Figure 5 illustrates relations which need to be taken into account in order to arrive at a correct description of the effect of heavy metals poisoning.
,
t+
Aging
f +
ip:
Deactivation Conditions
\
Poisoning Fouling
1
Active Metals
+/-
t
I
- Coke
Selectivity
Deactivation
7
Metals on
Actvi!;
2Catalyst Replacement
s
to Oil Ratio
Catalyst
Metals in feed
Figure 5. Dynamics of deactivation by metals
5.2. The Poisoning Power of Nitrogen From table 5 we observe that nitrogen in feedstock can have quite a big impact on the activity of a catalyst [lo, 321.The large effect of Nitrogen may be explained by blockage of a number of sites through coke formation, related to the adsorption of the nitrogen containing hydrocarbon molecules. Ho et al [41]show that the poisoning power of a nitrogen aromatic (polar) compound is primarily determined by a balance between its heaviness or size and basicity. The former may be measured by molecular weight, the latter by proton affinity. For example [lo]at end of run conditions of FCC pretreatment, the nitrogen left in feed can have a much higher poisoning power (smaller molecules containing nitrogen) than in the case of a non-treated nitrogen containing feedstock (nitrogen still in larger, less mobile molecules).
141
Formation of Coke Deposits Coke is a typical example of a reversible catalyst poison. The deactivation influence of coke depends very much on the nature of the coke, its structure and morphology and the exact location of its deposition on the catalyst surface [42, 43, 441. Coke formation follows the adsorption of coke precursors on the catalyst surface. The adsorption depends on the strength of the interaction and the volatility of the species. Polar Sulphur, Nitrogen and Oxygen containing compounds will tend to be adsorbed more strongly than neutral hydrocarbons [3]. Catalyst age distribution is also a factor here as the coke deactivation is more severe for the ralatively fresh catalyst, because of a larger surface area available for adsorption. 5.3. The
Various mechanisms of coke poisoning: active site coverage, pore filling as well as pore blockage have been observed in FCC [18, 19, 431 and Percolation theory concepts have been proposed for the modelling here of [45, 46, 47, 481. This approach provides a framework for describing diffusion and accessibility properties of randomly disordered structures. Basically, we could consider the FCC catalyst system as a combination of a shrinking core of sites not yet deactivated by coke and a progressing shell of large hydrocarbon molecules and metal contaminants, penetrating into the catalyst particle. The relative velocities of these fronts will be of great importance and will be strongly determined by the accessibility of the various functional sites of the catalyst [40].
enetration of reactants
of non-deactivated sites
Figure 6. Shell progression and shrinking core
142
Figure 7 summarizes the main relations, which determine the effect of coke on deactivation. Note that a poor coke selectivity (or low cat-to-oil ratio) will aggravate the poisoning effect of the fraction of the Conradson Carbon Residue, which is converted to coke. The amount of "soft" coke or hydrocarbons entrained to the regenerator without being stripped 13, 101 will have a significant effect on the overall coke selectivity and will depend @nthe surface area and pore size architecture of the aged catalyst [lo, 491.
-
Coke removed by stripping
adsorbed HC
feactio;
1-
Coke
Activity in Reactor
-
CCR on cat -Catalyst
to Oil Ratio
1\
CCR in feed
Figure 7. Dynamics of deactivation by coke
6. CONCLUDING REMARKS
Notwithstandingthe extensive work performed and voluminous literature in this field, we feel that there is still a need for further improvements in the quantification of the deactivation phenomena in FCC. Considering the many relationships and the complexity of the various mechanisms involved, a system dynamics approach [4] could be beneficial.
143
REFERENCES 1. J.L. Mauleon and J.B. Sigaud; "Characterization and selection of heavy feeds for upgrading through FCC", 12th WPC, Houston 1987, John Wiley & Sons Ltd. 2. P.O'Connor, A.W. Gevers, A. Humphries, L.A. Gerritsen and P.H. Desai; "Concepts for Future Residue Catalyst Development", ACS Symposium Series 452, Chapter 20, pg. 318, M.L. Occelli, editor, ACS Washington DC 1991. 3. E. Furimsky; Erdol und Kohle, Bd. 32, Heft 8, August 1979, pg. 383. 4. "Managerial Applications of Systems Dynamics"; MIT Press, Cambridge, Mass, 1986. E.B. Roberts, editor. 5. G.S. John and R.J. Mikovsky; Chem. Eng. Sci (1961) 15,161, 172. 6. A.W. Chester and W.A. Stover; Ind. Eng. Chem., Prod. Des. Dev. (1977) vol 16, no. 4. 7. W.S. Letzsch, R.E. Ritter and E.W. Vaughn; O&GJ, 130, January 1976, pg. 130. 8. L.A. Pine, P.J. Maher and W.A. Wachter; J. Catal (1984) 85,pg. 466. 9. J.S. Yoo, E.H. Burk, J.A. Karch and A.P. Voss; Ind. Eng. Chem. Res. (199O)a, pg. 1183. 10. P. O'Connor, A.C. Pouwels and J.R. Wilcox; "Evaluation of Resid FCC Catalysts" Symposium on Catalytic Cracking of Heavy Oils, paper 242E, 1992 AIChE, Annual Meeting, Miami Beach, 1-6 November 1992. 11. A. Voorhies; Ind. Eng. Chem. (1945) 37, pg. 318-322. 12. C. Fu and A.M. Schaffer; Ind. Eng. Chem. Prod. Res. Dev. (1985) 24, pg. 68-75. 13. D.F. Tolen; O&GJ, 30 March 1981, pg. 90. 14. E.L. Leuenberger; O&GJ, 15 July 1985, pg. 125. 15. D.J. Rawlence and K. Gosling; Catalysis Today (1991) 11pg. 47. 16. D.A. Keyworth, W.J. Turner and T.A. Reid; O&GJ, 14 March 1988, pg. 65. 17. R.A. Beyerlein, G.A. Tamborski, C.L. Marshall, B.L. Meyers, J.B. Hall and B.J. Higgins; "Monitoring FCC Deactivation Profile by Equilibrium Catalyst Separation", ACS Symposium Series 452, Chapter 8, pg. 109, M.L. Occelli, editor, ACS Washington DC 1991. 18. R. Man, F.Y.A. El-Kady and R. Marzin; Chem. Eng. Sci. (1985) 40, no. 2, 249. 19. P. O'Connor and F.W. van Houtert; Paper F-8, Akzo Catalyst Symposium 1986, The Netherlands, H.Th. Rijnten and H.J. Lovink, editors. 20. F.H.H. Khouw, M.J.R.C. Nieskens, M.J.H. Borley and K.H.W. Roebschlaeger; "The Shell Residue FCC Process Commercial Experiences and Future Developments", 1990 NPRA Annual Meeting, 25-27 March '90, paper AM-90-42. 21. R.A. Beyerlein, C.C. Feng, J.B. Hall, B.J. Higging and G.J. Ray; Symposium on Advances in FCC, ACS Chicago, August 1993, Paper in press. 22. P. O'Connor and A.C. Pouwels; "Realistic FCC Commercial Catalyst Testing in the Laboratory", 8th International Symposium on Large Chemical Plants, page 241, October 1992, Antwerpen, Belgium. Proceedings by Royal Flemish Society of Engineerings, G. Froment, editor. 23. L.A. Gerritsen, H.N.J. Wijngaards, J. Verwoert and P. O'Connor; Catalysis Today (1991) 11,pg. 61. 24. B.J. Meyers, T.H. Fleisch and C.L. Marshall; Applied Surface Science (1986), 26, pg. 503.
144
25. W.P. Hettinger, H.W. Beck, E.B. Cornelius, P.K. Doolin, R.A. Kmecak and S.M. Kovack; "Hydrothermal Screening of Reduced Crude Conversion Catalysts", Symposium on Advances in FCC, ACS Washington, August 1983, American Chem. SOC.Div. Petr. Chem. 28 (4), 920 (1983). 26. M.L. Occelli; "Metal Resistant FCC: Thirty Years of Research", ACS Symposium Series 452, Chapter 21, pg. 343, M.L. Occelli, editor, ACS Washington DC 1991. 27. A.W. Chester; Ind. Eng. Chem. Res. (1987) 26. pg. 863-869. 28. P.K. Doolin, J.F. Hoffman and M. Mitchell Jr.; Applied Catalysis (1991) 71, pg. 223. 29. L.A. Pine; J. Catal (1990) 125,pg. 514. 30. R.H. Nielsen and P.K. Doolin; "Metals Passivation" in Fluid Catalytic Cracking: Sciene and Technology Studies in Surface Science and Catalysis, vol. 76, 1993 Elsevier Science Publishes B.V. J.S. Magee and M.M. Mitchell Jr., editors. 31. W.S. Letzsch and D.N. Wallace; O&GJ, 29 November 1992, pg. 58. 32. J. Scherzer and D.P. Mac Arthur;"lnd. Eng. Chem. Res. (1988) 27, pg. 1571. 33. D.F. Tatterson and R.L. Mieville; "lnd. Eng. Chem. Res. (1988) 27, pg. 1595. 34. S.A. Roth, L.E. Iton, T.H. Fleisch, B.L. Meyers, C.L. Marshall and W.N. Delgass; J. Catal (1987) 108,pg. 214. 35. R.F. Wormsbecher, A.W. Peters and J.M. Maselli; J. Catal (1986) 100,pg. 130. 36. C. Lam, P. O'Connor and C.P. Smit; "The Advance Catalyst Series", Akzo Catalyst Symposium 1988, The Netherlands, H.J. Lovink, editor. 37. W.C. Cheng, M.V. Juskelis and W. Suarez; Applied Catalysis A. General (1993) 103, pg. 87. 38. Rajagopalan, W.C. Cheng, W. Suarez and C.C. Wear; "Resid FCC Catalyst Technology: Today and Future", 1993 NPRA Annual Meeting, March 1993, AM-93-53. 39. D.M. Stockwell, G.S. Koermer and W.M. Jaglowski; United States Patent, 5082814, 21st January 1992. 40. P. O'Connor and A.P. Humphries; "Accessibility of Functional Sites in FCC", Symposium on Advances in FCC, ACS Chicago, August 1993. American Chem. SOC.Div. Petr. Chem. (1993) 38,no. 3, pg. 598. 41. T.C. Ho, A.R. Katritzkyand S.J. Cato; Ind. Eng. Chem. Res. (1992)a, pg. 1589. 42. R.G. Menon; J. of Mol Catalysis (1990) 59,pg. 207. 43. J.W. Dean and D.B. Dadyburjor; lnd. Eng. Chem. Res. (1989) 28, pg. 271. 44. T.C. Ho; Ind. Eng. Chem. Res. (1992) 31, pg. 2281. 45. M.A. Uguina, D.P. Serrano, R. van Grieken and S. Venes; Applied Catalysis A: General (1993) !39, pg. 97. 46. G.F.A. Froment; "A quantitative Approach of Catalyst Deactivation by Coke Formation",Proceedings International Symposium on Catalyst Deactivation, Elsevier, Amsterdam 1980. 47. M.A. Sahimi and T.T. Tsotsis; J. Catal (1985) 96, pg. 552. 48. S.C. Reyes and L.E. Scriven; Ind. Eng. Chem. Res. (1991) 30,pg. 71. 49. P. O'Connor; "RFCC Operating Regimes and Catalyst Selection", paper to be published at Akzo Catalyst Symposium, June 1994, The Netherlands.
E.