ARTICLE pubs.acs.org/JPCA
Location of Si Vacancies and [Ti(OSi)4] and [Ti(OSi)3OH] Sites in the MFI Framework: A Large Cluster and Full Ab Initio Study Shuping Yuan,*,† Hongzong Si,† Aiping Fu,† Tianshu Chu,† Fenghui Tian,† Yun-Bo Duan,† and Jianguo Wang‡ †
Institute for Computational Science and Engineering, Laboratory of New Fiber Materials and Modern Textile, the Growing Base for State Key Laboratory, Qingdao University, Qingdao, Shandong 266071, P.R. China ‡ State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of Sciences, P.O. Box 165, Taiyuan 030001, P. R. China ABSTRACT: Titanium silicalite-1 (TS-1) is an important catalyst for selective oxidation reactions. However, the nature and structure of the active sites and the mechanistic details of the catalytic reactions over TS-1 have not been well-understood, leaving a continuous debate on the genesis of active sites on the TS-1 surface in the literature. In this work, the location of Si vacancies and [Ti(OSi)4] and [Ti(OSi)3OH] sites in the MFI (Framework Type Code of ZSM-5 (Zeolite Socony Mobile-Five)) framework has been studied using a full ab initio method with 40T clusters with a Si:Ti molar ratio of 39:1. It was shown that the former four energetically favorable sites for Si vacancies are T6, T12, T4, and T8 and for Ti centers of [Ti(OSi)4] are T10, T4, T8 and T11, being partially the same sites. Whether by replacing Si vacancies or substituting the fully coordinated Si sites, the most preferential site for Ti is T10, which indicates that the insertion mechanism does not affect the favorable sites of Ti in the MFI lattice. For the defective [Ti(OSi)3OH] sites, it was found that the Si vacancy at T6 with a Ti at its neighboring T9 site (T6-def-T9-Ti pair) is the most energetically favorable one, followed by a T6-def-T5-Ti pair with a small energy gap. These findings are significant to elucidate the nature of the active sites and the mechanism of reactions catalyzed by TS-1 and to design the TS-1 catalyst.
absorption fine structure (EXAFS) and X-ray absorption near-edge spectroscopy (XANES) studies of Bordiga et al.19,20 The formation of [TiO4] units in the MFI was also confirmed by several theoretical studies.21-23 In the MFI framework there are 12 crystallographically distinct tetrahedral sites (T sites) in the unit cell, which we will refer to as T1-T12. Various experimental techniques have been applied to investigate the distribution of Ti centers in the MFI lattice. The fact that only a very small amount of Ti, less than 3 wt % in TiO2, corresponding to a Si:Ti molar ratio of 39:1,24 can be incorporated in the MFI framework does not facilitate the extraction of the Ti contribution from the total experimental signal. Therefore, the experimental results from different groups are rather conflicting. In the year 2000, Marra et al.25 attempted to use low temperature X-ray diffraction (XRD) technique to determine the siting of Ti in TS-1; the results showed that the titanium has a preferential tendency to occupy sites T10 and T11 and to escape from T4 and T12. After one year, the same authors reported a direct experimental evidence26 for Ti siting in the MFI framework by powder neutron diffraction. It was pointed out that T6, T7, and T11 are the most populated sites, and T10 is also supported by weak evidence. Almost simultaneously, Hijar et al.27 and Henry et al.28 reported a nonrandom partitioning of Ti
1. INTRODUCTION Because of its excellent performance as catalyst in selective oxidation reactions with 30% aqueous H2O2 as an oxidant agent, titanium silicalite-1 (TS-1) has attracted much attention since 1983, when it was first synthesized by Taramasso et al.1 The catalysis of many transformations by TS-1 has been reported, which include aromatic hydroxylation,2 alkane oxidation,3 olefin epoxidation,4,5 cyclohexanone ammoximation,6,7 oxidation of ammonia to hydroxylamine,8,9 and the conversion of secondary amines to dialkylhydroxylamines10 and secondary alcohols to ketones.11 Among these reactions, the hydroxylation of phenol and the ammoximation of cyclohexanone to cyclohexanonoxime have been applied successfully in industrial processes.12 Numerous experimental and theoretical studies have been done to elucidate the structure of TS-1, the nature of the active species for the oxidation reaction,13-15 and the reaction mechanisms. Most of these works have been summarized in recent reviews.12,16 However, debates still exist on the structure and catalytic details of this catalyst. The catalytic active centers of TS-1 are now known to be Ti sites internal to the crystallite, which are, ideally, fully tetrahedrally bonded in the TS-1 lattice.17,18 The direct evidence for Ti incorporation in the MFI (Framework Type Code of ZSM-5 (Zeolite Socony Mobile-Five)) framework by replacing Si atoms in the tetrahedral positions was first presented by extended X-ray r 2011 American Chemical Society
Received: November 17, 2010 Revised: December 14, 2010 Published: January 10, 2011 940
dx.doi.org/10.1021/jp110977m | J. Phys. Chem. A 2011, 115, 940–947
The Journal of Physical Chemistry A
ARTICLE
among the 12 independent T sites of the MFI framework by powder neutron diffraction. Hijar et al. found that titanium is distributed among T3, T7, T8, T10, and T12, while Henry et al. suggested T8, T10, and T3 as the Ti preferentially substitution sites. Obviously, different experimental studies provide different selections of preferred sites for Ti substitution on the T sites of MFI. Compared with experimental works, theoretical studies on exploring the location of Ti centers in TS-1 began somewhat earlier.21,29-34 Millini et al.21 reported a local density approximation (LDA) study on Ti distribution in TS-1 in 1994. It was found that Ti atoms preferentially occupy T12, T3, T10, and T11 sites. Njo et al.29 reported a Metropolis Monte Carlo and molecular mechanics study on the same issue, and sites T2 and T12 were indicated to be populated by Ti. The molecular dynamics simulation by Oumi et al.30 identified T8 as the most preferred Ti site. By using a molecular mechanics force-field method, Sastre and Corma31 also confirmed T8 as the most favored Ti site, followed by T9, T7, and T6. Atoguchi and Yao32 carried out quantum mechanics (QM)/molecular mechanics (MM) calculations on a cluster containing all of the 12 distinct T sites and found T9 and T10 to be the preferred Ti sites. Recently, an embedded-cluster study identified T12 and T2 as the most and least preferred sites.33 More recently, a periodic density functional study on the siting of Ti within TS-1 was carried out by Gale;34 the favorable sites were proposed to be T10, T8, T4, and T11, which is in somewhat agreement with the experimental diffraction results of Hijar et al.27 and Henry et al.28 Up to now, in the computational studies, the preferential Ti sites in the MFI lattice were determined on the basis of the substitution reaction that Ti atoms replace fully coordinated silicon atoms in the MFI structure. The neutron powder diffraction studies by Lamberti et al.,26 Bordiga et al.,35 and Artioli et al.36 showed that the preferential location of Si vacancies in a Ti-free silicalite was in the same sites (T6, T7, T10, and T11) as that of Ti in TS-1. Combined with the finding by IR and other experiments19 that the insertion of Ti in the MFI framework progressively reduced the Si vacancies, it was suggested that the incorporation of the Ti atoms in the MFI lattice occurs via the insertion of titanium in defective sites of silicalite-1.12,26 While an improved diffraction study by Henry et al.28 showed that Si vacancies in TS-1 are favored at T1 and T5 sites, which are different from Ti sites (T8, T10, and T3). Thus, it is necessary to elucidate whether the incorporation of Ti in the MFI lattice occurs via the insertion of Ti in Si vacancies or in the fully coordinated Si sites. It is actually well-known that TS-1 samples, synthesized following the original patent,1 are rather defective materials showing internal defects. The characterization techniques have evidenced the coexistence of both regular [Ti(OSi)4] and defective [Ti(OSi)3OH] sites in TS-1 samples.12,26 This stimulated Wells et al.37 to carry out a density functional theory study of the propylene epoxidation mechanisms on both regular [Ti(OSi)4] and defective [Ti(OSi)3OH] sites. Therefore, the location of the defective [Ti(OSi)3OH] sites in TS-1 is as important as that of [Ti(OSi)4] sites. On the basis of the above-described background, this work investigated the siting of Si vacancies and the regular [Ti(OSi)4] and defective [Ti(OSi)3OH] sites in the MFI framework by using a full ab initio method with 40T clusters with a Si:Ti molar ratio of 39:1. The formation of the [Ti(OSi)4] sites by Ti insertion in fully coordinated Si sites as well as in Si vacancies was simultaneously investigated.
Figure 1. 17T clusters centered on the T12 site (1-a) and 17-2T; 17-3T and 17-4T model schemes with the free atoms shown in ball and stick and the fixed atoms in line during the structure optimizations (1-b, 1-c, and 1-d).
2. COMPUTATIONAL DETAILS Models. Although periodic calculations can treat the crystalline of zeolites accurately, it is limited for zeolites with large unit cells such as TS-1 due to the computational costs. Thus, small cluster models or larger clusters with hybrid methods have been frequently used for TS-1. In this work clusters with different sizes are employed. The structure of the clusters are taken from the lattice structure of Na-ZSM-5, in which there are 12 crystallographically distinct T sites (Tx, x = 1, 2, ..., 12) in the unit cell.38 Clusters containing 17 and 40 tetrahedral sites (17T and 40T, see Figures 1 and 2) centered at each of the 12 independent T sites are used to calculate the substitution energies of Si by Ti and to investigate the effect of the cluster sizes on the results. The 17T clusters include the first to fourth nearest neighboring shells of the central T atom, and the 40T clusters are obtained by enlarging the 17T models to the sixth nearest neighboring shell of the central T atom. For comparison, substitution energies on 5T clusters (Figure 3), which were often used in the past to investigate the isomorphous substitution of metal atom such as aluminum in the zeolite framework,39 are also calculated on each site. Each peripheral silicon atoms in the clusters is saturated by one or two terminal hydrogen atoms. The distance between the hydrogen atom and the corresponding silicon atom is 1.47 Å, and the orientation of Si-H bonds is along the pre-existing Si-O bonds. Methods. All of the calculations in this work are carried out by using the density functional method and Gaussian 03 package.40 The B3LYP hybride functional and polarized 6-31G(d, p) basis set, which have been widely applied in the studies of Ti-containing zeolites, were employed throughout. The substitution energy (SE), which corresponds to the energy change of the substitution reactions, has been frequently 941
dx.doi.org/10.1021/jp110977m |J. Phys. Chem. A 2011, 115, 940–947
The Journal of Physical Chemistry A
ARTICLE
Table 1. RSE (kcal/mol) for the Formation of [Ti(OSi)4] Sites by Ti Replacing a Fully Tetrahedral Si Site (Reaction-1) on the 12 Different T Sites of MFI Calculated by Different Models 5-2T
17-2T
17-3T
17-4T
40-2T
40-3T
40-4T
T1
3.50
24.77
14.61
11.98
4.94
7.95
8.20
T2
5.40
27.51
11.63
9.80
7.70
5.90
6.00
T3
5.69
29.30
17.43
14.40
9.43
11.59
11.18
T4
4.60
27.75
7.66
6.14
7.47
2.20
1.70
T5
4.15
25.98
16.07
14.00
6.08
10.11
10.62
T6
4.22
27.25
17.87
14.88
7.31
11.24
11.10
T7 T8
5.89 6.06
28.09 28.73
13.34 9.67
12.28 7.71
8.51 8.82
7.69 3.18
8.09 2.56
T9
3.50
22.22
9.92
9.79
5.06
6.25
7.40
T10a
0.00
0.00
0.00
0.00
0.00
0.00
0.00
T11
3.71
25.75
8.44
7.31
5.78
2.77
2.90
T12
4.66
26.75
16.08
14.38
6.58
10.08
10.59
a The SE for the T10 site is -81.92, -92.84, -94.18, -96.01, -72.95, -85.30, and -85.88 kcal/mol.
example, in Table 1), is used to directly measure the energetically favorable sites and the SE gap between different sites. The RSE could be calculated by: RSE ¼ ½EðTx-Ti-clusterÞ - EðTx-Si-clusterÞ
Figure 2. 40T clusters centered on the T12 site (2-a) and 40-2T; 40-3T and 40-4T model schemes with the free atoms shown in ball and stick and the fixed atoms in line during the structure optimizations (2-b, 2-c, and 2-d).
- ½EðT10-Ti-clusterÞ - EðT10-Si-clusterÞ ðFormula-2Þ It is notable that all energies reported do not include corrections for zero point energy due to their partial optimizations and basis-set-superposition-error (BSSE) which is expected to be roughly the same for each of the clusters examined and cancel in large extent in the SE and RSE calculation. It is found that full relaxation of zeolite clusters leads to structures that do not resemble acid sites in real zeolitic lattices.41 To model the “rigidity” of the zeolite lattice, some atoms were fixed at their experimental positions during optimization. Therefore, in this work, partial optimizations have been done for each cluster. For the smallest 5T clusters, only the terminal hydrogen atoms were fixed during optimization. With respect to the 17T and 40T models, different constraint schemes were adopted to test the region of the structural distortion resulting from Ti substitution on the zeolite framework.
Figure 3. 5T clusters centered on the T12 site (with Si or Ti on T12 site; for 3-a and 3-b).
used to measure the energetic preferential sites of heteroatoms in the zeolite framework. For Ti substituting a fully coordinated silicon atom in the MFI structure, the substitution reaction is:
3. RESULTS AND DISCUSSION In the first part, the preferential [Ti(OSi)4] sites in the MFI structure were determined on the basis of the substitution reaction of Reaction-1. It is obvious that larger clusters with more atoms relaxed during optimizations are better in accounting for long-range effects than small clusters. To find an appropriate model scheme, the effects of the cluster sizes and the constraint schemes on the calculated SE were investigated in detail.
Si-cluster þ TiðOHÞ4 f Ti-cluster þ SiðOHÞ4 ðReaction-1Þ and the SE is calculated by: SE ¼ ½EðTi-clusterÞ þ EðSiðOHÞ4 Þ - ½EðSi-clusterÞ þ EðTiðOHÞ4 Þ
ðFormula-1Þ
where the Si-cluster refers to clusters centered at each of the 12 dinstinct T sites and the Ti-cluster refers to the corresponding model with the central Si site substituted by a Ti atom; the E(Sicluster) or E(Ti-cluster) refers to the total electronic energy of the optimized Si or Ti centered clusters. In this work, the relative SE (RSE), which is defined as the energy difference between SE of each site (Tx, x = 1, 2, ..., 12) and the lowest SE site (T10, for
3.1. Preferential [Ti(OSi)4] Sites and Validation of Models and Methods. 3.1.1. Constraint Schemes and Model Sizes.
For zeolite clusters, to model the “rigidity” of the zeolite lattice, some atoms would be fixed at their experimental positions during optimization. In the present work, three constraint schemes with different sizes of region relaxed during optimizations were investigated. Table 1 presented the calculated RSE
942
dx.doi.org/10.1021/jp110977m |J. Phys. Chem. A 2011, 115, 940–947
The Journal of Physical Chemistry A
ARTICLE
(by Formula-2) for Reaction-1 on the 12 independent sites of the MFI. The 17-2T in Table 1 and Figure 1 refers to the constraint scheme that in the 17T clusters, the four nearest oxygen atoms and the second four nearest silicon atoms of each central T sites as well as the central T atom itself, totally 9 atoms, were relaxed during optimizations, while other atoms were fixed; and the 17-3T or 17-4T refers to the constraint scheme that in the 17T clusters the relaxed regions are extended to the third or fourth nearest neighboring shell of the central T atom, with totally 21 or 33 atoms being relaxed during optimizations. It is the same nomenclature for 5-2T, 40-2T, 40-3T, and 40-4T in Table 1 and Figures 2 and 3. Calculations for RSE of Reaction-1 on the 17T models were carried out with the three above-mentioned constraint schemes, 17-2T, 17-3T, and 17-4T. The results are presented in Table 1. It is shown that Reaction-1 is exothermically, which is in good agreement with other theoretical works.42 In addition, for all of the employed constraint schemes, the calculated lowest RSE is found in the T10 site, which suggests that T10 is the most favorable site for Ti substitution in the MFI structure. However, the most unfavorable site and the orders of the whole series for the 12 distinct sites are not consistent with each other. For 17-2T, the order of preference is T10 > T9 > T1 > T11 > T5 > T12 > T6 > T2 > T4 > T7 > T8 > T3, while for 17-3T the order showed a large difference: T10 > T4 > T11 > T8 > T9 > T2 > T7 > T1 > T5 > T12 > T3 > T6. With respect to the preferential order from 174T (T10 > T4 > T11 > T8 > T9 > T2 > T1 > T7 > T5 > T12 > T3 > T6), it shows little difference from that of the 17-3T in the middle of the sequence, while the most and least favored sites are very consistent. This indicates that the distortion of the zeolite structure by Ti insertion would spread to at least the third nearest neighboring shell of the Ti site. Therefore, it is important to allow more atoms relaxed during structure optimizations, which are, as we suggested, the first to fourth nearest neighboring shells of the central Ti. Results calculated from 5T clusters are also shown in Table 1 for comparison. It can be seen that the preferential sequence from 5-2T scheme is different from 17-2T, although the same atoms are relaxed in these two model schemes. This indicates that, in addition to the constraint schemes, the model sizes have important effect on calculated SE. It has been shown that, for Al(III) with a comparative ionic radius as Si(IV), 5T could give reliable results.39 Meanwhile with respect to Ti(IV) which has a much larger radius than Si(IV), its incorporation in the zeolite lattice led to a remote spread of the distortion in the zeolite structure. Thus, larger clusters are required to model the structural distortion resulting from the replacement of Si by Ti. As a result, 40T models are employed. There are two reasons to choose the 40T models: first, they contain more neighboring shells of the center T site (up to the sixth neighboring shell); second, in these models when a Si is replaced by a Ti, the molar Si:Ti ratio meets the least Si:Ti ratio of 39:1 of this zeolite. The RSE data for Reaction-1 calculated from the 40-4T model scheme are also shown in Table 1. It is indicated that the preferential order for Ti siting is: T10 > T4 > T8 > T11 > T2 > T9 > T7 > T1 > T12 > T5 > T6 > T3, which showed a little difference to the order of 17-4T, with a large energy gap between T2 and T11. This indicates that Ti centers would preferentially occupy sites T10, T4, T8, and T11. In addition, by comparing the sequences of the RSE calculated from the three employed constraint schemes on the 40T models, that is, 40-2T, 40-3T, and 40-4T, it can be found that 40-3T shows a large difference
from 40-2T, while 40-4T differs a little from 40-3T (only in the order of T8 and T11). This indicates that the RSE is converged with a 40-4T model scheme. The periodic approach is, when applicable, certainly the most accurate method and the only one able to account for the longrange effect.12 But it is limited for zeolites with large unit cells such as MFI due to the computational costs. The preferential Ti sites in the present work are in good agreement with one full ab initio periodic work on the siting of Ti in TS-1,34 in which T10, T4, T8, and T11 were confirmed to be the energetically favored sites for one Ti insertion in a unit cell and relaxation of the unit cell. This indicates that the clusters used in this work are large enough to include the framework distortion of the replacement of Si by Ti in the TS-1 unit cell. Our results are also partially consistent with the experimental results of Hijar et al.27 and Henry et al.28 The former two favored sites for Ti substitution are T10 and T4, which might be related to the MFI structure that sites T1-T3, T5-T9, T11, and T12 are located only at the walls of straight channels, whereas the walls of zigzag channels comprise all T sites, including T4 and T10.25 From 5T to 17T and 40T, the Si:Ti ratio in these clusters ranges from 4 to 16 and 39, and the RSE sequences showed a little difference if the same constraint scheme is employed to the structure optimizations. This indicates that Si:Ti ratio has less effect than cluster size and constraint scheme. By analyzing the Ti preferential sequences derived from different model schemes, it can be seen that T10 is always the most favored site for Ti substituting, and the SE of T8 and T4 are much more sensitive to the constraint schemes. As the size of the relaxed region increases and the model extends, the SE gap between T4 or T8 and T10 is lowered much more than other sites. As a result, T4 and T8 became more favorable for Ti siting in 40-4T than in any other investigated models. This is in good agreement with the findings by Deka et al.,33 that T4 and T8 sites are the most flexible sites, and is also consistent with Gale's34 results, that T10 is consistently a favored site, and T8 is more marginal depending on the optimization conditions imposed. It is also found that T8, T10, and T11 were often identified as favorable Ti sites by experimental studies,25-28 while T4 was only indicated by one work.17 A previous study43 on aluminum substitution in the zeolite structure showed that SE becomes closer as the size of the relaxed region increases. As expected, data in Table 1 indicated that the relaxation of more atoms in the clusters leads to a lower RSE, and the RSE is almost converged with the 40-4T model. 3.1.2. Structure of the Si- and Ti-Centered Clusters. The structure parameters of the optimized 40-4T clusters are presented in Table 2. It is found that the average T-O1 bond lengths in Ti substituted clusters increase by about 0.15 Å upon Ti insertion into the framework of the MFI zeolite. This is in agreement with the longer diameter of Ti than Si atom. The calculated Ti-O1 distances of 1.758-1.779 Å are about 0.02-0.04 Å shorter than the EXAFS experimental values of 1.80 þ 0.01 Å19 and in good agreement with 1.72-1.78 Å calculated in a theoretical study of Atoguchi and Yao.32 In Ti-substituted clusters, the Si-O bonds in the second nearest neighboring shell of the central Ti atom (O1-T2 bonds in Table 2) are also elongated by 0.005-0.011 Å as compared with the nonsubstituted Si clusters, while T2-O3 bonds are shortened within 0.003 Å. Compared with the changes for T-O1 (about 0.15 Å) and O1-T2 (0.005-0.011 Å) bond lengths, the 943
dx.doi.org/10.1021/jp110977m |J. Phys. Chem. A 2011, 115, 940–947
The Journal of Physical Chemistry A
ARTICLE
Table 2. Structure Parameters of the Optimized 40-4T Clusters Centered at 12 Different T Sitesa,b T-O1 (avar.)
O1-T2 (avar.)
T2-O3 (avar.)
O3-T4 (avar.)
T-O1-T2
Ti
1.765
1.618
1.606
1.611
146.3, 123.3, 128.1,148.9
Si
1.611
1.613
1.606
1.612
154.9, 131.6, 137.3, 155.2
Ti
1.767
1.620
1.602
1.610
139.8, 134.2, 133.8, 156.7
Si
1.609
1.612
1.604
1.608
155.8, 139.3, 141.9, 171.2
Ti
1.758
1.614
1.602
1.607
126.4, 142.7, 154.7, 142.4
Si
1.603
1.604
1.604
1.608
139.2, 158.3, 156.2, 155.1
T4
Ti
1.771
1.626
1.602
1.607
147.1, 148.9, 146.0, 130.3
T5
Si Ti
1.614 1.764
1.619 1.614
1.602 1.604
1.604 1.610
154.6, 153.5, 156.6, 141.4 134.8, 127.8, 135.2, 138.7
Si
1.610
1.610
1.606
1.610
139.6, 139.1, 141.6, 147.7
Ti
1.763
1.618
1.602
1.604
124.0, 150.3, 138.7, 130.7
Si
1.606
1.607
1.603
1.603
134.9, 162.2, 150.8, 138.9
Ti
1.768
1.618
1.604
1.612
133.6, 127.4, 133.8, 140.6
Si
1.609
1.609
1.604
1.609
143.0, 135.9, 142.2, 151.6
T8
Ti
1.769
1.624
1.602
1.608
149.1, 163.1, 128.8, 144.6
T9
Si Ti
1.610 1.768
1.615 1.619
1.605 1.610
1.605 1.614
161.2, 168.5 140.1, 155.2 142.0, 127.2, 122.3, 146.3
Si
1.615
1.616
1.611
1.614
147.9, 133.1, 131.7, 152.1
Ti
1.779
1.629
1.610
1.614
129.9, 129.8, 141.5, 135.2
Si
1.624
1.626
1.611
1.613
137.7, 134.6, 147.9, 141.0
Ti
1.773
1.626
1.606
1.610
140.2, 133.1, 143.8, 136.9
Si
1.616
1.620
1.608
1.608
147.4, 144.5, 148.9, 142.8
Ti
1.766
1.616
1.602
1.608
134.4, 134.9, 138.3, 130.8
Si
1.607
1.607
1.602
1.605
149.2, 140.2 155.9, 133.9
T1 T2 T3
T6 T7
T10 T11 T12 a
Bond lengths are in Å and angles in degrees. b T refers to the central Ti or Si atom in each cluster; O1 refers to oxygen atoms in the first nearest neighboring shell of T; and T2 refers to Si atoms in the second nearest neighboring shell of T. It is the same nomenclature for O3 and T4 in Table 2. T-O1 (avar.), O1-T2 (avar.), T2-O3 (avar.), and O3-T4 (avar.) refer to the average distances of all T-O1, O1-T2, T2-O3, and O3-T4 bonds, respectively, and T-O1-T2 refers to the angle of T-O1 and O1-T2 bonds.
variations in T2-O3 distance upon Ti incorporation decrease greatly. As to the O3-T4 bond, the substitution of Si by Ti causes a little shortening or lengthening of its distance, indicating less influence of the substitution to the remote neighbors of the central T atom. Therefore, to represent the structure distortion of the zeolite lattice upon Ti substitution, it is necessary to use clusters containing the first to fourth nearest neighboring shell of the central T atom, and these atoms should not be fixed during structure optimizations. In Table 2, all Ti-O1-T2 angles decrease as compared with the Si-O1-T2 angles in the original Si clusters to compensate the enlargement of Ti-O bond lengths. It is also seen that for Ti substituted in T10 site, the decrease of the Ti-O1-T2 angles are much smaller than in other sites. This might explain the easier substitution of Ti in T10 site. On the basis of the above discussion, it is suggested that the 404T model scheme is the best choice to investigate the siting of Ti in the MFI framework. Therefore, it is selected for the calculations of the following parts. 3.2. Siting of Si Vacancies in the MFI Lattice and the Preferential Ti Site from Si Vacancies. It has been shown by several independent characterization techniques (such as IR, UV-vis, XPS, EXAFS, and microcalorimetry)19,44-48 that the insertion of the Ti in the MFI lattice caused the progressive reduction of the framework defects. In a neutron diffraction study, Lamberti et al.26 have shown that Ti sites in TS-1 and Si vacancies in Ti-free silicalite-1 are preferentially hosted in the same four sites (T6, T7, T11, and T10). Therefore, it was suggested that the incorporation mechanism of Ti atoms in the
Figure 4. 40T clusters centered on T6 site with the central site being defected (4-a) and its neighboring T9 site being substituted by a Ti (T6def-T9-Ti pair, 4-b).
MFI framework occurs via the insertion of Ti in defective sites of silicalite-1. In this work the distribution of the defective sites (Si vacancies) in the MFI framework was investigated. On the basis of the results, the energetically favored Ti sites by Ti replacing Si vacancies are determined. The siting of Si vacancies is investigated by calculating the energy change of this reaction: Tx-Si-cluster þ 4H2 O f Tx-defective-cluster þ SiðOHÞ4 ðReaction-2Þ where Tx-Si-cluster is the cluster centered on each of the 12 distinct T site with its central Tx (x = 1, 2, ..., 12) Si site being fully 944
dx.doi.org/10.1021/jp110977m |J. Phys. Chem. A 2011, 115, 940–947
The Journal of Physical Chemistry A
ARTICLE
Table 3. Calculated RSE (kcal/mol) for a Hydroxyl Nest Substituting a Fully Coordinated Si Site (Reaction-2) on Each of the 12 Different Sites of MFI and the RSE of Forming the [Ti(OSi)4] Sites from Si Vacancies (Reaction-3) T1
T2
T3
T4
Reaction-2
6.29
3.07
8.23
0.76
Reaction-3
6.21
7.24
7.25
5.23
T5
T6
T7
T8
T9
T10
T11
4.74
0.00
5.23
0.87
4.34
4.29
3.80
0.44
10.29
15.40
7.16
5.99
7.36
0.00
3.40
14.43
Table 4. RSE (kcal/mol) for Ti Substituting a Si on Each of the Four Neighboring T Site of the Si Vacancies (Reaction-4) To Form a Ti and Defective Si Pair T6-def
T12-def
T1-Ti
T4-def
T8-def
13.96
T2-Ti
12.47
T3-Ti
15.03
T2-def
when replacing a Si in the MFI framework, while the whole sequence derived from two reactions (Reaction-1 and Reaction3) shows some difference. This indicates that T10 is not affected by the insertion mechanism as the most favorable sites for Ti in the MFI lattice. This might be related to its special location in the walls of zigzag channels of the MFI lattice. 3.3. Location of the Defective [Ti(OSi)3OH] Sites. The EXFAS and IR studies have evidenced the coexistence of both regular [Ti(OSi)4] and defective “open” [Ti(OSi)3OH] sites in TS-1 samples. Recently, a theoretical work by Wells et al.49 demonstrated that the microstructure aspect of the Ti active site is crucial for epoxidation reactivity and furthermore that Ti sites located near Si vacancies in Ti substituted silicate lattices provide a significant advantage for epoxidation. Up to now, no experimental and theoretical works on the location of this kind of sites have been reported. Therefore, it is important to clarify this type of sites in the TS-1 catalyst. Location of the defective [Ti(OSi)3OH] sites in the TS-1 was studied based on the results of the previous part of the siting of Si vacancies. For the former six defective sites (T6, T12, T4, T8, T2, and T11), the RSE for a Ti replacing each of the four neighboring Si of the hydroxyl nest was calculated. The reaction could be written as:
T11-def
2.52 9.39
3.18
9.08
3.71
T4-Ti T5-Ti
1.64
5.40
13.55
T6-Ti
16.82
T7-Ti T8-Ti T9-Ti
9.29
4.37
5.30 0.00
5.92 5.83
2.39
T10-Ti
6.48
T11-Ti
8.34
T12-Ti
3.09
15.42
3.90
coordinated, and the Tx-defective-cluster is the corresponding defective cluster with the lack of the central Si atom being balanced by a hydroxyl nest (Figure 4). The RSE presented in Table 3 corresponds to Reaction-2 and is calculated by:
Tx-defect-Ty-Si þ TiðOHÞ4 f Tx-defect-Ty-Ti þ SiðOHÞ4
RSE ¼ ½EðTx-defective-clusterÞ - EðTx-Si-clusterÞ
ðReaction-4Þ
- ½EðT6-defective-clusterÞ - EðT6-Si-clusterÞ ðFormula-3Þ ,with T6 having the lowest SE. On the basis of the results shown in Table 3, it can be found that the preferential order for Si vacancies is: T6 > T12 > T4 > T8 > T2 > T11 > T10 > T9 > T5 > T7 > T1 > T3, with the SE varying within 8.23 kcal/mol and a relatively large gap between T8 and T2 sites. This indicates that Si vacancies would mainly occupy T6, T12, T4, and T8 sites, and probably spread among all of the 12 distinct sites. With respect to the former four energetically favorable sites for Si vacancies, T4 and T8 are also favored by the Ti center. This might explain the experimental finding that the insertion of Ti in the MFI lattice caused the progressive reduction of the framework defects. The experimental work of Lamberti et al.26 also confirmed T6 as one of the preferential site for Si vacancies. If the framework Ti site is formed by Ti substituting a Si vacancy in the zeolite model, the substitution reaction would be: Tx-defective-cluster þ TiðOHÞ4 f Tx-Ti-cluster þ 4H2 O ðReaction-3Þ The calculated RSE for Reaction-3 (Table 3) on the 12 crystallographically distinct T sites suggested that the most preferred site for Ti is T10, with a relatively large gap between the second preferred site T11 and T10. In the Section 3.1, it was also found that T10 is the most energetically favorable site for Ti
T12
where Tx-defect-Ty-Si refers to clusters with Tx site being a hydroxyl nest and Ty site being a Si neighbored to it, which would be replaced by a Ti, while Tx-defect-Ty-Ti refers to the corresponding clusters with Si at Ty site being replaced by a Ti. The total energies of Tx-defect-Ty-Si and Tx-defect-Ty-Ti clusters were obtained by reoptimizing the corresponding defective clusters with up to the fourth neighboring shell of the Ty site relaxed. The calculated RSE for Reaction-4 on the 12 distinct T sites are presented in Table 4. It is indicated that, when Si vacancy is at T6, Ti preferentially occupies its neighboring T9 site (T6-def-T9-Ti pair). This is the most energetically stable pair, followed by T6-def-T5-Ti pair, with a rather small enegy gap (1.64 kcal/mol). In addition, T8-def-T9-Ti, T2-def-T1-Ti, T12-def-T12-Ti, T12-def-T3-Ti, and T11-def-T12-Ti pairs with a energy gap less than 4.0 kcal/mol to the most stable one would not be ruled out.
’ CONCLUSIONS Siting of Ti in the MFI framework has been explored by full ab initio calculations. It was found that T10, T4, T8, and T11 are the most energetically favored sites for Ti substitution on the T sites of MFI, which is in good agreement with the results of a full ab initio periodic study. The effects of the cluster sizes, Si:Ti ratio, and the constraint scheme for structure optimizations on the calculated RSE were investigated in detail. The results showed that cluster size and constraint scheme are two significant factors 945
dx.doi.org/10.1021/jp110977m |J. Phys. Chem. A 2011, 115, 940–947
The Journal of Physical Chemistry A
ARTICLE
to affect the preferential siting of Ti, while the Si:Ti ratio has less effect than cluster size and constraint scheme. In addition, the distortion of the zeolite structure by Ti insertion would spread to at least the third nearest neighboring shell of the Ti site. Therefore, large clusters such as 40T are required to represent the zeolite structure, and during structure optimizations, the relaxed atoms should include the first to fourth nearest neighboring shells of the central atom (40-4T model scheme). On this basis, siting of the defective sites in the MFI was investigated by 40-4T model scheme. It was shown that the former four energetically favorable sites for Si vacancies are T6, T12, T4, and T8. Among these sites, T4 and T8 are also favored by Ti center; this might explain the experimental finding that the insertion of Ti in the MFI framework progressively reduced the Si vacancies. Calculations also showed that whether by replacing Si vacancies or substituting the fully coordinated Si sites, the most preferential site for Ti incorporating is T10, which indicates that the insertion mechanism does not affect the favorable sites of Ti in the MFI lattice. With respect to the location of the defective [Ti(OSi)3OH] sites, the calculations showed that Si vacancy at T6 with a Ti at its neighboring T9 site (T6-def-T9-Ti pair) is the most favorable site, followed by T6-def-T5-Ti pair with a small energy gap.
(15) Bonino, F.; Damin, A.; Bordiga, S.; Lamberti, C.; Zecchina, A. Langmuir 2003, 19 (6), 2155–2161. (16) Ratnasamy, P.; Srinivas, D.; Kn€ ozinger, H. Adv. Catal. 2004, 48, 1–169. (17) Lamberti, C.; Bordiga, S.; Zecchina, A.; Carati, A.; Fitch, A. N.; Artioli, G.; Petrini, G.; Salvalaggio, M.; Marra, G. L. J. Catal. 1999, 183 (2), 222–231. (18) Blasco, T.; Camblor, M.; Corma, A.; Perez-Parriente, J. J. Am. Chem. Soc. 1993, 115 (25), 11806–11813. (19) Bordiga, S.; Coluccia, S.; Lamberti, C.; Marchese, L.; Zecchina, A.; Boscherini, F.; Buffa, F.; Genoni, F.; Leofanti, G.; Petrini, G.; Vlaic, G. J. Phys. Chem. 1994, 98 (15), 4125–4132. (20) Bordiga, S.; Boscherini, F.; Coluccia, S.; Genoni, F.; Lamberti, C.; Leofanti, G.; Marchese, L.; Petrini, G.; Vlaic, G.; Zecchina, A. Catal. Lett. 1994, 26, 195–208. (21) Millini, R.; Perego, G.; Seiti, K. Stud. Surf. Sci. Catal. 1994, 84, 2123–2129. (22) de Mann, A. J.; Sauer, M. J. J. Phys. Chem. 1996, 100 (12), 5025– 5034. (23) To, J.; Sokol, A. A.; French, S. A.; Catlow, C. R. A. J. Phys. Chem. C 2008, 112 (18), 7173–7185. (24) Perego, G.; Bellussi, G.; Corno, C.; Taramasso, M.; Buonuomo, F.; Esposito, A. Stud. Surf. Sci. Catal. 1986, 28, 129–136. (25) Marra, G. L.; Artioli, G.; Fitch, A. N.; Milanesio, M.; Lamberti, C. Microporous Mesoporous Mater. 2000, 40 (1-3), 85–94. (26) Lamberti, C.; Bordiga, S.; Zecchina, A.; Artioli, G.; Marra, G.; Spano, G. J. Am. Chem. Soc. 2001, 123 (10), 2204–2212. (27) Hijar, C. A.; Jacubinas, R. M.; Eckert, J.; Henson, N. J.; Hay, P. J.; Ott, K. C. J Phys. Chem. B 2000, 104 (51), 12157–12164. (28) Henry, P. F.; Weller, M. T.; Wilson, C. C. J. Phys. Chem. B 2001, 105 (31), 7452–7458. (29) Njo, S. L.; van Koningsveld, H.; van de Graaf, B. J. Phys. Chem. B 1997, 101 (48), 10065–10068. (30) Oumi, Y.; Matsuba, K.; Kubo, M.; Inui, T.; Miyamoto, A. Microporous Mesoporous Mater. 1995, 4 (1), 53–57. (31) Sastre, G.; Corma, A. Chem. Phys. Lett. 1999, 302, 447–453. (32) Atoguchi, T.; Yao, S. J. Mol. Catal. A 2003, 191 (2), 281–288. (33) Deka, R. Ch.; Nasluzov, V. A.; Shor, E. A. I.; Shor, A. M.; Vayssilov, G. N.; R€ osch, N. J. Phys. Chem. B 2005, 109 (51), 24304– 24310. (34) Gale, J. D. Solid State Sci. 2006, 8, 234–240. (35) Bordiga, S.; Roggero, I.; Ugliengo, P.; Zecchina, A.; Bolis, V.; Artioli, G.; Buzzoni, R.; Marra, G. L.; Rivetti, F.; Span o, G.; Lamberti, C. J. Chem. Soc., Dalton Trans. 2000, 3921–3929. (36) Artioli, G.; Lamberti, C.; Marra, G. L. Acta Crystallogr., Sect. B: Struct. Sci. 2000, 56, 2–10. (37) Wells, D. H.; Joshi, A. M., Jr.; Delgass, W. N.; Thomson, K. T. J. Phys. Chem. B 2006, 110 (30), 14627–14639. (38) van Koningsveld, H.; van Bekkum, H.; Jansen, J. C. Acta Crystallogr. 1987, B43, 127–132. (39) Yuan, S.; Wang, J.; Li, Y.; Peng, S. J. Mol. Catal. A 2001, 175 (1-2), 131–138. (40) Gaussian 03, Revision B.04; Gaussian, Inc.: Pittsburgh, PA, 2003. (41) Sauer, J.; Ugliengo, P.; Garrone, E.; Saunders, V. R. Chem. Rev. 1994, 94 (7), 2095–2160. (42) Sinclair, P. E.; Catlow, C. R. A. J. Phys. Chem. B 1999, 103 (7), 1084–1095. (43) Schr€oder, K.-P.; Sauer, J.; Leslie, M.; Catlow, C. R. A. Zeolites 1992, 12 (1), 20–23. (44) Lamberti, C.; Bordiga, S.; Arduino, D.; Zecchina, A.; Geobaldo, F.; Spano, G.; Genoni, F.; Petrini, G.; Carati, A.; Villain, F.; Vlaic, G. J. Phys. Chem. B 1998, 102 (33), 6382–6390. (45) Bolis, V.; Bordiga, S.; Lamberti, C.; Zecchina, A.; Carati, A.; Rivetti, F.; Spano, G.; Petrini, G. Langmuir 1999, 15 (18), 5753– 5764. (46) Bolis, V.; Bordiga, S.; Lamberti, C.; Zecchina, A.; Carati, A.; Rivetti, F.; Spano, G.; Petrini, G. Microporous Mesoporous Mater. 1999, 30 (1), 67–76.
’ AUTHOR INFORMATION Corresponding Author
*Tel.: þ86-532-85950690. Fax: þ86-532-85950768. E-mail address:
[email protected] (Shuping Yuan).
’ ACKNOWLEDGMENT The authors are grateful to National Natural Science Foundation of China (20403028, 20773071, 20703027) and the Open Foundation of the State Key Laboratory of Coal Conversion (9908). We also thank the Natural Science Foundation of Shandong Province (ZR2010BM024) and the Educational Commission of Shandong Province (J09LB06, J09LF16, J10LB06). ’ REFERENCES (1) Taramasso, M.; Perego, G.; Notari, B. U.S. Patent No. 4410501, 1983. (2) Yokoi, T.; Wu, P.; Tatsumi, T. Catal. Commun. 2003, 4 (1), 11–15. (3) Bravo-Suarez, J. J.; Bando, K. K.; Fujitani, T.; Oyama, S. T. J. Catal. 2008, 257 (1), 32–42. (4) Cumaranatunge, L.; Delgass, W. N. J. Catal. 2005, 232 (1), 38–42. (5) Liu, T.; Hacarlioglu, P.; Oyama, S. T.; Luo, M.-F.; Pan, X.-R.; Lu, J.-Q. J. Catal. 2009, 267 (2), 202–206. (6) Sooknoi, T.; Chitranuwatkul, V. J. Mol. Catal. A 2005, 236 (1-2), 220–226. (7) Saxena, S.; Basak, J.; Hardia, N.; Dixit, R.; Bhadauria, S.; Dwivedi, R.; Prasad, R.; Soni, A.; Okram, G. S.; Gupta, A. Chem. Eng. J. 2007, 132 (1-3), 61–66. (8) Mantegazza, M. A.; Leofanti, G.; Petrini, G.; Padovan, M.; Zecchina, A.; Bordiga, S. Stud. Surf. Sci. Catal. 1994, 82, 541–550. (9) Mantegazza, M. A.; Petrini, G.; Spano, G.; Bagatin, R.; Rivetti, F. J. Mol. Catal. A 1999, 146 (1-2), 223–228. (10) Lane, B. S.; Burgess, K. Chem. Rev. 2003, 103 (7), 2457–2474. (11) Notari, B. Adv. Catal. 1996, 41, 253–334. (12) Bordiga, S.; Bonino, F.; Damin, A.; Lamberti, C. Phys. Chem. Chem. Phys. 2007, 9 (35), 4854–4878. (13) Damin, A.; Bordiga, S.; Zecchina, A.; Lamberti, C. J. Chem. Phys. 2002, 117 (1), 226–237. (14) Bordiga, S.; Damin, A.; Bonino, F.; Zecchina, A.; Spano, G.; Rivetti, F.; Bolis, V.; Prestipino, C.; Lamberti, C. J. Phys. Chem. B 2002, 106 (38), 9892–9905. 946
dx.doi.org/10.1021/jp110977m |J. Phys. Chem. A 2011, 115, 940–947
The Journal of Physical Chemistry A
ARTICLE
(47) Le Noc, L.; Trong On, D.; Solomykina, S.; Echchahed, B.; Bel, F.; dit Moulin, C. C.; Bonneviot, L. Stud. Surf. Sci. Catal. 1996, 101, 611–620. (48) Trong On, D.; Le Noc, L.; Bonneviot, L. Chem. Commun. 1996, 299–300. (49) Wells, D. H.; Delgass, W. N., Jr.; Thomson, K. T. J. Am. Chem. Soc. 2004, 126 (9), 2956–2962.
947
dx.doi.org/10.1021/jp110977m |J. Phys. Chem. A 2011, 115, 940–947
