3282
Ind. Eng. Chem. Res. 1997, 36, 3282-3291
Some Aspects of the Mechanisms of Catalytic Reforming Reactions† P. Govind Menon* Department of Chemical Engineering and Technology, Royal Institute of Technology, S-10044 Stockholm, Sweden
Zolta´ n Paa´ l* Institute of Isotopes of the Hungarian Academy of Sciences, P.O. Box 77, H-1525 Budapest, Hungary
A brief review is given of the evolution of our ideas on the mechanism of catalytic reforming reactions over the last 40 years. The early bifunctional reaction mechanism, though criticized much, led first to an understanding and optimization of the reforming process and later to several other catalyst and process innovations. Kinetic and tracer studies on model catalysts using single hydrocarbons as reactants and the surface-science approach revealed the finer nuances of the mechanism involved. Simultaneously, the roles of S, Cl, coke, and hydrogen and of a second metal component like Re, Ir, or Sn in the industrial catalyst were also understood to a greater or lesser extent. All these results make the mechanism of reforming reactions a field of fascinating complexity in fundamental and industrial catalysis. Introduction Ever since the advent of catalytic reforming of naphtha into high-octane gasoline in the 1950s, the mechanism of aromatization of alkanes and other reforming reactions has been a subject of intensive study. An added impulse to this came from the rapid rise of petrochemical industries in the following decades and the dedicated use of catalytic reforming to produce C6C8 aromatics as petrochemical feedstock. The bifunctional catalyst for these applications was Pt/alumina, later Pt-Re/alumina, and other bimetallic and multimetallic combinations incorporating Re, Sn, Ge, Pb, etc. During the last 15 years the remarkable catalytic activity of ZSM-5 type of catalysts to aromatize light petroleum gases (LPG) such as propane and butane came as a surprising discovery. In more recent years, efforts are being focused to convert ethane and even methane (natural gas) into C6-C8 aromatics. Our concepts and ideas on the mechanism of catalytic reforming reactions have simultaneously undergone a process of evolution. There is also a greater awareness and appreciation today of the differences to be expected in mechanism when the reforming reactions are studied under ideal or near-ideal circumstances on a “clean” metal surface or catalyst under low pressure and when the reaction is carried out in an industrial reactor with a complex mixture of reactants under pressure on a sulfided, chlorided, and partially coked “real” catalyst. During the last 25 years, many valuable studies have been undertaken by Prof. Gilbert Froment and his coworkers at the University of Gent to understand the mechanisms of catalytic reforming reactions. Hence, we regard the Froment Festschrift to be the appropriate medium to present a brief state-of-the-art review of the mechanism of reforming reactions, in particular of the aromatization of C6-C8 alkanes to aromatics. The Early Bifunctional Mechanism and Its Criticism For the metal-acid Pt/alumina bifunctional catalyst, Mills et al. (1953) proposed a bifunctional mechanism * Author to whom correspondence is addressed. † Dedicated to Prof. G. F. Froment on his 65th birthday. S0888-5885(96)00606-9 CCC: $14.00
Scheme 1. Original “Classical” Bifunctional Mechanism for Reforming Reactions of n-Hexane and Methylcyclopentanea
a The hydrogenation-dehydrogenation reactions take place on the Pt metal sites (vertical) and the carbonium ion reactions on the acid sites of alumina (horizontal). Hydrocracking and coking may involve both functions and hence are shown with broken lines. Hydrogenolysis takes place only on the Pt sites.
shown in Scheme 1. This mechanism attributed all dehydrogenation/hydrogenation reactions to Pt metal sites and all skeletal rearrangement steps (ring closure, ring contraction, and enlargement) to carbonium ion reactions occurring at the acidic sites on alumina. In general, the reactions on the metal sites are considered quite fast and the carbonium ion reactions are considered as the slow rate-determining steps. The concerted step-by-step reactions on both metal sites and acid sites are allegedly necessary to convert methylcyclopentane (MCP) to benzene and n-hexane to its branched isomers or to benzene. This scheme describes well the reactions occurring in the commercial reformer (Parera et al., 1986; Parera and Figoli, 1995). However, it should not be regarded as the exclusive pathway of aromatization. Direct C6 © 1997 American Chemical Society
Ind. Eng. Chem. Res., Vol. 36, No. 8, 1997 3283
ring closuresas opposed to the C5-C6 ring enlargement stepswas proposed by various studies (Dautzenberg and Platteeuw, 1970; Davis, 1973, 1977; Shum et al., 1986). In view of the discovery of metal-catalyzed isomerization and C5 ring closure of alkanes, Gates et al. (1979) urged scientists “to alter the reaction scheme ... to include the additional reactions (particularly cyclization and isomerization) which occur on the metal surface alone, without involving acidic centers”. An alternative pathway for aromatization would be stepwise dehydrogenation of the open-chain feed followed by ring closure of some of the polyunsaturated intermediates, e.g., hexatriene. Originally, the formation of dienes during aromatization was demonstrated over chromia catalysts by Pines and Csicsery (1962). Rozengart et al. (1964) showed that heptadienes were intermediate in the heptene-to-toluene reaction over chromia catalyst, similarly hexadienes in the hexeneto-benzene reaction over molybdena (Rozengart et al., 1971). A similar role of heptatrienes was demonstrated with heptadiene aromatization (Rozengart et al., 1966). These ideas were pursued further also on metallic catalysts. Experiments carried out with heptenes on Ni/Al2O3 (Paa´l and Rozengart, 1966) and hexenes on Pt black (Paa´l and Te´te´nyi, 1967a) demonstrated the intermediate character of dienes in both cases. The idea of the stepwise aromatization route has appeared in several papers since then (e.g., Margitfalvi and Go¨bo¨lo¨s, 1988; Davis and Derouane, 1991; Derouane et al., 1992). Iglesia et al. (1990) regarded heptatriene as a likely intermediate of n-heptane dehydrocyclization over 4% Te/NaX, as a result of 13C tracer studies. The better detection of intermediates formed in low concentrations needed, however, special experimental techniques. Detection of Overactive Intermediates on “Real” Catalysts One of the best proofs for any reaction mechanism is the experimental detection and identification of the reaction intermediates postulated therein. The great commercial success of the catalytic reforming process served as a strong stimulus to check and confirm the bifunctional mechanism shown in Scheme 1. This work was pioneered by Vladimir Haensel and colleagues at Universal Oil Products (UOP), which was and still is the foremost licensor for the catalytic reforming process and catalyst. Haensel et al. (1964) could detect methylcyclopentene as a primary intermediate product in the conversion of methylcyclopentane to benzene at relatively high liquid hourly space velocity (LHSV) up to 120. Isolating and detecting cyclohexene as an intermediate during the conversion of cyclohexane to benzene proved to be a far more difficult task. Cyclohexene is extremely reactive in the presence of hydrogen under pressure on a Pt reforming catalyst. Finally, in a specially engineered reactor setup and at an LHSV of 32 000, Haensel et al. (1964) could detect the cyclohexene intermediate as well. This provided a striking experimental confirmation for the bifunctional reaction steps shown in Scheme 1 under the “real” conditions of the catalytic reforming process. Radiotracer experiments were the other technique applied to check and verify the formation of reactive intermediates. Te´te´nyi et al. (1962) demonstrated the formation of cyclohexene during cyclohexane dehydrogenation to benzene on monofunctional metal powders. Another study (Paa´l and Te´te´nyi, 1967b) verified that hexenes were produced from n-hexane over both mono-
Scheme 2. Extended Reaction Scheme for Reforming Reactions of n-Hexanea
a Metal-catalyzed stepwise dehydrogenation of the open-chain alkane followed by ring closure of polyunsaturated intermediates is included, together with metal-catalyzed isomerization as well as C5-cyclization and ring-opening reactions. Adapted from Paa´l et al. (1987).
functional Pt black, Ni black, and bifunctional Pt/Al2O3 as opposed to “direct” alkane f cyclohexane ring closure proposed by Kazansky et al. (1948). The application of labeled molecules made it possible to detect the formation of very low amounts of hexatriene produced during aromatization of [14C]-n-hexene over Pt- black (Paa´l and Te´te´nyi, 1968b). Using the mixture of [1-14C]-1-hexene and inactive 1,3,5-hexatriene, radioactivity appeared in the unreacted hexatriene which was not radioactive originally. It must have originated from the radioactive 1-hexene; hence, the 1-hexene f hexatriene reaction was demonstrated. Dautzenberg and Platteeuw (1970) subsequently proposed the formation of hexatriene over Pt/alumina and its cyclization in the gas phase. The gas-phase reaction was soon confirmed experimentally (Paa´l and Te´te´nyi, 1973) while studying the transformation of various hexadiene and hexatriene isomers and the effect of hydrogen on these reactions. This study also contained additional information by directly comparing the aromatization of various double-bond isomers of hexadienes and both geometric isomers of 1,3,5hexatriene. The same paper proposed that transconfigurations of the intermediate polyenes may be coke precursors while cis-isomers would lead directly to aromatics. Once hexatriene was formed, it would undergo a rapid cyclization step on the surface without desorption (Paa´l, 1992). Teplyakov and Bent (1996) recently provided direct experimental evidence of such surface species by combining temperature-programmed desorption/reaction of 1-hexene on Cu3Pt single crystal with surface reflection-absorption IR spectroscopy. The spectrum measured at a temperature preceding benzene desorption could be attributed to a hexa-σ-adsorbed 1,3,5-hexatriene. A new, more comprehensive scheme (Scheme 2) includes also the metal-catalyzed reactions (Paa´l, 1987). As far as the true ring closure step is concerned, the radiotracer method gave uncertain results whether
3284 Ind. Eng. Chem. Res., Vol. 36, No. 8, 1997 Table 1. Reactions of Hexatrienes over Pt Black (T ) 360 °C, He Carrier Gas, 5 mL Pulses onto 0.4 g of Pt)a trans-triene (pulse no.; conv, %)
cis-triene (pulse no.; conv, %)
1; 37.4 2; 34.1 5; 30.8 1; 100 2; 100 5; 99.6
