A Two-Layer ONIOM Study on Initial Reactions of Catalytic Cracking of

Mar 4, 2010 - of C4 olefin to produce propene and ethene.2-5 Zhu et al.4 have reported that .... 3. Results and Discussion. The energy profiles calcul...
0 downloads 0 Views 4MB Size
J. Phys. Chem. C 2010, 114, 5975–5984

5975

A Two-Layer ONIOM Study on Initial Reactions of Catalytic Cracking of 1-Butene To Produce Propene and Ethene over HZSM-5 and HFAU Zeolites Ying-Xin Sun, Jing Yang, Li-Feng Zhao, Jian-Xing Dai, and Huai Sun* School of Chemistry and Chemical Engineering, Shanghai Jiao Tong UniVersity, Shanghai, 200240, China ReceiVed: NoVember 7, 2009; ReVised Manuscript ReceiVed: December 28, 2009

Two-layer ONIOM calculations were carried out to study initial reactions of catalytic cracking of 1-butene to produce propene and ethene on HZSM-5 and HFAU zeolites. Direct cracking and dimerization cracking mechanisms were evaluated. The calculated data indicate that the dimerization cracking is more favorable than the direct cracking based on both kinetic and thermodynamic considerations. HZSM-5 is catalytically more effective than HFAU for the cracking reactions. ONIOM energy analysis shows that the effectiveness of zeolite is due to long-range van der Waals interaction energies that stabilize all reaction intermediates, and short-range interaction between the zeolite and the reacting species that reduce the activation energies. For the dimerization process, stepwise and concerted mechanisms are similar in energy changes. Generally speaking, dimerizations appear to be exothermic reactions with modest activation energies. The activation energies for isomerization, β-scission, and deprotonation are lower for larger molecular fragments than for smaller ones. 1. Introduction Catalytic cracking of paraffins or olefins on zeolites is one of most important processes in the petrochemical industry. Because traditional methods such as steam cracking of hydrocarbon feedstock cannot satisfy increasing industrial demand of propene,1 there has been a strong interest in catalytic cracking of C4 olefin to produce propene and ethene.2-5 Zhu et al.4 have reported that formation of ethene and propene by cracking of C4H8 oligomers has a strong dependence on zeolite structures. For instance, the propene selectivity is below 5% on the 12member-ring Y zeolite, about 12% on the 10-member-ring ZSM5, and up to 43% on small-pore SAPO-34 zeolite. It has been suggested4,5 that the higher propene selectivity in zeolite with smaller pore size is due to suppression of the intermolecular hydrogen transfer and secondary alkane-alkene transformations.6 High-level reaction pathways were speculated5 based on experimental data, but the mechanism of catalytic cracking of C4 olefin to produce ethene and propene is not clear. The catalytic cracking of C4 olefin is a very complicated process modulated by external conditions such as temperatures, pressures, and types of zeolites; there are many side reactions which have strong impact to the yields of products. It is difficult to map out the reaction mechanisms solely based on experimental measurements. To this end, computational chemistry is a powerful approach to explore the details of chemical reactions. Several theoretical approaches have been used to study catalytic reactions on zeolites, such as bare cluster model,7-15 density functional theory (DFT) calculation with periodic boundary condition,16 and combined quantum mechanics and molecular mechanics (QM/MM) approach.17-22 The bare cluster approach in which the zeolites are represented by small silicon oxide clusters is useful for describing local phenomena such as interactions of organic molecules with active sites, or bond cleavage and formation near the active sites. For instance, Svelle et al.14 investigated dimerization of linear alkenes catalyzed by * Corresponding author: tel, 86 21 5474 8987; fax, 86 21 5474 1297; e-mail address, [email protected].

acidic zeolites using a four tetrahedral site (4T) cluster via two different mechanisms: stepwise and concerted. Frash et al.9 studied β-scission reactions of hydrocarbons using a 3T cluster model. Because the zeolite environment is neglected in calculations, the bare cluster models cannot be used to compare different zeolites. A periodic DFT approach is the most natural approximation to a crystalline framework relative to the bare cluster model method; however an accurate description of dispersion energies is an active research area. Hybrid QM/MM method that combines quantum mechanics for describing chemical reactions near active sites and molecular mechanics force field for representing long-range interactions provides potential to study chemical reactions in complicated zeolite environment efficiently. There are many reported works along this direction in recent years. To name a few, Namuangruk et al.18 studied adsorption energies of ethene and butene isomers on HZSM-5 using Our-own-N-layered Integrated molecular Orbital + molecular Mechanics (ONIOM) method.23-25 Joshi and Thomson22 carried out a QM/MM investigation on C6 diene cyclization on HZSM-5. In this work, we carried out two-layer ONIOM calculations to study reaction mechanisms in catalytic cracking of 1-butene to produce propene and ethene. Relevant experimental and computational studies on interactions of hydrocarbon with zeolites provide foundation for proposing reaction mechanisms. For example, Kazansky et al.26 have studied adsorption of ethene on HMOR zeolite and subsequent oligomerization at room temperature. Namuangruk et al.27 investigated ethene dimerization over FAU zeolite using ONIOM method and evaluated two mechanisms: stepwise and concerted dimerization. Hay et al.28 have examined direct β-scission of the C-C bond of pentene using a cluster model and concluded that overall reaction is endothermic and the kinetic rather than thermodynamic factors plays a significant role in cracking of paraffin hydrocarbons. On the basis of these works, we proposed direct cracking and dimerization cracking mechanisms and investigated the structural and energetic aspects of the reaction pathways using the ONIOM method. In order to understand how zeolite structures influence

10.1021/jp910617m  2010 American Chemical Society Published on Web 03/04/2010

5976

J. Phys. Chem. C, Vol. 114, No. 13, 2010

Sun et al.

SCHEME 1: Initial Reaction Pathways of Cracking 1-Butene to Produce Propene and Ethene over Acidic Zeolite

the reaction pathways, we compared two different zeolites, 10member-ring HZSM-5 and 12-member-ring HFAU. The reaction network is very complicated; many reasonable reactions can be proposed. Focusing on initial and most favorable reactions only, we proposed four reaction pathways as illustrated in Scheme 1. The reaction pathways are grouped by formation of primary butoxide (1-butoxide, pathways I and II) and formation of secondary butoxide (2-butoxide, pathways III and IV). In each of these two groups, direct cracking (I and III) and dimerization cracking (II and IV) were investigated. These initial reactions yield desired products (ethene or propene) and alkoxides. To finish the catalytic cycles, the resulting alkoxides may be converted through isomerization, deprotonation, dimerization, and hydrogen transfer reactions. To completely explore all of the possible reactions is beyond the scope of this project. Therefore we calculated plausible reactions that are potentially more energetically favorable than others and that complete the catalytic cycles, as illustrated in Scheme 2. We are not aware of any theoretical reports on the mechanisms of cracking of 1-butene over zeolites and other porous materials. The elucidation of the reaction mechanisms provides insights into the fundamental steps of the reactions. On comparison of calculated results between two different zeolites, this study helps to understand how different zeolite structures influence the initial reactions of the cracking process, which is useful for optimizing the reaction conditions and designing more efficient catalysts for industrial production.

SCHEME 2: Plausible Reaction Pathways for Completing the Catalytic Cycles of 1-Butene Cracking over Acidic Zeolite

2. Models and Methods HZSM-5 and HFAU are represented by nanoclusters denoted as 140T and 120T, respectively, in this work. Figure 1 illustrates

Catalytic Cracking of 1-Butene

Figure 1. 140T nanocluster model of HZSM-5 zeolite divided into two regions: the inner 38T region indicated by colored balls are treated using the quantum mechanics DFT method and the outer region is treated by using UFF. The dash line indicates the direction of the straight channel (a) and zigzag channel (b).

the 140T HZSM-5 model which covers both straight and zigzag channels in the material. The T12 site at the cross section of straight and zigzag channels was selected as the active Brønsted acid site29 where the Al atom, acidic hydrogen Hb, and acidic oxygen O1 atoms are labeled. Figure 2 illustrates the 120T HFAU model which covers an area of two supercages connected to each other through the 12-member-ring window. The Brønsted acidic sits is denoted by labeled atoms.30 The atomic coordinates of the models were taken from lattice data of ZSM-5 and FAU zeolites.31,32 The dangling bonds resulting from cutting outside Si-O bonds were saturated by hydrogen atoms with the same direction of crystallographic Si-O bonds, and the Si-H bond lengths were fixed at 1.47 Å. A two-layer ONIOM33 calculation scheme was applied in this work. In this method, the computation model was divided into two regions. The “inner” region, which includes a zeolite fragment containing the active Brønsted acidic site and reacting molecules, was described by quantum mechanics (QM) calculations. The “outer” region, which was the extended zeolite framework, was treated by molecular mechanics (MM). The total energy of the system is expressed as33 All Inner Inner EONIOM2 ) EMM - EMM + EQM

where the superscript All means the whole system which include the “inner” and “outer” parts as explained above. In this scheme

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5977

Figure 2. 120T nanocluster model of HFAU zeolite divided into two regions, the inner 42T region indicated by colored balls is treated using the quantum mechanics DFT method and the outer region treated by using UFF: (a) front view; (b) side view.

the long-range van der Waals (VDW) interaction energies were described by the MM terms. The errors of ONIOM extrapolation were checked by an S values test.34,35 For the ONIOM energy surface to be continuous, the MM contribution must be continuous. The MM contribution can be examined using the S value All Inner SMM ) EMM - EMM

The S value must be independent of definitions of connectivity in the reaction center. For a transition state, for example, its S value calculated with the connectivity as the reactant structure must be the same as that calculated with the connectivity as the product structure. We investigated several representative transition states of β-scission of butoxides and octoxides on HZSM-5 and HFAU zeolites. The calculated S values between different selections of connectivity were zero. The calculated data are summarized in the Supporting Information. We applied universal force field (UFF)36 for describing interatomic interactions in the two-layer ONIOM calculations. The quantum mechanics density functional theory (DFT) using the B3LYP functional37 was used as default for the QM calculations. To reduce computational expenses, the inner QM region was further divided into two subdivisions which were treated using different basis sets. An active subregion, which includes a 5T cluster of the active site of zeolite and reacting

5978

J. Phys. Chem. C, Vol. 114, No. 13, 2010

Sun et al. TABLE 1: Comparison of ONIOM Energy Barrier Heights (in kJ/mol) for the Most Important Transition States (β-Scission of Butoxides and Octoxides) Calculated with and without Atomic Partial Charges in the MM Calculations

Figure 3. Calculated activation energies of stepwise dimerization (TS3 in Scheme 1) of 1-butene molecules using different sizes of the inner (QM) region on HZSM-5 and HFAU zeolite models.

molecules, was treated using the 6-31G(d,p) basis set; the rest of the inner region was treated using the 3-21G basis set. In geometry optimizations the active subregion was fully relaxed, while the rest of the computation model, including the lessactive inner region and outer region, was fixed. The optimized structures were tested by computing second derivatives of the ONIOM2 total energy with respect to nuclear coordinates to ensure that the optimized structures were minima on the potential energy surfaces. In order to estimate short-range dispersion energies, MP2 single-point energy calculations were performed on selected species based on DFT optimized structures. We denote data calculated using this method as ONIOM2(MP2// DFT:UFF) in this paper. All calculations reported in this work were performed using the Gaussian03 program package.38 The ONIOM method with B3LYP functional and UFF force field has been used successfully for studying chemical reaction mechanisms in zeolites by several research groups.22,39,40 It is known that B3LYP functional underestimates activation barriers of transition states and is unable to describe van der Waals complexes bound by medium-range interactions.41 New density functionals, such as M0542 and M0643-46 have been proposed to solve these problems. According to the Zhao et al.,41 B3LYP functional systematically underestimates barrier heights by an average of 4.4 kcal/mol for a collection of 76 barrier heights of hydrogen-transfer and non-hydrogen-transfer reactions. In this work, we did not use these newly developed functionals for the following reasons: (1) The reported underestimates are systematical so that comparisons of data obtained using the same method for different chemical species are more accurate than the absolute predictions. (2) The average underestimates are relatively small in comparison with most energy barrier heights obtained in present work. (3) The long-range van der Waals interactions are included using classical force field in our computational model. Since the charge parameters in UFF force field were not optimized to represent the electrostatic potentials in zeolites, we extended the sizes of the inner regions to reduce any artifact associated with this shortcoming. This was done by evaluating activation energy sensitive to electrostatic interactions using various sizes of the inner regions. Figure 3 shows the results calculated for the activation energy of stepwise dimerization of 1-butene with 1-butoxide on HZSM-5 and HFAU, respectively. These energies are sensitive to electrostatic potentials because the transition state corresponds to a charged species. As shown in the figure, the calculated total ONIOM2 energies decrease significantly as the sizes of the inner cluster increase.

TS2 with charge without charge ∆E TS7 with charge without charge ∆E TS5 with charge without charge ∆E TS10 with charge without charge ∆E

HZSM-5

HFAU

155.48 158.23 -2.75

179.43 181.17 -1.74

164.58 155.92 8.66

156.95 157.97 -1.02

120.42 111.11 9.31

140.25 142.21 -1.96

125.26 117.47 7.79

133.23 135.76 -2.53

The calculated data approach to converged values as the sizes of the inner regions are 38T for HZSM-5 and 42T for HFAU. Consequently, we applied the 38T and 42T as the inner regions for HZSM-5 and HFAU respectively. As the QM region is large enough, the contribution of atomic partial charges used in the MM calculations is minimal. This is demonstrated by calculated energy data for the most important transition states. As given in Table 1, barrier height differences obtained with and without atomic partial charges (Mulliken charges derived from the DFT calculations and subject to charge neutrality constraints in the model and real regions) are small, ranging from approximately 1 to 9 kJ/mol. 3. Results and Discussion The energy profiles calculated for the initial reactions (Scheme 1) are plotted in Figure 4. The energy profiles are denoted by solid lines and bold font faces for HZSM-5 and dash lines and normal font faces for HFAU. For each intermediate or transition state, energy data calculated relative to the total energy of the reactant (1-butene) and the zeolite (HZSM-5 or HFAU) at infinitive separation are labeled for both zeolites. For transition states (TS), activation energies are also given in pairs (∆EHZSM5, ∆EHFAU). The activation energies are calculated as energy differences between the transition states (denoted as TSn) and their previous intermediate states. Formation of Butoxides. The reactions start with one 1-butene molecule adsorbed on the active site, forming a weakly bonded C4-complex. The complex is subsequently converted into butoxide through a transition state. Depending on where the proton is transferred to, primary or secondary carbenium ions47 may be formed. The transition states of the protonation are denoted as TS1 and TS6, and the resulting butoxides are 1-butoxide and 2-butoxide, respectively. Table 2 lists the energy data calculated for relevant species: C4-complex, transition states TS1 and TS6, and two butoxides. The energy values calculated on both zeolite models are listed, and the energy data are decomposed in terms of QM and MM contributions. The adsorption energies obtained for C4-complex are (-75, -54) kJ/mol on HZSM-5 and HFAU, respectively. The QM contributions to the adsorption energies are similar on two zeolites, -43 kJ/mol on HZSM-5 and -48 kJ/mol on HFAU. However, MM contributions are quite different: -32 and -6 kJ/mol, respectively, indicating the long-range VDW

Catalytic Cracking of 1-Butene

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5979

Figure 4. Energy profiles calculated for initial reaction pathways illustrated in Scheme 1. The solid lines and data in bold font face are obtained using the HZSM-5 model; the dash lines and data in normal font face are obtained using the HFAU model. Energy data calculated for each species are relative to the total energy of the reactant (1-butene) and the zeolite (HZSM-5 or HFAU) at infinitive separation. For transition states, activation energies are given in pairs (∆EHZSM-5, ∆EHFAU).

interactions are stronger in HZSM-5. The ONIOM2(MP2// B3LYP:UFF) energies calculated on the same structures are (-101, -66) kJ/mol. Both DFT and MP2 results are comparable

with those reported by Namuangruk et al.18 and Pantu et al.48 The calculated activation energies for the primary carbenium ion are (100, 77) kJ/mol for HZSM-5 and HFAU, respectively,

5980

J. Phys. Chem. C, Vol. 114, No. 13, 2010

Sun et al.

TABLE 2: Relative ONIOM2 Energies and Decompositions in QM and MM Contributions for Species Found in Protonation of 1-Butene To Produce Butoxide on HZSM-5 and HFAU Zeolite (in kJ/mol) HZSM-5

HFAU

species

QM

MM

total

QM

MM

total

C4-complex TS1 1-butoxide TS6 2-butoxide

-43 37 -26 -5 -48

-32 -12 -28 -14 -19

-75 25 -54 -19 -67

-48 28 -68 -11 -74

-6 -5 -5 -4 -4

-54 23 -73 -15 -78

which are much higher than (56, 39) kJ/mol calculated for the second carbenium ions. The total energies listed in Table 2 show that the transition state (TS6) of the secondary carbenium is more stable than the transition state (TS1) of the primary carbenium, and 2-butoxide is more stable than 1-butoxide, on both zeolites. A close analysis of the energy components indicates that the stability of the secondary carbenium and 2-butoxide is mainly due to the QM contributions. The optimized structures of the C4 complexes, transition state TS6, and 2-butoxide on HZSM-5 are displayed in Figure 5. The 1-butene molecule is bonded to the surface through a π-hydrogen bond (Figure 5a). The Brønsted acid hydrogen is positioned to the center of the double bond of 1-butene, as the distances from the hydrogen to two sp2-carbon atoms are similar: 2.20 and 2.38 Å, slightly closer to the primary sp2 carbon atom. The optimized structural parameters are consistent with Gutmann’s rules,49 i.e., the O1sHb and C1dC2 bond lengths are increased. The optimized structure of transition state TS6 (Figure 5b) shows a carbenium ion feature47 as the proton Hb is closer to the reactant carbon C1 (1.22 Å) than the zeolite oxygen O1 (1.51 Å). Another sp2 carbon (C2) is getting close to another oxygen (O2) as the distance obtained is 2.56 Å. The resulting 2-butoxide is illustrated in Figure 5c. The C2-O2 bond is formed with a distance of 1.57 Å. Structural features obtained for HFAU are similar to those given in the Supporting Information.

Figure 5. Optimized structures of (a) C4-complex, (b) transition state TS6, and (c) 2-butoxide in HZSM-5 zeolite.

Figure 6. Optimized structure of transition state TS11 for isomerization between 1-butoxide and 2-butoxide on HZSM-5.

Figure 7. β-Scission that breaks the C(R)-C(β) and C-OZ (oxygen on zeolite) bonds and forms a shorter chain alkoxide and alkene.

The 1-butoxide and 2-butoxide may be converted into each other by intramolecular hydrogen transfer reaction. The transition state is denoted as TS11 in Scheme 1. In this isomerization process, the C-O bond is shifted to adjacent carbon and a proton is transferred in the opposite direction simultaneously. The optimized structure of TS11 on HZSM-5 is illustrated in Figure 6. The transferring hydrogen Hc is between two carbons (C1 and C2) with almost identical distances, 1.34 and 1.32 Å, a new C1-O2 (or C2-O2) bond may be formed as the distance is 2.40 (or 3.11) Å. Data calculated for HFAU as given in the Supporting Information are similar. The activation energies from 1-butoxide to 2-butoxide as given in Figure 4 are (99, 128) kJ/ mol on HZSM-5 and HFAU, respectively. Data in Table 2 show that the relative energy of 1-butoxide and 2-butoxide on HZSM-5 (-13 kJ/mol) is larger than that on HFAU (-5 kJ/mol). The MM contributions are 9 and 1 kJ/ mol on HZSM-5 and HFAU zeolites, respectively, which indicates that the long-range interactions are more favorable for the conversion of 1-butoxide into 2-butoxide on HFAU. On the other hand, QM contribution (-22 kJ/mol on HZSM-5 vs -6 kJ/mol on HFAU) is in the opposite direction, indicating the short-range interactions, which reflect the pore size and shape more sensitively, favor the conversion in HZSM-5 zeolite. β-Scission of Butoxides. Direct cracking of 1-butene proceeds by taking the β-scission process as illustrated in Figure 7. In this process, the OzsC and C(R)sC(β) bonds are broken, new CdC(R) and OzsC(β) bonds are formed simultaneously. This process generates ethene and ethoxide from 1-butoxide and propene and methoxide from 2-butoxide. The calculated β-scission activation energies are rather high: (158, 181) kJ/mol for 1-butoxide (TS2) and (156, 158) kJ/mol for 2-butoxide (TS7), as given in Figure 4. Similar study in previous literature by Frash et al.9 predicted a 240 kJ/mol for the β-scission of 1-butoxide, higher than our results, indicating the importance of including the environmental effect.

Catalytic Cracking of 1-Butene

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5981

TABLE 3: Relative ONIOM2 Energies and Decompositions in QM and MM Contributions for Transition States of the β-Scission of Butoxide on HZSM-5 and HFAU (in kJ/mol) HZSM-5

HFAU

species

QM

MM

total

QM

MM

total

TS2 TS7

132 108

-28 -20

104 88

113 85

-4 -5

109 80

The calculated ONIOM2 energies and energy components of the transition states, TS2 and TS7, are given in Table 3. The TS2 is generally less stable than TS7, and the differences are mainly due to QM contributions. It is of interest to note that the relative stabilities of each transition state between two different zeolites are similar. Moreover, the low MM energy in HZSM-5 is compensated by its high QM energies, which reflects the compromise between stronger VDW interactions and larger spatial hindrance in small pore zeolite. Figure 8 shows the optimized structures of the transition states TS2 and TS7 on HZSM-5. In both cases the cleavage is observed; the butoxide is separated into two moieties: ethene and ethoxide for TS2 or propene and methoxide for TS7. In both cases, the new alkene molecules are formed because the new CdC bond lengths, 1.35 and 1.36 Å, are very close to the normal CdC bond length of 1-butene (1.33 Å). However, new O-C bonds are not completely formed yet, as shown by the O1-C3 distances of 2.33 Å in TS2 and 2.14 Å for the O1-C4 bond in TS7. The distance between the C3 (or C4) atom to sp2-carbon atoms of the newly formed alkene molecules is in the range of 2.22-2.35 Å. After the transition state, the optimized O-C distance is reduced to ca. 1.6 Å, and the alkene products are desorbed with a small energy increase in the range of 20-30 kJ/mol, as shown in Figure 4. Similar structural data were obtained for HFAU as given in the Supporting Information. Dimerization. Alternative to direct cracking, the butoxide interacts with another 1-butene molecule to form a weakly bonded C8 complex. The C8 complex is subsequently converted into covalent bonded octoxide by forming a C-C bond between the positively charged carbon atom of butoxide to one of the sp2 carbon atoms of the second butene while the O-C bond is shifted to an adjacent sp2 carbon atom. Since two sp2 carbon atoms of the second 1-butene may interact with two types of butoxides, four different C8 octoxides can be formed as illustrated in Scheme 3. Because the primary carbenium ions are less stable than the secondary ones, we studied the more stable isomers, 3-octoxide and 5-methyl-3-heptoxide, in this work. As shown in Figure 4, the C8 complexes (C8-complex1 and C8-complex2) are 20-60 kJ/mol lower in energy than corresponding butoxides on different zeolites. The decrease of energy is important because it indicates the reaction is more likely to adsorb another reactant instead of going through a high-energy barrier of the β-scission in the direct cracking route. In subsequent transition states TS3 and TS8, the positively charged

Figure 8. Optimized structures of transition states TS2 and TS7 for β-scission of 1-butoxide and 2-butoxide, respectively, on HZSM-5.

carbon atom attacks one of the two sp2 carbon atoms of the second 1-butene to form a carbenium cation. The activation energies of the dimerization are modest, (51, 79) kJ/mol for TS3 (path II) and (68, 62) kJ/mol for TS8 (path IV). The octoxides may be formed in a concerted manner14 directly from two reactants as well. In this mechanism, before a butoxide is formed, another butene molecule is coadsorbed near the active center to form C8-complex3. The two butene molecules interact with the acidic center of zeolite to form an octoxide through a single transition state. Similar to the stepwise dimerization as discussed above, four types of octoxides may be formed. Figure 9 shows the optimized structures of two transition states (TS12 and TS13) of the concerted mechanism on HZSM-5. Each of the transition states corresponds to a structure in which the proton is transferred from zeolite to one 1-butene and a new C-C bond between two 1-butene molecules is formed simultaneously. As shown in the figure the acidic proton Hb is transferred to the C2 (1.24 Å) in TS12 and C1 (1.19 Å) in TS13. In both cases, the Hb-O1 distances are longer (1.55 in TS12, and 1.59 in TS13). Meanwhile, the new C-C bond may be formed between C1 and C5 (or C6) in TS12 and C2 and C5 (or C6) in TS13. Table 4 compares the calculated activation energies of stepwise and concerted dimerizations. The stepwise mechanism includes two transition states, protonation of the first 1-butene

SCHEME 3: Four Types of Octoxides May Be Formed from Dimerization of Two 1-Butene Molecules

5982

J. Phys. Chem. C, Vol. 114, No. 13, 2010

Sun et al.

Figure 9. Optimized structures of transition states TS12 and TS13 of concerted dimerization of two 1-butene molecules on HZSM-5.

TABLE 4: Comparison of ONIOM2 Activation Energies (∆Eq) of Stepwise and Concerted Dimerization of Two 1-Butene Molecules (in kJ/mol) 3-octoxide

HZSM-5 HFAU

5-methyl-3-heptoxide

stepwise TS1/TS3

concerted TS12

stepwise TS6/TS8

concerted TS13

100/51 77/79

112 69

56/68 39/62

79 43

TABLE 5: Relative ONIOM2 Energies and Decompositions in QM and MM Contributions for Octoxides (in kJ/mol) HZSM-5

HFAU

species

QM

MM

total

QM

MM

total

3-octoxide 2-octoxide 5-methyl-3-heptoxide 5-methyl-2-heptoxide

-136 -128 -133 -135

-35 -34 -35 -28

-171 -162 -168 -163

-156 -161 -149 -155

-10 -14 -9 -10

-166 -175 -158 -165

and dimerization with the second 1-butene (TS1, TS3 for producing 3-octoxide, TS6 and TS8 for 5-methyl-3-heptoxide). In the concert mechanism, the protonation and dimerization are done through one transition state (TS12 and TS13). From the activation energy data, one sees that the concerted mechanism is slightly more favorable than the stepwise mechanism on HFAU, but the opposite is found on HZSM-5. In addition, the energy barriers are much lower for the more branched octoxide, 5-methyl-3-heptoxide. The resulting 3-octoxide and branched 5-methyl-3-heptoxide are quite stable; the relative energies are in the range from -160 to -170 kJ/mol lower than the reference state (Figure 4). Isomerization and β-Scission of Octoxides. Since β-scission is a reverse reaction of dimerization, octoxides produced by the dimerization process must be isomerized in order to produce molecules different from the reactants. We studied isomerization by shifting the C-Oz bond from the third (3) carbon C (-Oz) to the second (2) carbon for both 3-octoxide and 5-methyl-3heptoxide, the isomerization produces 2-octoxide and 5-methyl2-heptoxide, respectively. The activation energies are (57, 70) kJ/mol for 3-octoxide, and (89, 89) kJ/mol for 5-methyl-3heptoxide. The energy components for the octoxides before and after the isomerization are listed in Table 5. Generally speaking, these octoxides are thermodynamically similar in stability. Comparison of data obtained for two different zeolites indicates that the MM contributions are overall lower on HZSM-5 than those on HFAU. However, the QM energies are higher in HZSM-5,

TABLE 6: Comparison of ONIOM2 Activation Energies (∆Eq) and Decomposition of the Activation Energies in Terms of QM and MM Contribution for β-Scission of Butoxides and Octoxides on HZSM-5 and HFAU (in kJ/mol) HZSM-5

HFAU

species

DFT

MM

total

DFT

MM

total

1-butoxide 2-butoxide 2-octoxide 5-methyl-2-heptoxide

158 156 117 126

0 -1 -6 -8

158 155 111 118

181 159 138 136

1 -1 4 0

182 158 142 136

presumably due to higher intramolecular constraints induced by smaller pore environment. Overall the total energies are similar between different zeolites. The transition states of β-scission of octoxides show similar features as discussed above for β-scission of butoxides. The transition states (TS5 and TS10) are intermediate states from octoxide to pentoxide and propene. The activation energies of the β-scission are lower than those obtained for butoxides, (111, 142) kJ/mol for TS5 and (117, 136) kJ/mol for TS10 (see Figure 4). Since β-scission is the rate-determining step, we analyzed the activation energies in terms of MM and QM components for both butoxides and octoxides in Table 6. The small MM energy contributions (0-8 kJ/mol) imply that the long-range VDW energies (MM contributions) in both transition states and previous intermediate states are similar. Therefore the energy differences are mainly from the QM contributions which only include short-range interactions between the zeolite and the reacting species. Comparison of the data calculated for different zeolites shows that the activation energies are generally lower in HZSM-5 than in HFAU, indicating HZSM-5 is more catalytically effective than HFAU. Plausible Reactions of Alkoxides. The energy profiles of plausible reactions that complete the catalytic reaction cycles are sketched in Figure 10. Similar to Figure 4, the energy data are calculated relative to the 1-butene and zeolites at infinite separation and the activation energies are calculated as differences between the transition states and previous intermediate states. Data obtained using both zeolite models are given. For path I, the newly formed ethoxide may be converted into ethene by direct deprotonation. The calculated activation energies are (102, 117) kJ/mol, which are lower than the high energy barrier (i.e., TS2 (158, 181) kJ/mol) obtained for β-scission reactions. Under similar thermodynamic conditions this reaction could happen to complete the catalytic reaction cycles. For path II, we also calculated the deprotonation of 1-pentoxide to form

Catalytic Cracking of 1-Butene

J. Phys. Chem. C, Vol. 114, No. 13, 2010 5983

Figure 10. Energy profiles of possible reactions for completing catalytic cycles of 1-butene cracking pathways I, II, III, and IV. The notations are the same as those given in Figure 4.

1-pentene. Modest energy barrier heights of (70, 120) kJ/mol are obtained. For path III, methoxide may be dimerized with another ethene molecule to form propoxide. As shown in Figure 10, the dimerization energy barriers are modest, (66, 92) kJ/mol. The resulting 1-propoxide may be deprontonated to yield propene and the catalytic cycle is finished. The activation energies calculated for deprotonation of 1-propoxide are (100, 95) kJ/ mol. In addition, we calculated reaction of methoxide with H2 molecule to produce CH4 and acidic zeolite. The corresponding activation energies are (150, 160) kJ/mol; more details are given in the Supporting Information. Finally for path IV, 2-methyl-1-butoxide may be converted into 2-methyl-2-butoxide or a ternary carbenium (tert-pentyl cation) by isomerization reactions. Stable structure of 2-methyl2-butoxide, characterized by the C-Oz bond length of 1.76 Å, was obtained on HFAU. We could not obtain stable structure for this molecule on HZSM-5. In contrast, tert-pentyl cation is very stable on both zeolites and its structure was characterized as minimum on the potential energy surface by frequency calculations. The calculated C-Oz bond lengths are 3.27 and 3.08 Å on HZSM-5 and HFAU, respectively. The activation energies for converting from 2-methyl-1-butoxide to tert-pentyl cation are (74, 94) kJ/mol as shown in Figure 10. The tert-pentyl cation is only 34 kJ/mol more stable than 2-methyl-2-butoxide on HFAU. The relative stability between tert-pentyl cation and 2-methyl-2-butoxide appears to be strongly correlated to the computational method, in addition to zeolite structure. Our ONIOM2(MP2//B3LYP:UFF) data show that the relative energy order is reversed: 2-methyl-2-butoxide is 14 kJ/ mol more stable than tert-pentyl cation in HFAU. This is consistent with data reported in the literature50 on tert-butyl cation stability. In addition, Boronat et al.20 reported that the tert-butyl cation could be converted into tert-butoxide through a transition state on Brønsted acid sites on HMOR zeolite.

4. Conclusions In this work, two-layer ONIOM calculations were carried out to characterize reaction mechanisms of catalytic cracking of 1-butene to produce propene and ethene on zeolites. Direct cracking and dimerization cracking mechanisms were proposed and evaluated. The calculated data show that the direct cracking is endothermic with high activation energies. In contrast, dimerization cracking is exothermic with lower activation energies. Comparison of ONIOM2 activation energies calculated for two representative zeolites shows that HZSM-5 is more catalytically effective, especially for the dimerization cracking mechanism. These features are clearly demonstrated in Figure 4 and are consistent with experimental observations.4 One of the advantages of the ONIOM method (and other QM/ MM methods) is that the environment effects can be analyzed based on energy contributions. For molecules examined in this work, the long-range VDW interaction energies (MM contributions) are generally more negative in HZSM-5 than those in HFAU. The extent of the MM stabilization is similar for intermediate and transition states of C4 and C8 species. Therefore, the resulting activation energies are mainly due to the QM energy contributions that include the short-range interaction between the zeolite and the reacting species. Our analysis of C4+C4 dimerization pathways shows both stepwise and concerted mechanisms are similar in energy changes. On different zeolites, either one of the two mechanisms may be slightly more favorable, but not decisively. With consideration of kinetic factor, the stepwise mechanism should be more favorable because it does not require two reactant molecules colliding at the reactive site simultaneously. Although reaction pathways studied in this work are rather limited, some additional conclusions may be drawn. Dimerization seems to be a generally exothermic reaction with relatively low activation energy on zeolites. This has been seen in this work not only by C4+C4 dimerization but also by C1+C2

5984

J. Phys. Chem. C, Vol. 114, No. 13, 2010

dimerization. In addition, the activation energies for isomerization, β-scission, and deprotonation are generally lower for larger molecular fragments than smaller ones. That means as long as the pore permits required space, larger molecular fragments would be formed before they are converted into smaller species. Finally, it is important to state that our work only considers the static, temperatureless conversions, which are changes of reaction enthalpy. The changes of reaction entropy are not discussed. Recent work by Hafner et al.51 has demonstrated that propane dehydrogenation catalyzed by acidic chabazite occurs via formation of various forms of propyl cations stabilized by entropy, while formation of an alkoxy species is a relatively rare event. Acknowledgment. Financial support from the National Science Foundation of China (No. 10676021) and the National Basic Research Program of China (No. 2007CB209701) is gratefully acknowledged. We also thank Dr. Wan-Bin Zhang for valuable advice and discussions. Supporting Information Available: Data of S value testing and additional structural parameters for optimized transition and intermediate states on the HZSM-5 and HFAU models. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Ladwig, P. K.; Asplin, J. E.; Stuntz, G. F.; Wachter, W. A.; Henry, B. E. US Patent 6 069 287, 2000, assigned to Exxon Research and Engineering Corporation. (2) Zhao, G. L.; Teng, J. W.; Xie, Z. K.; Jin, W. Q.; Yang, W. M.; Chen, Q. L.; Tang, Y. J. Catal. 2007, 248, 29–37. (3) Byggningsbacka, R.; Kumar, N.; Lindfors, L. -E. J. Catal. 1998, 178, 611–620. (4) Zhu, X. X.; Liu, S. L.; Song, Y. Q.; Xu, L. Y. Appl. Catal., A 2005, 288, 134–142. (5) Zhu, X. X.; Liu, S. L.; Song, Y. Q.; Xie, S. J.; Xu, L. Y. Appl. Catal., A 2005, 290, 191–199. (6) Rutenbeck, D.; Papp, H.; Ernst, H.; Schwieger, W. Appl. Catal., A 2001, 208, 153–161. (7) Peng, L. M.; Huo, H.; Liu, Y.; Grey, C. P. J. Am. Chem. Soc. 2007, 129, 335–346. (8) Fermann, J. T.; Moniz, T.; Kiowski, O.; Mcintire, T. J.; Auerbach, S. M.; Vreven, T.; Frisch, M. J. J. Chem. Theory Comput. 2005, 1, 1232– 1239. (9) Frash, M. V.; Kazansky, V. B.; Rigby, A. M.; van Santen, R. A. J. Phys. Chem. B 1998, 102, 2232–2238. (10) Boronat, M.; Viruela, P.; Corma, A. Phys. Chem. Chem. Phys. 2001, 3, 3235–3239. (11) Boronat, M.; Viruela, P.; Corma, A. J. Phys. Chem. A 1998, 102, 982–989. (12) Tantanak, D.; Limtrakul, J.; Gleeson, M. P. J. Chem. Inf. Model. 2005, 45, 1303–1312. (13) van Santen, R. A.; Kramer, G. J. Chem. ReV. 1995, 95, 637–660. (14) Svelle, S.; Kolboe, S.; Swang, O. J. Phys. Chem. B 2004, 108, 2953–2962. (15) Correa, R. J.; Mota, C. J. A. Phys. Chem. Chem. Phys. 2002, 4, 375–380. (16) Benco, L.; Hafner, J.; Hutschka, F.; Toulhoat, H. J. Phys. Chem. B 2003, 107, 9756–9762. (17) Limtrakul, J.; Nanok, T.; Jungsuttiwong, S.; Khongpracha, P.; Truong, T. N. Chem. Phys. Lett. 2001, 349, 161–166. (18) Namuangruk, S.; Tantanak, D.; Limtrakul, J. J. Mol. Catal. A: Chem. 2006, 256, 113–121.

Sun et al. (19) Sinclair, P. E.; de Vries, A.; Sherwood, P.; Catlow, C. R. A.; van Santen, R. A. J. Chem. Soc., Faraday Trans. 1998, 94, 3401–3408. (20) Boronat, M.; Viruela, P. M.; Corma, A. J. Am. Chem. Soc. 2004, 126, 3300–3309. (21) Nieminen, V.; Sierka, M.; Murzin, D. Yu.; Sauer, J. J. Catal. 2005, 231, 393–404. (22) Joshi, Y. V.; Thomson, K. T. J. Catal. 2005, 230, 440–463. (23) Vreven, T.; Morokuma, K. J. Comput. Chem. 2000, 21, 1419–1432. (24) Vreven, T.; Mennucci, B.; da Silva, C. O.; Morokuma, K.; Tomasi, J. J. Chem. Phys. 2001, 115, 62–72. (25) Vreven, T.; Byun, K. S.; Koma´romi, I.; Dapprich, S.; Montgomery, J. A.; Morokuma, K.; Frisch, M. J. J. Chem. Theory Comput. 2006, 2, 815– 826. (26) Kazansky, V. B.; Subbotina, I. R.; Jentoft, F. J. Catal. 2006, 240, 66–72. (27) Namuangruk, S.; Pantu, P.; Limtrakul, J. ChemPhysChem 2005, 6, 1333–1339. (28) Hay, P. J.; Redondo, A.; Guo, Y. Catal. Today 1999, 50, 517– 523. (29) Zygmunt, S. A.; Curtiss, L. A.; Zapol, P.; Iton, L. E. J. Phys. Chem. B 2000, 104, 1944–1949. (30) Hill, J. -R.; Freeman, C. M.; Delley, B J. Phys. Chem. A 1999, 103, 3772–3777. (31) de Vries, A. H.; Sherwood, P.; Collins, S. J.; Rigby, A. M.; Rigutto, M.; Kramer, G. J. J. Phys. Chem. B 1999, 103, 6133–6141. (32) Olson, D. H.; Dempsey, E. J. Catal. 1969, 13, 221–231. (33) Maseras, F.; Morokuma, K. J. Comput. Chem. 1995, 16, 1170– 1179. (34) Morokuma, K. Bull. Korean Chem. Soc. 2003, 24, 797–801. (35) Vreven, T.; Byun, K. S.; Koma´romi, I.; Dapprich, S.; Montgomery, J. A.; Morokuma, K.; Frisch, M. J. J. Chem. Theory Comput. 2006, 2, 815– 826. (36) Rappe, A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A., III; Skiff, W. M. J. Am. Chem. Soc. 1992, 114, 10024–10035. (37) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652. (38) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03 ReVision C. 02 ed.; Gaussian Inc.: Wallingford, CT, 2004. (39) Panyaburapa, W.; Nanok, T.; Limtrakul, J. J. Phys. Chem. C 2007, 111, 3433–3441. (40) Jansang, B.; Nanok, T.; Limtrakul, J. J. Phys. Chem. C 2008, 112, 540–547. (41) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157–167. (42) Zhao, Y.; Gonza´lez-Garcı´a, N.; Truhlar, D. G. J. Phys. Chem. A 2005, 109, 2012–2018. (43) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215–241. (44) Zhao, Y.; Truhlar, D. G. J. Chem. Phys. 2006, 125, 194101–194118. (45) Zhao, Y.; Truhlar, D. G. J. Phys. Chem. A 2006, 110, 13126–13130. (46) Zhao, Y.; Schultz, N. E.; Truhlar, D. G. J. Chem. Theory Comput. 2006, 2, 364–382. (47) Rigby, A. M.; Kramer, G. J.; van Santen, R. A. J. Catal. 1997, 170, 1–10. (48) Pantu, P.; Boekfa, B.; Limtrakul, J. J. Mol. Catal. A: Chem. 2007, 277, 171–179. (49) Gutmann, V. The Donor-Acceptor Approach to Molecular Interaction; Plenum Press: New York, 1978. (50) Tuma, C.; Sauer, J. Phys. Chem. Chem. Phys. 2006, 8, 3955–3965. (51) Bucˇko, T.; Benco, L.; Dubay, O.; Dellago, C.; Hafner, J. J. Chem. Phys. 2009, 131, 214509.

JP910617M