I n d . Eng. C h e m . Res. t987,26, 882-886
882
Catalytic Cracking of Alkanes on Large Pore, High Si02/A1203Zeolites in the Presence of Basic Nitrogen Compounds. Influence of Catalyst Structure and Composition in the Activity and Selectivity Avelino Corma* and Vicente Forn6s Instituto de Catdlisis y Petroleoqulmica, C.S.I.C., Madrid 28006, Spain
Juan B. M o n t h and Antonio V. Orchilli% Departamento Zngenierla Quimica, Universitat de Valdncia, 46100 Burjassot, Valencia, Spain
The influence of the nature and amount of basic, nitrogen-containing organic molecules (pyridine, 2,6-dimethylpyridine, and quinoline) in the feed on the activity and selectivity of a series of ultrastable Y zeolites during n-heptane cracking has been studied. The effect of the framework SiOz/Al2O3 ratio and procedure of dealumination on the resistance of the Y zeolite to deactivation are described and the results compared with those obtained with other zeolites such as Beta and H-ZSM-5. The observations have been extrapolated to a gas oil cracking, and it is predicted that the selectivity to gasoline and the RON decreases when increasing the nitrogen content in feed and the proton affinity of the poisons. Several authors have reported a severe reduction in the cracking activity of amorphous silica-alumina catalysts when basic molecules are present in the feed (Mills et al., 1950; Voge et al., 1951; Viland, 1957). This problem was partially solved with the advent of zeolite-containing cracking catalysts, due to their higher number of active acid sites. The problem has recently reappeared, due to the fact that FCC's catalytic crackers are being used for processing residues, containing high amounts of metals and sulfur and of basic nitrogen-containing molecules (Magee et al., 1979). Fu and Shaffer (1985) have published an extensive work on the influence of different nitrogencontaining basic organic molecules and on the performance of various commercial cracking catalysts. Besides the new generation of cracking catalysts for RON-enhanced (RON = research octane number) gasolines based on ultrastable Y zeolites is very sensitive to the presence in the feed of poisons for acid sites (Na+and basic molecules). Ultrastable zeolites are highly dealuminated Y zeolites which have low activity for catalyzing hydrogen-transfer reactions and produce, therefore, gasoline with high olefins content, i.e., with high RON (Pine et al., 1984; Rajagopalan and Peters, 1985; Andreasson and Upson, 1985). However, highly dealuminated Y zeolites have a small number of active acid sites, and their activity is strongly affected by the presence of poison in the feed. In the present work, the influence of the nature and amount of basic, nitrogen-containing molecules in the feed, on the activity and selectivity of a series of ultrastable Y zeolites obtained by different procedures and of other zeolites with high Si02/A1203framework ratios, has been studied using the cracking of n-heptane as test reaction.
Experimental Section Materials. Samples of LaHY zeolites were prepared by a conventional exchange procedure. A NaY zeolite (SK-40, SiO2/Al20, molecular ratio = 4.8) was exchanged with La3+,followed by deep bed calcination, until the level of exchange was 0%, 15%, 40%, 68%, 82%, and 99% (samples designated HY-0, HY-1, HY-2, HY-3, HY-4, and
* To whom correspondence
should be addressed.
0888-5885f 87f 2626-0882$01.50f 0
Table I. Characteristics of the Zeolites Used in This Studs zeolite Lap03,wt % uc, 8, Al/uca Si/Alb HY-0 0 24.40 18.79 9.2 3.28 24.46 25.53 6.5 HY-1 HY-2 24.59 40.16 3.7 8.29 12.83 24.68 50.28 2.8 HY-3 17.29 24.70 52.53 2.6 HY-4 LaY 20.57 24.76 59.28 2.5 HYD-1 24.30 7.68 24.0 24.43 22.33 7.6 HYD-2 24.54 34.29 4.6 HYD-3 80.0 HZSM-5 BETA 10.0 a Calculated from the Fitchner equation (Fitchner-Schmittler et al., 1984). uc = unit cell size. *Atomic ratio.
Lay, respectively). Then, the remaining Na+ of these samples was exchanged by NH4+until its content was less than 1% of the original one. A series of HY dealuminated zeolites (HYD) were prepared by SiC14treatment at temperatures in the range from 300 to 500 "C (Beyer et al., 1985), followed by repeated NH4+exchange and calcination. A H-ZSM-5 and a H-Beta zeolite were also used in this study for comparative purposes. The characteristics of the samples are given in Table I. Procedure. The catalytic experiments were performed at atmospheric pressure, 430 "C, and 150 s on-stream in a continuous flow glass tubular reactor with an internal diameter of 3.6 cm and 80 cm long. The catalytic bed (2 cm in length) consisted of the zeolite catalyst (0.5-0.75 mm) diluted with glass chips of the same size. The catalyst-to-oil ratios (gg-') used were 0.116 for HY-0, HY-1, HY-2, HY-3, HYD-1, and HYD-2; 0.174 for HY-4; 0.464 for Lay; 0.100 for HYD-3; 0.070 for HZSM-5; and 0.017 for Beta-10. The experimental technique for cracking and analysis of the reaction products has been described in detail elsewhere (Corma et al., 1984). Pyridine, quinoline, and 2,6-dimethylpyridine (GC purity) were added to nheptane (purity > 99.9) and fed into the reactor. The IR spectra of the poisoned catalysts were obtained with a Perkin-Elmer 580B spectrophotometer equipped with data station, at 150 "C and lo-* torr. The spectra in the 1400-1600-~m-~ range were recorded at room temperature. 0 1987 American Chemical Society
Ind. Eng. Chem. Res., Vol. 26, No. 5, 1987 883 Pyridine
Puinolinr
pKb. 8.8
pKb.S.1
P . A r 121.8
P.A=219.1
2,6 D.M.Pyridine
pKbd.3 ' P. A,. 231 .O Kcal mol-'
Figure 1. Basicity parameters of the nitrogen-containing organic molecules studied.
u
Y
z 4
m LT 0
m yl 4
I J
1700
1700
1600 V.
1500
1400
CM-'
IbOO
v,
1500
9400
em.'
Figure 3. IR spectra of the HY-0 zeolite after reaction with nheptane in the presence of pyridine (left) and 2,6-dimethylpyridine (right).
01 0.2 Nitrogen in feed ( w t % )
Figure 2. Activity measured by n-heptane conversion of the HY-0 zeolite vs. the nitrogen content in the feed: ( 0 )pyridine, (+) quinoline, (A)2,6-dimethylpyridine.
+@+ fi Figure 4. Scheme of the poisoning through the inductive effect.
Nitrogen on catalyst was analyzed by means of a C, N, H, analyzer equipped with a CAHN RG electrobalance.
Results and Discussion Influence of the Nature of the Base. Intuitively, one would expect that the higher the basicity of a given molecule, the higher should be ita poisoning effect for solid acid catalysts. However, as it has been pointed out (Fu and Schaffer, 1985), there are different ways to estimate basicity, the most common ones being the basicity constant (Kb)and the proton affinity (P.A.). The Kb is a measure of the basicity of a given molecule in aqueous solution, while the P.A. is determined by measuring the gas-phase proton-transfer reaction between a base and the NH4+ion. The values of K
0.05
01
0.3
05
Nitrogen on catalyst (wto/i.)
Figure 5. Activity measured by n-heptane conversion for zeolites for different amounts of nitroHY-0 (A),HY-1 (e),and LaY (0) gen-on-zeolite deposited during pyridine poisoning. Table 111. Number of Active Acid Sites Poisoned by a Molecule of Pyridine, on Y Zeolites with Different Framework Si02/A120sRatio molecules of ONNNA13+ ON"/ x (g of molecules Py x W0 SiO,/ catalvst ( e of cat.)-' A1,0," cat.)" of Pv 2.9 14 3.5 1.2 HY-0 3.2 1.2 HY-1 2.7 13 1.2 0.9 LaY 1.3 5 2.6 1.1 HYD-1 2.4 48 HYD-2 2.6 15 3.7 1.4 2.5 9 3.0 1.2 HYD-3 'Calculated from the unit cell size and Fitchner-Schmittler's et al. equation (1984).
by an inductive effect of the type presented in Figure 4, where the positive charge of the proton anchoring the pyridine strongly decreases, but part of the electronic density of the basic molecule is transmitted to the rest of the framework, affecting the density of the rest of the surface protons. By such a mechanism, it would be possible to change active sites (6+), in sites not acid enough (S'+) to catalyze the cracking reaction studied. Influence of the Amount of Poison. The results from Figure 2 show a strong decrease in the activity, measured by n-heptane conversion, with the first small amount of pyridine added, while the decrease is smaller for higher amounts of pyridine. A behavior of this type has frequently been used as a proof of the heterogeneity of solid acid surfaces. However, this conclusion implies that all the base fed has been adsorbed on the zeolite, something which does not necessarily occur. On the other hand, when the activity is related with the amount of base adsorbed on the zeolite, it can be seen (Figure 5 ) that the activity decreases linearly with the amount of pyridine on the zeolite catalysts. The same behavior is observed with the other two bases. The observed linearity indicates that the poisoning of the zeolite is not selective or, what is equivalent, that the surface of the zeolite is quite homogeneous for cracking n-heptane. This, in turn agrees with the finding (Corma et al., 1986a) that despite the fact that there is an acid strength distribution of acid sites, only a small fraction of all acid sites (the strong acid ONNN sites) are active for n-heptane cracking. Effect of the Framework Si02/A1203Ratio on Poison Susceptibility. In Table I11 the molecules of pyridine necessary to reduce the n-heptane conversion to zero, obtained by extrapolation of the straight lines in Figure 5 , are compared with the theoretical number of framework ONNN sites, calculated considering a random
20
30
LO
Al/u.c
Figure 6. n-Heptane cracking catalyst decay by 300 ppm nitrogen (pyridine) in the feed vs. number of framework aluminums per unit cell.
distribution of aluminums in the framework (Pine et al., 1984). The ratio between the two values shows that each molecule of pyridine poisons, in some cases, slightly more than one active site. It seems that the maximum number of active sites poisoned by one molecule of pyridine is achieved for a Si02/A1203 14 for which the density of ONNN sites is maximum (Pine et al., 1984). If our above assumption on the induced deactivation is correct, one should expect that the number of active sites poisoned by one molecule of pyridine would tend to one when increasing the framework Si02/A1203ratio, when the effect of the proton affinity on the decay would disappear. This could be observed in the series of zeolites dealuminated with SiC14, in which it is possible to see (Table 111) that the number of ONNN sites poisoned by one molecule of pyridine increases first up to Si02/A1203 14 and then decreases with further increase of the framework S O 2 / A1203 ratio. In Figure 6 we show the dependence the n-heptane cracking catalyst decay caused by 300 ppm basic nitrogen by pyridine addition to the feed on a series of HLaY and HYD-SiC14 dealuminated zeolites on the number of framework aluminums per unit cell. It can be seen that the decay of catalysts due to poisoning by a basic molecule decreases first with dealumination, reaches a minimum for e 2 4 A13+ions per unit cell ( N 14, framework Si02/A1203), and then increases sharply. These results also indicate that there must be an increase in the amount of acid sites active for cracking with increasing dealumination, up to 20-30 A13+ per unit cell. After this point, further dealumination provokes a decrease in the number of active sites. Two conclusions can be extracted from this graph. The first conclusion is that, even considering the poisoning by induction effect, the Y zeolite most resistant to poisoning with organic bases is that with the maximum density of ONNN sites. This implies that, as it is indeed observed in Figure 6, the activity of ultrastable Y zeolites will be strongly affected by the poisoning with the nitrogen-containing basic molecules present in the feed. The second conclusion is that, as far as n-heptane cracking is concerned, the behavior of the poisoning with respect to the Si02/A1203ratio is not strongly dependent of the dealumination procedure used here. Influence of the Structure of the Zeolite. It has been claimed that by adding small amounts of H-ZSM-5 zeolite in the cracking catalyst, it is possible to increase the motor octane number of the gasoline produced (Yanik et al., 1985). This can be an attractive solution for refineries having alkylation plants, which can handle the larger amount of C3 and C4gases produced by incorporation of
Ind. Eng. Chem. Res., Vol. 26, No. 5, 1987 885 HY-o CAT (24.39 AI
H -BETA (Si/AI=101
~
Q
I
o Pyridine ET 2,6DM Pyridine Quinoline o 2.4.6 TM Pyridine
/// 0 HZSMS ( S i / A l = 8 0 1 I
200
1
I
I
1
LOO 600 800 1000 Nitrogen i n feed ( p p m l
I
Figure 8. Effect of pyridine, quinoline, and 2,6-dimethylpyridine on (Cg C,)/(C, + C,) ratio observed during n-heptane cracking.
+
1200
Figure 7. n-Heptane cracking catalyst decay vs. nitrogen in the feed for different zeolites and poisons.
the HZSM-5 in the cracking catalyst. However, this zeolite can be quite sensitive to basic poisons due to its relatively high framework SiOz/A1,03 ratio. Nevertheless and due to the size of the pore, ZSM-5 zeolites can exclude the most voluminous basic products, and this may decrease their susceptibility to poisoning. The results from Figure 7 show that the HZSM-5 decays faster than the aluminated Y zeolites, for pyridine, quinoline, and 2,6-dimethylpyridine, which can penetrate inside the pores, due probably to the higher SiOz/A1,03 ratio of HZSM-5. However, if trialkylated pyridine, such as 2,4,6-trimethylpyridine which cannot penetrate in the HZSM-5 channels (Namba et al., 1984),is used as a poison, then it is observed that the HZSM-5 zeolite is practically not poisoned. In the case of commercial cracking catalysts, it remains the option to use big-pore, high SiOZ/AlzO3zeolites obtained by synthesis, one of which is the Beta zeolite (Wadlinger et al., 1975). This zeolite has a void structure similar to zeolite L, consisting of pores with 12-membered rings and lobes (Martens et al., 1985). The results presented in Figure 7 show that from the point of view of poisoning by basic nitrogen-containingorganic compounds, the H-Beta decays faster than a HYD zeolite with the same framework SiOz/A1,03 ratio, indicating that the former probably has, in the framework, a lower density of isolated aluminum atoms. Extrapolation of n -Heptane Poisoning Results to Gas Oil Cracking. We have reported (Corma et al., 1986b) that for a series of Y zeolite catalysts, there is a good correlation between the (C3 C4)/(Cz + C,) selectivity ratio observed during n-heptane cracking and the selectivity to gasoline during gas oil cracking. A good correlation was also observed between the paraffin-to-olefin ratio in the products of n-heptane cracking and the RON of the gasoline produced during gas oil cracking. Following this idea, the (C3 + C4)/(C2 + C,) ratio obtained with a HY ultrastable (HY-0) zeolite when poisoned with different amounts of pyridine, quinoline, and 2,6-dimethylpyridine in the feed is given in Figure 8. It can be seen that the (C, + C,)/(C, + C,) ratio decreases with increasing amounts of base in the feed, the decrease increasing with increasing proton affinity of the poisoning base. Moreover, in Figure 9 it can be seen that the paraffinteolefii ratio in the n-heptane cracking products increases with nitrogen in the feed and apparently also with increasing proton affinity of the base. From these results and if extrapolation to the above-described gas oil cracking is correct, we can conclude that an increase in the base content in the feed produces a decrease in the gasoline
+
0.06 0.12 0.18 0.24 Nitrogen in feed lwt%)
HY-0 CAT (24.39
: 3
A)
c)Quin.
2
A I
I
I
v
2.6DMPy I
0.18 0.24 Nitrogen in feed (wt%l 0.06
0.12
Figure 9. Effect of pyridine, quinoline, and 2,6-dimethylpyridine on paraffin/olefin ratio in the n-heptane cracking products.
selectivity and in its RON, the decrease increasing with increasing proton affinity of the basic poisoning product. In conclusion and taking n-heptane cracking as a reference, the following can be said: The deactivating power of basic molecules for catalytic cracking is directly related with their proton affinity. A basic molecule can deactivate more than one active site; the higher the proton affinity, the higher will be the intrinsic deactivating power. The surface of the zeolite is quite homogeneous from the point of view of alkane cracking, and only framework aluminums with zero aluminums in the next nearest neighbors can generate active sites for n-heptane cracking. As a consequence of the above point, the decrease in the activity due to poisoning is minimum for Y zeolites with N 24A1 per unit cell. Highly dealuminated ultrastable Y zeolites are very sensitive to poisoning by basic nitrogen-containing compounds. Therefore, if ultrastable zeolites have to be used in commercial cracking catalyst with feeds containing important amounts of basis products, additives have to be included in the matrix to preserve the zeolite activity. Even if the framework SiOz/A1203ratio is an important factor in the poisoning by basic products, the structure of the zeolite can be critical, especially for bulky poisoning molecules. If the results from n-heptane can be extrapolated to gas oil, it can be said that the selectivity to gasoline and the RON decreases with increasing nitrogen content in the feed and with the increasing proton affinity of the poisons. Registry No. Pyridine, 110-86-1; 2,6-dimethylpyridine, 10848-5; quinoline, 91-22-5; n-heptane, 142-82-5.
Literature Cited Andreasson, H. U.; Upson, L. L. Oil Gas J . 1985,83(31), 91. Beyer, H. K.; Belenkaya, Y. M.; Hauge, F.; Tielen, M.; G r o b e t , P. J. Jacobs, P. A. J. Chem. SOC.,Faraday Trans. 1 1985,81, 2889. Corma, A.; Montijn, J. B.; Orchill&, A. V. Ind. Eng. Chem. Prod. Res. Deu. 1984,23(3), 404. Corma, A.; FornBs, V.; Mont6n, J. B.; Orchill&, A. V. Znd. Eng. Chem. Prod. Res. Deu. 1986a,25(2), 231. Corma, A.; FornBs, V.; Melo, F. V. Actas X Simp. Zberoam. Catal. 1986b,2,647.
I n d . Eng. Chem. Res. 1987,26,886-894
886
Fichtner-Schmittler, H.; Lohse, U.; Engelhardt, G.; Patzelova, V. Cryst. Res. Technol. 1984, 19(1), K1-K3. Fu, Ch.; Schaffer, A. M. Id. Eng. Chem. Prod. Res. Deu. 1985,24(1), 68. Furimsky, E. Erdol Kohle 1982, 35, 455. Magee, J. S.; Ritter, R. E.; Rheaume, L. Hydrocarbon Process. 1979, 58(9), 123. Martens, J. A,; PBrez-Pariente, J.; Jacobs, P. A. Acta Phys. Chem. 1985, 31(1-2), 487. Mills, G. A.; Boedeker, E. R.; Oblad, A. G. J. Am. Chem. SOC.1950, 72, 1554. Namba, S.; Nakanishi, S.; Yashima, T. J . Catal. 1984,88, 505. Perrin, P. D. Dissociation of Organic Bases in Aqueous Solutions; Butterworths: London, 1965.
Pine, L. A.; Maher, P. J.; Wachter, W. A. J. Catal. 1984, 85, 466. Rajagopalan, K.; Peters, A. W. Prepr.-Am. Chem. Soc., Diu. Pet. Chem. 1985, 30(3), 538. Viland, C. K. Prepr.-Am. Chem. Soc., Diu. Pet. Chem. 1957,2(4), A-41. Voge, H. H.; Good, G. M.; Greensfelder, B. S. Proc. World Pet. Congr., 3rd 1951, 4, 124. Wadlinger, R. L.; Kerr, G. T.; Rosinski, E. J. US Patent 3308069, 1975. Yanik, S. J.; Demmel, E. J.; Humpries, A. P.; Campagna, R. J. Oil Gas J. 1985, May 13, 108.
Received for review June 6, 1986 Accepted November 14, 1986
Formation of Deposits from Thin Films of Mineral Oil Base Stocks on Cast Iron Spyros I. Tseregounis,* James A. Spearot, and Deborah J. Kitet Fuels and Lubricants Department, General Motors Research Laboratories, Warren, Michigan 48090-9055
A microoxidation test was used to investigate deposit formation from thin films of a mineral oil base stock on cast iron surfaces. The amount of deposits, the induction time for their appearance, and the rate of their formation were measured. The results show that the induction time increases linearly with the thickness of the film and decreases exponentially with temperature. The rate of deposit formation increases exponentially with temperature. The effects of oxygen diffusion through the thin film and oil evaporation are discussed. Comparison and correlation of results obtained from seven mineral lubricant base stocks indicated that the induction time and the amount of deposits depend on the molecular weight of the oil and its carbon-type distribution. Antioxidant additives also affect the formation of deposits. A zinc dialkyldithiophosphate additive increases the induction time and decreases the rate of deposit formation, while an ashless additive (2,6-di-tert-butylphenol) delays slightly the appearance of deposits. The oxidation and combustion of hydrocarbons (either fuels or lubricants) are responsible for the formation of solid, insoluble products which stick on the metal surfaces of internal combustion engines. Accumulation of these products (commonlycalled deposits) can cause engine wear and can have adverse effects on engine efficiency, performance, and durability (Stewart and Stuart, 1963; Anderson, 1968; Ebert, 1985). Autoxidation of hydrocarbons has long been considered an important factor in the production of sludge and deposits. For relatively high partial pressure of oxygen (>lo0 torr), hydrocarbon oxidation proceeds via the free-radical mechanism (Sheldon and Kochi, 1981): initiation
+ + RH
propagation
R' ROO' termination 2R00'
-
initiator
O2
RH
R'
ROO'
ROOH
+ R'
neutral nonradical products
where RH = hydrocarbon, R' = hydrocarbon radical, ROOH = hydroperoxide, and ROO' = peroxy radical. While liquid-phase oxidation species and their condensation-polymerization products are eminent deposit precursors, the exact nature of such deposita and the factors Presently a t t h e Department of Chemical Engineering, University of Massachusetts, Amherst, MA 01003. 0888-5885/87/2626-0886$01.50/0
which affect their formation are relatively unknown. Studies of deposit-forming tendencies of jet fuels by a number of investigators (Taylor 1969, 1974; Taylor and Wallace, 1968; Taylor and Frankenfeld 1978; Mayo et al., 1975; Lauer et al., 1984) showed that temperature; fuel composition and chemistry; oxidizing atmosphere; sulfur, nitrogen, and oxygen compounds; pressure; and unsaturated or aromatic hydrocarbons are important parameters which control the appearance and rate of formation of deposits. Oxidation studies of mineral oils and ester-type lubricants (Clark et al., 1984; Cho and Klaus, 1983; Lahijani et al., 1980; Lockwood and Klaus, 1980, 1981) revealed that the formation of sludge and deposit precursors follows the condensation-polymerization of the nonvolatile oil oxidation products. The formation of sludge depends on the volatility of the lubricant and its oxidation products, the presence of additives, natural impurities (sulfur or nitrogen compounds), metal dissolution, and the type of the metal surface involved. According to Naidu et al. (1984), the formation of deposits can be simply described by the following first-order sequence: evaporation
evaporation
t
t
lubricant
-+
primary oxidation products
-+
high molecular weight products
-+
deposits
where the polymerization of the primary oxidation products may proceed 10-100 times faster than their production. Recent work by Nepogod'ev et al. (1985) shows that the polycondensation reactions of the hydroxy acids formed 0 1987 American Chemical Society