1871
2007, 111, 1871-1873 Published on Web 01/12/2007
Theoretical Study of Molecular Hydrogen Adsorption in Mg-Exchanged Chabazite F. Javier Torres, Bartolomeo Civalleri,* Alexander Terentyev, Piero Ugliengo, and Cesare Pisani Dipartimento di Chimica IFM, and NIS - Nanostructured Interfaces and Surfaces - Centre of Excellence, UniVersita` di Torino, Via P. Giuria 7, 10125 Torino, Italy ReceiVed: NoVember 16, 2006; In Final Form: December 18, 2006
An enthalpy of adsorption of ca. -10 kJ/mol was theoretically predicted for the interaction of molecular hydrogen with Mg-exchanged low silica chabazite (Si/Al ) 5/1), which is significantly higher than the theoretical prediction for alkali-metal-exchanged chabazites (-3.0/-7.0 kJ/mol) and in acceptable agreement with a recent experimental finding of -17.5 kJ/mol obtained for a different zeolite, that is, (Mg,Na)-Y.
A key element in the hydrogen economy is an efficient method of molecular hydrogen storage that allows a proper bridge between production and use.1 In this sense, intensive research has been done to determine the most adequate storage option, and among the proposed methods, hydrogen physisorption in microporous materials represents one of the most interesting. Various materials have been the object of analysis for the role of the host in the adsorption process. Carbon-based materials,2,3 metal-organic frameworks,4,5 and zeolites6,7 are some of them, with the latter ones being interesting candidates, at least for fixed plants, because of their low cost and thermal and mechanical stability. However, reported experimental8,9 and theoretical10-13 studies have found that the interaction energy of hydrogen molecules with the polarizing centers of zeolites such as alkali-metal ions is not strong enough (in the -5.0/9.0 kJ/mol range) to ensure an adequate adsorption-desorption cycle. Very recently, G. Turnes Palomino et al.14 reported experimental evidence that the incorporation of Mg2+ in zeolite Y improves its molecular hydrogen sorption capacity dramatically. The adsorption of H2 on the Mg-partially exchanged Na-Y zeolite was studied by means of variable-temperature FTIR spectroscopy and the enthalpy of adsorption (∆Hads) was estimated to be -17.5 kJ/mol, which is the highest value, reported so far, for zeolitic materials. In this letter, we report on the theoretical investigation of the hydrogen adsorption on Mg-exchanged chabazite with Si/ Al ) 5/1 (MgCHA-5/1) to give support to the experimental observations and provide, from an atomistic point of view, insight into the role of the different energetic and geometrical aspects in the interaction. At variance with the Y zeolite used in the experiment, chabazite was adopted in the present computational work because of its smaller unit cell that allows one to save computer resources. Initially, the periodic model of an H2-free Mg-CHA-5/1 chabazite was derived, starting from the all-silica chabazite structure (i.e., unit cell consisting of twelve tetrahedra that form hexagonal prisms connected by fourmembered rings). The required Si/Al ratio was achieved by the * Corresponding author. E-mail:
[email protected].
10.1021/jp067631l CCC: $37.00
Figure 1. In the upper part, a schematic representation of MgCHA5/1 chabazite with D-1 and D-2 aluminum distributions is depicted. The SI, SII, and SIII positions considered as cationic sites for the Mg2+ are also shown. The clusters, cut of from the periodic structure and adopted as model systems, are depicted in the lower part.
substitution of 2 out of 12 silicon atoms by aluminum atoms. Two different aluminum distributions were considered, here referred as D-1 and D-2. As shown in the upper part of Figure 1, D-1 envisages two aluminum atoms farthest apart in the hexagonal prism, whereas for D-2 the aluminum atoms are at the two extreme vertices of the same six-membered ring. For both structures, a Mg2+ cation was added as a charge balancer. Three different positions for the Mg2+ cation were considered in the chabazite cavity, namely, SI, SII, and SIII, in which the Mg2+ cation is located in the four-, six-, and eight-membered rings of the framework, respectively (see Figure 1). For each cationic site, the structure was fully relaxed by using the © 2007 American Chemical Society
1872 J. Phys. Chem. C, Vol. 111, No. 5, 2007
Letters
TABLE 1: Optimal Distance from the Centers of Mass of the H2 Molecule to the Mg2+ Cation (RH2-Mg), Change in the H-H Interatomic Distance ∆dH-H, Shift in the H-H Anharmonic Stretching Frequency (∆ν), and Corrected Binding Energy (BEc) in the H2-MgCHA-5/1 Periodic Structures with Different Aluminum Distributionsa aluminum distribution
RH2-Mg
∆dH-Hb
∆νc
BEc
D-1 D-2
2.3 2.3
0.003 0.004
-66 -67
12.7 12.0
a Distances in Å, frequency shifts in cm-1, and energies in kJ/mol. Difference with respect to the H-H distance equal to 0.762 Å computed in the free molecule. c Difference with respect to the H-H anharmonic stretching frequency equal to 4138 cm-1 computed in the free molecule.
b
CRYSTAL06 code.15 The B3LYP Hamiltonian16 together with an all-electron double-ζ quality Gaussian-type basis set was employed as the level of theory for these calculations (referred as B3LYP/BSA in the following). The adopted BSA for the framework atoms, namely, Si, Al, and O, is the same as that used in our previous work dedicated to the study of molecular hydrogen interaction with Li2CHA-5/1 and LiNaCHA-5/1.17 For the Mg and H atoms, the basis set is an 8-411G(d) contraction for the former and the standard Dunning’s aug-cc-pVDZ basis for the latter. The energy of the optimal B3LYP/BSA geometries showed that the relative stability of the different Mg2+ sites followed the trend SII > SIII > SI for both D-1 and D-2 distributions. The difference in stability between SII and SIII is more than 100 kJ/mol; thus, only the former case was considered as a candidate for the H2 adsorption. The relative stability of D-1 and D-2 showed that D-2 is only 20 kJ/mol more stable than D-1; therefore, both distributions were considered for the study of the H2 interaction. The H2 molecule was positioned close to the Mg2+ ion for both cases, and only the H2 and Mg coordinates were reoptimized, while freezing the other degrees of freedom. This is justified because of the expected negligible changes in the framework positions due to the weak H2 interaction. Table 1 shows that the optimal distance between the H2 centers of mass and the Mg2+ was ∼2 Å in both cases. The H-H interatomic distance increased only slightly (see Table 1), indicating a nonchemical interaction.18 For the optimal geometries, the anharmonic H-H stretching frequency was evaluated by solving the one-dimensional Schro¨dinger equation by considering the H-H bond as an independent oscillator.19 The resulting H2 frequency shifts (computed with respect to the free H2) are also reported in Table 1. In both cases, bathochromic shifts close to -70 cm-1 were computed showing that the Mg2+ cation has a larger effect on the H2 stretching mode in comparison with Li+ and Na+, for which bathochromic shifts of -50 and -43 cm-1 were computed, respectively.17 This is already an indication of the stronger polarizing power of Mg2+ compared to Li+ and Na+, as expected. Compared to the experimental value14 of -110 cm-1, the B3LYP value appears in acceptable agreement. The
binding energies were computed as
BE[H2-MgCHA-5/1] ) E[MgCHA-5/1] + E[H2] - E[H2-MgCHA-5/1] (1) These values were corrected for the basis set superposition error by using the Boys-Bernardi counterpoise method.20 Because of the well-known deficiency of the density functional methods to cope with dispersive forces,21,22 the BEc values reported in Table 1 are expected to be somehow underestimated. To include, at least partially, the missed dispersion contribution to the binding energy, an ONIOM2-like approach was adopted, in which the MP2 method23 (capable to account for a fraction of dispersive interactions) was used as the HIGH level of theory. The details of the method were already discussed in our previous studies,10,11,17 and here we only report the ONIOM2 definition of binding energy
BE[ONIOM2] ) BEc[LOW,Real] + (BEc[HIGH,Model] - BEc[LOW,Model]) (2) where BEc[HIGH,Model], BEc[LOW,Model], and BEc[LOW,Real] are the BSSE corrected binding energies, obtained as shown in eq 1, for the model system at the HIGH level of theory and for both the real and model systems at the LOW level of theory. In the present study, the B3LYP/BSA level26 was adopted as the LOW level of theory, whereas B3LYP/aug-cc-pVDZ and MP2/augcc-pVDZ were used as HIGH1 and HIGH2 levels of theory, respectively. The MP2 calculations were carried out with the Gaussian03 software package.24 The reason that these combinations are adopted is twofold: (i) HIGH1 allows one to estimate the effect of the basis set quality on the interaction energy (from BSA to full aug-cc-pVDZ); (ii) HIGH2 allows one to estimate dispersive forces by means of the use of the perturbative method MP2, assuming that the intramolecular correlation remains the same at HIGH1 and HIGH2 levels, as proposed in the past for the CO adsorption on the (001) MgO surface.25 The D-1 and D-2 clusters adopted as the model systems for the two aluminum distributions are shown in the lower part of Figure 1. The computed binding energies at LOW, HIGH1, and HIGH2 levels of theory are reported in Table 2. Comparison between the [LOW:Real] and the [LOW:Model] shows that the two values are extremely close for the D-2 cluster, whereas they differ significantly for the D-1 case. The origin of this discrepancy can be traced back to the very large value of the dipole moment of D-1 (around 30 D) compared with that of D-2 (around 2 D). This is a consequence of the truncation occurring in the adopted clusters. Indeed, the two Al ions belonging to the four-membered ring in D-1 cause the high dipole moment, whereas in D-2, the intra-ring polarity imparted by the Al ions in the six-membered ring cancels out almost completely (see the lower part of Figure 1 for details). This is confirmed by the very close values for the [LOW:Real] level (i.e., the full periodic calculation) in which no dramatic
TABLE 2: BSSE Corrected Binding Energies for the Real and Model Systems at LOW, HIGH1, and HIGH2 Levels of Theorya aluminum distribution
[LOW:Real]
[LOW:Model]
[HIGH1:Model]
ONIOM2 B3LYP
[HIGH2:Model]
ONIOM2 MP2
D-1 D-2
12.7 12.0
17.8 11.5
17.9 11.8
12.8 12.3
23.0 17.0
17.9 17.5
a The ONIOM2 binding energies are also reported. Energies in kJ/mol. [LOW:Real] ) B3LYP/BSA:crystal. [LOW:Model] ) B3LYP/BSA: cluster.26 [HIGH1:Model] ) B3LYP/aug-cc-pVDZ:cluster. [HIGH2:Model] ) MP2/ aug-cc-pVDZ:cluster.
Letters differences in the framework polarity is expected for the two Al distributions because of the crystalline nature of the system. Data in Table 2 also allow one to estimate the effect of adopting the more flexible aug-cc-pVDZ basis set on all atoms, compared to the case in which it was employed for the H2 molecule only, as done for the periodic case. Comparing [LOW:Model] with [HIGH1:Model] results in negligible differences, showing that also for periodic calculations, which are very demanding and rather delicate to run when a basis set with diffuse fuctions is adopted, the present strategy to locally improve the basis set is worthwhile. Because distorsions in the electric moments of the cluster occur for both the [LOW:Model] and the [HIGH:Model], the final ONIOM2-B3LYP results for D-1 and D-2 are in very good agreement with each other. The dispersive contribution to the interaction energy can be estimated as the difference between ONIOM-MP2 and ONIOM2-B3LYP data (see Table 2), which accounts for about 5 kJ/mol, that is, some 30% of the total interaction energy. This value appears, on an absolute scale, reasonable and smaller than that computed for CO adsorbed on the (001) MgO surface.25 To compare the present theoretical energy of interaction with the experimental heat of adsorption reported in ref 14, the zero point energy correction and the thermal corrections at 135 K were added to the ONIOM2-MP2 data of Table 2. To get these corrections, a full harmonic frequency calculation was carried out in the periodic model of the D-1 case at the B3LYP/BSA level. The estimated heat of adsorption, ∆Hads, of -9.6 kJ/mol is in acceptable agreement with the experimental value of -17.5 kJ/mol obtained for (Mg,Na)-Y. The discrepancy can be attributed to two factors: (i) the adopted corrections to the pure interaction energies are rather approximate, and (ii) the comparison between theory and experiment cannot be strict because the latter was estimated from a variable-temperature FTIR on zeolite Y, in which not all Na ions were exchanged (that is, for a different zeolitic framework and cationic content). Even though the predicted ∆Hads is underestimated, it represents a significant improvement with respect to computed values for alkali-metal-exchanged high- and low-silica chabazites.10,17 Indeed, calculated heats of adsorption increase from -3 kJ/mol for Li-CHA-11/1 to -7 kJ/mol for Li2-CHA-5/1, and finally to -10 kJ/mol for Mg-CHA-5/1. Therefore, when considering the above points, the present computed value can be considered satisfactory. This result together with the computed anharmonic H2 frequency bathochromic shifts shows that the adopted Mg-CHA-5/1 models capture the essential features of the real system (Mg-NaY) on which the measurements were carried out and reinforce the conclusion of G. Turnes Palomino et al.14 about the remarkable H2 adsorptive power of the Mgexchanged microporous materials. Acknowledgment. This work was carried out in the frame of the research project “Innovative Materials for Hydrogen Storage” supported by Regione Piemonte.
J. Phys. Chem. C, Vol. 111, No. 5, 2007 1873 References and Notes (1) Crabtree, G.; Dresselhaus, M.; Buchanan, M. Phys. Today 2004, 39. (2) Naab, F. U.; Dhoubhadel, M.; Gilbert, J. R.; Savage, L. K.; Holland, O. W.; Duggan, J. L.; McDaniel, F. D. Phys. Lett. A 2006, 352, 152. (3) Bhatia, S. K. Langmuir 2006, 22, 1688. (4) Sun, D. F.; Ma, S. Q.; Ke, Y. X.; Collins, D. J.; Zhou, H. C. J. Am. Chem. Soc. 2006, 128, 3896. (5) Rosi, N. L.; Eckert, J.; Eddaoudi, M.; Vodak, D. T.; Kim, J.; O’Keeffe, M.; Yaghi, O. M. Science 2003, 300, 1127. (6) Zecchina, A.; Bordiga, S.; Vitillo, J. G.; Ricchiardi, G.; Lamberti, C.; Spoto, G.; Bjorgen, M.; Lillerud, K. P. J. Am. Chem. Soc. 2005, 127, 6361. (7) Kazansky, V. B.; Borovkov, V. Y.; Karge, H. G. Microporous Mesoporous Mater. 1998, 22, 251. (8) Otero Area´n, C.; Manoilova, O. V.; Bonelli, B.; Rodrı´guez Delgado, M.; Turnes Palomino, G.; Garrone, E. Chem. Phys. Lett. 2003, 370, 631. (9) Stephanie-Victoire, F.; Goulay, A. M.; Cohen de Lara, E. Langmuir 1998, 14, 7255. (10) Torres, F. J.; Vitillo, J. G.; Civalleri, B.; Ricchiardi, G.; Zecchina, A. J. Phys. Chem. C, in press. (11) Torres, F. J.; Civalleri, B.; Pisani, C.; Ugliengo, P. J. Phys. Chem. B 2006, 110, 10467. (12) Otero Area´n, C.; Turnes Palomino, G.; Garrone, E.; Nachtigallova, D.; Natchtigall, P. J. Phys. Chem. B 2006, 110, 395. (13) Natchtigall, P.; Garrone, E.; Turnes Palomino, G.; Rodrı´guez Delgado, M.; Nachtigallova, D.; Otero Area´n, C. Phys. Chem. Chem. Phys. 2006, 8, 2286. (14) Turnes Palomino, G.; Llop Carayol, M. R.; Otero Area´n, C. J. Mater. Chem. 2006, 28, 2884. (15) Dovesi, R.; Saunders, V. R.; Roetti, C.; Orlando, R.; ZicovichWilson, C. M.; Pascale, F.; Civalleri, B.; Doll, K.; Harrison, N. M.; J. Bush, I. J.; D’Arco, P.; Llunell, M. CRYSTAL06 user’s manual; Universita` di Torino: Torino, Italy, 2006. (16) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (17) Torres, F. J.; Civalleri, B.; Terentyev, A.; Ugliengo, P. Phys. Chem. Chem. Phys., submitted for publication. (18) Lochan, R. C.; Head-Gordon, M. Phys. Chem. Chem. Phys. 2006, 8, 1357. (19) Ugliengo, P. Anharm - a program to solVe the one-dimensional nuclear Schro¨dinger equation; 1989. (20) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553. (21) Pe´rez-Jorda, J.; Becke, A. D. Chem. Phys. Lett. 1995, 233, 134. (22) Wu, Q.; Yang, W. J. Chem. Phys. 2002, 116, 515. (23) Møller, C.; Plesset, M. Phys. ReV. 1934, 46, 618. (24) 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.; Bakken, V.; 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 B.05; Gaussian, Inc.: Wallingford, CT, 2004. (25) Ugliengo, P.; Damin, A. Chem. Phys. Lett. 2002, 366, 683. (26) For the cluster calculations, BSA is modified by using the 31G(p) basis set for the H atoms employed to saturate the model system.
