Molecular Heterogeneous Catalysis, Wiley (2006), 352729662X

176 Chapter 4

not correspond to the pre-transition state adsorption mode from which the reaction occurs. In the preferred associative reaction path, the CH3 + cation generated in the transition state is directly transferred to the oxygen atom of methanol without the formation of the methoxy intermediate. The reaction energies for the two reaction mechanisms involved in dimethyl ether formation are compared in Fig. 4.14. Note that with respect to the energies of the molecules in the gas phase, the activation barriers for the elementary reaction steps that correspond to the association reaction path are substantially lower than the barriers for the consecutive reaction path.

We conclude this section by presenting the sca olding e ect, whereby adsorbed polar molecules such as H2O or H2S can significantly lower the activation energies for di erent hydrocarbon conversion reactions. In order to illustrate this e ect, we will focus on the trans–alkylation reaction[17]. In this reaction, a methyl group from one aromatic mole– cule is transferred to another aromatic molecule. In the “alkoxy”-mediated reaction path the methoxy species are formed by the C–C bond cleavage of the methyl–phenyl bond.

The transition state for this C–C bond cleavage is shown in Fig. 4.6. Figure 4.15 shows the transition state for the same reaction but now in the presence of water[18]. The activation

energy for this C–C bond formation reaction from the adsorbed methoxy is lowered by 50 kJ/mol in the presence of water. The presence of water significantly stabilizes the charged carbenium ion that forms, thus significantly lowering the activation barrier. The presence of water alters the structure of the carbenium ion by increasing the angle between the CH3 + group and the benzene ring. The angle is nearly perpendicular when water is present (see Fig. 4.15) thus optimizing the interaction with the benzene π-electrons. The e ect of coadsorbing non-reacting polar molecules is, in this case, purely geometric. The water molecule acts as a sca old. Similar assistance e ects have been found in several other reactions such as the C–C bond formation reaction from methanol which is discussed later.

Figure 4.15. Transition state for alkylation of toluene by methoxy species in the presence of water[18]. Within lattice negatively charged oxygen atoms around Al are indicated.

Shape-Selective Microporous Catalysts, the Zeolites 177

4.2.2 Transition-State Selectivity. Alkylation of Toluene by Methanol Catalyzed by Mordenite

In the previous section, we explained the steric and energetic consequences of the zeolite micropore shape for the activation energy of an elementary zeolite-catalyzed reaction step. In this section we discuss transition-state selectivity whereby di erences in selectivity are ascribed to a more or less optimum match of the reaction transition state with the micropore cavity. We will demonstrate that the di erence in the selectivity for the reaction is determined by the probability that a preferred pre-transition state orientation can form rather than by di erences in the activation barrier for the reaction step in which protonation occurs. This result is analogous to the finding that the preferred reaction channel for enantioselective homogeneous catalysts proceeds through the adsorption complex with the most favorable free energy (see Section 2.4.4). Similarly, we will discuss the importance of the pre-transition-state complex in enzyme catalysis in Chapter 7.

To illustrate the influence of pre-transition-state control for a zeolite-catalyzed reac-

tion, we describe the results of a quantum-chemical study of the alkylation of toluene by methanol[18]. The selectivities to produce ortho-, meta- or para-xylene in the channel of

the Mordenite zeolite are studied. The study illustrates the e ect of the spatial constraint induced by the one-dimensional 12-ring micropore channel dimension on the selectivity of the zeolite catalyzed-reaction.

The reaction mechanism for the alkylation of toluene is well understood and is illustrated in Fig. 4.16 for the production of para-xylene.

Figure 4.16. Mechanism of the alkylation reaction of toluene by methanol catalyzed by an acidic zeolite.

Methanol is first activated in the zeolite to form the CH3 + intermediate as was described in Section 4.2.1. The CH3 + intermediate subsequently adds to the para position

on the adsorbed toluene to form para-xylene. The CH3 + intermediate can similarly attack the adsorbed toluene at the ortho and meta positions to give ortho- and meta-xylene,

respectively.

The reaction energy diagram for the alkylation of toluene by methanol in mordenite is presented in Fig. 4.17. In this reaction energy diagram the energy changes for the first two steps refer to the heat release that occurs from the adsorption of methanol and toluene, respectively. The interaction of toluene with the zeolite channel atoms is primarily controlled by the van der Waals interactions between the toluene atoms and channel oxygen atoms. DFT calculations tend to describe these interactions poorly. The van der Waals-type interactions have therefore been empirically estimated and added to the quantum-chemical interaction energies.

The key to understanding of the energy di erences of this zeolite-catalyzed reaction is an appreciation of the di erences in energy cost to reorient adsorbed CH3OH towards the

178 Chapter 4

para, meta or ortho-carbon atom of toluene that becomes alkylated. In mordenite, the energy di erences of these pre-transition-state structures are controlled by the repulsive interactions that arise due to discrepancies of intermediate shapes for ortho-, meta- and para-orientated pre-alkylation complex with the channel shape and size. With respect to

the ground-state energies of the pre-transition-state intermediates, methylation of toluene occurs with nearly similar energies to the ortho-, meta- and para-positions. Since the

para pre-transition-state structure is most stabilized, this reaction channel gives the lower

overall activation energy for formation of para-xylene compared with the reaction channels that give ortho- and meta-xylene.

The concept of the pre-transition states relates zeolite catalysis to enzyme catalysis. As we will see in Chapter 7, a major di erence between the zeolite and the enzyme is the limited stabilization of the intermediate in the zeolites which is due to the high rigidity of the zeolite framework. In contrast to enzymes, the zeolite lattice is rather inflexible. It will not adjust to the shape of the desired transition-state structure in previous figures, hence the barriers for proton activated reactions will remain high as compared with those of enzyme-activated reactions.

Figure 4.17. Zeolite transition-state selectivity. Toluene alkylation with methanol catalyzed by H-MOR showing the energies of the key reaction intermediates[18]. Reaction energy diagram for ortho-, meta- and para-xylene are compared.

4.2.3 Lewis Acid Catalysis

4.2.3.1 Lewis Acidity in Zeolites; Cations Compared with Oxy-Cations

The reactivity of a zeolite activated by ion exchange with a soft Lewis acid cation will be examined in detail by following C–H bond activation over a Zn2+ ion. We compare the results with those for the reactivity of the ZnOZn2+ oxycation, often also formed in experimental systems during the ion exchange-reaction of Zn2+ into zeolites. As an introduction to Chapter 7 on biocatalysis, we will also discuss the hydrolysis of acetonitrile by a Zn2+ ion exchanged into the micropore of a zeolite. A comparison will be made with the reactivity of Ga+ and polarization e ects due to a hard Lewis acid such as Mg2+.

Let us first analyze the interaction of probe molecules such as CO with the Zn2+ and related cations in some detail. The adsorption of CO can be used to help understand the di erences in the electrostatic polarizing properties of cations in zeolites. For nonreducible cations, there is an upwards shift of the CO frequency which is proportional to qr , where q and r are the charge and the radius of the cation, respectively. Rehybridization

of predominantly the σ-type orbitals in CO leads to a depopulation of the antibonding C–O orbital. This results in a strengthening of the CO bond and, hence, an upward shift of the CO vibrational frequency.

Figure 4.18. Hybridization model of CO σ-type valence orbitals. The relation with the molecular orbitals as computed for CO (see Fig. 3.4) is indicated. 1) Hybridized s,pz atomic orbitals on C, 2) Hybridized s,pz atomic orbitals on O.

As illustrated in Fig. 4.18, the antibonding nature of the CO 5σ orbital can be readily deduced from a CO chemical bonding picture based on hybridization of the C and O, 2s and 2pz atomic orbitals. In Chapter 3 we presented calculated CO molecular orbitals and energies (page 93) and in the Addendum to that chapter we gave a short introduction to hybridization

The linear combination of the 2s and 2pz atomic orbitals on each of the atoms leads to the formation of four molecular orbitals. A bonding and antibonding pair of molecular orbitals with a large di erence in energy is generated from the hybridized atomic orbitals oriented toward each other, called 3σ and 6σ orbitals. The hybridized atomic orbitals involved strongly overlap and, hence, occupation of the bonding orbital strongly contributes to the strength of the C–O bond. The energy di erence between the other pair of bonding and antibonding orbitals is much smaller since they are formed from the hybridized atomic orbitals that are not directed towards one another.

The results in Fig. 4.18 also show the higher occupied 5σ orbital as the lone-pair orbital localized on C, which is antibonding with respect to the C–O bond. The corresponding bonding combination is the 4σ orbital, that is, the lone pair orbital localized mainly on the oxygen atom.

Table 4.1. CO orbital energies and their relative shifts (eV)

180 Chapter 4

Table 4.1 and Fig. 4.19 compare DFT-computed orbitals of CO when adsorbed on Mg2+, Sr2+ and Zn2+ cations adsorbed to a four-ring structure of Siand Al-containing tetrahedra.

∆σ = E5σ − E4σ, ∆π = E2π − E1π , ∆σπ = E5σ − E1π

Figure 4.19. The interaction of CO with Mg2+, Sr2+ and Zn2+ coordinated to a four-ring cluster Si2Al2O4 (OH)4 2−: Local densities of states of CO orbitals are shown as a function of orbital energy:

(d) CO molecular orbitals; rCO=1.144 kcal/mol. The local densities of states projected on C and O are shown. The exact LDOS are delta functions. They have been artificially broadened for ease of visualization.

Table 4.1 compares the relative energy positions of the CO molecular orbitals and their respective energy di erences.

The interaction between the cation and CO is seen to lower all of the CO orbital energies. The reduction of the di erence in energy between CO 4σ and 5σ molecular orbitals is due to the stronger lone-pair CO 5σ interaction with the cation. This results

in a small rehybridization of the CO σ orbital. The increased CO bond strength relates to a reduction of the antibonding character of the 5σ orbital. The increased interaction with Zn2+ is clearly seen to be due to the additional interaction with the Zn dz2 orbitals.

One should also note the decreased di erence in energy between CO 1π and 5σ molecular orbitals for CO adsorbed to Sr2+ and Mg2+. The di erence in energy for CO adsorbed on Mg2+ is lower than that for Sr2+. The 5σ orbital directed towards the cation experiences a larger electrostatic attraction than the 1π orbital perpendicular to the CO–cation interaction axis. The decrease in 5σ–1π interaction is nearly proportional to the di erence in cation–carbon distance.

The bond energies decrease rapidly with bond distance. One expects for polarizable systems a dependence with radius of r−4. Inspection of the orbital pictures in Fig. 4.19 immediately indicates the di erence in the interaction between hard Lewis acid cations such as Mg2+ and Sr2+ and soft cations such as Zn2+.

The presence of cations in zeolites can significantly a ects the reactivity of coadsorbed small metal particles[19]. The influence of cations on transition-metal clusters is mainly electrostatic. The electrostatic field generated by the cation polarizes the metal particle. This polarization dramatically a ects the reactivity of the small metal cluster. Calculations analyzing the interaction of an H2 molecule with a Ir4 cluster in the presence and absence of an Mg2+ cation illustrates the importance of induced polarization e ects.[20].

Figure 4.20. Adsorption geometries[20] for hydrogen on Mg2+-promoted Ir4 clusters.

H2 weakly interacts with an isolated Ir4 tetrahedron in an end-on perpendicular adsorption mode. The adsorption of H2 to the same cluster is largely enhanced, however, if the Ir4 cluster interacts with an Mg2+ cation. Hydrogen adsorption is enhanced at positions where the electron density is most reduced due to the polarization of the Ir4 atoms by Mg2+. Most striking is the increase in electron density of the antibonding H2 orbital that results from structure 6 in Fig. 4.20 (see Table 4.2).

In the absence of Mg2+, the interaction between H2 and Ir4 is dominated by the strong Pauli repulsive interaction between the doubly occupied orbitals of H2 and Ir4. In the presence of Mg2+, however, the Ir4 is polarized by Mg2+, which reduces the electron

182 Chapter 4

density between H2 and Ir4. The resulting reduction of the Pauli repulsion allows the H2 molecule to approach the Ir4 plane in the parallel adsorption mode with strong orbital overlap. The result is a strong activation and weakening of the H2 bond. This study illustrates how polarization of metal particles can dramatically alter their chemical reactivity, especially with respect to closed shell systems such as H2 or CH bonds.

Table 4.2. The molecular orbital population and the interaction energy (∆E kJ/mol) for hydrogen

chemisorbed at di erent positions on the Mg2+ promoted Ir4 clusters shown in the structures of Fig. 4.20[20]

Ir4–H2

The interaction of a reacting molecule with a metal cation in the zeolite is not only determined by cationic chemical-bonding properties, but also by the negative lattice charge distribution that compensates for the positive charge of the cation. This is elegantly

demonstrated by infrared measurements of the absorption intensities for methane adsorbed on Zn2+ cations in low Al-ZSM-5 and high framework Al-zeolite Y[21]. In the latter,

the framework negative charge is the highest and hence the e ective positive charge on Zn2+ is the smallest. For this reason, the infrared intensity for the vibrational excitation of methane that is only spectroscopically allowed in the presence of the electrostatic field of the zeolite is highest in the low Al-ZSM-5. This is shown in Fig. 4.21.

In Fig. 4.21 one notes the large di erence in the intensity of the vibrational excitation around 2800 cm−1 in methane adsorbed on Zn2+ in ZSM-5 and zeolite Y. This excitation is not observed in the gas phase, where by symmetry it is vibrationally forbidden. The e ective charge of the Zn2+ cation polarizes the molecule, which results in symmetry breaking, thus allowing for the excitation of this symmetric CH4 mode. The lower e ective charge of Zn2+ in the faujasite structure zeolite Y with higher Al concentration results in a smaller polarization and hence a decreased relative intensity.

Cations prefer particular sites in the zeolite. The divalent Zn2+ cations prefer adsorption in a ring of framework tetrahedra with at least two Al framework cations, so that an overall charge neutral site is generated. There is also a relationship between the size

Figure 4.21. Methane adsorption by zinc-modified zeolite. Comparison between a low Al/Si framework ratio in ZSM-5 and a high Al/Si ratio in zeolite Y[21].

of the cation and the size of the (SixAl(1−x)O2)n ring system, which determines which adsorption is the most preferred. For Zn2+ coordination, the six-ring of (Si2/3Al2/3O2)6 is preferred. In some zeolites such six-rings have di erent local zeolite framework environments. The tendency to accommodate lattice deformations, due to Zn2+ attachment, may then vary, which will also a ect the preferred siting. Zn2+, for example, prefers the

six-ring lattice position whereas the larger ZnOZn2+ oxycation prefers the larger zeolite 8-ring[22] as its optimal site.

The activation of C–H bonds for di erent hydrocarbons can occur both at Zn2+ and ZnOZn2+ sites. We will first discuss hydrocarbon activation by Zn2+. The results presented here are based on quantum-chemical cluster calculations. The reaction energies involved in the overall catalytic cycle for the activation of ethane over a Zn2+ cation and a ZnOZn2+ oxycation adsorbed on a representative cluster chosen to model the ZSM-5 adsorption site are compared in Fig. 4.22.

The activation of an alkane by Zn2+ occurs through formation of a Zn–alkyl species and a zeolitic H+. The proton adsorbs on a basic lattice oxygen atom that connects a silicon with an aluminum lattice cation[24].

When the alkane molecule reacts with the ZnOZn2+ oxycation the proton binds to the oxycationic oxygen atom, which has a much higher reactivity than the zeolite lattice oxygen atom. As a consquence, the initial activation of the C–H bond at the ZnOZn2+ site is highly preferred over Zn2+.

In a subsequent reaction step, the alkene is generated from the Zn–alkyl complex, by a β- C–H cleavage reaction, which is endothermic. The catalytic cycle closes by recombination of the two hydrogen atoms to give H2. This final step is much more di cult for the ZnOZn2+ center than for Zn2+, because of the much larger [ZnOHZn] hydroxyl-bond energy compared with that of the zeolitic proton.

In the case of Zn2+, the basic zeolite oxygen atoms can be considered to act as a reactive ligand to Zn2+ assisting heterolytic cleavage reactions. This is quite common for reactive charged cations in zeolites and can be considered as a spillover e ect. The reduction of cations will not always lead to the generation of zeolitic protons.

The activation of H2 by Ga+ is proposed to proceed di erently since no protons are observed experimentally when H2 or alkane adsorbs. Homolytic hydrogen dissociation

Homolytic oxidative addition of Ga+ with H2 is a slow reaction (Eact = 240 kJ/mol, because the back-donative interaction with a Ga+ d-atomic orbital is very weak (see Chapter 3, page 129). This reaction is quite di erent from heterolytic dissociation with an activation barrier of only 60 kJ/mol for H2 on Zn2+ to form ZnH+ and a zeolite proton. Catalysis by Ga is complex in its chemistry. The GaO+ species can activate the C–H bond, but H2 recombination is slow. Also, GaO+ may reduce during reaction and GaH2 + can be formed. Both Ga+ and GaH2 + can activate the C–H bonds of alkanes. The chemistry is clarified by theoretical results that show that actually both systems for hydrocarbons heterolytic bond cleavage is the preferred reaction path[24b]. However, the intermediate GaHR+ is thermodynamically preferred over ZOH–GaR. Therefore, after initial heterolytic dissociation by Ga+ the intermediate GaHR+ is rapidly formed. The consecutive reaction to alkenes proceeds via GaHR+, explaining the absence of a proton signal in infrared experiments[24b]. According to Kazansky et al.[25], the reduction of GaO+ results in the formation of low-coordinated gallium or gallium hydrides;

They predicted for reaction (a) a reaction energy of −243 kJ/mol, reaction (b) a reaction energy of +130 kJ/mol and for the oxidative addition reaction (c) a reaction energy of −57 kJ/mol. For the last reaction an activation energy of at least 240 kJ/mol has been computed, to be compared with only 60 kJ/mol for reaction (a).

The reduction of cations such as Pd2+ and other transition metals proceed analogously to the C–H activation events discussed above for Zn2+. Hydrogen dissociatively adsorbs to form [PdH]+ and a zeolite proton. This is followed by the subsequent activation of a second H2 molecule to form PdH2 and another zeolitic proton. Hydrogen readily desorbs from Pd, leaving a reduced metal atom next to two zeolitic protons[19].

Cations such as Zn2+ or Ga+ behave as soft Lewis acids in the reactions discussed above with the formation of intermediate metal–alkyl or metal–hydride species. This implies an electron transfer between the ligand and cation. To illustrate further the Lewis acid nature of Zn2+, we analyze the mechanism for the hydrolysis of CH3CN in which there is no change in formal valency of Zn2+, and compare the energetics for this ionexchanged Zn2+ reaction with that for the zeolitic proton[26]. The overall reaction scheme is

CH3CN + H2O −→ CH3C(O)NH2

Acetonitrile adsorbs strongly on Zn2+. Its calculated interaction energy with the Zn2+ site in the zeolite model is −126 kJ/mol. In contrast, it interacts much more weakly with the zeolitic proton: Eads = −46 kJ/mol. The product molecule of the hydrolysis reaction, acetamide [CH3C(O)NH2], however, has an adsorption energy of −73 kJ/mol with Zn2+ as it only coordinates through the basic nitrogen atom of acetamide. Acetamide binds to the proton in the protonated zeolite at −90 kJ/mol. It adsorbs in a bidentate configuration where the carbonyl oxygen atom binds to the H+ and via the amide N–H group with the basic zeolite oxygen atom. The rate of the overall hydrolysis reaction of