I
C. G. MYERS, W. H. LANG, and P. B. WElSZ The Socony Mobil Research Department, Paulsboro Laboratory, Paulsboro, N. J.
Aging of Platinum Reforming Catalysts The reactions causing aging can be reversed-restoring high catalyst activity
P u m w M REFORMING CATALYSTS lose activity (age) during naphtha processing. Some of this aging is eliminated by pretreating to remove sulfur, nitrogen, and oxygen compounds from the naphtha charge. Even with pretreated naphtha, however, aging still occurs, and it is very severe in the low pressure operations which generally give the highest gasoline yields. This aging is presumably associated with the hydrocarbon naphtha components themselves. Therefore, aging caused by prototype hydrocarbons of distinctive structures was measured along with their chemical reactions in the reformer. This report discusses theoretical and practical aspects of the results. In- the work establishing the hydrocarbon types most responsible for aging, a platinum on low area silica-alumina catalyst was used. This had the advantage that the results were free from aging associated with chlorine loss, since neither the activity nor the acidity of silica-alumina based reforming catalysts is derived from combined chlorine. The results, however, suggest a mechanism by which hydrocarbons cause aging that seems quite independent of the detailed structure of the acid centers of the catalyst; the generalized mechanism is probably applicable to all platinum reforming catalysts. I n addition, the practical consequences of the work have been shown to apply to platinum-alumina catalysts. These catalysts all have acidic sites and hydrogenation-dehydrogenation sites (7, 6, 77); the acidic sites are probably rate controlling (5, 6, 70) a t the severities used-98 or higher leaded octane number. The acid sites isomerize and/ or crack olefin intermediates made a t the platinum dehydrogenation sites, the products being hydrogenated to saturates or dehydrogenated to aromatics a t the platinum sites (6, 77). I n addition, the platinum sites dehydrogenate 6-carbon-ring alicyclics directly to aromatics (7, 3, 6, 7 7 ) .
was General Chemical reagent grade, 1methylnaphthalene was Eastman practical grade, n-dodecane was from HumphreyWilkinson, and the petroleum naphtha was of straight run Mid-Continent origin. All other hydrocarbons were Phillips pure grade (994-010). Naphthalene, toluene, and n-pentane were used without special purification. The naphtha was pretreated with hydroen over cobalt molybdate on alumina ?American Cyanamid HDS-2 catalyst) to remove nitrosen and sulfur compounds. Cumene was percolated through silica gel and alumina. All other hydrocarbons were percolated through silica gel only. Purities are given in Table 11. Hydrogen was passed through Deoxo cylinders (palladium catalyst) to convert traces of oxygen to water and then dried over silica gel. The properties of the three reforming catalysts used are listed in Table I. Catalyst No. 2 is experimental, the other two are samples of commercial catalysts. Procedure. As catalysts age during naphtha reforming under fixed conditions, product octane number declines. Standard conditions were 250 p.s.i.g., 300" F. at reactor inlet, 10 moles of hydrogen per mole of naphtha, 2 volumes of naphtha per hour per volume of catalyst, and catalyst No. 1. Under these conditions, high octane number product arises primarily from dehydrogenation of 6-carbon-ring alicyclics to aromatics and from isomerization of 5-carbon-ring alicyclics to 6-carbonring structures followed by dehydrogenation to aromatics. Lesser contributions come from isomerization of n-paraffins or from their dehydrocyclization to aromatics. The contribution of any individual hydrocarbon to catalyst aging was evaluated by comparing product octane number from naphtha reforming under the standard conditions before and after a period of operation with the single hydrocarbon replacing the naphtha-other conditions unchanged. The changes in naphtha reforming activity effected by these individual hydrocarbons were compared to the change found in a like interval of continued naphtha processing. Product analysis depended on mass spectrometry, Podbielniak distillation, and infrared analysis. Octane numbers were obtained by the CFR F-1 Research Method using 3 ml. of tetraethyllead. In the direct study of aging rate as a function of hydrogen-naphtha ratio, catalyst activity was measured as in commercial operation-temperature required to
Table I.
Catalyst No.
Experimental
Materials. Cyclopentane was Phillips technical grade (95 %), toluene was Baker Analyzed (99 +yo), naphthalene
+
a
produce 10-pound Reid vapor pressure gasoline at a fixed octane level. A commercial platinum on alumina catalyst (catalyst No. 3) and an experimental platinum on silica-alumina catalyst (catalyst No. 2 ) were included in the study to test for generality of the results. All processing was bench scale (75 ml. of catalyst), single pass, and without hydrogen recycle. Catalyst Aging by Cyclopentane
When reformed a t 10 to 1 hydrogenhydrocarbon ratio, cyclopentane-but not n-pentane-aged the catalyst even while producing mainly n-pentane (Table 11, Figure 1). However, the cyclopentane also produced aromaticsparticularly naphthalene and other polynuclear aromatics-not produced by the n-pentane. Thus, cyclopentane produces a damaging intermediate not readily formed from n-pentane. Cyclopentadiene might well be this intermediate. I t is more readily formed-by dehydrogenation a t the platinum sitesfrom cyclopentane than n-pentane, and it polymerizes readily to polycyclics which could degrade and dehydrogenate to the observed naphthalene and other polynuclear aromatic products. Furthermore, the concept of cyclopentadiene and its polymers as intermediates leading to catalyst poisoning agrees with the finding of Shuikin and others (8) that cyclopentadiene, 1ethylcyclopentene-1 , and alkylcyclopentanes all poison the dehydrogenation centers of platinum, 5%, on carbon. Finally, the suggestion of cyclopentadiene as a n actual intermediate from cyclopentane is supported by the observation of Weisz (9) that 18 mole yo of methylcyclopentadiene was produced from methylcyclopentane over platinum on silica gel a t atmospheric pressure with a hydrogen-hydrocarbon ratio of 4. Cyclopentadiene is especially attractive as a conceptual intermediate because of its ready polymerization to polycyclics which could degrade to aro-
Two Catalysts W e r e Commercial, One Experimental Support Area, Type" sq. meters/g.
1
SA
2
S
3
Ab
100 640 410
Pt Concn., Wt. % 0.55
Minor Constituent Concn., Typea wt. yo A 13
0.35 0.58
A c1
S = silica gel; SA=silica-alumina; A = alumina; C1 = chlorine.
VOL. 53, NO. 4
0.13 0.67
Mostly eta alumina.
0
APRIL 1961
299
matics containing more than two condensed rings. This would explain the finding of phenanthrene, fluorene, pyrene, chrysene, perylene, and coronene as trace products of naphtha reforming. Moreover, dicyclopentadiene, in particular, might adsorb directly on the platinum sites by formation of complexes analogous to dicyclopentadienyl ruthenium (ruthenocene), a “sandwich” compound of a platinum series metal that has actually been prepared (72). This would represent a detailed picture of initial adsorption of a specific diolefin catalyst poison a t the platinum site.
Reversibility of the Poisoning Reactions
The preceding sketch indicating a path from cyclopentane to tricyclopentadiene shows each reaction step to be reversible for the following reasons. First, the conventional reforming mechanism implies that the platinum catalyzed dehydrogenation reactions are reversible. Secondly, the thermal polymerization of cyclopentadiene, at least, is reversible. The following evidence suggests that, in general, the main steps involved in catalyst aging by hydrocarbons are reversible.
Catalyst Reactivation b y Desorption of Poisons. Treating aged catalyst with
75 70
0
I
I
IO
20
,
,
30 40 50 DAYS ON ST9EAM
:I/,
60
Figure 1. Typical experiments in the study of catalyst aging by specific hydrocarbons Catalyst No. 1 ; standard conditions
Table II.
flowing nitrogen gas restored a large part of catalyst activity lost during a preceding period of naphtha reforming, the rate of the reactivation being relatively low (Table 11, Figure 1). During the reactivation, condit ons were the same as in the preceding reforming except that the naphtha pump was shut off and nitrogen replaced hydrogen at the same volumetric flow rate. Thus, a quite significant part of the catalyst poisons associated with aging is reversibly adsorbed on rate controlling sites (as to catalyst activity). These are
Different Hydrocarbons Aged the Catalysts Differently
Catalyst No. 1 (Table I); standard conditions (see Procedure) Initial Octane No., Change in F-1 Time on Octane No. 3 ml. Identified Reaction Run per Day Stream, Hydrocarbon TEL (No./Day)“ Products Days No. 90.0 7 -2.8 1 Tolueneb Toluene 6 2 89.2 +0.3 Tolueneb Toluene 0.0 6 93.5 Propane 3 Cumeneb -4.8 4 91.8 3 Naphthalenec 94.2 1.3 l-Methylnaphthalened - 15 5 92.1 6.3 fO.2 C1-Ca paraffins 6 n-Pentanee 2.8 -1.0 n-Heptanef 93.6 TolueneQ 7 4 . 0 92.0 2 Aromaticsg n-Dodecane 8 +o. 1 95.2 3.3 Isobutane Iso-octane 9 2 -4.4 97.8 ~ - C ~ Hnaphthalene, I~, 10 C yclopentaneh aromatics 1 -0.4 Methylcyclopentane 96.5 11 93.6 7 Toluene i-0.3 M ethylcyclohexanej 12 99.9 7 -1.0 13 Perc naphthai - 1.1 99.0 6.5 14 Naphtha 95.6 2.1 -3.6 15 Naphtha 7 94.6 -0.6 16 Naphtha -0.5 6.7 91.4 17 Naphtha 3.8 90.7 -1.6 18 Naphtha -0.4 9 84.0 19 Naphtha 0.0 4.2 81.0 20 Naphthak +0.3 3.2 77.3 21 NaphthaL 4-1.8 4 73.7 22 Naphtham +1.0 3.5 70.2 Naphthan 23 7 +0.4 24 91.6 None 90.1 7 +0.9 25 None 2.8 +5.5 79.7 26 None As pure as NBS standard hydrocarbons. 10 wt. % naphthalene in blida 2~0.1 no./day. Continent naphtha. Eastman material percolated through silica gel-still containing 4-5 mole % S compounds. e Contains 0.5 mole ’% cyclopentane. f Contains 0.2 mole 7% methyl’ dimethylbutanes. Contains 0.3 cyclohexane. Plus paraffins. Contains 2.4 mole % mole % cyclohexane trace of benzene: exp. was a t 500 p.s.i.g. Mid-Continent naphtha percolated through silica gel-unpercolated naphtha used elsewhere. k Naphtha processing following naphtha processing of line 22. Naphtha processing following naphthalene processing Naphtha Naphtha processing following methylnaphthalene processing of line 5 . of line 4. processing following toluene-nitrogen processing of line 1.
+
...
...
...
+
~~
300
INDUSTRIAL AND ENGINEERING
CHEMISTRY
probably the acid sites at the severity involved in this Lvork, (5, 6, 77). This concept of reversible adsorption of poisons on acid s tes, moreover, is compatible with xhe already established (7) reversibility of poison adsorption on silica-alumina catalysts (like the base of the reforming catalyst) during catalytic cracking. The nitrogen results establish a path through the gas phase by which poisons on the rate controlling acid sites can migrate to the platinum sites. and vice versa. The results do not establish this as the only path but they do make it logical. Reactivation by Hydrogen. Flowing hydrogen also reactivates aged platinum catalysts (2). I t was about twice as effective as nitrogen in a specific comparison (Table 11, Figure 1). This extra effectiveness comes, presumably, from hydrogenation of poisons which do not readily desorb, as such. during nitrogen treatment; and it implies that the poisons causing aging are unsaturated. The over-all rate of the hydrogenative reactivation-taking a week to regain only six octane numbers in activity-was much slower than the rates to be expected for hydrogenarion of unsaturates derived directly from naphtha components. This is not surprising since the reactivation must include transferring poisons from the acid sites controlling the reforming activity to the platinum sites before hydrogenation can occur. This first step might \vel1 be slower than the hydrogenation rare itself. I t might be either direct migration of all species of poisons from acid sites to platinum sites, or depolymerization of poisons like polycyclopentadienes to monomers followed by migration of the depolymerized species. Quite effective decoking of aged catalysts is possible with very high hydrogen partial pressures. At 6000 p.s.i.g., flowing hydrogen reduced catalyst coke from 17.1 wt.% (as C, based on catalyst) to 1.7 wt.% in 6 days. The hydrogen-to-carbon atom ratio of the “coke” rose from 0.5 before hydrogen treat to 1.5 after, showing that hydrogenation was actually involved (4). The catalyst here was platinum on silica-alumina that had been severely aged by gas oil hydro racking.
Reactivation by Hydrogen-Naphtha In three separate cases Mixtures. (Table 11, Figure l), processing with standard naphtha (10-to-1 hydrogennaphtha ratio) actually reactivated a catalyst after severe aging by reforming. The aging had been caused by methylnaphthalene in one case, and by a 10% blend of naphthalene in naphtha in another (both at 10-to-1 hydrogennaphtha ratio). I n the third case, the aging was caused by reforming toluene in admixture with nitrogen gas. The methylnaphthalene processing aged the catalyst so badly that subse-
P L A T I N U M REFORMING C A T A L Y S T S These initial poisons are reversibly adsorbed on the platinum sites, from which they migrate through the gas phase, at least, to the acid sites, upon which they are also reversibly adsorbed. Probably, it is the build-up of these initial poisons on the acid siteswhich are rate controlling as to reforming activity-which is reflected in the rapid aging encountered with fresh catalyst. O n either site, these initial poisons polymerize to polycyclic compounds having several double bonds per molecule and thought of as catalyst “coke.” Probably this polymerization on the rate controlling acid sites sets the rate of the gradual decline in reforming activity that follows the initial rapid deactivation. Moreover, because this polymerization is the rate determining step in that gradual decline, it must be slower than either the platinum catalyzed reactions making the initial poisons or the migration of those initial poisons to the acid sites. I n view of the conclusion reached earlier on the reversibility of the poisoning mechanism in general, the main reaction steps outlined above for the mechanism are all believed to be reversible. The whole poisoning mechanism thus becomes that shown below. The conclusion that migration of the initial poisons from the platinum sites to the acid sites is faster than their subsequent polymerization on the acid sites helps to select the detailed steps in nitrogen reactivation. Direct desorption of the polymer catalyst “coke” itself would be much slower than desorption of depolymerized coke because the vapor pressure of the polymers would be much lower. I n view of the belief that the polymerization reaction is reversible. therefore, depolymerization followed by desorption of the depolymerized material seems the more likely path during the nitrogen reactivation, depolymerization being rate controlling. In hydrogen reactivation, there is an extra drain-off of depolymerized poisons by hydrogenation. This lowers the supply of these poisons to the gas phase from the platinum sites. This, in turn, increases the concentration gradient driving their desorption from the acid sites;
quent naphtha processing produced only 73.7-octane number reformate at first (Table 11). Continued naphtha reforming, however, actually reactivated the catalyst, bringing reformate octane number u p to 81 after 4 days (line 22, Table 11). Reformate octane number then stayed constant a t 81 for 4.2 more days of naphtha reforming (line 20, Table 11). These results strongly suggest that poison forming reactions which cause aging are reversible. The last 4.2 days of naphtha reforming a t a constant octane number of 81 indicate the establishment of steady state with respect to aging. Moreover, these results (along with the other data of Table 11) strongly point to a steady state concentration of poisons which is lower than that corresponding to 77-octane number product but higher than that corresponding to 84-octane number product. I n the cases of poisoning by 10% naphthalene in naphtha and by toluene, the subsequent naphtha reforming was not carried on long enough to establish a steady state level of activity. I n each case, however, the mere fact of the reactivation itself fits in with the idea that an “equilibrium” poison concentration can be approached from the side either of fresh or of overly poisoned catalyst. Moreover, these effects are not limited to cases of poisoning by deliberately chosen specific hydrocarbons, used either alone or in naphtha blends. They have been observed in standard naphtha reforming. Here, milder reformingi.e., lower temperature-follow’ng severe reforming with the same naphtha produced a net reactivation of a badly aged catalyst. Poisoning Mechanism
The above conclusions suggest a generalized mechanism of catalyst aging. The first step is the formation, mainly at the platinum sites, of unsaturated reaction intermediates-indicated later to be mostly monocyclic diolefins along with some dicyclic polyolefins. Additional amounts may be made a t the acid sites by catalytic cracking of unsaturates being supplied continually by the platinum sites. PLATINUM SITES
,HYDROCARBON FEED (ADSORBED)
INITIAL (FAST)
\
-
nHp + POISONS (ADSORBED)
POLYMERIZED
(sLOW)-POI SONS (“COKE’) (ADSORBED)
POISONS
INITIAL
(CRACKING)
(ADSORBED)
IZED ~~,”,,”,”:,”,’~~~~; -(sLowEsT) PPOOISLOYNMSER(ZOKE‘) P
(ADSORBED)
fAnSORF?Fn\
A C I D S I T E S (RATE CONTROLLING AS TO R E F O R M I N G A C T I V I T Y )
The main reaction paths producing the unsaturated poisons causing aging are reversible
and this increases the gradient driving the depolymerization reaction on the acid sites themselves. As a result, depolymerization is speeded up. This appears as the extra effectiveness of hydrogen reactivation relative to nitrogen reactivation. Structure of the Initial Poisons
Catalyst Aging b y Toluene. When toluene was reformed in the presence of hydrogen, the net effect was to restore activity which had been lost by the catalyst during an earlier period of naphtha reforming (Table 11, Figure 1). With nitrogen replacing the hydrogen, toluene processing aged the catalyst badly (Table 11, Figure 1). This aging was not due to the nitrogen gas, per se, because nitrogen alone was beneficial (Table 11, Figure 1). Nor was it due to gross formation of nitrogen compounds because none were detected in the reaction product, and the used catalyst contained no nitrogen (within experimental error of 0.001%). T h e product liquid (99 wt.% on charge) was unreacted toluene only (by mass spectrometry). The results suggest, rather, that the toluene itself was the source of the observed aging. Cyclohexadiene is an intermediate which would be the first product of toluene hydrogenation a t the platinum site, is analogous to the cyclopentadiene that was needed to explain the cyclopentane results vis-a-vis the n-pentane results, polymerizes by the Diels Alder reaction to give the carbon atom structures of condensed polynuclear aromatics (“coke”), and has lost resonance stabilization relative to the parent aromatic. Even in the absence of deliberately added hydrogen, cyclohexadiene might be formed, the necessary hydrogen coming frcm conversion of part of the toluene to catalyst “coke.” The initial -but not the final-steps of this coke formation would, of course, have to differ from those postulated above for the case when added hydrogen is present. Methylcyclopentane a n d Methylcyclohexane. Methylcyclopentane aged the catalyst only one-tenth as rapidly as cyclopentane, and methylcyclohexanetested a t 500 p.s.i.g.-actually restored activity which had been lost during earlier naphtha reforming (Table 11). (Methylcyclohexane behaved like toluene in this respect; and, in fact, the liquid product was 4570 toluene.) These results suggest that alkylcyclqpentanes do not produce as high a concentration of harmful intermediates as cyclopentane does. This is reasonable because a drain-off reaction-producing benzene and alkylbenzenes-is available to alkylcyclopentanes which is not available to cyclopentane itself. That is, VOL. 53, NO. 4
APRJL 1961
301
alkylcyclopentanes can be isomerized to 6-membered rings a t the acid sites and then dehydrogenated to benzenes a t the platinum sites (6). Assume that the over-all conversion of methylcyclopentane to benzene is faster than the formation of methylcyclopentadiene. Then the concentration of merhylcyclopentadiene from methylcyclopentane processing might well be lower than that of cyclopentadiene from cyclopentane processing. This fits the aging results and an analogous argument applies to the methylcyclohexane results.
Catalyst Aging b y Other Aromatics. Even in the presence of hydrogen, naphthalene and 1-methylnaphthalene aged the catalyst rapidly and badly (Table 11). Some of the aging was probably caused by sulfur and/ or nitrogen impurities (Table 11, footnote d) in the case of methylnaphthalene, but it was undoubtedly hydrocarbon-derived in the naphthalene case. Partial hydrogenation of either hydrocarbon gives dihydronaphthalenes like those postulated as poisons during naphthalene formation from cyclopentane. As in the toluene case, these intermediates would have lost resonance stabilization-relative to the parent aromatic-as a result of the hydrogenation that produced them. Processing cumene in the presence of hydrogen produced benzene and propane, and was somewhat more harmful than processing toluene (Table 11, Figure 1). I n the cumene processing, cyclohexadiene is a n intermediate which is entirely analogous to the methylcyclohexadiene postulated for the toluene case. Added poisons can come from reactions of the propylene produced initially when the isopropyl group splits off from the cumene. Mono-olefins are generally recognized as sources of catalyst coke. They may therefore either be relatively weak initial poisons per se o r condense and dehydrogenate to the relatively stronger cyclodiolefin poisons. Catalyst Aging b y Paraffins. Results with n-pentane have already been discussed. n-Dodecane aged the catalyst badly (Table 11), producing alkylnaphthalenes by dehydrocyclization. Probable poisons are, again, alkyldihydronaphthalenes. n-Heptane aged the catalyst about one-quarter as rapidly as ndodecane, producing toluene by dehydrocyclization. Methylcyclohexadiene is again a probable poison here. Comparing the n-heptane results with the toluene results suggests that a higher concentration of methylcyclohexadiene is produced during n-heptane dehydrocyclization than by direct hydrogenation of toluene. This is certainly possible since thermodynamic limitations on the diene formation from toluene do not necessarily apply to the dehydrocyclization mechanism. 2,2,4-Trimethylpentane neither aged the catalyst nor produced aromatics
302
under comparable circumstances (10to-1 hydrogen-naphtha ratio). Like npentane (which also caused no aging in the presence of hydrogen), it cannot readily form cyclics because it has only 5 , carbon atoms in its longest chain. Aging Rate vs. Hydrogen-Naphtha Ratio
The; toluene results suggested that aging rate varies invereely with hydrogen-hydrocarbon ratio. This appears to be true of naphtha reforming in general, results with two quite different catalysts being shown in Figure 2. These results are consistent with the generalized poisoning mechanism. ‘With the essential steps of the mechanism all reversible, and with the dehydrogenation reactions (producing the initial poisons from the essentially saturated naphtha) faster than all others, the partial pressure, p z , of those poisons builds up to a steady state limited by thermodynamics. In effect, its value is controlled by the equilibrium : Pt Hydrocarbon feed , (gas, $HC) site initial poisons n Hz
+
(gas, P z ) (gas, PH2) Let fisc be the partial pressure of hydro~ the ~ partial pressure of carbon feed, f i be hydrogen, and K be equilibrium constant.
(With the initial poisons identified as diolefins or polyolefins, the value of n is two or more.) Total pressure, P, and hydrogennaphtha ratio, r, are defined approximately by p = PHZ +pHO (2)
+&
(3)
Under reforming conditions, p, is very much smaller than either PHC or P H ~ . I
I
I
\
1 I I I 10 20 30 40 HYDROGENlNAPHTHA (MOLE/MOLE)
Figure 2. That aging rate varies inversely with hydrogen-naphtha ratio i s consistent wiih the poisoning mechanism
0 Catalyst No. 2, 250 p.s.i.g., 2 vol. naphtha per hr. per vol. catalyst ( 2 LHSV), 99 O.N. (F-1 3 ml. TEL). A Catalyst No. 3, 200 p.r.i.g., 1 LHSV, 104 O.N. Catalyst No. 3, 100 p.r.i.g., 1 LHSV, 104 O.N.
+
INDUSTRIAL AND ENGINEERING CHEMISTRY
Neylecting p z in Equations 2 and 3 for this reason, and eliminating p ~ cand $H between Equations 1, 2, and 3 gives:
pz
=
rpn-a (1
+ 1>
pz
