J. Phys. Chem. B 2002, 106, 6163-6177
6163
Local States in Microporous Silica and Aluminum Silicate Materials. 1. Modeling Structure, Formation, and Transformation of Common Hydrogen Containing Defects Alexey A. Sokol* and C. Richard A. Catlow DaVy and Faraday Research Laboratory, The Royal Institution of Great Britain, 21 Albemarle Street, London W1S 4BS, U.K.
Juan M. Garce´ s and Alex Kuperman† The Dow Chemical Company, 1776 Building, Midland, Michigan 48674 ReceiVed: August 29, 2001; In Final Form: January 15, 2002
We develop a theoretical approach to point defects in zeolites as catalytically active framework sites, advance structural models for the most important defects and, by calculating the energy minimum configurations, study the main stages of the defect transformation in synthesis, postsynthetic treatment, and aging processes. Calculations have been focused on common hydrogen containing defects in zeolites including bridging hydroxyl and complexed silanol groups. The Brønsted acid site [AlH]Si, the vicinal disilanols and hydroxyl nests (or hydrogarnet defect) have been investigated using a periodic density functional theory approach employing numeric atomic orbital basis sets, as implemented in the Dmol3 code. We propose a new mechanism for the modification of zeolite frameworks based on the silanol inversion. We identified a novel, low energy path (with reaction energy ca. 21 kcal/mol) to an elusive trigonal aluminum species, which does not involve dehydroxylation processes. The inversion in the vicinal disilanol structure reported previously has been confirmed to lead to a lower energy configuration with two loosely bound single silanol groups, which we suggest cannot be associated with 1H NMR shifts of hydrogen bonded silanols. The inversion of the hydroxyl nest is shown to preserve a close trisilanol ring structure with extensive hydrogen bonding while opening this defect to an interaction with guest molecules in the pores of zeolites. On the basis of a low energy of ca. 0.5 kcal/mol calculated for the inversion of the hydroxyl nest, we propose that in zeolite materials T-site (silicon or aluminum) vacancies are realized in two major conformations: a traditional nest structure and a pair consisting of a trisilanol ring and a single silanol group.
1. Introduction Most important physical and chemical properties of materials are either directly determined or significantly influenced by local states in the form of structural and electronic defects. Microporous aluminum silicates, or zeolites, are no exception. These materials are utilized in a wide range of applications ranging from more traditional industrial areas of heterogeneous catalysis,1 ion exchange2 and solar cells3 to new frontiers in electronics,4 quantum wires5 and dots.6 Moreover, the experimental techniques used in the material characterization not only register the existing local structures but also modify old and introduce new defects. Hence, it is vital to understand the full extent of the defect physics and chemistry in these technologically important materials. Even a simple analysis shows that a number of defect species with similar properties can form in these materials. In this respect, characterization of the defects in zeolites becomes similar to the characterization of the crystal structure of zeolitic frameworks. One needs a guide that can be used in identification of particular species by their signature obtained from different experimental techniques. Modern solid-state theory can provide here a necessary tool and link to tie together the known facts and test new and old hypotheses.7 * Corresponding author. E-mail:
[email protected]. † Current address: Chevron Research and Technology Co., 100 Chevron Way, Richmond, CA 94802-0627.
The present is the first in a series of papers in which we aim to develop a consistent theoretical view on point defects in zeolites as catalytically active framework sites, advance structural models for most important defects and, by calculating the energy minimum configurations, study the main stages of the defect transformation in synthesis, postsynthetic treatment and aging processes. The experimental data available on zeolites are vast: from the fundamental contribution of Barrer to more recent work employing solid-state NMR,8,9 IR,10 molecular probe11 techniques applied to the zeolites and related materials. Moreover, despite the complexity of the problem we may now apply reliable theoretical methods to study zeolites as summarized in ref 12. Since the appearance of this monograph, chemically accurate ab initio electronic structure calculations have become more accessible in the field; further progress has been achieved using quantum mechanical models in the frame of hybrid QM/ MM embedding and periodic boundary conditions (see, e.g., refs 13-36). The latter approach is adopted in this work. 2. Defect Models For clarity, in the following discussion, we start by recalling a number of basic facts about zeolites.2,37-39 By definition, zeolites are microporous crystalline tectosilicates. The primary building unit of silicates is a tetrahedron with a silicon atom in the center (T-site) and four oxygen atoms at its apexes; in
10.1021/jp0133384 CCC: $22.00 © 2002 American Chemical Society Published on Web 05/25/2002
6164 J. Phys. Chem. B, Vol. 106, No. 24, 2002 tectosilicates, each oxygen atom is shared by two adjacent tetrahedra, so that the zeolite anion sublattice is built up of twocoordinated oxygen atoms. Some of the tetrahedrally coordinated silicon cations can be isomorphically replaced by a number of elements. Interlinked tetrahedra constitute the porous framework of zeolites dissected by channels, containing cavities or cages and pockets, thus forming unique open structures. This fact gives rise to the concept of an internal surface of these materials, since a large fraction of atoms in the zeolite bulk is accessible to the guest molecules. In catalytic applications, the prevalent type of substitution into the T-sites is by aluminum. The difference in valence between Si4+ and Al3+ (a negative charge on the framework within an ionic description) necessitates charge compensation that is accomplished by counterions, the extraframework species that may be (i) protons tightly attached to the framework oxygen thus forming hydroxyl groups in H-forms of zeolites; and (ii) more loosely tied to the framework metal (e.g., Na+, K+, Mg2+, Ca2+, Ni2+, La3+, Ru3+) and organic (such as tetrapropylamonium, TPA+) cations. Sometimes, extraframework anion species (F- and Cl- most commonly, but also sulfates and others) are also present and appreciably influence zeolite properties. Due to the hydrous natural environment, the synthetic conditions, or the hydrothermal treatment, zeolites also contain significant amounts of water (physi- and chemisorbed) unless special efforts (calcination) have been applied to remove it. All the extraframework species introduce an additional element of complexity, rendering zeolites compounds of variable composition and introducing disorder (although these are crystalline materials). Additionally, the presence of extraframework species increases the coordination of the oxygen atoms. If not prevented, atmospheric oxygen is invariably present throughout the pores both in molecular and atomic forms. Not all deviations from a rigid crystalline pattern will be considered as defects in the present work, only those that break the short-range order around the framework constituent ions, i.e., the framework point defects. Disorder, linear, two-dimensional and macro-defects are outside the scope of this study. The concentration and distribution of aluminum in zeolites differ from one zeolite to another, but at this stage many zeolites can be obtained as highly siliceous materials. This encourages us to use the all-silica zeolite as a reference system and treat the framework aluminum as a defect, even more so since in most zeolites the chemically active sites in the framework are based on the aluminum (or similar) substitution. 2.1. General Elemental Defects. The three following kinds of point defects are common in all solids: vacancies, interstitials and substitutions. Any complex atomic defect occurring in real materials could be described in terms of these three defects or combination of them. An additional level of complexity is introduced by the electronic degrees of freedom, which lead to such electronic defects as trapped polarons and excitons. Let us consider first the defect species that attract most attention in related silica materials (including dense minerals, glasses, metal/insulator/semiconductor interfaces etc.). We start our analysis by considering elemental defects in the Si and O framework sublattices. A trivial chemical exercise of removing and adding atoms (or ions) and electrons yields an exhaustive list of defect structures. At this stage we will not consider closely how the ensuing species can be embedded into the lattice, but rather try to highlight every important possible defect configuration that could be encountered in various areas of silica science as well as hypothetical species. Tables 1 and 2 sum-
Sokol et al. marize this approach. A distinction is made here for structural defects: between (i) intrinsic defects formed by framework Si and O atoms, (ii) isomorphous substitutions inserted into the framework sites (the major source of catalytic activity of these materials), and (iii) defects of “coordination” that provide the necessary conditions for modifying the zeolite framework. The electronic defects are treated as a separate class of local states, i.e., electrons and holes, either self-trapped or bound to isomorphous substitutions in the framework. The porosity of zeolites provides very favorable conditions for the introduction of defects; resulting defect concentrations are unusually high even compared to those in glasses.40 It was reported, for instance, that up to 30% of all Si ions in a zeolite mordenite41 can be involved in different defects. Large pores provide a large interstitial space for drift and diffusion of ions displaced from their equilibrium sites in the lattice, giving rise to a Frenkel defect pair as well as for isomorphous substitution of Si. In the case of an isovalent substitution (for example, a Ge ion for a Si ion), a simple framework defect is generated. Otherwise, as in the case of Al substitution, a more complex structure will appear, which should also include either a charge compensating extraframework species or the further alteration of the framework around the defect. The semi-ionic character of silica leaves open the question of the charge of the vacancy. Oxygen cannot exist as a separate species at the interstitial positions in the O2- state (the latter is stabilized at the lattice sites by the Madelung field). So it can be removed from the site either in an atomic form, leaving a neutral vacancy, or as O-, which generates a paramagnetic center in the framework (see below), or as O2- as a constituent part of some complex species. Two doubly charged oxygen vacancies can be counterbalanced by a silicon vacancy, forming a Schottky defect trio. Direct formation of vacancies is not energetically favored in zeolites as compared to the chemically related hydrogen containing defects considered next. However, they do appear as a result of chemical transformation of active sites. For example, zeolite calcination causes the healing of hydroxyl nest defects, which was proposed to lead to a peculiar structure of two “nonintact” siloxane bridges in place of a Si vacancy,42 which is a condensed Schottky trio. In fact, the concentration of Si vacancies, charged or neutral, should be significantly lower than the concentration of O vacancies, which can be easily understood: on removing oxygen from a lattice site, we break two SisO bonds, whereas Si vacancy formation is necessarily accompanied by the breakage of four SisO bonds. Elemental substitution of oxygen by other elements is also not a very frequent occurrence, and when this is the case, the species in question is usually a fluoride or a chloride ion; the negative charge compensation required also makes such defects energetically unfavorable. The presence of O vacancies in silicas is primarily deduced from the experimental observation of the E′ center, which is an electron trapped at a Si site. This electron occupies a dangling sp3 orbital (tSi•) and is clearly identified by electron spin resonance (ESR) techniques. To account for an unpaired electron at a Si site, we should consider the cleavage of one of the SisO bonds and further stabilization of the dangling O atom. Similar but still different ESR signals have been reported for this center, which have been rationalized as an effect of the environment; hence, a number of related species have been postulated in the experimental literature. For example, three models have been proposed for the E′ centers in glasses:40 (i) the E′R center is an E′ center coordinated to a peroxy species, formed by the interstitial O atom and lattice oxygen (Frenkel pair for-
Microporous Silica and Aluminum Silicate Materials
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6165
TABLE 1: Elemental Structural Defects in Silica
mation); (ii) the E′β center is associated with a proton that results from reaction of trigonal Si with atomic hydrogen; (iii) finally, a single-fold ionization of a neutral O vacancy gives rise to the E′γ center. The trigonal Si plays an essential part in these models, either as a precursor or as a counterpart species formed across the O vacancy. The neutral oxygen vacancy seems to be another firmly established defect in silica materials. In particular, two wellknown ultraviolet/optical absorption bands at 5.0 and 7.6 eV have been attributed to singlet-triplet and singlet-singlet excitations of this defect.43 The oxygen expelled from the lattice site is thought to give rise to a peroxy linkage (tSisOs OsSit). The latter defect may also result from insertion of the molecular oxygen into an oxygen vacancy. Such a process may be widespread in zeolites, as hinted by 18O adsorption studies, which exhibit single-step double exchange with the framework oxygen.44 The actual defect configuration is not known, but the defect should be a precursor of a peroxy radical (tSisOsO•‚‚‚Sit) that is observed in ESR as a species accompanying the E′R center.45 The defect has also been considered
to be responsible for the so-called blue luminescence of irradiated silica.46 Some fused silicas (e.g., Type III40), contain large amounts of hydrogen that can be released from their lattice upon irradiation, implying the presence of silanol groups (tSisOH). Irradiation leads to the presence of atomic hydrogen coupled with the nonbridging oxygen, or siloxy radical defects (tSisO•). The hydrogen, both atomic and molecular, reacts with oxygen vacancies, and a hydride defect related to the E′β center (tSi•‚‚‚HsSit) may appear. This defect is not likely to arise in zeolites under usual, hydrothermal conditions, but the possibility should be considered when a zeolite is used for hydrogen storage and is oxygen deficient. Finally, the vicinal silanol defect (tSisOH‚‚‚HOsSit) was envisaged as a result of reaction between a peroxy linkage defect and molecular hydrogen. Such defects should play an important role in zeolites, but the mechanism of their formation would be quite different arising probably from hydrolysis. The main question about the defect is its structure and bonding character: steric restraints make the hydroxyl groups interact strongly,
6166 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Sokol et al.
TABLE 2: Elemental Electronic Defects in Silica
and the structure of this defect is one of the topics of the present study. 2.2. Catalytically Active Centers. Using zeolites as catalysts has stimulated numerous studies of active centers in these materials. In the contemporary zeolite literature, it is usual to distinguish between two groups of acid sites, one reacting with bases by means of proton donation, the Brønsted acid sites, and another reacting with bases without involvement of protons, the Lewis acid sites. The negative charge induced in a framework by an aluminum substituted for silicon is compensated by protons in H-forms of zeolites, generating the celebrated model of a Brønsted acid site, the most important and well character-
ized hydrogen containing defect in zeolites. Its acid strength is characterized by the tendency to donate a proton. In the theoretical work of Sauer et al.47,48 the relation between the acid strength and the heat of deprotonation has been investigated for zeolites using molecular models, an approach that is now widely accepted. However, we note that both acid strength and deprotonation energy require further theoretical elucidation for solids. It is more difficult to give a generally accepted example of a framework Lewis acid site. The only clear experimental evidence of the latter is provided by electron hole centers in irradiated samples: Vn (or Xn) centers (tTsO˙ sT′t), the nature of the T and T′ sites determining the hyperfine splitting in their ESR spectra.49-52 In the absence of structural defects, the
framework oxygen atoms are considered as Lewis basic sites and the extraframework cations as Lewis acid sites. The trigonal silicon and aluminum have been proposed as plausible centers of Lewis activity in zeolites.49,53 The generation of this kind of defect should be facilitated by thermochemical treatments (dehydroxylation of zeolites upon ultrastabilization) when the framework oxygen atoms sited between aluminum and silicon atoms are either expelled (leaving a vacancy) or drastically displaced from the equilibrium position to form an alternative bond. The presence in any appreciable concentration and possible role of the trigonal framework cations still remains a contentious topic. A number of models have been proposed in the literature for the Brønsted and Lewis sites that involve both intra- and extraframework ions. For instance, according to Brunner et al.54 a nonframework tetrahedral aluminum arises from dealumination and forms a 2-ring with the framework silicon. The defect can either donate a proton or accept an electron, thus playing the role of a Brønsted and a Lewis acid site simultaneously. Based on molecular modeling, a number of generic models for Lewis acid sites that involve aluminum have been advanced and recently summarized in ref 55. These models include reconstruction of a -SisAlsSisAl- 4-ring of “flexible” frameworks into an aluminum 2-ring. The structure may also appear in the extraframework aluminum debris that result from dealumination treatment and upon interaction of extra- and intrantraframework aluminum.67 Similar 2-ring siliceous structures have been investigated using both semiempirical and ab initio models in the study of sol gels and silica synthesis.56
Microporous Silica and Aluminum Silicate Materials 2.3. Defects in Zeolites Introduced during Synthesis and Preparation. The history of a zeolite includes the following processes: synthesis, preparation, usage, aging, reactivation with subsequent recycling, until the zeolite has lost irrevocably its useful properties. All stages are characterized by the defects present in it, as in any other material. Each stage comprises several steps, some of which are used repeatedly. Three points are crucial: the Si/Al ratio in the material, the water content and irradiation, since these determine quite different kinds of defects. Synthesis of zeolites under hydrothermal conditions is discussed in detail by Barrer.2 The process of crystallization has been envisaged from gels and solutions to crystal growth. However, direct experimental evidence only started to appear in the 1980s. Zeolite synthesis proceeds, as a rule, from uniform mixtures of the oxides and hydroxides of silicon, aluminum, and other cationic species in the presence of water, acids (HF, HCl, etc.), bases (OH-), and templates. The number of Brønsted acid sites introduced into a zeolite during synthesis depends on the number of aluminum species that are not accompanied by counterbalancing cations (other than protons) incorporated in the framework after aluminum species. In H-forms of as-synthesized zeolites, the Brønsted acid site is the predominant defect. The transformation of a Brønsted acid site into a nonbridging hydroxyl group only associated with a silicon site and a trigonal aluminum has been discussed many times,2 but solid evidence for such a configuration was obtained only for boralites with trigonal boron instead of aluminum. An entirely different situation appears for aluminum deficient materials such as siliceous zeolites. For example, Groenen et al.57 have argued that the crystallization process in siliceous solutions is maintained through double-ring-silicate and monosilicate condensation. The distribution of different silicate units is controlled by the synthesis conditions. In particular, zeolite ZSM-5 crystallizes from a gel with a greater contribution of five-double-ring silicates, although it is necessary to have a certain number of monosilicates in the initial solution; otherwise the resulting framework will include the T-site vacancies. Another kind of defect can be observed when the condensation of silicate molecules is incomplete: two nonbridging oxygens or a hydroxyl group and a nonbridging oxygen or two hydroxyls take the place of the one bridging oxygen.58 Moreover, both kinds of defects are also formed upon post synthesis treatments. The T-site vacancies must appear in zeolites as a result of dealumination by strong acids, steaming and thermal treatment. A pair of hydroxyl groups appears in proximity to one another due to a TsOsT bridge hydrolysis. Therefore, both defects are intrinsic to both as-synthesized and as-prepared zeolites. The first complex defect based on the T-site vacancy has been proposed as the hydroxyl nest by Barrer59 and later by Kerr,60 a defect in which dangling oxygens around the T-site vacancy are saturated by protons:
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6167 After many years of study and argument this defect has been acknowledged and characterized to some extent. But the story of its transformation upon hydrothermal treatment and other influences is still not clear. It is established that annealing, calcination and steaming at elevated temperatures result in zeolite structural defects “healing” when the characteristic signal of hydroxyl nests disappears from the infrared and proton magnetic resonance spectra.61 There has been speculation on the mechanism of healing, but there is no direct experimental proof. The models comprise dehydration, dehydroxylation and deprotonation. From the early 1970s many suggestions appeared that the dehydration results in the formation of two so-called “nonintact” SisOsSi linkages:
which is initially plausible on simple chemical grounds, but to which there are two firm objections: (i) It is known that the annealing of hydroxyl nests is a reversible process, whereas regular SisOsSi linkages are very stable toward hydrolysis. Therefore, if these species are a counterpart of the hydroxyl nest defect, they should be in some way defective, and therefore seen in experiment. (ii) T-sites from the second coordination shell of any particular T-site are separated from one another by about 4.5-5.5 Å. A “nonintact” linkage of a couple of silicon ions at such positions should cause a large deformation of all adjacent bonds, which would be accompanied by a large energy penalty. Thus, in our view, the formation of the nonintact linkage upon the dehydration of hydroxyl nests is doubtful, and we need alternative models for this process. Dehydroxylation of one T-site leaves behind an E′ center or a diamagnetic trigonal silicon, depending on the charge of the hydroxyl group leaving the nest. Interaction of the species with one of the adjacent silanols could result either in a nonintact SisOsSi linkage with a loose atomic hydrogen in its vicinity, or a protonated SisOsSi bridge. So that for each pair of silanols at the hydroxyl nest we have
tSisOH‚‚‚HOsSit f tSi•‚‚‚HOsSit + • OH f tSisOsSit + •H + •OH or
tSisOH‚‚‚HOsSit f tSi+‚‚‚HOsSit + OH- f [tSisOHsSit]+ + OHIf the dehydroxylation is very severe, a pair of hydroxyl groups can be lost simultaneously, and formation of a neutral oxygen vacancy is a possibility:
tSisOH‚‚‚HOsSit f tSi•‚‚‚•Sit + 2•OH f tSisSit + 2•OH with the formation of two E′ centers and a subsequent
6168 J. Phys. Chem. B, Vol. 106, No. 24, 2002 condensation yielding a long SisSi bond. Finally, deprotonation should facilitate peroxy bridge formation:
tSisOH‚‚‚HOsSit f tSisO-‚‚‚-OsSit + 2H+ f tSisOsOsSit + 2e- + 2H+ Two electrons must be removed from the site at the second step of the reaction, which can be achieved either by the interaction with a Lewis base or by an excitation of the nonbridging oxygens. A localized hole and a delocalized electron trapped by the framework could follow from the latter mechanism. The reaction apparently occurs in the presence of large amounts of water in the material by means of the formation of a hydroxonium ion. An alternative path to the formation of a peroxy bridge can be envisaged as a dehydrogenation process in which the released atomic hydrogen forms a hydrogen molecule. Two types of nearby hydroxyl group pairs, or double hydroxyl defects, are expected to be present in siliceous materials: the first is realized in the form of geminal, tSi(OH)2, and the second as Vicinal, tSisOH‚‚‚HOsSit, silanol groups. The geminal silanols can exist in zeolites as structural defects resulting from synthesis under the hydrothermal conditions or from a severe dealumination treatment. The vicinal silanols can result from decomposition of organic templates (TPA in zeolite ZSM-5 provides a good example) upon calcination and/or acid leaching:54
Sokol et al. a dramatic increase in catalytic activity. Debra et al. 63 have shown that the T3 and T4 sites in mordenite are preferential for Al substitution. Taking into account that these sites form the walls separating the big 12-ring and the small, narrow 8-ring channels, it is clear that dealumination treatment should destroy such walls while forming the T-site vacancies.64 This and other models (with the destruction of the 6-ring) should include the condensation of two T-site vacancies into a wide gate between the two channels. The size of the resulting pore would be about 10 Å in diameter, which is still too small to explain the adsorption data. A peak in the pore diameter distribution (resulting from a mild cyclic dealumination) was reported to be in the range of 30-50 Å,65 which implies that at least three neighboring interchannel walls are modified by the dealumination process with subsequent condensation of the T-site vacancies. Utilizing zeolites in ion exchange with radioactive elements should lead to the occurrence of specific radiation defects in these materials. The kind of damage to zeolites brought about by irradiation, as in all materials, depends on the irradiation characteristics such as its nature, energy and intensity. Γ-irradiation does not produce appreciable concentrations of Frenkel pairs but rather electronic defects of the Vn type. Upon irradiation the electron is ejected from an oxygen atom potentially forming a self-trapped hole. This electron, in turn, is trapped either by the zeolite framework (at sites whose nature is unclear) or by protons, if present as bridging OH groups, giving rise to the formation of atomic hydrogen. Depending on the local environment of the excited oxygen, defined by the zeolite composition, further defects can be formed.66,67 3. Defect Transformation Schemes
At the initial stage of this process, two nonbridging oxygens (siloxy species) are associated with Na+ and TPA+. Upon thermal decomposition and ion exchange, the dehydration of the vicinal disilanols results in a SisOsSi linkage. If the initial defect (on the left-hand side of the reaction) is a part of a bigger defect complex, for example. A T-site vacancy, a “nonintact” SisOsSi bond (bridge) is formed, as in the case of the hydroxyl nest. Alternatively, an initial defect can be localized at a regular SisOsSi bridge, which is restored upon postsynthetic treatments. Thus the annealing of zeolites heals structural defects including hydroxyl nests and disilanol species. However, in full analogy with the hydroxyl nest defect, annealing can proceed via different routes. Besides dehydration, of particular importance is the double dehydroxylation of the vicinal disilanols upon which an oxygen vacancy is formed, and thus all the previously considered defects can be created. Nonbridging defects (present as tSisO- and tSisF‚‚‚ HOsSit) and T-site vacancies complement each other in siliceous zeolites.62 The concentration of nonbridging defects was found to increase with an increase in the synthesis pH or the concentration of F- or alkali cation species. The higher concentration of template (Pr4N+ in the latter case) and crystallization at lower temperatures favored the formation of the T-site vacancy. By calcining zeolites at 500 °C, the authors could decrease the concentration of nonbridging defects, but the T-site vacancy concentration remained unchanged. Structural defects can also contribute to the formation and stabilization of mesopores and other large-scale defects in zeolites, mordenite being a good example. It is known that the secondary mesopore structure in mordenites is responsible for
We conclude that the hydrogen containing defects including hydroxyl groups are most common and chemically important, and we will focus in this work on three particular species: (i) [Al,H]Si, the Brønsted acid site, (ii) [4H]Si, the hydroxyl nest defect, or a T-site vacancy, and (iii) the vicinal and geminal disilanols. The presence of all three defects in zeolites is established by experiment, but reliable atomistic models have been developed only for Brønsted acid sites and hydroxyl nest defects. The vicinal silanol defect is known only by its chemical composition (tSiOH HOSit) while its concentration is estimated on the basis of the Q3 (branching units) signal in the solid-state 29Si NMR spectra and the 1H NMR signal in the range of hydrogen bonded silanol groups.68,69 These defects relate to many local properties of zeolites including the Brønsted and Lewis acidity. However, a rationalization of the latter phenomenon has posed a serious problem in zeolite chemistry. Whereas the hydroxyl groups, in all three defects considered above, are to a larger or smaller extent responsible for Brønsted acidity of the zeolites, none of them seems to be a good electron acceptor capable of maintaining the Lewis acidity. Furthermore, the modification of zeolites widely used in industry and scientific research includes acid leaching, steaming, washing and calcination in oxidizing atmosphere. All the processes involved lead to the transformation of the defects considered, as has been confirmed many times by experiment, in which the characteristic signal of these defects either drastically changes or disappears altogether. Hence, we next examine the atomistic models of the three main defects, while concentrating in each particular case on the schemes for defect transformations. All paths for the defect transformation have been derived from our analysis of the experimental data, comparison with theoretical studies performed by other researchers (mostly in areas of related
Microporous Silica and Aluminum Silicate Materials
Figure 1. Transformation of Brønsted acid center upon post synthetic treatments.
materials such as silicate minerals and glasses), and our preliminary semiclassical and semiempirical calculations (see in particular our INDO results on the Brønsted acid site70). The transformation of both the Brønsted acid sites and the silanol based defects upon different treatments leads to a number of interesting charge neutral defects and related radical species. These include “nonintact” oxygen bridges, peroxy bridges, nonbridging siloxy groups and oxygen vacancies. Trigonal Al and Si sites as well as E′ centers and 5- and 6-coordinated framework cationic sites may not necessarily contain hydrogen as such but can result from the transformation of hydrogen containing defects or are their precursors; thus these defects will also be of interest in this study. Moreover, the Lewis acidity of zeolites can be explained by the presence and relative concentrations of such defects. 3.1. Brønsted Acid Sites. Paradoxically, the geometry of this defect as estimated from the XRD, NMR and EXAFS experiments is not very precise, which is a clear consequence of the high proton mobility and effects of disorder. The averaging over the AlsO and SisO bonds at the site leads to a prediction of corresponding distances in the AlsOH as well as SisOH groups that are too short. The only reliable geometrical parameters, which can presently be obtained from experiment are the AlsH and SisH distances. For example, Fenzke et al.71 have reported the AlsH distance: 2.37 ( 0.04 Å in the small cavities and 2.48 ( 0.04 Å in the big cavities of the zeolite HY. The hydroxyl stretching frequencies as obtained in the IR experiment are measured with good accuracy, although there remains the question of the temperature dependence and the line broadening in the spectrum due to the anharmonic effects and the electronphonon interactions. Postsynthetic hydrothermal treatment of H zeolites can lead to two competing processes, as shown in Figure 1. Both processes result in the generation of Lewis active centers. However, these centers have different locations and exhibit a different coordination of aluminum. 3.2. Silanol Groups. The terminal silanol is a typical surface feature of zeolites (like many other oxides) and thus remains beyond the scope of the present work, which concentrates on bulk defect species. We will concern ourselves with all other defects. Like the Brønsted acid sites, silanols are characterized in experiment by their vibrational properties and very little can be said about their geometry. We should note the work of Hunger et al.,72 who reported the distance between two protons at the adjacent silanol groups (thought to be of the vicinal type) to be 3.1 Å. Moreover, Koller et al.73 have found the inter-
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6169 oxygen distance in the hydrogen bonded silanol groups to be 2.7 Å; a peculiarity of the scheme proposed in their work was that, in fact, two hydrogen bonds were associated with an isolated siloxy species (SiO-) resulting from the framework hydroxylation next to lattice oxygen (with another bond across the six-membered ring channel). In the present work it will be argued that the hydrolysis of the framework leads to the formation of dioxygen species and that the distances reported by Koller et al. are characteristic of the hydroxyl nest defects. Hydroxyl Nest Defect. As can be seen from experiment, the calcination and other kinds of thermal treatment applied after zeolite dealumination lead to reversible changes in IR and NMR spectra. The signal presumably corresponding to the hydroxyl nests disappears and new features can be seen, which is evidence for the formation of new defects. In the following scheme, given in Figure 2, the emphasis is placed on three processes: dehydration, deprotonation and dehydroxylation. These processes are probably concomitant and could occur at the same defect site simultaneously. Vicinal Silanols. These defects can have the same mechanisms of transformation as the two previous species. However, we have specifically highlighted here electronic processes during defect formation as they lead to radical and generally speaking oxidizing chemical centers, which are of particular interest for us. Postsynthetic treatments involved in a thorough dehydration of zeolites require high temperatures, which could facilitate formation and stabilization of charged defects in radical processes on zeolites as outlined in the scheme. The main thrust of the remainder of this paper and following publications will be the evaluation of the energetics of the processes summarized in Figures 1-3 using computational electronic structure techniques. 4. Computational Scheme To investigate the structures and energetics of these (and other) defects, we have undertaken density functional theory (DFT) electronic structure Γ-point (k ) 0) calculations employing periodic boundary conditions as implemented in the DSOLID/Dmol3 code.74 The exchange and correlation functionals used include the local density (LD) and generalized gradient (GG) terms of Perdew and Wang 1991.75 Use of numerical atomic basis functions allows one to obtain an accurate and realistic representation of the charge density and the binding energies, which are close to the DFT limit using the current functionals.74 In particular, for geometry optimization, we used (i) the double numerical basis set that comprises two functions for each occupied valence orbital in a free atom, with polarization functions, (ii) the medium quality integration grid (a default option that provides a convergence of about 0.01-0.02 Å in the bond distances and usually better than 1 kcal/mol in the binding energies). The energies in the minimum configurations have then been calculated using a triple numerical basis set complemented by two polarization functions per atom and the “fine”52 mesh; the cutoff radius for basis functions was increased from 5.5 to 7.0 Å. In the present case, our calculations were based on a pure silica sodalite structure, which has a cubic lattice with 36 atoms in its unit cell. Defects are introduced into the unit cell of the all silica sodalite system and repeated periodically. We note that the interactions between the periodic images are small as the overlap between the defect states studied is negligible while the strongest dipole-dipole electrostatic interactions sum to zero due to the cubic symmetry of the lattice. However, the lattice deformation due to the defect is modeled only within the unit
6170 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Sokol et al.
Figure 2. Transformation of the hydroxyl nest defect upon postsynthetic treatments.
Figure 3. Transformation of vicinal silanol defect upon postsynthetic treatments.
cell, a deficiency, which can be improved only in larger supercell calculations planned for our future studies. The first stage of the calculations reported here was performed using (i) the frozen core approximation and (ii) the charge density obtained within the local density approximation with the generalized gradient correction to the energy and forces, applied a posteriori. The crystal structure of the siliceous sodalite was fully optimized, and the resulting lattice parameter a (8.878 Å, compared with 8.830 Å from experiment76 on the single crystals templated with ethylene glycol) was kept fixed in all
subsequent defect calculations, in which we undertook full geometry optimization with respect to all atomic coordinates. When both approximations were removed at the final stage of this study, the SisO bond distance in perfect sodalite increased from 1.62 to 1.64 Å (cf. experimental value of 1.60 Å), which determined the change in the binding energy of the material. The changes in other structural parameters had only a negligible effect. The energies of defect formation calculated here can be used as an indication of the feasibility of creating or transforming
Microporous Silica and Aluminum Silicate Materials
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6171
Figure 4. Brønsted acid site: structural model obtained from DFT calculations.
TABLE 3: Brønsted Acid Site, Bond Distances (Å) and Angles (deg) Obtained from DFT Calculations H1-O1 H1-Al1 H1-Si1
0.978 2.45 2.30
Si1-O1-Al1 H1-O1-Si1 H1-O1-Al1
134.0 114.4 111.4
Al1-O5 Al1-O4 Al1-O6 Al1-O1 Al1-Si1 Al1-Si3 Si1-Si2 Si2-Si3
1.730 1.720 1.749 1.895 3.35 3.18 3.19 3.07
Si1-O8 Si1-O2 Si1-O7 Si1-O1 Si2-O2 Si2-O3 Si3-O4 Si3-O3
1.617 1.599 1.616 1.720 1.645 1.625 1.602 1.609
one defect into another under equilibrium conditions. The dynamics, statistics and kinetics of the defect formation is based on the free energy including vibrational corrections to the internal energies and entropy factors, free energies along the reaction/diffusion paths (or at least in the transition states) and knowledge of the defect interaction energies. These calculations are indeed the first step in the investigation of defects in zeolites. Additional characterization of the defect properties (i.e., by calculation of the optical IR, ESR and NMR spectra) will be reported in our following publications. Nevertheless, the energies calculated in the present paper are expected to be the major factor controlling the feasibility of the defect reactions. 5. Results and Discussion 5.1. Common Hydrogen Containing Defects of Condensation. 5.1.1. Brønsted Acid Site. The atomic structure of the Brønsted acid site obtained in our DFT calculations is presented in Figure 4. All structural parameters, summarized in Table 3, compare well with calculations reported in the literature.12,13,20,70,77 In particular, the hydroxyl bond length is found to be 0.98 Å in our DFT calculations corresponding closely to the value of 0.97 Å in the PW91 DFT plane wave calculations of Shah et al.,35 which is probably closer to that corresponding to the saturated basis set with this density functional. We also calculate the SisO and AlsO bond lengths for a bridging hydroxyl in close agreement with Shah’s calculations (1.72 and 1.90 Å, respectively). However, the comparison of molecular cluster results obtained using higher correlated post Hartree-Fock methods with corresponding DFT results indicates a trend in overestimation of the semi-covalent/semi-ionic bond lengths by DFT. As mentioned before, the only structural characteristic of the Brønsted acid site directly available from experiment is the AlsH separation. In particular for the HY zeolite Fenzke et al.49 have reported 2.37 and 2.48 Å for sites in the small and large cavities, respectively. The DFT value of 2.45 Å agrees with experiment rather well if we assume that this feature is not very sensitive to the zeolite framework topology. Indeed, the framework of zeolite Y is built of sodalite cages, separated
by interfaces, whereas in sodalite itself these cages are closely packed, so that the local environment of the Brønsted site is similar in both cases. Regarding the spectroscopic characteristics of the defects, our results for one-electron spectra are only indicative as far as the optical properties are concerned: the DFT reproduces only the ground state of the defect. We note only that the hydrogen associated electronic states readily hybridize with the Si and Al contributions in the bottom part of the upper O 2p valence band and provide only a minor contribution at the top of the valence band. The important point here is the lowering of the bottom of the conduction band where there is a noticeable contribution of states localized broadly around the defect site. Bearing in mind the experimental assignment of the optical absorption band at 6.6 eV78 to hydroxyl groups, we should expect an appearance of the defect bands around this value due to all the neutral defects that contain spin-unpolarized hydroxyl groups considered below. The delocalized character of the states associated with these hydroxyl groups along with the high concentration of such defects in zeolites suggests that it is plausible that the corresponding defect states overlap with the formation of broad bands. 5.1.2. Hydroxyl Nest (Hydrogarnet) Defect. The hydroxyl nest, or hydrogarnet, or [4H]Si substitutional defects in zeolites have been intensively studied using semiclassical and semiempirical techniques. We also note a recent combined experimental and computational work on hydroxyl nests by Bordiga et al.79,80 in which hybrid QM/MM techniques have been employed to investigate hydrogen bonding effects on the vibrational properties in this defect. Periodic ab initio models of the hydroxyl nest defect have been derived for R-quartz and a number of silicate minerals. While the semiclassical models tend to overestimate the energy of the defect formation, 1.02 eV being the lowest value reported for grossular by Wright et al.,81 the local density approximation applied to the periodical model of this defect in R-quartz seems to overbind it. In particular Lin et al.82 have reported a negative defect formation energy of about -2.4 eV with respect to an interstitial water molecule and of about -0.5 eV with respect to free water. Our calculations show that this energy evolves as a fine balance between the water physisorption energy in the cavities of siliceous materials and the intensive hydrogen bonding within the defect. When we consider the defect formation as a process in which two isolated water molecules are brought to a defect site from the vacuum, we calculate the energy of formation to be negative, also being about -0.5 eV. The physisorption of one water molecule on the internal wall of a sodalite cage allows us to gain about 0.5 eV, which makes the defect unfavorable with the defect formation energy of ca. 0.5 eV, or 12.7 kcal/mol. The calculated energy of physisorption should be considered carefully. In fact, this value consists of two contributions: the physisorption energy itself, which should be compared with experiment (unfortunately, there are no such data available for highly siliceous zeolites) and the basis set superposition error, which can be quite large in its absolute value. Indeed, the binding energy within an isolated water molecule is calculated as ca. 10.5 eV and the basis superposition error estimated using the extended Gaussian-type basis set (the 6-311G** basis comparable for energetics with the DNP basis set used here) proves to be 0.1-0.3 eV depending on the values of the exponents used. The qualitatively different character of the numerical basis set precludes the direct use of these values but can be applied as a rough estimate for an upper bound on the basis superposition error. Thus nearly half of the value calculated perhaps
6172 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Sokol et al.
Figure 5. Hydroxyl nest defect: structural model obtained from DFT calculations.
Figure 6. Vicinal silanol defect: structural model obtained from DFT calculations.
should be subtracted from the energy of water physisorption. A similar estimate must then be carried out for silanol hydroxyl groups comprising the defect. To avoid this approximate procedure, we have used a direct comparison of the binding energies of the supercells that contain a physisorbed water molecule, the hydroxyl nest defect and that of pure siliceous sodalite. The earlier results of Lin et al.82 on the energetics and structure of hydroxyl nests in R-quartz show a number of features, which we did not find in our calculations using the sodalite structure. Particularly, Lin et al. calculated the positive energy of 0.9 eV for water insertion in the channels of quartz, which contrasts with our negative value of -0.56 eV for sodalite. Many calculations, including ours,77,83 have demonstrated that this defect is stabilized by extensive hydrogen bonding effects: up to six hydrogen bonds can form in its ground state. Lin et al. reported the alignment of all hydroxyl groups along the crystal c axis with the formation of just two internal hydrogen bonds. The gross overestimation of the hydrogen bond strength is a well established deficiency of the LDA approach. Thus in our view the close values for the defect formation energies in the two calculations is largely a coincidence. The geometry for the hydroxyl nest defect obtained in our DFT calculation is given in Figure 5. The hydroxyl bond length in our calculation is slightly overestimated, as with the Brønsted acid site, but an interesting trend is evident: the shorter the hydrogen bond in which the given hydrogen is involved, the longer the corresponding hydroxyl bond length, which naturally correlates with the hydrogen bond strength and a slight charge transfer toward the hydroxyl oxygen: the shortest hydrogen bond of 1.75 Å corresponds with the longest hydroxyl bond of 0.996 Å and the lowest effective charge on oxygen of -0.77 e. The generic structure of the defect, in fact, strongly resembles that obtained from results of semiclassical simulations58 and the first ab initio LDA molecular cluster calculations.84 The oxygen tetrahedron has a basal plane formed as a strongly bound threering and one hydroxyl group is situated over this basal plane with weak hydrogen bonds both to hydrogen and oxygen. The SisOsH bond angles vary between 115° and 118°, which is quite similar to the results of numerous molecular cluster calculations (see, for example, on silanol species the early review by Sauer85). Another interesting feature is a relatively small value for the SisO bond length in the silanols viz. 1.63-1.64 Å, which is rather close to all other SisO bond distances in the immediate vicinity of the defect. In contrast, pronounced elongation of this
bond is observed in the molecular cluster calculations where it probably relates to artificial effects of the unrealistic boundary conditions. Nevertheless, a slight polarizing effect of the defects introduced into the zeolite framework leads to bond elongation, which can therefore also account for the zeolite expansion observed at the first stages of dealumination. Often the hydrogen charge is used as a measure of a site acidity implying that the work required to withdraw a proton is a monotonic function of the electron population on hydrogen. Within such an approach, we observe in accord with chemical intuition that the hydroxyl nest defect should be a weaker acid site than the standard [H,Al]Si substitutional. However, this kind of rationalization has to be tested by modeling reactivity of the nest with conjugated bases. 5.1.3. Vicinal Disilanols. The structure of the vicinal silanol defect is the least clear among the common hydrogen defects considered so far, although the species in question seems to be the most important condensation defect of high silica zeolites. It can also result from hydrolysis of the SisOsSi bridges in any zeolite and thus relates to an important issue of zeolitic framework solubility and phase transformation. It is usually assumed that a hydrogen bond is formed between two hydroxyl groups in the defect structure. However, these two hydroxyl groups on two adjacent silicon sites come in close contact to each other in this defect, which is often not appreciated in the experimental literature. The hydrolysis of the SisOsSi bridge has been considered, for example by Koller51 as a part of the siloxy-silanol complex defect. The authors have used an interatomic potential based simulation to justify their model, which included the formation of a hydrogen bond between the two silanols. We have also applied semiclassical as well as semiempirical techniques to investigate the defect formation. Importantly, only the closed two-membered-ring structure resulted from these calculations performed on a number of crystalline silica materials: all attempts to separate hydroxyl groups failed; the geometry optimization leads to the formation of a ring and does not leave room for the hydrogen bonding suggested in the literature. To avoid any artificial effects of the preliminary modeling, we have started the geometry optimization using the DFT approach with the hydrogen bonded like structure. The results of our calculations are illustrated in Figure 6. As one can see, the two-member-ring structure has been reproduced by the ab initio modeling. The hydrogen bonding does not occur within the defect and there is only very weak interaction of one of the protons (H2) with bridging oxygens (O4 and O7) at a distance of 2.48 Å from one another. This interaction is responsible for
Microporous Silica and Aluminum Silicate Materials orientation of the disilanol ring in the plane of the 6-ring of which the linkage (-O4sSi1sO2sSi2sO7-) is a member. The essential feature of the defect preserved by the DFT calculation is a relatively short inter-oxygen distance of 2.13 Å (despite the significant increase compared to 1.5-1.8 Å predicted by semiempirical and semiclassical simulation techniques), the value being intermediate between typical interoxygen distances in peroxides (of 1.45-1.55 Å) and oxides (of 2.5-3.0 Å). This value could also be associated with the O23species, but the defect is not spin polarized and the only possible explanation lies with a shift of the charge density centered on oxygens toward the hydrogens of the two hydroxyl groups in effect reducing the Coulomb repulsion between the two oxygen ions. The close values of the SisO bond length of ca. 1.84 Å and the very ring structure of the defect suggest the multi-center character of the bonding in this system. The difference in Si1s O1 and Si1sO2 bond distances is clearly due to a weak interaction between protons and the O4 and O7 ions forming a 6-ring. Our modeling of this defect species using small (2 T-site) molecular models has yielded fully symmetrical configurations with equal corresponding bonds. The close structure of the ring and the short separation between the two oxygens has further been investigated using customized triple and quartic basis sets with higher moment polarization functions. We note that although small changes could be observed in bond lengths and angles, a ring structure has remained intact and thus is not a consequence of possible basis set superposition errors. The defect could be identified experimentally (although with difficulty) by the characteristic SisH separation distance of ca. 2.5 Å (the regular distance for an isolated, or nonbridging silanols being in the range 2.1-2.3 Å), which would not be overlapped by the signal from the Brønsted acid sites in the high silica zeolites. Thus both proton and 29Si cross-polarization solid-state NMR experiments seem to be good candidates for the identification of this species. Another peculiarity of the defect is the 5-fold coordination of the silicon, which has not been reported in the experimental literature. We note, however, that the regular interpretation of the 29Si NMR spectra with uniform shifts between Qn groups is not entirely applicable in this case. The hydroxyl groups on silicon in the vicinal silanol defect are not free; the electronic polarizing charge is concentrated nonuniformly around each site, and therefore it is not quite clear where the corresponding signal should appear in the spectrum. Thus a simulation of the NMR spectrum is needed in future studies. A similar situation occurs with the IR spectra as an unusually long SisO bond distance and the bridging character of the hydroxyl groups hinder any qualitative prediction even of the direction of the shift in the energy of the silanol hydroxyl stretching and bending vibrations. The effective charge on the hydrogens in the defect is 0.39 e, which resembles the results for the hydroxyl nest defect with the same kind of implications for the zeolite Brønsted acidity. However, in contrast to the nest defect, the hydroxyl groups in the vicinal species are of the terminal character; the hydrogen does not take part in the formation of hydrogen bonds, which makes it far more accessible to the guest species in the zeolite pores. The formation energy for the vicinal silanol defect viz. 34.4 kcal/mol has been calculated here with respect to the water sorbed on the sodalite cage wall similar to our calculation on the hydroxyl nest defect. The relatively low value implies a high concentration of the defect in real materials and underlines its importance for understanding the physicochemical properties of zeolites.
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6173
Figure 7. Energetics of silanol inversion in common hydrogen containing defects in zeolites. Energy in kcal/mol.
5.2. Silanol Inversion. 5.2.1. Mechanism and Energetics of Defect Transformation. As the first step in our investigation of defect transformations we have studied silanol inversion within three major hydrogen containing species considered above: the Brønsted acid site, the hydroxyl nest and the vicinal disilanols. This mechanism does not change the overall chemical composition of the defects but transforms these defects into relatively isolated silanol groups at the internal surface of zeolites and new defect species. The reaction could be facilitated in zeolites by the presence of water, and it gives rise to a number of catalytically interesting species, which can play the role of both Lewis and Brønsted acid sites in these materials. These sites are also primary candidates for ion exchange processes, in which transition metal ions can be grafted on to the internal surfaces of zeolites. The mechanism and energetics of silanol inversion resulting from our calculations are outlined in Figure 7. At the initial stage of the process a Si atom is in a tetrahedral coordination of oxygens. One of these oxygens is saturated with a proton (a silanol hydroxyl group) while the other three, siloxane oxygens are bonded to silicon atoms of the second coordination sphere. During inversion, the Si atom moves from its central, tetrahedral position through the basal plane of the siloxane oxygens into an interstitial space available in zeolites in the form of channels and cavities. The hydroxyl group follows the Si atom. The silanol hydroxyl group in the three main defects substitutes for the fourth regular siloxane oxygen. In the process of inversion this group leaves behind a vacant site, which is fully analogous to an oxygen vacancy. This site can be occupied by other guest species, thus further extending opportunities for modification of the zeolite framework. Due to the high computational cost, we have not investigated actual reaction paths and associated barriers. However, to an extent we have mimicked this process in the case of the vicinal disilanols described below and estimated the barrier to be under 1 eV. With these relatively low barrier and reaction energies we should expect the inversion to occur readily. First, we consider the transformation of the Brønsted acid site, in which a trigonal aluminum is formed. As mentioned, this center has been postulated as an important framework Lewis
6174 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Sokol et al. TABLE 6: Inversion of the Brønsted Acid Site, Bond Distances (Å) and Angles (deg) Obtained from DFT Calculations H1-O1 Si1-O1 Si1-O2 Si1-O7 Si1-O8 Al1-O4 Al1-O5 Al1-O6
Figure 8. Inversion of the Brønsted acid site: structural model obtained from DFT calculations.
TABLE 4: Hydroxyl Nest Defect, Bond Distances (Å) and Angles (deg) Obtained from DFT Calculations H1-O1 Si1-O1 H1-O1-Si1 H2-O2 Si2-O2 H2-O2 -Si2 H3-O3 Si3-O3 H3-O3-Si3 H4-O4 Si4-O4 H4-O4-Si4 Si4‚‚‚H1
0.996 1.636 115.4 0.986 1.641 118.2 0.988 1.637 113.8 0.969 1.632 117.3 4.02
H1‚‚‚O2 O1‚‚‚O2 H1‚‚‚H2 H2‚‚‚O3 O2‚‚‚O3 H2‚‚‚H3 H3‚‚‚O1 O3‚‚‚O1 H3‚‚‚H1 H4‚‚‚H1 H4‚‚‚H2 H4‚‚‚H3 Si4‚‚‚H2
1.75 2.70 2.00 1.87 2.74 2.10 1.80 2.71 1.93 2.55 3.47 3.15 3.76
Si1-O19 Si1-O5 Si1-O20 Si2-O6 Si2-O15 Si2-O16 Si3-O13 Si3-O14 Si3-O8 Si4-O7 Si4-O9 Si4-O10 Si4‚‚‚H3
1.632 1.631 1.643 1.629 1.632 1.635 1.635 1.632 1.625 1.629 1.642 1.636 4.10
TABLE 5: Vicinal Silanol Defect, Bond Distances (Å) and Angles (deg) Obtained from DFT Calculations O1‚‚‚O2
2.13
H1-O1
0.968
H2-O2
0.977
H1‚‚‚Si1 H1‚‚‚Si2 H2‚‚‚Si1 H2‚‚‚Si2 H1‚‚‚O2 H1-O1-Si1 H1-O1-Si2
2.49 2.51 3.03 120.3
Si1-O1 Si2-O1 Si1-O2 Si2-O2 Si1-O3 Si2-O6 Si1-O4 Si2-O7 Si1-O5 Si2-O8 H2‚‚‚O1 H2-O2-Si1 H2-O2-Si2
1.873 1.811 1.636 1.687
Si3-O8 Si4-O5 Si3-O9 Si4-O12 Si3-O10 Si4-O11 Si3-O13 Si4-O13
1.625
Si1‚‚‚Si2
2.99
1.624 1.639 1.633
1.644 3.11 123.9
acid site.49,53 We propose here a novel, low energy path for the formation of this species. Next, we concentrate on the silanol inversion in the vicinal disilanol ring structure, which proved to be an exothermic process. Finally, we describe the formation of the trisilanol ring resulting from inversion of the hydroxyl nest defect. 5.2.2. From the Brønsted to the Lewis Acid Site: Trigonal Aluminum. Silanol inversion in the Brønsted acid site leads to the formation of a pair of silanol and trigonal aluminum defects, shown in Figure 8 with structural parameters collected in Table 6. The O1sH1 bond length of 1.64 Å and H1sO1sSi1 bond angle of 117° are characteristic of a free silanol group while the OsAlsO bond angles within 2° of the 120° value confirm the trigonal configuration of the Al species. The increase of 1.12 Å in the separation of Si1 and Al atoms upon inversion results in the opening of an approach to the Al site for guest molecules, conjugated bases being of particular interest. High reactivity known to be associated with trigonal aluminum
0.972 1.642 1.631 1.633 1.650 1.690 1.708 1.701
H1-O1-Si1 H1‚‚‚Si1 H1‚‚‚Al1 O1‚‚‚Al1 Si1‚‚‚Al1 O4-Al1- O5 O5-Al1-O6 O6-Al1-O4
116.7 2.25 6.41 6.04 4.47 122.2 119.0 118.3
Si2-O2 Si2-O3 Si2-O9 Si2-O11 Si3-O3 Si3-O4 Si3-O10 Si3-O12
1.637 1.626 1.625 1.629 1.635 1.615 1.634 1.640
implies quick modification of the site upon interaction with guest molecules. As discussed in sections 2.2 and 2.3, the trigonal aluminum site has been postulated in the experimental literature as a major Lewis acid site in zeolites. This site is expected to form in the dehydroxylation processes, which are activated by high-temperature treatments used in the ultrastabilization and, more generally, the calcination of zeolites. In the process of dehydroxylation both three-coordinated Al and Si should be formed; however, there is still little evidence for the three-coordinated Si in zeolites.86-88 We have recently reported a relatively high energy of 72 kcal/mol for the defect pair formation in the concerted dehydroxylation-deprotonation process over two adjacent Brønsted acid sites,89 which compares unfavorably with 28 kcal/mol, which we calculate for the silanol inversion of the Brønsted acid site. In this context we propose the silanol inversion to be an important step in the zeolite dealumination. Indeed, the hydration of the resulting trigonal aluminum upon steaming, used in the ultrastabilization, should lead to highcoordinated Al species outlined in Figure 1 and considered in detail by Ruiz et al.90 The hydrolysis of AlsO bonds at a regular Brønsted acid site would result in a certain strain in the surrounding framework, a preliminary or concurrent inversion of the silanol should release this strain and facilitate further steps in the dealumination, which eventually would lead to the formation of hydroxyl nests. Alternatively, upon annealing in anhydrous conditions, silanol and trigonal aluminum can be formed as a metastable species. Some of these sites can be revealed at the surface termination of a zeolite, which can explain the fact that XPS, being a surface spectroscopy, has so far been the only experimental technique which has unambiguously determined trigonal aluminum in zeolites.91 Further experimental support should also be expected from novel NMR techniques, which have successfully been used in the recent determination of a number of distinct Al species in zeolites (see, for example, the work of Fyfe et al.92 and references therein). 5.2.3. InVersion of the Vicinal Disilanol Ring: Single Silanols. A number of alternative paths for the transformation of the vicinal disilanol pair defect can be envisaged. Two patterns can be recognized in the configuration of the resulting species: one involving the breaking of the ring structure and the other retaining the ring structure. The simplest example of the ring breaking mechanism is the dehydration of the vicinal silanol pair defect, which restores an intact SisOsSi linkage, the siloxane bridge. This process would reverse the formation of the vicinal silanol pair defect in high silica materials. In the presence of water in the pores of the zeolite, site deprotonation can also occur: the protons are transferred to the adjacent bridging oxygens by water molecules forming the transient H3O+ hydroxonium ions. It has been of particular interest to us to investigate the stability of the ring structure of the defect toward such a process since it could lead to the formation of the tSisO-‚‚‚HOsSit species proposed in the work of Koller
Microporous Silica and Aluminum Silicate Materials
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6175
Figure 10. Inversion of the hydroxyl nest: structural model obtained from DFT calculations.
Figure 9. Inversion of the vicinal disilanol ring: structural model obtained from DFT calculations.
et al.,51 as discussed previously. Moreover, the attachment of a proton to one of the adjacent oxygen ions can be the source of the isolated silanol groups and their derivatives in the bulk zeolite, the existence of which has been frequently doubted in the experimental literature. As mentioned, this study has been limited by the use of periodic boundary condition techniques; thus we could not follow the deprotonation process as such, but to an extent we could emulate it by choosing certain defect structures as starting points for the geometry optimization. In particular here, the geometry of the previously optimized vicinal silanol pair has been fixed except for one proton, which was moved to an adjacent oxygen site. The following geometry relaxation has proved to be a costly calculation, which included more than 200 optimization cycles. During the first stage, the ring structure has been preserved until a very shallow valley in the coordinate space has been found. Following this valley, the ring decomposition and the complete inversion of the silicon through the basal plane of three oxygens has occurred. If the initial proton transfer required an energy input of ca. 50 kcal/mol, the following relaxation in the first stage has reduced this value by ca. 25 kcal/mol, and the final descent has yielded another 50 kcal/mol. Thus the defect structure, which comprises two practically isolated silanol groups, lies 21.9 kcal/mol lower in energy than the vicinal silanol pair ring. Only one of the silanol groups in the resultant defect forms a weak hydrogen bond to an adjacent bridging oxygen ion (with the bond distance of ca. 1.98 Å). The weak character of the bond is also corroborated by a 122.5° bond angle and a 0.970 Å length of the hydroxyl bond, the values being in the characteristic range for free silanols within the given computational scheme. The resulting structure is shown in Figure 9. A slight difference in the geometries of the two silanols is also reflected in the effective charge distribution, the hydrogen bonded group being a little more polarized (with a charge transfer of about 0.02 e from the hydrogen to oxygen). The difference is even more pronounced in the SisO bond distances (one bond is 0.021 Å longer than the other) and the effective charge on silicon (a difference of about 0.03 e). An important distinction between the isolated silanols and the ring structure of the vicinal disilanol defect is the bridging character of the hydroxyl groups, which in the latter should give rise to characteristic shifts in the observed IR and NMR spectra. The low calculated energy of formation leaves no doubt as to
TABLE 7: Inversion of the Vicinal Disilanol Ring, Bond Distances (Å) and Angles (deg) Obtained from DFT Calculations H1-O1 Si1-O1 Si1-O3 Si1-O4 Si1-O5 H2-O2 Si2-O2 Si2-O7 Si2-O8 Si2-O9
0.970 1.643 1.657 1.629 1.617 0.970 1.622 1.631 1.636 1.624
H1-O1-Si1 H1‚‚‚Si1 Si6-O3 Si3-O4 Si5-O5 H2-O2-Si2 H2‚‚‚Si2 Si4-O7 Si8-O8 Si7-O9
115.7 2.24 1.643 1.629 1.620 122.5 2.29 1.619 1.617 1.615
H2‚‚‚O3 O2‚‚‚O3 H1‚‚‚H2 Si1‚‚‚Si2 Si3-O6 Si3-O12 Si3-O13 Si4-O6 Si4-O10 Si4-O11
1.98 2.71 4.64 4.16 1.625 1.627 1.626 1.633 1.627 1.638
the presence of the isolated silanols in the zeolite bulk; the corresponding IR and NMR signals originating from such groups should hardly differ from the signal of the terminal silanols present on the zeolite surface. The geometry of these species, predicted by our calculation, is practically identical to that found in molecular cluster modeling (e.g., Sauer93). We conclude that the traditional assignment of experimental signal shifts in the IR and NMR spectra associated with the vicinal disilanols should be reevaluated. We do not find any extensive hydrogen bond formation within the defect, which would lead to significant spectroscopic shifts from the free, terminal silanol groups. Therefore, a provision should be made in the interpretation of these spectra for two distinct features: ring structure and weakly bound silanol groups situated at the adjacent T-sites. Ab initio based simulation of these spectra is presently under way. 5.2.4. InVersion of the Hydroxyl Nest: Trisilanol Ring. Transformation of the hydroxyl nests outlined in Figure 2 involves various dehydration, dehydroxylation and dehydrogenation processes, all of which require at least 50 kcal/mol as will be discussed in following publications (also see ref 94). In contrast, the silanol inversion requires very little energy, 0.5 kcal/mol, and can be considered as the first step in the migration of the hydroxyl nest defects. The structure of the resulting defect is shown in Figure 10 with structural parameters given in Table 8. The 3-ring remains the major constituent element of the defect species, which should be responsible for characteristic NMR and IR shifts attributed to strongly hydrogen bonded species. Inversion of the silanol Si4sO4sH4 considerably reduces the perturbation of the trisilanol structure, the hydrogen bonds becoming shorter and more uniform (compare 1.75, 180, and 1.87 Å in the hydroxyl nest with 1.74, 1.74, and 1.83 Å in the trisilanol ring). The silanol inversion opens the defect to the interaction with guest molecules; thus the 3-ring structure can occur on the internal surface of zeolites, but it must also be present on their external surfaces. The trigonal Al species at the external surfaces, considered above, is a primary candidate
6176 J. Phys. Chem. B, Vol. 106, No. 24, 2002
Sokol et al.
TABLE 8: Inversion of the Hydroxyl Nest, Bond Distances (Å) and Angles (deg) Obtained from DFT Calculations H1-O1 Si1-O1 H1-O1-Si1 H2-O2 Si2-O2 H2-O2 -Si2 H3-O3 Si3-O3 H3-O3-Si3 H4-O4 Si4-O4 H4-O4-Si4
0.997 1.638 113.3 0.987 1.643 119.2 0.991 1.634 115.1 0.969 1.646 117.8
H1‚‚‚O2 O1‚‚‚O2 H1‚‚‚H2 H2‚‚‚O3 O2‚‚‚O3 H2‚‚‚H3 H3‚‚‚O1 O3‚‚‚O1 H3‚‚‚H1 Si4‚‚‚H1 Si4‚‚‚H2 Si4‚‚‚H3
1.74 2.69 1.99 1.83 2.69 2.11 1.74 2.65 1.90 5.00 4.60 4.76
Si1-O19 Si1-O5 Si1-O20 Si2-O6 Si2-O15 Si2-O16 Si3-O13 Si3-O14 Si3-O8 Si4-O7 Si4-O9 Si4-O10
1.629 1.629 1.626 1.629 1.634 1.631 1.635 1.639 1.634 1.622 1.642 1.651
for the transformation into such a trisilanol ring upon acid leaching and steaming that zeolites undergo during the dealumination treatments. The transformation of the nest upon thermal treatment, which would involve silanol inversion, could also explain a number of interesting features in the proton conduction experiments reported by Sayed.95-97 In particular, Sayed explained the enhancement in the proton conduction upon dealumination by a high mobility of protons over hydroxyl nests. The open character of the trisilanol structure seems to support Sayed’s arguments, while one of the activation energies reported could be the barrier for the silanol inversion in the hydroxyl nest. 6. Conclusion In this paper, we have shown the close relationship between elemental point defects studied in solid-state physics and catalytically important active centers in zeolites. The common hydrogen containing defects considered give rise to a number of new exciting species that can play the role of both Brønsted and Lewis active sites. We advance a novel mechanism of framework modification in zeolites, the silanol inversion. This mechanism leads to the formation of a trigonal Al site counterbalanced by a single silanol species and of a trisilanol ring at the Si vacancy site. The detailed results and discussion of the alternative transformation paths for the common hydrogen containing defects, which result in a number of new defect species, will be published shortly. Acknowledgment. We are grateful to the EPSRC for financial support for the computational resources used in this work. A.A.S. gratefully acknowledges financial support from the Dow Chemical Co. We are much obliged to Molecular Simulations Inc. who provided most of the computational and visualization software used. For many useful discussions we thank our colleagues M. McAdon, J. Ruiz, B. Schoemann, A. L. Shluger, P. E. Sinclair, D. W. Lewis, S. A. French, S. T. Bromley, and F. Cora`. References and Notes (1) Thomas, J. M. Sci. Am. 1992, 286, 82. (2) Barrer, R. M. Hydrothermal chemistry of zeolites; Academic Press: London and New York, 1982. (3) Liu, X.; Mao, Y.; Ruetten, S. A.; Thomas, J. K. Solar Energy Mater. Solar Cells 1995, 38, 199. (4) Simon, M. W.; Edwards, J. C.; Suib, S. L. J. Phys. Chem. 1995, 99, 4698. (5) Ozin G. A.; Ozkar, S. AdV. Matter. 1992, 4, 11. Jaskolski, W. Phys. Rep. 1996, 271, 1. (6) Dag, O ¨ .; Kuperman, A.; Ozin, G. A. AdV. Mater. 1995, 7, 72. (7) Stoneham A. M. Theory of defects in solids: Electronic structure of defects in insulators and semiconductors; Oxford University Press: Oxford, U.K., 1985. (8) Blumenfeld, A. L.; Fripiat, J. J. Top. Catal. 1997, 4, 119.
(9) Beyerlein, R. A.; Choi-Feng, C.; Hall, J. B.; Huggins, B. J.; Ray, G. J. Top. Catal. 1997, 4, 27. (10) Kustov, L. M. Top. Catal. 1997, 4, 131. (11) Kno¨ziger, H.; Huber, S. J. Chem. Soc., Faraday Trans. 1998, 94, 2047. (12) Catlow, C. R. A., Ed. Modelling of structure and reactiVity in zeolites; Academic Press: London, 1992. (13) Uglengo, P.; Garrone, E. J. Mol. Catal. 1989, 54, 439. (14) Garrone, E.; Uglengo, P. Mater. Chem. Phys. 1991, 29, 287. (15) Gale, J. D.; Catlow, C. R. A.; Carruthers, J. R. Chem. Phys. Lett. 1993, 216, 155. (16) White, J. C.; Hess, A. C. J. Phys. Chem. 1993, 97, 6398. (17) Ferrari, A. M.; Ugliengo, P.; Garrone, E. J. Phys. Chem. 1993, 97, 2671. (18) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders, V. R. Chem. ReV. 1994, 94, 2095. (19) Sauer, J. Zeolites and Related Microporous Materials: State of the Art 1994 1994, 84, 2039. (20) Nicholas, J. B.; Hess, A. C. J. Am. Chem. Soc. 1994, 116, 5428. (21) vanSanten, R. A. Catal. Today 1997, 38, 377. (22) Sinclair, P. E.; Catlow, C. R. A. Chem. Commun. 1997, 19, 1881. (23) Nicholas, J. B. Top. Catal. 1997, 4, 151. (24) Gonzales, N. O.; Bell, A. T.; Chakraborty, A. K. J. Phys. Chem. B 1997, 101, 10058. (25) Barich, D. H.; Nicholas, J. B.; Xu, T.; Haw, J. F. J. Am. Chem. Soc. 1998, 120, 12342. (26) Brandle, M.; Sauer, J.; Dovesi R.; Harrison, N. M. J. Chem. Phys. 1998, 109, 10379. (27) Civalleri, B.; Garrone, E.; Ugliengo, P. Chem. Phys. Lett. 1998, 294, 103. (28) Pacchioni, G.; Vitiello, M. Phys. ReV. B 1998, 58, 7745. (29) Demkov, A. A.; Sankey, O. F. Micropor. Mesopor. Mater. 1998, 21, 347. (30) Nicholas J. B. Top. Catal. 1999, 9, 181. (31) Civalleri, B.; Garrone, E.; Ugliengo, P. Chem. Phys. Lett. 1999, 299, 443. (32) Ricchiardi, G.; de Man, A.; Sauer, J. Phys. Chem. Chem. Phys. 2000, 2, 2195. (33) Schwarz, K.; Nusterer, E.; Blochl, P. E. Catal. Today 1999, 50, 501. (34) Teunissen, E. H.; Roetti, C.; Pisani, C.; de Man A. J. M.; Jansen, A. P. J.; Orlando, R.; van Santen, R. A.; Dovesi, R. Modelling Simul. Mater. Sci. Eng. 1994, 2, 921. (35) Shah, R.; Gale, J.; Payne, M. C. J. Phys. Chem. 1996, 100, 1688. (36) Haase, F.; Sauer, J. Micropor. Mesopor. Mater. 2000, 35-6, 379. (37) Breck, D. W. Zeolite molecular sieVes; Wiley: New York, 1974. (38) Dyer, A. An Introduction to Zeolite Molecular SieVes; Wiley & Sons: New York, 1988. (39) Gottardi, G.; Galli, E. Natural zeolites; Springer-Verlag: Berlin, Heidelberg, New York, Tokyo, 1985. (40) Griscom, D. L. J. Non-Cryst. Solids 1985, 73, 51. (41) Hannus, I.; Nagy, J. B.; Kiricsi, I.; Fonseca, A.; Fernandez, C.; Fejes, P. Z. Phys. Chem. 1995, 189, 229. (42) Beyer, H. K.; Belenykaja, I. M.; Mishin, I. W.; Borbely, G. In Structure and reactiVity of modified zeolites; Jacobs, P. A., et al., Ed.; Elsevier: Amsterdam, 1984; p 133. (43) Tohmon, R.; Mizuno, H.; Ohki, Y.; Sasagane, K.; Nagasawa, K.; Hama, Y. Phys. ReV. B 1989, 39, 1337. (44) Chang, Y.-F.; Somorjai, G. A.; Heinemann, H. J. Catal. 1995, 154, 24. (45) Tsai, T. E.; Griscom, D. L. Phys. ReV. Lett. 1991, 67, 2517. (46) Guzzi, M.; Martini, M.; Mattaini, M.; Pio, F.; Spinolo, G. Phys. ReV. B 1987, 35, 9407. (47) Sauer, J., Schirmer, W. In InnoVation in zeolite materials science; Grobert, P. J., Mortier, W. J., Vansant, E. F., Schulz-Ekloff, G., Eds.; Elsevier: Amsterdam, 1988; p 323. (48) Sauer, J. J. Mol. Catal. 1989, 54, 312. (49) Stamires, D. N.; Turkevich, J. J. Am. Chem. Soc. 1964, 86, 749. Stamires, D. N.; Turkevich, J. J. Am. Chem. Soc. 1964, 86, 757. (50) Vedrine, J. C.; Naccache, C. J. Phys. Chem. 1973, 77, 1606. (51) Abou-Kais, A.; Vedrine, J. C.; Massardier, J.; Dalmai-Imelik, G. J. Chem. Soc., Faraday Trans. 1974, 70, 1039. (52) Abou-Kais, A.; Vedrine, J. C.; Massardier, J. J. Chem. Soc., Faraday Trans. 1975, 71, 1697. (53) Uytterhoeven, J. B.; Christner, L. G.; Hall, W. K. J. Phys. Chem. 1965, 69, 2117. (54) Brunner, E.; Ernst, H.; Freude, D.; Hunger, M.; Pfeifer, H. In InnoVation in zeolite materials science; Grobert, P. J., Mortier, W. J., Vansant, E. F., Schulz-Ekloff, G., Eds.; Elsevier: Amsterdam, 1988; p 155. (55) Zhidomirov, G. M.; Yakovlev, A. L.; Milov, M. A.; Kachurovskaya, N. A.; Yudanov, I. V. Catal. Today 1999, 51, 397.
Microporous Silica and Aluminum Silicate Materials (56) Davis, L. P.; Burggraf, L. W. In Ultrastructure Processing of AdVanced Ceramics; Mackenzie, J. D., Ulrich, D. R., Eds.; Wiley & Sons: New York, 1988; p 367. (57) Groenen, E. J. J.; Kortbeek, A. G. T. G.; Mackay, M.; Sudmeijer, O. Zeolites 1986, 6, 403. (58) Testa, F.; Crea, F.; Nastro, A.; Aillo, R.; Mostowitcz, R.; Nagy, J. B. Zeolites 1991, 11, 633. (59) Barrer, R. M.; Makki, M. B. Can. J. Chem. 1964, 42, 1481. (60) Kerr, G. T. J. Phys. Chem. 1967, 71, 4155. Kerr, G. T. J. Phys. Chem. 1968, 72, 2594. Kerr, G. T. J. Catal. 1969, 15, 200. (61) Fejes, P.; Hannus, I.; Kirisci, I.; Pfeifer, H.; Freude, D.; Oehme, W. Zeolites 1985, 5, 45. (62) Chezeau, J. M.; Delmotte, L.; Guth, J. L.; Gabelica, Z. Zeolites 1991, 11, 598. (63) Debras, G.; Nagy, J. B.; Gabelica, Z.; Bodart, P.; Jacobs, P. A. Chem. Lett. 1983, 199. (64) Bodart, P.; Nagy, J. B.; Debras G.; Gabelica, Z.; Jacobs, P. A. J. Phys. Chem. 1986, 90, 5183. (65) Fernandes, L. D.; Bartl, P. E.; Monteiro, J. L. F.; da Silva, J. G.; de Menezes, S. C.; Cardoso, M. J. B. Zeolites 1994, 14, 533. (66) Shih, S. J. Catal. 1983, 79, 390. (67) Wichterlova´, B.; Nova´kova´, J., Pra´sˇil, Z. Zeolites 1988, 8, 117. (68) Engelhardt, G. In Introduction to zeolite science and practice; van Bekkum, H., et al., Eds.; Elsevier: Amsterdam, 1991; p 285. (69) Brunner, E.; Ernst, H.; Freude, D.; Hunger, M.; Pfeifer, H. In InnoVation in Zeolite Material Science; Grobet, P. J., et al., Eds. Elsevier: Amsterdam, 1988; p 155. (70) Sokol, A. A.; Catlow, C. R. A. In Computer Modelling of Electronic and Ionic Processes in Solids; Tennison, R. C., Kiv, A. E., Eds.; Kluwer: Amsterdam, 1997; p 125. (71) Fenzke, D.; Hunger, M.; Pfeifer, H. J. Magn. Reson. 1991, 95, 477. (72) Hunger, M.; Freude, D.; Pfeifer, H.; Schwieger, W. Chem. Phys. Lett. 1990, 167, 21. (73) Koller, H.; Lobo, R. F.; Burkett, S. L.; Davis, M. E. J. Phys. Chem. 1995, 99, 12588. (74) Delley, B. J. Chem. Phys. 1990, 92, 508. EOM DSolid User Guide; Molecular Simulations: San Diego, Sept 1996. Dmol3, Cerius2 3.5, User Guide; Molecular Simulations: San Diego, Sept 1997. (75) Perdew, J. P.; Zunger, A. Phys. ReV. B 1981, 23, 5048. Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Phys. ReV. B 1992, 46, 6671. Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244.
J. Phys. Chem. B, Vol. 106, No. 24, 2002 6177 (76) Richardson, J. W.; Pluth, J. J., Jr.; Smith, J. V.; Dytrych, W. J.; Bibby, D. M. J. Phys. Chem. 1988, 92, 243. (77) Sokol, A. A.; Catlow, C. R. A.; Garces, J. M.; Kuperman, A. J. Phys. Chem. B 1998, 102, 10647. (78) Denks, V. P. Phys. Solid State 1994, 36, 918. (79) Bordiga, S.; Roggero, I.; Ugliengo, P.; Zecchina, A.; Bolis, V.; Artioli, G.; Buzzoni, R.; Marra, G.; Rivetti, F.; Spano, G.; Lamberti, C. J. Chem. Soc., Dalton Trans. 2000, 21, 3921. (80) Bordiga, S.; Ugliengo, P.; Damin, A.; Lamberti, C.; Spoto, G., Zecchina, A.; Spano, G.; Buzzoni, R.; Dalloro, L.; Rivetti, F. Top. Catal. 2001, 15, 43. (81) Wright, K.; Freer, R.; Catlow, C. R. A. Phys. Chem. Minerals 1994, 20, 500. (82) Lin, J. S.; Payne, M. C.; Heine, V.; McConnel, J. D. C. Phys. Chem. Minerals 1994, 21, 150. McConnell, J. D. C.; Lin, J. C.; Heine, V. Phys. Chem. Minerals 1995, 22, 357. (83) Brodholt, J. P.; Refson, K. J. Geophys. Research (Solid Earth). 2000, 105 (B8) 18977. (84) Purton, J.; Jones, R.; Heggie, M.; O ¨ berg, S.; Catlow, C. R. A. Phys. Chem. Minerals 1992, 18, 389. (85) Sauer, J. J. Chem. ReV. 1989, 89, 199. (86) Kazansky, V. B. Catal. Today 1988, 3, 367. (87) Hunger, M.; Freude, D., Pfeifer, H. J. Chem. Soc., Faraday Trans. 1991, 87, 657. (88) Freude, D.; Ernst, H., Wolf, I. Solid State Nucl. Magn. Reson. 1994, 3, 271. (89) Sokol, A. A.; Catlow, C. R. A.; Garces, J. M.; Kuperman, A. AdV. Mater. 2000, 12, 1801. (90) Ruiz, J. M.; McAdon, M. H.; Garces, J. M. J. Phys. Chem. B 1997, 101, 1733. (91) Remy, M. J.; Genet, M. J.; Poncelet, G.; Lardinois, P. F.; Notte´, P. P. J. Phys. Chem. 1992, 96, 2614. (92) Fyfe, C. A.; Bretherton, J. L.; Lam, L. Y. J. Am. Chem. Soc. 2001, 123, 5285. (93) Sauer, J. J. Phys. Chem. 1987, 91, 2315. (94) Sokol, A. A.; Catlow, C. R. A.; Garces, J. M.; Kuperman, A. In Proceedings of the 12th International Zeolite Conference; Treacy, M. J., et al., Eds.; MRS: Baltimore, 1998;Vol. 1, p 457. (95) Sayed, M. B. J. Phys. Chem. Solids 1992, 53, 1041. (96) Sayed, M. B. Zeolites 1996, 16, 157. (97) Sayed, M. B. Micropor. Mesopor. Mater. 2000, 37, 107.