Molecular Heterogeneous Catalysis, Wiley (2006), 352729662X

226 Chapter 5

giving COcus and Ocus, respectively. The calculated binding energies, di usion barriers and activation energies for di erent surface reaction steps on the RuO2(110) surface are given in Table 5.1. One recognizes the small interaction energy of Oad as Ocus and the relative strong bonds of three-fold coordinated CObr and Obr . Figure 5.9b illustrates the di erent surface phases that are formed in thermodynamic equilibrium with gas-phase CO and O2. At high O2 pressure and low CO pressure, the ideally terminated RuO2

surface is stable (Obr /Ocus ). When the CO pressure increases, CO initially adsorbs as COcus and at very high pressure begins to substitute not only for Ocus but also for Obr .

The Dynamic Monte Carlo results show that maximum CO oxidation occurs at a surface that is to be described as a disordered phase. This phase occurs in the boundary region of the stable phases shown in Fig. 5.9b, in particular between the Obr /COcus and Obr /Ocus phases. At low CO partial pressure and relatively high O2 pressure the Obr /Ocus phase is stable, on which CO only rarely adsorbs and the CO2 rate is extremely low. When the CO pressure increases, at T = 600 K, PCO = 20 atm and PO2 = 1 atm, the average occupation numbers are

NCObr = 0.11, NCOcus = 0.70, NObr = 0.89 and NOcus = 0.29. The rate of CO2 formation is high and dominated by the reaction between COcus adsorbed to coordinatively un-

saturated Ru and Ocus , the coordinatively unsaturated apex O atom. At very high CO pressure the CObr /COcus surface is formed, close to the state of the reduced Ru surface. The rate of CO2 formation is now suppressed because the surface is poisoned with COad.

5.3.2 Atomic Orbital Hybridization at Surfaces; Hydration Energies

The di erences in chemisorption between transition metal and transition-metal oxide surfaces, as discussed above, are quite general and are very similar for the other oxides and for sulfides and other ionic solids. The electronic factors that control this behavior can be captured by relating the cations in the oxide to the metal atoms in well-defined coordination complexes.

By way of example, we probe the adsorption and dissociation of water. Water is also known to dissociate heterolytically on the RuO2 surface. The 5s and 5p electron energies of Ru (a second-row transition metal atom, located in Group VIII, with higher ionization energies) are lower than the corresponding 4s and 4p electronic energies of Ti because of the higher e ective nuclear charge of Ru. Hence the OH− anion will be more strongly bonded to the unoccupied lone-pair orbital of Ru4+ than Ti4+. The Ru-oxide surface will hence, be more Lewis acidic than that of Ti oxide. On RuO2 the RuO bonds are stronger than in TiO2 and, hence the Brønsted acidity of the hydroxylated RuO2 (110) surface is also greater. The overall result will be that H2O dissociates more easily on RuO2 than TiO2.

The chemical reactivity of transition-metal oxides or sulfides can be understood in terms of elementary quantum-chemical concepts by using the hybridization schemes introduced earlier for the metal–carbonyl clusters analyzed in Section 3.3.1. The following hybridization schemes are proposed for di erent geometric arrangements: six d2sp3-orbitals for octahedrally coordinated transition metals with three nonbonding d-orbitals; four-sp3 orbitals with five nonbonding d-atomic orbitals and also four dsp2 orbitals for planar four-coordinated metal atoms and two nonbonding d-atomic orbitals (see the Addendum in Chapter 3 for a background on hybridization).

This can be generalized to bonding in the oxides and sulfides, when each anion neighbor of a cation is considered to contribute two electrons to σ-type cation–anion covalent bond.

The electronic structure on the cation is very similar to that in the coordination com-

Catalysis by Oxides and Sulfides 227

Figure 5.10. Isolable orbital schemes indicating similarity in reactivity based electron count and coor– dination[17].

plexes. Therefore, the electronic structure of the surface cations can be deduced by using the same hybridization schemes but counting now two σ electrons for each oxygen or sulfur atom. For example, in Co2O3, Co3+ is coordinated to six oxygen atoms. Within this model description, the six Co d2sp3 orbitals strongly overlap with six σ-type oxygen orbitals, which results in six low-energy bonding and six high-energy antibonding σ-type directed orbitals. The 12 oxygen electrons will occupy the six bonding σ-type hybridized bonding orbitals. The six electrons left will occupy the three nonbonding d-atomic orbitals continue. At the (111) surface of Co2O3, one oxygen–cobalt bond will be broken. Co3+ now creates a σ-type empty d2sp3-orbital that acts as a dangling bond along the surface normal. In Co2+, this dangling bond would be occupied with one electron.

In Figure 5.10, orbital schemes are summarized for coordination complexes predicting the dangling bonds that arise when ligands are removed. A systematic analysis has been developed based on the total electron count and dependent on geometry [17]. There is a so-called isolable correspondence between di erent molecular systems with di erent coordinatively saturated structures having the classical 8, 18 or 16 valence electron count (Fig. 5.10, compounds 2, 6, 14 and 10). For example, column two in Fig. 5.10 shows the development of lone-pair orbitals according to the d2sp3 scheme. In the Cr complex six σ bonding orbitals with ligands occur. Each ligand contributes two electrons. The nonbonding d-atomic orbitals of Cr contain six electrons. On moving to Mn, a position lower in the column concerned, one electron is added to the system that in the MnL6 complex would have to occupy an antibonding σ-type orbital. The complex prefers instead to remove a ligand with formation of MnL5 and a lone–pair orbital occupied with one electron. Replacing Mn by Fe adds another electron. Double occupation of a lone-pair

Catalysis by Oxides and Sulfides 229

As can be seen in Fig. 5.11, the hydration enthalpy increases with decreasing cation size across a row in the periodic table. Hence an OH− group attached to Ni2+ is expected to be more strongly bonded than one to Ca2+. For similar surface topologies, the Ni2+(OH)− group is predicted to be less basic than Ca2+(OH)−. In the (100) surface of the rock-salt structure, the ion charge excesses on the corresponding oxygen atoms are the same for nickel oxide and calcium oxide, = 13 , indicative of a basic group. Ion charge excesses are only sensitive to surface topology, and cannot predict properties that are cation or anion dependent other than when the charge is changed.

In addition to the electrostatic e ect, there is the contribution to the chemical bond

which is the result of the distribution of electrons over the metal d-atomic orbitals, the ligand field e ect[19] . The contribution of this e ect is smallest when only five electrons

are present and distributed over the five d-atomic orbitals. The each electron occupies one atomic orbital (the high-spin state) thus resulting in a spherical electron distribution. When the electron distribution is nonspherical, additional stabilizing interactions occur due to the redistribution of electrons over the d-atomic orbitals. This e ect is a maximum for d-electron counts of three or eight.

Interestingly, the catalytic activity for di erent first-row metal oxides shows two max-

ima across the periodic table for catalytic reactions involving C–C and C–H activation. Dowden and Wells[20] showed that 3d5 (Mn2+, Fe3+), 3d0 (Ti4+) and 3d10 (Zn2+ ) cation-

containing oxides show the lowest activity, whereas 3d3 (Cr3+ ), 3d6 (Co3+), 3d7 (Co2+) and 3d8 (Ni2+) cation containing oxides have the highest activity (Fig. 5.12).

Figure 5.12. The catalytic activity of metal oxides along the first row of the periodic table. Adapted from D.A. Dowden[20].

In this section we have related the reactivity of oxides to cation properties such as charge, radius and ligand field stabilization. In Section 5.1 we have seen that di erences in the reactivity of anionic-oxygen atoms as well as electrostatic e ects are important also. In Section 5.6.8 we will emphasize reactivity di erences in selective oxidation further in relation to the type of surface. We will again see that it is important to distinguish between coordinatively saturated surfaces and surfaces formed by metal–oxygen bond cleavage.

230 Chapter 5

5.4 Medium E ects on Brønsted Acidity

In order to understand he physicochemical basis of Brønsted acidity, it is important to distinguish between homolytic and heterolytic dissociation and the corresponding energies related to each process. p

Ediss (HOMO) = EHX − EH − EX

Ediss (HETERO) = Ediss(HOMO) + I.P.H + E.A.X

The homolytic dissociation energy is the energy cost to separate a molecular bond into neutral atoms. The heterolytic dissociation energy is the energy required to split a molecular bond into oppositely charged ions. This involves the energy necessary to dissociate the molecule homolytically into neutral fragments [Ediss (HOMO)] plus the energy required to remove an electron from the more electropositive fragment (ionization potential I.P.H] and place it in the more electronegative fragment (electron a nity [E.A.X]) as is shown above.

The acidity of a molecule strongly depends on the medium in which it is studied. Most studies are typically carried out in water. The concentration of the hydronium ions (H3O+ ) that form in water determines the solution pH. The following equilibrium is key.

HX(aq) + H2O(e) H3O+ (aq) + X−(aq)

This equilibrium is controlled by the dissociation energy of HX and the stability of the solvated H3O+ and X− intermediates that form.

In the halides, the electron a nity does not vary much, hence the changes in the homolytic dissociation energy of HX determines the changes in the equilibrium constants. The same holds in comparing H2S with H2O. The results, however, are quite di erent if one compares the acidity across a row in the periodic table, such as H2O compared with HF or H2S compared with HCl.

Tables 5.2 and 5.3 present the energies for the homolytic dissociation of various HX species [Ediss(HOMO)]and the electron a nities [E.A.X])] for their corresponding anions (X−), respectively.

The di erence in electron a nity between OH and F or SH and Cl is much larger than the corresponding homolytic bond energies. HF in H2O is more acidic than H2O itself, because F− is more stable than OH−. The same holds for the di erence in acidity of H2S and HCl.

Catalysis by Oxides and Sulfides 231

The stability of the hydronium ion determines the pH of water or that of other acids dissolved in water. This changes when we compare the acidity in di erent solvents. Neat acids such as HF and H2SO4 are much more acidic than when they are dissolved in water. This is because di erent protonated species are present. For HF dissolved in HF there is the equilibrium

2HFsolv [HFH]+ + F−

The di erence from HF dissolved in water is the relative stability of [HFH]+ versus H3O+ . The deprotonation energies for HFH+ and H3O+ were calculated to be 652.7 and 957.2 kJ/mol, respectively. The di erences in relative stability imply a much smaller equilibrium concentration of [HFH]+ than [H3O]+ normalized on the same HF concentration. The much smaller deprotonation energy of HFH+ compared with H3O+ implies a much higher reactivity of H+ attached to HF than attached to H2O.

Such medium e ects imply a dramatic di erence between acid catalysis carried out over acidic surfaces in a gas-phase medium and that carried out in a polar solution phase medium in the absence of a heterogeneous solid acid surface. On a solid surface, the activation of an adsorbate occurs from a neutral state since the carbenium ion states are typically unstable and serve as transition states. On the other hand, protonation in a solution can occur from an ionized state, in which reactant molecules have already been protonated. This explains the much higher reactivity for protonation reactions which occur in polar media.

An interesting comparison for acid dissociation carried out in neat and in aqueous media was presented by Kazansky[21] for the sulfuric acid system.

The dissociation of sulfuric acid is exothermic in water (∆E = −83.6 kJ/mol), whereas in neat sulfuric acid the heat of autodissociation is slightly endothermic (∆E = +18.8 kJ/mol). Notwithstanding their low concentration, the catalytically active species in the solvent will be H3O+ or H3SO4 +. The high reactivity of H3SO4 + arises from the much lower deprotonation energy for H3SO4 + than H3O+ .

The essential di erence between proton transfer from a neutral complex and protonated

state becomes clear in comparing the activation of isobutene by the neutral (H2SO4)2 cluster shown in Fig. 5.13a and H3SO4 + shown in Fig. 5.13b[17]. There is a high activation

barrier for proton transfer in the neutral system (100 kJ/mol for the H2SO4 monomer as shown in Fig. 5.13a) as compared with the low activation barrier (14 kJ/mol as shown in Fig. 5.13b) and high exothermicity for protonation in the charged system.

The interaction energy of a carbenium ion with a neutral sulfuric acid molecule is only 64 kJ/mol, compared with a C–O bond dissociation energy of 760 kJ/mol for the neutral tert–butyl sulfuric acid. The QM results are corrected for solvation energy a ects by using a continuum model (see also Chapter 6. page 290) that describes the dielectric response of the solvent medium on the energy of dissociation for the reaction:

[t-butylH2SO4]solvated + −→ [t-butyl]s + + H2SO4solvated

The reaction now becomes exothermic at –57 kJ/mol and, hence, the protonated isobutene is a stable compound in solution. This result illustrates the importance of solvation on reactions in polar media with high dielectric constants.

232 Chapter 5

Figure 5.13a. Energy profile for isobutene interaction with dimeric sulfuric acid (H2SO4 )2. (a) π-

Complex; (b) transition state. (c) t-C4 H2SO4 .H2SO4 . The data on the reaction with monomeric sulfuric acid are given for comparison[22].

In contrast to the low activation barrier for the protonation of isobutene in sulfuric acid, the computed activation energies for isobutene protonation in a zeolite using comparable models are much higher, namely 70 kJ/mol. In Chapter 4, the formation of protonated isobutene in mordenite is considered in more detail. It is concluded that protonation is an activated process with respect to the adsorbed π-complex and that the protonated isobutene forms an alkoxy complex. This chemistry is very similar to that discussed here for protonation of isobutene with (H2SO4)2 (Fig. 5.13a).

Protonation in zeolites is also extensively discussed in Chapter 4. Zeolitic protons (see page 163) are part of the zeolite (SiO2/Al2O3) microporous system. The zeolite is made up of a rather rigid medium with a low dielectric constant (ε ≈ 2). Chemistry in zeolites is, therefore, closer to that which is carried out in a vacuum rather than in solution. Solvation e ects remain limited to the relatively small adjustments of the lattice atom positions that occur in the direct environment of adsorbed cations or anions. Electrostatic charges are

Catalysis by Oxides and Sulfides 233

Figure 5.13b. Energy profile for isobutene interaction with protonated sulfuric acid (H3 SO+4 ). (a) π- Complex; (b) transition state; (c) Ion–molecular complex[22].

only screened by polarization of nearby lattice oxygen atoms. In a protonation reaction, the interaction between positively charged reactant cation state and the negative charge on the zeolite lattice is a critical in establishing the energy or the transition-state energy. In zeolite catalysis, the protonated intermediate is always a highly activated complex and usually part of the transition state.

For reactions in H2O an analogous discussion applies as for reactions with the protonated sulfuric acid form. The reactive intermediate now, however, is H3O+. H3O+ has a much stronger OH bond than H3SO4 +, and hence its reactivity is much less. The species that is formed by reaction with alkene in an aqueous solution can best be compared with protonated alcohols forms typically known as alkyl oxonium ions[23]. Protonated alkene is hydrated in aqueous media. The formation of alkyloxonium ions instead of the carbenium ions in H2O can be viewed as due to the basicity of water. The major di erence between solid acids and acidic solutions arises because the hydrogen atoms in solid acids are part of strong covalent bonds and are not present as protons that are present in in the solution phase. In a solution there is a equilibrium between non-dissociated acid molecules and the

234 Chapter 5

ionized ions, solvated by the solvent. Their equilibrium defines the pH of the solution. In a solid acid there is no counterpart of the pH. Acidic chemistry evolves only upon contact with a reactant. The protonated intermediates often only exist as part of transition states. Acidity in solid acid catalysis has to be considered a kinetic phenomenon.

5.5 Acidity of Heteropolyacids

Heteropolyacids are unique molecular metal oxide clusters which contain acid, base and redox functionality that can be varied over a fairly broad range in order to tune their potential catalytic behavior. The polytungstate forms have Hammett acidity values of Ho = –13.1 which, ranks them as being more acidic than 100% H2SO4 (Ho = 12). They are, therefore, known as superacids. However, the use of Hammett indicators is based on color changes of the indicator molecule by protonation. In solution their is a equilibrium between protonated indicator molecules and non-dissociated acid molecules. On the solid acid the protonated indicator molecule is stabilized by the negative cgarge of the ionized surface. This is very di erent from solvation phenomena in solutions. Hence interpretation of Hammett indicator measurements on solid acids is not straightforward and immediate comparison with their use in liquid acids is not justified. We discussed an analogous situation in Chapter 4 for acid catalysis of zeolites. There we noted that adsorption e ects obscure intrinsic acidity similarities or di erences. The heteropolyacids are stable both in solution and in the solid state. They can also be readily supported on silica or other metal oxide supports. They have therefore been considered as possible alternatives to replace the highly corrosive and environmentally toxic HF and H2SO4 liquid acids presented in the previous section for low-temperature acid-catalyzed reactions such as isomerization and alkylation[24]. The most predominant form studied in the literature is the Keggin structure. The phosphotungstic form of the Keggin unit is shown in Fig. 5.14. The Keggin structure is comprised of primary, secondary and tertiary features. The primary Keggin unit (KU) is made up of a central metal oxide tetrahedron comprised of a central atom such as P or Si which is directly bound to four oxygen atoms. The charge imbalance between the central atom and its oxygen ligands leads a negatively charged anionic core.

Their central anionic tetrahedron core is surrounded by twelve outer metal oxide octahedra that form an outer metal oxide shell that acts to delocalize the charge on the central cluster. These outer metal–oxide octahedra are comprised of a metal atom that sits in the center, known as the addenda atom, surrounded by six oxygen atoms. The negative charge in this system is then delocalized over the outer metal oxide shell. The resulting cage structure contains three uniquely di erent oxygen atoms. The first are the terminally bound tungstenyl (W=O) oxygen atoms. The other two refer to bridging oxygen atoms from either the corneror edge-sharing octahedra that make up the outer shell. The heteropolyanion form is subsequently converted to the acid form by the addition of protons at external oxygen atoms of the cage. The protons can reside at either the corneror edge-sharing oxygen bridges or the terminal oxygen atoms. The calculated energy di erence indicates that there is typically a small site preference for the bridge sites but that the energy di erence between each of these sites is not very large (< 15 kJ/mol). The charge of the anion can essentially be considered to be delocalized over the entire outer shell of the Keggin structure. This delocalization gives rise to a lower proton a nity, which would imply a greater acidity for these materials.

In the solid state, the primary Keggin structures pack into secondary and tertiary

Catalysis by Oxides and Sulfides 235

forms. In the secondary structure, six water molecules of hydration come together to aid in assembling primary Keggin structures into a hexahydrate BCC form. The waters of hydration can form around an internal proton which exists in the form of the H5O2+ Zundel ion in the center of the hexahydrate cage. These “inner” protons may be considered inaccessible to reagents that cannot penetrate into the secondary structure.

Thermogravimetric analysis indicates that the waters of hydration are removed at temperatures above around 500 K. This can lead to a dehydrated secondary form. Ab initio calculations indicate[25] that the loss of secondary water molecules will begin to isolate protons on the bridging positions between two Keggin units (Fig. 5.16), which for the most part will be inaccessible for reaction.

Figure 5.14. The Keggin structure of the PW12 O40 3−, identifying the three types of oxygen atoms in the structures[25]. The hexahydrate heteropolyacid contains typically six water molecules per Keggin unit. The water molecules essentially help to assemble and pack the primary structure into an organized secondary hierarchy. A computed representation of the secondary H3 PW12O40 .6H2O hexahydrate form is given in Fig. 5.15.

The use of HPAs to catalyze specific reactions heterogeneously requires anchoring these molecular cages to a particular support. Silica currently appears to be the favorite used in the experimental literature. The supported Keggin structure can readily be modeled by examining just the interactions between the reactants and the primary Keggin structure. DFT calculations were carried out to establish the proton a nities for at various sites on the Keggin structure and for di erent compositions for both the addenda and the central atoms. The results for the adsorption of the proton on the primary PW12O40 3− Keggin structure are depicted in Fig. 5.14[26] and summarized in Table 5.4[26].

The results demonstrate that the proton attached to the Keggin unit with a charge of 3− is, as should be expected, the strongest interaction resulting in a proton a nity of 1591 kJ/mol. The presence of additional protons on the metal oxide cage ultimately compensates for the negative charge and lowers the proton a nity. The proton a nity of the third proton is lowest at 1077 kJ/mol. This is significantly lower than those for zeolites with low aluminum concentration. The a nity of the HPA for the second proton (1349 kJ/mol) is actually closer to the a nity found for zeolitic protons. The strength of these bonds is thus indicative of strong Brønsted acidity for the HPAs.

From Table 5.4, one can see that the whereas the proton prefers the bridging oxygen atoms, the di erence between the bridge and atop sites is fairly small, which would suggest that the proton may be quite mobile, having the ability to di use between di erent sites.