An Experimental and Theoretical Study of Reaction Steps Relevant to the Methanol-To-Hydrocarbons Reaction

Home , Methylcyclopropane, Rearrangement reaction, Transalkylation

An Experimental and Theoretical Study of Reaction Steps Relevant to the Methanol-to-Hydrocarbons Reaction

Stian Svelle

Dissertation for the degree of Doctor Scientiarum

Department of Chemistry Faculty of Mathematics and Natural Sciences University of Oslo

August 2004

Preface

The work that constitutes this thesis was carried out in the period August 2000 to August 2004 at the Department of Chemistry, Faculty of Mathematics and Natural Sciences, University of Oslo. The Norwegian degree of Doctor Scientiarum comprises two years of research and one year of courses and seminars. In addition, one year of teaching at undergraduate level has been performed. The work has been financed by the Department of Chemistry, University of Oslo. The Norwegian Research Council (NFR) is acknowledged for a grant of computer time at the Norwegian supercomputing facilities. My main supervisor has been Professor Stein Kolboe. Stein is greatly acknowledged for his inspiring and close guidance throughout this period. Professor Karl Petter Lillerud and Professor Unni Olsbye have functioned as co-supervisors and have contributed with invaluable help and discussions. Dr. Ole Swang is greatly acknowledged for providing a flying start to the quantum chemical calculations. Førsteamanuensis Per Ola Rønning deserves my gratitude for finding the time to help with the isotopic labeling experiments and for having constructed the reactor system. Professor Einar Uggerud and the ever patient Dr. Osamu Sekiguchi, should be thanked for a very successful cooperation with the gas-phase methylbenzene experiments. Finally I would like to express my gratitude towards my fellow students Bjørnar Arstad, Morten Bjørgen, Morten B. Jensen, and Anastasisa Virnovskaia for providing a pleasant environment, both socially and scientifically.

Stian Svelle Oslo, August 2004

Table of contents

LIST OF PAPERS ii

1. INTRODUCTION 1

1.1. Catalysis 1

1.2. Zeolites 2 1.2.1. Historical development 2 1.2.2. The structure of zeolites 3 1.2.3. Properties and applications of zeolites 5 1.2.4. Zeolite ZSM-5 (MFI) 6

1.3. Reactions relevant to this work 8 1.3.1. Methanol-to-hydrocarbons 8 1.3.2. Conversion of halomethanes to hydrocarbons 17 1.3.3. Conversion of light alkenes 18

2. EXPERIMENTAL METHODS 20

2.1. The reactor system 20 2.2. Product analysis 21 2.3. Procedure for determining isotopic distributions 22

3. COMPUTATIONAL METHODS 24

3.1. The cluster model 24

4. THIS WORK 28

4.1. Scope 28 4.2. Synopsis of results 29 4.3. Main conclusions 42

5. REFERENCES 44

APPENDIX 53

i List of papers

This thesis is based on the following manuscripts, referred to by their corresponding roman numerals in the text. The manuscripts are collected in the Appendix.

Paper I: A Theoretical Investigation of the Methylation of Methylbenzenes and Alkenes by Halomethanes over Acidic Zeolites. S. Svelle, S. Kolboe, O. Swang, J. Phys. Chem. B 107 (2003) 5251-5260.

Paper II: A Theoretical Investigation of the Methylation of Alkenes with Methanol over Acidic Zeolites. S. Svelle, B. Arstad, S. Kolboe, O. Swang, J. Phys. Chem. B 107 (2003) 9281-9289.

Paper III: Theoretical Investigation of the Dimerization of Linear Alkenes Catalyzed by Acidic Zeolites. S. Svelle, S. Kolboe, O. Swang, J. Phys. Chem. B 108 (2004) 2953-2962.

Paper IV: Kinetic Studies of Zeolite-catalyzed Methylation Reactions. 1. Coreaction of [12C]ethene and [13C]methanol. S. Svelle, P.O. Rønning, S. Kolboe, J. Catal. 224 (2004) 115- 123.

Paper V: Kinetic Studies of Zeolite-catalyzed Methylation Reactions. 2. Coreaction of [12C]propene or [12C]n-butene and [13C]methanol. S. Svelle, P.O. Rønning, S. Kolboe, Preliminary manuscript, to be submitted to J. Catal.

Paper VI: The Intermediates in the Methanol-to-Hydrocarbons (MTH) Reaction: A Gas Phase Study of The Reactivity of Polymethylbenzenium Cations. S. Svelle, M. Bjørgen, S. Kolboe, D. Kuck, U. Olsbye, O. Sekiguchi, E. Uggerud, Preliminary manuscript, to be submitted to Phys. Chem. Chem. Phys.

ii 1. Introduction

1.1. Catalysis Most industrial chemical processes, such as oil refining or polymer production, rely heavily on catalysis and would not be feasible without its implementation. In living organisms, biological catalysts, or enzymes, are involved in most molecular reactions. Hence, catalysis plays a key role in our every-day lives. Scientists in the early 19th century discovered that many chemical reactions could be accelerated by the presence of certain compounds, such as acids or metals. Only trace amounts of these substances were required for a pronounced effect to be seen, and these added substances were not themselves consumed in the reactions. The first attempt to rationalize these effects was made by the Swedish scientist Berzelius in 1836, and the concept of catalysis and catalysts was introduced [1]. In modern chemistry, a catalyst is defined as a substance that increases the rate at which a chemical system approaches equilibrium without being consumed in the process. Thus, the presence of a catalyst allows a reaction to proceed more efficiently or under milder conditions than would otherwise be possible. A catalyst will affect only the rate at which chemical equilibrium is approached and not the overall thermodynamics and the equilibrium concentrations. Such an enhanced rate of Energy reaction may be the result of either a lowering of the activation energy as

Activation illustrated in Figure 1.1, or an increase in the energy for uncatalyzed reaction number of collisions between the reactants. Activation energy for catalyzed The field of catalysis may be divided reaction into three parts, heterogeneous, Reactants homogeneous and enzymatic catalysis. In Reaction energy heterogeneous catalysis, the reactants and Products the catalysts are present in different physical

Reaction coordinate states, e.g. a solid catalysts and gaseous

Figure 1.1. Schematic illustration of the reactants. In homogeneous catalysis, the lowered activation energy resulting from the introduction of a catalyst to a reaction reactants and the catalyst constitute a single system. physical state, e.g. a liquid solution. In

1 enzymatic catalysis, the catalysts are biological macromolecules. The research presented in this thesis concerns heterogeneous catalysis exclusively, where gases are reacted over solid catalysts, and the investigated catalyst belongs to a class of materials known as zeolites.

1.2. Zeolites

1.2.1. Historical development The history of zeolites began in 1756 when the Swedish mineralogist Cronstedt discovered a suite of well-formed crystals in northern Sweden. He named these new minerals “zeolite” from the Greek words “ζειν” (zein – boiling) and “λιθοσ” (lithos – stone), in allusion to their frothing loss of water and steam upon heating. The mineral discovered by Cronstedt is now known as stilbite. Zeolites soon became recognized as minor, but ubiquitous constituents of every basalt formation and in many rocks of similar origin. Zeolites are often found as beautiful crystals in nature, and jewelry represented just about the only commercial interest in zeolites for nearly 200 years [2]. Some of the characteristic properties of zeolites were described quite early. In 1777, Fontana described the adsorption of charcoal [3], and in 1840 Damour observed that crystals of zeolites could be reversibly dehydrated with no apparent visual change [4]. In 1858 Eichhorn demonstrated that chabazite and natrolite exhibited reversible ion exchange [5]. Weigel and Steinhoff (1925) noted that dehydrated zeolite crystals would adsorb small organic molecules and reject larger ones [6]. This phenomenon was described in 1932 by McBain as “molecular sieving” [7], and zeolites are frequently referred to as molecular sieves. The first structures of zeolites were determined in 1930 by Taylor and Pauling [8-10]. These observations were greatly extended by the pioneering work of Barrer, which commenced in 1938. He presented the first classification of the then known zeolites based on molecular size considerations in 1945 [11] and in 1948 reported the first definitive syntheses of zeolites [12]. In 1949 workers at Union Carbide Corporation’s Linde Division, directed by Milton, synthesized zeolites by a low-temperature hydrothermal process, the approach most commonly employed for zeolite synthesis today. Milton and co-worker Breck synthesized a number of commercially significant zeolites, among which type A, X and Y [13]. These findings initiated the present large-scale applications of synthetic zeolites as desiccants, ion exchangers and in separation processes.

2 In 1962 Mobil Oil introduced the use of synthetic zeolite X as a cracking catalyst. Zeolite catalysis was taken one step further in the late sixties when researchers at Mobil were able to synthesize high silica zeolites such as ZSM-5 (see section 1.2.4) and zeolite β. Further development within the field of zeolites involved the inclusion of phosphorous or metals in the structures, leading to new zeolite-like (zeotype) materials, typically denoted SAPO, AlPO4 or MeAPO4 [13]. A very recent advance has been the development of techniques for delamination of zeolites and producing zeolite nanoparticles [14], and continuous efforts are being made to improve existing and discover new materials throughout the research communities. To date, approximately 40 zeolites have been discovered in nature and about 130 zeolite structures have been synthesized in laboratories. Moreover, a far greater number of hypothetical and still plausible frameworks may be envisaged [15].

1.2.2. The structure of zeolites

Zeolites are a class of materials built up from aluminum, silicon and oxygen. They also contain charge balancing cations and adsorbed water. Structurally, zeolites may be viewed as crystalline materials based on a three dimensional network of TO4 tetrahedra, where T is Si or Al, connected by sharing oxygen atoms at each tetrahedral corner. As illustrated in Figure 1.2, these primary structural units form larger secondary building units that are combined to form three-dimensional framework structures. These structures are microporous, in the sense that there are cavities and channels, or pores, of molecular dimensions in the frameworks. Zeolites are typically classified as 8-ring, 10-ring or 12-ring

Figure 1.2. Several tetrahedra (top left) may share corner structures, based on the oxygens and form larger, double six-rings (top right). number of tetrahedral atoms These may be combined to form a 12-ring and parts of the gmelinite framework (bottom). (T-atoms) comprising the

3 circumference of the pores. Hence, a 10-ring could be described as a 20-ring, if the oxygen atoms linking the T-atoms were included. Typical pore dimensions are 3.0 – 4.5 Å for small pore zeolites, 4.5 – 6.0 Å for medium pore zeolites and up to 8.0 Å for large pore zeolites [16]. 4- A zeolite built up exclusively from [SiO4] tetrahedra would constitute an overall electrically neutral lattice with chemical formula SiO2. Each substitution of a silicon 5- tetrahedron with an [AlO4] tetrahedron will therefore result in one negative electric charge in the framework, and these negative charges must be balanced by cations located within the pores of the zeolite in order to maintain overall electric neutrality. The general chemical formula is given in Equation 1.1,

M 2 / nO ⋅ Al2O3 ⋅ y SiO2 ⋅ w H 2O (1.1) where n is the valence of the cation M, y may vary from 2 to infinite and w is a measure of the number of water molecules inside the zeolite void. The charge compensating cations M may in principle be any cation, and if they are protons, the zeolite becomes a solid acid, giving rise to the use of zeolites as acidic catalysts. Proton exchange is often indicated by adding an H to the name of the zeolite, such as Figure 1.3. Brønsted site. H-ZSM-5 and H-Y. This thesis deals strictly with proton exchanged zeolite catalysts.

4 1.2.3. Properties and applications of zeolites Zeolites possess several unique properties that constitute the basis for their widespread applications.

• Zeolites have well-defined crystal structures • Zeolites have sharply defined pore sizes, giving rise to selectivity with respect to reactants, transition states and products • Zeolites have large surface areas, typically several hundred m2/g BET areas • Zeolites have mobile, exchangeable cations • Zeolites are thermally stable • Zeolites are non-toxic and not environmentally harmful • Metals may be deposited within the zeolite voids, giving rise to bifunctionality

The largest volume application of zeolites is as ion-exchangers in laundry detergents. “Hard” calcium and magnesium ions in the washing water are exchanged with “soft” sodium cations. This prevents the precipitation of calcium and magnesium surfactant salts, which would result in a less efficient cleaning. Zeolites have to a large extent replaced the less desirable phosphates as detergent builders. Zeolites are employed as adsorbents and in separation processes. The sharply defined size of the pore apertures makes zeolites particularly suited for applications such as the separation of normal- and iso-paraffins [17]. The use of zeolite A as a desiccant, or selective water remover, is familiar to most chemists. The first use of zeolites as catalysts occurred in 1959 when zeolite Y was used as an isomerization catalyst by Union Carbide [17]. Presently, the major employment of zeolites within catalytic chemistry is as acid cracking catalysts. It can be estimated that more than 99% of the world’s gasoline production from crude oil involves the use of zeolite catalysts, and faujasite-type zeolites (X and Y) account for at least 95% of the zeolite catalysis market, by volume [13]. Zeolites are often referred to as shape-selective catalysts [17]. This notion is inherently linked to the crystallinity and well-defined pore systems of these materials. Three types of shape selectivity are usually considered, reactant shape selectivity, product shape selectivity and transition state shape selectivity. The pore-openings of a zeolite may be such that certain reactants are occluded, whereas others are allowed to diffuse into the catalyst crystal and be

5 transformed into products. This is called reactant shape-selectivity. Another possibility may be that several products are formed inside fairly large voids or cages of the catalyst, but only those smaller than a certain critical size are permitted to leave the catalyst through narrow pore apertures. The larger product molecules are trapped inside the zeolite cages unless further reactions occur, resulting in the formation of smaller molecules. This is called product shape-selectivity. Finally, the zeolite topology may be such that there are spatial (steric) restrictions prohibiting the progress of reaction via a bulky activated complex, whereas other reactions, proceeding via geometrically smaller transition states, are allowed. Transition state selectivity is conceptually more complex than reactant and product shape selectivity and may be harder to assess than the two other types. Nonetheless, recent research tends to focus more and more on this aspect of shape selectivity. [18]

1.2.4. Zeolite ZSM-5 (MFI)

The second most used zeolites in catalysis, after faujasite type zeolites, is zeolite ZSM-5, Zeolite Socony Mobil, which is also designated as MFI, Mobil Five. ZSM-5 was originally patented by Argauer and Landolt at Mobil Oil in 1972 [19]. The structure was first reported by Kokotailo et al. in 1978 [20], and it was later refined by X-ray single crystal methods in 1981 [21]. The following data describe the structure [22]:

Unit cell symmetry: orthorhombic Unit cell constants: a = 20.07 Å, b = 19.92 Å, c = 13.42 Å Space group: Pnma 96 T-atoms per unit cell Framework density: 17.9 T atoms per 1000 Å3

Structural formula, sodium exchanged: Nan[AlnSi96-nO192] ~ 16 H2O n < 27

The pore system of ZSM-5 is three-dimensional and it consists of one set of straight channels running along [100] with dimensions 5.1 × 5.5 Å intersected by a set of zig-zag channels along [010] with dimensions 5.3 × 5.6 Å, as shown in Figure 1.4a. Movement in the [001] direction is cumbersome, and diffusing molecules are required to move back and forth between the two pore systems. At the channel intersections there are slightly larger voids,

6 which may accommodate molecules that do not easily diffuse in the channels themselves, see Figure 1.4b.

Figure 1.4. The pore system (a) and the channel intersections (b) of ZSM-5 [16,23].

Houzvicka et al. have estimated the diameter of these intersection voids to be close to 9 Å [23], whereas Davis and Zones have suggested a diameter closer to 10 Å [24]. More details on the structure are given in Figure 1.5.

Figure 1.5. From left to right: view down the straight channels of ZSM-5, view down the sinusoidal channels, view along the third, less penetrable direction. Eight unit cells have been rotated to generate the drawings.

7 1.3. Reactions relevant to this work

The main focus of this thesis has been to study the reaction in which methanol is converted into higher hydrocarbons over acidic zeolite catalysts, and a brief review of this process is given in section 1.3.1. below. The conversion of halomethanes to hydrocarbons is relevant by analogy, and has been studied using theoretical methods. A short introduction to this subject can be found in section 1.3.2. During the conversion of methanol to hydrocarbons, fairly large concentrations of alkenes desorb from the catalyst, and the greater part of this thesis concerns the interaction of these alkenes with methanol. As part of this, the reactivity of these alkenes without methanol present has been evaluated, using both theoretical and experimental methods. Relevant background information is given in section 1.3.3.

1.3.1. Methanol-to-hydrocarbons

It becomes ever clearer that the world reserves of oil eventually will be overshadowed by those of coal and natural gas. Unless revolutionary technology is developed, which might seem uncertain, a large part of our energy will come from coal and gas in the future. The methanol-to-hydrocarbons (MTH) process represents parts of a possible route for the upgrading of natural gas or coal to higher value products, such as gasoline or small alkenes [25,26]. These small alkenes may serve as feedstock for several petrochemical processes. Natural gas or coal may be transformed to synthesis gas (CO and H2) by steam reforming or gasification, which is subsequently reacted to form methanol over a

Cu/ZnO/Al2O3 catalyst. Methanol may then be converted into a mixture of hydrocarbons using acidic zeolite or zeotype catalysts. An overview is given in Figure 1.6.

Figure 1.6. Schematic view of the upgrading of coal, oil and gas [25].

Mobil Oil was the first to discover and develop this zeolite-based MTH technology, resulting in the methanol-to-gasoline (MTG) process; in which methanol is converted to gasoline over ZSM-5 derived catalysts. A typical MTG plant was originally operated in a fixed bed reactor at several bars of methanol and about 400 °C [26]. Later, fluidized bed technology was developed. The product distribution is given as a function of the inverse feed rate in Figure 1.7.

Figure 1.7. Product distribution from methanol over H-ZSM-5 as a function of the inverse feed rate or contact time [26].

9 The first step of the methanol-to-hydrocarbons reaction is the dehydration of methanol to form dimethylether and water. By weight, water constitutes 56% of the products in a complete MTH reaction. At intermediate feed rates, more than half of the hydrocarbon product is alkenes. At the lowest feed rates, a high-octane gasoline blend of aromatics and alkanes is formed. However, current environmental considerations have led to a decrease in the desired content of aromatics in gasoline, and the MTG raw product cannot presently be used without additional blending. Later, Norsk Hydro and UOP jointly developed the methanol-to-olefins (MTO) variant of the reaction, where ethene and propene are the main products formed over SAPO- 34 zeotype catalyst systems. The H-SAPO-34 catalyst is a silicoaluminophosphate zeotype material with fairly large, cages (about 7 × 10 Å) connected by 8-ring windows (3.8 Å diameter) [27]. SAPO-34 is iso-structural to chabazite, meaning that the framework structures are identical. The narrow pore openings give rise to product shape selectivity, and only small, linear alkenes may diffuse through the apertures and out of the catalyst crystals. The MTO reaction is carried out at slightly higher temperature in a fluidized bed reactor with continuous catalyst regeneration. The latest addition to this field is the methanol-to-propylene (MTP) alternative, currently offered by Lurgi [28]. Presently, the economic incentive is greater for running an MTO plant than an MTG operation. The value of gasoline made from natural gas by MTG technology is slightly less than that of the methanol; whereas the value of alkenes produced from gas is approximately double that of methanol. Despite the apparently great potential for industrial application of the MTH reaction, the only full-scale commercial operation to date was the MTG plant operated on New Zealand, where gasoline production was started in 1986 and later shut down due to drops in the price of oil relative to that of methanol [29]. However, an MTO plant is expected to come on-stream in Nigeria in 2006, as part of a natural gas-to-polymers project [30].

The reaction mechanism of the MTH reaction has been the subject of vast amounts of research, and it has also been the focus of this thesis. Several recent reviews that cover this subject can be found in the literature [25,26,31-33]. Initial research focused on possible routes for the formation of initial carbon-carbon bonds from C1 units, i.e. methanol or dimethylether. More than 20 possible mechanisms have been proposed, encompassing a variety of reactive intermediates. Among them are oxonium ylides [34-36], carbocations [37,38], carbenes [39- 41], and free radicals [42,43]. However, most of these proposals require the participation of active sites usually not associated with zeolites, such as strongly basic sites for deprotonation

10 of oxonium ions or carbene formation and superacidity for carbonium ion formation [33]. Electron spin resonance (ESR) investigations have provided little evidence for the participation of radicals [44]. Also, several mechanistic proposals have later been examined using theoretical methods, most often resulting in quite high barriers [45-47]. The issue of C–C bond formation directly from methanol was recently readdressed by Song et al [48]. It was realized that if the rate of direct methanol conversion is very low, it might possibly in fact be zero in the absence of hydrocarbons. When using reactants, carrier gases and catalysts extensively purified from hydrocarbon impurities (methanol usually contains traces of ethanol, the catalyst may contain traces of uncombusted template, and carrier gases may also contain ppm levels of hydrocarbons), the initial rate of methanol conversion was reduced by orders of magnitude. The following points may summarize the findings of Song et al. concerning direct methanol conversion.

• It seems exceedingly clear that direct formation of C–C bonds directly from C1 units is of no importance during steady state conversion. • It is unlikely that direct conversion is important during an induction period with normal feedstocks. Any such reactions are probably overshadowed by the coreaction of methanol with hydrocarbon impurities from various sources.

The hypothesis stating that the rate of direct methanol conversion is exactly zero is of course difficult to prove. This issue is of little practical importance, but remains an intriguing question for the devoted scientist [49,50]. Some characteristics of the MTH reaction were described quite early. The MTH reaction is autocatalytic [26,37,51], in the sense that the presence of small amounts of products leads to an enhanced rate of conversion, until steady state is eventually reached. Linked to this, an induction period [37,50,52,53] is often observed, meaning that the level of methanol conversion increases with time during the initial stages of the reaction. It was also recognized that continuous production of hydrocarbons directly from C1 units could not be a major part of the MTH reaction during steady state conversion. In 1986, Dessau, from Mobil’s laboratories, stated that “asking where the first olefin comes from is analogous to asking where the first peroxide comes from in an autooxidation reaction” [54]. Based on this, Dessau and co-workers [54,55] proposed the indirect reaction scheme shown in Figure 1.8.

Figure 1.8. Scheme for steady state methanol conversion pathway, as proposed by Dessau [54].

According to the scheme in Figure 1.8, after the necessary alkenes somehow are formed during the induction period, all alkenes after that are produced via repeated methylations, oligomerization and cracking. Aromatics and alkenes are end products of cyclization reactions and hydrogen transfers (aromatization). This mechanism has been referred to as a consecutive mechanism [25]. An alternative to this alkene-based scheme exists, focusing on cyclic species as the propagator of an indirect alkene formation route. This pathway is commonly referred to as the hydrocarbon pool mechanism. The profound influence of aromatics on the MTH reaction was first noted by Mole and co-workers [56,57]. It was observed that adding small amounts of toluene or para-xylene resulted in an enhanced rate of methanol conversion; the effect was called aromatic co-catalysis. Mole presented the rather elaborate reaction scheme shown in Figure 1.9 to rationalize the results. The key features are: Protonation and subsequent deprotonation of a methylbenzene leads to the formation of an exocyclic double bond

Figure 1.9. Left side: Protonation and proton loss from a methyl group leads to the formation of an exocyclic double bond. Right side: Methylation on the exocyclic double bond followed by de-ethylation results in ethene formation [57].

12 (prototropic shift). Methylation on the exocyclic double bond (side-chain methylation) leads to the formation of higher alkyl chains, which are split off as alkenes. The validity of the alkene based reaction scheme in Figure 1.8 was investigated by Dahl and Kolboe [58-60]. Carbon-13 labeled methanol was coreacted with ethene or propene over an H-SAPO-34 catalyst. Both ethene and propene displayed little reactivity when coreacted with methanol. Most of the products were formed exclusively from methanol, and the alkenes were basically inert under the reaction conditions investigated. However, later studies, and the results presented in this thesis, have shown that propene has a greater reactivity on H-ZSM-5 type catalysts [61,62]. The low reactivity of the alkenes disagrees with the consecutive mechanism shown in Figure 1.8, and an indirect, parallel mechanistic scheme was suggested; see Figure 1.10. The hydrocarbon pool, (CH2)n, was said to represent an adsorbate which may have many characteristics in common with ordinary coke and might easily contain less hydrogen than indicated [58-60]. The chemical structure of the pool was not further specified.

Figure 1.10. The hydrocarbon pool scheme, as originally proposed by Dahl and Kolboe [58].

13 13

Subsequently, [ C]methanol e n e

has been coreacted with benzene or B M lbenz a r

(a) t hy toluene, and it was concluded that an t Te

arene, or some arene derivative, is B benzene l M entame P involved in forming a substantial part, Tri xamethy lenes or all, of the propene [63]. Arstad and e H Xy Kolboe investigated the stability of species trapped inside the H-SAPO-34 9 1011121314151617181920 cavities during methanol conversion, (b) and their results further pointed towards polymethylbenzenes as key components of the hydrocarbon pool [64,65]. The reaction was thermally 9 1011121314151617181920 quenched after short times on stream, with or without flushing with carrier (c) B gas after the feed was stopped. The B HM spent catalyst was dissolved in PM hydrochloric acid, thereby liberating the material retained inside the cages 9 1011121314151617181920 Retention time / minutes for GC-MS analysis. Figure 1.11 a) and b) show the build-up of Figure 1.11. Total ion chromatograms of the CCl extract of SAPO-34 dissolved in HCl. The polymethylbenzenes as the major 4 MTH reaction was run at 325 °C. a) 30 and b) components of the retained material. 120 sec. on stream without flushing. c) 120 sec. on stream followed by 160 sec. of flushing with Upon flushing with carrier gas after carrier gas [64]. the feed was stopped, c), the concentrations of hexa- and pentamethylbenzene in the pores were reduced, whereas the lower homologues were unaltered or increased, and it was suggested that penta- and hexamethylbenzene shed two or three methylgroups to form ethene or propene units and a lower methylbenzene homologue [64]. Bjørgen et al. investigated the reactivity of polymethylbenzenes over the large pore zeolite H-β, which allows

Figure 1.12. The molecules as large as hexamethylbenzene to leave and enter the heptamethylbenzenium cation. pore system [66-69]. Experiments relying on extensive use of

14 isotopically labeled reactants were carried out, and the heptamethylbenzenium cation (Figure 1.12) was identified as a key intermediate. The reaction scheme for monomolecular alkene formation from the heptamethylbenzenium cation shown in Figure 1.13 was in accord with the observations made. This scheme was adapted from a similar mechanistic scheme, known as the paring mechanism, which was proposed by Sullivan et al. [70] in 1961 to rationalize the product distribution observed when hexamethylbenzene was reacted over a bifunctional nickel sulfide on silica-alumina catalyst or over the purely acidic silica-alumina support.

* * * * * * * *CH3OH

C6H6 H-Zeolite *CH3OH + * * * * -

* * Zeolite

* * * * *

* * * *

+ *

H-Zeolite + -

* Zeolite

* * + * * -H

* * * Zeolite

* *

+ Zeolite- *

- - *

Zeolite +

* + *

Ze* olite

* + *

* *

* -

* Zeolite

* * *

-H *

* * *

* +

* * * * +

*CH3OH *CH OH * 3 -H *

Zeolite-

* * H-Zeolite * H-Zeolite H-Zeolite

Figure 1.13. Monomolecular rearrangements of the heptamethylbenzenium cation leading to propene and iso-butene formation [67]. Adapted from the paring reaction scheme, as originally proposed by Sullivan et al [70].

In parallel to the work of Kolboe and co-workes, Haw and co-workers have conducted a series of investigations that shed further light on the mechanism of the MTH reaction [71,72]. Haw and co-workers investigated the MTH chemistry on an H-ZSM-5 catalyst [73,74]. By using NMR techniques and a pulse-quench catalytic reactor, it was shown that methylated cyclopentenyl cations were stable at certain reaction conditions. The 1,3- dimethylcyclopentenyl cation (Figure 1.14) was found to be an intermediate in the synthesis

15 of toluene. It was also suggested that methylated cyclopentenyl cations might function as reaction centers for alkene formation, via side-chain methylations and skeletal isomerizations leading to cations with higher alkyl Figure 1.14. The1,3- dimethyl-cyclopentenyl substituents, which may be eliminated as ethene or propene cation [74]. Subsequent studies identified methylbenzenes as the organic reaction centers for methanol to hydrocarbons catalysis on H-SAPO-34 catalysts [75-78], in accord with the reports from Arstad and Kolboe mentioned above [64,65]. These studies were extended to include zeolite H-beta, and by investigating the reactivity of several polymethylbenzenes and butylbenzene isomers, the mechanistic scheme shown in Figure 1.15 for propene formation was proposed [79,80]. Ethene and isobutene formation may proceed analogously, invoking

Figure 1.15. A detailed side-chain methylation route leading to propene [79]. only one or three deprotonation/methylation steps, respectively. In contrast to the scheme proposed by Mole and co-workers (Figure 1.9), this scheme includes the heptamethylbenzenium cation as an important intermediate. This carbenium ion is a gem- dimethylated ion, which cannot loose its positive charge by removing a proton bonded directly to the aromatic ring. Additional spectroscopic evidence in favor of side-chain methylation on H-ZSM-5 catalysts has been obtained by Hunger and co-workers [81], who also have employed in situ NMR to examine several aspects of MTH chemistry [82-84]. It should also be noted that a lower homologue of the heptamethylbenzenium ion, the gem- dimethyl isomer of the pentamethylbenzenium ion has been observed in zeolite H-ZSM-5

16 [85]. The plausibility of the side-chain methylation route via several gem-dimethylated benzenium ions was later confirmed by computational methods [86]. The differences between a paring type mechanism and side-chain methylation are subtle, and there is no definitive proof available in favor of either pathway being the exclusive mechanism. As with so many aspects within heterogeneous catalysis, this issue probably depends on many factors, such as reaction conditions, catalyst acid site density, catalyst acid strength, and catalyst topology. Indeed, the two indirect mechanisms, i.e. a paring type reaction and side-chain methylation, might not be mutually exclusive. Regardless of the exact details, there now appears to be growing consensus in favor of the indirect hydrocarbon pool mechanism.

1.3.2. Conversion of halomethanes to hydrocarbons

The conversion of halomethanes to hydrocarbons over acidic zeolites is a reaction with clear analogies to the MTH reaction. The products from this reaction are hydrocarbons and the corresponding hydrogen halide; see Equation 1.2. X denotes the halogen.

n CH 3 X → (CH 2 ) n + HX (1.2)

The product distribution obtained from halomethanes (most frequently chloromethane) is similar to that obtained from methanol, and this is often considered to imply that similar reaction pathways are operative on the catalyst surfaces [87-89]. The main focus of previous research has been on the production of gasoline by reacting chloromethane over H-ZSM-5 catalysts. Good levels of catalyst activity and lifetime have been reported [89,90]. In a very recent publication, Lorkovic et al. describe the conversion of bromomethane over a modified ZSM-5 catalyst [91]. Some efforts have also been directed at maximizing the yields of light alkenes. Sun et al. used a phosphorous modified Mg-ZSM-5 catalyst to convert chloromethane into mainly ethene, propene, and butenes [92]. Jens et al. reacted chloromethane over several acidic zeolites at high feed rates (WHSV = 24 h-1 and T = 400 °C for H-ZSM-5), and fairly high yields of light alkenes were observed [93].

17 It should be noted that around 4% halogenated aliphatics are typically reported, the most abundant halogenated products being 2-chloropropane and chlorobutanes [94]. An extensive list of references covering the literature on halomethane conversion is given in paper I.

The potential advantage of CuCl halomethanes over methanol is the KCl CH4 + O2 H2O circumvention of the production of synthesis LaCl3 gas, as halomethanes may be produced HCl directly from methane. This is possible CH 3Cl through the oxohydrochlorination process (OHC) [90], where methane is reacted with Hydrocarbon products H-Zeolite hydrogen chloride and oxygen to form chloromethane and water over a supported Figure 1.16. Cyclic pathway for the copper chloride catalyst; see Figure 1.16. conversion of methane to hydrocarbons. In Chlorination may also be effected over a this cycle, oxohydrochlorination of methane to form chloromethane is combined with range of super-acid catalysts or by conversion of chloromethane to monohalogenation of methane over hydrocarbons over an acidic zeolite catalyst [90]. supported platinum catalysts [95]. A drawback associated with the use of halomethanes as intermediates in the upgrading of natural gas to higher hydrocarbons is the unavoidable production of HX. Even if these acids are recycled, they are highly corrosive and difficult to handle, especially in an industrial scale. Also, the halomethanes themselves and the abovementioned content of halogenated aliphatics in the raw product are harmful, and this aspect requires particular attention.

1.3.3. Conversion of light alkenes

Alkene reactions, such as dimerization and cracking, are quite relevant to the MTH reaction. These reactions may take place simultaneously with other reaction steps, e.g. methylation, and it is difficult to distinguish between different reaction routes. Moreover, oligomerization of light alkenes such as propene or butenes to higher molecular weight products is an extensively studied field within petrochemistry in its own right [96]. As ever more stringent limits are set for the content of aromatics in engine fuels,

18 alkene oligomerization represents an important route for the production of high-octane environmentally friendly liquid fuels. The traditional acid catalyst for alkene oligomerization is the solid phosphoric acid (SPA) catalyst, which consists of phosphoric acid impregnated on kieselguhr [97]. Several zeolite based processes are also available, such as the Mobil Olefin to Gasoline and Distillate (MOGD) process and the Shell Polygasoline and Kero (SPGK) process [98,99]. Reaction mechanisms invoking carbenium ions as intermediates have been very successful in rationalizing observed reactivity trends and product distributions [100]. However, the acid strength of zeolites is generally accepted to be closer to that of concentrated sulfuric acid rather than super acids, and zeolites are therefore unable to protonate simple alkenes to form persistent cations. Hence, such carbenium ion mechanisms are inadequate for a complete fundamental understanding of the oligomerization reactions. Spectroscopic methods have been useful for obtaining mechanistic data. Infrared (IR) and nuclear magnetic resonance (NMR) spectroscopic studies indicate that the alkene + alkene reaction proceeds via an initial formation of a π-complex (adsorption), followed by the activation of one alkene on an acidic site, commonly referred to as alkoxide formation, and finally carbon-carbon bond formation in a coupling step [101-106]. Computational methods have also given insight. It is primarily cracking, or alkene β- scission, which has been the focus of most theoretical reports and always via mechanisms involving alkoxides as intermediates [107-111].

19 2. Experimental methods

A general overview of the experimental techniques used is given in this section. Specific details concerning the reaction conditions are described in the appropriate papers.

2.1. The reactor system

The reactor system used for the experiments described in this thesis was designed by Rønning [112]. The catalytic reactions were performed in a U-shaped, fixed bed Pyrex microreactor (3 mm ID) with a glass sinter to support the catalyst. A layer of packed quartz wool separated the catalyst and the glass sinter. Typically, 2.5 mg of catalyst was used, in order to achieve the very high feed rates required to obtain the desired data. The catalyst powder was pressed to wafers and subsequently crushed and sieved, thereby obtaining catalyst particles in the size range 250 – 420 µm. Despite having only a thin layer of catalyst on the quartz wool/glass sinter in the reactor, reactant by-pass was negligible. Control experiments carried out at the highest feed rates that were used with 2-butanol as feed, resulted in complete dehydration of the alcohol to form butenes, thus verifying that by-pass was indeed insignificant. Blank experiments gave insignificant alcohol dehydration.

Figure 2.1. Flowchart of the reactor system [112].

20 The flowchart of the reactor system is given in Figure 2.1. Porter P-150 ball flow meters were used to control and measure the carrier gas flows (He or N2) via lines 1 and 2. Alcohols were fed by passing part of the carrier gas through a vessel containing the desired alcohol (saturation evaporators), thus saturating the carrier gas. The partial pressure of the alcohol was controlled by adjusting the temperature of the saturation evaporator and the fraction of carrier gas fed through each line. Alkene reactants were fed through the auxiliary lines (lines 3 or 4) using a needle valve flow regulator and a Top-Trak model 822-2 mass flow meter. The reaction temperature was measured with a stainless steel-sheathed thermocouple (0.5 mm diameter) placed in the catalyst bed. In order to reduce the expenditure of [13C]methanol in the isotopic labeling experiments, the feed was only admitted to the catalyst for 15 minutes prior to each analysis. After taking a sample of the effluent for analysis, the feed was stopped (maintaining the carrier gas flow) and the conditions adjusted to those desired for the next analysis. Separate tests indicated that admitting methanol for 15 minutes prior to analyzing the effluent was sufficient to reach steady state activity, thereby avoiding any initial transient effects.

2.2. Product analysis

Product analysis was performed using gas chromatography. Quantitative effluent composition was determined using an on-line Carlo Erba GC6000 Vega with flame ionization detector (FID) equipped with a Supelco SPB-5 column (60 m × 0.53 mm × 3 µm). Additional analyses were performed on a Siemens Sichromat 2-8 or an HP 6890 equipped with a

Chrompack PLOT column (Al2O3/KCl, 50 m × 0.53 mm × 10 µm), both with FID. The isotopic composition of the products was determined using an HP 6890 GC with an HP 5973 mass sensitive detector (GC-MS). Using cryostatic cooling, the HP-5MS column (60 m × 250 µm × 0.25 µm) gave adequate separation of all the compounds of interest in this work.

21 2.3. Procedure for determining isotopic distributions

A central part of this work concerns the coreaction of isotopically labeled methanol with ordinary, unlabeled alkenes. The procedure used for determining the isotopic composition of the products formed in these experiments was developed by Rønning and Mikkelsen, and has been extensively described previously [112,113]. The kinetic isotope effect has been assumed to be insignificant. Only ions corresponding to an intact carbon skeleton have been used as basis for the calculations. In order to extract the isotopic composition of a compound it is necessary to know the mass spectrum of the ordinary 12C compound. Standard spectra were obtained by reacting ordinary methanol over the catalyst. Corrections for the naturally occurring content of 13C is easily carried out by using Equation 2.1.

N Aobs (i) − ∑ Acorr (i − n) ⋅ Pn A (i) = n=1 (2.1) corr 0.9889 N

13 Acorr(i): Single ion peak area with mass number m/z = i, corrected for naturally occurring C

Aobs(i): Observed single ion peak area 13 Pn: The statistical probability of an ion with N carbons containing n C atoms, based on the number of possible permutations and the natural 13C content

To ensure reliable isotopic analysis of the products new standard mass spectra were recorded at intervals. The GC-MS system showed great stability, and variations in the standard spectra were negligible. Single ion chromatograms were extracted from the total mass spectra and integrated. The area of a single ion peak is the sum of the areas of the ions with the same mass, but with a different number of hydrogen atoms and labeled carbons. For instance, the area of the m/z = 12 12 13 27 peak in the mass spectrum of ethene is the sum of contributions from CH- CH2, CH- 12CH and 13C-13CH. The general expression for the observed peak area for a single ion is given in Equation 2.2.

22 N Aobs (i) = Asum D12C (i)X 12C + ∑ Asum D12C (i − n)X (n) (2.2) n=1

12 D12C(i): The fraction of ions with mass number i in a pure C spectrum 12 X12C: The fraction of ions containing C atoms only X(n): The fraction of ions containing n 13C atoms Other symbols as before

Equation 2.2 expresses the observed single ion peak areas as linear combinations of the fractions of 13C atoms in the molecules and a set of linear equations may be formulated for a given species and solved using a multivariable linear regression analysis implemented in a specialized Excel spreadsheet. This will yield the fractions of ions containing a specific number of 13C atoms as the results. Once these fractions are known, the total content of 13C atoms in a given compound may be calculated. The reliability of the calculated isotopic distribution is expressed by the correlation coefficient from the linear regression and by a root mean square parameter, given as the squares of differences between observed and calculated single ion peak areas. The root mean square parameter is very sensitive and provides a good indication of the degree of success of the model fitting. Generally, the reported isotopic distributions are trustworthy, and any errors arising from the calculation procedure are probably overshadowed by other uncertainties in the experimental setup.

23 3. Computational methods

A considerable part of this thesis involves the use of quantum chemical methods to study individual zeolite catalyzed reaction steps. Whenever modeling of heterogeneous catalysis is attempted, a central issue is how to describe the catalyst. Quite often the catalyst is amorphous and ill defined, making this model selection a formidable task, but this is fortunately not the case for zeolite catalysts, which are crystalline materials of known structure. Even so, it is very rarely (never) possible to design a model of the catalyst that includes all its characteristic properties and still is applicable to calculations. In the current work, a well-proven 19-atom fragment has been adopted as the model of the infinite lattice of a zeolite catalyst [114-123]. The following section will address the applicability and the limitations of the model used.

3.1. The cluster model

In the present context, a cluster model is a molecular fragment describing a part of a much larger system. The cluster used in the quantum chemical modeling performed as part of this thesis consists of four tetrahedrally coordinated atoms (4 T-atoms), three silicons and one aluminum to generate the acidic site. Figure 3.1 shows how the cluster can be considered to be extracted from a larger part of the catalyst framework. The severed bonds are saturated by hydrogen atoms.

abAcidic proton

Si Si Si

8-ring pore mouth cluster model

Figure 3.1. Schematic illustration of the selection of the cluster (a), adapted from [116]. Optimized geometry of the cluster model at the B3LYP/6-31G(d) level of theory (b).

24 The cluster depicted in Figure 3.1b has been used in papers I-III, and all the calculations have been carried out using the same level of theory for geometry optimizations. The B3LYP hybrid density theory functional [124,125] was combined with the 6-31G(d) basis set [126], which is a double-ζ basis set with polarization functions on the heavy atoms (all atoms except hydrogen and helium). LANL2DZdp [127,128] effective core potentials (ECPs) were used on the halogen atoms in paper I. These ECPs are the counterpart to the 6- 31G(d) basis set. All stationary points were investigated by performing analytic frequency calculations and ensuring that the correct number of imaginary frequencies (negative eigenvalues in the Hessian) was at hand, i.e. zero imaginary frequencies for energy minima and one imaginary frequency for transition states. All atoms have been allowed to relax freely during optimizations. Single point energy calculations have been carried out using different basis sets and/or the MP2 computational method. Although considerable shifts in energies were found, the trends in energy did not depend on the level of theory. The Gaussian 98 suite of programs has been used for all calculations [129]. The obvious incentive for using a fairly small cluster to model the zeolite catalyst is to minimize the computational cost. Even though modern supercomputers are fast, they still leave a lot be desired when it comes to brute computing force. The 4 T-atom cluster represents a compromise between the accuracy of the computational scheme, and hence time consumption, and the size of the cluster. An estimated 40,000 CPU-hours have been used on the various Norwegian supercomputing facilities during this work. It is equally obvious that the cluster leads to an incomplete modeling of two key properties of a ‘real’ zeolite catalyst. First, steric effects are poorly described, as arbitrarily large molecules or assemblages of molecules may be adsorbed onto the cluster. Second, the electrostatic field that is present inside the zeolite pores is not properly included. However, it is possible to perform qualitative explorations of general reaction mechanisms of reactions that occur locally on one single active site. It is also possible to make qualitative comparisons of the energy profiles for similar reactions, such as comparing the barrier for the methylation of ethene with that for the methylation of propene. If the reactions are too different, energy comparisons might not be valid. Moreover, the fairly small size of the model system is an advantage when it comes to selecting the level of sophistication of the computational scheme to be used. Analytic force constants are readily available for the 4 T-atom cluster during optimizations, which is the single most helpful tool for locating transition states and ensuring geometry convergence. Also, calculating the analytic force constants eventually allows an

25 analysis of the vibrational frequencies at the stationary points, which is required to classify the stationary points found as minima, transition states or higher order saddle points. Two other approaches to quantum chemical modeling of zeolite catalyzed reactions have provided substantial insight to the applicability of cluster models. With quantum chemical-molecular mechanics (QM/MM, or more generally QM-Pot), only the active site is treated with quantum mechanics, whereas the surrounding parts of the catalyst are modeled using classical force fields [116,130-136]. Periodic methods are well suited for zeolite catalysis because a repeating unit (typically one or two unit cells) is subjected to calculation and no arbitrary cut-offs are necessary. This approach usually relies on DFT combined with plane wave basis sets [117,118,137-143]. Both techniques provide a better description of the steric effects and the electrostatic field neglected by the cluster approach. Several studies provide data for comparing results obtained with QM/MM or periodic models with results from cluster calculations [116-118,134], and the following observations have been made. 1) The basic molecular reaction mechanisms are the same. 2) The energies of charged species and transition states, which are often carbenium ion like, are overestimated, often by as much as 20 – 50% by the cluster model [118]. This is mainly due to the omission of the electrostatic field. In a study of the methylation of toluene with methanol, Vos et al. found that the barrier decreased with 40 kJ/mol when going from a cluster model to a periodic model [118]. Other, similar cases have also been reported [141,142]. 3) However, in the absence of steric limitations, i.e. if the reacting species fit well inside the pore system under investigation, these shifts in energy relative to the neutral reactants and products are uniform [117,142]. This means that qualitative comparisons of cluster approach barriers are valid. An inherent limitation to DFT computational methods is the poor description of dispersion forces [144]. This inevitably leads to an underestimation of the strength of the interaction between adsorbate and adsorbent, and this effect is evident in papers I-III, where all the adsorption energies are unrealistically low. The single point energy calculations carried out with the MP2 method gave more realistic results in this respect. Finally, the choice of allowing all atoms to relax fully during geometry optimizations should be commented on. It is not uncommon to place constraints on the position of some atoms, angles or dihedrals of the cluster to simulate the rigidity of the zeolite framework, as it is clear that an unconstrained cluster allows the silicon atoms to move more than what is strictly possible in a crystal structure. This also helps to prevent the cluster from collapsing or folding up and the reactants from moving towards the edges or even to the underside of the cluster. The 4 T-atom cluster, however, is relatively well behaved in this respect, and thus

26 constraints are not necessary in order to get intuitively reasonable stationary points. Also, applying constraints will lead to only partially defined states, because of bogus negative eigenvalues. Frequency calculations on partially relaxed (non-stationary) systems also lead to errors in the zero-point energy corrections [145].

27 4. This work

4.1. Scope

The primary objective of the present work is to obtain new insight into the reaction mechanism of the zeolite catalyzed methanol-to-hydrocarbons (MTH) reaction. It was decided to use both experimental and computational techniques to reach this goal. In a series of previous publications, Kolboe and co-workers had demonstrated the very successful use of isotopic labeling to obtain mechanistic data concerning the MTH reaction. In particular, the coreaction of ordinary ethene or propene with [13C]methanol had been investigated over both SAPO-34 and H-ZSM-5 catalysts [59,60,62]. An investigation of the n-butene + methanol system was therefore initiated. Over time, it became apparent that it was possible to determine the rate for the methylation of n-butene by methanol. The ethene and propene systems were therefore reexamined in order to collect kinetic information also for those cases. With the development of user-friendly quantum chemistry programs such as the Gaussian suite of programs, the possibility of applying quantum chemical methods to many types of problems has become readily available even for non-experts. When performing mechanistic studies, there is quite often a considerable synergy effect when combining experimental and computational approaches. The methylation reactions mentioned above turned out to be an issue well suited for quantum chemical investigations. The incentive for examining the halomethane reactivity was the clear analogy to the MTH reaction system. Alkene dimerization was also a reaction readily examined with quantum chemistry. As discussed in the introduction of this thesis, polymethylbenzenes, or their cationic counterparts, are suspected to be key intermediates in the MTH reaction. It was therefore decided to investigate the intrinsic reactivity of these species in the gas-phase by employing sophisticated mass spectrometric (MS) techniques in collaboration with the MS group at the Department of Chemistry, University of Oslo. The data thus obtained will also be compared with results from an ongoing computational study on gas phase polymethylbenzenium reactivity [146].

28 4.2. Synopsis of results

Paper 1: A Theoretical Investigation of the Methylation of Methylbenzenes and Alkenes by Halomethanes over Acidic Zeolites.

Paper I is the first theoretical study related to the mechanism of halomethane conversion over acidic zeolites. The work was initially undertaken as an extension of the investigation of the methylation of alkenes by methanol presented in paper II. The scope of the study was manifold. We wanted to identify any trends in reactivity among the three halomethanes, chloromethane, bromomethane, and iodomethane; to obtain a fundamental insight into the reactivity of the halomethanes; to distinguish between a stepwise and a concerted reaction mechanism; and to compare the results with results from investigations on methanol reactivity. Adsorbed reactants, intermediates, transition states, and products were optimized for the methylation of ethene, propene, and toluene by all three halomethanes via both a stepwise and a concerted reaction mechanism; see Figure 4.1. The stepwise mechanism proceeds via surface bound methoxide groups as intermediates. The calculated activation energies are given in Table 4.1 for concerted methylation and Table 4.2 for stepwise methylation.

CH3X + H H

O O O O Si Al Si Si Al Si

+ HX

CH3X

H CH3 H

O O O O O O Si Al Si Si Al Si Si Al Si

Figure 4.1. Concerted (top) and stepwise (bottom) methylation of toluene.

29 A comparison of the barrier heights Table 4.1. Calculated activation energies for concerted methylation. B3LYP/6-31G(d) involved indicates that the concerted + ZPE level of theory. mechanism will be dominating; the two barriers of the two-step mechanism are in all Reactant Barrier for concerted cases investigated significantly higher than the methylation single barrier of the concerted route. ethene + chloromethane 123 A weak trend in reactivity was found bromomethane 118 among the halomethanes. The order of iodomethane 117 propene + reactivity appears to follow the expected chloromethane 108 leaving group trend: MeI > MeBr > MeCl. bromomethane 101 iodomethane 97 The differences were, however, quite small toluene + and some minor discrepancies were found. chloromethane 128 bromomethane 120 The calculated trend does conform to the iodomethane 115 expected result, but several properties of the halomethanes, such as the polarizability and Table 4.2. Calculated activation energies the dipole moment, are different and a for stepwise methylation. B3LYP/6-31G(d) uniform shift in barrier heights might not be + ZPE level of theory. expected if the study were extended to more Reactant Barrier for Barrier for realistic models. Hence, the data do not allow methoxide stepwise formation C-C bond a definitive conclusion regarding the order of formation reactivity among the halomethanes. chloromethane 166 bromomethane 155 Additionally, the calculated activation iodomethane 149 energies indicate that propene is more easily ethene 169 propene 156 methylated than ethene, whereas toluene is toluene 171 about as reactive or slightly less reactive than ethene. The reactions investigated are initiated by the interaction of the halogen atom of the halomethane molecule with the acidic proton of the cluster model. Figure 4.2 displays the transition state for the formation of a methoxide group from chloromethane. The chlorine- carbon bond is severely stretched as a formal methyl cation approaches the zeolite oxygen atom. In contrast to what is observed when methanol is the reactant [114,121], there is no complete protonation in the transition state. This holds also for the concerted methylations and for all three halomethanes.

30 The barriers calculated for these reactions were considerably lower than those found for the analogous reactions involving methanol [114]. However, it may not be concluded that the halomethanes are intrinsically more reactive than methanol without some reservations. As already mentioned, there is no complete proton transfer in the halomethane reactions, and this Figure 4.2. Transition state for the formation of a surface subtle difference might render a bound methoxide and HCl from chloromethane, i.e., the first step of the two-step mechanism. Geometry comparison between the two optimization performed at the B3LYP/6-31G(d) level of cases invalid. theory.

31 Paper II: A Theoretical Investigation of the Methylation of Alkenes with Methanol over Acidic Zeolites.

Paper II is parallel to a report by Arstad et al. on the methylation of a series of methylbenzenes [114]. Here the objective was to probe for trends in reactivity in a series of seven alkenes and to obtain data for comparison with the results concerning the methylation of polymethylbenzenes. Only the concerted mechanism of methylation was investigated. The calculated barrier heights are listed in Table 4.3, and the results show that the activation energies for the methylation reactions decrease as the alkenes become larger and the double bond more substituted.

Table 4.3. Calculated barrier heights for the investigated methylation reactions. B3LYP/6- 31G(d) + ZPE level of theory.

Object of methylation Investigated products Barrier height (kJ/mol) ethene propene, 183 cyclopropane propene trans-2-butene 169 1-butene trans-2-pentene 168 trans-2-butene 2-methyl-2-butene 162 cis-2-butene 2-methyl-2-butene 161 iso-butene 2-methyl-2-butene 156 2-methyl-2-butene 2,3-dimethyl-2-butene 154

Somewhat surprisingly, energy minima describing cationic alkene products and a water molecule were found for five of the seven alkenes investigated. The exceptions were methylation of ethene and isobutene. The decrease in the activation energies with the size of the alkenes could be well rationalized by considering the relative stability of these carbenium ion intermediates. In one instance, a barrier for the deprotonation reaction was located. As expected, this barrier is very low, about 6 kJ/mol. The deprotonation energy is 152 kJ/mol (see Figure 4.3). Hence, such charged aliphatic species will not have appreciable lifetime under experimental conditions. Nonetheless, these stationary points are quite descriptive for the reaction pathway of the methylation reactions. The complete energy profile for the methylation of t-2-butene is displayed in Figure 4.3. Close inspection of the geometry of the transition state reveals that the formal methyl cation appears to approach the center of the alkene double bond rather than becoming attached to one single carbon atom. The prominence of this feature, a nearly triangular arrangement in the transition state, did vary with the double bond substitution pattern, but it

32 Energy

+163 kJ/mol

Transition state +106 kJ/mol

Protonated products

0 kJ/mol

Adsorbed reactants - 46 kJ/mol

Adsorbed products

Reaction coordinate

Figure 4.3. Stationary points on the reaction path for the methylation of t-2-butene. was found for all the alkenes that were investigated. When aromatic species are methylated by methanol, no such three-ring arrangement is found, as the formal methyl cation clearly approaches one of the ring carbons [114]. The activation energies reported in Table 4.1 are in the very same range as those previously reported for the methylation of polymethylbenzenes [114], indicating a similar reactivity towards methanol for the two classes of compounds. The effect of the cluster 175 B3LYP/6-31G* B3LYP/6-311G**//B3LYP/6-31G* acidity was investigated by l) MP2/6-31G*//B3LYP/6-31G* o B3LYP/6-31G* + ZPE 170 modifying the composition of the -OH J/m k cluster model. By replacing the ( 165 gy –OH group bonded to the -F aluminum atom with –F or a –CF3 160 group, the proton affinity of the -CF 155 3 anionic cluster was tuned over a Activation ener range of 50 kJ/mol. The 150 1200 1220 1240 1260 1280 1300 1320 methylation of propene was Cluster proton affinity (kJ/mol) selected as a test reaction, and the Figure 4.4. The calculated activation energy for the barrier was found to vary with ~ 15 methylation of propene depends on the proton affinity kJ/mol; see Figure 4.4. of the anionic cluster model.

33 Paper III: Theoretical Investigation of the Dimerization of Linear Alkenes Catalyzed by Acidic Zeolites.

During the experimental work that resulted in papers IV and V it became apparent that alkene interconversion reactions, such as dimerization, isomerization, and cracking, occurred in competition with the methylation of alkenes under certain reaction conditions. It was therefore natural to extend the theoretical studies of the alkene + methanol systems to also include the alkene + alkene reactions. The approach selected was similar to that chosen for the halomethane + alkene investigation described in paper I. The dimerization of a series of alkenes, ethene, propene, 1-butene, and t-2-butene was submitted to calculation, and both a stepwise and a concerted mechanism were studied. In the concerted pathway, protonation and C-C bond formation occur simultaneously, whereas the stepwise mechanism proceeds via alkoxide formation followed by C-C bond formation. The transition states for the steps in which the C-C bond is formed are shown for the two mechanisms in Figure 4.5.

a) b)

Figure 4.5. Transition states for the steps in which the new C-C bond is formed via the concerted (a) and the stepwise (b) mechanisms for ethene dimerization.

In the transition state for concerted dimerization, the acidic proton is partially transferred from a zeolite oxygen atom to one of the carbon atoms of the alkene originally coordinated to the acidic site. Simultaneously, the other carbon of the double bond is attacked by the π-electrons of the second alkene, resulting in the formation of the new C-C bond. For the unsymmetrically substituted alkenes, there are two possible sites of protonation, leading to formation of either a formally primary or secondary carbenium ion in the transition state. Both options were explored, whereas only the intuitively most stable possibilities were explored for the stepwise mechanism. The calculated barrier heights are listed in Table 4.4.

34 Table 4.4. Calculated activation energies for dimerization of alkenes. B3LYP/6-31G(d) + ZPE level of theory.

Reactant Barrier for concerted dimerization Barriers for stepwise dimerization primary ion-like secondary ion-like barrier 1 barrier 2 transition state transition state alkoxide C-C bond formation formation ethene 127 97 167 propene 122 97 82 143 1-butene 120 102 79 137 t-2-butene 114 96 139

The lowest barriers are found for the alkoxide formation, and the barriers for concerted dimerization are always lower than the second barrier of the stepwise route. Surprisingly, a counter-intuitive trend is found among the reactants for concerted dimerization via transition states that resemble secondary carbenium ions. A full description of the energy profile for ethene dimerization is shown in Figure 4.6. Strikingly, the highest lying point on the potential energy surface is found for concerted dimerization. It appears that the main cause for the high barrier of the second step of the stepwise mechanism is the exothermicity of the preceding alkoxide formation rather than an intrinsic instability of that transition state.

TS for concerted TS for TS for dimerization alkoxide + alkoxide alkene formation + 91 + 76 + 66

alkene (g) alkene (g) HZeo + 97 + 127 + 167 alkene (ads) 0 kJ/mol alkene (g) HZeo -31 -35 -57 alkene (ads) alkene (ads) alkene (g) HZeo alkoxideZeo -92 Adsorbed -88 alkene alkene (ads) product BLUE: Stepwise alkoxideZeo - 140 RED: Concerted Alkoxide product - 174

Figure 4.6. Energy profile for ethene dimerization. B3LYP/6-31G(d) + ZPE level of theory.

35 Paper IV: Kinetic Studies of Zeolite-catalyzed Methylation Reactions. 1. Coreaction of [12C]ethene and [13C]methanol.

Paper IV is an experimental investigation of the reaction kinetics of the methylation of ethene by methanol. A main motivation for performing the experiments discussed in Paper IV was to obtain experimental data for comparison with the computational description of the alkene methylation reactions outlined in Paper II. A sizable series of experiments were performed prior to those discussed in Paper IV to find the reaction conditions best suited for investigating the methylation reaction in the absence of complicating side reactions. Both methanol and ethene are converted into other products when fed alone to an H-ZSM-5 catalyst. By operating with moderate reactant partial pressures (10 – 100 mbar), at a reaction temperature around 350 °C, and at extremely high feed rates (WHSV ~ 200 h-1) and thereby low conversions, the side reactions were suppressed to insignificance. In order to achieve the required high feed rates, but still operating at a realistic reaction temperature and catalyst acid site density, merely 2.5 mg of a commercial H- ZSM-5 catalyst, generously supplied by Süd Chemie, was used. Despite having only a thin layer of catalyst, reactant by-pass was negligible. This was verified by reacting 2-butanol over the catalyst at the highest feed rates employed. An essentially complete dehydration of the alcohol to butenes was observed. Blank experiments gave insignificant alcohol dehydration. Figures 4.7 and 4.8 show the product distribution and the isotopic composition of propene, the methylation product, as a function of varying the feed rate in the range 29 – 294 h-1. Quite clearly, propene is the dominating coreaction product and the isotopic composition is in accord with what expected when 100

) propene

propene is formed by methylation of % alkanes

C 80 n-butenes ( ethene. By altering the partial isobutene C 5

ities C pressures of both reactants, the 60 6+ iv t reaction orders for the methylation 40 of ethene to form propene were 20

found to be approximately one with oduct selec r P respect to methanol and zero with 0 0.00 0.01 0.02 0.03 0.04 respect to ethene. Extrapolation of 1/WHSV = CT (h) the reaction rate to CT = 0 gave the Figure 4.7. Product selectivities. 50 mbar ethene coreacted with 50 mbar methanol; reaction -4 apparent rate constant, k = 2.6 × 10 temperature = 350 °C; WHSV varied from 29.4-294 1

36 mol/(g h mbar) and the pre- 100 5 exponential, A = 3.5 × 10 mol/(g h 80 on (%) i

t No. of mbar). An Arrhenius plot based on u 13

b C-atoms i 60 r t zero the rate of formation of the mono- one dis

r 40 two labeled propene isomer was three 20 constructed, and an apparent opome ot s activation barrier of 109 kJ/mol was I 0 0.0034 0.0084 0.016 0.034 found. The reaction was considered 1/WHSV = CT (h) to be a concerted process; the Figure 4.8. Isotopic composition of propene in the methylation proceeds upon effluent. Reaction conditions as in Figure 4.7. interaction of a methanol molecule physisorbed onto an acidic site with an ethene molecule located next to it on a siliceous part of the zeolite catalyst. Hence, the adsorption enthalpy for ethene in silicalite, 25 kJ/mol [147] was used as a correction in order to obtain an estimate for the true barrier. This approach gave an intrinsic barrier height of about 135 kJ/mol. Multiple methylation reactions were also observed. The most abundant linear butene isotopomer contained two labeled carbons at short contact times. Up to quadruple methylation 12 13 could be discerned, leading to formation of C2 C4 hexene isotopomers. The extrapolated selectivities towards alkene products larger than propene and the corresponding isotopic distributions indicated that sequential methylations might occur before desorption and complete migration of the intermediate products out of the zeolite crystals and into the bulk gas phase. The coreaction results also indicated that, in addition to methylation, another mechanism is operative, leading to products very rich in 13C. It was speculated that a small fraction of each alkene homologue can be formed via the hydrocarbon pool mechanism, giving rise to alkenes with high contents of labeled carbon atoms. The isotopic distribution shown in Figure 4.8 would then be composed of two superimposed distributions, one caused by methylation and the other being the result of propene formation through the hydrocarbon pool mechanism. Small amounts of aromatics, which have been shown to function as pool species on other catalysts [64,65], were always detected, and these compounds typically contained about 85% 13C, fairly randomly distributed within the molecules. Under ordinary MTH reaction conditions, formation of higher alkenes via homologation, starting with ethene, is not an important reaction.

37 Paper V: Kinetic Studies of Zeolite-catalyzed Methylation Reactions. 2. Coreaction of [12C]propene or [12C]n-butene and [13C]methanol.

Paper V is a preliminary manuscript. The data described in Paper V is the natural extension of the work presented in Paper IV. In contrast to what was found in the ethene + methanol coreaction system, both propene and the linear butenes (lumped as n-butene) display significant reactivity in the absence of methanol co-feed, even at the very high feed rates employed. Therefore, a brief examination of the reactivity of propene and n-butene alone was conducted prior to the coreaction studies. The degree of propene conversion and the observed product selectivities are shown as functions of temperature in Figure 4.9. The most striking feature is the clear decrease in the conversion when the reaction temperature is increased. This might be attributed to a decrease in the surface coverage at elevated temperatures. The conversion of propene requires the interaction of (at least) two molecules and such bimolecular events will be disfavored by decreases in the concentration of adsorbed species. The conversion was, however, insignificant below ca. 200 °C. Isobutene and n-butene are dominating among the products formed.

60 14

50 12 ) 10 Conv % 40 Conversion

ersion (%) C s (C 8 2 e 30 n-butene iviti

t 6 isobutene C 20 5 4 Selec C 6+ 10 2

0 0 250 275 300 325 350 375 400 425

Reaction temperature (°C)

Figure 4.9. Observerd product selectivities and level of conversion. 20 mbar of propene reacted alone; contact time = WHSV-1 = 0.012 h; reaction temperature varied from 250 – 425 °C.

Figure 4.10 displays the degree of n-butene conversion and the product yields as a function of reaction temperature. The yields of products other than isobutene decrease when the temperature is raised, whereas isobutene displays the opposite behavior. This causes the

38 15 20

12 16 Co )

nv Conversion % 9 12 ersion

C C 3 (

s isobutene d l

(%) C 6 8 5 Yie C 6+ 3 4

0 0 300 350 400 450 500 Reaction temperature (°C)

Figure 4.10. Observerd product yields and level of conversion. 13 mbar of n-butene reacted alone; contact time = WHSV-1 = 0.019 h; reaction temperature varied from 290 – 510 °C. conversion to pass through a minimum at about 350 °C. Again, the degree of conversion was insignificant below 200 °C.

The data from the coreaction experiments were quite analogous to those discussed for the coreaction of ethene and methanol in Paper IV. The expected methylation products were always dominating (i.e. n-butene from propene and pentenes from n-butene), and the isotopic data revealed that methylation was indeed the main mode of product formation. Bearing in mind that the alkenes display significant reactivity when fed to the catalyst alone at these conditions, it appears that the presence of methanol strongly suppresses the alkene interconversion reactions. However, the higher reactivity of propene and n-butene led to a slightly more complicated picture. The first and zero order behavior with respect to the alkene pressure and the methanol pressure, respectively, were only partially retained. The alkenes are able to compete for acidic sites, as manifested by a substantial share of hexenes being formed by propene dimerization rather than triple methylation when methanol and propene were coreacted. Also, the conversions in these coreaction systems were higher than when ethene was the coreactant. In the ethene study (Paper IV) the conversion of the feed mixture was about 0.3% at the highest feed rate investigated. The corresponding values were 3% and 9% when propene and n-butene were coreacted with methanol, i.e. a 10 and 30 fold increase in conversion. Also in this case the major goal was to obtain kinetic data for the methylation reactions. Figures 4.11 a) and b) show the extrapolation of the rates of conversion to CT = 0 at 350 °C. By multiplying with the fractional limiting isotopic and product selectivities, dividing

39 30 10 a) b) 25

n n on on i i o o 8 i i s s s s r r 20 ) ) ) ) e e -1 -1 -1 -1 v v n 6 n nver nver *h *h *h *h o o -1 -1 15 o o -1 -1 c c g g g g c c * * (g* (g* 4 of of of of (g (g 10 e e t t a a R R Rate 2 Rate 5

0 0 0.00 0.01 0.02 0.03 0.04 0.00 0.01 0.02 0.03 0.04 0.05 1/WHSV = CT (h) 1/WHSV = CT (h)

Figure 4.11. Rates of conversion. a) 20 mbar of propene coreacted with 50 mbar methanol; reaction temperature 350 °C; total gas flow varied from 10 to 100 mL/min. b) 13 mbar of n- butene coreacted with 50 mbar methanol; reactions temperature 350 °C; total gas flow varied from 10 to 100 mL/min. with the alkene partial pressures, and converting to molar units the following apparent rate -3 -2 constants were found at 350 °C: kpropene = 4.5 × 10 mol/(g h mbar) and kn-butene = 1.3 × 10 mol/(g h mbar). Hence, the following ratios are obtained for the apparent rate constants: kethene

: kpropene : kn-butene = 1 : 17 : 50. The reaction temperature was varied in the ranges 290 – 410 °C and 350 – 450 °C for the methylation of propene and n-butene, respectively. The product distributions did not depend strongly on the reaction temperature. Arrhenius plots based on the rates of formation of the mono-labeled methylation products were constructed and the apparent activation energies were found to be 109 kJ/mol, 69 kJ/mol, and 54 kJ/mol for the methylation of ethene, propene, and n-butene, respectively. These values may be corrected with the appropriate enthalpies of alkene adsorption in silicalite, which are reported to be 25 kJ/mol, 39 kJ/mol, and 45 kJ/mol, in the same order as before [147,148]. This yields the following estimates for the intrinsic activation energies: ~ 135 kJ/mol, ~ 110 kJ/mol, and ~ 100 kJ/mol. For comparison, the activation energy for the conversion of methanol alone has been found to be 135 kJ/mol, based on data obtained in the range 180 – 250 °C [52]. The internal consistency of these data is not perfect. However, it should be kept in mind that the method employed for assessing the activation energies is not strictly rigorous and this will limit the quality of the data obtained. Indeed, the Arrhenius plots displayed some deviation from linearity, and the assumption of first and zero order behavior with respect to the alkene and methanol pressures over the entire temperature ranges might not be strictly valid. Better data might be obtained by investigating the effects of reactant partial pressures at several different reaction temperatures.

40 Paper VI: The Intermediates in the Methanol-to-Hydrocarbons (MTH) Reaction: A Gas Phase Study of The Reactivity of Polymethylbenzenium Cations.

Paper VI is a preliminary manuscript. As discussed in the Introduction part of this thesis, methylbenzenes and their cationic counterparts have been identified as key intermediates in the methanol-to-hydrocarbons reaction. The paring mechanism hypothesizes that monomolecular alkene loss from polymethylbenzenium cations is an important reaction in the MTH reaction system. It was therefore decided to investigate the intrinsic reactivity of polymethylbenzenium cations in an environment free from the surrounding catalyst framework, i.e. in the gas-phase. This was achieved by employing a specialized mass spectrometer. In contrast to a ‘regular’ mass spectrometer (such as the MS-detector found in a GC-MS instrument) the equipment used in this study permits us to study the fragmentation of even-electron cations of a given single mass produced via either exclusive protonation + + + (CH5 /C2H3 as proton donor) or methylation (CH3ClCH3 as methyl cation donor). The species submitted to mass spectrometric analysis are shown in Figure 4.12. All compounds were protonated and methylated, thereby producing a wide range of ions. These ions have a rich and complicated chemistry, and extensive fragmentation is observed. Loss of ethene, propene, and butene was observed throughout the series of investigated compounds, and the feasibility of the paring mechanism has thus been demonstrated. However, loss of dihydrogen, methyl radicals, or methane is always much more prominent for the polymethylbenzenium ions, a feature incompatible with efficient alkene production. A comparison between mass spectrometric data and results from catalytic studies is not straightforward, because the ions produced in the mass spectrometer are far more energetic.

0 1 2a 2b 3

4 5 6 7

Figure 4.12. Species investigated by mass spectrometric techniques in paper VI.

41 4.3. Main conclusions

• Computational results indicate that methylation of alkenes and methylbenzenes with halomethanes proceeds via a concerted, and not a stepwise, mechanism. • A weak trend in reactivity was found among the halomethanes, indicating the expected order of reactivity: iodomethane > bromomethane > chloromethane. • The calculated activation energies for methylation with halomethanes are considerably lower than the corresponding values for methylation with methanol, but this might be caused by shortcomings in the model used.

• The calculated activation energies for methylation of alkenes with methanol decrease with the size and complexity of the alkenes. These barriers are in the very same range as those calculated for the methylation of polymethylbenzenes. • Very shallow energy minima describing adsorbed carbenium ions and water were in most cases found after the transition state for methylation of alkenes with methanol. The barrier heights mentioned above could be well rationalized by considering the relative stabilities of these species. • The activation energy for propene methylation by methanol varies with ~ 15 kJ/mol when the cluster acidity (anionic cluster proton affinity) is tuned over a range of 50 kJ/mol.

• Calculations have shown that alkene dimerization may proceed via both a concerted and a stepwise reaction mechanism. Attempts to discriminate between the two mechanisms were inconclusive. • The highest barriers were found for the dimerization of ethene. Propene, 1-butene, t-2- butene are more reactive; the calculated activation energies for dimerization are 13 – 30 kJ/mol lower than those found for ethene.

• Isotopically labeled methanol has been coreacted with ordinary ethene, propene, or linear butenes (n-butene) over an H-ZSM-5 acidic catalyst. At high feed rates, the methylation product is dominating, as evidenced by the composition of the effluent and the isotopic distributions.

42 • The methylation reactions display approximately first order behavior with respect to the alkene partial pressure and zero order behavior with respect to the methanol partial pressure over the investigated ranges. • Under ordinary MTH reaction conditions, formation of higher alkenes via homologation, starting with ethene, is not an important reaction. However, propene and n-butene are more reactive. • Experimental kinetic data have been obtained for the methylation of ethene, propene, and linear butenes (n-butene) by methanol over an H-ZSM-5 acidic catalyst. The ratios

between the apparent rate constants at 350 °C are: kethene : kpropene : kn-butene = 1 : 17 : 50. • Arrhenius plots based on the observed rate of methylation have been constructed and the apparent activation barriers were estimated to be 109 kJ/mol, 69 kJ/mol, and 54 kJ/mol for the methylation of ethene, propene, and n-butene, respectively. When corrected with the appropriate enthalpies of alkene adsorption the following estimates are found for the intrinsic activation energies: ~ 135 kJ/mol, ~ 110 kJ/mol, and ~ 100 kJ/mol.

• The gas phase reactivity of a wide range of polymethylbenzenium cations have been investigated using advanced mass spectrometric techniques. • Alkene loss is observed for all species, thus demonstrating the feasibility of the paring mechanism. However, loss of dihydrogen, methyl radicals, or methane is always much more prominent for the polymethylbenzenium ions, a feature incompatible with efficient alkene production. • The gas phase data provided limited insight to MTH chemistry, because the ions produced in the mass spectrometer are far more energetic than the species present in the catalyst pores.

43 5. References

[1] G.C. Bond, in: P.W. Atkins, J.S.E. Holker, A.K. Holliday (Eds.), Heterogeneous Catalysis: Principles and Applications, Oxford Chem. Ser. 34, Clarendon Press, Oxford, 1987, p. 1. [2] F.A. Mumpton, in: L.B. Sand, F.A. Mumpton (Eds.), Natural Zeolites. Occurrence, Properties, Use, Pergamon Press, Oxford, 1978, p. 3. [3] F. Fontana, Memorie Mat. Fis. Soc. Ital. Sci. 1 (1777) 679. [4] A. Damour, Ann. Mines 17 (1840) 191. [5] H. Eichhorn, Poggendorf Ann. Phys. Chem. 105 (1858) 126. [6] O. Weigel, E. Steinhoff, Z. Kristallogr. 61 (1925) 125. [7] J.W. McBain, The Sorption of Gases and Vapors by Solids, Rutledge & Sons, London, 1932. [8] L. Pauling, Proc. Nat. Acad. Sci. 16 (1930) 453. [9] L. Pauling, Z. Kristallogr. 17 (1930) 213. [10] W.H. Taylor, Z. Kristallogr. 17 (1930) 1. [11] R.M. Barrer, J. Soc. Chem. Ind. 64 (1945) 130. [12] R.M. Barrer, J. Chem. Soc. (1948) 2158. [13] E.M. Flanigen, in: H. van Bekkum, E.M. Flanigen, P.A. Jacobs, J.C. Jansen (Eds.), Introduction to Zeolite Science and Practice, Stud. Surf. Sci. Catal. 137, Elsevier, Amsterdam, 2001, p. 11. [14] A. Corma, J. Catal. 216 (2003) 298. [15] D.W. Breck in: Zeolite Molecular Sieves. Structure, Chemistry, and Use, Robert E. Krieger Publishing Company, Malabar, USA, 1984, p. 47. [16] A. Dyer, in: An Introduction to Zeolite Molecular Sieves, John Wiley & Sons, New York, 1988, p. 1. [17] M. Guisnet, J.-P. Wilson in: M. Guisnet, J.-P. Wilson (Eds.), Zeolites for Cleaner Technologies, Catal. Sci. Ser. 3, Imperial College Press, London, 2002, p. 1. [18] L.A. Clark, M. Sierka, J. Sauer, J. Am. Chem. Soc. 126 (2004) 936. [19] R.J. Argauer, G.R. Landolt, US Patent 3 703 886 (1972), to Mobil Oil Corporation. [20] G.T. Kokotailo, S.L. Lawton, D.H. Olson, Nature 272 (1978) 437. [21] D.H. Olson, G.T. Kokotailo, S.L. Lawton, J. Phys. Chem. 85 (1981) 2238.

44 [22] C. Baerlocher, W.M. Meier, D.H. Olson, in: Atlas of Zeolite Framework Types, Elsevier, Amsterdam, 2001, p. 184. [23] J. Houzvicka, J.G. Nienhuis, S. Hansildaar, V. Ponec, Appl. Catal. A 165 (1997) 443. [24] M.E. Davis, S.I. Zones in: M.L. Occelli, H. Kessler (Eds.), Synthesis of Porous Materials. Zeolites, Clays, and Nanostructures, Chem. Ind. Ser. 69, Marcel Dekker, New York, 1996, p. 1. [25] M. Stöcker, Microporous Mesoporous Mater. 29 (1999) 1. [26] C.D. Chang, Catal. Rev. Sci. Eng. 25 (1983) 1. [27] C. Baerlocher, W.M. Meier, D.H. Olson, in: Atlas of Zeolite Framework Types, Elsevier, Amsterdam, 2001, p. 96. [28] http://www.chemicals-technology.com/projects/tjeldbergoddenethpro (accessed July 2004). [29] J. Cobb, in: G. Connell (Ed.), New Zealand Synfuel: The Story of the World’s First Natural Gas to Gasoline Plant, Cobb/Horwood Publications, Auckland, New Zealand, 1995. [30] http://www.chemicals-technology.com/projects/eurochem (accessed July 2004). [31] C.D. Chang in: C. Song, J.M. Garcés, Y. Sugi (Eds.), Shape Selective Catalysis: Chemical Synthesis and Hydrocarbon Processing, Am. Chem. Soc. Symp. Ser. 738, American Chemical Society, Washington DC, 2000, p. 96. [32] P. Barger, in: M. Guisnet, J.-P. Wilson (Eds.), Zeolites for Cleaner Technologies, Catal. Sci. Ser. 3, Imperial College Press, London, 2002, p. 239. [33] J.F. Haw, D.M. Marcus, in: S.M. Auerbach, K.A. Carrado, P.K. Dutta (Eds.), Handbook of Zeolite Science and Technology, Marcel Dekker, New York, 2003, p. 833. [34] J.P. van den Berg, J.P. Wolthuizen, J.H.C. van Hooff, in: L.V. Rees (Ed.), Proceedings 5th International Zeolite Conference (Naples), Heyden, London, 1980, p. 649. [35] G.A. Olah, Pure Appl. Chem. 53 (1981) 201. [36] G.J. Hutchings, F. Gottschalk, M.V. Michéle, R. Hunter, J. Chem. Soc., Faraday Trans. 83 (1987) 571. [37] Y. Ono, T. Mori, J. Chem. Soc., Faraday Trans. 77 (1981) 2209. [38] D. Kagi, J. Catal. 69 (1981) 242. [39] C.D. Chang, A.J. Silvestri, J. Catal. 47 (1977) 249. [40] C.D. Chang, J. Catal. 69 (1981) 244. [41] E.A. Swabb, B.C. Gates, Ind. Eng. Chem. Fund. 11 (1972) 540.

45 [42] W. Zatorski, S. Kryzanowski, Acta Phys. Chem. 29 (1978) 347. [43] J.K.A. Clarke, R. Darcy, B.F. Hegarty, E. O’Donoghue, V. Amir-Ebrahimi, J.J. Rooney, J. Chem. Soc., Chem. Commun. (1986) 425. [44] S. Kolboe, in: A. Holmen K.-J. Jens, S. Kolboe (Eds.), Natural Gas Conversion, Stud. Surf. Sci. Catal. 61, Elsevier, Amsterdam, 1991, p. 413. [45] J. Sauer, M. Sierka, F. Haase, in: D.G. Truhlar, K. Morokuma (Eds.), Transition State Modeling for Catalysis, Am. Chem. Soc. Symp. Ser. 721, American Chemical Society, Washington DC, 1999, p. 358. [46] S.R. Blaszkowski, R.A. van Santen, J. Phys. Chem. 99 (1995) 11728. [47] S.R. Blaszkowski, R.A. van Santen, J. Am. Chem. Soc. 119 (1997) 5020. [48] W. Song, D.M. Marcus, H. Fu, J.O. Ehresmann, J.F. Haw, J. Am. Chem. Soc. 124 (2002) 3844. [49] S. Kolboe, in: L. Guszi, F. Solymosi, P. Tetenyi (Eds.), New Frontiers in Catalysis. Proceedings of the 10th International Congress on Catalysis, Stud. Surf. Sci. Catal. 75, Elsevier, Amsterdam, 1993, p. 449. [50] W. Wang, A. Buchholz, M. Seiler, M. Hunger, J. Am. Chem. Soc. 125 (2003) 15260. [51] N.Y. Chen, W.J. Reagan, J. Catal. 59 (1979) 123. [52] S. Kolboe, in: D.M. Bibby, C.D. Chang, R.F. Howe, S. Yurchak (Eds.), Methane Conversion, Stud. Surf. Sci. Catal. 36, Elsevier, Amsterdam, 1988, p. 189. [53] S. Kolboe, Acta Chem. Scand. A40 (1986) 711. [54] R.M. Dessau, J. Catal. 99 (1986) 111. [55] R.M. Dessau, R.B. LaPierre, J. Catal. 78 (1982) 136. [56] T. Mole, J. Whiteside, D. Seddon, J. Catal. 82 (1983) 261. [57] T. Mole, G. Bett, D, Seddon, J. Catal. 84 (1983) 435. [58] I.M. Dahl, S. Kolboe, Catal. Lett. 20 (1993) 329. [59] I.M. Dahl, S. Kolboe, J. Catal. 149 (1994) 458. [60] I.M. Dahl, S. Kolboe, J. Catal. 161 (1996) 304. [61] S. Kolboe, I.M. Dahl, in: H.K. Beyer, H.G. Karge, I. Kiricsi, J.B. Nagy (Eds.), Catalysis by Microporous Materials, Stud. Surf. Sci. Catal. 94, Elsevier, Amsterdam, 1995, p. 427. [62] P.O. Rønning, Ø. Mikkelsen, S. Kolboe in: M.M.J. Treacy, B.K. Marcus, M.E. Bisher, J.B. Higgins (Eds.), Proceedings of the 12th International Zeolite Conference, vol. II, Materials Research Society, Warrendale, USA, 1999, p. 1057.

46 [63] Ø. Mikkelsen, P.O. Rønning, S. Kolboe, Microporous Mesoporous Mater. 40 (2000) 95. [64] B. Arstad, S. Kolboe, Catal. Lett. 71 (2001) 209. [65] B. Arstad, S. Kolboe, J. Am. Chem. Soc. 123 (2001) 8137. [66] M. Bjørgen, U. Olsbye, S. Kolboe, J. Catal. 215 (2003) 30. [67] M. Bjørgen, U. Olsbye, D. Petersen, S. Kolboe, J. Catal. 221 (2004) 1. [68] M. Bjørgen, U. Olsbye, S. Svelle, S. Kolboe, Catal. Lett. 93 (2004) 37. [69] M. Bjørgen, F. Bonino, S. Kolboe, K.-P. Lillerud, A. Zecchina, S. Bordiga, J. Am. Chem. Soc. 125 (2003) 15863. [70] R.F. Sullivan, C.J. Egan, G.E. Langlois, R.P. Sieg, J. Am. Chem. Soc. 83 (1961) 1156. [71] J.F. Haw, Phys. Chem. Chem. Phys. 4 (2002) 5431. [72] J.F. Haw, W. Song, D.M. Marcus, J.B. Nicholas, Acc. Chem. Res. 36 (2003) 317. [73] P.W. Goguen, T. Xu, D.H. Barich, T.W. Skloss, W. Song, Z. Wang, J.B. Nicholas, J.F. Haw, J. Am. Chem. Soc. 120 (1998) 2650. [74] J.F. Haw, J.B. Nicholas, W. Song, F. Deng, Z. Wang, T. Xu, C.S. Heneghan, J. Am. Chem. Soc. 122 (2000) 4763. [75] W. Song, J.F. Haw, J.B. Nicholas, C.S. Heneghan, J. Am. Chem. Soc. 122 (2000) 10726. [76] W. Song, H. Fu, J.F. Haw, J. Phys. Chem. B, 105 (2001) 12839. [77] W. Song, H. Fu, J.F. Haw, J. Am. Chem. Soc. 123 (2001) 4749. [78] H. Fu, W. Song, J.F. Haw, Catal. Lett. 76 (2001) 89. [79] A. Sassi, M.A. Wildman, H.J. Ahn, P. Prasad, J.B. Nicholas, J.F. Haw, J. Phys. Chem. B 106 (2002) 2294. [80] A. Sassi, M.A. Wildman, J.F. Haw, J. Phys. Chem. B 106 (2002) 8768. [81] M. Seiler, W. Wang, A. Buchholz, M. Hunger, Catal. Lett. 88 (2003) 187. [82] M. Hunger, T. Horvath, J. Am. Chem. Soc. 118 (1996) 12302. [83] M. Seiler, W. Wang, M. Hunger, J. Phys. Chem. B 105 (2001) 8143. [84] W. Wang, M. Seiler, M. Hunger, J. Phys. Chem. B 105 (2001) 12553. [85] T. Xu, D.H. Barich, P.W. Goguen, W. Song, Z. Wang, J.B. Nicholas, J.F. Haw, J. Am. Chem. Soc. 120 (1998) 4025. [86] B. Arstad, J.B. Nicholas, J.F. Haw, J. Am. Chem. Soc. 126 (2004) 2991. [87] V.N. Romanikov, K.G. Ione, Kinet. Catal. 25 (1984) 75. [88] J.R. Anderson, Appl. Catal. 47 (1989) 177.

47 [89] C.E. Taylor, in: A. Corma, F.V. Melo, S. Mendioroz, J.L.G. Fierro (Eds.), Proceedings of the 12th International Congress on Catalysis, Stud. Surf. Sci. Catal. 130D, Elsevier, Amsterdam, 2000, p. 3633. [90] C.E. Taylor, R.P. Noceti, R.R. Schehl, in: D.M. Bibby, C.D. Chang, R.F. Howe, S. Yurchak (Eds.), Methane Conversion, Stud. Surf. Sci. Catal. 36, Elsevier, Amsterdam, 1988, p. 483. [91] I. Lorkovic, M. Noy, M. Weiss, J. Sherman, E. McFarland, G.D. Stucky, P.C. Ford, Chem. Commun. (2004) 566. [92] Y. Sun, S.M. Campbell, J.H. Lunsford, G.E. Lewis, D. Palke, L.-M. Tau, J. Catal. 143 (1993) 32. [93] K.-J. Jens, S. Halvorsen, E.B. Ofstad, in: D.M. Bibby, C.D. Chang, R.F. Howe, S. Yurchak (Eds.), Methane Conversion, Stud. Surf. Sci. Catal. 36, Elsevier, Amsterdam, 1988, p 491. [94] C.M. White, L.J. Douglas, J.P. Hackett, R.R. Anderson, Energy & Fuels, 6 (1992) 76. [95] G.A. Olah, B. Gupta, M. Farina, J.D. Felberg, W.M. Ip, A. Husain, R. Karpeles, K. Lammertsma, A.K. Melhotra, N.J. Trivedi, J. Am. Chem. Soc. 107 (1985) 7097. [96] C.T. O’Connor, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis – Vol. 4, VCH Verlagsgesellschaft mbH, Weinheim, Germany, 1997, p 2380. [97] L. Schmerling, V.N. Ipatieff, Adv. Catal. 2 (1950) 21. [98] S.A. Tabak, F.J. Krambeck, W.E. Garwood, AlChE J. 32 (1986) 1526. [99] C. Butcher, The Chemical Engineer: The Bulletin of the Institution of Chemical Engineers 489 (1991) 12. [100] J. Bandiera, Y. Ben Taarit, Appl. Catal. A 132 (1995) 157. [101] G. Spoto, S. Bordiga, G. Ricchiardi, D. Scarano, A. Zecchina, E. Borello, J. Chem. Soc., Faraday Trans. 90 (1994) 2827. [102] F. Geobaldo, G. Spoto, S. Bordiga, C. Lamberti, A. Zecchina, J. Chem. Soc., Faraday Trans. 93 (1997) 1243. [103] J.F. Haw, B.R. Richardson, I.S. Oshiro, N.D. Lazo, J.A. Speed, J. Am. Chem. Soc. 111 (1989) 2052. [104] H. Ishikawa, E. Yoda, J.N. Kondo, F. Wakabayashi, K. Domen, J. Phys. Chem. 103 (1999) 5681. [105] J.N. Kondo, H. Ishikawa, E. Yoda, F. Wakabayashi, K. Domen, J. Phys. Chem. 103 (1999) 8538.

48 [106] M. Trombetta, G. Busca, S. Rossini, V. Piccoli, U. Cornaro, J. Catal. 168 (1997) 349. [107] P.J. Hay, A. Redondo, Y. Guo, Catal. Today 50 (1999) 517. [108] M.V. Frash, V.B. Kazansky, A.M. Rigby, R.A. van Santen, J. Phys. Chem. B 102 (1998) 2232. [109] A.M. Rigy, G.J. Kramer, R.A. van Santen, J. Catal. 170 (1997) 1. [110] V.B. Kazansky, Catal. Today 51 (1999) 419. [111] M.V. Frash, R.A. van Santen, Top. Catal. 9 (1999) 191. [112] P.O. Rønning, Ph. D. thesis, Department of Chemistry, University of Oslo, (1998). [113] Ø. Mikkelsen, Ph. D. thesis, Department of Chemistry, University of Oslo, (1998). [114] B. Arstad, S. Kolboe, O. Swang, J. Phys. Chem. B 106 (2002) 12722. [115] B. Arstad, S. Kolboe, O. Swang, J. Phys. Chem. B 108 (2004) 2300. [116] P.E. Sinclair, A. de Vries, R. Sherwood, C.R.A. Catlow, R.A. van Santen, J. Chem. Soc., Faraday Trans. 94 (1998) 3401. [117] M. Boronat, C.M. Zicovich-Wilson, A. Corma, P. Viruela, Phys. Chem. Chem. Phys. 1 (1999) 537. [118] A.M. Vos, X. Rozanska, R.A. Schoonheydt, R.A. van Santen, F. Hutschka, J. Hafner, J. Am. Chem. Soc. 123 (2001) 2799. [119] X. Rozanska, X. Saintigny, R.A. van Santen, F. Hutschka, J. Catal. 202 (2001) 141. [120] M. Boronat, C.M. Zicovich-Wilson, P. Viruela, A. Corma, J. Phys. Chem. B 105 (2001) 11169. [121] A.M. Vos, K.H.L. Nulens, F. De Proft, R.A. Schoonheydt, P. Geerlings, J. Phys. Chem. B 106 (2002) 2026. [122] X. Rozanska, X. Saintigny, R.A. van Santen, S. Clémendot, F. Hutschka, J. Catal. 208 (2002) 89. [123] A.M. Vos, R.A. Schoonheydt, F. De Proft, P. Geerlings, J. Phys. Chem. B 107 (2003) 2001. [124] A.D. Becke, J. Chem. Phys. 98 (1993) 5648. [125] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785. [126] P.C. Hariharan, J.A. Pople, Chem. Phys. Lett. 16 (1972) 217. [127] P.J. Hay, W.R. Wadt, J. Chem. Phys. 82 (1985) 284. [128] C.E. Check, T.O. Faust, J.M. Bailey, B.J. Wright, T.M. Gilbert, L.S. Sunderlin, J. Phys. Chem. A 105 (2001) 8111. [129] Gaussian 98 (Revision A.10), M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, V.G. Zakrzewski, J.A. Montgomery, Jr., R.E. Stratmann,

49 J.C. Burant, S. Dapprich, J.M. Millam, A.D. Daniels, K.N. Kudin, M.C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G.A. Petersson, P.Y. Ayala, Q. Cui, K. Morokuma, P. Salvador, J.J. Dannenberg, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J. Cioslowski, J.V. Ortiz, A.G. Baboul, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, J.L. Andres, C. Gonzalez, M. Head-Gordon, E.S. Replogle, J.A. Pople, Gaussian, Inc., Pittsburgh PA, 2001. [130] M. Brändle, J. Sauer, J. Mol. Catal. A: Chem. 119 (1997) 19. [131] J. Sauer, M. Sierka, J. Comp. Chem. 21 (2000) 1470. [132] L.A. Clark, M. Sierka, J. Sauer, in: R. Aiello, G. Giordano, F. Testa (Eds.), Impact of Zeolites and Other Porous Materials on the New Technologies at the Beginning of the New Millennium, Stud. Surf. Sci. Catal. 142, Elsevier, Amsterdam, 2002, p. 643. [133] U. Eichler, M. Brändle, J. Sauer, J. Phys. Chem. B 101 (1997) 10035. [134] M. Boronat, C.M. Zichovich-Wilson, P. Viruela, A. Corma, Chem. Eur. J. 7 (2001) 1295. [135] M. Boronat, P.M. Viruela, A. Corma, J. Am. Chem. Soc. 126 (2004) 3300. [136] M.E. Franke, M. Sierka, U. Simon, J. Sauer, Phys. Chem. Chem. Phys. 4 (2002) 5207. [137] R. Shah, J.D. Gale, M.C. Payne, J. Phys. Chem. B 101 (1997) 4787. [138] X. Rozanska, T. Demuth, F. Hutschka, J. Hafner, R.A. van Santen, J. Phys. Chem. B 106 (2002) 3248. [139] X. Rozanska, R.A. van Santen, F. Hutschka, J. Hafner, J. Catal. 215 (2003) 20. [140] X. Rozanska, R.A. van Santen, T. Demuth, F. Hutschka, J. Hafner, J. Phys. Chem. B 107 (2003) 1309. [141] X. Rozanska, R.A. van Santen, F. Hutschka, J. Phys. Chem. B 106 (2002) 4652. [142] X. Rozanska, R.A. van Santen, F. Hutschka, J. Hafner, J. Am. Chem. Soc. 123 (2001) 7655. [143] T. Demuth, P. Raybaud, S. Lacombe, H. Toulhoat, J. Catal. 222 (2004) 323. [144] J. Sauer, P. Ugliengo, E. Garrone, V.R. Saunders, Chem. Rev. 94 (1994) 2095. [145] X. Rozanska, R.A. van Santen, in: S.M. Auerbach, K.A. Carrado, P.K. Dutta (Eds.), Handbook of Zeolite Science and Technology, Marcel Dekker, New York, 2003, p. 785. [146] B. Arstad, S. Kolboe, O. Swang, J. Phys. Org. Chem. in press, 2004.

50 [147] V.R. Choudhary, S. Mayadevi, Zeolites 17 (1996) 501. [148] P. Pascual, P. Ungerer, B. Tavitian, A. Boutin, J. Phys. Chem. B 108 (2004) 393.

51 52

Appendix: Papers I - VI

Paper I

A Theoretical Investigation of the Methylation of Methylbenzenes and Alkenes by Halomethanes over Acidic Zeolites

S. Svelle, S. Kolboe, O. Swang, J. Phys. Chem. B 107 (2003) 5251-5260

J. Phys. Chem. B 2003, 107, 5251-5260 5251

A Theoretical Investigation of the Methylation of Methylbenzenes and Alkenes by Halomethanes over Acidic Zeolites

Stian Svelle,† Stein Kolboe,*,† Unni Olsbye,† and Ole Swang‡ Department of Chemistry, UniVersity of Oslo, P.O. Box 1033 Blindern, N-0315 Oslo, Norway, and SINTEF Applied Chemistry, Department of Hydrocarbon Process Chemistry, P.O. Box 124 Blindern, N-0134 Oslo, Norway ReceiVed: January 24, 2003; In Final Form: March 17, 2003

The reactivity of chloromethane, bromomethane, and iodomethane over acidic zeolite catalysts has been probed, using the methylation of ethene, propene, and toluene as model reactions. Adsorbed reactants, transition states, and adsorbed products have been investigated. A cluster model has been used to represent the zeolite. Both the associative (direct) mechanism and the dissociative mechanism (via the formation of surface methoxide groups) have been studied. Quantum chemistry predicts that the associative pathway is favored over the dissociative mechanism. A weak trend in reactivity is found among the halomethanes, indicating the following order of reactivity: MeI > MeBr > MeCl. The results are compared with recent theoretical studies on methanol reactivity.

1. Introduction water as byproduct, which constitutes 55.6% (by weight) of the The proven world reserves of natural gas (mainly methane) products in methanol conversion, is thereby avoided. are of equal magnitude to those of oil.1 Therefore, activation Early patents, mainly describing the conversion of methanol of methane and subsequent conversion of activated methane to to hydrocarbons over acidic zeolites, also mention the possibility higher hydrocarbons or other value-added products is a research of converting halomethanes.10,11 Some reports also exist in the area to which considerable effort has been devoted.2 Currently, regular scientific literature, mainly focusing on chloromethane most such routes of gas upgrading proceed via the formation as reactant.12-22 Utilizing bromomethane or iodomethane as 23 of synthesis gas (CO + H2). This step may be achieved by steam possible reactants is occasionally mentioned, but little experi- reforming of methane using a supported nickel catalyst.3 The mental data are available for comparison between the three. reaction temperature employed is 800-900 °C, which might Murray et al.24 have found that the order of reactivity follows result in a considerable loss of energy during heat transfer the expected leaving group trend (MeI > MeBr > MeCl). processes. Synthesis gas may be reacted into higher hydrocar- However, the catalysts used were alkali-exchanged zeolites, and bons using Fischer-Tropsch technology, employing various a comparison with purely proton-exchanged zeolites may not 4 metal catalysts. An alternative route involves Cu/Zn/Al2O3 be valid. catalyzed synthesis of methanol from synthesis gas5 and There does not appear to be a clear consensus on whether subsequent conversion of methanol to hydrocarbons (MTH).6,7 halogen atoms are incorporated in the hydrocarbons produced Depending on product demand, methanol can be reacted into a or not. It has been reported that up to 4% of the product from gasoline range mixture of hydrocarbons (methanol to gasoline, chloromethane conversion contained halogen,13,14,18,22 while MTG) using the acidic H-ZSM-5 catalyst or to a product mixture Lersch and Bandermann16 did not detect any chlorinated consisting mainly of ethene and propene (methanol to olefins, products. Minimizing the amount of halogenated products is MTO) by employing acidic H-SAPO-34 catalysts.6,7 beneficial from an industrial and environmental viewpoint. Circumvention of the costly formation of synthesis gas might Many articles on the subject point out the similarities in be achieved, however, by transforming methane directly into product distribution, and probably also in reaction mechanism, monosubstituted halomethanes. This is possible through the between methanol and halomethane conversion.12,13,15,16,18,21 8 oxyhydrochlorination (OHC) process, where methane is reacted These similarities will also be drawn upon in the present work. with hydrogen chloride and oxygen to form chloromethane and Recently, great progress has been made regarding the mecha- water over a supported copper chloride catalyst. Chloromethane nistic description of methanol conversion. Dahl and Kolboe25 may also be produced by monohalogenation of methane over have formulated a “hydrocarbon pool” mechanism, where 9 supported platinum metal catalysts. In the second step, chlo- methanol is continuously added to an adsorbate inside the zeolite romethane, or any halomethane, may be converted to hydro- pores, followed by production of alkenes via subsequent carbons and hydrogen halides over acidic microporous catalysts, dealkylations of the adsorbate. The specific nature of the for instance H-SAPO-34 or H-ZSM-5. The produced hydrogen hydrocarbon pool has been clarified in a series of publications halide may then be separated from the hydrocarbon product and from Kolboe and co-workers,26 Haw and co-workers,27 and 8 recycled back to the halogenation process step. Formation of Hunger and co-workers.28 It appears that polymethylbenzenes, polymethylnaphthalenes, polymethylated cyclopentenyl ions, * Corresponding author. E-mail: [email protected]. † University of Oslo. and, in some cases, aliphatic species can constitute the hydro- ‡ SINTEF Applied Chemistry. carbon pool. There may also be a facile interconversion of these 10.1021/jp030101u CCC: $25.00 © 2003 American Chemical Society Published on Web 05/09/2003 5252 J. Phys. Chem. B, Vol. 107, No. 22, 2003 Svelle et al.

SCHEME 1: Catalytic Cycle of the “Hydrocarbon Pool” in the Gaussian98 program were applied for hydrogen, oxygen, Mechanisma carbon, aluminum, and silicon. LANL2DZdp effective core potentials40,41 (ECPs) were used on chlorine, bromine, and iodine. Additionally, single-point electronic energies were calculated for the optimized geometries using MP2 and basis sets as above. Single-point calculations were also carried out using B3LYP and the full 6-311G(d,p) basis sets (no ECPs). The 6-311G(d,p) basis set for iodine was taken from ref 42. All basis sets not included in Gaussian98 were obtained from the EMSL Web site.43 The ultrafine integration grid was used for geometry optimizations in order to ensure convergence. a In the present work, we have investigated the methylation of B3LYP_ECP and MP2_ECP will in the following be used as compounds known to exist inside the zeolite pores during halomethane acronyms to indicate when ECPs have been used. conversion. The zeolite catalyst has been modeled using a cluster consisting of four tetrahedral atoms, that is, three silicon atoms hydrocarbon pool species. The hydrocarbon pool mechanism and one aluminum atom, to generate the acidic site.36,37 To is clarified in Scheme 1. reduce the effect of using a finite cluster model, care was taken If indeed the mechanistic similarities between methanol and to ensure that reactants and products were coordinated similarly halomethane conversion are as profound as suggested above, to the cluster for all reactions. For all stationary points, the main difference must be the addition of C1 entities to the vibrational spectra were calculated to ensure that the correct hydrocarbon pool, that is, methylation steps. Dealkylation steps number of imaginary frequencies was at hand, that is, one should be independent of the original reactant. In this report imaginary frequency for transition states and zero for energy we present a description of the methylation of ethene, propene,29 minima. For the transition states, the normal modes correspond- and toluene by chloromethane, bromomethane, and iodomethane, ing to the imaginary frequencies were visualized to confirm that using quantum chemical methods. A 19 atom cluster model they indeed corresponded to the expected motion of atoms. containing four tetrahedral atoms (denoted 4T) has been used Internal reaction coordinate (IRC) calculations, as implemented to represent the zeolite catalyst. Within the cluster approach, a in Gaussian98, were in some cases carried out. Such calculations small fragment is used to simulate the Bro¨nsted acidic site. follow reaction paths in order to investigate the minima Calculations with such clusters have shown them to be sufficient connected by the transition states. A step size of 0.3 amu-1/2 to qualitatively describe chemical rearrangements that occur bohr and the ultrafine integration grid were used. locally on the active site.30 Structure specific effects and effects caused by the electrostatic field present in the zeolite micropores 3. Results and Discussion are, however, not well described. The cluster employed in this work has been used by several workers to model reactions Methylation of ethene, propene, and toluene has been chosen similar to those described here.31-37 Rozanska31,33 et al. have as model reactions to probe the relative reactivity of chlo- studied the isomerization and transalkylation of toluene and romethane, bromomethane, and iodomethane over acidic zeolite xylenes and found that the relative order of activation energies catalysts. The reactions result in formation of propene (from is conserved when comparing results from cluster studies with ethene), trans-2-butene (from propene), and p-xylene (from calculations relying on periodical boundary conditions combined toluene) together with a hydrogen halide molecule. The other with plane wave basis sets. Also, alkylation of arenes32,34,36 and possibilities, resulting in cis-2-butene from propene and m-or alkenes37 by methanol and reactions of sulfur containing o-xylene from toluene, have not been investigated. Adsorption compounds35 have been successfully investigated using the 4T modes, transition barriers, and the reaction products have been atom cluster. analyzed, both energetically and geometrically. Recent theoreti- 37 36 Adsorbed reactants, transition states, and adsorbed products cal reports on methylation of alkenes and methylbenzenes have been optimized, and both the associative (direct) and with methanol describe the formation of charged hydrocarbon dissociative (via the formation of framework methoxide groups) species as shallow energy minima on the potential energy methylation mechanisms have been examined. In total, 15 surface (PES) in the presence of water. A test calculation was reaction steps, encompassing 45 stationary points, have thus performed for the methylation of propene by chloromethane, been modeled. The order of reactivity among the halomethanes and a stationary point (without imaginary frequencies) describing has been assessed, and comparison with recent relevant theoreti- an adsorbed charged hydrocarbon species was found also in cal work on methanol reactivity31-37 has been made. No attempts this case. The hydrocarbon fragment was bridged, similar to - methyl substituted protonated cyclopropane, as has been ob- have been made to describe initial formation of C C bonds. 37 This reaction step has been claimed to be of little importance served previously. However, the coordination of the hydro- for methanol conversion,38 as it is clear that the hydrocarbon carbon species to the cluster was somewhat unsatisfying in the pool mechanism prevails once a reasonable amount of hydro- sense that the hydrogen atoms available for return to the catalyst carbon adsorbate has been built up inside the zeolite pores. We were oriented away from the zeolite oxygen atoms and into expect this argument to be valid for halomethane conversion vacuum. Further, the barrier for reorientation and deprotonation as well. is expected to be very low. Nonetheless, it seems fairly clear that methylation reactions have to proceed via geometries closely resembling carbocations, whether these are stationary points or 2. Computational Details not.37 All calculations were done using the Gaussian98 program Two different likely mechanistic pathways have been inves- package.39 The B3LYP hybrid density functional was employed tigated: The associative mechanism, which is described in for all geometry optimizations. No geometric constraints were subsection 3.1, and the dissociative mechanism, described in used in the optimizations. The 6-31G(d) basis sets incorporated subsection 3.2. In the following, when energies are discussed, Conversion of Halomethanes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 22, 2003 5253

chloromethane on zeolite acid sites has been examined previously by Su and Jaumain,45 using FT-IR spectroscopy. They found that the interaction between the chlorine and the acidic proton caused the zeolite hydroxyl to be stretched sufficiently to cause a measurable shift in the IR frequency, which correlates well with the calculated increase in bond length reported here. No discussion on the side-on nature of the adsorption was given by Su and Jaumain.45 The double bonds of ethene and propene are coordinated to the methyl carbon atom of the halomethane and to the cluster itself. The XCm axis is directed toward the double bond, and the four atoms are all in the same plane, as can clearly be seen from the XC1C2Cm dihedral angles. Toluene is adsorbed in a similar manner, except that the methyl groups of the halomethanes are not pointing toward any specific region of the Figure 1. Stationary points along the reaction path for associative toluene molecule; see Figure 2. Therefore, only the XC C (C methylation of ethene by chloromethane. m 1 1 is the ring carbon in the para position to the toluene methyl group) angle is given in Table 2. The coadsorption energies are between 30 and 40 kJ/mol. A weak trend is observed; it appears that increasing the halogen atom size results in a slightly weaker interaction. Replacing ethene or propene with toluene makes the adsorption somewhat stronger. The adsorption energies are fairly modest, compared to energies close to 90 kJ/mol calculated in exactly the same manner for coadsorption of methanol and a series of different hydrocarbons on the cluster.36,37 This means that, in experimental practice, higher partial pressures must be applied for the halomethanes than for methanol in order to reach the same degree of surface coverage. All adsorption energies become considerably larger when the MP2 level of theory is used. This has also been observed in the previous studies on methanol reactivity,36,37 and it can be ascribed to the superior description of long-range van der Waals interactions by the MP2 scheme Figure 2. Stationary points along the reaction path for associative 46 methylation of toluene by chloromethane. compared to the B3LYP methodology. 3.1.2. Transition States and ActiVation Barriers. In the we refer to the B3LYP_ECP/6-31G(d) + ZPE values, unless transition states, the XCm bond is lengthened and the methyl otherwise stated. group is moving from the halogen toward the hydrocarbon to + 3.1. Associative Methylation. In this pathway, the halom- be methylated. Umbrella-like inversion of the formal CH3 ethane and hydrocarbon reactant are initially coadsorbed on the group has proceeded beyond the planar inversion point. This is cluster. In a single reaction step, the methyl group is transferred evident from the methyl dihedral angles in Table 2. Very from the halomethane to the hydrocarbon and a hydrogen halide significantly, the halomethanes are not protonated. However, is formed. The resulting product hydrocarbon (with one more the Oz1Ha1 bonds are considerably stretched, and the Ha1X C-atom than the reactant) and the hydrogen halide molecule distances correspondingly shortened, but no complete proton are then coadsorbed. Figure 1 shows the stationary points for transfer has occurred. An IRC calculation was performed for the associative methylation of ethene by chloromethane, and the methylation of propene by chloromethane, and this con- Figure 2 displays the same information for methylation of firmed that the correct transition state had been found. The IRC toluene by chloromethane. Energies and geometric parameters calculation also revealed that the acidic proton is transferred related to associative methylation are listed in Tables 1 and 2, from the zeolite oxygen to the halogen after the bond between respectively. the hydrocarbon and the methyl cation is formed. 3.1.1. Adsorbed Reactants. The energetically most favored For ethene, it appears that the methyl cation is approaching adsorption mode for the halomethanes can be labeled as side- the center of the double bond in a three-ring arrangement, which on adsorption, as can be seen in Figure 1. The large diffuse is indicated by the almost identical distances from Cm to the electron cloud of the halogen atom is associated with the acidic carbon atoms of the double bond. This nearly symmetrical proton of the zeolite model. The Oz1Ha1 bond is slightly coordination has been described previously in a similar cluster stretched, from 0.97 Å for the free cluster model to 0.98 Å upon study in which methanol was the methylating agent.37 It was coordination of a halomethane to the acidic site. The Ha1X then concluded that the reaction pathway from the transition distance depends on halogen size, and increasing the halogen state down to propene proceeds via a protonated cyclopropane, atom size also causes the halomethane methyl carbon to be lifted which undergoes a single-step isomerization to form a secondary up from the cluster, as can be seen from the CmAl distances. propyl cation. Deprotonation then yields propene. The similari- One of the hydrogen atoms of the methyl group is coordinated ties between the mechanistic details disclosed here and those to a zeolite oxygen atom. This weak interaction is indicated by described for the methanol reactions are so profound that this the HmOz2 distances in Table 2. The side-on adsorption causes conclusion must also be valid when a halomethane is reacting the XCm bonds to be stretched by 0.01-0.03 Å, relative to the with ethene. For propene, the CmC1 and CmC2 distances are quite bond lengths calculated in the gas phase. Adsorption of different, and it seems that the methyl cation is being attached 5254 J. Phys. Chem. B, Vol. 107, No. 22, 2003 Svelle et al.

TABLE 1: Energies of Stationary Points for Associative Methylation Reactions and Gas-Phase Reaction Energiesa energy of gas-phase transition adsorbed gas-phase gas-phase calc gas-phase exp gas-phase reactant state products products reaction energy reaction enthalpy reaction enthalpyb B3LYP_ECP/6-31G(d) + ZPE ethene + chloromethane 35 123 -53 7 -28 -27 -41 bromomethane 34 118 -43 12 -22 -21 -34 iodomethane 31 117 -32 14 -17 -17 -20 propene + chloromethane 36 108 -51 9 -27 -25 -40 bromomethane 35 101 -40 14 -21 -19 -33 iodomethane 31 97 -30 15 -17 -15 -19 toluene + chloromethane 39 128 -44 13 -25 -22 -41 bromomethane 38 120 -33 18 -19 -16 -34 iodomethane 34 115 -23 19 -15 -12 -20 B3LYP_ECP/6-31G(d) ethene + chloromethane 42 124 -50 21 -21 bromomethane 41 118 -39 26 -15 iodomethane 37 116 -28 27 -10 propene + chloromethane 42 110 -48 23 -19 bromomethane 41 102 -37 28 -13 iodomethane 37 97 -25 30 -8 toluene + chloromethane 47 133 -35 32 -15 bromomethane 45 123 -24 36 -9 iodomethane 41 116 -14 37 -4 B3LYP/6-311G(d,p)// B3LYP_ECP/6-31G(d) ethene + chloromethane 35 126 -62 2 -34 bromomethane 35 122 -48 11 -24 iodomethane 35 125 -30 19 -17 propene + chloromethane 37 115 -59 5 -31 bromomethane 35 109 -42 14 -22 iodomethane 33 100 -29 19 -14 toluene + chloromethane 32 129 -53 6 -27 bromomethane 40 130 -29 23 -17 iodomethane 34 118 -18 24 -9 MP2_ECP/6-31G(d)// B3LYP_ECP/6-31G(d) ethene + chloromethane 56 138 -58 30 -26 bromomethane 55 125 -47 37 -18 iodomethane 50 117 -35 39 -11 propene + chloromethane 59 129 -59 33 -26 bromomethane 59 124 -46 41 -18 iodomethane 55 118 -34 44 -11 toluene + chloromethane 68 161 -53 41 -27 bromomethane 67 148 -41 48 -19 iodomethane 60 135 -35 49 -12 a All values are relative to the adsorbed reactants and are in kJ/mol. Enthalpies are at 298 K. b Experimentally determined gas-phase reaction enthalpies taken from ref 44. to the least substituted carbon atom of the double bond and atom and the nature of the hydrocarbon. On the basis of the that a secondary butyl cation is formed, which agrees well with activation energies, listed in column 2 of Table 1, two issues the expected outcome of an electrophilic substitution reaction. merit further discussion. First, propene is more easily methylated Subsequent deprotonation results in formation of 2-butene. than ethene, and toluene is predicted to have a reactivity similar When toluene is methylated, there is no three-ring arrangement. to that of ethene. The same order has been observed in similar A bond is formed directly between the methyl cation and the studies with methanol as methylating agent.36,37 The higher carbon atom in the para position to the toluene methyl group. reactivity of propene with respect to ethene found both here The XCmC1 angles, which are nearly 180°, are reported in Table and in the literature may be rationalized by considering the 2. This is again analogous to the geometry of the transition state cation expected to form immediately after the transition state. for methylbenzene methylation with methanol.32,34,36 For ethene, highly unstable protonated cyclopropane is formed, The barrier height and the geometry of the transition state while an energetically favored secondary butyl cation is formed may be expected to depend on both the nature of the halogen after the methylation of propene, and the barrier is accordingly Conversion of Halomethanes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 22, 2003 5255

TABLE 2: Geometric Parameters for Associative Methylationa ethene + propene + toluene + MeCl MeBr MeI MeCl MeBr MeI MeCl MeBr MeI adsorbed reactants Oz1Ha1 0.98 0.98 0.98 0.98 0.98 0.98 0.98 0.98 0.98 Ha1X 2.25 2.41 2.63 2.25 2.41 2.63 2.23 2.39 2.62 b XCm 1.83 1.99 2.18 1.83 1.99 2.18 1.83 1.99 2.18 XAl 4.34 4.50 4.69 4.34 4.51 4.68 4.44 4.59 4.75 CmAl 4.12 4.17 4.23 4.10 4.18 4.22 4.37 4.41 4.41 c CmC1 3.60 3.62 3.66 3.60 3.62 3.69 3.86 3.86 3.87 c CmC2 3.68 3.69 3.75 3.70 3.63 3.67 C1C2 1.33 1.33 1.33 1.34 1.34 1.34 HmOz2 2.47 2.46 2.47 2.46 2.44 2.47 2.44 2.43 2.45 Oz1Ha1X 167.0 167.8 166.5 166.8 167.7 165.9 171.0 171.1 169.8 XC1C2Cm 1.1 1.1 1.5 2.6 2.6 3.5 XCmC1 147.5 148.6 147.6 dihedral methyld 31.9 31.2 31.2 31.9 31.2 31.2 31.7 31.1 30.9 transition states Oz1Ha1 1.03 1.03 1.02 1.03 1.02 1.01 1.03 1.02 1.01 Ha1X 1.90 2.09 2.33 1.93 2.14 2.40 1.93 2.14 2.39 XCm 2.67 2.86 3.08 2.60 2.76 2.93 2.65 2.80 2.98 XAl 4.08 4.27 4.49 4.14 4.37 4.60 4.22 4.42 4.65 CmAl 3.71 3.76 3.81 3.66 3.71 3.78 4.05 4.12 4.20 c CmC1 2.08 2.06 2.04 2.05 1.99 1.99 1.94 1.93 1.94 c CmC2 2.09 2.07 2.06 2.31 2.41 2.49 C1C2 1.36 1.36 1.37 1.37 1.37 1.37 HmOz2 2.21 2.21 2.22 2.30 2.33 2.35 2.24 2.27 2.31 Oz1Ha1X 174.7 175.2 175.5 174.8 175.2 174.6 177.5 177.8 177.4 XC1C2Cm 2.4 2.6 2.8 2.3 2.9 2.9 XCmC1 175.1 175.3 175.6 dihedral methyl -19.6 -21.5 -22.9 -16.1 -16.9 -16.5 -18.7 -19.2 -19.0 imaginary frequency of TS 327i 316i 313i 341i 336i 359i 354i 354i 367i adsorbed products Oz1Ha1 1.69 1.71 1.80 1.70 1.72 1.83 1.71 1.73 1.84 Ha1X 1.33 1.49 1.67 1.33 1.49 1.66 1.33 1.49 1.66 XCm 5.20 5.37 5.59 4.32 4.53 4.92 4.42 4.68 6.02 XAl 4.04 4.22 4.49 4.09 4.26 4.53 4.06 4.25 4.49 e Ha2C1 2.13 2.14 2.15 2.15 2.15 2.21 e Ha2C2 2.27 2.27 2.27 2.20 2.20 2.17 Oz2Ha2 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 f Ha2C3 2.20 2.21 2.14 f Ha2C4 2.32 2.32 2.44 C3C4 1.40 1.40 1.40 C1C2 1.34 1.34 1.34 1.35 1.35 1.34 Oz1Ha1X 175.1 176.1 176.7 175.9 176.7 176.9 175.4 176.0 176.6 Oz2C1C2Ha2 0.6 0.6 0.9 0.8 0.7 0.9 2.3 2.2 1.9

a -1 b Distances in Å; angles in deg; frequencies in cm . Atom labels are defined in Figures 1 and 2. The calculated gas-phase XCm distances are c 1.80 Å for MeCl, 1.98 Å for MeBr, and 2.17 Å for MeI. For propene, C1 is the hydrogen substituted carbon and C2 is the methyl substituted carbon of the double bond. d This is the dihedral angle defined by the four atoms of the halomethane methyl group, HHHC. It describes the degree e f of inversion of the methyl group. For propene, formed from ethene, C2 is the methyl substituted carbon of the double bond. For p-xylene, formed from toluene, C3 and C4 are two hydrogen substituted ring carbons. The acidic proton is coordinated to the C3C4 bond. lower. This is also reflected in a shift in geometry toward a to be equal within 1 kJ/mol. Drawing definitive conclusions on more reactant-like transition state for the methylation of propene, the order of reactivity among the halomethanes on the basis of as predicted from the Hammond postulate.47 The fairly high these small differences may not be warranted. Both the activation barrier for toluene cannot be as easily rationalized. polarizabilities and dipole moments of the three halomethanes It might be speculated that the partial localization of aromatic are different, and the expected downward shift in barrier height electrons in the transition state causes the barrier to be relatively upon inclusion of the electrostatic field present in a real zeolite high, as some aromaticity is lost. catalyst might not be uniform. The calculated trend does, Second, the barriers for methylation by iodomethane are however, conform with the order of reactivity determined somewhat lower (by 1-5 kJ/mol) than those for methylation experimentally when alkali-exchanged zeolites were employed.24 by bromomethane, which in turn are lower (by 5-8 kJ/mol) Notably, there is a substantial increase in the barrier height for than those for chloromethane. This same trend was observed methylation of toluene calculated at the MP2_ECP/6-31G(d)// for all three hydrocarbon reactants and at every level of theory B3LYP_ECP/6-31G(d) level of theory, by 20-30 kJ/mol, employed, apart from two minor exceptions, found when the relative to the value at the B3LYP_ECP/6-31G(d) + ZPE level. B3LYP/6-311G(d,p)//B3LYP_ECP/6-31G(d) methodology was This shifts the barrier for toluene methylation well above the used. First, for methylation of ethene, the reactivity of bro- barrier for ethene methylation and illustrates that quantitative momethane is shifted above that of iodomethane, but the assessments of the results should be made cautiously. difference between the barriers for all three halomethanes is From the geometric parameters for the transition states listed only 4 kJ/mol. Second, for the methylation of toluene, the in Table 2, some effects of the halogen atom type may be reactivities of chloromethane and bromomethane are predicted discerned, and it is possible to correlate the small differences 5256 J. Phys. Chem. B, Vol. 107, No. 22, 2003 Svelle et al. in activation barriers with geometric trends. A lower barrier height again implicates that the reaction has passed a shorter distance along the reaction coordinate at the transition state, and the geometry should to a greater degree resemble that of the reactants. This is in part observed. The Oz1Ha1 bond length decreases when the barrier increases. Similar theoretical studies of the methylation of propene with methanol, where methanol is protonated in the transition state, have shown that the activation energy is linked to the ease of removal of the acidic proton from the catalyst model.37 In the methanol case, it seems reasonable to assume that a major contribution to the barrier lies in the breaking of the Oz1Ha1 bond in the protonation and the resulting separation of charges. When halomethanes are the methylating agent, there is no complete protonation, but it can be concluded that, for iodine, the XCm bond is sufficiently Figure 3. Stationary points along the reaction path for the formation activated with only a moderate stretch of the Oz1Ha1 bond. For of a surface methoxide from chloromethane. bromine, and especially chlorine, a stronger interaction with the acidic proton is required for XCm activation. The proton is therefore removed further from the cluster, and the activation energy is accordingly higher. There is no detectable systematic trend in the distances from the methyl cation to the hydrocarbon (CmC1 and CmC2) when going form chloromethane to bromomethane to iodomethane. 3.1.3. Adsorbed Products and Reaction Energies. The reaction products, a hydrogen halide molecule and a hydrocarbon with one more carbon atom than the reactant, are also coadsorbed on the cluster. The hydrocarbon is adsorbed on the acidic site, while the hydrogen halide is adsorbed end on and forms a hydrogen bond to one of the zeolite oxygen atoms. The Oz1Ha1 distance, which describes this hydrogen bond, is in every case shortest for hydrogen chloride and longest for hydrogen iodide, opposite of what might be expected on the basis of the Figure 4. Stationary points along the reaction path for the methylation differences in acid strength. Hydrogen iodide is the strongest of ethene by a surface methoxide. acid, in the gas phase, both experimentally and at the framework methyl group is transferred in an activated reaction B3LYP_ECP/6-31G(d) + ZPE level of theory.48 Increasing step. The net overall reaction is the same as that for the halogen atom size might also result in an increase in the steric associative mechanism. Figures 3 and 4 depict the stationary repulsion between the halogen and the cluster, and this could points for the formation of surface methoxide from chloro- explain the observed trend in the O H distances. Indeed, the z1 a1 methane and methylation of ethene by the methoxide species, sum of the O H and H X distances is in every case about z1 a1 a1 respectively. Energies and geometric parameters for methoxide 0.2 Å shorter than the sum of the ionic radii of oxygen and the formation are listed in Tables 3 and 4, and energies and relevant halogen. geometric details for methylations with the surface methoxide The coadsorption energies of the products are larger than will be found in Tables 5 and 6. those for the reactants, and this causes the cluster reaction 3.2.1. Formation of Zeolite Bonded Methoxide Groups. The energies to be more exothermic than the gas-phase reaction adsorption modes of the halomethanes appear to be virtually energies. For the products, there are two factors that mainly independent of the presence of a coadsorbate. Side-on adsorp- contribute to the adsorption energy: the hydrogen halide tion, with the halogen atom coordinated to the acidic site and associated to the zeolite oxygen and the acidic proton coordi- one of the hydrogen atoms of the methyl group in interaction nated to the electrons of the hydrocarbon double bond. This with a zeolite oxygen atom, is again favored. The distances compares to one such interaction for the reactants, that is, the describing the adsorbed halomethanes, listed in Table 4, are halogen atom coordinated to the acidic site. On the basis of the very nearly identical to those found in Table 2 for the number of proper interactions between adsorbate and adsorbent, halomethanes with a coadsorbate present. As can be seen from it is then fairly easy to explain the stronger adsorption of the the first column of Table 3, the adsorption energies are similar products relative to the reactants. for all halomethanes, only a slightly smaller energy is found The trend in experimental gas-phase reaction enthalpies, listed when the atomic number of the halogen increases. The same in column 7 of Table 1, is nicely reproduced by the calculations, trend was found for the coadsorption energies described in both in the adsorbed state and in the gas phase. The levels of subsection 3.1.1. The difference between the adsorption energy theory employed are, however, not sufficient to quantitatively of a halomethane alone and the adsorption energy found when reproduce the experimental reaction enthalpies. the halomethane and a hydrocarbon are coadsorbed should be 3.2. Dissociative Methylation. In the dissociative mechanism, a measure of the strength of the binding of the hydrocarbon in the halomethane is first adsorbed alone on the cluster. Then, in the coadsorption mode. A comparison of the energies listed in a single step, the halomethane dissociates as the methyl group column 1 of Table 1 and column 1 of Table 3 reveals that the is transferred to a zeolitic oxygen, and a hydrogen halide and a B3LYP functional with the employed basis sets inadequately framework bonded methoxide are formed. Later on, the describes this weak adsorption. There is no substantial difference hydrogen halide is displaced by a hydrocarbon, to which the in the adsorption energies found with or without a hydrocarbon Conversion of Halomethanes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 22, 2003 5257

TABLE 3: Energies of Stationary Points for Formation of TABLE 5: Energies of Stationary Points for Methylation of Methoxide and Adsorbed Hydrogen Halidea Ethene, Propene, and Toluene by a Zeolite Framework Methoxide Speciesa energy of energy of gas-phase transition adsorbed reactant state products gas-phase reactant transition state adsorbed product B3LYP_ECP/6-31G(d) + ZPE B3LYP_ECP/6-31G(d) + ZPE chloromethane 32 166 17 ethene 5 169 -65 bromomethane 31 155 27 propene 5 156 -63 iodomethane 28 149 37 toluene 8 171 -59 B3LYP_ECP/6-31G(d) B3LYP_ECP/6-31G(d) chloromethane 36 168 20 ethene 7 173 -65 bromomethane 35 156 31 propene 8 162 -63 iodomethane 32 149 43 toluene 11 179 -55 B3LYP/6-311G(d,p)// B3LYP_ECP/6-31G(d) B3LYP/6-311G(d,p)//B3LYP_ECP/6-31G(d) chloromethane 35 170 24 ethene 6 171 -78 bromomethane 36 159 42 propene 6 158 -77 iodomethane 34 150 54 toluene 6 172 -68 MP2_ECP/6-31G(d)// B3LYP_ECP/6-31G(d) MP2_ECP/6-31G(d)//B3LYP_ECP/6-31G(d) chloromethane 45 198 21 ethene 15 181 -74 bromomethane 44 181 33 propene 17 171 -75 iodomethane 40 169 45 toluene 26 190 -75 a All values are relative to the adsorbed halomethane and are in kJ/ a All values are relative to the energies of the hydrocarbons to be mol. methylated adsorbed on the methoxide substituted cluster and are in kJ/mol. TABLE 4: Geometric Parameters for the Formation of a Surface Methoxide Species and a Hydrogen Halide Molecule TABLE 6: Geometric Parameters for Methylation of from Chloromethane, Bromomethane, and Iodomethanea Ethene, Propene, and Toluene by a Surface Methoxide Speciesa MeCl MeBr MeI ethene propene toluene adsorbed reactants Oz1Ha1 0.98 0.98 0.98 adsorbed reactants Ha1X 2.26 2.42 2.65 Oz1Cm 1.45 1.45 1.45 b XCm 1.82 1.99 2.18 CmC1 3.97 4.02 3.84 b XAl 4.34 4.48 4.67 CmC2 4.13 4.01 CmAl 4.12 4.14 4.20 C1C2 1.33 1.34 HmOz2 2.39 2.39 2.40 CmAl 3.00 3.00 2.97 CmOz2 3.45 3.45 3.47 Oz1CmC1 133.0 Oz1Ha1X 166.2 166.3 165.0 Oz1C1C2Hm 16.2 15.7 XCmOz2 98.4 99.6 99.6 CmC1C1Hm 1.5 1.8 dihedral methylb 31.8 31.2 31.1 dihedral methylc 34.5 34.5 34.3 transition states transition states Oz1Ha1 1.01 1.01 1.01 Oz1Cm 2.25 2.22 2.25 Ha1X 2.12 2.31 2.57 CmC1 2.11 2.07 2.02 XCm 2.47 2.64 2.85 CmC2 2.15 2.29 CmOz2 1.93 1.92 1.91 C1C2 1.36 1.36 XAl 4.35 4.53 4.76 CmAl 3.22 3.21 3.12 CmAl 2.88 2.89 2.91 HmOz2 1.93 1.98 2.42 Oz1Ha1X 163.6 163.2 162.3 Oz1C1C2Hm 3.0 2.0 XCmOz2 146.0 147.1 148.5 Oz1CmC1 173.9 dihedral methyl -7.6 -9.0 -10.3 dihedral methyl -15.4 -14.1 -15.6 imaginary frequency of TS 494i 475i 462i imaginary frequency of TS 384i 380i 405i adsorbed products adsorbed products Oz1Ha1 1.73 1.76 1.88 Oz2Ha2 0.99 0.99 0.98 d Ha1X 1.33 1.48 1.66 Ha2C1 2.14 2.19 d XCm 3.93 4.12 4.42 Ha2C2 2.31 2.21 HmX 2.87 3.05 3.38 C1C2 1.34 1.34 Oz2Cm 1.46 1.46 1.45 Oz2C1C2Ha2 3.1 3.0 e XAl 4.04 4.22 4.51 Ha2C3 2.30 e CmAl 2.96 2.96 2.96 Ha2C4 2.33 CmHmX 164.2 165.7 160.4 C3C4 1.40 Oz1Ha1X 172.6 173.4 173.6 Oz2C3C4Ha2 0.6 dihedral methyl -34.2 -34.2 -34.3 a Distances in Å; angles and dihedral angles in deg; frequencies in a -1 b Distances in Å; angles and dihedral angles in deg; frequencies in cm . Atom labels are defined in Figures 4 and 2. For propene, C1 is -1 b cm . Atom labels are defined in Figure 3. This is the dihedral angle the hydrogen substituted carbon and C2 is the methyl substituted carbon defined by the four atoms of the halomethane methyl group, HHHC. of the double bond. c This is the dihedral angle defined by the four atoms of the halomethane methyl group, HHHC. d For propene, formed coadsorbate. At the MP2 level of theory, however, a difference from ethene, C2 is the methyl substituted carbon of the double bond. e was found, indicating that the interaction energies between the For p-xylene, formed from toluene, C3 and C4 are two hydrogen hydrocarbons and the cluster are in the range 10-20 kJ/mol, substituted ring carbons. The acidic proton is coordinated to the C3C4 depending on hydrocarbon size. bond. The transition states for surface methoxide formation are methylation, described above. The methyl cation is about qualitatively quite similar to those found for the associative halfway between the halogen atom and one of the zeolite oxygen 5258 J. Phys. Chem. B, Vol. 107, No. 22, 2003 Svelle et al. atoms. The halogen does clearly interact with the acidic site, but there is no complete protonation. IRC calculations showed that a bond was formed between the methyl cation and the zeolite oxygen before the acidic proton was completely transferred to the halogen. The normal mode of the transition state is defined by X, Cm, and Oz2, and as indicated by the XCmOz2 angles, there is considerable deviation from linearity. In fact, the transition state may be considered to be a six-ring arrangement of the involved atoms (XCmOz2AlOz1Ha1). The activation barriers for methoxide formation follow the same trend as was found for associative methylation, that is, the highest barrier is found for chloromethane and the lowest for iodomethane. The differences are somewhat more pronounced than previously, and they are largest at the MP2_ECP/ 6-31G(d)//B3LYP_ECP/6-31G(d) level of theory. The activation energies found are, however, substantially higher than those found for associative methylation, at every level of theory used. Notably, at the MP2_ECP/6-31G(d)//B3LYP_ECP/6-31G(d) level of theory, there is a considerable increase in the calculated barriers. Figure 5. Energy diagram for the complete reaction pathway of dissociative methylation. The energy difference between P1 and R2 Both the XCm and Oz1Ha1 distances are shorter than what corresponds to the difference in adsorption energy between the hydrogen was found in the case of associative methylation. Also, the halide and the hydrocarbon. This step in energy may be considered an degree of inversion of the methyl cation is considerably less effect of the limited cluster size. MeX ) halomethane, HC ) hydrocarbon, HZeo ) protonated cluster, HX ) hydrogen halide, pronounced. This illustrates how dissociation of the halomethane ) + ) molecule has proceeded further at the transition state for MethoxideZeo methoxide substituted cluster, HC 1 methylated hydrocarbon. Energies correspond to methylation of ethene by associative methylation than in the case for methoxide formation. chloromethane, in kJ/mol. Formation of the surface methoxide and a hydrogen halide molecule is an endothermic reaction. The trend in reaction MP2 is used. At the MP2 level it is also possible to order the energy among the different halomethanes is the same as that adsorption energies of the three hydrocarbons: The larger the for associative methylation; the reaction is thermodynamically hydrocarbon, the stronger is the adsorption. This is closely less unfavorable for chloromethane than for bromomethane and related to the assessment of the strength of the interaction iodomethane. between a hydrocarbon and a cluster with a halomethane already 3.2.2. Methylation by the Methoxide Species. Because of the coordinated to the acidic site, as discussed above. It is clear limited size of the cluster model employed in this study, the that the energetically dominating part of any adsorption on the further progress of the dissociative reaction mechanism has to cluster involves coordination to the acidic site. proceed via the displacement of the hydrogen halide into the The transition states for methylation from the methoxide gas phase and adsorption of the hydrocarbon reactant onto the species described here closely resemble that reported for cluster with the methoxide group. As depicted in Figure 4, for methylation of benzene, also by a methoxide species.31,33,34 The ethene and propene, the hydrocarbon is partly coordinated to methoxide leaves the zeolite oxygen as a methyl cation, which the cluster itself, as well as to the methoxide group. The CmHm is moving toward the hydrocarbon. Inversion of the methyl axis is clearly pointing directly at the alkene double bond, and cation has not fully progressed to the planar inversion point. In the CmC1C2Hm dihedral angle is nearly zero, showing that the a manner similar to the associative methylation described above, four atoms are in the same plane. For toluene, the adsorption the methyl cation is coordinated to the center of the alkene mode is similar, but the interaction between the methoxide double bonds. For ethene, the progress of the reaction is again hydrogen (Hm) and the π-system is less obvious. This is the expected to involve formation of a protonated cyclopropane, same arrangement as reported for benzene adsorption on a which is isomerized into a secondary propyl cation. The order cluster with a methoxide group.31,33,34 A second, distinctively of reactivity among the hydrocarbons is the same as that for different adsorption mode also exists, where the hydrocarbon the associative step, and the same argumentation can be used is located directly above the methoxide group, and the π-elec- to rationalize the results. All calculated barriers are, however, trons of the hydrocarbons are coordinated to Cm and not to Hm. significantly higher than those in the associative case. For the alkenes, the Oz1C1C1Cm dihedral angles are then very At the end point of the reaction pathway for the dissociative nearly zero. Energetically, these two possibilities are indistin- methylation, the hydrocarbon with one more carbon atom than guishable, and it turns out that adding the ZPE corrections to the initial reactant is adsorbed onto the acidic site, as displayed the electronic energies (B3LYP_ECP/6-31G(d)) in one case in Figure 4. These stationary points are basically identical to reverses the order of stability of the two arrangements. The those described for the adsorbed products after associative adsorption mode presented here has been selected because it to methylation, except that no hydrogen halide is present. the greatest degree resembles previously published results on 3.3. Implications for Catalysis and Comparison with similar reactions.31,33,34 Selecting the on-top starting point for Methanol Conversion. Figure 5 displays how the results from the reactions would not result in any change in the activation the two steps of the dissociative mechanism can be connected or reaction energies presented here, nor would it affect any to describe the entire reaction path. We have studied nine predictions on relative reactivities. different total reactions; each hydrocarbon can react with each The calculated adsorption energies (column 1 of Table 5) halomethane. It is very clear that the associative pathway is are very small when the B3LYP functional is used, but the favored over the dissociative one. Both barriers of the dissocia- predicted strength of the interaction increases considerably when tive mechanism are higher than the single barrier of the Conversion of Halomethanes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 22, 2003 5259 associative pathway. Xia et al.49 have investigated the mecha- transition state is greater for methanol. Therefore, it seems nism of chloromethane conversion over H-ZSM-5 using in situ reasonable to assume that the methanol activation energies are FTIR spectroscopy, and they proposed a mechanism involving falsely shifted upward relative to those for the halomethanes. surface methoxide and alkoxide groups as important intermedi- In conclusion, a comparison of the absolute barrier heights for ates. Methoxide formation from halomethanes on acidic zeolites the two types of reactants based on the cluster approach is not has also been observed experimentally by others.19,24,50-52 The necessarily valid and may lead to wrong suppositions, as the predicted barriers for both formation and further reactions of mechanisms are qualitatively different. Indeed, experimental the methoxide group presented here are certainly not prohibi- work indicates a lower activity for chloromethane conversion tively high and therefore do not definitively rule out reactions over acidic zeolite catalysts relative to methanol conversion at involving such species, but the quantum chemical analysis comparable conditions.16,21 strongly favors the associative mechanism. We speculate that As mentioned above, Vos et al.34 have reported an activation this pathway is less prone to detection by spectroscopy than barrier of 195 kJ/mol for methylation of benzene by a framework surface methoxides, which, once formed, might have consider- methoxide group, which is higher than the barrier for toluene able lifetimes within the catalyst pores. Also, most such reported here (171 kJ/mol). This is consistent with theoretical spectroscopic studies have focused on the initial stages of work by Arstad et al.,36 who have shown that adding methyl halomethane conversion,19,24,49-52 when the concentrations of groups to a benzene ring causes the reactivity toward methy- hydrocarbons available for the associative mechanism are low. lation by methanol, that is, electrophilic aromatic substitution, Similar to what is seen in methanol conversion,6,7 an induction to increase. period before maximum activity is reached has also been observed for halomethane conversion.53,54 It may well be that 4. Conclusions the methoxide groups play a part in the initial buildup of Methylation of ethene, propene, and toluene by chlo- hydrocarbons inside the zeolite pores (i.e. the hydrocarbon pool) romethane, bromomethane, and iodomethane has been inves- and that associative methylation prevails once steady-state tigated theoretically using the cluster approach. Both the conversion has been reached. associative mechanism and the dissociative mechanism have The above conclusion differs from a similar theoretical been studied. Both barriers of the dissociative pathway are analysis of the methylation of benzene by methanol via both higher than the single barrier of associative methylation, and the associative and dissociative mechanisms presented by Vos the associative mechanism will be favored. This result differs et al.34 They found that formation of the framework bound from a similar comparison of the two mechanisms employing methoxide resulted in a fairly high activation barrier of 220 kJ/ methanol as the alkylating agent.34 It was then found that the mol but that the barrier for the second step of the dissociative methoxide groups, once formed, are more reactive than the pathway had a lower barrier, 186 kJ/mol, than the single barrier original reactant.34 Only small reactivity differences were found of the associative pathway, 195 kJ/mol. Further, Vos et al.34 among the halomethanes, indicating that iodomethane is more presented a kinetic analysis of the results, which indicated that reactive than bromomethane, which in turn is more reactive than the two-step mechanism was competitive with the single-step chloromethane. Although this is in agreement with experimental route at elevated temperatures (400 K). It is interesting to note results employing alkali-exchanged zeolites,24 the differences that when halomethanes are precursors for the methoxide group, in barriers are so small that a definitive conclusion may not be the order is reversed relative to methanol, and the precursors warranted. Single-point energies have been calculated at various are more reactive toward hydrocarbons than the methoxides are. levels of theory, and the trends presented are independent of The transition states for methoxide formation from halom- methodology. ethanes presented here are significantly different from the transition state for methoxide formation from methanol described Acknowledgment. Thanks are due to the Norwegian Re- 34 search Council for a grant of computer time through the NOTUR by Vos et al., where the OCmOz2 angle (corresponding to the project (accounts NN2878K and NN2147K). XCmOz2 angle here) is close to 180°, in contrast to angles close to 150° found here. When methanol is the reactant and the 4T cluster is employed, it seems that the energy gained by arranging References and Notes the three atoms involved in the normal mode of the transition (1) British Petroleum statistical review of world energy June 2002. state in a linear fashion outweighs what could be gained by http://www.bp.com/centres/energy2002/index.asp (accessed Dec. 2002). V associating one of the protons on the methanol oxygen atom to (2) See, for instance: Natural Gas Con ersion VI; Iglesia, E., Spivey, J. J., Fleisch, T. H., Eds.; Stud. Surf. Sci. Catal. 136; Elsevier Science: a cluster oxygen atom. A comparable and equally important Amsterdam, 2001. difference between the methanol32,34,36,37 and halomethane cases (3) Kochloefl, K. Methane Steam Reforming. In Handbook of Het- can be found when comparing the transition states for direct erogeneous Catalysis, Vol 4; Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; VCH Verlagsgesellschaft mbH: Weinheim, Germany, 1997; pp 1819- methylation. When methanol is the methylating agent, there is 1831. a complete protonation in the transition state, whereas this is (4) Ponec, V. Carbon Monoxide and Carbon Dioxide Hydrogenation. not the case for the halomethanes. This has to be kept in mind In Handbook of Heterogeneous Catalysis, Vol 4; Ertl, G., Kno¨zinger, H., when comparing the barriers for associative methylation by Weitkamp, J., Eds.; VCH Verlagsgesellschaft mbH: Weinheim, Germany, 1997; pp 1876-1894. 37 methanol, which have been found to be 183 kJ/mol (ethene), (5) Hansen, J. B. Methanol Synthesis. In Handbook of Heterogeneous 169 kJ/mol (propene),37 and 187 kJ/mol (toluene),36 at an Catalysis, Vol 4; Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; VCH identical level of theory and using the same cluster as in the Verlagsgesellschaft mbH: Weinheim, Germany, 1997; pp 1856-1876. (6) Sto¨cker, M. Microporous Mesoporous Mater. 1999, 29,3-48. present work. As outlined above, the acidity of the cluster model (7) Chang, C. D. The Methanol-to-Hydrocarbons Reaction: A Mecha- is crucial in determining barrier height, and we have previously nistic Perspective. In Shape SelectiVe Catalysis; Song, C., Garce´s,J.M., argued that the cluster employed does not fully mimic the acidity Sugi, Y., Eds.; ACS Symposium Series 738; American Chemical Society: - of a real zeolite.37 This deficiency of the cluster approach affects Washington, DC, 2000; pp 96 114. (8) (a) Noceti, R. P.; Taylor, C. E. U.S. Patent 4,769,504, Sept. 6, 1998. the barriers for methanol reactions more severely than those (b) Taylor, C. E.; Noceti, R. P. U.S. Patent 5,019,652, May 28, 1991. (c) for halomethanes, because the degree of protonation in the Taylor, C. E.; Noceti, R. P. U.S. Patent 5,139,991, Aug. 18, 1992. 5260 J. Phys. Chem. B, Vol. 107, No. 22, 2003 Svelle et al.

(9) Olah, G. A.; Gupta, B.; Farina, M.; Felberg, J. D.; Ip, W. M.; (34) Vos, A. M.; Nulens, K. H. L.; De Proft, F.; Schoonheydt, R. A.; Husain, A.; Karpeles, R.; Lammertsma, K.; Melhotra, A. K.; Trivedi, N. J. Geerlings, P. J. Phys. Chem. B 2002, 106, 2026-2034. J. Am. Chem. Soc. 1985, 107, 7097-7105. (35) Rozanska, X.; Saintigny, X.; van Santen, R. A.; Cle´mendot, S.; (10) Butter, A. A.; Jurewicz, A. T.; Kaeding, W. W. U.S. Patent 3,- Hutschka, F. J. Catal. 2002, 208,89-99. 894,407, July 8, 1975. (36) Arstad, B.; Kolboe, S.; Swang, O. J. Phys. Chem. B 2002, 106, (11) Kaiser, S. W. International Patent WO 86/04577, Aug. 14, 1986. 12722-12726. (12) Romannikov, V. N.; Ione, K. G. Kinet. Catal. 1984, 25,75-80. (37) Svelle, S.; Arstad, B.; Kolboe, S.; Swang, O. J. Phys. Chem. B, (13) Taylor, C. E.; Noceti, R. P.; Schehl, R. R. Direct Conversion of submitted for publication. Methane to Liquid Hydrocarbons Through Chlorocarbon Interdiates. In (38) (a) Dessau, R. M.; LaPierre, R. B. J. Catal. 1982, 78, 136-141. Methane ConVersion; Bibby, D. M., Chang, C. D., Howe, R. F., Yurchak, (b) Kolboe, S. Acta Chem. Scand. 1986, A40, 711-713. (c) Dessau, R. M. S., Eds.; Stud. Surf. Sci. Catal. 36; Elsevier Science: Amsterdam, 1988; J. Catal. 1986, 99, 111-116. (d) Song, W.; Marcus, D. M.; Fu, H.; pp 483-489. Ehresmann, J. O.; Haw, J. F. J. Am. Chem. Soc. 2002, 124, 3844-3845. (14) Jens, K.-J.; Halvorsen, S.; Bauman Ofstad, E. Methane Conversion (39) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, M. A.; Via Methyl chloride: Condensation of Methyl chloride to Light Hydro- Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; carbons. In Methane ConVersion; Bibby, D. M., Chang, C. D., Howe, R. Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. F., Yurchak, S., Eds.; Stud. Surf. Sci. Catal. 36; Elsevier Science: D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, Amsterdam, 1988; pp 491-495. M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; (15) Anderson, J. R. Appl. Catal. 1989, 47, 177-196. Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, (16) Lersch, P.; Bandermann, F. Katalyse 1989, 118, 183-201. D. K.; Rabuck, D. K.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; (17) Lersch, P.; Bandermann, F. Appl. Catal. 1991, 75, 133-152. Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, (18) White, C. M.; Douglas, L. J.; Hackett, J. P.; Anderson, R. R. Energy I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Fuels 1992, 6,76-82. Peng, C. Y.; Nanayakkara, A.; Gonzales, C.; Challacombe, M.; Gill, P. M. (19) Sun, Y.; Cambell, S. M.; Lunsford, J. H.; Lewis, G. E.; Palke, D.; W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, Tau, L.-M. J. Catal. 1993, 143,32-44. M.; Replogle, E. S.; Pople, J. A. Gaussian 98 (reVision A.11.); Gaussian, (20) Chich, B. ReV. Roum. Chim. 1997, 42, 1165-1170. Inc.: Pittsburgh, PA, 1998. (21) Taylor, C. E. Conversion of Substituted Methanes over ZSM- (40) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 284-298. Catalysts. In Proceedings of the 12th International Congress on Catalysis; (41) Check, C. E.; Faust, T. O.; Bailey, J. M.; Wright, B. J.; Gilbert, T. Corma, A., Melo, F. V., Mendioroz, S., Fierro, J. L. G., Eds.; Stud. Surf. M.; Sunderlin, L. S. J. Phys. Chem. A 2001, 105, 8111-8116. Sci. Catal. 130D; Elsevier Science: Amsterdam, 2000; pp 3633-3638. (42) Glukhovtsev, M. N.; Pross, A.; McGrath, M. P.; Radom, L. J. Chem. (22) Jaumain, S.; Su, B.-L. Direct Catalytic Conversion of Chlo- Phys. 1995, 103, 1878-1885. romethane to Higher Hydrocarbons over Various Protonic and Cationic (43) Basis sets were obtained from the Extensible Computational Zeolite Catalysts as Studied by in-situ FTIR and Catalytic Testing. In Chemistry Environment Basis Set Database, Version 10/29/02, as developed Proceedings of the 12th International Congress on Catalysis; Corma, A., and distributed by the Molecular Science Computing Facility, Environmental Melo, F. V., Mendioroz, S., Fierro, J. L. G., Eds.; Stud. Surf. Sci. Catal. and Molecular Sciences Laboratory, which is part of the Pacific Northwest 130B; Elsevier Science: Amsterdam, 2000; pp 1607-1612. Laboratory, P.O. Box 999, Richland, WA 99352, USA, and funded by the (23) Brophy, J. H.; Font Freide, J. J. H. M.; Tomkinson, J. D. U.S. Department of Energy. The Pacific Northwest Laboratory is a International Patent WO 85/02608, June 20, 1985. multiprogram laboratory operated by Battelle Memorial Institute for (24) Murray, D. K.; Chang, J.-W.; Haw, J. F. J. Am. Chem. Soc. 1993, the U.S. Department of Energy under Contract DE-AC06-76RLO 1830. 115, 4732-4741. Contact David Feller or Karen Schuchardt for further information. (25) (a) Dahl, I. M.; Kolboe, S. Catal. Lett. 1993, 20, 329-336. (b) http://www.emsl.pnl.gov:2080/forms/basisform.html (accessed Oct. 2002). Dahl, I. M.; Kolboe, S. J. Catal. 1994, 149, 458-464. (c) Dahl, I. M.; (44) Thermodynamic data were taken from the NIST database; http:// Kolboe, S. J. Catal. 1996, 161, 304-309. webbook.nist.gov/chemistry (accessed July 2002). (26) (a) Mikkelsen, Ø.; Rønning, P. O.; Kolboe, S. Microporous (45) Su, B.-L.; Jaumain, D. Chloromethane as a Probe Molecule to Mesoporous Mater. 2000, 40,95-113. (b) Arstad, B.; Kolboe, S. Catal. Characterize the Bro¨nsted Acidity of Zeolites: An in-situ FTIR Study. In Lett. 2001, 71, 209-212. (c) Arstad, B.; Kolboe, S. J. Am. Chem. Soc. Proceedings of the 12th International Zeolite Conference; Treacy, M. M. 2001, 123, 8137-8138. (d) Bjørgen. M.; Olsbye, U.; Kolboe, S. J. Catal., J., Marcus, B. K., Bisher, M. E., Higgins, J. B., Eds.; Materials Research in press. Society: Warrendale, PA, 1998; Vol. 4, pp 2689-2696. (27) (a) Haw, J. F.; Nicholas, J. B.; Song, W.; Deng, F.; Wang, Z.; Xu, (46) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders, V. R. Chem. ReV. T.; Heneghan, C. S. J. Am. Chem. Soc. 2000, 122, 4763-4775. (b) Goguen, 1994, 94, 2095-2160. P. W.; Xu, T.; Barich, D. H.; Skloss, T. W.; Song, W.; Wang, Z.; Nicholas, (47) Hammond, G. S. J. Am. Chem. Soc. 1955, 77, 334-338. J. B.; Haw, J. F. J. Am. Chem. Soc. 1998, 120, 2650-2651. (c) Sassi, A.; (48) The experimental proton affinities of chloride, bromide, and iodide Wildman, M. A.; Ahn, H. J.; Prasad, P.; Nicholas, J. B.; Haw, J. F. J. Phys. are 1395, 1353, and 1315 kJ/mol.44 Calculations at the B3LYP_ECP/6- Chem. B 2002, 106, 2294-2303. (d) Song, W.; Haw, J. F.; Nicholas, J. B.; 31G(d) + ZPE level of theory yield 1356, 1322, and 1294 kJ/mol. Heneghan, C. S. J. Am. Chem. Soc. 2000, 122, 10726-10727. (e) Song, (49) Xia, X.-R.; Bi, Y.-L.; Wu, T.-H.; Zhen, K.-J. Catal. Lett. 1995, W.; Fu, H.; Haw, J. F. J. Phys. Chem. B 2001, 105, 12839-12843. 33,75-90. (28) (a) Seiler, M.; Schenk, U.; Hunger, M. Catal. Lett. 1999, 62, 139- (50) Ko´nya, Z.; Hannus, I.; Kiricsi, I. Appl. Catal. B 1996, 8, 391- 145. (b) Hunger, M.; Seiler, M.; Buchholz, A. Catal. Lett. 2001, 74,61-68. 404. (29) Alkylation of propene by chloromethane has been investigated (51) Su, B.-L.; Jaumain, D.; Ngalula, K.; Briend, M. Effect of Acido- experimentally by Kaeding and Butter. See: Kaeding, W. W.; Butter, S. Basicity of Beta Zeolites on the Conversion of Chloromethane to Hydro- A. U.S. Patent 3,906,054, Sept. 16, 1975. carbons as Studied by FTIR and TPD-MS. In Proceedings of the 12th (30) Frash, M. V.; van Santen, R. A. Top. Catal. 1999, 9, 191-205. International Zeolite Conference; Treacy, M. M. J., Marcus, B. K., Bisher, (31) Rozanska, X.; Saintigny, X.; van Santen, R. A.; Hutschka, F. J. M. E., Higgins, J. B., Eds.; Materials Research Society: Warrendale, PA, Catal. 2001, 202, 141-155. 1998; Vol. 4, pp 2681-2688. (32) Vos, A. M.; Rozanska, X.; Schoonheydt, R. A.; van Santen, R. A.; (52) Bosa´cˇek, V. J. Phys. Chem. 1993, 97, 10732-10737. Hutschka, F.; Hafner, J. J. Am. Chem. Soc. 2001, 123, 2799-2809. (53) Levik, O. I. Thesis, Norwegian Institute of Technology (now (33) Rozanska, X.; van Santen, R. A.; Hutschka, F.; Hafner, J. J. Am. Norwegian University of Science and Technology), 1990. Chem. Soc. 2001, 123, 7655-7667. (54) Olsbye, U.; Dahl, I. M. To be submitted for publication. Paper II

A Theoretical Investigation of the Methylation of Alkenes with Methanol over Acidic Zeolites

S. Svelle, B. Arstad, S. Kolboe, O. Swang, J. Phys. Chem. B 107 (2003) 9281-9289

J. Phys. Chem. B 2003, 107, 9281-9289 9281

A Theoretical Investigation of the Methylation of Alkenes with Methanol over Acidic Zeolites

Stian Svelle,† Bjørnar Arstad,† Stein Kolboe,*,† and Ole Swang‡ Department of Chemistry, UniVersity of Oslo, P.O. Box 1033 Blindern, N-0315 Oslo, Norway, and Department of Hydrocarbon Process Chemistry, SINTEF Applied Chemistry, P.O. Box 124 Blindern, N-0134 Oslo, Norway ReceiVed: October 8, 2002

A quantum chemical investigation of the methylation of alkenes by methanol over acidic zeolite catalysts has been performed. A cluster model consisting of four tetrahedrally coordinated atoms (T-atoms) has been used to represent the zeolite. The activation barrier for methylation is dependent on the size and substitution pattern of the alkene reactant. The results can be rationalized by considering the relative energies of the carbocations that describe the reaction path. Implications for methanol-to-hydrocarbon chemistry are discussed. The effect of zeolite acidity on the activation barrier has also been probed. The barrier decreases when the catalyst acidity is increased.

1. Introduction coordinated atoms (T-atoms), 19 atoms total) as a model catalyst. Vos et al.21,22 considered the methylation of benzene and toluene Since the discovery of zeolite-catalyzed conversion of to form the next-higher homologue. We have previously methanol to hydrocarbons (MTH), a great research effort has analyzed the reactivity of toluene, 1,2,4,5-tetramethylbenzene, 1,2 been made to resolve mechanistic issues. Initially, focus was and hexamethylbenzene using quantum chemistry.23 The results - directed at the formation of initial C C bonds and whether indicated that any methylbenzene inside a zeolite pore in the ethene or propene was the initial alkene formed. This reaction presence of methanol will be completely methylated to form 3-5 step has, however, been claimed to be of little interest, or the heptamethylbenzenium ion (if sufficient space is available), 6 even insignificant, because it is clear that, once hydrocarbons because the methylation activation barrier decreases as the are formed, a completely different mechanistic scheme is valid. number of methyl groups increases. To our best knowledge, no During the first years of research, a scheme was proposed theoretical work on the methylation of alkenes by methanol over where initially formed lower alkenes are methylated to form zeolites exists in the literature, nor does there seem to be any - higher homologues, which, in turn, are cracked into mainly C2 proper compilation of experimental kinetic data available for C4 alkenes. These either desorb as products or are methylated these reactions. again. Aromatics were considered to be end products of In this work, we have investigated the methylation of seven cyclization and dehydrogenation steps and believed to be coke different alkenes, from ethene to 2-methyl-2-butene, using DFT. 3,5,7,8 precursors. The aforementioned 19-atom cluster has been used to model A second, more recent scheme, which is termed the “hydro- the catalyst. The results have direct bearing on MTH chemistry, - carbon pool” mechanism,9 11 also exists. Initially, a mechanistic because it allows an evaluation of parts of the initially proposed cycle that is based on continuous alkylations of unspecified, mechanistic scheme. Together with previously published results, hydrogen-poor adsorbed species, which could, in turn, split off a direct comparison of the intrinsic reactivity of alkenes versus low alkenes, was suggested. Experiments where benzene or aromatics is now possible. This information is difficult to obtain toluene and 13C-methanol were co-reacted have been performed, experimentally. The findings presented here are also relevant and it was clear that ethene, propene, and the arenes in the for the formation of aromatic compounds in MTH systems, effluent had indistinguishable isotopic distributions.12 This because this, by necessity, must proceed via the formation of strongly indicated that methanol conversion could proceed via relatively long-chained alkenes. Furthermore, we present an repeated methylations and dealkylations of aromatic reaction evaluation of the effect of cluster acidity on activation barriers. centers. Several later studies have confirmed and clarified this This important point lies at the very center of acid catalysis hypothesis, and it seems that methylbenzenes, methylnaphtha- research. lenes, and certain methylated cyclic carbenium ions can constitute the hydrocarbon pool.13-20 There may also be a facile 2. Computational Details interconversion of these hydrocarbon pool species. All calculations were done with the Gaussian98 program Very recently, theoretical work that is relevant to the MTH package.24 All geometries were optimized using the B3LYP reaction mechanism has been published. Activation energies for hybrid density functional and 6-31G* basis sets. No geometric important reaction steps have been investigated using density constraints were used. In addition, single-point energies were functional theory (DFT) and a fairly small cluster (4 tetragonally calculated for these geometries, using the B3LYP/6-311G** and MP2/6-31G* levels of theory. A few single-point energy * Author to whom correspondence should be addressed. E-mail: calculations were also performed using the MP2/6-311G** level [email protected]. † University of Oslo. of theory. The zeolite catalyst has been modeled using a cluster ‡ SINTEF Applied Chemistry. consisting of four T-atoms, i.e., three Si atoms and one Al atom, 10.1021/jp022201q CCC: $25.00 © 2003 American Chemical Society Published on Web 08/12/2003 9282 J. Phys. Chem. B, Vol. 107, No. 35, 2003 Svelle et al.

TABLE 1: Products Formed in the Methylation Reactions Investigated object of methylation investigated products ethene propene, cyclopropane propene trans-2-butene 1-butene trans-2-pentene trans-2-butene 2-methyl-2-butene cis-2-butene 2-methyl-2-butene isobutene 2-methyl-2-butene 2-methyl-2-butene 2,3-dimethyl-2-butene to generate the acidic site.23 This cluster has been used previously to model similar reactions.21-23 To reduce the effect of using a finite cluster model, care has been taken to ensure that the reactants and products were coordinated to the cluster in a similar manner for all reactions. For all stationary points, Figure 1. Stationary points on the reaction path for the methylation vibrational spectra were calculated to ensure that the correct of trans-2-butene. number of imaginary frequencies was at hand, i.e., one imaginary frequency for transition states and zero for energy cluster. Methanol is adsorbed end-on onto the acidic site, minima. For the transition states, the normal modes correspond- forming two hydrogen bonds. This is the same adsorption mode ing to the imaginary frequencies were visualized, to confirm as described earlier by several workers.21,22,27-29 The methyl that they indeed corresponded to the expected motion of atoms. group of methanol does not point directly toward the alkene Intrinsic reaction coordinate (IRC) calculations were, in some double bond; rather, the alkene is located next to the methanol cases, performed to investigate the minima connected by the molecule. Different reactant positions, with respect to alkene transition states. The transition states were also investigated by methyl groups and also the relative positioning of the two perturbing the geometries very slightly along the reaction reactants, were investigated. All energy differences among the coordinate, corresponding to the negative eigenvalue in the stationary points thus found were very small, i.e., <2 kJ/mol. Hessian, and using the geometries thus produced as starting The reactant geometries were chosen to be as qualitatively points for energy minimizations. This quasi-IRC approach similar as possible. Upon coadsorption of methanol and an appeared to be a more robust procedure than the IRC function- alkene on the cluster, the Oz1Hz distance is considerably ality in Gaussian98 and gave qualitatively identical results as stretched, from 0.974 Å to 1.03 Å. The acidic proton is located IRC calculations, but at considerably smaller computational cost. at a distance of 1.54-1.55 Å from the methanol O atom, and the Oz2Hm4 distance is 1.83 Å. These values are essentially 3. Results and Discussion identical to those reported for the coadsorption of methanol and The methylation of ethene, propene, 1-butene, trans-2-butene, arenes, using the B3LYP functional,22,23 and are slightly longer cis-2-butene, isobutene, and 2-methyl-2-butene has been inves- (∼0.05 Å) than the values calculated using the MPWPW91 tigated on a 4 T-atom cluster model. Adsorbed reactants, functional.21 transition states, protonated products, and final products were The coadsorption energies (the first column in Table 2) are optimized as described above. For most of the reactions, the essentially the same for all alkenes investigated. The absolute carbenium ion intermediate (see discussion below) could be values are 5-15 kJ/mol smaller than those previously reported deprotonated to yield several different product alkenes. In such for the coadsorption of methanol and methylarenes.23 The cases, the thermodynamically most stable alkene was chosen calculated adsorption energy is larger when the zero-point as the product. The products thus selected are listed in Table 1. energy is excluded (the B3LYP/6-31G* electronic energy). Such The reaction mechanism described here is associative, and it a difference can be attributed to the very different vibrational is similar to that reported for the methylation of aromatic modes of the separated, gas-phase reactants and the reactants compounds.21-23 Theoretical studies on the methylation of in the coadsorbed state. There is, furthermore, a substantial arenes indicate that the associative mechanism is at least as likely increase in the calculated adsorption energy when ab initio MP2 as a dissociative pathway, because the first step of the methodology (MP2/6-31G*//B3LYP/6-31G*) is used rather than dissociative mechanism (i.e., methoxide formation) was found DFT. This is expected, because it is well-known that DFT is to have a higher activation energy than single-step, associative ill-suited for accurate descriptions of long-range van der Waals methylation.22 This result also applies to alkene methylation. interactions.30,31 There is a decrease in adsorption energy (∼10 Also, experimental results favor the associative mechanism for kJ/mol) when going from the B3LYP/6-31G* level of theory this related reaction.25 We have not investigated the dissociative (without ZPE correction) to the B3LYP/6-311G**// B3LYP/6- pathway. 31G* level of theory. An increase in the basis set size used in Figure 1 shows the stationary points on the reaction pathway the MP2 single-point energy calculations, from 6-31G* to for the methylation of trans-2-butene and is representative of 6-311G**, has little effect on the adsorption energies. the other reactions. Table 2 lists the energies of the stationary 3.2. Transition States. In the transition state, methanol is points for all reactions considered, calculated gas-phase reaction protonated and the O-C bond of methanol is significantly energies, and calculated reaction enthalpies. Experimental stretched. A formal methyl cation leaves the methanol and + reaction enthalpies are included for comparison. When energies moves toward the alkene double bond. This CH3 group is are discussed, we refer to the B3LYP/6-31G* + ZPE values, almost planar; the H atoms have just moved past the inversion unless otherwise stated. Important geometric parameters are point. The degree of inversion is indicated by the Hm1Hm2Hm3Cm listed in Table 3. The atom labels are defined in Figures 2 and dihedral angles in Table 3. The methyl cation is, in every case, 3. only slightly above the plane defined by Om and the alkene 3.1. Reactants and Adsorption Energies. The starting point double bond. This is clear from the OmClessCmoreCm dihedral for all reactions is methanol and an alkene coadsorbed onto the angles. The imaginary frequencies of the transition states are Methylation of Alkenes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 35, 2003 9283

TABLE 2: Energies of the Stationary Points for Methylation Reactions and Gas-Phase Reaction Energiesa energy (kJ/mol) gas-phase reaction object of gas-phase transition protonated adsorbed gas-phase gas-phase enthalpy (kJ/mol) methylation reactants state alkene products products reaction energy calculated experimentb B3LYP/6-31G* + ZPE ethenec 87 183 N/Ad -56 45 -42 -41 -73 ethenee -11 82 -6 -6 -40 propene 88 169 114 -53 46 -42 -39 -72 1-butene 88 168 112 -53 46 -42 -39 -72 trans-2-butene 88 162 106 -46 53 -34 -32 -75 cis-2-butene 87 161 117 -53 47 -41 -38 -73 isobutene 88 156 N/Ad -45 55 -34 -31 -65 2-methyl-2-butene 87 154 97 -40 61 -26 -23 -70 B3LYP/6-31G* ethenec 98 184 N/Ad -56 60 -38 ethenee -16 92 -6 propene 98 170 115 -53 70 -36 1-butene 99 170 114 -53 62 -37 trans-2-butene 98 164 108 -46 68 -29 cis-2-butene 98 164 120 -51 63 -35 isobutene 99 159 N/Ad -44 70 -28 2-methyl-2-butene 97 156 100 -39 77 -20 B3LYP/6-311G**//B3LYP/6-31G* ethenec 89 184 N/Ad -68 36 -53 ethenee -26 73 -16 propene 88 171 114 -67 38 -50 1-butene 88 171 114 -65 39 -49 trans-2-butene 88 165 104 -58 45 -43 cis-2-butene 89 163 116 -63 41 -48 isobutene 88 158 N/Ad -57 46 -42 2-methyl-2-butene 88 155 95 -53 54 -34 MP2/6-31G*//B3LYP/6-31G* ethenec 111 (105)f 184 (186) N/Ad -65 (-77) 55 (35) -56 (-70) ethenee -43 73 -38 propene 114 172 95 -63 59 -55 1-butene 112 168 90 -70 56 -56 trans-2-butene 117 162 83 -62 65 -52 cis-2-butene 109 157 81 -76 51 -58 isobutene 116 164 N/Ad -59 68 -48 2-methyl-2-butene 110 (106) 147 (148) 73 (72) -65 (-82) 66 (46) -44 (-60) a All values relative to the adsorbed reactants. Enthalpies calculated at 298 K. b Experimental reaction enthalpies (298 K) from ref 26. c Product formed is propene. d All attempts to optimize these structures failed. e Product formed is cyclopropane. Gas-phase reactants, adsorbed reactants, and transition state are the same as when propene is the product formed. f Numbers in parentheses have been calculated at the MP2/6-311G**// B3LYP/6-31G* level of theory.

Figure 2. Schematic representation of the transition state for methylation of alkenes. Same atom labels apply for the adsorbed reactants. Cless is the less-substituted carbon atom, whereas Cmore is the more- substituted carbon atom. Figure 3. Schematic representation of the adsorbed products. Cless is the less-substituted carbon atom, whereas Cmore is the more-substituted in the same range as those reported for similar studies on the carbon atom. methylation of arenes,22 i.e., ∼350i cm-1. The manner in which the methyl cation is coordinated to the to find a substantially unsymmetrical transition state; however, alkene double bond in the transition state is of considerable these efforts either failed or resulted in a shift of geometry interest. For the methylation of ethene, there is a very nearly toward the nearly symmetrical arrangement presented above. symmetrical arrangement, and the CmCless and CmCmore distances The transition state was then investigated further by performing are identical (∆R ) 0.003 Å; see Figure 2 for atom label IRC calculations and employing the previously outlined quasi- definitions). This indicates that a protonated cyclopropane IRC procedure. These calculations did indeed proceed via species is being formed, rather than a primary propyl cation, geometries that were very similar to that of protonated cyclo- which is the expected intermediate if the methyl cation is propane. However, no stationary point with such a geometry attached to only one of the ethene C atoms. Efforts were made was found, and the adsorbate resembling protonated cyclopro- 9284 J. Phys. Chem. B, Vol. 107, No. 35, 2003 Svelle et al.

TABLE 3: Geometric Parameters (Distances, Angles, and Dihedral Angles) )a )b ) ) ) ) ) ) C2 C2 C3 1-C4 t-2-C4 c-2-C4 i-C4 C5 Reactantsc distance (Å) d Oz1Hz 1.03 1.03 1.03 1.03 1.03 1.03 1.03 Oz2Hm4 1.83 1.83 1.83 1.83 1.83 1.83 1.83 HzOm 1.54 1.54 1.55 1.54 1.55 1.54 1.55 Hm4Om 0.99 0.99 0.99 0.99 0.99 0.99 0.99 OmCm 1.44 1.44 1.43 1.44 1.43 1.44 1.43 CmCless 4.06 3.94 4.11 3.93 4.10 3.91 4.07 CmCmore 4.18 4.20 4.14 4.10 4.12 4.23 4.10 ClessCmore 1.33 1.34 1.34 1.34 1.34 1.34 1.34 CmAl 4.42 4.43 4.38 4.44 4.34 4.42 4.39 angle (deg) Hm1Hm2Hm3Cm 35.7 35.8 35.9 35.8 35.9 16.7 35.9 OmClessCmoreCm 15.7 16.2 13.5 15.8 12.8 35.8 8.9 Transition Statesc distance (Å) d Oz1Hz 1.88 1.85 1.85 1.86 1.81 1.82 1.84 Oz2Hm4 1.93 1.92 1.91 1.89 1.97 1.88 1.87 HzOm 0.99 0.99 0.99 0.99 0.99 0.99 0.99 Hm4Om 0.98 0.98 0.98 0.99 0.98 0.99 0.99 OmCm 2.27 2.23 2.23 2.21 2.21 2.17 2.17 CmCless 2.14 2.07 2.07 2.19 2.21 2.08 2.14 CmCmore 2.14 2.35 2.35 2.23 2.22 2.50 2.38 ClessCmore 1.36 1.36 1.37 1.37 1.37 1.37 1.37 CmAl 3.50 3.49 3.50 3.50 3.50 3.52 3.51 angle (deg) Hm1Hm2Hm3Cm -14.8 -12.9 -12.7 -11.6 11.3 -10.1 -10.0 OmClessCmoreCm 2.2 3.0 3.0 2.2 0.9 2.3 1.7 Productse distance (Å) ff Oz1Hz 1.04 1.03 1.04 1.04 1.04 1.04 Oz2Hw4 1.85 1.87 1.88 1.89 1.89 1.88 HzOm 1.53 1.55 1.53 1.53 1.53 1.52 Hw1Om 0.99 0.99 0.99 0.99 0.99 0.99 Hw2Cw 0.98 0.98 0.98 0.98 0.98 0.98 g Hw2Cless 2.33 2.34 2.29 2.41 2.27 2.32 h Hw2Cmore 2.47 2.34 2.36 2.33 2.43 2.24 i ClessCmore 1.34 1.53 1.34 1.34 1.35 1.36 angle (deg) OmCmoreClessHw2 1.1 3.7 0.2 4.3 3.6 0.2 a Product formed is propene. b Product formed is cyclopropane. c Atom labels are defined in Figure 2. d The starting point and transition state are the same as those for the formation of propene. e Atom labels are defined in Figure 3. f The same product (2-methyl-2-butene) is formed as that for the methylation of trans-2-butene. g Distance to the methyl-substituted C atom of the double bond. h Distance to the ethyl-substituted C atom of the double bond. i C-C single bond in cyclopropane. pane was immediately deprotonated to form cyclopropane. for the isomerization of 108-109 s-1 at 600 K. The predicted Cyclopropane is not observed as a product under regular MTH ring opening should thus occur readily under normal reaction conditions,1 nor is it formed when the methylation of ethene is conditions. studied experimentally at extreme feed rates.32 It must, therefore, For the other reactions considered, direct formation of be concluded that primarily formed protonated cyclopropane secondary carbenium ions (methylation of propene, 1-butene, isomerizes into an open-chained propyl cation before deproto- trans-2-butene, and cis-2-butene) or tertiary carbenium ions nation can occur. Koch et al.33,34 published a full description of (methylation of isobutene and 2-methyl-2-butene) is possible all protonated propene isomers, using the MP2/6-311G** level without hydrogen shifts. The reactions are therefore considerably of theory for geometry optimizations. They concluded that no less complicated, and a description of the reaction path is quite primary propyl cation exists as an energy minimum, and that straightforward. A bond is formed between the methyl cation the only stable species are the corner-protonated cyclopropane and the less-substituted C atom of the double bond. This is cation and the secondary propyl cation. Furthermore, they found indicated by the CmCless and CmCmore distances listed in Table that corner-protonated cyclopropane isomerizes to the secondary 3. The main charge is then located on the more-substituted C cation in a single-step reaction. From the data reported by Koch atom, Cmore. When the alkene double bond is symmetrically et al., we have calculated that the barrier is 55.6 kJ/mol (MP4- substituted with methyl groups, these distances are basically (FC)/6-311G**//MP2/6-311G** + ZPE correction). The ge- identical, as was the case for ethene. The detailed geometries ometry optimizations were redone by us, using B3LYP/6-31G* of the cationic species formed will be discussed in the next methodology. No primary propyl cation could be optimized as section. an energy minimum at our level of theory either, and the The calculated activation barriers for the methylation reactions isomerization barrier was now 41.0 kJ/mol (B3LYP/6-31G* + are listed in the second data column of Table 2. The trend ZPE correction). This is a small activation energy, compared observed can be rationalized on the basis of the expected to that of the first reaction step, and, with a typical pre- energies of the carbocations formed. According to the Hammond exponential of 1012-1013 s-1, this corresponds to a rate constant postulate, the activation energies for similar, endothermic Methylation of Alkenes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 35, 2003 9285 reactions should mimic those of the products. The methylation of ethene leads to the formation of a very unstable cyclopropane carbonium ion, and the barrier is accordingly high. The methylation of propene leads to the formation of a secondary butyl cation, and the barrier is considerably lower than that for ethene. The small difference in barrier between the methylation of propene and 1-butene can be considered to be an effect of the increased charge-stabilizing effect of an ethyl substituent, Figure 4. (a) Front and (b) side view (90°) of the adsorbed cation relative to a methyl group. The barriers for methylation of cis- formed after the methylation of propene. 2-butene and trans-2-butene are very similar and lower than that for 1-butene. The similarity is expected, because the where the optimizations failed). This was confirmed by the reactions both lead to the formation of secondary pentyl cations. vibrational spectra. The cation formed via the transition state The lower barrier, relative to that of 1-butene, can be ascribed for the methylation of propene is shown in detail in Figure 4. to the favorable effects of having substituents on both ends of Table 4 lists selected geometric parameters. The atom labels the double bond. Simple inspection of the Mulliken charges are defined in Figure 5. An extensive description of all the reveals that, in the transition state, and also in the resulting cationic species is given as Supporting Information. Whether cation, there is considerable charge on both C atoms of the cations in zeolites are transition states or energy minima in quantum chemistry calculations is a matter of some debate.35-43 double bond; this observation agrees well with chemical Kazansky and co-workers37-43 have shown that the chemisorp- intuition. These charges are best stabilized by having methyl tion of alkenes proceeds via transition states that closely groups on both C atoms. The activation energies for the resemble carbenium ions and have concluded that free carbe- methylation of isobutene and 2-methyl-2-butene are lower still, nium ions do not exist as energy minima when an alkene is because of the formation of tertiary carbenium ions and the coordinated to cluster models (in the absence of water). Vos increased stabilizing effect of the larger alkyl substituents. This and co-workers21,22 investigated the methylation of arenes along line of reasoning is strongly supported by the calculated energies the very same reaction mechanism as that in this work and did of the adsorbed cations (third data column in Table 2), which not report any charged intermediates within the cluster approach. follow the activation energies. Note that the cation formed after However, Arstad et al.23 very recently published contradictive the methylation of cis-2-butene is coordinated to the cluster in results, where coadsorbed water and protonated arenes were a manner that is different from the coordination of the other found to be energy minima, using the same cluster as Vos et cations, and the calculated energy is therefore not directly al. In that case,23 removal of the water molecule resulted in the comparable to that of the others. formation of a C-O bond between the hydrocarbon and the The geometries of the transition states are also in reasonable cluster. agreement with the Hammond postulate, because a lower barrier All cations are stabilized through hydrogen bonds with the implies a geometry that is more similar to that at the starting cluster and the water molecule. These stabilizing interactions point. The average distance from the methyl cation to the alkene are indicated by the fairly short H O and H O distances. The double bond increases, and the distances from the methanol O 2 m 1 z1 C1H1 and C1H2 bonds are correspondingly stretched. Because atom to the methyl cation tend to decrease as the barrier height of these very favorable interactions between the cations and decreases, although these trends are not perfect. The variations the cluster, water is partially displaced and only one hydrogen in the degree of inversion of the methyl cation also support bond between water and the cluster is formed. The other water this idea. hydrogen is pointing away from the cluster. This is an effect of The order of the activation energies is the same for all levels the limited cluster size and may be considered to be a cluster of theory employed, except for the increase for isobutene that artifact. In a real zeolite catalyst, there would be several zeolitic is found when using MP2/6-31G*//B3LYP/6-31G* methodol- O atoms available for the formation of hydrogen bonds. The ogy. In this case, the barrier for the methylation of isobutene is presence of water seems to be crucial for the existence of shifted above that for methylation of cis- and trans-2-butene. cationic species as energy minima in at least two ways. First, However, this does not mean that the rationalization discussed the water molecule functions as a wedge, thus separating the above, which is based on carbenium ion stability, is incorrect. cation and the cluster and preventing deprotonation. Second, It is important to bear in mind that the activation barriers the strong interaction between one water H molecule and one presented in Table 2 are calculated as the difference between zeolite O atom effectively reduces the proton affinity of the the energy of the coadsorbed reactants and the energies of the cluster, thereby also reducing the probability of deprotonation. transition states. In the case of the four butenes, the stoichiom- Jeanvoine et al.44 found a similar effect when using molecular etries are the same and a comparison of the absolute energies dynamics (MD) to study water in chabazite. They found that is possible. This analysis reveals that the absolute energy of the presence of a second water molecule was required to the transition state for the methylation of isobutene is lower facilitate water protonation and hydroxonium ion formation. than that for the other three butenes at every level of theory, Termath et al.45 and Sˇtich et al.46 also found that increasing the but it also reveals that the starting point lies lower still, when loading of molecules enhances the probability of protonation the MP2/6-31G*//B3LYP/6-31G* scheme is used, so that the when employing MD to study water and methanol in zeolites, barrier becomes greater. However, it is not possible to arrive at respectively. a decisive conclusion regarding the relative reactivities of the Formally, the positive charge is located on C3 (Figure 5), butene isomers, although the bulk of the results suggest that and the substituents on this atom are arranged in a very nearly the reactivity is greatest for isobutene. planar, trigonal fashion; thus, the C2R2R1C3 dihedral angle, in 3.3. Protonated Products. The immediate methylation every case, is almost zero. The butyl and pentyl cations products are carbenium ions and water, as outlined in the described in Table 4 have sharp C1C2C3 angles and short C1C3 preceding section. These species were optimized as energy distances, which are only slightly longer than the C1C2 distance. minima (except for the methylation of ethene and isobutene, The C2C3 distance is only ∼0.1 Å longer than the distance of 9286 J. Phys. Chem. B, Vol. 107, No. 35, 2003 Svelle et al.

TABLE 4: Geometric Parameters (Distances, Angles, and Dihedral Angles) for Adsorbed Cationic Species ) ) ) ) ) ) ) C2 C3 1-C4 t-2-C4 c-2-C4 i-C4 C5 Protonated Productsa b b R1 N/A -H -H -CH3 -CH3 N/A -CH3 R2 -CH3 -CH2CH3 -H -H -CH3 R3 -H -H -H -CH3 -CH3 c R4 -H -H -CH3 -H -H distance (Å) H1Oz1 1.92 1.91 1.89 1.95 2.06 d H2Om 1.86 1.87 1.87 2.03 C1H1 1.11 1.11 1.12 1.12 1.10 C1H2 1.13 1.13 1.12 1.09 1.11 Hw1Oz2 1.73 1.73 1.74 1.83 1.76 C1C2 1.64 1.65 1.70 1.67 1.64 C2C3 1.41 1.41 1.41 1.41 1.43 C1C3 1.83 1.83 1.79 1.84 2.04 angle (deg) C1C2C3 73.3 73.2 69.5 72.8 83.1 C2R2R1C3 6.4 7.0 7.8 4.3 2.0 distance (Å) e R4Oz1 3.33 3.26 4.36 3.48 e R4Oz3 2.22 2.20 3.04 2.23 e C2R4 1.09 1.09 1.09 1.09

a b c Atom labels are defined in Figure 5. All attempts to optimize these structures failed. R4 is, in most cases, the proton that is transferred back d e to the zeolite to yield the described products. The three H atoms on C1 are located at distances of 2.66, 2.69, and 2.88 Å from the Om atom. For the cation formed after the methylation of trans-2-butene, there is a different arrangement; the proton that is to be returned points away from the cluster, and a methyl group points down toward the cluster instead. the energy is substantially lowered at the MP2/6-31G*//B3LYP/ 6-31G* level of theory. This may be explained by the superior description of long-range interactions inherent in this method. The hexyl cation formed after the methylation of 2-methyl- 2-butene is less bridged than the other ions. The C1C3 distance is longer and the C1C2C3 angle is less sharp. This is probably because the formal charge on C3, which, in this case, is a tertiary C atom, is fairly well stabilized. Hence, the tendency for formation of a bridged species is reduced. It is noteworthy that this cation, at the MP2/6-31G*//B3LYP/6-31G* level of theory, lies only 73 kJ/mol above the adsorbed reactants.

The R4Oz3 distances listed in Table 4 are, for several species, Figure 5. Schematic representation of an adsorbed cation. quite short (R4 is, in most cases, the proton that is transferred the original double bond. It therefore becomes a matter of back to the zeolite to yield the described products). Even though personal preference whether these species should be considered this proton points directly toward a zeolite O atom, spontaneous carbenium ions or methyl-substituted cyclopropane carbonium deprotonation does not occur. The geometry around C2 partially ions. Sieber et al.47 used quantum chemistry to give an extensive explains this observation. During deprotonation, C2 will gradu- + ally adopt sp2 geometry. This represents considerable distortions; gas-phase description of C4H9 cations, using MP2(full)/6- 31G** for geometry optimizations, and Faˇrcas¸iu and Norton48 both the H1Oz1 and H2Om interactions will be at least partially published a similar investigation of the secondary and tertiary weakened. This seems to be the main reason these structures pentyl cations, using MP2/6-31G** for optimizations. These are energy minima. authors concluded that the open-chain, secondary C4 and C5 The failure to optimize the protonated cyclopropane car- cations are transition states rather than energy minima. The bonium ion and water on the cluster was not surprising. This secondary cations are always stabilized as more- or less-bridged molecule is the smallest of the series, and it, therefore, has the species. C1C2C3 angles of 62.5° and 66.4° are reported for gas- greatest mobility on the cluster. It is, thus, more prone to phase geometries of cations very similar to those formed after deprotonation. Also, it is considerably less stable energetically the methylation of cis- and trans-2-butene, respectively. These than the other species investigated. It was, however, less obvious angles are even sharper than those listed in Table 4. The C2C3 that all attempts to optimize the cation formed after the distances produced by these high-level calculations are in the methylation of isobutene should fail. The potential energy range 1.397-1.403 Å, which is slightly shorter than the surface (PES) is extremely flat around a geometry that corre- distances presented in this work. The geometries presented for sponds to the hypothesized cation (see Figure S6 of the the energy minima in refs 47 and 48 are thus in good agreement Supporting Information). No energy minimum could be found, with the results presented here. however, even though two of four convergence criteria were The adsorption mode of the cation formed after the meth- fulfilled at one point. The electronic energy was then 99 kJ/ ylation of cis-2-butene is different from that of the others. The mol greater than that of the adsorbed reactants, which fits well C1 methyl group is rotated and there is no proper interaction with the main trend in Table 2. The geometry was similar to between any of the hydrogen atoms on C1 and the water that of the cations reported for the other methylations, i.e., a molecule; the shortest distance is 2.66 Å. The energy is therefore bridged species. It is not possible to exclude the existence of a higher than that of any of the other cationic species. However, shallow energy minimum; however, it may be that, in this Methylation of Alkenes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 35, 2003 9287 particular case, the cation is an inflection point on the PES, rather than a minimum. The deprotonation of the butyl cation formed after the methylation of propene was investigated more thoroughly, and an activation energy for proton loss of 6 kJ/mol (8 kJ/mol without ZPE correction) was found. An identical value (6 kJ/mol) was found at the B3LYP/6-311G**//B3LYP/6-31G* level of theory, and the barrier was 23 kJ/mol when using MP2/ 6-31G*//B3LYP/6-31G* methodology. The deprotonation transition state describes a slight movement of the methyl group Figure 6. Three cluster models with different terminating groups and acidity ((a) -OH cluster, (b) -F cluster, and (c) -CF3 cluster). bonded to C3 toward the water molecule. The quasi-IRC approach was applied and revealed that, in a concerted reaction, We have previously investigated the reactivity of methyl- one of the protons of the methyl group is transferred to the water substituted arenes with the same methods as those used here. molecule and the H atom (which is the water proton pointing w1 The barriers for methylation of toluene, 1,2,4,5-tetramethyl- toward a zeolite O atom) is returned to the cluster. This results benzene, and hexamethylbenzene were found to be 187, 171, in the formation of 1-butene and water. The very low barrier and 169 kJ/mol (B3LYP/6-31G* + ZPE correction), respec- indicates that proton loss will occur instantaneously under tively.23 This is in the very same range as the barriers presented experimental conditions. Indeed, there have been no experi- here, and it must be concluded that arenes and alkenes have mental observations of aliphatic carbenium ions in zeolites. The very similar activity in methylation reactions. One important stationary points described previously are, nonetheless, quite factor that has not been considered so far is the pre-exponential. descriptive for the reaction pathway of the methylation reactions This factor may be somewhat larger for arenes than for alkenes, studied and serve to elaborate the role of carbenium-ion-like because an aromatic ring formally has three double bonds, which species in zeolites. can all be methylated. This fact means that the arene has several 3.4. Products and Reaction Energies. The final products reactive sites, compared to the lone double bond of an alkene. (water and the deprotonated alkene with one more C atom than 3.6. Effect of Zeolite Cluster Proton Affinity. The effect the reactant) are also coadsorbed onto the cluster. The water of cluster acidity or proton affinity was investigated by molecule is hydrogen-bonded to the acidic site, and the alkene constructing two zeolite clusters that are different from the is associated with the water molecule. The reaction energies model discussed so far. We substituted the terminal -OH group for the adsorbed products are more exothermic than the gas- that is bonded to the Al T-atom with -F and -CF3 groups. phase energies. This, of course, indicates that the products are The overall compositions of these new models are then more strongly adsorbed than the reactants, by ∼15 kJ/mol. As + - + - H [(F)Al(OSiH3)3] and H [(F3C)Al(OSiH3)3] , as shown in expected, the differences between calculated reaction enthalpies Figure 6. The proton affinity of the regular, -OH-substituted at 298 K and the zero-point corrected reaction energies (both deprotonated cluster is 1256 kJ/mol, and our approach allowed at B3LYP/6-31G*) are very small. Increasing the size of the a tuning of the proton affinities over a range of 40 kJ/mol. basis set used in combination with the B3LYP functional causes Zeolite acid strength is claimed to be in the range of strong 35 - the reaction energies to move somewhat closer to the experi- acids, rather than that of superacids. The HSO4 anion has mental values. Going from B3LYP/6-311G**//B3LYP/6-31G* an experimentally determined gas-phase proton affinity of 1282 to MP2/6-31G*//B3LYP/6-31G* methodology represents a kJ/mol,26 which compares well with the acidity of our cluster small improvement, but only at the MP2/6-311G**//B3LYP/ model. On the other hand, Nicholas and Haw51 determined that 6-31G* level of theory is it possible to achieve fair agreement a hydrocarbon with a proton affinity of >874 kJ/mol will be with the experimental values. It seems that this fairly advanced protonated and form a stable carbenium ion within the zeolite method is required to reproduce the reaction energies of pores. This value is considerably lower than a gas-phase proton nonisodesmic reactions properly. The small variations in the affinity, because of the electrostatic stabilization of the ion pair. experimental values are not reproduced by any of the methods. Studies relying on periodical boundary conditions and plane 3.5. Bearings on Methanol-to-Hydrocarbon Chemistry. wave basis sets, which include interactions with the entire zeolite The results presented above have important relevance to framework, have shown that charged species and transition states methanol-to-hydrocarbon (MTH) chemistry in several ways. are stabilized by 20%-50%,21,52,53 compared to the cluster Ethene has a considerably higher barrier for methylation than approach. Although the cluster model is a good qualitative model any of the other alkenes investigated, and a low reactivity is of an acidic site, it seems reasonable to assume that the model predicted. This has been confirmed by experimental results by does not fully simulate the quantitative acid strength of a real Dahl and Kolboe.9,10 They observed experimentally that ethene zeolite catalyst. shows little reactivity when co-reacted with methanol over The methylation of propene was selected as a probe reaction, SAPO-34, which is a zeolite-like catalyst, and it was concluded and the activation barriers (calculated as before) are displayed that ethene is not likely to have an important role as a reaction in Table 5. Selected geometric parameters for the stationary intermediate. The next-higher homologue, propene, has a lower points are listed in Table 6. It is clear that the activation energy barrier, and this trend is upheld when comparing propene with is strongly dependent on the cluster proton affinity. The the butenes and the butenes with 2-methyl-2-butene. The MTH difference of 40 kJ/mol in proton affinity results in a shift in reaction is described as autocatalytic, and this is most often activation energy of 13 kJ/mol, corresponding to a change by a observed as a large increase in the initial methanol conversion factor of 14 in the reaction rate at 600 K. This relationship is when any hydrocarbon is added to the methanol feedstock.4,7,49,50 the same for all methods used, although the absolute value of The results presented here further elucidate this aspect of MTH the cluster proton affinity is sensitive to the calculation chemistry, because we have shown that the product from one methodology. This correlation between catalyst acidity and reaction step (methylation) is more reactive than the initial activation energy is probably strongest when the zeolite proton reactant. is completely transferred to the reactants in the transition state, 9288 J. Phys. Chem. B, Vol. 107, No. 35, 2003 Svelle et al.

TABLE 5: Proton Affinities of Three Differently Substituted The increased acid strength is also reflected in the geometries Zeolite Clusters and Activation Energies for the Methylation of the stationary points. For the adsorbed reactants, the Oz1Hz of Propene distance follows the cluster proton acidity nicely. Notably, this cluster cluster proton activation energy reaction energy distance is the same for all three clusters when no adsorbate is substituent affinity (kJ/mol) (kJ/mol) (kJ/mol) present, i.e., 0.974 Å. The HzOm distance is shortened when B3LYP/6-31G* + ZPE the proton acidity decreases, indicating the formation of a -OH 1256 169 -53 stronger hydrogen bond between the zeolite and methanol. The - - F 1242 161 55 same trends are observed for the adsorbed products. These -CF 1217 156 -56 3 effects are less pronounced for the transition states, where there B3LYP/6-31G* are no clear variations that can be related to the acid strength. -OH 1289 170 -53 -F 1274 161 -55 4. Conclusions -CF3 1249 156 -56 B3LYP/6-311G**//B3LYP/6-31G* A quantum chemical investigation of the methylation of -OH 1295 171 -67 different alkenes by methanol on a zeolite cluster model has -F 1275 161 -69 been performed. An associative mechanism has been assumed. - - CF3 1251 155 73 The activation barrier decreases as the size of the alkene MP2/6-31G*//B3LYP/6-31G* increases, i.e., the product from one reaction step is more - - OH 1284 172 64 reactive than the initial reactant. The activation barriers are in -F 1267 164 -65 the same range as those reported for methylation of arenes. -CF3 1239 159 -65 Therefore, in regard to methylation by methanol, the results do TABLE 6: Geometric Parameters (Distances, Angles, and not indicate a significant difference in activity between the two Dihedral Angles) for the Methylation of Propene with Three groups of hydrocarbons. - - - Different Cluster Models ( OH, F, and CF3) Charged intermediates have been optimized as shallow energy - - - OH F CF3 minima for most of the reactions investigated. These carbo- Reactantsa cations are always bridged species, similar to the expected distance (Å) structure of protonated cyclopropane derivatives. The geometries Oz1Hz 1.03 1.04 1.05 are in good agreement with high-level gas-phase calculations Oz2Hm4 1.83 1.83 1.82 that have been reported in the literature. The presence of water H O 1.54 1.52 1.49 z m is important to stabilize the charged species on the cluster model. Hm4Om 0.99 0.99 0.99 OmCm 1.44 1.44 1.44 A barrier for deprotonation was, in one case, calculated and CmCless 3.94 3.93 3.93 found to be merely 6 kJ/mol, indicating immediate deprotona- CmCmore 4.20 4.19 4.19 tion. Although these species will not be persistent under ClessCmore 1.34 1.34 1.34 experimental conditions, they describe the reaction path for the CmAl 4.43 4.40 4.36 angle (deg) studied methylation reactions. The lifetime of these species will be so short that obtaining experimental proof of their existence Hm1Hm2Hm3Cm 35.8 35.6 35.6 OmClessCmoreCm 16.2 16.5 16.8 may be impossible. Transition Statesa The activation barrier for methylation of propene has been distance (Å) found to vary significantly, by 13 kJ/mol, when the cluster Oz1Hz 1.85 1.88 1.88 proton affinity is tuned over a range of 40 kJ/mol. Oz2Hm4 1.92 1.89 1.92 HzOm 0.99 0.99 0.99 Acknowledgment. Thanks are due to the Norwegian Re- Hm4Om 0.98 0.99 0.98 O C 2.23 2.22 2.22 search Council for financial support (through Grant Nos. m m 135867/431 and 149326/431) and a grant of computer time CmCless 2.07 2.08 2.08 CmCmore 2.35 2.36 2.36 through the NOTUR project (Account Nos. NN2147K and ClessCmore 1.36 1.36 1.36 NN2878K). CmAl 3.49 3.46 3.48 angle (deg) Supporting Information Available: Thorough descriptions H H H C -12.9 -12.2 -12.3 m1 m2 m3 m of the geometries and illustrations of the adsorbed cationic OmClessCmoreCm 3.0 2.9 3.0 b species and a description of the PES around the cation formed Products after the methylation of isobutene (PDF). This material is distance (Å) available free of charge via the Internet at http://pubs.acs.org. Oz1Hz 1.04 1.04 1.05 Oz2Hw4 1.88 1.86 1.85 HzOm 1.53 1.51 1.48 References and Notes Hw1Om 0.99 0.99 0.99 (1) Sto¨cker, M. Microporous Mesoporous Mater. 1999, 29,3-48. Hw2Cw 0.98 0.98 0.98 H C 2.29 2.29 2.29 (2) Chang, C. D. The Methanol-to-Hydrocarbons Reaction: A Mecha- w2 less nistic Perspective. In Shape SelectiVe Catalysis; Song, C., Garce´s,J.M., Hw2Cmore 2.36 2.36 2.35 Sugi, Y., Eds.; ACS Symposium Series 738; American Chemical Society: ClessCmore 1.34 1.34 1.34 Washington, DC, 2000; pp 96-114. angle (deg) (3) Dessau, R. M.; LaPierre, R. B. J. Catal. 1982, 78, 136-141. OmCmoreClessHw2 0.2 0.6 0.3 (4) Kolboe, S. Acta Chem. Scand., Ser. A 1986, A40, 711-713. - a Atom labels are defined in Figure 2. b Atom labels are defined in (5) Dessau, R. M. J. Catal. 1986, 99, 111 116. (6) Song, W.; Marcus, D. M.; Fu, H.; Ehresmann, J. O.; Haw, J. F. J. Figure 3. Am. Chem. Soc. 2002, 124, 3844-3845. - which is the case for the selected model reaction. We suspect (7) Chen, N. Y.; Reagan, W. J. J. Catal. 1979, 59, 123 129. (8) Chang, C. D. Catal. ReV.sSci. Eng. 1983, 25,1-118. that, for a reaction involving a lesser degree of proton transfer (9) Dahl, I. M.; Kolboe, S. Catal. Lett. 1993, 20, 329-336. in the transition state, any such dependence would be reduced. (10) Dahl, I. M.; Kolboe, S. J. Catal. 1994, 149, 458-464. Methylation of Alkenes: A DFT Study J. Phys. Chem. B, Vol. 107, No. 35, 2003 9289

(11) Dahl, I. M.; Kolboe, S. J. Catal. 1996, 161, 304-309. (32) Svelle, S.; Rønning, P. O.; Kolboe, S., in preparation. (12) Mikkelsen, Ø.; Rønning, P. O.; Kolboe, S. Microporous Mesoporous (33) Koch, W.; Schleyer, P. v. R.; Buzek, P.; Liu, B. Croat. Chem. Acta Mater. 2000, 40,95-113. 1992, 65, 655-672. (13) Arstad, B.; Kolboe, S. Catal. Lett. 2001, 71, 209-212. (34) Koch, W.; Liu, B.; Schleyer, P. v. R. J. Am. Chem. Soc. 1989, (14) Arstad, B.; Kolboe, S. J. Am. Chem. Soc. 2001, 123, 8137-8138. 111, 3479-3480. (15) Bjørgen, M.; Olsbye, U.; Kolboe, S. J. Catal. 2003, 215,30-44. (35) Martens, J. A.; Jacobs, P. A. An Introduction to Acid Catalysis (16) Haw, J. F.; Nicholas, J. B.; Song, W.; Deng, F.; Wang, Z.; Xu, T.; with Zeolites in Hydrocarbon Reactions. In An Introduction to Zeolite Heneghan, C. S. J. Am. Chem. Soc. 2000, 122, 4763-4775. Science and Practice; van Bekkum, H., Flanigen, E. M., Jacobs, P. A., (17) Goguen, P. W.; Xu, T.; Barich, D. H.; Skloss, T. W.; Song, W.; Jansen, J. C., Eds.; Studies in Surface Science and Catalysis 137; Elsevier Wang, Z.; Nicholas, J. B.; Haw, J. F. J. Am. Chem. Soc. 1998, 120, 2650- Science: Amsterdam, 2001; pp 633-672. 2651. (36) Frash, M. V.; van Santen, R. A. Top. Catal. 1999, 9, 191-205. (18) Sassi, A.; Wildman, M. A.; Ahn, H. J.; Prasad, P.; Nicholas, J. B.; (37) Kazansky, V. B.; Senchenya, I. N. J. Catal. 1989, 119, 108-120. - Haw, J. F. J. Phys. Chem. B 2002, 106, 2294 2303. (38) Kazansky, V. B.; Frash, M. V.; van Santen, R. A. Appl. Catal. A (19) Song, W.; Haw, J. F.; Nicholas, J. B.; Heneghan, C. S. J. Am. Chem. 1996, 146, 225-247. Soc. 2000, 122, 10726-10727. - (39) Kazansky, V. B.; Frash, M. V.; van Santen, R. A. Quantum- (20) Song, W.; Fu, H.; Haw, J. F. J. Phys. Chem. B 2001, 105, 12839 Chemical Study of the Nonclassical Carbonium Ion-Like Transition States 12843. in Isobutane Cracking on Zeolites. In 11th International Congress on (21) Vos, A. M.; Rozanska, X.; Schoonheydt, R. A.; van Santen, R. A.; - V - Catalysis 40th Anni ersary, Part B; Hightower, J. W., Delgass, W. N., Hutschka, F.; Hafner, J. J. Am. Chem. Soc. 2001, 123, 2799 2809. Iglesia, E., Bell, A. T., Eds.; Studies in Surface Science and Catalysis 101; (22) Vos, A. M.; Nulens, K. H. L.; De Proft, F.; Schoonheydt, R. A.; - - Elsevier Science: Amsterdam, 1996; pp 1233 1242. Geerlings, P. J. Phys. Chem. B 2002, 106, 2026 2034. (40) Frash, M. V.; Solkan, V. N.; Kazansky, V. B. J. Chem. Soc., (23) Arstad, B.; Swang, O.; Kolboe, S. J. Phys. Chem. B 2002, 106, - - Faraday Trans. 1997, 93, 515 520. 12722 12726. (41) Frash, M. V.; Kazansky, V. B.; Rigby, A. M.; van Santen, R. A. J. (24) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, Phys. Chem. B 1997, 101, 5346-5351. M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; (42) Frash, M. V.; Kazansky, V. B.; Rigby, A. M.; van Santen, R. A. J. Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. Phys. Chem. B 1998, 102, 2232-2238. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, (43) Kazansky, V. B. Catal. Today 1999, 51, 419-434. M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; (44) Jeanvoine, Y.; AÄ ngya´n, J. G.; Kresse, G.; Hafner, J. J. Phys. Chem. Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, - D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; B 1998, 102, 7307 7310. (45) Termath, V.; Haase, F.; Sauer, J.; Hutter, J.; Parrinello, M. J. Am. Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, - I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Chem. Soc. 1998, 120, 8512 8516. (46) Sˇtich, I.; Gale, J. D.; Terakura, K.; Payne, M. C. J. Am. Chem. Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. - W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, Soc. 1999, 121, 3292 3302. M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.11; Gaussian, (47) Sieber, S.; Buzek, P.; Schleyer, P. v. R.; Koch, W.; Carneiro, J. - Inc.: Pittsburgh, PA, 1998. W. d. M. J. Am. Chem. Soc. 1993, 115, 259 270. (25) Ivanova, I. I.; Corma, A. J. Phys. Chem. B 1997, 101, 547-551. (48) Faˇrcas¸iu, D.; Norton, S. H. J. Org. Chem. 1997, 62, 5374-5379. (26) Thermodynamic data were taken from the NIST database (http:// (49) Mole, T.; Whiteside, J. A.; Seddon, D. J. Catal. 1983, 82, 261- webbook.nist.gov/chemistry). 266. (27) Zicovich-Wilson, C. M.; Viruela, P.; Corma, A. J. Phys. Chem. (50) Mole, T.; Bett, G.; Seddon, D. J. Catal. 1983, 84, 435-445. 1995, 99, 13224-13231. (51) Nicholas, J. B.; Haw, J. F. J. Am. Chem. Soc. 1998, 120, 11804- (28) Blaszkowski, S. R.; van Santen, R. A. J. Phys. Chem. 1995, 99, 11805. 11728-11738. (52) Rozanska, X.; van Santen, R. A.; Hutschka, F. J. Phys. Chem. B (29) Blaszkowski, S. R.; van Santen, R. A. J. Phys. Chem. B 1997, 101, 2002, 106, 4652-4657. 2292-2305. (53) Sauer, J.; Sierka, M.; Haase, F. Acidic Catalysis by Zeolites: Ab (30) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders, V. R. Chem. ReV. Initio Modeling of Transition States. In Transition State Modeling for 1994, 94, 2095-2160. Catalysis; Truhlar, D. G., Morokuma, K., Eds.; ACS Symposium Series (31) Kurita, N.; Sekino, H. Chem. Phys. Lett. 2001, 348, 139-146. 721; American Chemical Society: Washington, DC, 1999; pp 358-367.

Paper III

Theoretical Investigation of the Dimerization of Linear Alkenes Catalyzed by Acidic Zeolites

S. Svelle, S. Kolboe, O. Swang, J. Phys. Chem. B 108 (2004) 2953-2962

J. Phys. Chem. B 2004, 108, 2953-2962 2953

Theoretical Investigation of the Dimerization of Linear Alkenes Catalyzed by Acidic Zeolites

Stian Svelle,*,† Stein Kolboe,† and Ole Swang‡ Department of Chemistry, UniVersity of Oslo, P.O. Box 1033 Blindern, N-0315 Oslo, Norway, and SINTEF Applied Chemistry, Department of Hydrocarbon Process Chemistry, P.O. Box 124 Blindern, N-0134 Oslo, Norway ReceiVed: October 22, 2003; In Final Form: December 19, 2003

The zeolite-catalyzed dimerization of ethene, propene, 1-butene, and trans-2-butene has been modeled using quantum chemical methods. Reactants, transition states, and products have been investigated. A cluster model consisting of four tetrahedrally coordinated atoms (T-atoms) has been used to represent the catalyst. Two different mechanism types have been evaluated: concerted and stepwise. In the concerted pathway, protonation and C-C bond formation occur simultaneously. The stepwise mechanism proceeds via alkoxide formation followed by C-C bond formation. The order of reactivity among the different alkene reactants has been assessed. Quantum chemistry predicts that the activation energy of the concerted mechanism lies between the two barriers of the stepwise mechanism. More detailed knowledge concerning the stability of alkoxide species relative to physisorbed alkenes will be necessary for discrimination between the two mechanistic proposals. Implications for the reverse reaction, the â-scission of alkenes, are briefly discussed.

1. Introduction and concerted, have been explored. The stepwise mechanism proceeds via the initial formation of alkoxides as intermediates, Dimerization and oligomerization of lower alkenes to form which subsequently react further to form a new C-C bond. higher molecular weight products may be an attractive route 1 The concerted mechanism involves simultaneous protonation for the production of high octane number gasoline. In particular, and C-C bond formation. Very little theoretical data focusing the dimerization of butenes has become increasingly interesting on the dimerization reaction can be found in the literature. The because methyl tert-butyl ether (MTBE), a well-known high reverse reaction, alkene â-scission, has been investigated to a octane oxygenate blending component, is required to be replaced - limited extent,12 16 and only via mechanisms involving alkoxide by environmentally more beneficial alternatives.2 The traditional intermediates. The concerted dimerization mechanism presented catalyst for propene and butene oligomerization is phosphoric here appears to be novel. It has been of primary interest to acid impregnated on a solid kieselguhr carrier, known as the compare the reactivity of the various alkenes submitted to solid phosphoric acid (SPA) catalyst.3 The SPA catalyst displays calculation and also to efficiently compare the two mechanisms excellent activity and selectivity. However, safely disposing of for alkene dimerization. Attempts were made also to investigate the spent catalyst, which contains organic tars and is highly the dimerization of 2-methylpropene (iso-butene), but it was acidic, is challenging. Regenerable zeolites have been investi- not possible to model this reaction via the mechanisms outlined gated as alternatives and been found to be well suited as catalysts above. Rather, attempts to find transition states gave tert-butyl for alkene dimerization and oligomerization.4-9 Depending on carbenium ions as energy minima, as described also by Sinclair reaction conditions, the products may be in the gasoline range,8 et al.17 in a QM/MM-study. There have been no experimental but lubricants can also be manufactured.5 observations of simple alkyl carbenium ions in zeolites, and Apart from these practical applications, zeolite-catalyzed these computational results might be artifacts. Also, two iso- alkene dimerization is also interesting from a more fundamental butene molecules are considerably bulkier than the other pairs viewpoint, as alkene dimerization is an elementary reaction of reactants, and we speculate that the limit to the usefulness which may occur in many zeolite-catalyzed hydrocarbon of our cluster model (see below) has been reached or transcended transformations. In particular, during the conversion of methanol in the iso-butene case. It therefore seems that the iso-butene to hydrocarbons (MTH), the concentration of alkenes escaping reaction requires substantial further investigations, beyond the the zeolite pore systems can be quite high.10 It is therefore of scope of the present work, before a satisfactory description may interest to evaluate the reactivity of these alkenes with quantum be given. chemical methods. We have previously investigated the reaction of alkenes with a methanol molecule,11 and in this report we A 19-atom cluster model containing four tetrahedral atoms study the combination of two alkene molecules. Increased insight (denoted 4T) has been used to represent the zeolite catalyst. in mechanistic details relevant to the MTH reaction was the Within the cluster approach, a small fragment is used to simulate original motivation behind the present work. the Brønsted acidic site. Calculations with such clusters have proven them to be adequate for qualitative descriptions of Dimerization of ethene, propene, 1-butene, and trans-2-butene 16 has been investigated. Two different mechanism types, stepwise chemical rearrangements that occur locally on the active site. Structure-specific effects and effects caused by the electrostatic field present in the zeolite micropores are, however, not well * Corresponding author. E-mail: [email protected]. † University of Oslo. described. The same cluster as employed in this work has been ‡ SINTEF Applied Chemistry. used by several workers to model reactions similar to those 10.1021/jp0371985 CCC: $27.50 © 2004 American Chemical Society Published on Web 02/10/2004 2954 J. Phys. Chem. B, Vol. 108, No. 9, 2004 Svelle et al.

Figure 1. Energy profiles of the concerted and stepwise dimerization mechanisms. described here: Methylation of alkenes11 and methylbenzenes18 of atoms. Internal reaction coordinate (IRC) calculations, as by methanol and halomethanes19 and also the ethylation and implemented in Gaussian98, were in some cases carried out. iso-propylation of methylbenzenes20 have been investigated. Such calculations follow reaction paths in both directions from Moreover, Rozanska et al.21,22 studied the isomerization and a given transition state in order to investigate the minima transalkylation of toluene and xylenes and found that the relative connected by the transition state. A stepsize of 0.3 amu-1/2 bohr order of the activation energies is conserved when comparing and the ultrafine integration grid were used. The transition states results obtained when using the same cluster as here with were also investigated by perturbing the geometries very slightly calculations relying on periodic boundary conditions (zeolite along the reaction coordinate corresponding to the negative mordenite) combined with plane wave basis sets. eigenvalue in the Hessian, and using the geometries thus produced as starting points for energy minimizations. This quasi- 2. Computational Details IRC approach occasionally yields information not readily All calculations were done using the Gaussian98 program derived from strictly rigorous IRC-calculations. In combination, 23 package. The B3LYP hybrid density functional combined with these two techniques did confirm that the transition states found 6-31G(d) basis sets were employed for all geometry optimiza- involve the rupture and formation of the required bonds, thus tions. No geometric constraints were used in the optimizations. properly describing the desired reactions. In the following, when The ultrafine integration grid was used in order to ensure energies are discussed, we refer to the B3LYP/6-31G(d) + ZPE convergence. Additionally, single-point electronic energies were values, unless otherwise stated. calculated for the optimized geometries using B3LYP and MP2 combined with larger, triple-ú basis sets with polarization 3. Results and Discussion functions on all atoms. These schemes are denoted B3LYP/ Dimerization of ethene, propene, 1-butene, and trans-2-butene cc-pVTZ//B3LYP/6-31G(d) and MP2/6-311G(d,p)//B3LYP/ has been submitted to quantum chemical analysis. Two different 6-31G(d). mechanisms have been evaluated, a stepwise pathway (subsec- The zeolite catalyst has been modeled by a widely used cluster tion 3.1) and a concerted mechanism (subsection 3.2). Figure 1 consisting of four tetrahedral atoms, i.e., three silicon and one displays the energy profiles of both mechanisms and labels the aluminum atom to generate the acidic site.18 For most stationary stationary points described in the following discussion. These points, there is more than one possible orientation of the reactant labels will be used for comparison of the potential energy relative to the cluster. These possibilities are usually indistin- surfaces (PES) of the two mechanisms. All transition states guishable in energy, but care has been taken to select stationary explored are carbenium ion-like. For the concerted mechanism, points in which reactants and products were coordinated we investigated the formation of primary, as well as secondary, similarly to the cluster for all reactions. This makes comparisons carbenium ion-like species in the transition states. For the for trends easier, but should otherwise not influence the results. stepwise mechanism, only the intuitively most stable possibilities For all stationary points, vibrational spectra were calculated to were analyzed. As the reactions proceed along the reaction ensure that the correct number of imaginary frequencies was at coordinate from the dimerization transition states toward the hand, i.e., one imaginary frequency for transition states and zero products, (at least) two possibilities can be envisaged: immedi- for energy minima. For the transition states, the normal modes ate deprotonation to yield a neutral alkene or formation of an corresponding to the imaginary frequencies were visualized to alkoxide species. Both possibilities have been investigated and confirm that they indeed corresponded to the expected motion the corresponding energies are reported. These aspects are dis- Dimerization of Linear Alkenes: A DFT Study J. Phys. Chem. B, Vol. 108, No. 9, 2004 2955

TABLE 1: Products Investigated for the Various Implications for the reverse reaction, the â-scission, are Dimerization Reactions discussed briefly in subsection 3.4. reactant alkoxidea alkene product 3.1. Stepwise Dimerization. In this pathway, one alkene ethene 1-butoxide 1-butene molecule is initially adsorbed on the cluster, and an alkoxide propene (s)b 4-methyl-2-pentoxide 4-methyl-2-pentene species is formed. Subsequently, a second alkene is adsorbed propene (p)c 2-hexoxide 2-hexene onto the cluster with the alkoxide group, and in the second step 1-butene (s) 5-methyl-3-heptoxide 5-methyl-3-heptene the alkoxide leaves the zeolite wall and a C-C bond is formed. 1-butene (p) 3-octoxide 3-octene Figure 2 shows the stationary points for the formation of trans-2-butene 3,4-dimethyl-2-hexoxide 3,4-dimethyl-2-hexene ethoxide from ethene, and Figure 3 displays the reaction of the a The last number designates the carbon atom bonded to the zeolite ethoxide group with the second ethene molecule. Energies for oxygen. Branches are considered as substituents on a linear carbon the stepwise mechanism are listed in Table 2. Geometric b ) chain. (s) formally secondary carbenium ion in the transition state. parameters for alkoxide formation and the further reaction are c (p) ) formally primary carbenium ion in the transition state. listed in Tables 3 and 4, respectively. cussed thoroughly in the following sections. For most reactions, The formation of alkoxide species (σ-complex) from alkenes deprotonation immediately after the transition state could result coordinated to the zeolite acidic site (as π-complex), or alkene in several different alkenes. In these cases, the thermodynami- chemisorption, has been the subject of numerous previous cally most stable products were selected. Possible hydride- or theoretical investigations.17,24-31 Our results are in general methyl-shifts were not considered. Table 1 lists the products accord with the bulk of previously published data and will be formed in the investigated reactions. discussed in brevity. In the transition state for alkoxide

Figure 2. Stationary points for formation of ethoxide group from ethene.

Figure 3. Stationary points for the coupling of an ethoxide group with ethene. 2956 J. Phys. Chem. B, Vol. 108, No. 9, 2004 Svelle et al.

TABLE 2: Energy Parameters for Stepwise Dimerizationa Energy ofb

R1 TS1S P1S R2S TS2S Pσ Pπ ∆E(π-σ) Eact1S Eact2S B3LYP/6-31G(d) +ZPE ethene -31 +66 -88 -92 +76 -174 -139 -57 +97 +167 propene -34 +48 -81 -86 +57 -150 -122 -46 +82 +143 1-butene -35 +45 -80 -84 +53 -137 -120 -45 +79 +137 trans-2-butene -34 +61 -63 -66 +72 -91 -94 -29 +96 +139 B3LYP/6-31G(d) ethene -36 +70 -107 -113 +61 -208 -161 -70 +106 +175 propene -39 +54 -97 -104 +49 -180 -141 -57 +93 +153 1-butene -39 +53 -95 -102 +46 -166 -139 -56 +93 +148 trans-2-butene -39 +66 -80 -85 +66 -123 -113 -40 +105 +151 B3LYP/cc-pVTZ//B3LYP/6-31G(d) ethene -27 +94 -65 -66 +100 -146 -130 -38 +122 +167 propene -30 +77 -57 -59 +87 -121 -109 -27 +107 +145 1-butene -30 +78 -55 -55 +84 -105 -106 -25 +108 +139 trans-2-butene -29 +89 -41 -40 +99 -64 -82 -12 +118 +140 MP2/6-311G(d,p)//B3LYP/6-31G(d) ethene -38 +99 -91 -107 +80 -210 -179 -53 +137 +187 propene -45 +82 -93 -113 +62 -211 -181 -48 +127 +174 1-butene -47 +81 -93 -115 +50 -202 -181 -46 +128 +165 trans-2-butene -49 +89 -82 -105 +63 -181 -172 -33 +139 +168 a Labels defined in Figure 1; energies in kJ/mol. b Relative to two gas-phase alkene reactants and the cluster at infinite separation.

TABLE 3: Geometric Parameters for Formation of TABLE 4: Geometric Parameters for the Coupling of an Alkoxide Groupsa Alkoxide Group with an Alkenea ethene propene 1-butene trans-2-butene ethene propene 1-butene trans-2-butene adsorbed reactants adsorbed reactants Oz1H1 0.98 0.99 0.99 0.99 Oz1C1 1.47 1.49 1.49 1.49 H1C1 2.29 2.14 2.15 2.19 Oz2H1 4.23 2.82 2.73 2.72 H1C2 2.25 2.31 2.30 2.21 Oz3H2 3.65 b C1C2 1.34 1.34 1.34 1.34 C1C2 4.11 4.00 4.07 4.49 C C b 3.95 3.98 4.50 4.12 transition states 1 3 C2C3 1.33 1.34 1.34 1.34 Oz1H1 1.34 1.42 1.42 1.48 AlC2 4.55 5.02 5.24 5.39 H1C1 1.30 1.25 1.25 1.23 AlC3 4.96 4.68 5.37 5.29 C1C2 1.40 1.40 1.41 1.41 Oz1C2C3C1 136 140 142 142 Oz2C2 2.18 2.37 2.39 2.40 Oz1H1C1 174 176 176 176 transition states b dihedral-C2 176 176 176 177 Oz1C1 2.36 2.53 2.58 2.67 O H 2.23 2.09 2.09 2.09 a Atom labels defined in Figure 2; distances in Å, angles in degrees. z2 1 b Oz3H2 2.28 This is the dihedral angles formed by C2 and the three substituents C1C2 2.36 2.59 2.55 2.77 on this carbon, indicating the deviation from planarity (180°) around C1C3 2.33 2.78 2.75 2.70 the carbon formally bearing the positive charge in the transition state. C2C3 1.35 1.35 1.35 1.35 AlC2 4.62 5.33 5.35 6.39 AlC3 4.40 4.66 4.69 4.76 formation, the acidic proton is about halfway between the zeolite C1H1Oz2 134 163 165 164 oxygen atom and the alkene carbon atom. Simultaneously, the C1H2Oz3 138 other carbon atom of the alkene double bond is moving closer Oz1C2C3C1 165 178 178 177 to another zeolite oxygen atom, as the C-O bond of the alkoxide a Atom labels defined in Figure 3; distances in Å, angles in degrees. b is forming. As can be seen from Table 3, going from ethene to For propene/1-butene the methyl/ethyl group is attached to C2. propene/1-butene to trans-2-butene results in the acidic proton chemisorption is effectively the same as for 1-butene. In fact, being farther from the zeolite oxygen in the transition state and the calculated barrier becomes minutely higher for 1-butene closer to the alkene carbon, i.e., the proton transfer is gradually when higher-level computational schemes are used. The barrier more pronounced in this order. Linked to this, the Oz2C2 for alkoxide formation from trans-2-butene is markedly higher distances become longer as the degree of proton transfer than for propene and 1-butene, it is actually quite close to that increases. A combination of steric effects and the varying of ethene. This was somewhat unexpected, but agrees quite well stabilities of the carbenium ion-like transition states upon with the work of Correa et al.,25 in which the barrier for trans- increased methyl group substitution on the alkene double bond 2-butene chemisorption is reported to be 13 kJ/mol higher than appears to be a reasonable explanation for these trends. The that found for 1-butene. It seems plausible that this predicted double bond is slightly stretched in the transition state. reduction in reactivity for trans-2-butene compared to 1-butene The activation barrier decreases by about 15 kJ/mol when is caused by steric limitations, because, on the basis of inductive going from ethene to propene, which is expected on the basis effect argumentation, trans-2-butene is expected to be equally of the stability of a formally primary carbenium ion (from or slightly more reactive than propene and 1-butene. The trends ethene) compared with a secondary ion (from propene). No in the calculated barriers are reproduced by all computational significant stabilizing effect can be observed by replacing a schemes, although considerable shifts in the energies are found methyl group with an ethyl group: the barrier for propene when comparing different methods. For instance, the MP2/ Dimerization of Linear Alkenes: A DFT Study J. Phys. Chem. B, Vol. 108, No. 9, 2004 2957

6-311G(d,p)//B3LYP/6-31G(d) approach yields activation ener- predicted to be considerably more reactive; the barrier is 24 gies that are close to 30 kJ/mol higher that those found when kJ/mol lower than for ethene. The barriers for 1-butene and using B3LYP/6-31G(d) electronic energies. trans-2-butene are insignificantly different; both lying slightly The geometries of the formed alkoxides are considered trivial, below that found for propene. This trend is reproduced at every and no geometric data for these species are listed in Table 3 level of theory employed. All barriers are considerably higher (see Supporting Information for details). The chemisorption than those found for the first step of this mechanism, the reactions are all found to be quite exothermic, more so for ethene chemisorption. Notably, they are in the same range as the than for propene and 1-butene (for which the energies are equal) barriers for the reverse of the chemisorption, i.e., the release of than for trans-2-butene. Again, the results differ considerably the alkoxides as physisorbed alkenes. from one method to another. Interestingly, adding the B3LYP/ The transition states described above are similar to the one 6-31G(d) ZPE correction to the B3LYP/cc-pVTZ//B3LYP/ described by Hay et al.12 in a study of pentene â-scission, relying 6-31G(d) electronic energy predicts a thermoneutral reaction on a 3T cluster and HF/6-31G(d) optimizations. Their study energy for trans-2-butene chemisorption. These reaction energies corresponds to the combination of an ethoxide with a propene are of importance with respect to the comparison of the two molecule. An activation energy of +188 kJ/mol (HF/6-31G(d) mechanistic proposals, and will be further elaborated in subsec- + ZPE) or +163 kJ/mol (B3LYP/6-31G(d)//HF/6-31G(d) + tion 3.3. ZPE) can be derived for the ethoxide/propene coupling step. 12 Further progress of the stepwise mechanism requires the Hay et al. found Oz1C1 and C1C2 distances about 0.04 Å longer physisorption of a second alkene to the cluster on which an than those found here. Other workers describe similar transition alkoxide group is already formed. In the stationary points thus states for alkene â-scission, always relying on less advanced found, the second alkene is located next to the alkoxide group, descriptions of the catalyst acidic site compared to those - interacting with the cluster as well as the alkoxide itself. The employed in the present study.13 16 adsorption energy is built up from a sum of weak van der Waals As mentioned above, there are several possibilities for the forces acting between adsorbate and adsorbent. As is evident progress of the reaction after the transition state in which the when comparing the P1S (cluster with alkoxide) energies with C-C bond is formed. Direct deprotonation will lead to a the R2S (cluster with alkoxide plus second alkene) energies listed physisorbed alkene product, whereas coordination of the charged in Table 2, the adsorption energies are poorly predicted by all carbon atom to a zeolite oxygen will lead to the formation of DFT schemes employed. Typically, the values are in the range an alkoxide. This issue was investigated by employing the IRC 0-5 kJ/mol. More realistic values were found when MP2/ and quasi-IRC approaches described in Section 2. The rigorous 6-311G(d,p)//B3LYP/6-31G(d) was used. This shortcoming of IRC procedure gave limited information, as these calculations DFT methods in describing weak interactions of this kind is terminated after only a modest shift in geometry beyond that well-known.32 of the transition state. All transition states through which C-C The next step along the reaction coordinate of the stepwise bonds are formed were therefore submitted to the quasi-IRC mechanism is the transition state that describes the formation procedure. For most cases, this resulted in direct deprotonation of the new C-C bond. The Oz1C1 bond, which connects the and alkene formation. Three exceptions were found, however: alkoxide with the cluster, is gradually stretched, and at the same For ethene dimerization, via both mechanisms, and for the time C1 approaches the alkene double bond. For propene and stepwise dimerization of trans-2-butene, deprotonation leading 1-butene, where the alkene double bond is unsymmetrically to the formation of cyclopropane derivatives was observed (see substituted, C1 is becoming connected to least substituted carbon discussion below). Evidently, the outcome of these calculations, (denoted C2). Formally, this places the charge on the most i.e., formation of alkenes, cyclopropanes, or alkoxides, is very substituted carbon (denoted C3). The C1C2 distances are about sensitive to the exact positioning of the methyl groups and other 0.2 Å longer than the C1C3 distances. For ethene and trans-2- alkyl substituents on the alkenes in the transition states. For butene, these distances are more similar, and C1 evidently example, during the dimerization of 1-butene, an ethyl group approaches the center of the double bond. This ensures that the might be directed either toward or away from the cluster, charge formally is evenly distributed between the equivalent resulting in two geometrically different transition states indis- carbon atoms C2 and C3 in the transition states. The Oz1C2C3C1 tinguishable in energy. Such a difference in substituent orienta- dihedral angles are close to 180°, indicating that the four atoms tion might possibly lead to a different site of deprotonation or lie in the same plane. An exception was found for ethene preferred formation of an alkoxide or even a cyclopropane dimerization, for which the angle was 165°. However, the derivative. It is not feasible to explore every possible such transition state for ethene dimerization is slightly different from orientation. We therefore opted to investigate as products species that of the other reactions, as there are two hydrogens on C1 known to exist in the zeolite pores during such hydrocarbon that form hydrogen bonds with zeolitic oxygen atoms. This is transformations, i.e., alkoxides and alkenes. The thermodynami- indicated by the Oz2H1 and Oz3H3 distances listed in Table 4, cally most stable possibilities, not involving hydride- or methyl and might be the cause of the difference. There is a steady shifts, were selected (Table 1). increase in the Oz1C1 distances in the order ethene < propene Frash et al.13 also considered the formation of cyclopropane < 1-butene < trans-2-butene. The distances from C1 to the derivatives, which later on undergo a ring-opening step leading alkene double bond also tend to increase in the same order, to an alkoxide (for the case of â-scission). However, they could except that the values found for propene and 1-butene are nearly only find cyclopropane derivatives on the PES when B3LYP/ identical. These trends are probably the result of steric require- 6-31G(d) was used for optimizations; these species were not ments, rather than one transition state being earlier or later than involved when HF/6-31G(d) was employed. Thus, the prediction another; there is simply an elongation of the key distances of of cyclopropanes as intermediates was suspected to be a the transition states as the alkenes become larger. computational artifact.13 On the basis of the data reported by Activation energies for the step in which the new C-C bond Frash et al.,13 a transition state for the ring-opening of is formed (Eact2S) are listed in data column 10 in Table 2. methylcyclopropane to form 1-butoxide was optimized, and a Ethene dimerization results in the highest barrier. Propene is fairly high barrier of 151 kJ/mol was found for this step (159 2958 J. Phys. Chem. B, Vol. 108, No. 9, 2004 Svelle et al.

Figure 4. Stationary points for concerted dimerization of two ethene molecules. kJ/mol without ZPE correction, 167 kJ/mol at the B3LYP/ Also, the omission of the surrounding pore structure, and thus cc-pVTZ//B3LYP/6-31G(d) level, and 203 kJ/mol at the MP2/ also steric constraints, might allow the largest alkoxides and 6-311G(d,p)//B3LYP/6-31G(d) level of theory). We did not alkenes to adopt unreasonably favorable positions. succeed in finding a transition state directly connecting meth- 3.2. Concerted Dimerization. In the concerted mechanism, ylcyclopropane and 1-butene. The calculated activation energies protonation and C-C bond formation occur simultaneously in for the ring-opening are higher than the barrier for concerted a single step. This is shown in Figure 4, and the energetic and ethene dimerization at every level of theory (see below) and geometric specifics of this pathway are listed in Tables 5 and also higher than the highest barrier of the stepwise mechanism 6, respectively. The starting point is identical to that of the at the B3LYP/cc-pVTZ//B3LYP/6-31G(d) and the MP2/ stepwise mechanism. One alkene is physisorbed onto the acidic 6-311G(d,p)//B3LYP/6-31G(d) levels of theory. On the basis site in the previously described side-on manner. Then, rather of this, the detection of cyclopropanes should be facile during than forming an alkoxide, a second alkene is physisorbed next experimental studies of alkene dimerization, because of the large to the first, on a siliceous part of the zeolite. The adsorption ring-opening barrier. We have not been able to find any ex- energy of this second alkene is very small and not well described perimental reports describing extensive formation of cyclopro- by DFT methods. Reasonable results are obtained only when panes during alkene dimerization/oligomerization, and it is MP2 is used, quite analogous to what was found above when tempting to conclude that their theoretically predicted formation an alkene was adsorbed on a cluster with an alkoxide group is indeed a computational artifact or that the cluster approach already attached. At the MP2/6-311G(d,p)//B3LYP/6-31G(d) grossly overestimates the ring-opening barrier. If this last pos- level of theory, there is a small increase in the adsorption sibility were the case, cyclopropanes might very well be short- energies in the order of the size of the alkenes. lived intermediates during alkene dimerization and â-scission. In the transition state for concerted dimerization, the acidic The above discussion on product formation also applies to proton is partially transferred from a zeolite oxygen atom to the concerted mechanism, and it is thus a matter purely of one of the carbon atoms of the originally π-coordinated alkene. convenience that the alkoxide is depicted as the product in Simultaneously, the other carbon of the double bond is attacked Figure 3 and the physisorbed alkene is the product in Figure 4 by the π-electrons of the second alkene, resulting in the (see below). Details on the geometries of both the π- and formation of the new C-C bond. For the unsymmetrically σ-coordinated products can be found in the Supporting Informa- substituted alkenes, there are two possible sites of protonation, tion. leading to formation of either a formally primary or secondary All reactions are quite strongly exothermic, and the formation carbenium ion in the transition state. Both options were explored. of an alkoxide is energetically more favorable than formation The reactions proceeding via formally primary ions are best of a π-coordinated alkene. There are, however, a few exceptions discussed separately from those involving secondary ions, before to this. In particular, for 3,4-dimethyl-2-hexene formed after a comparison is made between the two possibilities. dimerization of trans-2-butene, the physisorbed alkene lies lower For ethene, propene, and 1-butene, the concerted mechanism in energy than the alkoxide at the B3LYP/6-31G(d) + ZPE and may proceed via transition states involving species resembling B3LYP/cc-pVTZ// B3LYP/6-31G(d) levels of theory. This primary carbenium ions. As indicated by the Oz1H1 and H1C1 appears to be a trend in the sense that increasing the bulk of distances listed in Table 6, the degree of proton transfer from the hydrocarbon fragment favors the physisorbed species relative the zeolite to the alkene is greater for ethene dimerization than to the alkoxide. It should be noted, however, that as the size of for propene and 1-butene, which are almost equal in this respect. the hydrocarbon increases, the shortcomings of the cluster Also, the distance from C2, which formally bears the positive approach become more pronounced. In particular, some un- charge, to the second alkene double bond is smaller for ethene physical interactions between the hydrocarbon fragments and than for propene/1-butene. It seems that the progress of reaction the terminating -SiH3 groups cannot be completely avoided. is greater at the transition state for ethene than for propene and Dimerization of Linear Alkenes: A DFT Study J. Phys. Chem. B, Vol. 108, No. 9, 2004 2959

TABLE 5: Energy Parameters for Concerted Dimerizationa Energy ofb

R1 R2C TSC Pσ Pπ EactC B3LYP/6-31G(d) +ZPE ethene -31 -35 +91 -174 -139 +127 propene (s) -34 -39 +58 -150 -122 +97 propene (p) +83 -148 -123 +122 1-butene (s) -35 -40 +62 -137 -120 +102 1-butene (p) +81 -140 -125 +120 trans-2-butene -34 -38 +76 -91 -94 +114 B3LYP/6-31G(d) ethene -36 -44 +87 -208 -161 +131 propene (s) -39 -47 +58 -180 -141 +105 propene (p) +82 -179 -143 +129 1-butene (s) -39 -47 +58 -166 -139 +105 1-butene (p) +82 -170 -144 +129 trans-2-butene -39 -45 +71 -123 -113 +116 B3LYP/cc-pVTZ//B3LYP/6-31G(d) ethene -27 -28 +121 -146 -130 +149 propene (s) -30 -31 +90 -121 -109 +121 propene (p) +115 -120 -112 +146 1-butene (s) -30 -30 +94 -105 -106 +124 1-butene (p) +115 -109 -112 +145 trans-2-butene -29 -29 +105 -64 -81 +134 MP2/6-311G(d,p)//B3LYP/6-31G(d) ethene -38 -52 +113 -210 -179 +165 propene (s) -45 -59 +74 -211 -181 +133 propene (p) +90 -204 -174 +150 1-butene (s) -47 -65 +75 -202 -181 +140 1-butene (p) +86 -197 -176 +152 trans-2-butene -49 -67 +70 -180 -172 +137 a Labels defined in Figure 1; energies in kJ/mol. b Relative to two gas-phase alkene reactants and the cluster at infinite separation. 1-butene. The highest activation barrier is found for ethene intuitive result. One feasible explanation for the unexpectedly dimerization, and this is especially pronounced when MP2/ low activity of trans-2-butene relative to propene/1-butene lies 6-311G(d,p)//B3LYP/6-31G(d) methodology is used. The bar- in the reaction energies, which are considerably less exothermic riers found for propene and 1-butene are basically identical. This for trans-2-butene than for propene and 1-butene, both when is consistent with the observed variations in the key distances the π- and σ-complexes are considered. According to the in the transition states described above. A large part of the Brønsted-Polanyi relation,33 a less exothermic reaction should activation energy is associated with the removal of the acidic result in a higher barrier, which is exactly what is found here. proton from the cluster,11 which is most pronounced in the Also, steric limitations in the transition state raising the energy ethene transition state, thereby raising the barrier. Also, the short cannot be excluded. C2C3 and C2C4 distances found for ethene implies greater charge For propene and 1-butene protonation leading to both primary localization on C2, further enhancing the relative instability of and secondary carbenium ion-like transition states is possible, the transition state. The Mulliken charges on C2 are +0.31e for and as very much expected, the secondary option is clearly ethene and +0.26e for propene and 1-butene, thus confirming preferred, by 20-30 kJ/mol. A difference in barrier of 25 kJ/ this notion. mol corresponds to a factor of 150 in relative reaction rates at Transition states involving formally secondary carbenium ions 600 K, signaling that protonation at the most favored site is are possible for propene, 1-butene, and trans-2-butene. Again, nearly completely dominant. The degree of proton transfer in trends in geometric and energetic details can be found. When the transition states is always greater for the secondary alterna- inspecting the activation barriers obtained at the various levels tives, and this is also reflected in the more severely stretched of theory, it seems that the barrier for propene is slightly lower alkene double bonds (the C1C2 distances) for these options. The or possibly equal to that found for 1-butene, whereas trans-2- double bond of the other alkene (C3C4) is only moderately butene is predicted to have the lowest reactivity. Thus, the bulk stretched in every instance. Some features found in the transition of the results indicate the following order of reactivity in this states of the concerted mechanism are independent of the site case: propene g 1-butene > trans-2-butene. This somewhat of protonation. The most noteworthy aspect is the anti-periplanar unexpected result is opposite of the order found above when geometry, most commonly found for E2 elimination reactions. primary carbenium ions were involved as transition states. The Indeed, when considering the dimerization in the reverse degree of proton transfer from the zeolite, indicated by the Oz1H1 direction, i.e., alkene cracking, the reaction is rightfully labeled distance, increases when going from propene to 1-butene to as an E2 reaction. The anionic zeolite cluster serves as the base, trans-2-butene. Also, the C1C3 and C1C4 distances (C1 now bears taking the H1 proton from the alkene. Simultaneously, the second the positive charge) decrease in the same order. As before, it alkene is lost as a leaving group on the other side of the forming seems that a high activation barrier is coupled with a late double bond. The geometry around the carbon atom from which transition state. Interestingly, the Mulliken charges on C1 are the leaving group is lost is nearly planar, as indicated by the +0.32e for propene, +0.31e for 1-butene, and +0.29e for trans- dihedral angles listed in Table 6. Another aspect worth pointing 2-butene, indicating that the charge delocalization in the out is the nearly isosceles triangular arrangement of the carbon transition state is greatest for trans-2-butene, which is the atom formally bearing the positive charge and the two carbons 2960 J. Phys. Chem. B, Vol. 108, No. 9, 2004 Svelle et al.

TABLE 6: Geometric Parameters for Concerted Dimerizationa Primaryb Secondaryc ethene propene 1-butene propene 1-butene trans-2-butene adsorbed reactants Oz1H1 0.98 0.99 0.99 0.99 H1C1 2.27 2.31 2.26 2.19 H1C2 2.26 2.16 2.19 2.21 C1C2 1.34 1.34 1.34 1.34 C1C3 3.90 4.22 4.23 4.18 C1C4 4.13 3.92 3.94 4.01 C2C3 4.24 5.13 5.24 4.80 C2C4 4.81 4.53 4.69 4.72 C3C4 1.33 1.34 1.34 1.34 AlC1 4.51 4.36 4.22 4.24 AlC2 4.65 4.61 4.64 4.70 AlC3 4.56 5.24 5.24 5.50 AlC4 5.12 4.66 4.66 6.22 transition states Oz1H1 1.60 1.54 1.54 1.63 1.80 1.87 H1C1 1.20 1.23 1.24 1.94 1.99 1.90 H1C2 1.93 1.89 1.88 1.19 1.14 1.14 C1C2 1.41 1.41 1.40 1.42 1.44 1.45 C2C3 2.51 2.66 2.67 3.56 3.32 3.59 C2C4 2.53 2.50 2.51 3.39 3.31 3.16 C1C3 3.33 3.42 3.44 2.90 2.86 2.64 C1C4 3.29 3.28 3.29 2.78 2.58 2.62 C3C4 1.35 1.35 1.35 1.35 1.35 1.35 AlC1 3.48 3.50 3.51 3.63 3.63 3.58 AlC2 3.58 3.62 3.62 3.52 3.67 3.79 AlC3 4.95 4.90 4.93 4.98 6.34 5.56 AlC4 5.72 5.59 5.62 5.69 5.66 6.01 Oz2H2 2.03 2.11 2.11 2.07 2.22 2.18 Oz2H1C2 150 151 151 145 117 124 d dihedral-C1 173 171 170 d dihedral-C2 172 173 173 a Atom labels defined in Figure 4; distances in Å, angles in degrees. b Formally primary carbenium ion in the transition state; the charge is c d located on C2. Formally secondary carbenium ion in the transition state; the charge is located on C1. These are the dihedral angles formed by C1/C2 and the three substituents on these carbons, indicating the deviation from planarity (180°) around the carbon formally bearing the positive charge in the transition state. of the second double bond, indicated by the almost equal desired. The first barrier of the stepwise mechanism is always distances from C2 (primary) or C1 (secondary) to C3 and C4 lower than the single barrier of the concerted pathway, whereas found in Table 6. This characteristic was also observed in the the second barrier of the stepwise mechanism is considerably second transition state of the stepwise mechanism. Arstad et higher than that of the concerted route. With this in mind, one al.20 have very recently published a theoretical cluster study on could argue that concerted dimerization will dominate. However, zeolite-catalyzed arene alkylation via a mechanism analogous if the chemisorption reaction, for which the lowest barrier is to the concerted mechanism described here, and no such three- found, is driven to completion in the sense that every acidic ring arrangement was found in the alkylation transition states site is occupied by an alkoxide, the concerted route becomes in that case. It therefore seems that this feature is unique to blocked. reactions involving localized double bonds. The potential energy surfaces for both mechanisms are 3.3. Stepwise versus Concerted Dimerization. Any attempts described graphically in Figure 1 and numerically in Tables 2 to extract quantitative information based on the current set of and 5. Strikingly, the highest-lying points on the PES are found calculations warrant an initial discussion of the limitations for the concerted mechanism. Further, the energy differences inherent to the selected cluster model. There are two shortcom- between the topmost points for the two mechanisms are ings that require particular attention. First, due to the limited considerably smaller than the differences in activation barriers. size of the model surface, combined with our relying mainly On the basis of these observations, one could argue that the on DFT methods, the strengths of weak (dispersiVe) interactions main cause of the high barrier found for the second step of the will be underestimated.32 This causes, for instance, the energy stepwise mechanism is the exothermicity of the alkoxide differences between R1 and R2c and also P1s and R2s (see Figure formation rather than an intrinsic instability of the second 1) to become unrealistically small. However, this effect is transition state as such. Simply put, after having formed the expected to be fairly similar for all adsorbed states, thus alkoxide, one appears to be positioned in an energy well, from becoming evened out when comparing the two mechanism which further progress of reaction requires the overcoming of types. Second, the omission of the major part of the zeolite wall a fairly high barrier. Thus, the relative energies of the phys- in relatively close proximity to the active site might underes- isorbed and chemisorbed alkenes (∆E(π-σ)) should be sub- timate the effects of steric limitations related to the pore wall jected to further scrutiny, as this issue appears to be decisive curvature or any other geometric peculiarity. Despite these when it comes to discrimination between the two reaction drawbacks of the selected model, one aim of the present study mechanisms. We speculate that the predicted stability of the was to decide which is the prevailing dimerization mechanism. alkoxides may be related to an incomplete description of steric Unfortunately, this effort was less conclusive than originally effects. A proper inclusion of these catalyst properties would Dimerization of Linear Alkenes: A DFT Study J. Phys. Chem. B, Vol. 108, No. 9, 2004 2961 probably destabilize the alkoxides, which are covalently bound highest for 1-butene and almost identical for 2-hexene and quite closely to the surface and thus very sensitive to the surface 3-octene. (iii) The concerted transition states involving primary structure. The physisorbed alkenes are located farther from the carbenium ions are always higher in energy than those involving zeolite surface and also exhibit greater mobility and will very secondary ions. (iv) The order of reactivity for the remaining likely adopt reasonably stable positions regardless of the alkenes are 3,4-dimethyl-2-hexene > 5-methyl-3-heptene > geometric specifics. Some data supporting this view can be 4-methyl-2-pentene > 1-butene, irrespective of reaction mech- found in the literature. Most quantum chemical investigations anism. (v) This holds both when the π- and σ-complexes are on the alkoxide chemisorption (even those in which geo- considered to be the reactant and at every level of theory metrically constrained clusters have been employed) do conclude employed. that the reaction is exothermic,17,25-30 as has also been found in the present study. There are, however, a few interesting 4. Conclusions 24,31 exceptions. Boronat et al. have combined cluster with The dimerization of ethene, propene, 1-butene, and trans-2- periodic calculations specific to the structure of zeolite theta-1 butene has been modeled using quantum chemical methods. and found the reaction to be endothermic, by as much as 60 Both concerted (simultaneous protonation and C-C bond kJ/mol, depending on alkene and calculation methodology. formation) and stepwise (via alkoxides) dimerization have been 34 Rozanska et al. have used periodic methods and found that evaluated. Formation of alkoxides, i.e., alkene chemisorption, steric constraints are very important with respect to the has the lowest activation barrier of the investigated reaction energetics of isobutene adsorption, and that the relative stabilities steps. The barrier of the second step of the stepwise mechanism, of tert-butoxide and a free tert-butyl carbenium ion critically the C-C bond formation step, has a higher activation energy depend on zeolite catalyst structure. In conclusion, we identify than the single barrier of the concerted mechanism. For the alkene chemisorption step as a key reaction in determining concerted dimerization via transition states resembling formally which is the dominating mechanism. Further work, based on primary carbenium ions, the following order of reactivity has the qualitative description of alkene dimerization presented here, been found: trans-2-butene > 1-butene g propene > ethene. is necessary before this issue can be undisputedly settled. Somewhat unexpectedly, the opposite order of reactivity is Some experimental reports focusing on the mechanism of predicted for concerted dimerization via formally secondary alkene dimerization can be found in the literature. Zecchina and carbenium ion-like transition states: propene g 1-butene > co-workers35,36 have studied the dimerization of ethene and trans-2-butene. The stepwise mechanism, where only the most propene over H-mordenite and H-ZSM-5 using IR spectros- stable transition states (secondary) were investigated, resulted copy. A stepwise mechanism was assumed, and this allowed in the following ordering: 1-butene g trans-2-butene > propene the observations made to be satisfactory explained. However, > ethene. Attempts to determine which is the prevailing reaction monomeric alkoxide species were not specifically observed prior mechanism have been inconclusive. More detailed knowledge to dimer/oligomer formation, and a concerted mechanism was concerning the stability of alkoxide species relative to phys- not discussed. Haw et al.37 probed the oligomerization of isorbed alkenes will be necessary for discrimination between propene over HY catalysts using in situ NMR spectroscopy and the two mechanistic proposals. did indeed observe sec-propoxide species. Domen and co- workers38,39 have used IR spectroscopy to investigate iso-butene Acknowledgment. Thanks are due to the Norwegian Re- oligomerization (which has not been studied here, but is search Council for a grant of computer time through the NOTUR considered relevant). Also in this case a monomeric alkoxide project (accounts NN2878K and NN2147K). intermediate was assumed, but could not be observed. However, Trombetta et al.40 did observe tert-butoxide groups during iso- Supporting Information Available: Cartesian coordinates butene oligomerization over H-ZSM-5 at low temperatures, and absolute energies for all stationary points. Pictures of some but no alkoxides were observed when 1-butene or trans-2-butene transition states with displacement vectors. This material is were investigated. The fairly high reactivity of the intermediate available free of charge via the Internet at http://pubs.acs.org. alkoxide groups indicated by these experimental studies tends to contradict the quantum chemical predictions presented here, References and Notes where the barrier for further reaction of the alkoxides are quite (1) O’Connor, C. T.; Kojima, M. Catal. Today 1990, 6, 329-349. high. This further emphasizes the need for a better understanding (2) Golombok, M.; de Bruijn, J. Ind. Eng. Chem. Res. 2000, 39, 267- of the stabilities of the alkoxides, to establish the depth of the 271. (3) Schmerling, L.; Ipatieff, V. N. AdV. Catal. 1950, 2,21-78. energy well, if any, in which these species are located on the (4) Gigstad, I.; Kolboe, S. Propene Oligomerization over Dealuminated PES. Of course, the difficulties in observing the monomeric Mordenite. In Progress in Zeolite and Microporous Materials; Chon, H., alkoxides are in accord with the concerted mechanism, for which Ihm, S.-K., Uh, Y. S., Eds.; Stud. Surf. Sci. Catal. 105; Elsevier Science: - no such intermediate is necessary. Amsterdam, 1997; 965 972. (5) Knifton, J. F.; Sanderson, J. R.; Dai, P. E. Catal. Lett. 1994, 28, 3.4. â-Scission. Because of the incomplete description of the 223-230. PES prior to product desorption from the catalyst surface after (6) Hsia Chen, C. S.; Bridger, R. F. J. Catal. 1996, 161, 687-693. (7) van Niekerk M. J.; O’Connor, C. T.; Fletcher, J. C. Q. Ind. Eng. dimerization, only some information can be derived for the Chem. Res. 1996, 35, 697-702. reverse reaction, the â-scission reaction. It is possible to assess (8) Quann, R. J.; Green, L. A.; Tabak, S. A.; Krambeck, F. J. Ind. the barriers for cracking of the alkenes listed in Table 1 by Eng. Chem. Res. 1988, 27, 565-570. evaluating the energy differences between the dimerization (9) Kojima, M.; Rautenbach, M. W.; O’Connor, C. T. Ind. Eng. Chem. Res. 1988, 27, 248-252. products (as π-orσ-complexes) and the transition state of either (10) Sto¨cker, M. Microporous Mesoporous Mater. 1999, 29,3-48. mechanism. This information can be extracted from Tables 2 (11) Svelle, S.; Arstad, B.; Kolboe, S.; Swang, O. J. Phys. Chem. B and 5, and when doing so the following observations can be 2003, 107, 9281-9289. - made. (i) Since the dimerization reactions are exothermic, the (12) Hay, P. J.; Redondo, A.; Guo, Y. Catal. Today 1999, 50, 517 523. barriers for â-scission are quite high. (ii) Cracking via the (13) Frash, M. V.; Kazansky, V. B.; Rigby, A. M.; van Santen, R. A. J. concerted mechanism and formally primary carbenium ions are Phys. Chem. B. 1998, 102, 2232-2238. 2962 J. Phys. Chem. B, Vol. 108, No. 9, 2004 Svelle et al.

(14) Rigby, A. M.; Kramer, G. J.; van Santen, R. A. J. Catal. 1997, (25) Correa, R. J.; Mota, C. J. A. Phys. Chem. Chem. Phys. 2002, 4, 170,1-10. 375-380. (15) Kazansky, V. B. Catal. Today 1999, 51, 419-434. (26) Evleth, E. M.; Kassab, E.; Jessri, H.; Allavena, M.; Montero, L.; (16) Frash, M. V.; van Santen, R. A. Top. Catal. 1999, 9, 191-205. Sierra, L. R. J. Phys. Chem. B 1996, 100, 11368-11374. (17) Sinclair, P. E.; de Vries, A.; Sherwood, P.; Catlow, C. R. A.; van (27) Viruela-Martı´n, P.; Zicovich-Wilson, C. M.; Corma, A. J. Phys. Santen, R. A. J. Chem. Soc., Faraday Trans. 1998, 94, 3401-3408. Chem. B 1993, 97, 13713-13719. (18) Arstad, B.; Kolboe, S.; Swang, O. J. Phys. Chem. B 2002, 106, (28) Boronat, M.; Viruela, P.; Corma, A. J. Phys. Chem. A 1998, 102, 12722-12726. 982-989. (19) Svelle, S.; Kolboe, S.; Swang, O. J. Phys. Chem. B 2003, 107, (29) Rigby, A. M.; Frash, M. V. J. Mol. Catal. A 1997, 126,61-72. - 5251 5260. (30) Kazansky, V. B.; Senchenya, I. N. J. Catal. 1989, 119, 108-120. (20) Arstad, B.; Kolboe, S.; Swang, O. J. Phys. Chem. B, accepted for (31) Boronat, M.; Zicovich-Wilson, C. M.; Viruela, P.; Corma, A. Chem. publication. Eur. J. 2001, 7, 1295-1303. (21) Rozanska, X.; Saintigny, X.; van Santen, R. A.; Hutschka, F. J. V - (32) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders: V. R. Chem. Re . Catal. 2001, 202, 141 155. 1994, 94, 2095-2160. (22) Rozanska, X.; van Santen, R. A.; Hutschka, F.; Hafner, J. J. Am. - (33) (a) Brønsted, N. Chem. ReV. 1928, 5, 231-338. (b) Evans, M. G.; Chem. Soc. 2001, 123, 7655 7667. - (23) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, M. A.; Polanyi, N. P. Trans. Faraday Soc. 1938, 34,11 29. (34) Rozanska, X.; van Santen, R. A.; Demuth, T.; Hutschka, F.; Hafner, Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; - Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. J. J. Phys. Chem. B 2003, 107, 1309 1315. (35) Spoto, G.; Bordiga, S.; Ricchiardi, G.; Scarano, D.; Zecchina, A.; D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, - M., Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Borello, E. J. Chem. Soc., Faraday Trans. 1994, 90, 2827 2835. Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, (36) Geobaldo, F.; Spoto, G.; Bordiga, S.; Lamberti, C.; Zecchina, A. - D. K.; Rabuck, D. K.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; J. Chem. Soc., Faraday Trans. 1997, 93, 1243 1249. Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, (37) Haw, J. F.; Richardson, B. R.; Oshiro, I. S.; Lazo, N. D.; Speed, J. I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; A. J. Am. Chem. Soc. 1989, 111, 2052-2058. Peng, C. Y.; Nanayakkara, A.; Gonzales, C.; Challacombe, M.; Gill, P. M. (38) Ishikawa, H.; Yoda, E.; Kondo, J. N.; Wakabayashi, F.; Domen, W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, K. J. Phys. Chem. B 1999, 103, 5681-5686. M.; Replogle, E. S.; Pople, J. A. Gaussian98, revision A.11; Gaussian, (39) Kondo, J. N.; Ishikawa, H.; Yoda, E.; Wakabayashi, F.; Domen, Inc.: Pittsburgh, PA, 1998. K. J. Phys. Chem. B 1999, 103, 8538-8543. (24) Boronat, M.; Zicovich-Wilson, C. M.; Viruela, P.; Corma, A. J. (40) Trombetta, M.; Busca, G.; Rossini, S.; Piccoli, V.; Cornaro, U. J. Phys. Chem. B 2001, 105, 11169-11177. Catal. 1997, 168, 349-363. Paper IV

Kinetic Studies of Zeolite-catalyzed Methylation Reactions. 1. Coreaction of [12C]ethene and [13C]methanol

S. Svelle, P.O. Rønning, S. Kolboe, J. Catal. 224 (2004) 115-123

Journal of Catalysis 224 (2004) 115–123 www.elsevier.com/locate/jcat

Kinetic studies of zeolite-catalyzed methylation reactions 1. Coreaction of [12C]ethene and [13C]methanol

Stian Svelle,a Per Ola Rønning,b and Stein Kolboe a,∗

a Department of Chemistry, University of Oslo, P.O. Box 1033 Blindern, N-0315 Oslo, Norway b Oslo University College, Faculty of Engineering, Cort Adelers gate 30, 0254 Oslo, Norway Received 16 January 2004; accepted 17 February 2004

Abstract Coreaction of [13C]methanol and ethene has been carried out over H-ZSM-5 (Si/Al = 45). The catalyst has very small crystals. Most ◦ experiments were carried out at 350 C employing the partial pressures pmethanol = 50 mbar and pethene = 50 mbar. The reactor efﬂuents were analyzed using gas chromatography. Isotopic analysis was carried out using GC-MS. A range of feed rates has been used, up to − WHSV = 292 h 1. This allowed extrapolation of results to zero contact time, giving information about the primary product distribution and the primary isotopic composition of the products. It also allowed the determination of the rate of ethene methylation by methanol. At 12 13 very high feed rates the dominant coreaction product is propene (approaching 90%). At the highest feed rates the C2 C1 isotopomer constituted about 85% of the propene molecules. The reaction order for the methylation of ethene to form propene has been found to be one with respect to ethene and zero with respect to methanol. Measurements have been carried out over an extended range of temperatures, and an Arrhenius plot has been constructed. The apparent activation energy for the methylation of ethene was determined to be 109 kJ/mol. When corrected for the appropriate heat of adsorption for ethene, an intrinsic activation energy of 135 kJ/mol was found. Dimerization of ethene was insigniﬁcant under the investigated reaction conditions. Small amounts of aromatics (mainly xylenes) were always detected. These compounds were very rich in 13C, containing about 85% labeled carbons.  2004 Elsevier Inc. All rights reserved.

Keywords: H-ZSM-5; Zeolite; Methanol; Ethene; Methylation; Kinetics; Activation energy; Hydrocarbon pool; Methanol-to-hydrocarbons; Isotopic labeling

1. Introduction considerable effort, the reaction mechanism of the MTH reaction can still not be regarded a fully settled issue. Since The ability of protonated ZSM-5 (H-ZSM-5) to convert there are extensive reviews, no attempt is made here to give methanol to hydrocarbons (MTH) in the range C2–C10 and an overview of the large number of papers on the MTH re- water was discovered and published in 1976. Later it has action and the various mechanistic proposals. been found that also other protonated zeolites may have this Over the last few years, the “hydrocarbon pool” mech- ability, but they usually deactivate much faster. Industrial anism has gained acceptance. As originally proposed by processes (methanol to gasoline; MTG, and methanol to Dahl and Kolboe [4–6], the reaction proceeds via an ad- olefins; MTO) utilizing this reaction have been developed. sorbate that continually adds reactants and splits off prod- Since the discovery of this reaction a very large number ucts, in particular C2–C4 alkenes. Experiments where ben- of papers have been published covering virtually every as- zene or toluene and [13C]methanol were coreacted have been pect of the reaction—a majority being devoted to clarify the performed, and it was clear that ethene, propene, and the mechanistic aspects. During the years some 20–30 different arenes in the effluent had indistinguishable isotopic distri- mechanisms have been proposed. An early comprehensive butions [7]. This strongly indicated that methanol conver- review was given by Chang, the discoverer of the reaction sion could proceed via repeated methylations and dealky- system [1]. Recently, two quite comprehensive reviews have lations of aromatic reaction centers. The specific nature of been given by Stöcker [2] and by Chang [3]. Despite the very the hydrocarbon pool has been clarified in a series of publications from Kolboe and co-workers [8–11], Haw and * Corresponding author. Fax. +47 22 85 54 41. co-workers [12–17] and Hunger and co-workers [18,19]. E-mail address: [email protected] (S. Kolboe). It has become clear that polymethylbenzenes, polymethyl-

0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved. doi:10.1016/j.jcat.2004.02.022 116 S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123 naphthalenes, and polymethylated cyclopentenyl ions are the We have also determined the apparent activation energy main constituents of the hydrocarbon pool. for the methylation step and obtained an estimate for the At the same time it is clear that alkenes, which are im- “true” activation energy. The results are compared with re- portant components in the product stream, may be methy- cent theoretical reports on the zeolite-catalyzed methylation lated by methanol, once or several times, thereby creating of a wide range of alkenes [22] and methylbenzenes [23]. alkenes that are easily cracked to smaller alkenes that are Further experimental work with alkenes other than ethene is again methylated. These reactions might constitute a com- in progress. peting reaction pathway. Knowledge of the rate of alkene methylation is therefore an important issue. The importance of this alkene methylation/cracking pathway was in particu- 2. Experimental lar emphasized by Dessau and co-workers [20,21]. We have previously investigated the reaction system 2.1. Catalyst where [13C]methanol is coreacted with ethene or propene over a SAPO-34 catalyst [4,5], the most promising system The H-ZSM-5 sample used was a gift from Süd-Chemie / for an MTO process. This catalyst is very selective toward AG. The Si Al ratio is 45. The crystal size is very small, ethene and propene formation. The pronounced selectivity about 50 nm as determined by X-ray line broadening and TEM. This catalyst is quite resistant toward deactivation and for C2–C4 formation displayed by the SAPO-34 catalyst is due to the narrow 8-ring pores, which prevent any branched the experiments have been performed to minimize and even out any changes in activity over time. A 10 h experiment molecules to diffuse out of the catalyst particles into the sur- gave about a 15–20% decrease in catalyst activity. Conver- rounding gas phase. This study gave no information on the sions, and thus also rates, have been corrected for this small rate of ethene methylation by methanol, and the very narrow deactivation according to the method outlined by Dahl and pore structure might influence the reactions in a special way. Kolboe [5,6], but our conclusions do not depend upon this For this reason we consider it necessary to extend the inves- correction. tigation to include also the ZSM-5 catalyst where a much higher diffusivity is present; in addition it is the MTH cata- 2.2. Reagents lyst archetype. The rate of alkene methylation by methanol in acidic zeo- Ethene with a stated purity > 98% was purchased from type catalysts is not easily monitored, and to our knowledge Fluka. Gas chromatography (GC-FID)1 showed ethane to no such measurements have been carried out. The reason be the main hydrocarbon impurity (∼ 0.06%). No higher for the difficulty of performing these measurements is that alkenes could be detected (detection limit: 0.002%). [13C] alkene interconversion reactions (addition, metathesis and Methanol was supplied by ISOTEC at a stated chemical and cracking) are not easily distinguished from the reactions isotopic purity > 99%. GC-MS analysis2 showed that traces + caused by methanol. The alkene alkene reactions might of ethanol (about 0.2% v/v) were present in the methanol. even eclipse the methanol methylation. In addition to this The isotopic purity was investigated by converting the la- difficulty it must also be kept in mind that the reactions beled methanol to hydrocarbons over a zeolite catalyst (in caused by the hydrocarbon pool might well be just as fast, order to remove water and avoid the complications caused by + or faster than the methanol alkene reaction. the unknown and quite large content of 18O) and analyzing 13 By utilizing a reaction system consisting of [ C]metha- the isotopic composition of the main products as described 12 nol and [ C]alkene, and choosing the conditions so that below. The average content of 13C atoms was thus found to only a minimal conversion to products take place, such infor- be 98%. mation may be obtained. Measuring the rate of methylation of ethene by methanol is the main objective of this paper. In 2.3. Reaction conditions the present work, we have studied the coreaction of ethene and 13C-labeled methanol. All catalytic reactions were performed in a fixed-bed Preliminary experiments indicated that 350 ◦C is the op- Pyrex microreactor (3 mm i.d.). Ethene was fed as a gas, timal temperature for extracting the desired information. Ex- using a needle valve flow regulator and a mass-flow me- treme feed rates have been employed in order to investigate ter. Methanol was fed by passing part of the carrier gas (He, the primarily formed products and to explicitly study the in- > 99.996%) through a vessel containing the methanol, thus dividual reaction steps that can occur. Methylation of ethene saturating the carrier gas. The partial pressures of ethene and to form propene was then by far the most prominent reac- methanol were individually varied from 10 to 100 mbar. To- tion. Dimerization of ethene to form butenes was at best a tal pressure equaled atmospheric pressure. In order to reduce very minor, and negligible, reaction. In addition to methylation, a separate reaction pathway, leading to products quite 1 Analysis performed with a Siemens Sichromat 2-8 (FID) equipped rich in 13C was found to be operative. In our experimental with a Chrompack PLOT column (Al2O3/KCl, 50 m × 0.53 mm × 10 µm). range, the methylation is zero order with respect to methanol 2 Analysis performed with a Thermoquest GC-MS equipped with a and first order with respect to ethene. J&W DB-Wax column (30 m × 0.25 mm × 25 µm). S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123 117 the expenditure of [13C]methanol, the feed was admitted to 3. Results the catalyst for 15 min prior to each analysis. After taking a sample of the effluent for analysis, the feed was stopped The main objective for the present work where [13C]me- (maintaining the carrier gas flow) and the conditions ware thanol and [12C]ethene are coreacted is to collect kinetic data adjusted to those desired for the next analysis. Separate tests for the methylation of ethene by methanol and to further indicated that admitting methanol for 15 min prior to analyz- monitor the role of ethene in the MTH reaction. Attention ing the effluent was sufficient to reach steady-state activity. will focus on experiments where high feed rates have been The reaction temperature was measured with a thermocou- employed, leading to very low conversions to products. At ple (0.5 mm diameter) placed in the catalyst bed. Most ex- lower feed rates where higher degrees of reactant conver- periments were done at 350 ◦C, but reaction temperatures sions are obtained, secondary reactions will eventually dom- between 305 and 410 ◦C were investigated. Total gas flow inate, thus making information about the primary processes through the reactor was 10–100 mL/min giving feed rates unavailable. By using a very small amount of catalyst and (WHSV) in the range 18–292 h−1. 2.5 mg of catalyst was high gas flows, it was possible to achieve such low conver- used to reach the extremely high feed rates necessary to ob- sions at reaction temperatures and a catalyst acid site density tain information about the primary reaction products and to that are quite similar to those usually employed for practical minimize secondary reactions. Despite having only a thin MTH purposes. layer of catalyst on the glass sinter in the reactor, reactant The reactivities of the individual reactants were investi- bypass was negligible. Control experiments carried out at gated prior to the cofeeding experiments. Both methanol and the highest feed rates with 2-butanol as feed, resulted in ethene displayed low reactivity when fed alone, and prod- complete dehydration of the alcohol to form butenes, thus uct formation was negligible compared to the coreaction verifying that bypass was indeed insignificant. experiments except at the lowest feed rate. The conversion rate of methanol increases very markedly with increasing conversion (the well-known autocatalysis effect). The ratios 2.4. Analysis between the amount of propene formed in the coreaction experiments and that formed when neat methanol was fed Product analysis was performed using gas chromatogra- (50 mbar methanol and 2.5 mg catalyst in both cases) were phy. Quantitative effluent composition was determined us- ≈ 400, 250, and 3.1 for the carrier gas feed rates: 30, 20, ing an on-line Carlo Erba GC6000 Vega with flame ion- and10mL/min. Propene was the main product. At still ization detector (FID) equipped with a Supelco SPB-5 col- lower feed rates (lower total gas flows) the conversion of × × umn (60 m 0.53 mm 3 µm). Additional analyses were neat methanol appears to approach the conversions observed performed on a Siemens Sichromat 2-8 or an HP 6890 in the coreaction experiments. Measurements at the high- equipped with a Chrompack PLOT column (Al2O3/KCl, est feed rates may be rendered invalid by the most minute × × 50 m 0.53 mm 10 µm), both with FID. This setup al- traces of arene impurities in the methanol. A somewhat more lowed separation of all C1–C6 alkenes and alkanes. Products detailed treatment of methanol conversion is given in Sec- up to C9 are eluted, if present. tion 3.5. Isotopic composition of the products was determined using an HP 6890 GC with an HP 5973 mass-sensitive detec- 3.1. The effect of feed rate (contact time) tor (GC-MS). Using cryostatic cooling the HP-5MS column (60 m × 250 µm × 0.25 µm) gave adequate separation of all Afeedmixconsistingof50mbar[13C]methanol and the compounds of interest in this work. Ethene was always 50 mbar [12C]ethene was coreacted at a constant reaction 12 dominated by the C2H4 isotopomer from unconverted feed and not amenable for further analysis.

2.5. Calculations

The computational method used for determining the isotopic composition of the products has been outlined previously [4,24]. In order to extract the isotopic composition of a compound it is necessary to know the mass spectrum of the ordinary 12C compound (correction for natural 1.1% 13Cis easily carried out). Standard spectra were obtained by reacting ordinary methanol over the catalyst. To ensure reliable isotopic analysis of the products new standard mass spectra were recorded at intervals. The GC-MS system showed Fig. 1. Conversion of feed mixture: 50 mbar ethene coreacted with 50 mbar ◦ great stability, and variations in the standard spectra were methanol; reaction temperature = 350 C; WHSV varied from 29.4 to − negligible. 294 h 1. 118 S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123

not the only one. Fig. 3 shows the selectivities for the main products. The propene dominance is clearly seen. Extrap- olated to zero CT a propene selectivity of nearly 90% is observed. Since the experiment was carried out with [13C]methanol and [12C]alkene, the isotopic composition of the hydrocarbon products is of interest. The distribution of the various isotopomers of propene when the contact time is varied is shown in Fig. 4. The isotopomer with one 13Candtwo12C atoms is always dominating, and at the lowest CTs it reaches about 85%, which may also be the value obtained by extrapolation to zero CT. It is noteworthy that the second most 13 Fig. 2. Conversion rate: 50 mbar ethene coreacted with 50 mbar methanol; common isotopomer is the C3 species. Fig. 4 suggests that ◦ − reaction temperature = 350 C; WHSV varied from 29.4 to 294 h 1.The the amount of this isotopomer may be extrapolated to about dotted curve shows the extrapolation to CT = 0. 8% at CT = 0. The figure also shows that when the contact time is increased, the amount of this isotopomer increases strongly, and that at the highest contact time, CT = 0.034 h, 12 13 it is present in nearly the same amount as the C2 C1 species. Other experiments not discussed here, and carried out at considerably higher contact times, have shown that this increase continues until the larger part of the methanol has reacted. Under slightly different conditions the following isotopomer distribution was found at a feed rate correspond- 12 13 12 13 13 ingtoCT≈ 0.1h: C2 C1; 22%, C1 C2; 22%, C3; 55%. In this case 96% of the ethene in the effluent still had the original (natural) isotopic composition, but 69% of the methanol was converted to hydrocarbons [24]. The experiment shows that under conditions with a rather high conversion the larger part of the methanol conversion does not involve the ethene. Fig. 3. Product selectivities: 50 mbar ethene coreacted with 50 mbar meth- ◦ − anol; reaction temperature = 350 C; WHSV varied from 29.4 to 294 h 1. The dotted curve shows the extrapolation to CT = 0. temperature of 350 ◦C. 2.5 mg catalyst was used, and the total gas flow was varied from 10 to 100 mL/min, thus varying the WHSV between 29.4 and 294 h−1. The “contact time” (CT), defined as 1/WHSV, is a quantity better suited for the further discussion, and it will be used throughout. The space velocity range just mentioned corresponds to the CT range 0.034–0.0034 h. The conversion to hydrocarbon products obtained in this experiment is displayed in Fig. 1. The degree of conversion is obtained by taking all C atoms in hydrocarbons, except ethene, and then dividing by the total number of C atoms in a given analysis; i.e., ethene, methanol, and dimethyl ether are considered to be unconverted reactant. Note that the maximum conversion in Fig. 1 is about 7.5%. Know- ing the conversion and the feed rate, the rate of hydrocarbon formation (C atoms) is obtained by dividing the conversion by CT. The result is displayed in Fig. 2. A clear increase in reaction rate with increasing CT is observed, showing the well-known autocatalysis effect [1–3]. The data in Fig. 2 al- Fig. 4. Isotopic composition of alkene products in the effluent: 50 mbar low extrapolation to infinite feed rate, or zero contact time. [12C]ethene coreacted with 50 mbar [13C]methanol; reaction tempera- ◦ − Even though propene was the dominating product, it was ture = 350 C; WHSV varied from 29.4 to 294 h 1. S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123 119

Fig. 5. Rate of propene formation vs methanol partial pressure: 50 mbar = ethene coreacted with 20–100 mbar methanol; reaction temperature Fig. 7. Rate of propene formation vs ethene partial pressure: 10–105 mbar ◦ = ◦ 350 C; total gas flow 100 mL/min. ethene coreacted with 50 mbar methanol; reaction temperature = 350 C; total gas flow = 100 mL/min. 3.2. The effect of methanol partial pressure way for product formation, leading to alkenes with higher The effect of varying the methanol concentration in the contents of labeled atoms is also operative. The joint p/m- feed mix was investigated by employing the following re- xylene product fraction was always very rich in 13C. action conditions: Methanol partial pressure was varied between 20 and 100 mbar (WHSV = 199–448 h−1); ethene 3.3. The effect of ethene partial pressure partial pressure was fixed at 50 mbar; 2.5 mg catalyst was used; total gas flow through the reactor was held constant at The effect of varying the ethene partial pressure was 100 mL/min; the reaction temperature was 350 ◦C. probed in an experiment analogous to that described in Sec- The product distribution was independent of the methanol tion 3.2: Ethene pressure was varied between 10 and 105 partial pressure, i.e., the same as shown in Fig. 3 at CT = mbar (WHSV = 183–442 h−1); methanol partial pressure 0.0034 h, and will not be described further. The concentra- was fixed at 50 mbar; 2.5 mg catalyst was used; total gas flow tion of hydrocarbons in the effluent was independent of the through the reactor was held constant at 100 mL/min; the methanol partial pressure. With a constant gas flow rate this reaction temperature was 350 ◦C. Again, no significant sys- implies that the rate of hydrocarbon formation was indepen- tematic trend in product distribution was found. The concen- dent of the methanol pressure, i.e., the reaction order with tration of product hydrocarbons in the effluent was propor- respect to methanol is zero. This is shown in Fig. 5. tional to the ethene partial pressure. Hence, with a constant The isotopic compositions of the products in the efflu- gas flow, the rate was proportional to the ethene pressure. ent were analyzed as above and the results are displayed The reaction is first order with respect to ethene, as is seen in Fig. 6. Evidently, the isotopic distributions of the vari- in Fig. 7. ous products are virtually unaffected by the changes in the Isotopic analysis of the products was also carried out methanol pressure. The isotopic distributions that are seen here. No striking effect of varying the ethene pressure was from the histograms in Fig. 6 closely resemble those re- observed, but the fraction of isotopomers with only 13C ported for high feed rates in Fig. 4. Methylation reactions declined slightly with increasing ethene partial pressure. In 13 are again dominating, but a second, less dominating path- the case of propene, there was almost 10% of the C3 iso-

Fig. 6. Isotopic composition of alkene products in the effluent vs methanol partial pressure. 50 mbar [12C]ethene coreacted with 20–100 mbar [13C]methanol; ◦ reaction temperature = 350 C; total gas flow = 100 mL/min. Additional data points omitted. 120 S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123 topomer at pethene = 10 mbar. When pethene = 105 mbar 3.5. The minor products this fraction fell to about 6%. The minor products displayed a similar trend. It was shown above in Fig. 3 that propene is the main product, which at the lowest CT values constitutes nearly 3.4. The effect of reaction temperature 90% of the products. The remaining 10% comprise many products. They all contain more 13C than propene, but their → The reaction temperature was varied in order to investi- limiting behavior when CT 0 appears rather diverse and gate whether different temperatures would lead to changes in merits a short discussion. For this reason a part of Fig. 3 is product selectivities or isotopomer distributions, and to con- reproduced with a much expanded ordinate scale in Fig. 9. struct an Arrhenius plot and determine the apparent activa- Butenes are separated in two groups. n-Butenes are lumped tion energy for the methylation of ethene to propene. A feed separate from isobutene because the two groups have widely mixconsistingof50mbar[13C]methanol and 50 mbar different isotopic distributions. C5 molecules are lumped, [12C]ethene was coreacted over 2.5 mg catalyst at a total but isotopic data have been obtained only for 2-methyl-2- gas flow of 100 mL/min (WHSV = 292 h−1). The tempera- butene. C6+ molecules are lumped. Isotopic data are avail- ture was varied between 305 and 410 ◦C. This range covers able for one hexene isomer. The isotopic distributions of the minute amounts of p-xylene and 1,2,4-trimethylbenzene realistic and practical MTH conditions. present in the effluent could be investigated down to CT = Reactant conversions, rates of products formation, and 0.0067 h. This analysis was possible because of the simple isotopic compositions were measured. The effect of increas- mass spectra of the polymethylbenzenes, which are domi- ing the temperature, while keeping the contact time constant, nated by the molecular ion. Isobutene was not adequately on product and the isotopic distributions was much the same ◦ separated from 1-butene to allow rigorous determination of as that of increasing the contact time at 350 C. The product ◦ the isotopic composition. However, an estimate could be ex- distribution and conversion at 400 C was virtually identical ◦ tracted and this is listed in Table 1 together with the results to the one obtained at 350 C at CT values somewhere in the for the aromatics at CT = 0.0067 h. Random distributions range 0.017–0.022 h. for the observed total contents of 13C atoms are included Our primary interest is to determine the activation energy for comparison. The experimental distributions are seen to for methylation of ethene by methanol, i.e., the activation be fairly close to randomness. The all-13C isotopomers are, energy for formation of the propene isotopomer with one however, in all cases in clear excess relative to the random 13C atom. An Arrhenius plot showing the rate of formation distribution. For all three hydrocarbons 13C atoms constitute of this isotopomer is given in Fig. 8. Even at 410 ◦C, this more than 80% of the carbon atoms. Measurements at the isotopomer still constitutes more than 60% of the propene longer CTs indicated little variation when CT was changed. molecules, but at this high temperature one might expect In the CT → 0 limit the n-butenes, in particular, and also that part of the propene molecules take part in secondary the pentenes remain a sizable fraction of the product hydro- reactions. The apparent activation energy 103 kJ/mol that carbons, whereas the C6+ fraction is reduced to insignifi- is extracted from the Arrhenius plot is then somewhat on cance. The isobutene and alkane fractions are also reduced, the low side, but even if the activation energy is esti- but perhaps less clearly so than C6+. mated on basis of the total amount of propene, the estimate Fig. 4 shows that at all but the longest contact times does not rise beyond 114 kJ/mol. The best estimate for 12 13 the C2 C2 n-butene isotopomer, corresponding to double the apparent activation energy might be the mean value, methylation, is clearly dominating. The second most com- 109 kJ/mol. 13 mon species is the C4 isotopomer, which at the longest

Fig. 8. Arrhenius plot for the formation of singly labeled propene: 50 mbar ethene coreacted with 50 mbar methanol; CT = 0.0034 h; reaction temper- Fig. 9. Product selectivities with expanded Y axis. Propene is outside the ◦ ature varied from 305 to 410 C. scale. Reaction conditions as described in Fig. 3. S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123 121

Table 1 and propene, but no kinetic data were given, and their ap- Isotopic composition of isobutene, p/m-xylene, and trimethylbenzene proach only permitted the determination of the total label Number of Isobutene p/m-Xylene Trimethylbenzene content. Iglesia et al. [31] combined [13C]methanol with or- 13Catoms dinary propene over H-ZSM-5 and did find a large share of 02(0) 0 (0) 0 (0) singly labeled butenes, but again the kinetics was not in- 13(2) 1 (0) 0 (0) vestigated. Fairly recently, Rønning et al. [32] published an 213(14) 1 (0) 1 (0) investigation of the [12C]ethene/[13C]methanol system, but 333(40) 2 (1) 1 (0) 449(43) 6 (3) 2 (1) the conversions were fairly high and the reaction conditions 512(12) 6 (4) were thus not well suited for kinetic studies. 622(27) 13 (14) The results described in the previous section show that by 724(36) 22 (29) choosing the appropriate reaction conditions, methylation of 833(21) 26 (34) ethene may be made, by far, the most prominent reaction, 929(18) and that the rate of methylation can be measured. The lim- Total 13C content: 81 82 82 iting rate of formation of the propene isotopomer with one Random distributions for the given total label contents are displayed in 13C atom gives the rate of methylation. Remembering that parentheses. 50 mbar [12C]ethene coreacted with 50 mbar [13C]methanol; ◦ the reaction is of zero order with respect to methanol and reaction temperature = 350 C; CT = 0.0067 h. first order with respect to ethene, the kinetic equation describing the methylation of ethene with methanol is given by contact times becomes the most prominent. Other experi- Eq. (1). ments that are not discussed here at still longer contact times r = kp0 p1 (1) have shown that molecules with 4 and 3 13Catomseven- methanol ethene tually become completely dominating at higher methanol The extrapolated total reaction rate was shown in Fig. 2 ◦ conversions [24]. to be 0.70 gproduct/(gcatalyst h) at 350 Candpethene = = The C5 molecules show a similar behavior, but this time pmethanol 50 mbar. The limiting selectivity propene was 12 13 the most common isotopomer at small CTs is the C2 C3 shown in Fig. 3 to be 0.90, and the limiting isotopic selec- 12 13 species. At slightly longer contact times molecules with 5 tivity to the C2 C isotopomer was 0.85; thus the rate of and 4 13C atoms dominate. formation of the clean methylation product is (0.90 × 0.85 × Except at the longest contact time, i.e., CT = 0.034 h, 0.70 = 0.5355) 0.54 g/(g h) or 0.013 mol/(g h). The value − the sum of all products formed when methanol was the only of k then is 2.6 × 10 4 mol/(g h mbar). reagent was smaller than any of the products discussed in By varying the reaction temperature an apparent activa- Fig. 9. tion energy of 109 kJ/mol was found. Assuming that the reaction orders are unchanged over the investigated temperature range, this observed activation energy should be 4. Discussion equal to the sum of the true activation energy and the ethene adsorption enthalpy. We have previously published The results described above indicate that two routes for a theoretical report on the mechanism of zeolite-catalyzed product formation are operative when ethene and methanol methylation of alkenes, using a cluster consisting of four are coreacted over H-ZSM-5. Quite clearly, ethene is methy- tetrahedral atoms (three Si and one Al) to model the cata- 13 12 lated to form C1 C2 propene. Multiple methylations to lyst and utilizing density-functional theory (DFT) to carry higher alkenes are also observed. In addition, another mech- out the quantum chemical calculations [22]. According to anism is operative, leading to products very rich in 13C. the calculations the reaction takes place when a methanol molecule is adsorbed end-on onto an acidic site while the 4.1. Methylation of ethene alkene is adsorbed in close proximity on a siliceous part of the zeolite, and it proceeds in a concerted one-step process. Homologation of alkenes via methylation was fairly The same mechanism has been proposed for methylation early proposed to be key reaction steps when methanol is of arenes by methanol [23]. Hence, the theory suggests converted to hydrocarbons over acidic zeolites [20,21,25]. that the methylation reaction takes place by a Langmuir– Several reports on the coreaction of methanol and various Hinshelwood–Hougen–Watson (LHHW) reaction mecha- alkenes exist, verifying that such reactions may occur. Wu nism type. Based on this mechanism, the experimentally and Kaeding [26] coreacted unlabeled ethene and methanol determined adsorption enthalpy of ethene in silicalite-1, over H-ZSM-5, and an enhanced propene production was the completely siliceous MFI polymorph, may be used as observed, indicative of ethene methylation. Behrsing et a correction in order to obtain the true activation energy. al. [27] coreacted unlabeled hexenes with [13C]methanol Choudhary and Mayadevi [33] found an isosteric heat of over H-ZSM-5 and found considerable amounts of mono- adsorption of 25 kJ/mol for ethene on silicalite-1, and this 14 labeled C7 products. Tau and co-workers [28–30] used C value was found to be virtually independent of the ethene labeled reactants to investigate the methylation of ethene coverage. Assuming that the heat of adsorption is unaffected 122 S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123 by the presence of methanol as a coadsorbate, the intrin- To what extent does the reaction proceed via ethene sic activation barrier for the H-ZSM-5-catalyzed methyla- methylation when ethene is not added as a coreactant? The tion of ethene by methanol should be ∼ 135 kJ/mol. This data presented here show that although the methylation of is substantially lower than the value reported in our DFT ethene does proceed at a nonnegligible rate, it was eclipsed study [22], where an activation barrier of 183 kJ/mol was by other pathways once there was an appreciable amount of found for methylation of ethene. This discrepancy is, how- hydrocarbon products. It was shown (Sections 3.1 and 3.5) ever, quite expected. It is well known that the DFT/cluster that although the number of 12C atoms in the feed was dou- approach severely overestimates the energies of transition ble that of 13C, there was a large excess of 13C atoms in the states and charged species, in some cases by as much as hydrocarbon products when the CT was increased. 100 kJ/mol [34,35]. Bearing this in mind, the agreement be- When methanol alone is reacted over a ZSM-5 catalyst, tween experiment and theory is acceptable. ethene is usually a relatively small product. Using the same ◦ Equation (1), as written, is not a LHHW kinetic equa- catalyst at 350 C, as here, and a methanol partial pressure tion. The adsorption terms are missing. The experimentally 100 mbar only 6% of the products were ethene at 40% con- found zero and first-order behavior with respect to methanol version [24]. The ethene partial pressure in the effluent was and ethene means that in our experimental range, methanol then 1.2 mbar. Since the methylation rate is first order with is adsorbed on all acid sites, and ethene is extremely sparsely respect to ethene pressure, the methylation rate was then = adsorbed. Equation (1) can only be expected to describe only 2.5% of the rate obtained when pethene 50 mbar. It the reaction as long as this is valid; i.e., the equation may may therefore be concluded that ethene methylation is only a minor pathway in the MTH reaction. However, if the re- break down when pmetanol 20 mbar, and when pethene 100 mbar. actant concentrations are increased by one or two orders of The rate constant k of Eq. (1) is given by the Arrhe- magnitude this conclusion may be less valid. nius expression k = Aexp(−Eapparent/RT ).Thevalueofk at 350 ◦Cis2.6 × 10−4 mol/(g h mbar); hence A has the 4.2. Butenes and pentenes value 3.5 × 105 mol/(g h mbar) and k(T ) = 3.5 × 105 × − It was shown in Section 3.5 that n-butenes and pentenes exp(−13100/T K 1)mol/(g h mbar). remain a sizable fraction of the product hydrocarbons in The Si/Al ratio of the catalyst used here is 45, which the CT → 0 limit. The most common isotopomers have means that the stoichiometric formula of the zeolite may be two, respectively three, 13C atoms, as one should expect written (SiO ) (AlO )H, giving a molar mass 2760 g/mol; 2 45 2 if they are obtained by methylation once or twice of the i.e., there is 3.6 × 10−4 mol/g. There is one proton per propene. A question arises. Are these products primary prod- molecular unit, so this is also the molar content of active ucts, or are they secondary products formed by methylation sites per gram catalyst. The preexponential factor A there- of propene after its desorption from the site where it was fore has the value (3.5 × 105/3.6 × 10−4 = 9.7 × 108) formed? A = . × 8 / = . × 8 −1 = 9 7 10 mol (mol h mbar) 9 7 10 (h mbar) From the 13C atom contents in butene and pentene it × 5 −1 2.7 10 (s mbar) . is tempting to conclude that they are formed by secondary Somewhat simplistically one may say that the preexpo- and tertiary methylations. This may, however, be wrong. nential factor expresses the rate that would be obtained if In a series of consecutive (irreversible) first-order reactions the activation energy were negligible, and it expresses the A→B→C→D (ethene→propene→butene→pentene), it is maximum rate that might be observed for a given reaction. known that in the limit of short reaction time, t, [B] is pro- The maximum rate that may be imagined for this reaction is portional to t, [C] is proportional to t2,and[D]tot3.Inthe given by the rate by which the ethene molecules hit an ac- limit of t → 0 one will therefore have the concentration ra- tive site. It is therefore of interest to compare the value of tios [C]/[B]→0and[D]/[C]→0. The observed ratios did A, given above, with this collision frequency. According to not display this limiting behavior. simple kinetic theory of gases, the number of collisions per If, however, the subsequent methylations of propene take = 1/2 unit time and unit area is given by Zw p/(2πmkT) .If place before the propene molecules can leave the zeolite 2 the area of an active site is taken to be 0.5 nm , the formula crystal where they were formed and join the bulk gas phase ◦ gives that at 350 C, i.e., 628 K and pethene = 1.00 mbar the the concentration ratios become independent of the contact − collision frequency per active site is 1.0 × 106 s 1,which time, CT, because the mean residence time within a crystal − compares very well with the value A = 2.7×105 (s mbar) 1. is independent of the flow rate. The values obtained above for the rate constant as a func- Alternatively, parts of the butenes and pentenes might be tion of temperature, k(T ), and the preexponential, A,ex- primary products formed from the hydrocarbon pool. Such a pressed per acid site, should in principle be transferable to formation mechanism would also be in concordance with the other H-ZSM-5 catalysts operating at different partial pres- finding that the alkene molecules appear to come from two sures, reaction temperatures, and Si/Al ratios. However, it sources. The fraction of propene isotopomers with two or should be kept in mind that these rate parameters might not three 13C atoms would also fit in with this explanation, see be valid for other catalyst frameworks than MFI. Fig. 4. Their composition is in agreement with the known S. Svelle et al. / Journal of Catalysis 224 (2004) 115–123 123 isotopic composition of the arenes, Table 1. No conclu- Under ordinary MTH reaction conditions, formation of sion is presently warranted on this issue. Fig. 4 shows that higher alkenes via homologation, starting with ethene, is not linear extrapolation to the CT→ 0 limit indicates that the an important reaction. 12 13 13 two n-butene isotopomers, C1 C3 and C4, constitute 20–25% of the n-butenes. The two 2-methyl-2-butene iso- 12 13 13 topomers, C1 C4 and C5, constitute about 45% of the Acknowledgments 2-methyl-2-butene. These isotopomers could not be formed by repeated methylations of ethene. Fig. 9 shows that the We thank Süd-Chemie AG for the gift of the ZSM-5 sample and Dr. Ø. Mikkelsen for fruitful discussions. limiting value of the fraction of C6+ products is zero, so the 13C atom rich n-butene and 2-methyl-2-butene isotopomers are not cracking products of C +. Fig. 4 suggests that the 6 References real isotopomer limiting fractions are not obtained correctly 13 by linear extrapolation; the C content is even higher. We [1] C.D. Chang, Catal. Rev.-Sci. Eng. 25 (1983) 1. do not know if this is an artifact, whose origin is unknown, [2] M. Stöcker, Micropor. Mesopor. Mater. 29 (1999) 3. or if it is true. Even if it is a real effect, the conclusions ar- [3] C.D. Chang, in: C. Song, J.M. Garcés, Y. Sugi (Eds.), Shape Selective rived at above become still more valid. Catalysis, Am. Chem. Society, Washington, DC, 2000, p. 96. Interestingly, the isotopic composition of isobutene was [4] I.M. Dahl, S. Kolboe, Catal. Lett. 20 (1993) 329. [5] I.M. Dahl, S. Kolboe, J. Catal. 149 (1994) 458. completely different from that of the linear butenes. The [6] I.M. Dahl, S. Kolboe, J. Catal. 161 (1996) 304. 13 content of C atoms is much higher, and the isotopomer [7] Ø. Mikkelsen, P.O. Rønning, S. Kolboe, Micropor. Mesopor. Mater. 40 distribution is random. It seems probable that isobutene is (2000) 95. formed through the hydrocarbon pool mechanism. This is in [8] B. Arstad, S. Kolboe, Catal. Lett. 71 (2001) 209. excellent accord with data reported by Bjørgen et al. [11,36], [9] B. Arstad, S. Kolboe, J. Am. Chem. Soc. 123 (2001) 8137. which show that isobutene/isobutane is the most prominent [10] S. Kolboe, in: A. Galarneau, F. Di Renzo, F. Fajula, J. Vedrine (Eds.), Zeolites and Mesoporous Materials at the Dawn of the 21st Century, product formed when polymethylbenzenes are reacted over Elsevier Science, Amsterdam, 2001, p. 3946. zeolite H-beta. [11] M. Bjørgen, U. Olsbye, S. Kolboe, J. Catal. 215 (2003) 30. [12] J.F. Haw, J.B. Nicholas, W. Song, F. Deng, Z. Wang, T. Xu, C.S. He- neghan, J. Am. Chem. Soc. 122 (2000) 4763. 5. Conclusions [13] J.F. Haw, W. Song, D.M. Marcus, J.B. Nicholas, Acc. Chem. Res. 36 (2003) 317. The coreaction of ethene and [13C]methanol over [14] P.W. Gougen, T. Xu, D.H. Barich, T.W. Skloss, W. Song, Z. Wang, J.B. Nicholas, J. Am. Chem. Soc. 120 (1998) 2650. H-ZSM-5 has been studied at extremely high feed rates. [15] A. Sassi, M.A. Wildman, H.J. Ahn, P. Prasad, J.B. Nicholas, C.S. By using a microreactor, it has been possible to collect data Heneghan, J. Phys. Chem. B 106 (2002) 2294. at very low conversions, but at a realistic reaction temper- [16] W. Song, J.F. Haw, J.B. Nicholas, C.S. Heneghan, J. Am. Chem. ature and catalyst acid site density. The isotopic labeling Soc. 122 (2000) 10726. showed that methylation of ethene to form singly labeled [17] W. Song, H. Fu, J.F. Haw, J. Phys. Chem. B 105 (2001) 12839. [18] M. Seiler, U. Schenk, M. Hunger, Catal. Lett. 62 (1999) 139. propene was dominating at the lowest conversions. Further [19] M. Hunger, M. Seiler, A. Buchholz, Catal. Lett. 74 (2001) 62. methylations to form doubly labeled butenes and triply la- [20] R.M. Dessau, R.B. LaPierre, J. Catal. 78 (1982) 136. beled pentenes were also observed. However, more than [21] R.M. Dessau, J. Catal. 99 (1986) 111. 20% of the linear butenes, and more than 40% of the pen- [22] S. Svelle, B. Arstad, S. Kolboe, O. Swang, J. Phys. Chem. B 107 tene, 2-methyl-2-butene, contain more than 2, respectively (2003) 9281. 3, 13C atoms, and are not formed by multiple methylations [23] B. Arstad, S. Kolboe, O. Swang, J. Phys. Chem. B 106 (2002) 12722. [24] P.O. Rønning, Ph.D. thesis, University of Oslo, 1998. of ethene. In the limit CT → 0theC + product fraction be- 6 [25] N.Y. Chen, W.J. Reagan, J. Catal. 59 (1979) 123. comes zero, and in this limiting case, butenes and pentenes [26] M.M. Wu, W.W. Kaeding, J. Catal. 88 (1984) 478. are therefore not the result of cracking of higher alkenes. [27] T. Behrsing, T. Mole, P. Smart, R.J. Western, J. Catal. 102 (1986) 151. The methylation of ethene is zero order with respect to [28] L.-M. Tau, A.W. Fort, S. Bao, B.H. Davis, Fuel Process. Technol. 26 methanol partial pressure and first order with respect to (1990) 209. ethene partial pressure. An observed activation barrier for [29] L.-M. Tau, B.H. Davis, Energy Fuels 7 (1993) 249. the methylation of ethene of 109 kJ/mol has been found. [30] L.-M. Tau, B.H. Davis, Fuel Process. Technol. 33 (1993) 1. [31] E. Iglesia, T. Wang, S.Y. Yu, in: A. Parmaliana, D. Sanfilippo, F. Frus- Correction for the adsorption energy of ethene [33] yields an teri, A. Vaccari, F. Arena (Eds.), Natural Gas Conversion V, Elsevier intrinsic activation energy of about 135 kJ/mol. Dimeriza- Science, Amsterdam, 1998, p. 527. tion of ethene is a very minor reaction. Superimposed on the [32] P.O. Rønning, Ø. Mikkelsen, S. Kolboe, in: Proceedings of the 12th methylation pattern is a mechanism and reaction path lead- International Zeolite Conference, vol. II, Materials Research Society, ing to alkene products very rich in 13C atoms. This mech- Warrendale, PA, 1999, p. 1057. anism becomes dominating when the conversion exceeds a [33] V.R. Choudhary, S. Mayadevi, Zeolites 17 (1996) 501. [34] A.M. Vos, X. Rozanska, R.A. Schoonheydt, R.A. van Santen, F. Huts- few percent and hydrocarbons other than ethene are avail- chka, J. Hafner, J. Am. Chem. Soc. 123 (2001) 2799. able for reaction with methanol. The results are concurrent [35] X. Rozanska, R.A. van Santen, F. Hutschka, J. Phys. Chem. B 106 with the formation of alkenes with high 13C contents through (2002) 4652. the hydrocarbon pool mechanism. [36] M. Bjørgen, U. Olsbye, D. Petersen, S. Kolboe, J. Catal. 221 (2004) 1.

Paper V

Kinetic Studies of Zeolite-catalyzed Methylation Reactions. 2. Coreaction of [12C]propene or [12C]n-butene and [13C]methanol

S. Svelle, P.O. Rønning, S. Kolboe, Preliminary manuscript, to be submitted to J. Catal.

Kinetic studies of zeolite catalyzed methylation reactions

2. Co-reaction of [12C]propene or [12C]n-butene and [13C]methanol

Stian Svelle*, Per Ola Rønning† and Stein Kolboe*°

Department of Chemistry, University of Oslo, P.O.Box 1033 Blindern, N-0315 Oslo, Norway and Oslo University College, Faculty of Engineering, Cort Adelers gate 30, 0254 Oslo,

Norway

e-mail: [email protected]

Phone: +47 22 85 54 74

Fax: +47 22 85 54 41

° Corresponding author

* University of Oslo

† Oslo University College

ABSTRACT: The co-reaction of propene or n-butene and methanol over an H-ZSM-5 acidic zeolite catalyst has been investigated using isotopically labeled reagents. The main objective has been to obtain kinetic data for the methylation of the propene and n-butene. This study is an extension of our previous investigation of the co-reaction of ethene and methanol [1]. At the very high feed rates employed here, the methylation products are dominating, and the isotopic composition is in accord with a methylation formation mechanism. Arrhenius plots have been constructed, and the activation energies, when corrected for the appropriate heats of alkene adsorption, were ~ 110 kJ/mol for the methylation of propene and ~ 100 kJ/mol for the methylation of n-butene. The results are compared with resent computational studies of the methylation of alkenes. The origin of the products not formed via methylation is briefly discussed. A short survey of the reactivity of propene and n-butene without methanol co-feed is presented. It has been found that alkene interconversion reactions are strongly suppressed by the presence of methanol.

Keywords: H-ZSM-5; Propene methylation; Butene methylation; Methanol-to-hydrocarbons;

Propene; Linear butenes; Isobutene; Isotopic labeling

1. Introduction

The methanol-to-hydrocarbons (MTH) process represents a possible route for the upgrading of natural gas or coal to higher value products. Natural gas may be transformed to synthesis gas (CO and H2), which is subsequently reacted to form methanol. The methanol may then be converted into a mixture of hydrocarbons using acidic zeolite or zeotype catalysts. Mobil Oil was the first to discover and develop this zeolite-based technology, resulting in the methanol-to-gasoline (MTG) process; in which methanol is converted to gasoline over ZSM-5 derived catalysts. Later, Norsk Hydro and UOP jointly developed the methanol-to-olefins (MTO) variant of the reaction, where ethene and propene are the main products formed over SAPO-34 zeotype catalyst systems. The latest addition to this field is the methanol-to-propylene (MTP) alternative, currently offered by Lurgi. Despite the apparently great potential for industrial application of the MTH-reaction, the only full-scale commercial operation to date was the MTG-plant operated by on New Zealand, where gasoline production was started in 1986 and later shut down due to a falling price of oil relative to that of methanol [2]. However, an MTO plant is expected to come on-stream in

Nigeria in 2006, as part of a natural gas-to-polymers project [3].

In contrast to the maturity of the MTH-reaction when it comes to practical applications, the present level of understanding of the underlying reaction mechanism still leaves something to be desired. Initial research focused on the formation of carbon-carbon bonds directly from C1-units (methanol, dimethylether, or trimethyloxonium ions), but these efforts were all inconclusive, and this reaction is now considered to be of little importance during steady state conversion of methanol to hydrocarbons [4-7]. More indirect mechanism types are currently favored, and experimental and theoretical results in favor of the

“hydrocarbon pool” mechanism, originally proposed by Dahl and Kolboe [8-10], are mounting. The hydrocarbon pool mechanism proceeds via continuous methanol addition to a

3 hydrogen poor adsorbate inside the zeolite pores, from which lower alkenes are split off at a later stage of the catalytic cycle. The exact chemical nature of the hydrocarbon pool has been investigated by Kolboe and co-workers [11-15], Haw and co-workers [16-22], and Hunger and co-workers [23,24] and may depend on catalyst type and reaction conditions.

Polymethylbenzenes, polymethylnaphtalenes, and polymethylated cyclopentadienyl cations have all been shown to function as hydrocarbon pool species.

In the beginning of the 1980s, Dessau and co-workers proposed an indirect reaction mechanism for the methanol conversion based on alkenes as the key intermediates [4,6]. A considerable part of the hydrocarbons in the product stream are alkenes, and chain growth via alkene methylations followed by cracking to yield smaller alkenes might constitute a competing mechanistic scheme. Knowledge of the rate of alkene methylation (and ultimately other species present in the zeolite pores) is therefore an essential step for discrimination between the various proposed mechanisms.

We have previously investigated the methylation of ethene by methanol to form propene, at reaction conditions directly comparable to those employed in the present study

[1]. A central issue in that study was to find the reaction conditions best suited for measuring rate data for the methylation, i.e. suppressing any side reactions such as alkene dimerization, alkene cracking, cyclizations, and hydrogen transfer, but still operating at realistic a reaction temperature and catalyst acid site density. An intrinsic activation energy for ethene methylation of ~ 135 kJ/mol was found, and the reaction was of first order with respect to ethene partial pressure and zero order with respect to methanol. The apparent rate constant, k, was 2.6 × 10PP-4 mol/(g h mbar) at 350 °C and the pre-exponential, A, was 3.5 × 105 mol/(g h mbar).

In this paper we report new results on the kinetics of the methylation of propene and linear butenes by methanol. Isotopically labeled methanol has been used in order to follow the

4 individual reaction steps. In contrast to ethene, both propene and the linear butenes display non-negligible activity when reacted without methanol co-feed even at the very high feed rates employed here. It was therefore necessary to address this issue in some detail in order to better understand the results from the co-reaction experiments. Hence, the first part of this report concerns the reactivity of propene and the linear butenes alone, and the methylation reactions are addressed in following parts.

2. Experimental

The experimental setup and the calculation procedures employed have been described in detail previously [1]. A brief summary is given below.

2.1. Catalyst, reagents, and catalytic testing. The H-ZSM-5 catalyst used is a gift from Süd-

Chemie A.G. The Si/Al ratio is 45, and the sample consists of very small crystals (~ 50 nm).

[13C]methanol was supplied either by ISOTEC or Cambridge Isotope Laboratories. Natural isotope abundance propene was supplied by Fluka. Gas chromatographic analysis showed that propane (~ 0.07%) and 1-butene (~ 0.05%) were the main impurities. 1-Butene > 99% (Fluka) was used for the co-reaction experiments, whereas 2-butanol > 99% (Fluka) was used to study the reactivity of linear butenes without methanol co-feed. 2-Butanol is immediately dehydrated, forming linear butenes in situ. Regardless of the original butene source (2-butanol or 1-butene) the three linear butene isomers in the effluent were always in internal thermodynamic equilibrium, and we will therefore refer to this feedstock as n-butene in the following.

A fixed bed Pyrex microreactor (3 mm ID) was used for the catalytic experiments.

Propene and 1-butene were fed as gases, whereas methanol and 2-butanol were fed by passing

5 part of the carrier gas (He, >99.996%) through a vessel containing the desired alcohol, thus saturating the carrier gas. The feed rate (WHSV) was varied by varying the total gas flow through the reactor. Typical reaction conditions for the co-reaction of propene and methanol were: Methanol partial pressure = 50 mbar, propene partial pressure 20 mbar, total gas flow =

100 mL/min, reaction temperature = 350 °C, and catalyst mass = 2.5 mg, resulting in a feed rate (WHSV) of 237 h-1. When n-butene was studied, the alkene partial pressure was slightly lower than for propene, typically 13 mbar. The effects of varying the feed rate, reactant partial pressures, and reaction temperatures have been investigated. A key point has been to study the catalytic system at low conversion, i.e. high feed rates, but at the same time using a realistic reaction temperature and catalyst acid site density.

2.2. Analysis and calculations. The effluent was analyzed using a Carlo Erba Vega GC-FID equipped with a Supelco SPB-5 column (60 m × 0.53 mm × 3 µm). C1-C4 alkanes/alkenes were separated on a Siemens Sichromat 2-8 (FID) equipped with a Chrompack PLOT column

(Al2O3/KCl, 50 m × 0.53 mm × 10 µm). The isotopic distributions were determined based on analyses made with an HP 6890 GC-MS, using a HP-5MS column (60 m × 250 µm × 0.25

µm) combined with cryostatic cooling.

The computational method used for determining the isotopic composition of the products has been outlined previously [1,25].

3. Results

The main objective of the present work was to study and obtain kinetic data for the methylation of propene and linear butenes by methanol. In order to obtain kinetic data, it was necessary to operate at very high feed rates and correspondingly low conversions. At higher

6 conversions, i.e. lower feed rates, secondary reactions rapidly become dominant, and any information concerning the primary reaction steps becomes obscure. The feed rate is usually given as the weight hourly space velocity (WHSV), in units of gram feed per gram catalyst per hour. The inverse quantity, the “contact time” (CT = 1/WHSV) is better suited for the following discussion. Using CT rather than WHSV allows extrapolations of rates and selectivities to CT = 0 (or infinite feed rate) to be made, thereby obtaining an estimate for the primary or intrinsic quantities.

Unlike the previously published results concerning the co-reaction of ethene and methanol [1], both propene and n-butene display significant reactivity when fed alone, even at the shortest contact times employed. Therefore, a short survey of the reaction pattern of the alkenes alone will be given before the co-reaction data. As previously described, the reactivity of methanol alone was insignificant [1].

3.1. Propene without methanol co-feed. The effects of varying the contact time on the propene conversion were investigated by reacting 20 mbar of propene over 2.5 mg catalyst at

350 °C. The total gas flow through the reactor was varied from 10 - 100 mL/min, thus varying the contact time from 0.012 – 0.12 h (WHSV 81.8 – 8.2 hPP-1). The influence on the propene conversion, which increases from 4% to 28%, is displayed in Fig. 1. The non-linear increase with CT agrees well with the equilibrium content (C%) of propene in a C3-C6 alkene mixture, which is about 35% at an initial pressure of 20 mbar [26], corresponding to 65% conversion.

The level of reactivity is in stark contrast to what was observed when ethene was investigated in an analogous manner, the only difference being a higher ethene partial pressure of 50 mbar instead of 20 mbar [1]. The ethene conversion was below 0.1% even at the longest CT. The product distribution from propene was virtually unaffected by the changes in feed rate, and the effluent comprised 47 C% n-butenes, 32 C% isobutene, 15 C% C5, mainly pentenes, and 4

35 C% C , mainly hexenes. Hexenes %) 6

C 30 ( 25 constitute the expected outcome of a

opene 20 pr primary dimerization reaction, but this 15 on of i 10 s product fraction was only detected in small er 5 amounts. Clearly the propene molecules

Conv 0 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 1/WHSV = CT (h) undergo several secondary reaction steps

Fig. 1. Conversion of propene versus contact before the products leave the reactor, even time. 20 mbar of propene reacted alone; reaction temperature 350 °C; total gas flow at such high feed rates. I addition, traces of varied from 10 to 100 mL/min. butanes and ethene were detected. The rate of propene conversion, which is obtained by dividing the conversion by CT, extrapolates to about 3.5 g/(gcatalyst h) at CT = 0 for the chosen conditions.

The degree of propene conversion depends strongly on the partial pressure. This is displayed in Fig. 2, where the propene partial pressure is varied from 5 to 80 mbar. The conversion increases from less that 1% to 14% in this range. A linear increase of the conversion with pressure shows a second order reaction. Considering that product formation from propene necessarily involves an interaction of (at least) two propene molecules on the

catalyst surface, this behavior might be 16 %) C 14 expected. Possibly, the conversion 12 ne ( 10

ope increases slightly less than linearly with

pr 8 of

n 6 pressure, so the reaction order may be

sio 4 er 2 somewhat lower than 2.

Conv 0 0 102030405060708090 Significant changes in the product Propene partial pressure (mbar) distribution were observed when the feed Fig. 2. Conversion of propene versus propene partial pressure. 5 - 80 mbar of propene partial pressure was altered. At 5 mbar, reacted alone; reaction temperature 350 °C; total gas flow 100 mL/min. isobutene and n-butene are dominating,

8 constituting about 40 and 50 C% of the products, respectively. Modest amounts of pentenes (7

C%) and ethene (3 C%) were also detected. At 80 mbar n-butene is still dominating (39 C%),

C5 has become the second largest fraction (24 C%) followed by isobutene (21 C%), C6+ (16

C%). Ethene and butanes occur in trace amounts.

Table 1 lists the propene conversion and product selectivities as a function of reaction temperature. The most striking feature of Table 1 is the pronounced increase in conversion when the temperature is lowered. This might seem counterintuitive, but can be attributed to an increase in the surface coverage at low temperatures. This point will be elaborated in the

Discussion. The increase in the conversion when going from 275 °C to 250 °C is smaller than for the other steps in Table 1, indicating that the conversion is approaching a maximum value before it drops off as the temperature becomes too low for any reaction to occur. Product formation is insignificant below about 200 °C.

Table 1. Propene conversion and product selectivities (in C%) versus reaction temperature. 20 mbar of propene reacted alone; contact time 0.012 h; reaction temperature varied from 250 to 425 °C.

Reaction C2 n-C4 i-C4 C5 C6+ Conversion temperature (%) (°C) 250 0.1 41 14 35 9.8 12.8 275 0.1 49 19 26 5.8 12.1 300 0.2 53 23 22 2.6 8.9 325 0.4 53 27 18 2.5 6.2 350 0.9 49 31 14 5.1 3.1 350 0.8 50 33 14 3.1 3.8 375 2 47 32 13 6.0 1.9 400 5 45 30 12 8.3 1.2 425 11 42 24 12 11 0.8

3.2. n-Butene without methanol co-feed. The reactivity of n-butene was probed in experiments analogous to those described for propene above. In order to accurately control the reaction conditions, 2-butanol was used as a precursor for n-butene rather than gaseous 1-

30 C 3 25 isobutene C 20 5 C C% 6+ 15 conversion

0 0.00 0.01 0.02 0.03 0.04 0.05 0.06 Contact time = 1/WHSV (h)

Fig. 3. Conversion of n-butene and product selectivities (in C%) versus contact time. 13 mbar of n-butene reacted alone; reaction temperature 350 °C; total gas flow varied from 25 to 125 mL/min.

butene. The 2-butanol is virtually instantaneously transformed into water and linear butenes.

Separate experiments showed that the presence of additional water does not influence the butene conversion. There are three isomeric linear butenes, but their interconversion is so fast that an essential equilibrium is always observed. There is one major difference in the reactivity of n-butene compared to propene; the linear butenes may undergo a skeletal isomerization, leading to the formation of isobutene, and this results in a slightly more complicated reaction pattern. Conversion and selectivities as a function of contact time are given in Fig. 3. n-Butene displays a higher reactivity than propene. The rate of conversion

(conversion divided by CT) can be extracted from the data in Fig. 3 and is about 10 g/(g h), i.e. about 3 times that of propene, or – if compared at equal partial pressures – a factor about

4.5. Fig. 3 shows only small changes in the product composition when the CT is varied.

Clearer effects on the product distribution are seen when the n-butene partial pressure is altered; see Table 2. At the lowest pressures, isobutene becomes dominant among the products, the selectivity reaching nearly 70 C% at pn-butene = 1 mbar. The conversion decreases with decreasing pressure, but it does not fall off linearly towards zero, as was the case for propene. A linear regression line gives 8.3% conversion at pn-butene = 0.

Table 2. n-Butene conversion and product selectivities (in C%) versus partial pressure. 1 - 16 mbar of n-butene reacted alone; reaction temperature 350 °C; total gas flow 100 mL/min.

Partial C3 i-C4 C5 C6+ Conversion pressure (%) (mbar) 1 15 69 16 - 8.6 2 20 60 20 - 9.3 4 22 54 24 - 9.8 6 31 39 29 1.5 13 10 33 32 31 4.1 15 16 34 28 31 6.7 17

The conversion and product distribution at various reaction temperatures are given in

Fig. 4. When the temperature is raised above 350 °C, isobutene quickly becomes dominating, constituting 80 C% of the products at 500 °C. In the temperature range 300 – 500 °C the conversion was essentially constant, although increasing when the temperature falls from 350 to 300 °C. Below about 200 °C the conversion is insignificant.

100

C 3 60 isobutene C 5 C% C 40 6+ conversion

0 300 350 400 450 500 Reaction temperature (°C)

Fig. 4. Conversion of n-butene and product selectivities (in C%) versus reaction temperature. 13 mbar of n-butene reacted alone; total gas flow 75 mL/min; reaction temperature varied from 300 to 500 °C.

3.3. Co-reaction of propene and methanol.

3.3.1. Influence of the contact time. A feed mix consisting of 20 mbar [12C]propene and 50 mbar [13C]methanol was reacted at 350 °C over 2.5 mg of H-ZSM-5 catalyst. The contact time was varied from 0.0042 to 0.042 h (WHSV from 237 to 24 h-1) by adjusting the total gas flow. The conversion, determined by considering propene, methanol, and dimethyl ether as unconverted feed, is shown in Fig. 5a. It ranges from 2.5 C% to 25 C%, which is considerably higher than observed when ethene and methanol were co-reacted at comparable conditions

[1], when the conversion ranged from almost zero to about 7%. By dividing the conversion with CT, the rate of conversion is obtained; see Fig 5b. Extrapolation to CT = 0 yields a rate of 8.0 g/(g h). The corresponding limiting rate was 0.7 g/(g h) when methanol and ethene

were co-reacted at an ethene partial 30 a) pressure of 50 mbar [1]. The rate 25

20 decreases with increasing CT, as on (C%) i 15 the conversion increases and the 10 Convers 5 effective reactant concentrations

0 decrease. Hence, no autocatalysis is 10 b)

8 on

i seen in the present case. The ) ers -1 6 *h -1 onv product selectivities are shown in c g

* 4 (g of e Fig. 6. If methylation of propene 2 Rat

0 were the only reaction occurring, 0.00 0.01 0.02 0.03 0.04 1/WHSV = CT (h) the n-butene selectivity would reach

Fig. 5. Conversion (a) and rate of conversion (b) of 100%, but this is clearly not the feed mixture versus contact time. 20 mbar of propene co-reacted with 50 mbar methanol; reaction case. n-Butene does, however, temperature 350 °C; total gas flow varied from 10 to 100 mL/min. become dominating at the shortest

100 C 2 %) isobutene C 80 n-butene C 5 ities ( 60 C v 6+

20 oduct selecti Pr 0 0.00 0.01 0.02 0.03 0.04 1/WHSV = CT (h)

Fig. 6. Product selectivities (in C%) versus contact time. 20 mbar of propene co-reacted with 50 mbar methanol; reaction temperature 350 °C; total gas flow varied from 10 to 100 mL/min. contact times, but other products could always be detected. Extrapolated to CT = 0, an n- butene selectivity of about 70 C% is observed. The other limiting selectivities appear to be

15% C5, 8% isobutene, and 7% C6+. Ethene and alkanes (butanes) were always present in insignificant amounts.

Because of the quite high reactivity of propene alone it was essential to use isotopically labeled reactants in order to analyze the data. The isotopomer distributions of the alkene products are given in Fig. 7. For trans-2-butene, which is representative of the linear

12 13 butene isomers, the C3 C1 isotopomer is in clear excess, indicating that methylation is the main pathway for n-butene formation in the co-reaction system. The other n-butene isotopomers are always present in smaller amounts, even at the shortest CTs, and the

12 13 12 12 13 extrapolated value for C3 C1 is about 90%. The C4 and C2 C2 isotopomers both extrapolate to ca. 5% at CT = 0. When the CT is increased, the methylation pattern becomes obscured by secondary reactions, and the isotopomer distributions approach randomness. The total content of 13C-atoms in n-butene is close to 25% at the shortest CTs.

12 13 The isotopic distribution of 2-methyl-2-butene is strongly dominated by the C3 C2 species. The amount of this isotopomer exceeds by far that to be expected for a random distribution. A double methylation appears to be the main formation mechanism. The share of 13

100 ethene No. of 80 13C atoms zero 60 one two 40 three four 20 five six

) 0 100 % 2-methyl-2-butene ( t-2-butene

on 80 80 i t u b i

r 60 60 st i

d 40 40

omer 20 20 op t o

s 0 0 I 50 isobutene hexene 50

40 40

30 30

20 20

10 10

0 0 0.00 0.01 0.02 0.03 0.04 0.00 0.01 0.02 0.03 0.04

1/WHSV = CT (h)

Fig. 7. Isotopic composition of alkene products versus contact time. Note the scale differences. 20 mbar of propene co-reacted with 50 mbar methanol; reaction temperature 350 °C; total gas flow varied from 10 to 100 mL/min.

pentenes not composed of two 13C and three 12C-atoms and thus apparently not formed via

12 13 methylations is slightly greater than for the n-butenes. The limiting amounts of the C2 C3

12 13 and C4 C1 isotopomers are in both cases slightly above 10%. At the highest feed rates, 41% of the carbons in the pentenes are labeled.

Isotopic data are available for one hexene isomer. No clear differences could be seen when comparing the mass spectrum of this isomer with the other, less prominent or less well-

12 13 separated isomers. The C3 C3 isotopomer, corresponding to triple methylation of propene, is dominant. Also when ethene and methanol were co-reacted isotopic distributions indicating triple and even quadruple methylations were observed [1]. The second most abundant hexene isotopomer is built up exclusively from 12C-atoms. This can only be attributed to propene dimerization. It should, however, be noted that the relative content of hexenes becomes very small at the shortest CTs, extrapolating to 0 – 5% only. Propene dimerization is therefore a rather unimportant reaction compared to methylation at the conditions chosen in these experiments.

In addition to the major alkene products, small amounts of aromatics were always detectable in the effluent. The isotopic composition of the joint p/m-xylene product fraction could be determined due to the simple mass spectrum, which is dominated by the molecular ion peak. The distribution is broad; it is slightly broader than a random distribution based on the total 13C content, in the sense that the share of the all 12C and all 13C isotopomers are in excess compared to the random distribution. The total content of labeled carbons was high, increasing from 61% at long CTs to 67% at the shortest CT.

No obvious pattern is seen in the isotopic distribution of the ethene molecules in the

13 effluent, which is close to random. The relative abundance of the C2 isotopomer increases as the CT becomes shorter. Similarly to what was seen for p/m-xylene, the total 13C content is high; it increases from 62% at long CTs to 79% at the shortest CT.

The isotopic composition of isobutene is strikingly different from that of n-butene.

The distribution is close to random, and it may be concluded that isobutene is not formed via methylation of propene. Also, the total content of 13C is considerably higher, it ranges from

37% at CT = 0.0042 h to 47% at CT = 0.042 h, signifying that n-butene is an unlikely immediate precursor for isobutene is this co-reaction system.

Since the isotopic compositions of the products are known, it is possible to evaluate the degree of conversion of propene and methanol individually. By applying this approach to the data collected at the shortest contact time (CT = 0.0042 h), the propene conversion is determined to be 3% and the methanol conversion 2%. Thus, the propene conversion is roughly the same as it was when propene was investigated without methanol co-feed (4%).

Even so, the isotopic data show that the reaction steps by which propene is consumed are completely different when methanol is present (methylation) compared to when it is not

(alkene interconversions).

3.3.2. Influence of the propene partial pressure. The effects of varying the propene partial pressure on the reaction rates and isotopic compositions of the products formed in the coreaction were investigated by conducting the following experiment: The propene partial pressures were varied in the range 5 to 100 mbar. The methanol partial pressure was kept at

50 mbar with a constant total gas flow of 100 mL/min and 2.5 mg of catalyst. The total conversion of the feed mixture depends on the propene partial pressure; it increased from 1% at 5 mbar to 6% at 100 mbar. 10

) 9 h Correspondingly, the rate of mono- 8 g/g ( eled 7 ab labeled n-butene formation increases l - 6 o 5 4 with the propene partial pressure as formation mon f 3

tene 2 shown in Fig. 8. At higher propene u b

Rate o 1 n- 0 pressures, the methylation reaction 0 20406080100 Propene partial pressure (mbar) becomes less important (the isotopic Fig. 8. Rate of mono-labeled n-butene formation and product selectivities decrease, see versus propene partial pressure. 5 – 100 mbar of propene co-reacted with 50 mbar methanol; reaction below). temperature 350 °C; total gas flow = 100 mL/min.

Significant effects were also seen on the product distribution in this experiment. The selectivity towards n-butenes decreases when the propene pressure is increased, it falls from

65 C% at 5 mbar to 36 C% at 100 mbar. The C6+ fraction increases from 8 C% to 30 C% in the same interval. Only modest shifts were found for the other products, and the selectivities resemble those presented for short CTs in Fig. 6. Isotopic data are given in Fig. 9. The

100 ethene No. of 80 13C atoms zero 60 one two 40 three four 20 five six

) 0 100 %

( n-butene 2-methyl-2-butene n

o 80 80 i t bu i

r 60 60 st

di 40 40

20 20 pomer o ot

s 0 0 I isobutene hexene 80 80

60 60

40 40

20 20

0 0 0 20 40 60 80 100 0 20406080100 Propene partial pressure (mbar)

Fig. 9. Isotopic composition of alkene products versus propene partial pressure. 5 - 100 mbar of propene co-reacted with 50 mbar methanol; reaction temperature 350 °C; total gas flow = 100 mL/min.

17 isotopic composition of the products is dictated by the feed composition. At low propene pressures, the distributions are the same as those described at the shortest CTs in Fig. 7.

However, at 100 mbar of propene, the all 12C isotopomer is the most abundant for all the alkene products except n-butene. Methylation is the most prominent mechanism for n-butene

12 12 13 formation, but the amounts of the C4 and C3 C1 isotopomers are about equal at the highest propene partial pressure investigated. The total content of labeled carbons in the n-butenes decreases as the pressure increases, from 29% to 14%. A similar, further enhanced, trend is seen for 2-methyl-2-butene and the hexenes as the shares of 13C atoms decrease from 51% to

12% and from 59% to 8%, respectively. Ethene is richer in 13C than the other alkenes

13 analyzed. When the propene pressure is 5 mbar, the abundance of the C2 isotopomer is 83%.

The total 13C content falls from 90% to 36% when the propene pressure is increased from 5 to

100 mbar. As before, the isotopomer distribution is slightly broader than the random distribution. The isotopic distribution of isobutene is also close to random. The total content of labeled atoms varies from 58% to 9% over the investigated pressure interval. The aromatics, exemplified by p/m-xylene (not shown), contained mainly 13C atoms, falling from

85% to 46% as the propene pressure increased from 5 to 100 mbar.

3.3.3. Influence of the methanol partial pressure. The methanol partial pressure was varied from 10 to 90 mbar in an experiment analogous to the one described above for the propene pressure effects. Adjusting the methanol partial pressure had considerably less pronounced effects on conversions, product selectivities, and isotopic distributions than in the propene case. Fig. 10 shows the rate of n-butene formation against the methanol pressure. The slight increase in the rate above 20 mbar is probably within the experimental uncertainty, and the reaction may be considered to be zero order with respect to methanol at pressures above 20 mbar. Whether the data point at 10 mbar is an outlier or reality is unknown, but the rate will

8 eventually decrease when the methanol

h) 7 d /g

le 6 partial pressure becomes lower; there e on (g

i 5 t cannot be any methylation without no-lab 4 mo forma 3 f e methanol. The data point at p = 0 o methanol 2 ten u 1 Rate mbar is the rate of n-butene formation n-b 0 0 20406080100 Methanol partial pressure (mbar) from propene alone and does therefore

Fig. 10. Rate of n-butene formation versus not correspond to any methylation methanol partial pressure. 10 – 90 mbar of methanol co-reacted with 20 mbar propene; reaction. The product selectivities were reaction temperature 350 °C; total gas flow = 100 mL/min. virtually unchanged in the investigated interval and thus identical to those displayed for short contact times in Fig. 6. Modest effects were seen on the isotopic compositions over the investigated range. Lowering the methanol pressure, and thus making propene-propene reactions more likely, resulted in a decrease in the

13C content of the products. The general isotopic pattern was usually not changed, but in the case of the hexenes, the share of the all 12C isotopomer went from about 20% to 41%. This isotopomer corresponds to a simple propene dimerization.

3.3.4. Influence of the reaction temperature. In an attempt to determine the activation energy for the methylation of propene, reaction temperatures between 290 and 410 °C were investigated. Reaction conditions were the same as above. The propene partial pressure was

20 mbar; methanol pressure was 50 mbar; total gas flow was 100 mL/min, resulting in a contact time of 0.0042 h (WHSV = 237 h-1). Reactant conversions, rates of product formation, and isotopic compositions were determined as before. The conversion to products increased from 0.5% at 290 °C to 6.5% at 400 °C. A slight decrease in the selectivity towards n-butene was found at elevated temperatures, coupled with enhanced selectivities towards isobutene,

C5, and C6+. Pronounced temperature effects were also seen when propene (and n-butene) was

19 studied with no methanol present, as described above. It was then found that the conversion and thus the extent of alkene interconversion reactions increased when the temperature was lowered. This effect is not clearly seen in the co-reaction system, but the isotopic data do reveal that alkene + alkene reactions are of greater relative importance at low reaction

12 13 temperatures. For the n-butenes, the C3 C1 isotopomer, indicative of methylation, is always in excess. It constitutes 72% of the n-butenes even at 290 °C. At the same temperature, the

12 C4 isotopomer is second most abundant at 23%. As was seen from Fig. 9, at 350 °C the

12 13 12 share of C3 C1 was ~ 80% and C4 only about 4%. For 2-methyl-2-butene, double methylation is dominating above 330 °C. Below this temperature, other formation routes,

12 13 12 resulting in greater shares of the C4 C1 and C5 isotopomers, become increasingly important. A similar, but even more pronounced trend is seen for the hexenes. More than 70% of the hexenes are built up exclusively from unlabeled carbons at 290 °C. Correspondingly; the total 13C content is very low at 290 °C, merely 7%. This figure rises to 54% at 400 °C.

The isotopic data for ethene differ markedly from those for the other alkenes. Ethene contains slightly less than 80 % labeled carbons throughout the investigated temperature range.

Isobutene contains between 15% and 56% 13C, quite similar to what was observed for the higher alkenes, i.e., pentene and

) 2.5 hexene. Little variation was seen in the ion

at 2.0 m r

isotopic composition of the arenes. fo 1.5 ne *h)

te 1.0 p/m-Xylene contained from 60% to bu g/(g 0.5 n- 13 70% C, broadly distributed. of e

t 0.0 a r

A main motivation for the ln( -0.5 0.0015 0.0016 0.0017 0.0018 -1 present work is to determine the 1/T (K ) Fig. 11. Arrhenius plot for the formation of mono- apparent activation energy and the pre- labeled n-butene. 20 mbar propene co-reacted with 50 mbar methanol; CT = 0.0042 h-1; reaction exponential factor for the methylation temperature varied from 290 to 410 °C.

20 reaction by constructing an Arrhenius plot. Such a plot, based on the rate of formation of

12 13 C3 C1 isotopomer only, is given in Fig. 11, and an apparent activation energy of 69 kJ/mol can be extracted

3.4. Co-reaction of n-butene and methanol.

3.4.1. The influence of contact time. The co-reactivity of n-butene and methanol is rather analogous to that described above for propene and methanol and previously for ethene and methanol [1]. The following reaction conditions were employed: 13 mbar of [12C]n-butene and 50 mbar [13C]methanol were co-reacted at 350 °C over 2.5 mg of H-ZSM-5 catalyst. The contact time was varied from 0.0044 to 0.044 h (WHSV from 226 to 23 h-1) by adjusting the total gas flow between 10 and 100 mL/min. Conversions, product selectivities (Table 3), reaction rates (Fig. 12), and isotopic distributions (Fig. 13) were measured as before. The conversion of the feed mixture ranged from 10 to 54 C% and was thus slightly higher than when propene and methanol were co-reacted. As can be seen from Fig. 12, extrapolation of

Table 3. Conversion of feed mixture and product selectivites (in C%) versus contact time. 13 mbar of n-butene co-reacted with 50 mbar methanol; reactions temperature 350 °C; total gas flow varied from 10 to 100 mL/min.

CT (h) C2 C3 isobutene butanes C5 C6+ Conversion (%) 0.0044 0.9 17 14 2 46 19 8.9 0.0044 0.9 17 14 3 47 19 10.5 0.0055 0.8 17 14 3 45 20 13.1 0.0074 0.9 19 14 3 43 21 15.9 0.0074 1 20 14 3 41 21 18.9 0.0110 1.1 21 14 4 38 22 24.4 0.0221 1.2 24 13 5 32 25 38 0.0221 1 24 14 5 32 24 40.2 0.0294 1.3 24 13 5 31 26 42 0.0442 1.4 25 13 6 29 26 49.7 0.0442 1.3 25 14 6 28 26 53.8

30 the rate of conversion to CT = 0 yields 25 on

si 20 a value of 25 g/(g h). Pentenes are ) er -1 nv *h

o 15 -1

c dominating among the products, see

*g of (g 10 e Table 3, and the selectivity extrapolates

Rat 5

0 to about 55 C%. C3, C5, and C6+ 0.00 0.01 0.02 0.03 0.04 0.05 Contact time, 1/WHSV (h) become dominant at longer contact

times. Fig. 12. Rate of conversion of feed mixture versus contact time. 13 mbar of n-butene co-reacted with The isotopic data are analogous 50 mbar methanol; reactions temperature 350 °C; total gas flow varied from 10 to 100 mL/min. to those presented above for the

12 13 methylation of propene. For 2-methyl-2-butene, the C4 C1 isotopomer corresponding to methylation of n-butene, is by far the most prominent, and the abundance of this isotopomer extrapolates to about 85% at CT = 0. Double methylation is the prominent pathway for hexene formation, as is evident from the large content of the doubly labeled isotopomer.

Trace amounts of aromatics (methylbenzenes) were detected also in this experiment. The total content of 13C atoms in p/m-xylene ranged from 58% at CT = 0.0044 h to 42 % at CT = 0.044 h, and the isotopic distribution was close to the random distribution. Ethene was also fairly rich in 13C, the total content fell from 63 to 37% as the CT increased from 0.0044 to 0.044 h.

The distribution was close to random. Propene contained considerably fewer labeled carbons than ethene, but also in this case the distribution was fairly close to random. Overall, isobutene contained only ~ 15% labeled carbons over the investigated range of feed rates, and

12 the relative abundance of the C4 isotopomer was in excess of that expected based on a random distribution. This might indicate that a fraction of the isobutene may be formed in a monomolecular isomerization reaction of n-butene, even when methanol is present as a coreactant. The isotopic composition of trans-2-butene was also analyzed, and as expected the n-

22 butene fraction consists mainly of unconverted reactant. The total content of 13C in t-2-butene increased from 1.7% at CT = 0.0044 h to 5 % at CT = 0.044 h.

60 60 ethene propene 50 50

40 40

30 30

20 20

10 10

0 0 (%) isobutene 2-methyl-2-butene on i

t 80 80 bu i r 60 60 st

di 40 40

20 20 opomer ot s I 0 0 0.00 0.01 0.02 0.03 0.04 0.05 hexene

80 No. of 13C atoms 60 zero one 40 two three four 20 five six 0.00 0.01 0.02 0.03 0.04 0.05 1/WHSV = CT (h)

Fig. 13. Isotopic composition of alkene products versus contact time. Note the scale differences. 13 mbar of n-butene co-reacted with 50 mbar methanol; reaction temperature 350 °C; total gas flow varied from 10 to 100 mL/min.

3.4.2. Effect of the butene partial pressure. Also the effects of varying the butene pressure on reaction rates and isotopic composition of the products formed in the co-reaction were investigated. The experimental conditions were somewhat different from the conditions applied above. The temperature and amount of catalyst are identical at 350 °C and 2.5 mg, but

23 the methanol partial pressure was in this case 100 mbar rather than 50 mbar, and the total gas flow was 75 mL/min rather than 100 mL/min. The butene pressure was varied in the range 3 to 200 mbar.

Also here, like the 300

/gh) a)

g 250 methanol/propene co-reaction system, the on (

i 200 ers 150 nv

rate of conversion (Fig. 14a) and the rate o

c 100 of te

of mono labeled pentene formation (Fig a 50 R 0

14b) depend strongly on the alkene gh)

/ b) 40 g pressure. In analogy with the 30 onolabeled rmation ( o methanol/propene co-reaction system, the m 20 te of

a 10 R entene f product selectivities showed some p 0 0 50 100 150 200 dependence on the n-butene pressure, but n-butene partial pressure (mbar) the changes were modest. The most Fig. 14. Rate of feed mix conversion (a) and rate of mono-labeled pentene formation (b). 3- prominent change was seen for the 200 mbar of n-butene co-reacted with 100 mbar methanol; reaction temperature 350 °C; total pentene selectivity, which fell from about gas flow 75 mL/min. 41% at very low n-butene pressures to about 34% at a n-butene pressure of 200 mbar. The product spectrum is therefore quite similar to that given in Table 3. The experiments were, like the preceding ones, carried out with

[13C]methanol and ordinary 1-butene. The isotopic composition of the products in the effluent could then be determined. The results are displayed in Fig. 15. Ethene and isobutene were not analyzed.

Like the corresponding co-reaction experiments with methanol/propene, it is seen that the isotopic composition of all products is undergoing a marked change when the butene pressure changes. The gaseous flow rate remained constant, and is practically the same as the highest flow rate used in subsection 3.4.1 (75 vs.100 mL/min). The isotopic composition was

100 No. of propene 13 80 C atoms zero one 60 two

) three 40

(% four five on 20 ti six u b i 0 str pentene hexene di 80 er

pom 60 o t o s

I 40

0 0 50 100 150 200 0 50 100 150 200 n-butene partial pressure (mbar)

Fig. 15. Isotopic composition of alkene products versus n-butene partial pressure. 3-200 mbar of n-butene co-reacted with 100 mbar methanol; reaction temperature 350 °C; total gas flow 75 mL/min. in that case quite close to the extrapolated limiting composition where isotopic scrambling due to secondary reactions is essentially absent. It is therefore admissible to assume that the observed isotopic composition also in this case quite closely reflects the primary composition.

It is therefore clear that at the highest butene pressures alkene/alkene interconversion reactions are less successfully suppressed by the methanol. The dominating reaction is

12 13 formation of pentenes, and at low butene pressures the C4 C1 isotopomer, which is obtained by methylation of butene, constitutes 80% of the pentene molecules. Up to a butene pressure

12 of 25 mbar the content of the C5 isotopomer is negligible. The remainder of the pentene molecules contains more than one 13C atom. At high butene pressures, the content of 12C

12 atoms increases strongly, and at 200 mbar the C5 isotopomer constitutes 50% of the pentene molecules. This isotopomer is formed by alkene/alkene interconversion reactions.

Less directly, Fig. 16, which displays the separate methanol and butene conversion rates versus butene pressure, suggests the same message. The methanol conversion rate initially increases very strongly with increasing butene pressure, but the increase soon tapers

h) off, and above 40-60 mbar the increase /g 200

(g methanol 175

on n-butene becomes quite moderate. The butene si 150 er 125 nv conversion rate is seen to increase co 100

ant 75 t

ac 50 e

essentially linearly with butene r f 25 o

e t 0 a 0 50 100 150 200

pressure. These data were obtained from R n-butene partial pressure (mbar) the known rates of product formation Fig. 16. Rate of conversion of the individual (Fig. 14) and the known isotopic reactants 3-200 mbar of n-butene co-reacted with 100 mbar methanol; reaction temperature 350 °C; composition of the main products. total gas flow 75 mL/min.

The hexene isotopic distribution is seen to exhibit an evolution similar to that for

12 13 pentene, Fig. 15. At the lowest butene pressures the C4 C2 isotopomer, corresponding to a double methylation, constitutes nearly 80%. At 50 mbar this isotopomer still constitutes about

60%, but now there is more than 20% of the species with only one 13C atom, and more than

5% with the all 12C species. At 200 mbar butene pressure only 20% of the doubly labeled isotopomer remains.

Propene, which is always a quite important product in this reaction system with about

20 C% of the products, displays an isotopic distribution that is always rather close to a random distribution. With increasing butene pressure 12C atoms constitute a larger and larger fraction. At 3 mbar the propene contains 55% 13C, at 100 mbar there is only 10%. Propene is probably a product of complex alkene interconversion reactions, as are also the hexene isotopomers that do not contain 2 13C atoms.

The n-butene that emerged from the reactor mostly had the natural content, about

12 13 4.5%, of the C3 C1 isotopomer, and thus did not appear to have taken part in the reactions.

However, at the lowest butene pressures, 3 and 6 mbar, the analysis indicated the presence of about 1% of each of the isotopomers with two or more 13C atoms, indicating that some butene, like propene and pentene, is formed in reactions involving methanol.

3.4.3. Effects of the methanol partial pressure. The influence of the methanol pressure was investigated by varying the partial pressure from about 10 mbar to 100 mbar in separate experiments. The resulting effects on the rate of methylation product formation, which is shown in Fig. 17, are seen to be moderate. 6

h) 5

A reaction order close to zero is observed, /g g 4 labeled however, the slightly positive slope is ion ( at 3 m

probably a reality, at least for pressures mono- 2 ne for below about 50 mbar. The isotopic 1 Rate of pente 0 compositions of the products were 0 20406080100 Methanol partial pressure (mbar) measured, and no noteworthy shifts were Fig. 17. Rate of mono-labeled pentene observed in the investigated pressure formation (b). 12-100 mbar of methanol coreacted with 13 mbar n-butene; reaction range. temperature 350 °C; total gas flow 75 mL/min.

3.4.5. Effects of temperature variations. Determination of the activation energy of a reaction is a central part of the reaction study. We have therefore studied the rate of formation

12 13 of the C4 C1 isotopomer in the temperature range 350-450 °C at the conditions: 100 mbar methanol, 13 mbar butene and a flow rate 75 mL/min. Only minor effects were seen for the isotopic distributions in this temperature range. An Arrhenius plot based on the rate of formation of mono-labeled pentene was constructed (not shown) and an apparent activation barrier of 54 kJ/mol was found.

4. Discussion

4.1. Conversion of alkenes. Substantial research effort has been devoted to the conversion of light alkenes over acidic zeolites [27,28]. However, most of this work has been focused on either oligomerization in order to produce larger hydrocarbons, or, in the case of n-butene, the skeletal rearrangement reaction leading to isobutene. Hence, very little literature data has been obtained at conditions comparable to those employed in the present study, i.e. low partial pressures, high feed rates, and fairly high temperatures.

The key observations from the experiments where propene and n-butene were reacted alone can be summarized as follows: The propene conversion decreased at high temperatures.

The yield of products other than isobutene from n-butene also decreased at high temperatures.

Alkene interconversion reactions display a reaction order greater than one.

Quite clearly, the formation of products with more or fewer carbons than the original alkene reactant requires the initial interaction of (at least) two propene or n-butene molecules.

This consideration may explain both the observed increase in conversion at increasing alkene partial pressures and the increased yields at lower reaction temperatures. Lower temperatures and higher partial pressures inevitably lead to greater surface coverages, which will favor bimolecular events. The decrease in propene conversion at elevated temperatures could also be the result of thermodynamic limitations as suggested by Bandiera and Ben Taarit [29], but the equilibrium content of propene in a C2-C6 alkene mixture at 425 °C, which is around 35%

[26], is far above the observed 99% (see Table 1). We therefore consider the kinetic reasoning outlined above to provide a more satisfactory rationalization of the present data on the degree of propene conversion and also the decrease in the yield of products other than isobutene from n-butene at elevated temperatures.

The skeletal isomerization of n-butene to form isobutene has been a subject of considerable debate, in particular concerning the issue whether isobutene is formed in a monomolecular, pseudo-monomolecular, or bimolecular manner [28]. Isobutene displays an opposite behavior compared to the other major products; the yield increases at elevated temperatures, and the selectivity increases markedly at low alkene pressures. Hence, our results, obtained at very specialized reaction conditions, indicate that this is a monomolecular reaction under appropriate conditions.

It should be emphasized that even though the degree of alkene conversion is similar with or without methanol present, the isotopic analysis shows that the reactions in which the alkenes are consumed differ strongly. Thus, data for alkene conversion collected without methanol co-feed present cannot provide a direct measure of the alkene reactivity when methanol is present. Alkene interconversions are evidently strongly suppressed by the presence of methanol.

4.2. Methylation. The present report is an extension of our previous study on the kinetics of the methylation of ethene [1]. The value of the apparent rate constant for the methylation of

-4 ethene kethene was then found to be 2.6 × 10 mol/(g h mbar) at 350 °C, and the activation barrier (corrected for the appropriate heat of ethene adsorption) was ~ 135 kJ/mol.

Extrapolation of the rates of conversion to CT = 0 (Figs. 5b and 12) gave 8 and 25 g/(g h) in the propene and n-butene co-reaction systems, respectively. The rates of methylation are obtained by multiplying these values with the respective limiting fractional values for the product and isotopic selectivities. This procedure yields 5.0 g/(g h) and 11.7 g/(g h) for the methylation of propene and n-butene. Figs. 8 and 14 show that the rates of methylation of propene and n-butene may be taken to be first order with respect to alkene pressure up to 30 –

40 mbar, although a clear leveling off is seen at much higher pressures. More comparable data

29 are obtained by dividing with the pressures employed and finally converting to molar units.

-3 -2 The values then become: kpropene = 4.5 × 10 mol/(g h mbar) and kn-butene = 1.3 × 10 mol/(g h mbar), and the following ratios may be obtained: kethene : kpropene : kn-butene = 1 : 17 : 50. These rate constants are apparent rate constants that include the alkene adsorption constants, and the ratios do not strictly apply to the intrinsic rates of methylation.

We have previously investigated alkene methylation with theoretical methods, using a cluster consisting of four tetrahedral atoms to represent the zeolite combined with the

B3LYP/6-31G(d) + ZPE computational scheme [30]. The ratios derived from the calculated activation barriers are 1 : 15 : 39, if the average of the barriers for trans-2-butene, cis-2- butene, and 1-butene are used for n-butene, and the pre-exponentials are set to be equal. This apparent agreement with theory and experiment must be considered to be rather fortuitous.

Apparent activation energies were derived from the Arrhenius plots, and 69 kJ/mol was found for the methylation of propene (Fig 11) and 54 kJ/mol for n-butene. Again assuming that the methylation reactions are approximately first order with respect to the alkene partial pressures and zero order with respect to methanol, the corresponding heat of alkene adsorption should be added to these values in order to obtain the intrinsic activation energies. Pascual et al. used Monte Carlo simulations and force field methods to calculate adsorption isotherms for several alkenes, and a heat of adsorption of 39 kJ/mol was found for propene and values around 45 kJ/mol for the linear butenes on silicalite [31]. Hence, the activation barriers become ~ 110 kJ/mol and ~ 100 kJ/mol.

These estimates may be rather inaccurate. The assumption of a zero order reaction with respect to the methanol partial pressure may be doubtful at the higher temperatures. The data points in Fig. 11 suggest that they represent a curve with slightly negative curvature, as will be the case if the above assumption does not hold. The data for n-butene methylation give a curve that is very similar to Fig. 11. Furthermore, also the Arrhenius plot that was used in

30 the preceding study of the methylation of ethene indicated a slight deviation from linearity

[1]. The activation energy estimates are therefore likely to be low.

Other sources of error could be: At high temperatures, when the conversion is appreciable, some of the alkene originally formed via methylation may be consumed in further reactions, thus resulting in the measured rate being too low. Moreover, it seems to some extent unreasonable to assume that the entire amount of the mono-labeled isotopomers has been formed through methylation at low temperatures. Alkene interconversions are, at low temperatures, non-negligible relative to methylation, as is evident from the considerable share of the all 12C isotopomers at 290 °C. Considering the total contents of labeled atoms in the higher alkenes (exemplified by the hexenes), it is reasonable to assume that cracking of these would produce some of the mono labeled isotopomer. These effects also make the activation energy estimates too low.

4.3. The minor co-reaction products. Both when propene and n-butene were co-reacted with methanol, non-negligible concentrations of products other than the simple methylation products were always detected. Moreover, the selectivities towards some of these products did not extrapolate to zero at CT = 0. With respect to the mechanism of the MTH-reaction, it is of interest to evaluate possible routes of formation for these minor products. It should, however, be kept in mind that the reaction conditions employed here are quite different from those usually employed during steady state conversion of methanol to hydrocarbons, and the following discussion might therefore not always be valid.

4.3.1. Ethene. Ethene was always detected in trace amounts in all the co-reaction experiments. It was always fairly rich in 13C and the isotopic distribution was close to random. Three routes can be envisaged for ethene formation: A direct formation from

31 methanol/dimethylether, alkene cracking, or the hydrocarbon pool mechanism. Direct

13 formation from methanol might seem reasonable considering the high share of the C2 isotopomer detected, but this possibility may be discarded because no detectable amount of ethene was formed when methanol was reacted alone. A direct formation mechanism will not depend on the presence of products in the catalyst pores, i.e. not show any autocatalytic behavior, and the yield of ethene should be unaffected by an alkene co-reactant. Alkene cracking may also be ruled out; the discrepancy between the total content of 13C in ethene and the higher alkenes (hexene) is too large. Also, ethene is an unlikely alkene cracking product

[32]. This leaves the hydrocarbon pool mechanism as the only plausible route for ethene formation. The total label content and the isotopic distribution of ethene match those observed for p/m-xylene, which may be taken to represent the isotopic composition of the pool species reasonably well.

4.3.2. Propene and isobutene. For propene formation (in the n-butene + methanol system) alkene cracking appears to be the most probable formation mechanism. The total label content is considerably lower than for ethene and the aromatics and more in line that observed for the higher alkenes. Also, the isotopic distribution is fairly close to random, indicating a complex

(equilibrating) formation pathway. A similar conclusion may be drawn for isobutene. When n-butene is the co-reactant, and thus present in considerable concentrations, there appears to be a supplemental production of isobutene via a monomolecular skeletal isomerization of the n-butene feed.

4.3.3. Hexene. In the propene + methanol system, two pathways are easily discernible for the hexene formation, propene dimerization and triple methylation. When n-butene is the coreactant, double methylation is dominant. The other isotopomers are, however, always present

32 as minor components in both systems and do not easily extrapolate to zero at CT = 0. This behavior has been discussed previously [1].

5. Conclusions

[13C]Methanol and [12C]propene or [12C]n-butene have been co-reacted over an H-

ZSM-5 catalyst. The primary concern has been to obtain kinetic data for the respective methylation reaction. The previously observed (for ethene methylation [1]) zero order behavior with respect to methanol pressure and first order behavior with respect to the alkene pressure is only partially retained. Deviations are seen over the investigated partial pressure ranges. Estimated activation energies of ~ 110 kJ/mol and ~ 100 kJ/mol have been found for the methylation of propene and n-butene, respectively. These estimates are probably too low.

Alkene + alkene reactions are not insignificant for these systems, especially at higher propene or n-butene partial pressures, even though alkene interconversions are strongly suppressed by the presence of methanol.

References

[1] S. Svelle, P.O. Rønning, S. Kolboe, J. Catal. 224 (2004) 115.

[2] P. Barger, in: M. Guisnet, J.-P. Gilson (Eds.), Zeolites for Cleaner Technologies,

Catalytic Science Series – Vol. 3, Imperial College Press, London, 2002, p. 239.

[3] http://www.chemicals-technology.com/projects/eurochem/index.html (accessed July

2004).

[4] R.M. Dessau, R.B. LaPierre, J. Catal. 78 (1982) 136

[5] S. Kolboe, Acta Chem. Scand., Ser. A A40 (1986) 711.

[6] R.M. Dessau, J. Catal. 99 (1986) 111.

[7] W. Song, D.M. Marcus, H. Fu, J.O. Ehresmann, J.F. Haw, J. Am. Chem. Soc. 124

(2002) 3844.

[8] I.M. Dahl, S. Kolboe, Catal. Lett. 20 (1993) 329.

[9] I.M. Dahl, S. Kolboe, J. Catal. 149 (1994) 458.

[10] I.M. Dahl, S. Kolboe, J. Catal. 161 (1996) 304.

[11] B. Arstad, S. Kolboe, Catal. Lett. 71 (2001) 209.

[12] B. Arstad, S. Kolboe, J. Am. Chem. Soc. 123 (2001) 8137.

[13] S. Kolboe, in: A. Galarneau, F. Di Renzo, F. Fajula, J. Vedrine (Eds.), Zeolites and

Mesoporous Materials at the Dawn of the 21st Century, Elsevier Science, Amsterdam,

2001, p. 3946.

[14] M. Bjørgen, U. Olsbye, S. Kolboe, J. Catal. 215 (2003) 30.

[15] M. Bjørgen, U. Olsbye, D. Petersen, S. Kolboe, J. Catal. 221 (2004) 1.

[16] J.F. Haw, J.B. Nicholas, W. Song, F. Deng, Z. Wang, T. Xu, C.S. Heneghan, J. Am.

Chem. Soc. 122 (2000) 4763.

[17] J.F. Haw, W. Song, D.M. Marcus, J.B. Nicholas, Acc. Chem. Res. 36 (2003) 317.

[18] P.W. Goguen, T. Xu, D.H. Barich, T.W. Skloss, W. Song, Z. Wang, J.B. Nicholas, J.

Am. Chem. Soc. 120 (1998) 2650.

[19] A. Sassi, M.A. Wildman, H.J. Ahn, P. Prasad, J.B. Nicholas, C.S. Heneghan, J. Phys.

Chem. B 106, (2002) 2294.

[20] W. Song, J.F. Haw, J.B. Nicholas, C.S. Heneghan, J. Am. Chem. Soc. 122 (2000)

10726.

[21] W. Song, H. Fu, J.F. Haw, J. Phys. Chem. B 105 (2001) 12839.

[22] D.M. Marcus, W. Song, S.M. Abubakar, E. Jani, A. Sassi, J.F. Haw, Langmuir, 20

(2004) 5946.

[23] M. Seiler, U. Schenk, M. Hunger, Catal. Lett. 62 (1999) 139.

[24] M. Hunger, M. Seiler, A. Buchholz, Catal. Lett. 74 (2001) 62.

[25] P.O. Rønning, Ph.D. thesis, Department of Chemistry, University of Oslo, 1998.

[26] Data from: D.R. Stull, E.F. Westrum jr., G.C. Sinke, The Thermodynamics of Organic

Compounds, John Wiley & Sons, New York, 1969.

[27] C.T. O’Connor, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of

Heterogeneous Catalysis – Vol. 4, VCH Verlagsgesellschaft mbH, Weinheim,

Germany, 1997, p 2380.

[28] P. Mériaudeau, C. Naccache, Adv. Catal. 44 (2000) 505.

[29] J. Bandiera, Y. Ben Taarit, Appl. Catal. A: Gen. 132 (1995) 157.

[30] S. Svelle, S. Kolboe, O. Swang, J. Phys. Chem. B 107 (2003) 9281.

[31] P. Pascual, P. Ungerer, B. Tavitian, A. Boutin, J. Phys. Chem. B 108 (2004) 393.

[32] D.M. Brouwer, in: R. Prins, G.C.A Schuit (Eds.), Chemistry and Chemical

Engineering of Catalytic Processes, in: NATO ASI Ser., Ser. E, Vol 39, Sijthoff &

Noordhoff, The Netherlands, 1980, p. 137.

Paper VI

The Intermediates in the Methanol-to-Hydrocarbons (MTH) Reaction: A Gas Phase Study of The Reactivity of Polymethylbenzenium Cations

S. Svelle, M. Bjørgen, S. Kolboe, D. Kuck, U. Olsbye, O. Sekiguchi, E. Uggerud, Preliminary manuscript, to be submitted to Phys. Chem. Chem. Phys.

The Intermediates in the Methanol-to-Hydrocarbons (MTH) Reaction: A

Gas Phase Study of The Reactivity of Polymethylbenzenium Cations

STIAN SVELLE†, MORTEN BJØRGEN†, STEIN KOLBOE†, DIETMAR KUCK°, UNNI

OLSBYE†, OSAMU SEKIGUCHI†, EINAR UGGERUD†*

†University of Oslo, Department of Chemistry, P.O. Box 1033 Blindern, N-0315 Oslo, Norway

°Universität Bielefeld, Fakultät für Chemie, Postfach 10 01 31, D-33501 Bielefeld, Germany.

E-mail: [email protected]

*Corresponding author

Key words: gas phase ion chemistry, mass spectrometry, methylbenzene, methanol-to- hydrocarbons, zeolite, heptamethylbenzenium

RUNNING TITLE: Gas-phase study of polymethylbenzenium cations

Dedicated to Professor Vladimir E. Bondybey on the occasion of his 65th birthday.

1 Abstract: A full series of polymethylbenzenium ions has been generated in the gas phase by chemical ionization and the decomposition mass spectra of each member have been recorded.

While loss of H2 dominates for the lower homologues, loss of methane and methyl radical dominate for the higher homologues. Loss of larger fragments, in particular ethene, ethane and propene are also observed. The data are discussed with respect to a proposed reaction cycle in the zeolite-catalyzed methanol-to-hydrocarbons (MTH) reaction. The gas phase ions are considerable more energetic than those resident in a zeolite catalyst.

2 1. Introduction

Protonated zeolites have found widespread application in several acid catalyzed hydrocarbon transformation processes. Early research often invoked carbocationic species as intermediates in order to rationalize the observations made, but it is now generally accepted that zeolite acid strength is more in line with conventional acids rather than super acids, as there are only a few observations of persistent carbenium ions in zeolites. Recently, Bjørgen et al. [1,2] have shown that tetra-, penta-, and hexamethylbenzene are protonated on ring carbons when introduced into the wide pore β-zeolite, whereas the lower polymethylbenzene homologues do not form such cations. These polymethylbenzenes form, upon protonation, species demonstrated to be key intermediates in several reactions catalyzed by acidic zeolites.

Isomerization of methylbenzenes, via methyl shifts, most likely proceeds via these polymethylbenzenium type species having molecular structures corresponding to so-called

Wheland/Pfeiffer intermediates [3]. Transalkylation or disproportionation may also involve arenium cations, although mechanisms involving diphenylmethane are presently favored [3].

DFT/MP2-modeling performed by Arstad et al. [4] has indicated that polymethylbenzenium cations are energy minima on the potential energy surface during the methylation of methylbenzenes with methanol. This could also hold for other potential methylating agents.

Also, mounting evidence pointing towards methylbenzenes as crucial intermediates in the reaction where methanol is converted into a mixture of hydrocarbons (MTH) has been collected over the last few years [5,6]. Increased insight into the specifics of the MTH- reaction mechanism has been the main motivation for the present work.

Our understanding of the mechanism of the MTH-reaction has become greatly improved over the last decade, as the so-called hydrocarbon pool mechanism [7-10] has gained increasing acceptance. This is an indirect mechanism, believed to proceed via repeated methylations and subsequent alkene loss from organic reaction centers trapped within the

3 zeolite voids. The exact chemical nature of the hydrocarbon pool is incompletely known and might vary with the reaction conditions and catalyst type, but polymethylbenzenes, and in particular the higher homologues, have been shown to function as such reaction centers [11-

17].

The manner in which alkenes are lost from the polymethylbenzenes is a central issue in MTH-chemistry. Haw and co-workers [17,18] have suggested the mechanism outlined in

Scheme 1: Deprotonation of a polymethylbenzenium species results in the formation of an exo-cyclic double bond, which can by methylated, once or several times. The resulting alkyl groups will then be split off as alkenes, thereby regenerating a polymethylbenzene and completing a catalytic cycle. This multi-molecular mechanism type has been named the side- chain methylation mechanism.

In parallel, ongoing work by Kolboe and co-workers [19-21] has indicated that a monomolecular mechanism might lead to alkene formation from polymethylbenzenium species. Extensive isotopic labeling experiments were carried out by Bjørgen et al. [20], and the results were in accordance with the pathway for alkene formation outlined in Scheme 2. A fairly complex series of ring contractions and/or expansions leads to the growth of higher alkyl groups, which are lost as alkenes. A similar reaction scheme was proposed by Sullivan et al. [22] in 1961, to explain the formation of aliphatics from hexamethylbenzene on a bifunctional catalyst. In line with the naming introduced by Sullivan et al., we shall designate all such monomolecular routes leading to alkene loss from polymethylbenzenium species as the paring mechanism, regardless of the specific elementary steps involved.

In order to further elucidate the role of polymethylbenzenium cationic species in the

MTH-reaction, we found it worthwhile to investigate their intrinsic reactivity in an environment free from the surrounding catalyst framework, i.e. in the gas-phase. Hence, a series of protonated or methylated polymethylbenzenes (Scheme 3) was investigated using

4 mass analyzed kinetic energy (MIKE) spectrometry. Some reports on the gas-phase reactivity of cationic polymethylbenzenes already exist, in particular those published by Kuck and co- workers [23-26]. Also, Arstad et al. recently published a detailed theoretical study of cationic xylenium ions [27]. The results reported here are compared with relevant literature data and discussed with respect to the mechanism of the MTH-reaction.

2. Experimental

2.1. Chemicals. Compounds 0-6 were commercially available, at purities > 98 %. Compound

7 (1,2,3,3,4,5-hexamethyl-6-methylene-1,4-cyclohexadiene, denoted HMMC) was synthesized as the end product of a Friedel-Craft alkylation of pentamethylbenzene with chloromethane. Details have been given previously [21].

2.2. Calculations. Thermochemical data were obtained for compounds 7, [7 + H]+, and [7 +

+ CH3] using the G3MP2 computational scheme as implemented in Gaussian 03 [28]. These species have Cs symmetry.

2.3. Mass spectrometry. All mass spectra were recorded using a Fisons Prospec-Q, which is a hybrid mass spectrometer with EBEHQ configuration normally running at an acceleration voltage of 8 kV. For chemical ionization (CI) a tight ion source was used to ensure high- pressure conditions. For the protonation experiments methane was used as reagent gas

+ + providing the proton donors CH5 and C2H5 in approximately equal abundances. For the methylation experiments chloromethane was used as reagent gas providing mainly the methyl

+ + donor CH3ClCH3 plus minute amounts of CH3ClH . The precursor ion of interest was selected using the first two stages (EB). Ionic products from spontaneous decomposition in

5 the field free region following B (third field free region) were recorded with an orthogonal detector positioned in the fourth field free region, by scanning the second electric sector

(MIKE spectrometry). These spectra obtained in this manner are entitled metastable ion (MI) decomposition spectra. In separate experiments collisionally induced dissociation (CID) was achieved by bringing the selected precursor ions to collide with He in the third field free region. The He pressure inside the collision cell was set to attenuate the intensity of the precursor ion peak to ca. 30%. Relative abundances of fragment ions from metastable ion decomposition can in some cases be different from the relative peak heights reported in Table

1: a) detector response generally decreases with the translational energy of an ion, b) peak width and probability of detection strongly depends on translational energy release, and c) there will always be some unintended CID resulting from collisions with residual gas in the analyzer region, this effect is amplified in cases where only few ions give rise to MI.

3. Results

Compounds 0 – 7 (the numbering conveniently corresponds to the number of ring substituents) were all subjected to protonation and methylation, producing a wide series of homologue polymethylbenzenium cations in the gas phase. The MI spectra are summarized in

Table 1. The fragmentations of benzenium, toluenium [23] and xylenium [25] ions have been investigated previously, whereas the higher polymethylbenzenium ions have not, to the best of our knowledge, yet been studied using mass spectrometry. By employing both protonation and methylation, it is possible to produce the same — at least nominally — polymethylbenzenium ion from different starting materials, e.g. methylation of toluene and protonation of xylene will yield ions having the same stoichiometry. The energy of the ions will depend on their mode of formation, as will their structures. This aspect will be elaborated on in the discussion section.

6 + C6H7 (benzenium). Dehydrogenation of protonated benzene:

+ + C6H7 → C6H5 + H2 (1) was reported in 1974 [29] In addition to the predominant H2 loss, the MI spectrum of [0 +

H]+ (Table 1) shows one additional fragmentation reaction, namely loss of 28 Da, which must be ethene. This and a less intense peak for loss of 27 Da (vinyl) has considerable CID contribution.

+ + C7H9 (toluenium). The ion [1 + H] was easily formed by proton transfer. The

+ corresponding [0 + CH3] species is produced in the reaction between benzene and

+ CH3ClCH3 . In the latter case there is, however, a slight problem that the product species at m/z 93 is present in quite small amounts. This situation, as well as the presence of a stronger

+. signal for C7H8 (m/z 92), has the consequence that the ion beam with m/z 93 includes 10 %

13 12 +. of the C1 C6 isotopomer of C7H8 .

+ + The dominating peaks in the MI spectra of both [1 + H] and [0 + CH3] are due to

H2 loss. This reflects the observation that the mass spectrum of the ion source contents

(“normal” mass spectrum) is dominated by a peak at m/z 91. The second most abundant signal

+ + in the MI spectra of [1 + H] and [0 + CH3] (Table 1) is at m/z 77, corresponding to the loss of a methane molecule. This is a well-known reaction for the toluenium ion, and has been described by a number of workers. [26,29-32]. Smaller peaks are also seen at m/z 65 (- C2H4),

63 (- C2H6), 51 (- C3H6), and 39 (- C4H6). In addition, an extremely weak, not quantified signal is just visible as a weak shoulder to the methane loss peak corresponding to CH3 loss at m/z 78. This homolytic bond cleavage reaction requires 373 kJmol-1, formally a violation of the “even electron rule”. As we will learn in the forthcoming, compared to H2 loss the CH3

7 and CH4 losses become increasingly more favorable for the higher methylbenzenes, the

“forbidden” methyl loss even overtaking methane loss for the heptamethylbenzenium ion.

+ C8H11 (xylenium). The fragmentation of protonated p-xylene was extensively studied by

Mormann and Kuck [25] who reported loss of methane, ethene and ethane. They proposed a consistent reaction scheme with extensive isomerization prior to decomposition. In the present work, both p-xylene (2a) and o-xylene (2b) were investigated, and the MI spectra of [2a +

H]+ and [2b + H]+ are quite similar. Moreover, there are some differences from that of

+ methylated toluene [1 + CH3] (Table 1). The fragmentation patterns of all three spectra are, however, dominated by peaks at m/z 91, indicative of methane loss. Loss of H2 is also

+ observed, but in contrast to the unimolecular reactions from the lower homologues C6H7 and

+ + C7H9 this process is not the dominating. For [1 + CH3] the relative intensity for H2 loss is larger than for [2a + H]+ and [2b + H]+. For all three ions, [2a + H]+, [2b + H]+ and [1 +

+ CH3] , we observe peaks at m/z 79 (- C2H4), 77 (- C2H6), 65 (- C3H6), 63 (- C3H8), and 51 (-

C4H8). The relative heights of all these peaks are, however, variable among the precursors.

The high mass fragments are consistently more prominent in for the ions produced via methylation. This is mainly due to differences in energy (vide infra). We will also mention

+ that in addition to the MI spectra, we also recorded the CID spectra of the three C8H11 parent ions. First, the CID spectra are clearly distinct from the MI spectra, in particular by displaying the peaks for CH4 loss and H2 loss at the expense of higher mass neutrals. Second, the spectra

+ + of [2a + H] and [1 + CH3] are virtually identical with respect to H2 loss, while that of

+ [2b + H] is similar, having slightly less intense H2 loss relative to CH4 loss. All in all, these data represent a safety measure against the idea that the MI spectra include any significant contribution from unintentional CID, and also shows that the three structures (dynamical

8 mixtures of structures) are the same.

+ + C9H13 (mesitylenium). Protonation of mesitylene (1,3,5-trimethylbenzene) gave [3 + H] ,

+ + while methylation of o- and p-xylene gave [2a + CH3] and [2b + CH3] , respectively. The major unimolecular process was found to be methane loss in all cases, as evident from the dominating peaks at m/z 105. H2 loss is now strongly reduced. Except for the relative

+ + intensities of the H2 loss, the MI spectra of [2a + CH3] and [2b + CH3] are identical within experimental error. For the ions formed via methylation, fragmentation is somewhat more pronounced for losses of Cn units with n ≥ 2) i.e. the fragment ions m/z 93 (- C2H4), 91 (-

+ C2H6), 79 (- C3H6), 77 (- C3H8), and 65 (- C4H8) than for [3 + H]

+ C10H15 (durenium). The MI spectra of protonated durene (1,2,4,5-tetramethylbenzene), [4

+ + + H] , and methylated mesitylene [3 + CH3] were compared. The fragmentation patterns are analogous to those described so far: Methane loss results in the principal peak seen at m/z

119, and smaller peaks are also seen at m/z 107 (- C2H4), m/z 105 (- C2H6), m/z 91 (- C3H8),

+ and m/z 77 (- C4H10). Methyl loss is now becoming more prominent. The spectra of [4 + H]

+ and [3 + CH3] appear similar, except for an intensity difference for the loss of ethene.

+ + + C11H17 . Protonation of pentamethylbenzene (PMBz) gave [5 + H] , while [4 + CH3] was produced in the reaction between dimethylchloronium and durene. Methane loss (m/z 133) is the principal peak in both MI spectra also in this case. Additional appreciable signals are also seen at m/z 134 (- CH3), 121 (- C2H4), and 119/118. The signal at m/z 119/118 is a composite peak built up from the combined losses of C2H6 and C2H7, which might also be sequential loss of CH4 and CH3. Comparison of relative peak heights reveals only minor differences in

9 + + the fragmentation of [5 + H] compared to [4 + CH3] ; there is possibly an increased tendency

+ + for fragmentation other than methane loss for [4 + CH3] relative to [5 + H] . It worthwhile to notice that the mass spectra are considerably less dominated by methane loss than those described so far, with methyl loss becoming more significant.

+ C12H19 (hexamethylbenzenium). When hexamethylbenzene (HMBz) was subject to

+ + + + reaction with CH5 /C2H5 the ion [6 + H] is produced in high quantity, while [5 + CH3] is

+ the result of reacting CH3ClCH3 with pentamethylbenzene. At this stage, on the way to increasingly methylated systems, we observe a most noticeable shift in fragmentation pattern.

While methane loss has been the principal C1 fragment for all the lower homologues, methyl radical loss (m/z 148) now becomes equally predominant. Moreover, ethene loss is now less significant at m/z 135, overshadowed by the broad signal centered at m/z 133/132. It is also highly interesting to watch how comparable the intensity distributions for [6 + H]+ and [5 +

+ CH3] are. In addition to the major signals due to loss of CH3 and C2H6, smaller peaks are seen at m/z 121 (- C3H6), 117 (formally - C3H10, more likely - CH4 - C2H6), 107 (- C4H8),

105 (- C4H10), and 91 (- C5H12). Naturally, as the ions become larger, the list of mass losses becomes longer (Table 1).

+ C13H21 (heptamethylbenzenium). Protonation of HMMC leads to the heptamethylbenzenium ion [7 + H]+, for which no proton is likely to be bonded directly to the aromatic ring. Instead, there are two methyl groups bonded to one of the ring carbons. This distinctive structural feature is reflected in the observed fragmentation pattern, since methyl loss is clearly dominating. We also notice the corresponding fragment ion with m/z 162 to be a major component in the ion source spectrum. There is also a clear tendency for loss of

10 alkenes from the parent ion, evident from the major large peaks at m/z 149 (- C2H4), 135 (-

C3H6), and 121 (- C4H8 or/and - C2H4 - C2H4). The only additional signal is seen at m/z

147/146.

As was seen for [7 + H]+, loss of methyl (m/z 162) is the main peak also for

+ + [6 + CH3] . Methane loss is slightly more visible than for [7 + H] . The loss of ethene,

+ propene, and butene is prominent also for [6 + CH3] , but the relative peak heights are shifted

+ + considerably compared to [7 + H] . The spectrum of [6 + CH3] also contains several minor peaks not seen for [7 + H]+.

+ C14H23 . Methylation of HMMC is likely to occur mainly at the exocyclic double bond,

+ resulting in an ion containing an ethyl group. Hence, [7 + CH3] has an initial structure somewhat different from the previously described ions. Loss of the ethyl group in the form of ethene is the principal feature of the fragmentation pattern, giving rise to an ion with m/z 163.

Also, two peaks are seen at m/z 149 (- C3H6) and 149 (- C4H8). These are the only

+ fragmentation reactions observed for [7 + CH3] , as loss of methane or methyl appears to be insignificant for this ion.

4. Discussion

4.1. Internal energy of the metastable ions. The conditions in the ion source are not precisely known, so we do not know the internal energy distribution of the reacting ions. It is evident that thermodynamical equilibrium is not accomplished in the low-pressure plasma of the ion source, and it is likely that the ions leave the ion source without suffering many thermalizing collisions with neutral molecules. The detailed dynamics of the processes in

11 + + which MH and MCH3 are formed determine how much of the available energy that end up in the ions. Proton transfer is considered to be efficient in bringing of the order of 80% the enthalpy released into the MH+ ion [33-37], so we assume that a large fraction of the difference in proton affinity between the polymethylbenzene (M) and the corresponding base of the protonating molecule (methane or ethene) is transferred along with the proton:

+ + CH5 + M → MH + CH4 (2),

+ + C2H5 + M → MH + C2H4 (3).

+ There is also chance for protonation directly by CH4 , which has the potential for bringing even more energy into the MH+ ions. With regards to the methylation process, less is known about energy deposition during methyl cation transfer,

+ + CH3ClCH3 + M → MCH3 + CH3Cl (4).

Since it is likely that a straightforward SN2 type mechanism is in operation, it will be assumed that also here a majority of the available energy will end up in the ionic product. Table 2 lists the enthalpy changes for both proton transfer (PT) and methyl cation transfer (MT). The reader will notice that for the lower homologues the methyl cation transfer energy comes in- between the two proton transfer enthalpies, whereas the methyl cation transfer energy is closer to the lowest proton transfer energy for the higher homologues. This indicates a much broader, probably bimodal distribution for proton transfer.

Since the majority of high-energy ions have decomposed already in the ion source, the population of protonated molecules has been depleted by a larger relative fraction of its high- energy population upon leaving the ion source compared to a population of ions made by methylation. Consequently, we expect the average energy of metastable ions in the former

+ case to be lower. For C8H11 ions this is exactly what we observe. The higher energy process

+ of ethylene loss (vide infra) is more pronounced in methylated toluene [1 + CH3] , than in the

12 protonated xylenes [2 + H]+. If the energetic threshold of the processes in question lies in the

+ + region where the initial energy distribution of MCH3 has a maximum, while the MH has a minimum, this tendency will be even stronger. Upon finalizing this section, it should also be mentioned that cul-de-sac isomerization reactions preceding unimolecular decomposition may give rise to bimodal energy distributions [38]. As will be discussed in greater detail below, such isomerization reactions may indeed take place to a large degree in the systems discussed here.

4.2. Thermochemistry. Dehydrogenation of protonated benzene:

+ + C6H7 → C6H5 + H2 (5) was reported in 1974 [29] to occur with negligible translational energy release. Consequently, there is no reverse barrier, as evidenced in MP2 calculations [39]. The calculated forward barrier of 294 kJmol-1 is in good agreement with the experimental thermochemical minimum of 272 kJmol-1. The absence of a reverse barrier is in perfect agreement with the fact that

LUMO of the phenyl cation has the character of an empty p-orbital that points radially out from the periphery of the ring. As a result, this orbital has strong σ-electron accepting properties and the benzenium corresponds to a side-on adduct with H2 having C2v symmetry

[40], and H2 loss occurs with monotonically increasing potential energy towards the

+ dissociation limit. Isolable arenium ions are known, and the benzenium ion itself, C6H7 , has been inferred from NMR of super acidic solutions [41], and on the basis of X-ray

+ crystallography [42]. In addition to the predominant H2 loss, the MI spectrum of [0 + H]

(Table 1) shows one additional fragmentation reaction, namely loss of 28 Da, presumably ethene. Using available literature data [43], the thermochemical threshold for C2H4 loss is

-1 about 150 kJmol higher than that for H2 loss.

13 The thermochemical requirements for the three most abundant reactions from [1 + H]+

+ + and [0 + CH3] /C7H9 are:

+ + -1 C7H9 → C7H7 + H2 ; ∆H° = 54 kJmol (6)

+ + -1 C7H9 → C6H5 + CH4 ; ∆H° = 278 kJmol (7)

+ + -1 C7H9 → C5H5 + C2H4 ; ∆H° = 308 kJmol (8)

+ It has been suggested that formation of C7H7 could be accompanied by ring expansion, although it also consistent with a 1,2-H2 loss from the ipso protonated isomer. It the latter

-1 + case the reaction is endothermic by 54 kJmol , but irrespective of C7H7 isomer there is a significant reverse barrier to the reaction [29]. The methane loss reaction has been described by a number of workers. [25, 44-51]. Starting from the ipso protonated isomer a mechanism that parallels the H2 loss from benzenium (see above) can be envisaged. It is known that ca.

40 % of the methane lost includes a ring carbon, as shown by 13C labeling. [52]. This hints to a slow and essentially reversible ring expansion/contraction mechanism. The loss of C2H4 must by necessity involve at least one ring carbon atom from the parent ion. We assume cyclopentadienyl cation for the ionic product.

+ As reported above the major losses from fragmentation of C8H11 are

+ + -1 C8H11 → C8H9 + H2 ; ∆H° = 90 kJmol (9)

+ + -1 C8H11 → C7H7 + CH4 ; ∆H° = -33 kJmol (10)

+ + -1 C8H11 → C6H7 + C2H4 ; ∆H° = 161 kJmol (11)

+ + -1 C8H11 → C6H5 + C2H6 ; ∆H° = 311 kJmol (12)

+ + -1 C8H11 → C5H5 + C3H6 ; ∆H° = 276 kJmol (13)

14 A reaction scheme consistent with experiment was proposed by Mormann and Kuck

[26]. Extensive isomerization, probably including ring expansion/contraction, takes place prior to decomposition. This was also the conclusion of Arstad et al. based on B3LYP and

MP2 calculations [27].

It is noteworthy that the fragmentation patterns of ions made form methylation of the lower homologue are slightly different from that made by protonation of the higher. The fact

+ + that [1 + CH3] is richer in high energy fragments than [2 + H] indicates higher energy content of the metastable ions. This is a general trend, and the effect is particularly evident

+ + when comparing the spectra of [6 + CH3] and [7 + H] . Due to the non-aromatic structure of

7, protonation occurs with the transfer of much more energy than when 6 is methylated, and the loss of ethene and propene becomes of increasing relative importance. However, the data in the two last columns of Table 2 indicate that these energy differences will be smaller for the series of species from 2 to 6, as the enthalpy changes for methyl transfer and proton

+ transfer (from C2H5 ) become fairly equal.

4.3. Reactivity trends with size and complexity. Experimental reports concerning the reactivity of methylbenzenes in a zeolitic environment have shown that their reactivity depends on the number of methyl substituents. Arstad and Kolboe [11] have demonstrated that HMBz (or the ionic counterpart) decomposes much more rapidly than the tri-, tetra-, and pentamethylbenzene homologues after they have been synthesized inside the cages of the

SAPO-34 zeotype catalyst. Moreover, Song et al. [15] have suggested that the ethene/propene ratio in the product stream depends on the degree of methyl substitution on the benzene rings located in the catalyst pores; the lower homologues mainly yielding ethene and vice versa. It is therefore necessary to assess any trends in the gas phase reactivity of the ions studied here.

When analyzing the results displayed in Table 3, it must be considered that the data are

15 normalized with respect to the tallest peak, and that detector response is variable, as mentioned in the experimental section. With this in mind, the following points may be made:

1. The tendency for methane loss increases relative to dihydrogen loss with increasing

number of methyl groups (for number of substituents n ≤ 6).

2. For n = 7, methyl loss overtakes methane loss, reflecting the lower ionization energy

of the higher methylbenzene in combination with the absence of geminal hydrogen

atoms necessary for methane loss.

3. Ethane loss competes favorably with ethene loss for n ≤ 6.

4. When considering the loss of ethane from ions formed either via protonation of a

methylbenzene (not HMMC) or methylation of a methylbenzene, there is a reasonably

steady increase in the relative abundances with the number of methyl groups. This

might be correlated with an increased likelihood of having two methyl groups in ipso

positions, which is most probably required for ethane loss.

5. Butene loss (which could also be double ethene loss) is a prominent reaction only for

[7 + H]+.

6. The data do not show any increase in the propene/ethene ratio with the number of

methyl groups.

+ 7. For [7 + CH3] , alkene loss (C2H4, C3H6 and C4H8) is the only fragmentation

observed. The loss of the ethyl group as ethene is a well-known reaction for ethyl-

substituted benzenes. Arstad et al. have shown that the presence of a substiutent larger

than methyl (i.e. ethyl) results in a significant lowering of the barriers for propene and

butene loss compared to species with methylsubstituents exclusively [53]. This may be

rationalized by considering that the presence of an ethyl group allows transition states

16 involving secondary carbenium ion-like structures, rather than formally primary

carbenium ion-like structures.

4.4. Bearings on MTH-chemistry. As stated in the Introduction, the unimolecular reactivity of polymethylbenzenium ions is very likely of relevance to the mechanism of hydrocarbon formation from methanol in acidic zeolites. There are, however, two factors that must be taken into account at this point before attempting to draw firm conclusions. First, the validity of a comparison of the present set of results to the reactions actually occurring within the zeolite pores should be considered. In addition to any effects caused by steric limitations and/or the electrostatic field imposed by a surrounding zeolite on potential monomolecular ion reactions, it is conceivable that the zeolite itself is a vital participant in the reactions leading to alkene loss from polymethylbenzenium ions. For instance, it has been shown, using theoretical methods, that alkyl groups may be lost from benzenium ions in E2-type eliminations, wherein a zeolitic oxygen serves as the base [54]. If then the interaction of the organic cations and the zeolite wall is essential for alkene loss, or if the mechanism outlined in Scheme 1 is prevalent, the current data would be of minor relevance to MTH-chemistry.

This is a rather fundamental reservation, but the data discussed here will allow us reach our initial goal, which is to evaluate to what extent the monomolecular reactivity of the polymethylbenzenium ions is compatible with what is otherwise known about the MTH- reaction.

Second (and less decisive), gas phase molecules made by the processes described here are by far higher in internal energy than the corresponding ions within a zeolite cavity. A precise temperature cannot be given since the metastable ions are not a canonical ensemble, but the effect is substantial. Furthermore, interaction with surrounding molecules/zeolite

17 framework will speed up dissipation of the energy. For this reason the zeolite ions, once formed, will give rise to fewer reactions, favoring those of lowest critical energy.

The present set of data unequivocally demonstrates that alkenes may be split off from cationic polymethylbenzenes in monomolecular reactions, which is the expected outcome if the paring mechanism, as outlined in the introduction, is operative. Thus, it has been demonstrated that the paring mechanism is indeed feasible and not an intrinsic impossibility.

However, for all ions studied here (except methylated HMMC, see above) methane, methyl, or dihydrogen loss dominate, a feature incompatible with efficient production of C2 and higher hydrocarbons. Moreover, the experimental data and the thermochemical considerations show that methane, dihydrogen, and methyl loss are the least energy demanding processes of those observed. If these observations are pursued, two interesting implications for MTH- chemistry arise.

First, the loss of dihydrogen, methane, and ethane most likely involves the removal of two substituents in geminal positions on the ring. The resulting cationic product will be a substituted phenyl cation, a species consisting of a six-membered ring with no substituents on one of the ring carbons. Such species are usually not considered to play a prominent role in the MTH-reaction. They will, if present in a zeolite, display great reactivity, possibly serving as powerful hydride abstractors. Other possibilities might be intramolecular rearrangements or bond formation directly with the zeolite walls.

Second, the extensive loss of methyl radicals from the higher homologues in particular, merits a short discussion. Free radicals were suggested to participate in the MTH- reaction by Zatorski and Kryzyzanowski [55] as early as in 1978. Clarke et al. performed ESR experiments showing that free radicals may be present when dimethylether (which is formed from methanol during the MTH-reaction) is converted over zeolite catalysts, and it was suggested that dimethylether could be a source of methyl radicals [56]. However, it has been

18 concluded that NO does not act as a radical scavenger during methanol conversion [57]. Also, the ESR-signals are quenched by the presence of methanol [58], and the present consensus is against a radical mechanism [59]. Even so, when Kolboe recorded the ESR-signals throughout the induction period over a ZSM-5 catalyst, during which the hydrocarbon pool is built up within the zeolite voids and appreciable activity is reached, it was found that when the methanol feed was stopped, a clearly detectable ESR-signal due to organic free radicals appeared [58]. In light of the present data, it appears reasonable to assume that these ESR- signals were caused by the loss of methyl radicals from hydrocarbon pool species

(polymethylbenzenes) retained inside the catalyst pores after the methanol feed had been stopped.

5. Conclusions

The gas-phase reactivity of a wide series of polymethylbenzenium cations has been investigated using mass spectrometry. These ions have a rich and complicated chemistry and extensive fragmentation is observed. The initial scope of the study has been to evaluate the paring mechanism, which hypothesizes that monomolecular alkene loss from polymethylbenzenium cations is an important reaction in the zeolite-catalyzed methanol-to- hydrocarbons (MTH) reaction system. Loss of ethene, propene, and butene is observed, and the feasiblity of the paring mechanism has thus been demonstrated. However, loss of dihydrogen, methyl radicals, or methane is always much more prominent for the polymethylbenzenium ions. A comparison between mass spectrometric data and results from catalytic studies is not straightforward, because the ions produced in the mass spectrometer are far more energetic.

19 Acknowledgement. Thanks are due to the Norwegian Research Council for a grant of computer time through the NOTUR project (accounts NN2878K) and to Dr. Bjørnar Arstad for valuable discussions concerning his quantum chemical data.

References

1 M. Bjørgen, F. Bonino, S. Kolboe, K.-P. Lillerud, A. Zecchina and S. Bordiga, J. Am.

Chem. Soc., 2003, 125, 15863-15868.

2 M. Bjørgen, F. Bonino, S. Kolboe, K.-P. Lillerud, A. Zecchina and S. Bordiga, Angew.

Chem., Int. Ed., submitted.

3 J. A. Martens and P. A. Jacobs, in An Introduction to Zeolite Science and Practice, ed.

H. van Bekkum, E. M. Flanigen, P. A. Jacobs and J. C. Jansen, Amsterdam, 2001, p.

633-672.

4 B. Arstad, S. Kolboe and O. Swang, J. Phys. Chem. B, 2002, 106, 12722-12726.

5 S. Kolboe, in Zeolites and Mesoporous Materials at the Dawn of the 21st Century, ed.

A. Galarneau, F. Di Renzo, F. Fajula and J. Vedrinep, Amsterdam, 2001, p. 3946-

3954.

6 J. F. Haw, W. Song, D. M. Marcus and J. B. Nicholas, Acc. Chem. Res., 2003, 36,

317-326.

7 I.M. Dahl, S. Kolboe, Catal. Lett., 1993, 20, 329-336.

8 I.M. Dahl, S. Kolboe, J. Catal., 1994, 149, 458-464.

9 I.M. Dahl, S. Kolboe, J. Catal., 1996, 61, 304-309.

10 Ø. Mikkelsen, P.O. Rønning, S. Kolboe, Microporous Mesoporous Mater., 2000, 40,

95-113.

11 B. Arstad and S. Kolboe, Catal. Lett., 2001, 71, 209-212.

20 12 B. Arstad and S. Kolboe, J. Am. Chem. Soc., 2001, 123, 8137-8138.

13 P.W. Gougen, T. Xu, D.H. Barich, T.W. Skloss, W. Song, Z. Wang and J.B. Nicholas,

J. Am. Chem. Soc., 1998, 120, 2650-2651.

14 W. Song, J.F. Haw, J.B. Nicholas and C.S. Heneghan, J. Am. Chem. Soc. 2000, 122,

10726-10727.

15 W. Song, H. Fu and J. F. Haw, J. Am. Chem. Soc., 2001, 123, 4749-4757.

16 W. Song, H. Fu and J. F. Haw, J. Phys. Chem. B, 2001, 105, 12839-12843.

17 A. Sassi, M.A. Wildman, H.J. Ahn, P. Prasad, J.B. Nicholas and J. F. Haw, J. Phys.

Chem. B, 2002, 106, 2294-2303.

18 B. Arstad, J. B. Nicholas and J. F. Haw, J. Am. Chem. Soc., 2004, 126, 2991-3001.

19 M. Bjørgen, U. Olsbye and S. Kolboe, J. Catal., 2003, 215, 30-44.

20 M. Bjørgen, U. Olsbye, D. Petersen and S. Kolboe, J. Catal., 2004, 221, 1-10.

21 M. Bjørgen, U. Olsbye, S. Svelle and Stein Kolboe, Catal. Lett., 2004, 93, 37-40.

22 R. F. Sullivan, C. J. Egan, G. E. Langlois and R. P. Sieg, J. Am. Chem. Soc., 1961, 83,

1156-1160.

23 D. Kuck, J. Schneider and H.-F. Grützmacher, J. Chem Soc. Perkin Trans. II, 1985,

689-696.

24 D. Kuck, Mass Spectrom. Rev., 1990, 9, 583-630.

25 M. Mormann and D. Kuck, Int. J. Mass Spectrom., 2002, 219, 497-514.

26 M. Mormann and D. Kuck, J. Mass Spectrom., 1999, 34, 384-394.

27 B. Arstad, S. Kolboe and O Swang, J. Phys. Org. Chem., accepted for publication.

28 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R.

Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M.

Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N.

Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J.

21 Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.

E. Knox, H. P. Hratchian, J. B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R. E.

Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y.

Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S.

Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K.

Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J.

Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L.

Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M.

Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, and J.

A. Pople, Gaussian 03, Revision B.04, Gaussian, Inc., Pittsburgh PA, 2003.

29 D. H. Williams and G. Hvistendahl, J. Am. Chem. Soc., 1974, 96, 6755–6757.

30 M. S. B. Munson and F. H. Field, J. Am. Chem. Soc., 1967, 89, 1047-1052.

31 F. H. Field, J. Am. Chem. Soc., 1967, 89, 5328-5334.

32 G. Hvistendahl and D. H. Williams, J. Chem. Soc. Perkin Trans. II, 1975, 881-885.

33 H.-H. Bueker and E. Uggerud, J. Phys. Chem., 1995, 99, 5945-5949.

34 H.-H. Bueker, T. Helgaker, K. Ruud and E. Uggerud, J. Phys. Chem. 1996, 100,

15388-15392

35 T. S. Zwier, V. M. Bierbaum, G. B. Ellison and S. R. Leone, J. Chem. Phys., 1980, 72,

5426-

36 J. C. Weisshaar, T. S. Zwier and S. R. Leone, J. Chem. Phys., 1981, 75, 4873-4884.

37 R. D. Bowen and A. G. Harrison, Org. Mass Spectrom., 1981, 16, 159-166.

38 S. Hammerum, T. Vulpius and H.-E. Audier, Org. Mass Spectrom., 1992, 27, 369-

376.

39 E. del Río, R. López and T. L. Sordo, J. Phys. Chem. A, 1997, 101, 10090–10094.

22 40 M. N. Glukhovtsev, A. Pross, A. Nicolaides and L. Radom, J. Chem. Soc., Chem.

Commun., 1995, 2347-2348.

41 G. A. Olah, J. S. Staral, G. Asencio, G. Liang, D. A. Forsyth and G. D. Mateescu, J.

Am. Chem. Soc., 1978, 100, 6299-6308.

42 D. Stasko and C. A. Reed, J. Am. Chem. Soc., 2002, 124, 1148-1149.

43 NIST Standard Reference Database Number 69 - March, 2003 Release,

http://webbook.nist.gov/chemistry/

44 H. E. Audier, C. Monteiro, P. Mourgues and D. Berthomieu, Org. Mass Spectrom.,

1990, 25, 245-246.

45 D. Berthomieu, V. Brenner, G. Ohanessian, J. P. Denhez, P. Millié and H. E. Audier,

J. Phys. Chem., 1995, 99, 712-720.

46 J. P. Denhez, H. E. Audier and D. Berthomieu, Rapid Commun. Mass Spectrom.,

1995, 9, 1210-1212.

47 C. Matthias, S. Anlauf, K. Weniger and D. Kuck, Int. J. Mass Spectrom., 2000, 199,

155-187.

48 D. Kuck and C. Matthias, J. Am. Chem. Soc., 1992, 114, 1901-1903.

49 C. Matthias, and D. Kuck, Org. Mass Spectrom., 1993, 28, 1073-1081.

50 C. Matthias, K. Weniger and D. Kuck, Eur. Mass Spectrom., 1995, 1, 445-455.

51 C. Matthias and D. Kuck, Int. J. Mass Spectrom., 2002, 217, 131-151.

52 D. Kuck, Int. J. Mass Spectrom., 2002, 213, 101-144.

53 B. Arstad, S. Kolboe and O. Swang, in preparation.

54 B. Arstad, S. Kolboe and O. Swang, J. Phys. Chem. B, 2004, 108, 2300-2308.

55 W. Zatorski and S. Kryzanowski, Acta Phys. Chem., 1978, 24, 347-352.

56 J. K. A. Clarke, R. Darcy, B. F. Hegarty, E. O’Donoghue, V. Abir-Ebrahimi and J. J.

Rooney, J. Chem. Soc., Chem. Comm., 1986, 425-426.

23 57 G. J. Hutchings, in Natural Gas Conversion, ed. A. Holmen, K.-J. Jens and S. Kolboe,

Amsterdam, 1991, p. 405-412.

58 S. Kolboe, in Natural Gas Conversion, ed. A. Holmen, K.-J. Jens and S. Kolboe,

Amsterdam, 1991, p. 413-420.

59 M. Stöcker, Microporous Mesoporous Mater., 1999, 29, 3-48.

24 + + Table 1. MIKE spectra of adducts between H /CH3 and methylbenzenes. The numbers are relative peaks heights normalized to the tallest peak in each spectrum. For reasons explained in the experimental section these peak heights can in some instances be quite different from the relative abundances. An asterix indicates that overlap with a broad and more intense neighboring peak does not allow for precise quantification of the signal. Some likely fragments are displayed. Other possibilities may exist, in particular for the larger mass losses.

3 4 6 4 6 4 6 or 3 4 3 4 4 or or 2 7 8 10 H H H H H H H

H 4 2 2 2 3 3 2 2 H H H C -H 2 3 H C 3 -CH -CH -CH - -CH -C -C -C -C -C -C - C -C -C - -C

Compound m/z -2 -15 -16 -27 -28 -30 -31 -32 -40 -42 -44 46 -47 + a a a Benzene (0) H 79 100 11 13 5 + CH3 93 100 33 10 7 Toluene (1) H+ 93 100 44 11 5 8 + CH3 107 39 100 19 10 10 6 o-Xylene (2a) H+ 107 35 100 11 8 7 5 + CH3 121 100 14 20 13 12 + p-Xylene (2b) H 107 21 100 11 6 8 * + 25 CH3 121 6 100 16 17 13 12 Mesitylene (3) H+ 121 6 100 10 13 9 8 CH + 135 10 100 17 11 5 7 3 Durene (4) H+ 135 10 100 7 12 4 7 + b CH3 149 36 100 14 19 * 5 6 PMBz (5) H+ 149 27 100 10 16b * 5 5 + b b CH3 163 100 100 19 30 * 7 8 HMBz (6) H+ 163 100 100 14 28b * 10 9b + b b CH3 177 100 33 27 18 * 18 7 + HMMC (7) H 177 100 13 14 < 5 9 40 + CH3 191 100 4 a) The relative areas of the peaks resulting from ethene and vinyl losses is around 10 :1. Some CID is inferred. b) The peak is composite, making it difficult to quantify the relative contribution from two neighboring peaks...... Table 1. contd.

6 4 10 12 10 or 8 H H 4

H H H 2 4 4 5 5 H H 4 2 -C -C -C -C -C C -C - Compound m/z -54 -56 -58 -60 -66 -68 -70 -72 -80 -82 -84 -86 -94 -96 -98 -100 Benzene (0) H+ 79 + CH3 93 4 Toluene (1) H+ 93 5 + CH3 107 * 7 4 o-Xylene (2a) H+ 107 * 4 * 3 + CH3 121 7 * * 5 * 4 p-Xylene (2b) H+ 107 * 4 * 3 + CH3 121 5 * * 4 * 3 Mesitylene (3) H+ 121 3 * * 2 * 2 +

CH3 135 3 4 2 * 1 * 1 26 Durene (4) H+ 135 4 4 1 * 1 * 1 + CH3 149 * 6 * 3 2 2 PMBz (5) H+ 149 * 5 * 2 1 1 + CH3 163 * 5 * 5 1 2 1 HMBz (6) H+ 163 5 6 * 7 * 3 2 + CH3 177 8 5 * 4 * 4 * 2 HMMC (7) H+ 177 33 + CH3 191 4

Table 2. Thermochemical data (kJmol-1). Compound [M] ∆Hf°[M] IE [M] PA [M] ∆Hf°[MH+] ∆H°, PTa ∆H°, MTb Benzene (0) 83 892 750 863 199, 70 112 Toluene (1) 50 852 784 796 233, 104 122 o-Xylene (2a) 19 826 796 753 245, 116 165 p-Xylene (2b) 18 815 794 754 243,116 165 Mesitylene (3) -16 811 836 678 285, 156 169 c Durene (4) -43 818 846 641 295, 166 171 Pentamethylbenzene (5) -67 765 851 612 300, 171 166 Hexamethylbenzene (6) -77 758 861 592 310,181 178 HMMC (7) (-5) (955) (570) (404, 275)

27 CH4 -75 551 904

C2H4 52 680 902

CH3ClCH3+ 743

CH3Cl -82 1087

CH3 146 950

C7H7+(benzylium) 900

C7H7+(tropylium) 850

C H + 1149 6 5 C H -84 2 6 C H 20 3 4 Cl 121

Data taken from NIST web page, those in parentheses are quantum chemical data obtained at the G3MP2 level of theory. a) STP enthalpy change for proton transfer to compound from methonium ion, CH5+(first entry) and C2H5+(second entry) b) STP enthalpy change for methyl cation transfer to compound from dimethylchloronium ion, CH3ClCH3+. c) Data for the 1,2,3,5-tetramethylbenzene isomer was used.

28 + MeOH +

- H2O HZeo Zeo- HZeo

+ MeOH + +

- H2O - Zeo Zeo- HZeo

Scheme 1

29 + +4 MeOH HZeo Zeo- +3 MeOH -H+

+ + Zeo- HZeo Zeo- -H+

+ + +

- - Zeo + - Zeo Zeo- Zeo

Scheme 2

30 0 1 2a 2b 3

4 5 6 7

Scheme 3