J. Phys. Chem. 1996, 100, 6663-6671
6663
Computational Studies of Water Adsorption in the Zeolite H-ZSM-5 S. A. Zygmunt,† L. A. Curtiss,*,‡ L. E. Iton,‡ and M. K. Erhardt† Department of Physics and Astronomy, Valparaiso UniVersity, Valparaiso, Indiana 46383, and Argonne National Laboratory, Argonne, Illinois 60439 ReceiVed: October 3, 1995; In Final Form: January 26, 1996X
Ab initio molecular orbital calculations using Hartree-Fock theory and Møller-Plesset perturbation theory have been used to study the interaction of H2O with the Brønsted acid site in the zeolite H-ZSM-5. Aluminosilicate clusters with up to 28 T atoms (T ) Si, Al) were used as models for the zeolite framework. Full optimization of a 3 T atom cluster at the MP2/6-31G(d) level indicates that the “ion-pair” structure, Z-‚‚‚HOH2+, formed by proton transfer from the acid site of the zeolite (ZH) to the adsorbed H2O molecule, is a transition state, while the “neutral” adsorption structure, ZH‚‚‚OH2, is a local energy minimum. Partial optimization of a larger 8 T cluster at the HF/6-31G(d) level also gave results suggesting that the ion-pair structure is a transition state. Calculations were carried out to obtain corrections for high levels of theory, zero-point energies, and larger cluster size. The resulting energy difference between the neutral and ion-pair structure is small (less than 5 kcal/mol and possibly close to zero). The interaction energy of ZH‚‚‚OH2 is 13-14 kcal/mol, in agreement with experiment. We find that addition of a second H2O molecule to Z-‚‚‚HOH2+ in the 3 T atom cluster stabilizes the ion-pair structure, Z-‚‚‚H(OH2)2+, making it a local energy minimum. Finally, calculated vibrational frequencies for a 3 T atom cluster are used to help interpret experimental IR absorption spectra.
I. Introduction Brønsted acid chemistry is a dominant feature of zeolite catalysis in many important industrial applications. A reliable theoretical treatment of the proton affinity of H-ZSM-5 zeolite was obtained in previous work using high-level ab initio calculations on large cluster models.1 Proton transfer to a strong adsorbed base has also been studied theoretically in the interaction of ammonia with the Brønsted acid site.2-5 The interaction of weak bases, e.g., H2O, presents a more equivocal situation. What kind of equilibrium structure is formed when H2O is adsorbed at the Brønsted acid site in H-ZSM-5? Experimental evidence in the form of 1H NMR spectroscopy has suggested that the neutral hydrogen-bonded adsorbate and the ion-pair structure, formed by proton transfer from the acid site to H2O, coexist at 4 K, with the neutral complex being more prevalent than the ion-pair structure.6,7 Studies using IR spectroscopy have suggested that the ion-pair structure is most stable.8-11 One ab initio theoretical study12 also gave evidence to support this conclusion. However, more recent ab initio studies13,14 and experimental IR studies15,16 have led to a reinterpretation of the IR spectrum in a manner consistent with a neutral hydrogen-bonded adsorbate. These studies suggest that the ion-pair structure should be regarded as a transition state and not a true equilibrium geometry. Two recent reviews have included detailed discussions of the interpretation of the IR spectrum.4,17 Almost all of the theoretical studies have used small clusters to represent the acid site, and very few have accounted for electron correlation in their optimization of molecular geometries. In the light of the conflicting experimental data on the nature of water adsorption in H-ZSM-5, further ab initio computational studies using larger clusters and higher levels of theory are clearly desirable. * Author to whom correspondence should be mailed. Email:
[email protected]. † Valparaiso University. Email:
[email protected]. ‡ Argonne National Laboratory. X Abstract published in AdVance ACS Abstracts, March 15, 1996.
0022-3654/96/20100-6663$12.00/0
In this paper we report on a study of the interaction of water with an acid site in H-ZSM-5 incorporating large cluster size, electron correlation, and local geometry optimization in a unified way. The calculations presented here extend our earlier study of the adsorption of H2O on a 2 T atom (T ) Si, Al) cluster model of H-ZSM-518 and a preliminary report of some aspects of this work.19 In section II we describe the theoretical methods, and in section III we present results for the geometry and interaction energy of the neutral H2O adsorption structure, the relative energy of the ion-pair structure resulting from proton transfer, and the vibrational frequencies of the H2O adsorption structure. Results for the addition of a second H2O molecule are also presented. The calculated frequencies of these structures are used to help interpret the published experimental IR spectra. The evidence from the 1H NMR studies is also discussed in light of our results. II. Theoretical Methods The theoretical calculations presented here are based on ab initio molecular orbital theory using Hartree-Fock (HF) theory and Møller-Plesset (MP) perturbation theory.20,21 We used four aluminosilicate clusters of increasing size to model the interaction of H2O with the Brønsted acid site in H-ZSM-5. These clusters include 3, 8, 18, and 28 T atoms and have a total of 14, 34, 69, and 101 atoms, respectively. Each cluster includes one Al atom and a charge-balancing proton to maintain a neutral zeolite framework and is terminated by H atoms at the periphery. Their stoichiometries are H9Si2AlO2, H19Si7AlO7, H29Si17AlO22, and H33Si27AlO40, respectively. We performed calculations on the zeolite framework clusters as well as adsorption complexes between these clusters and an H2O molecule. In addition, we investigated the effect of higher H2O pressure by adding a second H2O molecule to the 3 T cluster model. The 8, 18, and 28 T clusters have been discussed in detail in ref 1, where they were used to assess long-range electrostatic effects of the zeolite framework on the calculated proton affinity of the T(12)-O(24)T(12) site of the H-ZSM-5 zeolite framework (MFI structure).1 © 1996 American Chemical Society
6664 J. Phys. Chem., Vol. 100, No. 16, 1996
Zygmunt et al.
The O(24) site at the center of the 8, 18, and 28 T atom clusters is a likely acid site in ZSM-5. Other studies have used detailed embedding schemes to estimate the magnitude of the important long-range effects in cluster models of adsorption in zeolites.22,23 For calculations based on the 3 T atom cluster, the geometry was fully optimized. The 3 T atom ion-pair complex (Z-‚‚‚HOH2+) was optimized with Cs point group symmetry, while all other 3 T structures were optimized with C1 symmetry. Corrections using the calculated zero-point energies for the 3 T atom clusters were used as estimates for the structures based on the larger cluster models. For the 8 T atom cluster, which had C1 symmetry, two different methods were used to model the O(24) site in H-ZSM5. In the first method, the atoms were assumed to be fixed at positions given by X-ray diffraction studies of H-ZSM-5,24 with an OH distance of 0.98 Å for the acid site. Hydrogen atoms were used to tie off the peripheral Si atoms at an Si-H distance of 1.43 Å (a value taken from ab initio calculations on silanol). Previous computational studies of the proton affinity of H-ZSM51 have questioned the ability of such a rigid cluster model to adequately represent the zeolite framework. Its major defect is an inability to model the structure of the SiO(H)Al bridge that forms the Brønsted acid site, since the X-ray diffraction data correspond to a structure very low in aluminum content. Thus, a second approach, based on the constant-volume relaxation (CVR) method,1 was used to account for the local effects of atomic relaxation around the acid site in the otherwise unperturbed H-ZSM-5 framework. The atoms of the central O3SiOHAlO3 unit near the acid site were fully relaxed, while atoms outside this central unit were fixed at positions determined from X-ray diffraction data. We have found this constrained relaxation scheme to be a useful model for the effect of an adsorbate on the local structure of the zeolite acid site. Similar methods were used in a semiempirical study by Redondo and Hay25 and a density-functional study by Cook et al.26 In both rigid and CVR cluster methods, the six intermolecular degrees of freedom between the cluster and the adsorbate molecule were fully optimized. In the 8 T atom calculations, the H2O and H3O+ geometries were held fixed at the optimized geometries of the isolated molecules. After optimization of the 8 T atom structures with H2O and H3O+, the effect of more distant atoms was then included by embedding the 8 T clusters in successively larger fragments of crystalline H-ZSM-5 to obtain first the 18 T atom and then the 28 T atom clusters. This procedure was done both for the rigid and the CVR cluster methods. The intermolecular degrees of freedom were not reoptimized in the 18 and 28 T atom clusters. We tested the accuracy of this method by reoptimizing these six coordinates in the 18 T atom complex with H3O+ at the HF/3-21G level and comparing them to those obtained in the 8 T atom cluster (rigid cluster model). Changes in bond lengths and angles on the order of 3-5% were noted, but the total energy after reoptimization decreased by less than 1 kcal/mol. Four levels of theory were used in this study. These include HF/3-21G, HF/6-31G(d), MP2/6-31G(d), and G2(MP2,SVP). G2(MP2,SVP) theory27,28 is a modification of G2(MP2) theory,29 in which the 6-31G(d) basis set is used instead of the 6-311G(d,p) basis set for the quadratic configuration interaction (QCISD(T))30 calculations. Using G2(MP2,SVP) theory, the energy of a molecule or cluster at 0 K can be written
E0 ) E(MP2/6-31G(d)) + ∆QCI + ∆BS + HLC + ∆ZPE (1) where
∆QCI ) E(QCISD(T)/6-31G(d)) - E(MP2/6-31G(d))
(2)
∆BS ) E(MP2/6-311+G(3df,2p)) - E(MP2/6-31G(d)) (3) E(MP2/6-31G(d)) is the electronic energy at the MP2/6-31G(d) level of theory. The HLC is a higher level correction for remaining basis set deficencies and ∆ZPE is the zero-point energy correction. The HLC cancels out in all of the reaction energies considered in this paper. G2(MP2,SVP) theory is based on MP2/6-31G(d) geometries and scaled (0.893) HF/6-31G(d) zeropoint energies. G2(MP2,SVP) theory was found to have an average absolute deviation of 1.63 kcal/mol in an assessment of the 125 reaction energies in the G2 test set.31 This theory was applied rigorously to the 3 T atom clusters. The basis set (∆BS), correlation (∆QCI), and zero-point (∆ZPE) energy corrections were also used as estimates for the 8 T cluster. III. Results and Discussion A. Reaction Energies. The reaction energies of H2O interacting at the hydroxyl site of aluminosilicate clusters of 3, 8, 18, and 28 T atoms representing H-ZSM-5 are listed in Tables 1 and 2. The quantities calculated in this study were ∆Eion, ∆Eneut, ∆Erel, and ∆Edesorp. The ∆Eion is the complexation energy of the “ion-pair” structure:
∆Eion ) E(Z-) + E(H3O+) - E(Z-‚‚‚HOH2+)
(4)
where Z- is the unprotonated zeolitic cluster and Z-‚‚‚HOH2+ is the ion-pair structure. The ∆Eneut is the complexation energy of the “neutral” (hydrogen-bonded) structure:
∆Eneut ) E(ZH) + E(H2O) - E(ZH‚‚‚OH2)
(5)
where ZH is the protonated zeolitic cluster and ZH‚‚‚OH2 is the neutral structure. The ∆Erel is the energy difference between the neutral and ion-pair structures:
∆Erel ) E(ZH‚‚‚OH2) - E(Z-‚‚‚HOH2+)
(6)
where a positive value would indicate that the ion-pair structure is more stable. The ∆Edesorp is the energy required to remove the H2O molecule from the most stable structure (ion-pair or neutral):
∆Edesorp ) E(ZH) + E(H2O) - min[E(Z-‚‚‚HOH2+), E(ZH‚‚‚OH2)] (7) Note that if the neutral structure is more stable than the ionpair structure, then ∆Edesorp ) ∆Eneut, which is what we found in both cases, as seen in Tables 1 and 2. The total energies of ZH, Z-, Z-‚‚‚HOH2+, and ZH‚‚‚OH2, from which the reaction energies in Table 2 are obtained, are listed in Table 3. The energies in Table 1 are based on rigid zeolite cluster calculations, while those in Table 2 are based on calculations which include relaxation of the local region of the cluster near the hydroxyl site, as described in the previous section. Results for the 3 T atom cluster, for which full optimizations were carried out, are also included in Table 2. The structures of the ion-pair and hydrogen-bonded configurations for the 3 T cluster are illustrated in Figure 1. At both HF/6-31G(d) and MP2/631G(d) levels, the neutral structure is a local minimum on the potential energy surface, while the ion-pair structure is a transition state associated with proton transfer between oxygen sites adjacent to the Al atom. The corresponding structures for the 8, 18, and 28 T clusters are illustrated in Figure 2, where the terminating H atoms at the periphery of the these clusters are not shown. For the 8 T atom cluster, the HF/6-31G(d) calculations show that the lowest energy structure is again a
Water Adsorption in the Zeolite H-ZSM-5
J. Phys. Chem., Vol. 100, No. 16, 1996 6665
TABLE 1: Reaction Energies (in kcal/mol) for Clusters with Rigid Structuresa ∆Eneut ) ∆Edesorp
∆Eion
method/basis
3
8
18
28
HF/3-21G HF/6-31G(d)
20.1
31.9 17.9
35.9 21.2
35.1 20.0
3
∆Erel
8
18
28
3
8
18
28
147.1 127.1
151.6 128.6
153.0 127.7
-12.7
-7.1 -10.1
-3.2 -8.5
-1.1 -6.9
a All results are from rigid clusters (3, 8, 18, 28 T atoms) with only the six intermolecular coordinates optimized. For the 18 T and 28 T clusters the 8 T optimized parameters are used. Zero-point energies not included in values.
TABLE 2: Reaction Energies (in kcal/mol) for Clusters Including Geometry Relaxationa ∆Eneut ) ∆Edesorp method/basis HF/3-21G HF/6-31G(d) MP2/6-31G(d)
3 15.2 22.1
∆Eion
8
18
28
31.7 14.7
34.8 17.0
34.2 16.7
3 138.4 150.1
∆Erel
8
18
28
154.6 131.1
154.2 130.9
154.6 129.7
3 -14.5 -6.2
8
18
28
-14.1 -16.9
-13.0 -14.7
-10.4 -13.3
a All results are from the CVR procedure as described in text, except for the 3 T cluster, which is a full geometry optimization. The 3 T cluster is illustrated in Figure 1; the other clusters (8, 18, 28 T atoms) are illustrated in Figure 2. For the 18 T and 28 T clusters the 8 T optimized parameters are used. Zero-point energies not included in values.
TABLE 3: Total Energies (in hartrees) for Clusters in Table 2a 3T HF/3-21G
HF/6-31G(d)
MP2/6-31G(d)
ZZH ZH‚‚‚OH2 Z-‚‚‚HOH2+ ZZH ZH‚‚‚OH2 Z-‚‚‚HOH2+ ZZH ZH‚‚‚OH2 Z-‚‚‚HOH2+
-974.338 36 -974.836 32 -1050.871 32 -1050.848 21 -974.918 11 -975.410 21 -1051.642 24 -1051.632 38
8T
18 T
28 T
-2784.716 76 -2785.240 28 -2860.876 80 -2860.854 36 -2799.554 73 -2800.045 63 -2876.079 85 -2876.052 97
-6783.058 69 -6783.574 97 -6859.216 42 -6859.195 64 -6819.126 69 -6819.610 18 -6895.648 09 -6895.624 60
-11001.846 91 -11002.360 60 -11078.001 11 -11077.984 49 -11060.226 22 -11060.706 12 -11136.743 46 -11136.722 21
a H O: E[MP2/6-31G(d)] ) -76.196 85, E[HF/6-31G(d)] ) -76.010 75, E[HF/3-21G] ) -75.585 96. H O+: E[MP2/6-31G(d)]) -76.475 11, 2 3 E[HF/6-31G(d)] ) -76.289 34, E[HF/3-21G] ) -75.891 23.
Figure 1. Optimized structures of (a) ion-pair and (b) hydrogen-bonded (neutral) configurations for H2O interacting with the hydroxyl site in an aluminosilicate cluster with 3 T atoms. Terminating H atoms are unlabeled.
hydrogen-bonded adsorption complex between H2O and the zeolite framework. Although we have calculated vibrational frequencies only for the 3 T atom cluster, the ion-pair structure is also indicated to be a transition state in the larger clusters, since removing the rigid structure constraint on H3O+ in the 8 T atom cluster CVR optimization at the HF/6-31G(d) level resulted in transfer of a proton with no apparent barrier from H3O+ to the acid site, i.e., Z-‚‚‚HOH2+ f ZH‚‚‚OH2. The results in Tables 1 and 2 indicate that the HF/3-21G calculations give binding energies consistently 15-25 kcal/mol too large compared to the 6-31G(d) basis and are thus not reliable for quantitative studies. This has also been noted in our previous study18 using smaller clusters. At the HF/6-31G(d) level, the rigid cluster method gives ∆Eneut, ∆Eion, and ∆Erel values that differ from those of the CVR method by up to 7 kcal/mol. The discrepancy between the two methods is particularly significant for ∆Erel. Finally, both the neutral and
ion-pair structures are stabilized by correlation effects as treated by MP2 theory. A comparison of HF/6-31G(d) and MP2/631G(d) calculations shows that these complexes are stabilized by 5-10 kcal/mol in the 3 T atom cluster with full optimization. There are several other factors which must be taken into account in order to obtain accurate values of ∆Edesorp and ∆Erel to compare with experiment and make predictions. These include (a) electron correlation effects beyond MP2 theory; (b) effects of larger basis sets, i.e., larger than 6-31G(d); (c) zeropoint energy effects; and (d) long-range electrostatic effects from the zeolite framework that are not included in the cluster calculation. We have applied G2(MP2,SVP) theory to the 3 T atom cluster (MP2/6-31G(d) geometries), as described in section 2, to take account of the first three factors. The higher level correlation effects are assessed by ∆QCI (eq 2); the larger basis set effects are assessed by ∆BS (eq 3); and the zero-point energy effects are assessed by ∆ZPE. The long-range electrostatic effect is estimated from the dependence of the reaction energies on cluster size using the rigid cluster results in Table 1. For the 3 T cluster this cluster size correction, ∆CS, is obtained as the difference between the reaction energies of the 3 T cluster and the 28 T cluster models. Our proton affinity study1 suggests that this correction is largely due to long-range electrostatic effects. Table 4 lists the ∆Edesorp and ∆Erel values based on 3 T cluster calculations including estimates for the four factors discussed above. In addition to using the 3 T atom cluster to obtain accurate values for ∆Eneut and ∆Erel, we also used the 8 T atom CVR cluster (HF/6-31G(d) energies in Table 2). Estimates for the correlation effects, basis set effects, and zero-point effects obtained from the 3 T atom cluster were applied to the 8 T
6666 J. Phys. Chem., Vol. 100, No. 16, 1996
Zygmunt et al.
Figure 2. Z-‚‚‚HOH2+ and ZH‚‚‚OH2 complexes for 8 T (a, b), 18 T (c, d), and 28 T (e, f) clusters. Terminating H atoms not shown.
energies. The cluster size effect, ∆CS, for the 8 T cluster is estimated from the difference in the reaction energies of the rigid 8 T and 28 T clusters in Table 1. The 8 T atom CVR cluster has the advantage of being a larger cluster than the 3 T atom cluster and should be a more realistic representation of the local relaxation that occurs at the interaction site. However, it has the disadvantage of not including the internal relaxation of the H3O+ in the ion-pair transition state, which may be significant. The 3 T cluster includes this effect since it is based on a full optimization at the MP2/6-31G(d) level. Table 4 lists the ∆Edesorp and ∆Erel values based on the 8 T cluster including
estimates for the four factors (a), (b), (c), and (d). We now discuss the results obtained from the 3 T and 8 T clusters. The G2(MP2,SVP) value for the desorption energy from the 3 T cluster at 0 K is 12.8 kcal/mol. Using the G2(MP2,SVP) corrections and a correction for MP2 theory based on the 3 T cluster, the 8 T cluster has a similar desorption energy, 12.3 kcal/mol. Correlation effects beyond the MP2 level, ∆QCI, decrease the desorption energy by 0.9 kcal/mol in the 3 T cluster. Correction for extension to a larger basis set, ∆BS, gives a 5.5 kcal/mol decrease in the desorption energy. The scaled HF/631G(d) zero-point energy correction from the 3 T cluster, ∆ZPE,
Water Adsorption in the Zeolite H-ZSM-5
J. Phys. Chem., Vol. 100, No. 16, 1996 6667
TABLE 4: Desorption and Relative Energies, ∆Edesorp and ∆Erel (in kcal/mol), from 3 T and 8 T Cluster Using Corrections for Cluster Size (CS), Correlation Effects (MP2,QCI), Basis Set (BS), and Zero-Point Energies (ZPE) ∆Edesorp method
correction
3T
a
14.7
HF/6-31G(d) MP2/6-31G(d)a
22.1 ∆MP2b ∆QCIc ∆BSc ∆ZPEc
G2(MP2,SVP) ∆CSe estimatef
8T
-0.9 -5.5 -2.9 12.8 -0.1 12.7
6.9 -0.9 -5.5 -2.9 12.3d 2.1 14.4
ion-pair structure (Z-‚‚‚HOH2+)
∆Erel 3T
TABLE 5: Selected Geometrical Parameters and Relative Energies of ZH‚‚‚OH2 and Z-‚‚‚HOH2+ for the 3 T Clustera
8T -16.9
-6.2 -1.6 0.1 1.9 -5.8 5.8 0.0
8.3 -1.6 0.1 1.9 -8.2d 3.2 -5.0
neutral structure (ZH‚‚‚OH2)
∆Erel
r(OzH) r(OwH) r(OwH′) ∠HOwH ∠OzAlOz r(OzHz) r(OzH) r(OwHz) r(OwH) r(OwH′) ∠HOwH′ ∠OzAlOz
HF/6-31G(d)
MP2/6-31G(d)
1.42 1.05 (0.97) 0.95 (0.97) 95.7(113.1) 96.8 0.97 2.01 1.75 0.96 (0.95) 0.95 (0.95) 106.6 (105.5) 98.5 -14.5
1.36 1.11 (0.99) 0.98 (0.99) 93.6 (111.4) 95.4 1.02 1.81 1.61 0.99 (0.97) 0.97 (0.97) 105.6 (104.1) 97.2 -6.2
a
From Table 2. HF/6-31G(d) value is from constant volume relaxation (CVR) with rigid H2O and H3O+ structures. MP2/6-31G(d) value is from a full optimization. b From difference in HF/6-31G(d) and MP2/6-31G(d) results on 3 T cluster in Table 2. c From calculations on MP2/6-31G(d) optimized geometry of 3 T cluster (see text). d This is an approximate G2(MP2,SVP) value for 8 T based on 3 T corrections. e Estimated correction for the effect of larger cluster size. From difference between 8 T and 28 T clusters (HF/6-31G(d)) for 8 T and from difference between 3 T and 28 T clusters (HF/6-31G(d)) for 3 T (Table 1). f Estimate based on G2(MP2,SVP) energy and correction for larger cluster size.
gives an additional 2.9 kcal/mol decrease. The final G2(MP2,SVP) desorption energy of 12.8 kcal/mol should be accurate to 2-3 kcal/mol for the 3 T cluster. The long-range electrostatic effect as estimated from the rigid clusters, ∆CS, is -0.1 kcal/mol for the 3 T cluster and 2.1 kcal/mol for the 8 T cluster, giving final estimates of the desorption energy of 12.7 and 14.4 kcal/mol, respectively. Thermal corrections will affect these values, but it is unclear whether the usual ideal gas corrections used in theoretical enthalpy calculations32 are valid in the restricted environment of the zeolite interior. Calculation of the basis set superposition error (BSSE) in the desorption energy for the 3 T rigid cluster by the counterpoise method gives a BSSE of 5.5 kcal/mol at the MP2/6-31G(d) level. This is similar to the correction obtained from using the 6-311+G(3df,2p) basis set [∆BS in Table 4] in G2(MP2, SVP) theory. Thus, we conclude that BSSE is probably negligible at the G2(MP2,SVP) level for these systems. A previous theoretical estimate33 for the desorption energy in a 2 T atom cluster was 11.2 kcal/mol, but this approach included no correction terms besides ∆(MP2). Our earlier study,18 using a 2 T atom cluster model including ∆MP2, ∆QCI, and ∆BS corrections, gave a value of 10.8 kcal/mol. Both of these results neglected the ∆ZPE correction and allowed for only a single O‚‚‚H hydrogen bond between the ZH framework and the H2O molecule. Another theoretical study using a 3 T atom cluster12 resulted in a value of 14.8 kcal/mol, but it did not account for the ∆CS, ∆QCI, and ∆BS terms calculated here. The experimental enthalpy of adsorption, which was obtained from a Clausius-Clapeyron analysis of p(T) data between 357 and 435 K for a coverage of less than one molecule per acid site, was found to be 12 ( 1 kcal/mol.8 The result presented here, in light of the uncertainties in the theoretical methods and the neglected thermal corrections, is consistent with experiment. The G2(MP2,SVP) value for the relative energy (∆Erel) from the 3 T cluster at 0 K is -5.8 kcal/mol (Table 4). Using the G2(MP2,SVP) corrections and a correction for MP2 theory based on the 3 T cluster, the 8 T cluster has a ∆Erel of -8.2 kcal/mol (Table 4). The corrections from G2(MP2,SVP) theory are all less than 2 kcal/mol for ∆Erel. The ∆Erel values based on G2(MP2,SVP) corrections for both the 3 T and 8 T clusters indicate that the neutral structure is more stable than the ionic
a Geometrical parameters defined in Figure 1. Bond lengths in Å, angles in degrees, and energies in kcal/mol; theoretical values for isolated molecules in parentheses.
structure. The estimate for the long-range electrostatic effect from the rigid clusters, ∆CS, is 5.8 kcal/mol for the 3 T cluster and 3.2 kcal/mol for the 8 T cluster giving the final estimates for ∆Erel of 0.0 and -5.0 kcal/mol, respectively. Thus, the longrange electrostatic effects favor the ion-pair structure over the neutral structure and decrease the barrier for proton transfer. If the internal relaxation of the H3O+ were included the barrier for the 8 T cluster would decrease further. The results for ∆Erel suggest that the barrier is small (less than 5 kcal/mol and possibly close to zero). The ∆Erel cannot be directly compared to an experimental value, but its small magnitude is consistent with the observed ease of H-D exchange between the H-ZSM-5 acid site and gas phase D2O at 293 K.34 Sauer et al.35 concluded in an ab initio theoretical study using a small zeolite cluster that the proton transfer barrier was a vanishingly small 0.1 kcal/mol, while in the related case of CH3OH, Haase and Sauer36 calculated a barrier of less than 1 kcal/mol. Bates and Dwyer used a densityfunctional method to calculate a proton transfer barrier for CH3OH of 1.4 kcal/mol.37 B. Structures. A comparison of the calculated geometrical parameters for the 3 T ZH‚‚‚OH2 and Z-‚‚‚HOH2+ structures, fully optimized at HF and MP2 levels of theory using the 6-31G(d) basis set, is shown in Table 5. As previously mentioned, the neutral structure is a local energy minimum, while the ion-pair structure is a transition state. In Table 5, note particularly the significant influence of electron correlation on the hydrogen-bond distances r(OzH) and r(OwHz) in the neutral structure. For the 8 T atom cluster, HF/6-31G(d) calculations also show that the lowest energy structure is a hydrogen-bonded adsorption structure between H2O and the zeolite framework, as shown in Figure 2b. The adsorbate is anchored to the framework by O‚‚‚H linkages of 1.70 and 2.06 Å. These bond lengths are within 0.05 Å of the HF/6-31G(d) results for the 3 T atom cluster shown in Table 5, and they are nearly the same as those found in a previous study35 using a double-zeta plus polarization (DZP) basis set at the HF level and a 3 T atom zeolite cluster. The influence of the adsorbed molecules on the 3 T and 8 T atom zeolite cluster geometry in the vicinity of the acid site can be seen in Table 6. These results show that when H2O is adsorbed at the acid site, there is a slight lengthening of the framework OzHz bond and a slight contraction of the SiOz and AlOz bonds adjacent to the acid site. The SiOzAl bond angle also contracts, consistent with a small displacement of the bridging Oz atom toward the adsorbate. The effect of the adsorbate is to polarize the framework OzHz bond and to donate
6668 J. Phys. Chem., Vol. 100, No. 16, 1996 negative charge to the framework (see Table 6). This is consistent with the downward shift in OzHz stretching frequency upon adsorption, and its enhanced IR absorption intensity (see section III C) is similar to other hydrogen-bonded complexes.38 In the transition state, the effect of the H3O+ adsorbate on the zeolite cluster geometry is similar to that of H2O but stronger. The SiOz and AlOz bonds contract by 0.04 Å or more, and the negative charge on the framework Oz atom increases, as shown in Table 6. One qualitative difference between the two cases is that H3O+ adsorption causes the SiOzAl bond angle to increase by about 1.5-2.0°, while H2O causes it to decrease by about the same amount. C. Vibrational Frequency Analysis. Since much of the disputed evidence about the nature of the adsorbed H2O structure has come from IR spectroscopy, we calculated vibrational frequencies for the 3 T atom clusters to compare with those observed in experimental spectra. While assigning the peaks in the experimental IR spectra has been a matter of some controversy, the main spectral features are clear.10 In H-ZSM5, sharp peaks occur at 3610 and 3726 cm-1 which have been identified as νOH stretching modes associated respectively with the SiO(H)Al acid site and the terminal SiOH groups which are present to some degree in all zeolites. Below about 2000 cm-1 lie additional peaks associated with the zeolite lattice modes, which overlap the in-plane bending mode (δOH) and the out-of-plane bending mode (γOH) of the acid site hydroxyl groups, making them difficult to study with IR spectroscopy, although they can be resolved using inelastic neutron scattering techniques. Upon adsorption of H2O, the peak at 3610 cm-1 is sharply reduced in intensity (but not totally eliminated), while the peak at 3726 cm-1 is also reduced, although less sharply. This shows that although H2O adsorbs at both types of OH sites, adsorption at the Brønsted acid site predominates. Upon adsorption, six distinctly new features appear in IR difference spectra:10,15 weak peaks at 3695 and 3543 cm-1; a weak shoulder near 3380 cm-1; two broad peaks, centered at 2885 and 2457 cm-1; and a somewhat sharper peak at 1630 cm-1. These peak positions depend slightly on H2O pressure and Si/Al ratio, which explains minor differences in published experimental studies. It seems clear that the features in the 3400-3700 cm-1 range arise from the weakly perturbed OH stretching modes presumably associated with the adsorbate molecule, so the main interpretive task lies in assigning the three lower frequency peaks. Some workers, in support of an ionpair adsorption structure, have claimed that the 2885 (A) and 2457 cm-1 (B) peaks correspond to symmetric and asymmetric OH stretching modes of an adsorbed H3O+, while the 1630 cm-1 (C) feature represents the associated HOH bending mode.10,11 Others, who favor the hydrogen-bonded adsorption structure, believe that these three peaks arise due to Fermi resonances between the broad, downshifted νOH stretching mode of the acid site hydroxyl and overtones of the in-plane δOH and out-ofplane γOH bending modes of the hydroxyl groups (2δ, 2γ)4,13,16 which are shifted to higher frequencies upon adsorption, or an overtone and combination of these modes (2δ, δ + γ).17 They interpret these three peaks as the A,B,C triplet well-known in other hydrogen-bonded systems.39 The IR spectrum of CH3OH and CH3CN adsorbed in H-ZSM-5 also shows these three peaks and they too have been interpreted in terms of Fermi resonances.4,36,40 The calculated vibrational frequencies of the ZH‚‚‚OH2 structure (3 T atom cluster) shown in Table 7 support a modified form of the latter interpretation. We do not consider the ionpair adsorption structure since our calculations show it to be a transition state. The hydrogen-bonded neutral complex has
Zygmunt et al. TABLE 6: Influence of H2O and H3O+ Adsorption on Selected Geometrical Parameters of 3 T and 8 T ZH Clustersa ZH b
r(OzHz) r(SiOz) r(AlOz) ∠SiOzAl q(Oz) q(Hz) q(Si) q(Al) q(ads)d
Z-‚‚‚HOH2+
ZH‚‚‚OH2 c
b
c
3T
8T
3T
8T
0.98 1.72 2.01 128.1 -0.95 0.53 0.93 0.91
0.95 1.65 1.83 131.5 -0.90 0.55 1.69 1.35
1.02 1.71 1.95 126.9 -1.02 0.59 0.94 0.90 0.04
0.98 1.64 1.81 130.0 -0.97 0.62 1.69 1.36 0.04
3 Tb
8 Tc
1.68 1.87 129.8 -1.06
1.57 1.72 133.4 -0.95
0.95 0.91 0.74
1.67 1.36 0.82
a Bond lengths in Å, bond angles in degrees, and atomic charges in units of e. Charges from Mulliken population analysis. b MP2/6-31G(d) level of theory with full optimization. See Figure 1 for labels. c HF/ 6-31G(d) level of theory with CVR optimization. See Figure 2 for labels. d Sum of Mulliken charges of adsorbate atoms.
MP2/6-31G(d) frequencies of 3612 and 3269 cm-1 (scaled41 by 0.9427) which correspond to the two OH stretching frequencies of the adsorbed H2O. These compare reasonably well to the experimental values shown in Table 7. In addition, the MP2/ 6-31G(d) acidic νOH stretching mode has a greatly enhanced intensity (1350 km/mol in ZH‚‚‚OH2 compared to 203 km/mol in ZH) and is strongly downshifted to 2784 cm-1 (from 3480 cm-1 in ZH). This is close to the first overtone of the calculated δOH in-plane bending mode (2 × 1325 cm-1 ) 2650 cm-1) and suggests that a strong Fermi resonance between the two modes might be expected to produce a broad, double-peaked structure in the IR spectrum. This is in good agreement with the two experimental peaks at 2885 and 2457 cm-1, since the minimum between the two peaks is predicted by Fermi resonance theory to occur at the overtone frequency (2650 cm-1) and is observed at roughly 2675 cm-1. This interpretation of these two peaks is in qualitative agreement with previous work.4,13 The third peak at 1630 cm-1 cannot be explained from our calculations in terms of a Fermi resonance shift. The MP2/6-31G(d) γOH out-of-plane frequency is at 1045 cm-1. Both the overtone (2γ ) 2090 cm-1) and the combination modes (δ + γ ) 2370 cm-1) lie at least 400 cm-1 below the calculated νOH stretching frequency, thus making the Fermi resonance explanation for the origin of the experimental IR peak at 1630 cm-1 seem dubious. It should be noted that Sauer’s calculation for the adsorption structure with CH3OH at a similar level of theory (MP2/DZP) gave a scaled νOH frequency of 2548 cm-1, making the Fermi resonance interpretation of the C peak in the IR spectrum seem more plausible for adsorbed CH3OH. We have included Sauer’s4,36 calculated frequencies for the CH3OH adsorption structure in Table 7. The difference between the case of hydrogen bonding in this study and in the systems which exhibit the A,B,C triplet peaks seems to be that the resonant interaction between νOH and the overtone 2γOH is much weaker for the ZH‚‚‚OH2 structure. This behavior is analogous to some hydrogen-bonded systems which exhibit distinct A and B features in their IR spectra, but no clear C peak.39 A detailed theoretical analysis of the vibrational spectrum of the hydrogen-bonded crystal CsHSeO4 suggests that the C peak in some hydrogen-bonded solids is due to a very weak Fermi resonance between νOH and the combination band γOH + νTO, where νTO is a lattice mode of the solid with the proper symmetry.42 The combination frequency γOH + νTO is very close to the observed C peak position because the resonance is weak. Pelmenschikov et al.43 recognized this possibility in their study of CD3CN adsorption in H-ZSM-5, and it may also play a role in H2O adsorption.
Water Adsorption in the Zeolite H-ZSM-5
J. Phys. Chem., Vol. 100, No. 16, 1996 6669
TABLE 7: Vibrational Frequencies (cm-1) for ZH‚‚‚OH2/CH3OH from IR Spectroscopy and ab Initio Calculations assignment
this worka (ZH‚‚‚OH2)
Sauerb (ZH..CH3OH)
asym OH stretch (H2O)
3612 (112)
sym OH stretch (H2O) HOH bend (H2O) νOH (ZH)
3269 (596) 1636 (46) 2784 (1350)
2548
δOH (ZH)
1325 (224)
1353
γOH (ZH)
1045 (22)
1015
expt.c (ZH‚‚‚OH2)
expt.d (ZH‚‚‚OH2)
3695 3543 3380 1630
3700 ∼3600
2885f
2900f
2457f
2470f
e
1620
a Obtained from MP2/6-31G(d) results using scale factor of 0.9427; intensities (km/mol) in parentheses. b References 4 and 36. Obtained from MP2/DZP results using scale factor of 0.954 . cReference 8. d Reference 15. e A broad shoulder was found in this region but no peak frequency was reported. f We assign these peaks to Fermi resonance between νOH fundamental and 2δOH overtone modes.
A more likely explanation for the 1630 cm-1 feature in the H2O structure is provided by the MP2/6-31G(d) HOH bending mode of the adsorbed H2O at 1636 cm-1 (see Table 6). The experimental peak at 1630 cm-1 is narrower than the two peaks at 2885 and 2457 cm-1, which indicates that the effect of Fermi resonance, if any, is slight. In addition, the experimental difference spectra at very low H2O pressure (10-4-10-5 mbar) show the presence of the 1630 cm-1 feature eVen before the peaks at 2885 and 2457 cm-1 become apparent.10 Thus, this feature may be an IR-active mode associated with the H2O molecule which is only weakly perturbed upon adsorption, consistent with the behavior of the MP2/6-31G(d) HOH bending mode, which is shifted by less than 1 cm-1 compared to the same mode for a free H2O molecule. Additional support for this interpretation is provided by the results of a very thorough IR spectroscopic study by Parker et al.15 of H2O and D2O adsorption on H-ZSM-5 and D-ZSM-5. The peak in question, measured at 1620 cm-1 for H2O adsorbed on H-ZSM-5, persisted with decreased intensity for H2O adsorbed on D-ZSM-5, was barely noticeable for D2O adsorbed on H-ZSM-5, and was completely absent for D2O adsorbed on D-ZSM-5. Thus, the peak intensity decreased as the deuteration of the framework and adsorbate increased. The authors of this study also noted the appearance of a peak at 1426 cm-1 as deuteration caused the peak at 1620 cm-1 to weaken and pointed out that these values are close to the experimental values for the HOH (1611 cm-1) and DOH (1398 cm-1) deformation modes determined in an IR study of matrix-isolated water dimers.44 The behavior of these two peaks is evidence for our assignment of this mode. In addition, the factor of 30 difference between our calculated intensities for the 1636 and 2784 cm-1 vibrations (see Table 7) is qualitatively consistent with the IR spectra presented in ref 15. While the MP2/6-31G(d) vibrational frequencies of the ZH‚‚‚OH2 structure are consistent with IR difference spectra measured at low H2O pressures, the variable-pressure IR study of ref 10 showed dramatic changes in the IR difference spectra for H2O equilibrium pressures above 10-2 mbar. In order to investigate the effect of increased pressure on the equilibrium structure of the H2O adsorption complex, we added a second H2O molecule to the 3 T Z-‚‚‚HOH2+ structure and performed a full geometry optimization. While the Z-‚‚‚HOH2+ ion-pair structure is a transition state, our calculations at both HF/631G(d) and MP2/6-31G(d) levels of theory show that the Z-‚‚‚H(OH2)2+ structure, shown in Figure 3, is a local minimum. The presence of a second H2O molecule apparently stabilizes the ion-pair structure in which a proton has been transferred from the acid site to the adsorbate. This is consistent with the fact that the water dimer has an HF/6-31G(d) proton affinity at 0 K of 194.8 kcal/mol, significantly greater than the value of 167.1 kcal/mol for the isolated water molecule. Thus, for
Figure 3. Z-‚‚‚H(OH2)2+ complex for 3 T cluster. Hydrogen bond distances at MP2/6-31G(d) level of theory are shown.
coverages greater than one molecule per acid site, this result suggests that Z-‚‚‚H(OH2)2+, or larger Z-‚‚‚H(H2O)n+ ion-pair complexes, may be more likely to form. The results of vibrational frequency calculations for the Z-‚‚‚H(OH2)2+ structure help to explain the observed changes in the IR difference spectrum of H-ZSM-5 as H2O pressure increases.10 As the pressure increases from 10-2 to 1 mbar, the Fermi resonance peaks at 2885 and 2457 cm-1 disappear, and a very broad and intense peak appears near 3208 cm-1 (with shoulders near 3300 and 3000 cm-1), along with a peak at 1700 cm-1. These spectral changes reflect a significant structural change in the H2O adsorption complex. The calculation for the Z-‚‚‚H(OH2)2+ structure gives scaled MP2/6-31G(d) frequencies of 1734 and 2913 cm-1 but provided no evidence supporting Fermi resonance in the region 2400-2900 cm-1. The 2913 cm-1 frequency has a large absorption intensity. The 1734 cm-1 frequency is associated with an HOH bending mode of the H3O fragment of the H(OH2)2+ ion, while the 2913 cm-1 frequency corresponds to a perturbed OH stretching mode on the same fragment. The calculation also yielded two frequencies at 2305 and 2562 cm-1, corresponding to asymmetric and symmetric OH stretching modes, respectively, on the H3O fragment of the adsorbed H(OH2)2+ ion. These modes involve the protons hydrogen bonded to the zeolite and have large absorption intensities but did not appear as peaks in the experimental spectra of ref 10. One possible explanation for this discrepancy is that larger (n > 2) Z-‚‚‚H(OH2)n+ ion-pair complexes, which we have not investigated, may predominate in H-ZSM-5 at 1 mbar of H2O pressure. This possibility has also been mentioned by Jentys et al.10 If correct, this could also explain the difference between the calculated frequency of 2913 cm-1 and the experimental peak at 3208 cm-1. Another possible explanation for the discrepancies between the theoreti-
6670 J. Phys. Chem., Vol. 100, No. 16, 1996 cal and experimental frequencies at higher pressures is that the 3 T cluster may not be large enough for an adsorbate as large as H(OH2)2+. Evidence for this is that the hydrogen attached to the Al in the 3 T cluster is only 1.84 Å away from one of the terminal hydrogens of H(OH2)2+, suggesting that it may be triply-anchored to the zeolite framework if the hydrogen were replaced by an oxygen. In future studies we will examine Z-‚‚‚H(OH2)2+ as well as the corresponding neutral structure ZH‚‚‚(OH2)2 in larger clusters. The 1H wide-line NMR spectra measured at 4 K have been used by Batamack et al. to discriminate the species existing in zeolite H-ZSM-5 and other solid acids upon water adsorption.6,7 For H-ZSM-5, at loadings up to one H2O per hydroxyl bridge, they concluded that the neutral hydrogen-bonded structure for the adsorbed water is the minimum energy state, in accord with the results obtained in our calculations. However, they also deduced that there was a coexisting population (20%) of species with the ionic structure and that the concentrations of the different species were virtually independent of temperature. The latter result implies that the energy difference between the ionpair and neutral species is extremely small. This is consistent with our estimate based on the 3 T cluster showing essential degeneracy of the ion-pair and neutral species, but the 5.0 kcal/ mol energy difference estimate based on the larger 8 T cluster would be too large to accommodate this interpretation of the NMR results. We consider the 5.0 kcal/mol to be an upper limit on ∆Erel and allow that the value could be very small. The calculations do show, however, that the ion-pair structure is a transition state and not a local minimum. We can suggest two alternative rationalizations for the NMR observations. First, we note that Batamack et al.6 also found that the relative fraction of the ion-pair species increased with increasing water loading for their H-ZSM-5 sample with the fewest defects. It is possible that the species identified as the ion-pair species in the 1H wide-line NMR spectra measured at 4 K is actually the Z-‚‚‚H(OH2)2+ species stabilized by two water molecules as found in our calculations (or even the higher complexes for more highly hydrated systems). This also implies that dual hydration, i.e., two waters per acid site, is occurring at loadings below one H2O per bridging hydroxyl group in the experiments, although the reason for this is not clear at this time. In future work we will investigate the energy of the interaction of the acid site and a water dimer and compare it to the ion-pair state in clusters larger than 3 T to provide insight into this matter. The alternative explanation relies on the recognition that there can be inequivalent Brønsted acid sites in the zeolite. The calculations were performed specifically for protonation at the T(12)-O(24)-T(12) site of the ZSM-5 zeolite framework (MFI structure). We have previously demonstrated that the proton affinity is site-dependent, so the energy difference between the ionic and neutral structures for the adsorbed water is expected to be site-dependent also. It is therefore possible that at a crystallographically inequivalent Brønsted acid site the relative stability of the neutral and ionic structures is reversed. The fraction of these sites manifested in the NMR spectra could then reflect their fractional population (occupancy) in the zeolite sample and would in fact be temperature invariant. The argument presented in the previous paragraph is one of heterogeneity in the Brønsted acid sites. This could be a sampledependent property, and we note that the reported species distribution as a function of water loading differed for H-ZSM-5 samples with different defect concentrations. Although we cannot address this result directly in the context of our calculations, it is clear that defects present an additional basis
Zygmunt et al. for heterogeneity in the Brønsted acid sites beyond the intrinsic heterogeneity presented by crystallographic site inequivalence. IV. Conclusions Using aluminosilicate cluster models of up to 100 atoms, we have carried out an ab initio molecular orbital study of the adsorption of H2O at an acid site in the H-ZSM-5 zeolite. The following conclusions can be drawn from this study: 1. The ion-pair structure is a transition state and the neutral structure is a local minimum from calculations on the 3 T atom cluster at the MP2/6-31G(d) level. The energy difference represents the barrier for proton transfer between oxygen sites. Calculations on a larger 8 T atom cluster representing the O(24) site, with local geometry relaxation included, also suggest that it is a transition state, although these calculations were done at the HF/6-31G(d) level and did not include internal relaxation of the H3O+. 2. Calculations including effects of larger cluster size, larger basis sets, zero-point energies, and higher level of electron correlation were used to obtain more accurate values for the water desorption energy and the barrier. The calculations give a water desorption energy for the O(24) site of 13-14 kcal/ mol. This is in agreement with the published experimental value for water desorption in ZSM-5. The calculations give a small proton transfer barrier (less than 5 kcal/mol and possibly close to zero). This is consistent with the observed ease of H/D exchange between the acid site and adsorbed D2O. 3. In calculating the desorption energy and the barrier, the correlation effect is found to be very important (7-8 kcal/mol), and most of it is accounted for by the MP2 theory which differs by less than 1 kcal/mol from the more sophisticated QCISD(T) technique. 4. The calculated vibrational frequencies of the H2O adsorption structure in the 3 T atom cluster are able to account for the main features of the experimental IR spectrum and provide evidence for a slightly different interpretation of the spectrum than previously proposed. 5. While the the ion-pair structure is a transition state when only one H2O molecule interacts with the Brønsted acid site in the 3 T cluster, the addition of a second H2O molecule to the first H2O molecule stabilizes the ion-pair structure, making it a local minimum. Calculated vibrational frequencies of the resulting equilibrium ion-pair structure, Z-‚‚‚H(OH2)2+, show significant changes from those of the ZH‚‚‚OH2 structure that are consistent with experimental IR spectra at pressures above 10-2 mbar. This Z-‚‚‚H(OH2)2+ species is also invoked in one possible reinterpretation of published low-temperature 1H wideline NMR spectra. An alternative interpretation based on heterogeneity of the Brønsted acid sites is also presented. Note Added in Proof. After the manuscript was submitted for publication, experimental results were published (Smith, A. K. C.; et al. 1996, 271, 799) which give crystallographic evidence for the coexistence of H2O and H3O+ adsorbates in H-SAPO-34. Their result is consistent with our alternative interpretation of the 1H-NMR data presented in section III.C, although we reiterate that since proton affinities may vary among crystallographically distinct Bronsted acid sites in a particular structure, there is no a priori reason for the H2O binding distribution to be the same in a zeolite and a SAPO material. Acknowledgment. This work was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences, under Contract No. W-31-109ENG-38. We acknowledge a grant of computer time at the
Water Adsorption in the Zeolite H-ZSM-5 National Energy Research Supercomputer Center. Acknowledgment is also made to the donors of the Petroleum Research Fund, adminstered by the American Chemical Society, for partial support of this research (S.A.Z. and M.K.E). References and Notes (1) Brand, H. V.; Curtiss, L. A.; Iton, L. E. J. Phys. Chem. 1993, 97, 12773. (2) Brand, H. V.; Curtiss, L. A.; Iton, L. E. J. Phys. Chem. 1992, 96, 7725. (3) Teunissen, E. H.; Jansen, A. P.; van Santen, R. A. J. Phys. Chem. 1995, 99, 1873. (4) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders, V. R. Chem. ReV. 1994, 94, 2095. (5) Kassab, E.; Fouquet, J.; Allavena, M.; Allavena, E. M. J. Phys. Chem. 1993, 97, 9034. (6) Batamack, P.; Doremieux-Morin, C.; Fraissard, J.; Freude, D. J. Phys. Chem. 1991, 95, 3790. (7) Batamack P.; Doremieux-Morin, C.; Vincent, R.; Fraissard, J. Chem. Phys. Lett. 1991, 180, 545. (8) Ison, A.; Gorte, R. J. J. Catal. 1984, 89, 150. (9) Aronson, M. T.; Gorte, R. J.; Farneth, W. E. J. Catal. 1987, 105, 455. (10) Jentys, A.; Warecka, G.; Derewinski, M.; Lercher, J. J. Phys. Chem. 1989, 93, 4837. (11) Marchese, L.; Chen, J.; Wright, P.; Thomas, J. J. Phys. Chem. 1993, 97, 8109. (12) Sauer, J. C.; Ko¨lmel, C.; Haase, F.; Ahlrichs, R. In Proceedings of the Ninth International Zeolite Conference, Montreal, 1992; von Ballmoos, R., Higgins, J. B., Treacy, M. M., Eds.; Reed Publishing: Stoneham, MA, 1993; Vol. I, p 679. (13) Pelmenschikov, A. G.; van Santen, R. A. J. Phys. Chem. 1993, 97, 10678. (14) Haase, F.; Sauer, J. J. Phys. Chem. 1994, 98, 3083. (15) Parker, L. M.; Bibby, D. M.; Burns, G. R. Zeolites 1993, 13, 107. (16) Zecchina, A.; Buzzoni, R.; Bordiga, S.; Geobaldo, F.; Scarano, D.; Ricchiardi, G.; Spoto, G. In Zeolites: A Refined Tool for Designing Catalytic Sites, Proceedings of the International Zeolite Symposium, Quebec 1995; Bonneviot, L., Kaliaguine, S., Eds.; Elsevier: Amsterdam, 1995; p 213. (17) van Santen, R. A.; Kramer, G. J. Chem. ReV. 1995, 92, 637. (18) Zygmunt, S. A.; Brand, H. V.; Lucas, D. J.; Iton, L. E.; Curtiss, L. A. J. Mol. Struct. (THEOCHEM) 1994, 314, 113. (19) Zygmunt, S. A.; Curtiss, L. A.; Iton, L. E. In Zeolites: A Refined Tool for Designing Catalytic Sites, Proceedings of the International Zeolite Symposium, Quebec 1995; Bonneviot, L., Kaliaguine, S., Eds.; Elsevier: Amsterdam, 1995; p 101.
J. Phys. Chem., Vol. 100, No. 16, 1996 6671 (20) Hehre, W. J.; Radom, L.; Pople, J. A.; Schleyer, P. v. R. Ab Initio Molecular Orbital Theory; John Wiley: New York, 1987. (21) Frisch, M. J.; Trucks, G. W.; Head-Gordon, M.; Gill, P. M. W.; Wong, M. W.; Foresman, J. B.; Johnson, B. G.; Schlegel, H. B.; Robb, M. A.; Replogle, E. S.; Gomperts, R.; Andres, J. L.; Raghavachari, K.; Binkley, J. S.; Gonzalez, C.; Martin, R. L.; Fox, D. J.; Defrees, D. J.; Baker, J.; Stewart, J. J. P.; Pople, J. A. Gaussian 92, ReVision G.4; Gaussian, Inc.; Pittsburgh, PA, 1992. (22) Teunissen, E. H.; Jansen, A. P. J.; van Santen, R. A.; Orlando, R.; Dovesi, R. J. Chem. Phys. 1994, 101, 5865. (23) Greatbanks, S. P.; Sherwood, P.; Hillier, I. H.; Hall, R. J.; Burton, N. A.; Gould, I. R. Chem. Phys. Lett. 1995, 234, 367. (24) van Koningsveld, H.; Van Bekkum, H.; Jansen, J. C. Acta Crystallogr. 1987, B43, 127. (25) Redondo, A.; Hay, P. J. J. Phys. Chem. 1993, 97, 11754. (26) Cook, S. J.; Chakraborty, A. K.; Bell, A. T.; Theodorou, D. N. J. Phys. Chem. 1993, 97, 6679. (27) Smith, B. J.; Radom, L. J. Phys. Chem., 1995, 99, 6468. (28) Curtiss, L. A.; Redfern, P. C.; Smith, B. J.; Radom, L. J. Chem. Phys., in press. (29) Curtiss, L. A.; Raghavachari, K.; Pople, J. A. J. Phys. Chem. 1993, 98, 1293. (30) Pople, J. A.; Head-Gordon, M.; Raghavachari, K. J. Phys. Chem. 1987, 87, 5968. (31) Curtiss, L. A.; Raghavachari, K.; Trucks, G. W.; Pople, J. A. J. Chem. Phys. 1991, 94, 7221. (32) Del Bene, J. E.; Mettee, H. D.; Frisch, M. J.; Luke, B. T.; Pople, J. A. J. Phys. Chem. 1983, 87, 3279. (33) Kassab, E.; Seiti, K.; Allavena, M. J. Phys. Chem. 1990, 95, 9425. (34) Grady, M. C.; Gorte, R. J. J. Phys. Chem. 1985, 89, 1305. (35) Sauer, J.; Horn, H.; Haser, M.; Ahlrichs, R. Chem. Phys. Lett. 1990, 173, 26. (36) Haase, F.; Sauer, J. J. Am. Chem. Soc. 1995, 117, 3780. (37) Bates, S.; Dwyer, J. J. Mol. Struct. (THEOCHEM) 1994, 306, 57. (38) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88, 899. (39) Claydon, M. F.; Sheppard, N. Chem. Commun. 1969, 1431. (40) Kubelkova, L.; Kotrla, J.; Florian, J. J. Phys. Chem. 1995, 99, 10285. (41) Pople, J. A.; Scott, A. P.; Wong, M. W.; Radom, L. Isr. J. Chem. 1993, 33, 345. (42) Ratajczak, H.; Yaremko, A. M.; Baran, J. J. Mol. Struct. 1992, 275, 235. (43) Pelmenschikov, A. G.; van Santen, R. A.; Ja¨nchen, J.; Meijer, E. J. Phys. Chem. 1993, 97, 11071. (44) Ayers, G. P.; Pullin, A. D. E. Spectrochim. Acta 1976, 32A, 1641.
JP952913Z
