Andzelm - Dft Methanol Conv To Hc In Zeolite

DFT Study of Methanol Conversion to Hydrocarbons in a Zeolite Catalyst JAN ANDZELM, NIRANJAN GOVIND, GEORGE FITZGERALD, AMITESH MAITI Accelrys Inc., 9685 Scranton Road, San Diego, California 92121 USA Received 19 October 2001; accepted 9 March 2002 DOI 10.1002/qua.10417

ABSTRACT: First-principles calculations using the density functional theory code DMol3 were performed to investigate important pathways in the methanol-to-gasoline conversion process over zeolite catalysts. Reaction paths and energy barriers involving the COO bond cleavage and the first COC bond formation were explored using allelectron periodic supercell calculations and newly implemented algorithms for the optimization of intermediates and transition states. The simulations indicate that the formation of surface ylide involves prohibitively high barriers, whereas surface methoxyl species can easily react with methanol to form ethanol. © 2002 Wiley Periodicals, Inc. Int J Quantum Chem 91: 467– 473, 2003

Introduction

T

he conversion of methanol to gasoline (MTG) in zeolites is an important industrial process that has attracted considerable interest from both industrial and academic researchers. The main interest is in the mechanism of the first COC bond formation and cleavage of the COO bond of methanol. Several reaction mechanisms have been proposed, and some have supporting experimental evidence [1–3]. A commonly accepted mechanism involves the initial physisorption of methanol at a Brønsted acid site of the Al-substituted zeolite [4]. As the concentration of methanol increases, clusters of hydrogen-bonded methanol molecules are Correspondence to: J. Andzelm; e-mail: JWA@ACCELERYS. COM

International Journal of Quantum Chemistry, Vol 91, 467– 473 (2003) © 2002 Wiley Periodicals, Inc.

formed in the zeolite cage [5, 6]. At that stage dimethyl ether can be formed [7], as confirmed by experiment [1]. However, it is unclear whether dimethyl ether is a necessary intermediate for the first COC bond formation [8]. As for the cleavage of the COO bond of methanol, it can occur through the formation of surface methoxyl species [5, 9 –11]. This reactive species can then be a starting point for the formation of the initial COC bond in a reaction leading to the formation of ethanol or ethyl-methylether [12]. Those intermediates can be easily dehydrated, thereby producing ethylene, which upon further reactions can lead to the formation of higher olefins, alkanes, etc. [12]. More recently, yet another mechanism for the COC bond formation was proposed: this involves the formation of a surface ylide neighboring a Brønsted acid site [8, 11]. In principle, the surface ylide species could form from the

ANDZELM ET AL. methoxyl species by the transfer of a proton to a neighboring bridging oxygen. However, it is also possible that the ylide carbon atom is incorporated into the zeolite framework, leading to a significant reorganization of a large number of atoms around the Brønsted acid site. Most first-principles computational studies employ cluster models to represent the neighborhood of the Brønsted acid site of a zeolite [3, 5, 8 –11]. There are only a few studies using periodic boundary conditions to represent realistic crystalline environment for the MTG reaction [4, 6, 7]. The plane wave code CASTEP, for instance, has previously been used to study the physisorption and clustering of methanol molecules leading to the formation of dimethyl ether [6, 7]. Here we report the first periodic calculations investigating the formation of a surface methoxyl species and the formation of an initial COC bond within a zeolite cage, ultimately leading to ethanol. The possibility of ylide formation close to a Brønsted acid site is investigated, as well. Density functional theory (DFT) calculations were performed using the DMol3 [13, 14] program. Recently developed algorithms to find transition states [15] and to optimize structures of a crystal [16] were employed to investigate reaction mechanisms. The accuracy of DMol3 to predict structures of adsorbed species in zeolites was verified in calculations on hydrogen-bonded model systems and bridging hydroxyl groups within the zeolite cage.

Computational Details The DFT calculations reported here were performed using the DMol3 program from Accelrys (Material Studio 2.0) [13, 14]. The electronic wavefunctions are expanded in atom-centered basis functions defined on a dense numerical grid. We used the double-numeric-polarized (DNP) [13] basis set, and a fine level of integration grid, amounting to approximately 5500 grid points per atom. The DNP, all-electron basis set is composed of two numerical functions per valence orbital, supplemented with a polarization function. Each basis function was restricted to within a cutoff radius of Rcut ⫽ 4.0 Å, thereby allowing for efficient calculations without a significant loss of accuracy. The electron density was approximated using a multipolar expansion up to hexadecapole. Typical overbinding associated with local density functionals was rectified through the use of the gradient-corrected Perdew-Burke-Ernzerhof (PBE) functional [17]. 468

The Brillouin-zone integration was performed using a 2 ⫻ 2 ⫻ 2 Monkhorst–Pack (MP) grid [18]. All geometries were optimized using the recently developed scheme based on delocalized internal coordinates generalized to periodic boundary conditions [16]. It was already demonstrated that for systems such as zeolites this algorithm can be several times more efficient than the one based on Cartesian coordinates. Methanol and water molecules in zeolite pores present additional challenge for the optimization algorithm, because such species are typically not connected to the zeolite framework. However, the DMol3 optimizer allows for a treatment of such disconnected fragments without having to introduce any artificial connecting bonds. Several reaction paths investigated here require approximate scanning of the potential energy surface. That was accomplished by using internal constraints that were imposed on connected or disconnected fragments. Precise determination of the transition states (TS) involving several species in concerted-like reactions in zeolite is obviously a formidable challenge. The traditional method of calculating a Hessian and following the reaction mode can be very costly for any zeolites of practical importance. Moreover, the success of such calculations is not guaranteed unless the initial structure is already close to the transition state. In this article we employ a recently developed method [15] that blends a generalization of the synchronous transit (ST) method of Halgren and Lipscomb with the conjugate gradient (CG) technique. In this approach only structures of reactants and products are needed to locate TS via a series of ST/CG steps. This method was found to be robust, efficient, and accurate in calculations for many complicated TS of reactions in the gas as well as the condensed phase.

Results HYDROGEN BONDING The mechanism of MTG reaction is determined to a large extent by the network of hydrogen bonds connecting methanol, water molecules, and the H atom at the Brønsted acid site. A quantitatively accurate treatment of hydrogen bonds is, therefore, important for this work. To test the accuracy of DMol3 in describing H-bond strengths, the interaction energies between two methanol molecules and between two water molecules were computed using the PBE/DNP/Rcut ⫽ 4.0 Å settings, as described in the previous section. The calculated valVOL. 91, NO. 3

METHANOL CONVERSION TABLE I ______________________________________________________________________________________________ Relative energies and selected geometrical parameters for the four bridging OH groups in H-Faujasite (FAU).a

Occupation site O1H O2H O3H O4H a b

Energy (kJ/mol)

rAlH (Å)

␣SiO(H)A1 (deg)

PBE/DNP

VWN/DNPb

2.54 (2.48 ⫾ 0.4) 2.43 2.46 (2.40 ⫾ 0.4) 2.44

130.1 (135.7) 141.1 (144.6) 136.8 (139.8) 135.9 (141.9)

0 7.8 4.1 9.4

0 9.8 4.9 7.9

Experimental values are given in parentheses [22, 23]. Results of Ref. [23].

ues of 5.8 and 5.7 kcal/mol for methanol and water dimers compare well with experimentally estimated ranges of 4.6 –5.9 and 5.0 –5.4 kcal/mol [19], respectively. This good agreement is due to the fact that the interaction is dominated by the electrostatics, and DFT predicts dipole moments for both methanol and water with reasonable accuracy [20]. Dispersion forces that are missing in DFT [21] and contribute perhaps as much as 25% to the total interaction energy of water dimer [19] are apparently compensated by other terms in the interaction energy. The standard procedure in ab initio calculations of interaction energies uses the so-called basis set superposition error (BSSE) approach [19] to account for incomplete atomic basis sets. The DMol3 program uses numerical functions that are far more complete than the traditional Gaussian functions, and therefore we expect BSSE contribution to be small. We verified this by performing DMol3 calculations on a water dimer using a much larger basis set (five numerical functions per every valence orbital, supplemented with three sets of polarization functions, diffuse functions, and Rcut ⫽ 8 Å). This calculation affects the interaction energy by less than 0.6 kcal/mol. Therefore, indirectly, we demonstrate that any BSSE contributions or further basis set extensions have a small effect of a fraction of kcal/mol on the interaction energy. We are confident that the PBE/DNP/Rcut ⫽ 4.0 Å DMol3 calculations are an accurate and practical approach to study complicated reactions in the unit cell of zeolites. ZEOLITE MODELS AND BRIDGING HYDROXYL GROUPS Most commercially used zeolite structures are too large for a detailed study of all reactions involved in the MTG process. The unit cell of a ZSM-5

catalyst, for instance, has 288 atoms. In this article we use the Ferrierite (FER) structure that has eightand 10-ring channels—somewhat similar to the 10rings channels of ZSM-5. The primitive cell of FER contains only 54 atoms and makes the calculations feasible. Recent calculations by Haase and Sauer [4] on the physisorption of a single methanol molecule in the FER and ZSM-5 zeolites show similar complexes in both zeolites. The methanol adsorption energies in both zeolites differ by less than 4 kcal/ mol, which is only 15% of the adsorption energy. Most results from the calculations on a FER zeolite are therefore expected to be valid for a commercial zeolite such as ZSM-5. To assess the accuracy of the DMol3 settings in a periodic geometry of a zeolite, we performed calculations on bridging hydroxyl groups at the alumino-silicate Brønsted acid site of Faujasite (FAU) and FER zeolites. Experimental studies [22] on the FAU zeolite have revealed that only three of the four possible bridging OH groups are observed, and their relative occupations are as follows: 3:1:1.6:0 for O1H:O2H:O3H:O4H sites, respectively. In the work by Hill et al. [23] DMol3 was applied at the Vosko-Wilk-Nusait (VWN)/ DNP level using unit cell parameters obtained from the shell model. Hill et al. [23] found excellent correlation between the relative energies of the four bridging OH groups and the relative experimental occupations. In the current work full geometry optimization was performed for the primitive rhombohedral cell with 145 atoms and fixed cell parameters. Table I summarizes our results as compared with the results of Ref. [23]. It is evident that the relative energies from our calculations correspond well with the experimental site occupations as well as with previous DMol3 calculations. The bond lengths and bond angles associated with bridging hydroxyl groups

INTERNATIONAL JOURNAL OF QUANTUM CHEMISTRY

469

ANDZELM ET AL.

2 FIGURE 1. The SN pathway for surface methylation with single methanol. Dotted lines represent hydrogen bonds. (a) Methanol-hydrogen complex at Brønsted acid site; (b) transition state; (c) surface methoxyl and water hydrogenbonded to zeolite.

are within 0.06 Å and 6° of experimental values. The O1H site, which is open to the largest channel of a zeolite, is a preferable location for the hydroxyl group in both FAU and FER zeolites. In the case of FER zeolite, the relative energies are 0, 7.5, 9.5, and 14.2 kJ/mol for the O1H, O2H, O3H, and O4H sites, respectively. The O1H, being the lowest energy site, was selected as the Brønsted acid site for our MTG study with the FER zeolite. FORMATION OF SURFACE METHOXYL SPECIES A single molecule of methanol forms multiple hydrogen bonds with the Brønsted acid site of FER zeolite. There are several adsorption sites possible that are within 1 kcal/mol; Figure 1(a) displays the most stable structure, with an adsorption energy of 18.5 kcal/mol. This value compares well the experimental estimates of the heat of adsorption in acidic zeolites, ranging from 15 to 27 kcal/mol [5, 24]. No protonation of methanol by the Brønsted acid site was found, in agreement with the recent study by Haase and Sauer [4]. The calculated transition state for the methylation of a surface oxygen at the alumino-silicate Brønsted acid site of FER zeolite is presented in Fig. 1(b). Clearly this is a concerted reaction involving breaking of the COO bond in methanol and bond formation between C and surface oxygen. Simultaneously the proton from the Brønsted acid site is transferred to the hydroxyl 470

group, thereby forming a water molecule [Fig. 2 1(c)]. This is a strained SN -type reaction with an activation barrier of 54 kcal/mol. Other DFT or ab initio cluster studies [9 –11] have reported a barrier of 44 to 65 kcal/mol. Interestingly, the presence of a second methanol molecule lowers the above activation barrier to 44 kcal/mol, as the 2 TS now corresponds to an unstrained SN pathway (see Fig. 2). At the TS the surface oxygen, methyl carbon, and oxygen of the leaving water molecule are roughly collinear [Fig. 2(b)]. The water molecule is formed as a result of proton transfer to the methanol hydroxyl group from the methoxonium ion. Our computed barrier of 44 kcal/mol is comparable with the 32– 46 kcal/mol barrier predicted by the MP2/6-31G*//HF/3-21G studies of Sinclair and Catlow [5] on clusters of various sizes. This reaction is facilitated by the hydrogenbonded methanol and methoxonium ion [Fig. 2(a)]. Methoxonium ion is formed spontaneously when a methanol molecule captures the proton from the Brønsted acid site. No barrier for that reaction is found, in agreement with the work by Sandre et al. [7]. Clearly, the presence of the second methanol molecule in the zeolite cage facilitates methoxonium ion formation because with one methanol molecule, no spontaneous deprotonation of zeolite was found. Further analysis of other possible pathways for surface methoxyl formation is given in a separate study. VOL. 91, NO. 3

METHANOL CONVERSION

2 FIGURE 2. The SN pathway for surface methylation with two methanol molecules: (a) methoxonium–methanol com-

plex; (b) transition state; (c) surface methoxyl and methanol–water complex.

REACTIONS OF SURFACE METHOXYL SPECIES Surface methoxyl species can react with methanol or dimethyl ether to form ethanol or methylethyl-ether, respectively [12]. Those are the first species containing a COC bond. In this study we investigated the formation of ethanol. Using the structure presented in Figure 3(a) as a reactant for the COC bond formation, we investigated several

possible pathways, which are discussed in details elsewhere. The most successful pathway involved water as an important catalytic agent, which through the network of hydrogen bonds stabilized ethanol molecule and facilitated transfer of the proton from the methyl group of the methanol to the Brønsted acid site of zeolite [Figs. 3(b) and 3(c)]. The importance of the water molecule in assisting in this reaction was already noticed by Blasz-

FIGURE 3. Ethanol formation via a methoxyl/water–mediated mechanism. (a) Surface methoxyl and water–methanol complex; (b) transition state; (c) ethanol, water, and hydrogen bonded to zeolite.

INTERNATIONAL JOURNAL OF QUANTUM CHEMISTRY

471

ANDZELM ET AL.

FIGURE 4. Formation of the surface ylide species: (a) surface methoxyl species; (b) transition state; (c) surface ylide species.

kowski and van Santen [12]. The transition state 2 [Fig. 3(b)] indicates that this is an SN -type reaction involving CH3 species already separated from the zeolite surface with structure close to the trigonal planar. Our computed barrier for that reaction is 25.4 kcal/mol, which is significantly less than the 60 kcal/mol barrier reported from the cluster calculations of Ref. [12]. The net reaction is exothermic and the final product adsorbed via hydrogen bonds [Fig. 3(c)] to the zeolite surface has an energy lower than the reactant by 49.2 kcal/mol. The corresponding reaction energy from Ref. [12] is about 24 kcal/ mol. This disparity can be attributed to differences in (1) the choice of the model: cluster models were used in Ref. [12], as compared with the periodic models used in this work; and (2) the method: self-consistent GGA calculations were performed here, whereas only perturbative GGA corrections were used in Ref. [12]. From comparing activation barriers for methoxyl formation we can conclude that the rate-limiting step is the breaking of the COO bond of methanol and the formation of the surface methoxyl species [Fig. 2(b)]. The activation energy for such reaction, 44 kcal/mol, is significantly higher than the barrier for the first COC bond formation of 25 kcal/mol [Fig. 3(b)], as found in this work. Finally, we investigated a recent hypothesis [8, 25] that the surface ylide (CH2) may provide a reactive C atom for the first COC bond formation. According to that hypothesis, the methoxyl species loses a proton, which is transferred to the neighboring Brønsted acid site. Our initial attempts to find a stable structure with a surface ylide separated from 472

a proton were not successful. The proton always comes back, without any barrier to form a stable methoxyl species. However, if there is a rearrangement of the zeolite framework and the ylide (CH2) species is inserted into the AlOO bond, a stable intermediate structure is indeed possible. Cluster calculations [8, 25] reveal that the barrier for such reaction methoxyl f ylide built in surface is about 50 kcal/mol. We expect a small cluster model to be inadequate to study reactions that may significantly affect the entire framework of a zeolite. We investigated this reaction (Fig. 4) using a consistent periodic zeolite model. The calculated barrier of 78.5 kcal/mol is significantly larger than our previously calculated barriers for methoxyl (44 kcal/mol) and the ethanol (25 kcal/mol) formation. Therefore, we can rule out the possibility of a surface ylide formation. One can notice a significant reorganization of the zeolite framework upon surface ylide formation, underlining a need for consistent calculations in a periodic environment.

Conclusion Using a DMol3 PBE/DNP approach we performed calculations on periodic FER zeolite to elucidate the mechanism of key reactions involved in the first COO bond breaking and COC bond formation for the MTG process. This is a consistent approach that allows us to study significant reorganization of the zeolite cages, which may occur if an adsorbed species is built into the zeolite frameVOL. 91, NO. 3

METHANOL CONVERSION work. The main findings of this work are summarized below: ▪ ▪

▪

▪

▪

▪

▪

The current approach allows us to calculate hydrogen bonding with a reasonable accuracy of ⬃0.5 kcal/mol for interaction energies. Periodic DFT calculations can describe well the neighborhood of a Brønsted acid site in alumino-silicate Faujasite zeolite. The experimentally known positions of the bridging hydroxyl groups and their relative stabilities are correctly reproduced in our calculations. Single methanol molecule is adsorbed in zeolite cages via hydrogen bonds, whereas the presence of the two methanol molecules allows for spontaneous formation of the methoxonium ion. Formation of surface methoxyl species occurs 2 in SN -type concerted reaction with a barrier of 44 kcal/mol if two methanol molecules are present. Surface methoxyl species can undergo a reaction to form an ethanol molecule. The barrier for such a reaction is 25 kcal/mol if the water molecule is present. Formation of the ylide that is built into the zeolite cage surface has to be ruled out as the barrier for such a reaction exceeds 78 kcal/ mol. The synchronous transit coupled with conjugated gradient refinement to optimize transition states [15] and delocalized internals to optimize structures of reaction intermediates [16], under periodic boundary conditions and internal constraints, allowed fast and accurate exploration of reaction pathways and associated heats and barriers.

Further details of the calculations and structures of transition states will be published elsewhere. The next step in our study of MTG process is likely to include investigations on the role of ether and the process of dehydration of ethanol to form ethylene. ACKNOWLEDGMENT The authors are grateful to Prof. Joachim Sauer (Humboldt University, Berlin) and Prof. Richard

Catlow (Royal Institute of Great Britain, London) for valuable discussions. We thank our colleagues at Accelrys for help in this study and acknowledge members of Accelrys’s Catalysis 2000 Consortium for their support of this work.

References 1. Chang, C. D.; Silvestri, A. J. J Catal 1977, 47, 249. 2. Sauer, J. Chem Rev 1989, 89, 199. 3. Hutchings, G. J.; Hunter, R. Catal Today 1990, 6, 279. 4. Haase, F.; Sauer, J. Micro Meso Mat 2000, 35, 379. 5. Sinclair, P. E.; Catlow, C. R. A. J Chem Soc Faraday Trans 1996, 92, 2099. 6. Shah, R.; Gale, J. D.; Payne, M. C. J Phys Chem 1996, 100, 11688. 7. Sandre, E.; Payne, M. C.; Gale, J. D. Chem Commun 1998, 22, 2445. 8. Hutchings, G. J.; Watson, G. W.; Willock, D. J. Micro Meso Mat 1999, 29, 67. 9. Blaszkowski, S. R.; van Santen, R. A. J Phys Chem 1995, 99, 11728. 10. Zicovich-Wilson, C. M.; Viruela, P.; Corma, A. J Phys Chem 1995, 99, 13224. 11. Sinclair, P. E.; Catlow, C. R. A. J Chem Soc Faraday Trans 1997, 93, 333. 12. Blaszkowski, S. R.; van Santen, R. A. J Am Chem Soc 1997, 119, 5020. 13. Delley, B. J Chem Phys 1990, 92, 508; J Phys Chem 1996, 100, 6107; J Chem Phys 2000, 113, 7756. 14. DMol3, Materials Studio version 2.0 from Accelrys: http:// www.accelrys.com/mstudio/dmol3.html. 15. Govind, N.; Fitzgerald, G.; King-Smith, D. to be published. 16. Andzelm, J.; King-Smith, D.; Fitzgerald, G. Chem Phys Lett 2001, 335, 321, and work that is to be published. 17. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys Rev Lett 1996, 77, 3865. 18. Monkhorst, H. J.; Pack, J. D. Phys Rev B 1976, 13, 5188. 19. Szalewicz, K.; Jeziorski, B. In Scheiner, S., Ed. Molecular Interactions; John Wiley & Sons: New York, 1997; p. 3. 20. Klamt, A.; Jonas, V.; Burger, T.; Lohrenz, J. J Chem Phys J Phys Chem 1998, 102, 5074. 21. Kristian, S.; Pulay, P. Chem Phys Lett 1994, 175, 229. 22. Czjzek, M.; Jobic, H.; Fitch, A.; Vogt, T. J Phys Chem 1992, 96, 1535. 23. Hill, J.-R.; Freeman, C.; Delley, B. J Phys Chem A 1999, 103, 3772. 24. Haase, F.; Sauer, J. J Am Chem Soc 1995, 117, 3780. 25. Sinclair, P. E.; Catlow, C. R. A. J Phys Chem 1997, 101, 295.

INTERNATIONAL JOURNAL OF QUANTUM CHEMISTRY

473