MODERN CRACKING CATALYST AND RESIDUE PROCESSING CHALLENGES
Dr. Gerald M. Woltermann, Sr. Research
Dr. Glenn Dodwell, Research Associate
Dr. Bruce Lerner, Staff Chemist
Iselin, New Jersey
Presented at the
March 17-19, 1996
San Antonio, Texas
BASF has introduced two new FCC catalysts designed for the RESID market. In lab testing MILLENNIUM and ULTRIUM both exhibit excellent selectivity and stability in the presence of contaminant metals and both are excellent in upgrading bottoms to usable products. Commercial experience confirms these lab results. In this paper, we will present both lab and commercial data comparing the two catalysts with state of the art competitive resid catalysts. We will also present microscopy, spectroscopic and porosimetry data which elucidates the mechanisms by which the two catalysts passivate nickel and vanadium.
Refineries are under increasing pressure to process a variety of feeds including those with high boiling point components and high levels of contaminant metals. Processing such feeds makes increasing demands on the catalyst to passivate contaminant metals and crack higher boiling range molecules, while maintaining the activity/stability and selectivity required of a modern FCC catalyst. Specifically, resid catalysts need to produce low coke and preserve zeolite integrity in the presence of significant concentrations of nickel and vanadium.
Nickel and vanadium usually occur in petroleum feedstocks as porphyrins(Ref. 1) or other organic salts. Porphyrins are highly conjugated organometallic compounds which will either thermally crack or condense to coke upon laying down upon the catalyst surface. The bare metal is then free to migrate further into the microsphere or transfer to other particles. Such metals on the surface of refractory oxides can promote unwanted side reactions resulting in coke and hydrogen formation and can result in destruction of zeolite and active matrix.
The two major metal contaminants behave differently and are deleterious to catalyst performance in disparate ways. Nickel can exist in both an oxidized (+2 valence state) and reduced (0 valence state) form. The latter is more active in the promotion of dehydrogenation reactions leading to coke and hydrogen formation and is more mobile within a catalyst particle. The mobility leads to agglomeration of the metal during the many reduction oxidation cycles experienced in an FCC unit. Agglomeration reduces the available nickel surface and decreases its activity for dehydrogenation. The oxidized form of nickel is more likely to remain immobile and embed itself in an oxide structure such as alumina forming solid state compounds such as spinel . Formation of these solid state oxides greatly decreases the kinetics of metal reduction to the zero valent state. Cadet et. al. (Ref. 2) have shown that nickel aluminate can also catalyze dehydrogenation reactions to produce coke and hydrogen. Thus, catalysts with high surface area alumina matrices are more likely to provide excellent nickel supports maximizing metal activity. In contrast to nickel, vanadium can exist in several oxidization states between +3 and +5. It does not reduce to the zero valence state under FCC conditions and does not have the propensity to sinter to large particles as is the case with nickel. Even at vanadium levels greater than 2%, formation of crystalline V205 has not been observed(Ref. 3) .Vanadium, unlike nickel, can transfer between particles, either by vapor phase transfer of vanadic acid or by solid solid transfer during particle collisions(Ref. 4,5). Vanadic acid is highly mobile due to its high vapor pressure at FCC temperatures and is known to form low melting eutectics with a number of solid state compounds. Because of these properties, even moderate levels of vanadium concentration can cause severe loss in zeolite and active matrix surface area. In the +5 oxidation state vanadium is a moderate promoter of dehydrogenation reactions(Ref. 6).
Because of the different behavior of the two metals, strategies for their passivation can differ significantly. In the case of nickel, for instance, one can trap the nickel in a solid state compound making it unavailable for reduction to the zero valent state and inaccessible or inactive for subsequent dehydrogenation reactions. Antimony addition for example ties up nickel as relatively inactive nickel antimonate. Another strategy would be to provide a surface for the nickel on which it is both easily reduced and highly mobile. Such a surface would allow formation of large nickel/nickel oxide particles which are relatively inactive for dehydrogenation reactions(Ref. 7). Vanadium on the other hand is best passivated by adding metal oxides which will readily form metal vanadate. Metal vanadate formation reduces metal mobility while at the same time preventing the vanadium from catalyzing dehydrogenation reactions. Rare earth compounds, as well as alkaline earth compounds, can perform this task. The stabilities of the initial metal oxides and product metal vanadates are critical to the viability of this method of passivation. Another tactic to suppress the metal activity would involve absorption of vanadium onto a refractory oxide surface. This support would then sinter, burying the vanadium and preventing dehydrogenation reactions or attack on the zeolite or active matrix.
Test methodology of FCC catalysts for metals tolerance is important in evaluating eventual commercial performance. It is not possible to exactly mimic the metals lay down and dispersion mechanism of a commercial FCC unit in lab testing because of time constraints and scale. Several methods have been developed over the years to try to measure a catalysts ability to passivate metals. Perhaps the simplest is the Mitchell or modified Mitchell method developed at Gulf Research(Ref. 8). In this technique, which has many variations from lab to lab, a dried catalyst sample is impregnated with a metal vanadate or octoate compound dissolved in cyclohexane, xylene, or toluene. The amount of solution used is enough to impregnate the catalyst to its puddle point, that is, the point where additional drops will result in formation of a puddle of solvent. The solvent is then evaporated at room temperature, carbonized at 600 F and the organic burned off at 1150 F leaving the bare metal deposited on the catalyst. The catalyst is then steamed under some partial pressure of oxygen before evaluation. The method is simple, quick and reproducible. However, it results in too much metals activity, unrealistic dispersion of the nickel and virtually none of the many redox cycles observed in a real unit. In the cracked-on method, a feed doped with some level of metal naphthenate is cracked on the catalyst over a number of oxidation reduction cycles, usually between 20- 50. The metallated catalyst is then steamed under partial oxygen pressure and evaluated. This method is more time consuming and less reproducible. However, it has advantage of subjecting the metals to repeated oxidation-reduction cycles and better mimics commercial metals activity and distribution. Young has reviewed various metallation techniques and their value. He suggests in some cases that Mitchell method ranking of catalysts can be reversed from the rankings determined by cracked-on evaluation. We use both methods in our evaluations.
Results and Discussion
1. TPR Results
Figure 1 shows temperature programmed reduction (TPR) results for nickel on pure silica, ULTRIUM and MILLENNIUM. In TPR, the materials are calcined in air and then reduced under hydrogen or other reducing gases as a function of temperature. The higher the temperature required for reduction, the less amenable the supported metal is to reduction. Clearly, the MILLENNIUM traps nickel in sites that inhibit its reduction by hydrogen, while nickel on silica reduces quite easily at moderate temperatures. ULTRIUM seems to have nickel present in both types of sites and has both easily reduced and non-reducible nickel. Bolt, et. al. , have shown reduction temperatures of nickel in the 850-900 C region are characteristic of nickel aluminates with the nickel in the tetrahedral holes of the spinel structure.
2. Elemental Mapping Studies
Microtoned slices of metal impregnated microspheres of MILLEMNIUM and ULTRIUM were subjected to elemental mapping in the electron microscope. These results confirmed and expanded the observations from the TPR. On MILLENNIUM, after Mitchell method impregnation, the nickel and vanadium are dispersed throughout the particle and are highly concentrated in areas of high alumina concentration. Mitchell method metals impregnation tends to overdisperse nickel as compared to the cyclic deposition which occurs in commercial units. However, similar results were observed for commercial equilibrium catalyst (Ecat) samples of MILLENNIUM containing 9000+ ppm of nickel with only slight surface concentration of nickel. ULTRIUM, on the other hand, shows significant nickel concentration and particle size on the silica rich surface of the catalyst even after Mitchell method impregnation. Vanadium is associated with magnesium, which is present in the vanadium trap component of ULTRIUM. The catalysts interact with nickel and vanadium in significantly different ways.
3. Passivation Mechanism
Both the TPR and elemental mapping show MILLENNIUM to selectively adsorb both metals onto its alumina surface. However, as pointed out earlier, sorption onto an alumina rich surface is not enough for good passivation. MILLENNIUM contains a chemically treated large pore alumina rich matrix component. The chemical treatment facilitates the chemisorption of both nickel and vanadium. After adsorption and under the influence of steam, the large pores collapse around the metals preventing migration or access to them by reactant molecules. Figure 2 shows the surface area versus pore size of MILLENNIUM at a variety of metals loadings on both Mitchell method impregnated samples and Ecat. As can be seen as vanadium loading increases, the pore area rapidly decreases after deactivation. It is likely that the pore collapses around both nickel and vanadium, thereby both immobilizing and passivating the metals.
Unlike MILLENNIUM, ULTRIUM passivates nickel both by agglomeration on the surface and formation of nickel aluminate. It appears from the extreme surface concentration of nickel on ULTRIUM that the former mechanism is prevalent. The inhibiting effect of nickel agglomeration on metals activity has been shown in many studies(Ref. 7) and can be quite dramatic in its impact. Vanadium, on the other hand, is clearly associated with magnesium in the vanadium trap portion of the catalyst. Because vanadium is involved in interparticle transfer, the trapping agent can either be in the same or separate particles from the active component. Magnesium is known to form vanadate and unlike the other alkaline earths, reductively eliminates sulfur. Hence, formation of magnesium sulfate does not permanently deactivate the magnesium with respect to vanadate formation (see below).
4. Catalytic Data
Both MILLENNIUM and ULTRIUM were tested under a variety of conditions versus state of the art competitive catalysts. Figure 3 compares MAT coke and hydrogen yields versus activity trends for a data composite of competitive catalyst to ULTRIUM and MILLENNIUM. Activity is defined as Conversion/(100-Conversion). Metals were cracked on using 20 cycles and the metallated catalysts post steamed at 1450 F for four hours in 90% steam-10% air. Both BASF catalysts show reduced coke and hydrogen yield implying superior metals tolerance compared to the competitors with ULTRIUM showing the best selectivity. Similar results occur when the catalysts are tested using Mitchell method impregnation in the resid feed MAT as shown in figure 4. Figure 4 also shows MILLENNIUM to have a small advantage in bottoms cracking compared to ULTRIUM. In figure 5, we see the superior metals tolerance of MILLENNIUM compared to competitive resid catalysts in the cracking of resid feed in an fixed fluid bed (FFB) unit. All three test methods show superior metals tolerance for the BASF resid catalysts. Figure 6, meanwhile, demonstrates the superior activity maintenance of BASF resid catalysts after Mitchell method impregnation with 5000 ppm vanadium.
5. Sulfur Tolerance
One of the problems with some commercial vanadium traps is the propensity to deactivate in the presence of sulfur. Hence, these materials look excellent in lab tests that ignore sulfur effects but fail commercially because most resid feeds contain an appreciable amount of sulfur. By way of example, figure 7a shows MAT activity versus cat/oil for a control catalyst compared to the same catalyst containing 5% SrTiO3. Strontium readily forms a stable vanadate but just as readily forms an even more stable sulfate. When the SrTiO3 containing catalyst is metallated in the absence of sulfur, the improvement in activity is excellent. However, when the feed used for metallation is spiked with thiophenic sulfur, the performance reduces to base level. Clearly the SrTiO3 is totally deactivated in the presence of sulfur. Figure 7b shows that the traps in MILLENNIUM and ULTRIUM are not affected by sulfur. They maintain their activity advantage whether or not sulfur is present.
6. Commercial Performance
MILLENNIUM has been tested commercially in several units. The most illustrative data has been gained from refinery A, which feeds a high nickel, moderate vanadium containing resid blend to the cat cracker. Figure 8 shows MAT coke yields of equilibrium catalyst from that unit. The nickel level increased dramatically with increasing MILLENNIUM in the unit inventory. Nevertheless, MAT coke yeild decreased over this same period, demonstrating the nickel tolerance of MILLENNIUM. Over the same period of time, the vanadium level on catalyst remained almost constant, but zeolite surface area retention increased significantly. Overall, the refinery was able to significantly increase feed throughput, and lower fresh catalyst addition rates while reducing the amount of bottoms. Both refinery and Ecat data attest to the improved metals tolerance and bottoms cracking ability of MILLENNIUM compared to the competitive resid catalyst used previously.
ULTRIUM also is in several units and data illustrative of its performance is shown in figures 9-11. In this case ULTRIUM replaced REDUXION and showed lower coke and hydrogen yields in the presence of 5000 ppm metals. No loss in bottoms cracking activity was observed despite replacing a good matrix activity catalyst.
Spectroscopic and microscopy data indicate MILLENNIUM traps nickel and vanadium on its chemically treated alumina rich matrix component. The pores collapse around the trapped metals and prevent them from catalyzing dehydrogenation reactions or migrating to the zeolite. ULTRIUM mainly passivates nickel by agglomeration on the catalyst surface, while trapping vanadium by forming a magnesium vanadate. Both catalysts show superior metals tolerance in a variety of laboratory testing procedures with ULTRIUM showing clearly best in class performance with respect to nickel passivation. Commercial trials of both MILLENNIUM and ULTRIUM have born out the positive catalytic traits shown in the laboratory tests.
1. Nielsen, R.H. et. al., "Metals Passivation" in Fluid Cracking Science and Technology, p339, Elsevier (1993)
2. Cadet V. et. al.,Appl. Catal., 68, 263 (1991)
3. Occeli, M. "Laser Raman and X-Ray Photoelectron Characterization of Vanadium Contaminated Components of Fluid Cracking Catalysts" in Fluid Catalytic Cracking II: Concepts in Catalyst Design, p 252, ACS (1991)
4. Wormsbecher, R.F. et. al., Symposium on Deactivation and Testing of Hydrocarbon Conversion Catalysts Preprint, Div. Petr.Chem., ACS Meeting Chicago Ill. 1995
5. Lerner, Bruce and Deeba Michelle, Symposium on Deactivation and Testing of Hydrocarbon Conversion Catalysts Preprints Div. Petr. Chem., ACS Meeting Chicago Ill. 1995
6. Boock, L.T. et. al., Symposium on Deactivation and Testing of Hydrocarbon Conversion Catalysts Preprints, Div. Petr. Chem., ACS Meeting Chicago Ill. 1995
7. Feely, J.S. "Genesis and Catalysis of PdNix Clusters in Y Zeolites" and references therein, PhD Thesis, Northwestern University, 1991
8. Mitchell, B.R., Ind. Eng..Chem. Prod. Res. Dev., 19, 209(1980)
9. Young, G.W., "Realistic Assessment of FCC Catalyst Performance in the Laboratory" in Fluid Cracking Science and Technology, p. 257, Elsevier (1993)
10. Bolt P.H. et.al., J. Catal., 151, 300(1995)