J. Phys. Chem. 1994, 98, 3083-3085
3083
'H NMR Chemical Shifts of Ammonia, Methanol, and Water Molecules Interacting with Bransted Acid Sites of Zeolite Catalysts: Ab-Initio Calculations Frank Haase and Joachim Sauer' Max-Planck-Gesellschaft, Quantum Chemistry Group at the Humboldt University Berlin, Jiigerstrasse IOII 1, D- 101 I7 Berlin, Germany Received: November 29, 1993; In Final Form: February 3, 1994'
The theoretically predicted IH shift of ammonia adsorbed on a zeolite cluster which containes two complete coordination spheres of 0 and Si around the central aluminum is 6.8 ppm. It corresponds well to the data observed for a single adsorbed NH4+ ion. The calculated 'Hchemical shifts of the two possible structure types, the neutral and the ion-pair complex, of a single adsorbed methanol molecule are 5.7 and 15.3 ppm, respectively. For complexeswith a single water molecule thesevalues are 4.5 and 12.7 ppm, respectively. Additional calculations on the ion-pair complex with a second adsorbed neutral methanol or water molecule indicate the formation of an oxonium species only for higher loadings on the Bransted sites.
Introduction The transfer of a proton from surface hydroxyls of zeolites onto substrate molecules (hydrocarbons) is the initial step of chemical reactions catalyzed by Bransted sites. Despite considerable effort, a detailed understanding of the interaction of even such simple nucleophilic molecules like methanol, water, or ammonia with acidic sites of zeolites is still lacking. There are two possible structure types of the adsorption complex: a neutral hydrogen-bonded complex of the base molecule with the bridging hydroxyl group (Figure 1, left-handside) and an ion-pair complex which consists of an hydronium, methoxonium, and ammonium ion coordinated via two or more protons onto the negatively charged framework site (Figure 1, right-hand side). Up to now, it is not clear whether both structures are local minima on the potential energy surface (PES) separated by an energy barrier or whether only one of these structures corresponds to a minimum on thePES. Intheformercaseanequilibriumbetweentheneutral and the ion-pair structure can be assumed. In the latter case of a single potential well the minimum may correspond to either an ion-pair complex, i.e., the proton transfer occurrs in a direct process without an activation barrier, or a neutral adsorption structure which implies that the protonation does not take place. Though IR and MAS NMR spectroscopies have frequently been used in studying such interactions, an unambiguous assignment of vibrational bands or NMR peaks often proved difficult. Most of these studies assumed a single-well potential with the ion-pair structure as a local minimum.'-' While in the case of ammonia this was not doubted, very recent work on the adsorption of water favored the interpretation in terms of the neutral adsorption structure."' A third groupof authors assuma an equilibrium between these two structure^.^^-^^ In such a situation, it can be very helpful to theoretically predict the IH NMR chemicalshifts for these two types of adsorptionstructures. We performed ab-initio calculationsof 'H NMR chemical shifts of models of the bridging hydroxyl site and of their complexes with water, methanol, and ammonia. As chemical shifts are structure sensitive, it was important to make these calculations on completely optimized structures which have been obtained before and will be reported elsewhere together with predictions of the vibrational spectra.16 Calculations The calculations were performed using the TURBOMOLE program package." All structures were fully optimized at the Abstract published in Aduonce ACS Absrrocts, March 15, 1994.
Figure 1. SCF optimized neutral (left) and ion-pair structures (right) of the complexes of cluster 3with ammonia (top) and methanol (bottom).
SCF level employing a double-zeta plus polarization basis set (dzp) on all atoms except oxygen, which was described by a triplezeta plus polarization basis set (tzp). The calculations of the nuclear shielding constants used the semidirect GIAO-SCF module of TURBOMOLE.18 It is based on the coupled perturbed HartreeFock algorithm suggested by Ditchfield19 and uses gauge including atomic orbitals (GIAO). We employed a TZP basis for all atoms which originates from Huzinaga's 10s6p sets but was fully reoptimized using analytical gradients of the energy with respect to the basis set parameters.20 Models Three models of increasing size were adopted for the bridging hydroxyl site. The smallest-HO(H)Al(OH)3 (l)-consist simply of a A104 tetrahedron saturated with hydrogen atoms and the bridging hydroxyl proton. In the second cluster-H3SiO(H)A1(OH)20SiH3 (2a)-the two oxygen atoms acting as adsorption sites are bound to SiHp groups; i.e., there is a partial
0022-3654/94/2098-3083304.50~00 1994 American Chemical Society
3084 The Journal of Physical Chemistry, Vol. 98, No. 12, I994
TABLE 1: Absolute 'H Shielding Constants u ( pm) of the Reference Molecules Metbaad, Ammonia, rad d t e r from GIAO Calculations on Different Theoretical Levels and Experimental Shifts 6 (ppm) Used for Conversion reference u(SCF/TZP) u(SCF/QZ2P)O u(MP2/QZ2P)' G(exp) CHsOH
NH3 H20
32.71 32.39 3 1.68
32.06 32.01 30.96
32.05 31.91 3 1.07
0.026 0.w 0.73d
* The calculation employed the quadrupole-zeta double-polarization basis set (QZ2P) at geemetries optimized at the MP2/TZ2P level.= Reference 28. Reference 29; there are other values for gascous NH3 of 0 . 0 P and -0.31 ppm.'O Reference 29. second coordination sphere. Because in the adsorption complex of ammonia all three protons of NH3 approach the zeolite cluster (thoughonly twoofthemactually formH bonds),for thismolecule cluster 2bHpSiO(H)A10H(OSiH3)2-was used, which contains three Si atoms in the second coordination sphere. The third cluster-H&O(H)Al(OSiH3)3 (3)--contains a completesecond coordination sphere of four silicon atoms. The optimized structures of the complexes of cluster 3 with ammonia and methanol are shown in Figure 1. The optimized adsorption structures of water exhibit the same features as those of methanol. The vibrational mode analysis'6 revealed that (i) only in the case of ammonia both structures, neutral adsorption complex and ion-pair complex, are local minima and (ii) in the case of water and methanol the ion-pair structures represent saddle points (transition structures) and only the neutral complex is a minimum on the PES. Inclusion of correlation at the secondorder Maller-Plesset perturbation theory (MP2) level had a significant effect on the relative stability of the two structures (it makes the stability difference small) but did not change either the nature of the stationary points or the details of the structures. Hence, the SCF optimized complexes are representative of the adsorption structures and can be used for calculations of the NMR chemical shifts.
Chemical Shifts The conversion from the calculated absolute shieldingconstants ~ was ~ done s according to the formula
u to the relative shifts 8
G,,,(complex)
= bTMs(int ref)
+ u(int ref) - u(comp1ex)
As internal reference of the calculations for the adsorption complexes of a particular molecule we used the molecule itself, i.e., methanol, water, and ammonia. The calculated absolute shielding constants of the internal reference molecules are listed in Table 1 together with the experimental ~ T M Svalue used for conversion. In addition, we give the absolute shielding constants obtained from SCF-GIAO and MP2-GIA02' calculations with larger basis sets22 to check the performance of the SCF-GIAO method for the moleculesunder study. Because the SCF-GIAO/ TZPvalues deviate by only 2% from the values obtained by MP2GIAO, it can be concluded that the method adopted performs well for the chosen references. To assess the reliability of our calculated IH shifts of molecules in H-bond situations, we performed calculations on methanol clusters. Their structures were available from SCF optimizat i o n which ~ ~ ~ employed the same tzp(O)/dzp(C,H) basis set as used to obtain the structures of the present study. For the methanol tetramer a structure was considered which consists of a three-memberedring and a fourth methanol molecule H-bonded to one oxygen atom of the ring, extending the coordination of that oxygen atom to four. It may be more representative of the situation in liquid methanol than the simple ring which is more stable in the gas phase. The calculated 'H shift of the proton on the 4-fold coordinated oxygen site is 4.3 ppm, close to the hydroxyl resonance of 4.7 ppm observed in NMR experiments on liquid methanol.2
Letters
TABLE 2 lH NMR Chemical Shifts 6 ( m) of the Neutral c ~ m g k x e s(NC)and ~on-pllirComplexesyi) of w3rrith Clusters 1-3 60
cluster model 1 (NC) 1 (IP)
0.8 9.3 0.9 10.3 1.3 9.9
1.9 9.3 0.9 10.3 1.3 9.9
G(av)b
3.1 2.4 1.6 3.0 1.0 3.0
11.6 8.2 10.3 4.2 11.1 4.4.
4.4 7.3 3.4 7.0 3.7 6.8
2b (NC) (IP) 3 (NC) 3 (IP) In the case of NC the first three signals are caused by the three ammonia protons pointing away from the cluster, and the fourth signal originatesfrom the proton of the zeolitecluster; in the IP complexca there are two equivalent signals due to the H-bond protons and two signals from the noninteracting proton. Arithmetic mean value.
TABLE 3: 'H NMR Cbmierl Shifts 6 (ppm) of the NC and IP Complexes of CH3OH 011 Clusters 1-3 NC
model 1 2a
3
SiO(H)AI 5.3 2.8 2.6
HOCH3 9.1 7.9 8.8
av 7.2 5.4 5.7
IP 17.5 15.9 15.3
TABLE 4 'H NMR Chemical Shifts 6 (ppm) of the NC and IP Complexes of HzO with Clusters 1-3 60
cluster model 1") 1 (IP) 2. (NC) 2a ( W 3 "1 3 UP)
1.2 19.2 1.6 18.1 1.7 17.4
6.0 19.2 3.9 18.1 3.1 17.4
6(av)b 8.7 2.6 7.4 3.0 8.0 3.3
5.3 13.7 4.3 13.1 4.5 12.7
a The NC complex gives rise to three signals: the first two are due to the two water protons (one involved in the H bond and the other pointing away from the cluster) and one from the zeolitic cluster proton; in the IP complexes there are two equivalent signals due to the H bond protons and one signal of the noninteracting proton. Arithmetic mean value.
Tables 2,3, and 4 summarize the 'H NMR chemical shifts of the adsorption complexes of ammonia, methanol, and water, respectively. While the adsorption of methanol and water on molecular sieves still is a subject of discussion and will remain, there is no doubt that ammonium ions are formed on adsorption of ammonia on bridging hydroxyl sites. Proton resonances between 6.5 and 7 ppm were assigned to NH4+ ions in the supercage and sodalith cage of NH4-Y for loadings of one molecule or less.6.7 Other authors report resonances at 7.0 ppm (NH4-Y)24 and 6.0 ppm (NH4-rh~).25The lH chemical shifts of 7.3-6.8 ppm calculated for the NH4+ ion-pair complexes (Table 2) correspond very well with the experimental findings. Their dependenceon the cluster size is small, which reflects that already small clusters describe the adsorption structure well and that electrostatic effects from the larger clusters have minor influence on the chemical shift of the ammonium protons. For methanol on H-rho a shift of 10-12 ppm has been reported.26 To our knowledge,this is the only experimental result available for the loading of one methanol molecule per Bransted site as assumed in the calculations. The value obtained for the ion-pair complex of the largest model of 15.3 ppm (Table 3) still exceeds the experimental shift by more than 3-5 ppm. For higher methanol loadings the experimental 1H shifts can be divided into two groups: 4.8-5.9 ppm for zeolites HY and HL and different SAPO$ and 9.4 and 10.5 ppm for loadings of six and two methanol molecules per acid site in H-ZSM-5, respectively.2.4 Because these values cannot be compared with the calculatedvalues of Table 3, wecarried out additional calculations on a complex consisting of cluster 3 and two adsorbed methanol molecules. The optimized structures of the neutral and ion-pair
The Journal of Physical Chemistry, Vo1. 98, No. 12, 1994 3085
Letters
data are due to a single adsorbed neutral molecule or a single protonated species, It should be noted that for both methanol and water the observed shifts are in between the calculated shifts of the neutral and ion-pair structures. Therefore, one could conclude that an equilibrium between these two complexes exist. However, we have no indication for this because only one minimum, the neutral H-bonded complex, was located on the PES. The ion-pair structures proved to be saddle points which are very likely candidates for the transition structures of the observed deuterium exchangereaction between water or methanol on deuterated acid sites. Hence, the observed shifts will be weighted averages of the initial and final states of this reaction. These, however, are symmetry equivalent and both correspond to the neutral complex. The experimental data for higher loadings of methanol and water can be explained by formation of an oxonium ion assisted by an additional neutral molecule (in the case of water maybe three or four). This is in accord with the fact that the proton affinity of clusters of water and methanol molecules is larger than that of a single water or methanol molecule.
Figure 2. SCF optimized neutral (top) and ion-pair structures (bottom) of complexes of cluster 3 with methanol and an additional methanol molecule. n
A
h
Acknowledgment. We are grateful to Prof. R.Ahlrichs and his group at the University of Karlsruhe for providing the most recent versions of the TURBOMOLE code, particularly the SHEILA program for chemical shifts. F.H. wants to thank U. Schneider for helpful discussions. Financial support from the “Fonds der Chemischen Industrie” is gratefully acknowledged. References and Notes (1) Marchese, L.; Wright, P. A.; Thomas, J. M. J. Phys. Chem. 1993, 97, 8109. (2) Anderson, M. W.; Barrie, P. J.; Klinowski, J. J . Phys. Chem. 1991, 95, 235. (3) Hunger, M.; Freude, D.; Pfeifer, H. J. Chem. Soc., Faraday Trans. 1991, 87, 657. (4) Mirth, G.; Lercher, J.; Anderson, M. W.; Klinowski, J. J . Chem. SOC.,Faraday Trans. 1990,86, 3039.
(5) Jentys, A.; Warecka, G.;Derewinski, M.; Lercher, J. J. Phys. Chem.
--.
W
Figure 3. SCF equilibrium geometry of the ion-pair complex of cluster 1 with water and an additional water molecule.31
complex are shown in Figure 2. Averaging the three hydroxyl 1H shifts yields values of 11.0 and 4.3 ppm for the ion-pair and neutral complex, respectively. While the former value agrees well with the observationof two methanol molecules per Br~rnsted site on H-ZSM-5,2 the latter is close to the lower values of the first group of the less acidic zeolites.2 The NMR data for the adsorption of water are more consistent than those for the adsorption of methanol. For loadings of one water molecule per bridging hydroxyl the following values are reported: 6.2 ppm (H-Y),3 a broad signal of 4.3-5.8 ppm (HY),12 7.1 ppm (H-ZSM-5),13 and a broad peak at 7 ppm (Hrho).26 WhilethecalculatedNMRshifts for theion-pair structure (Table 4) are approximately twice as large as the experimental values, the calculated shifts of the physisorption complexes are smaller. Toestimate the effect of higher loadingson the calculated shiftsof the ion-pair structure, we performed additional *HNMR shift calculations on a complex of cluster 1 with one hydronium ion and one water molecule (Figure 3).31 The average of the calculated hydroxyl shifts is 9.5 ppm. Clearly, this value will decreasefurther when additionalwater molecules will be adsorbed. In summary, the calculated lH chemical shifts for methanol and water indicate that it is hardly conceivable that the observed
1989. -- - - ,93. 4837. -(6) Jacobs, W.P. J.H.;deHaan, J. W.;vandeVen,L. J.M.;vanSanten, R. A. J. Phys. Chem. 1993,97, 10394. (7) Jacobs, W. P. J. H. Ph.D. Thesis, Eindhoven, 1993. (8) Pelmenschikov, A. G.; van Santen, R. A. J. Phys. Chem. 1993,97, 10678. (9) Pelmenschikov, A. G.; van Santen, R. A.; JBnchen, J.; Meijer, E. J . Phys. Chem. 1993, 97, 11071. (10) Pelmenschikov, A. G.;van Wolput, J. H. M. C.; Jiinchen, J.; van Santen, R. A. J. Phys. Chem., in press. (11) Parker, L. M.; Bibby, D. M.; Burns, G.R. Zeolites 1993, 13, 107. (12) Batamack, P.; Doremieux-Morin. C.; Fraissard, J. J . Chim. Phvs. 1992, 89, 423. (13) Batamack, P.; Doremieux-Morin, C.; Fraissard, J.; Frcude, D. J. Phys. Chem. 1991, 95, 3790. (14) Batamack,P.; Doremieux-Morin,C.; Vincent, R.; Fraissard, J. Chem. Phys. Left. 1991, 180, 545. (15) Kubelkova, L.; Novakova, J.; Nedomova, K. J. Carol. 1990, 124, 441. (16) Haase, F.; Sauer, J. Manuscript in preparation. (17) TURBOMOLE: Ahlrichs,R.;Bir,M.;Hher, M.; Horn,H.;Kblmel, C. Chem. Phys. Lett. 1989,154. 165. TURBOMOLE and TurboNMR are
commercially available from BIOSYM Technologies Inc., San Diego, CA. (18) HBser, M.; Ahlrichs, R.; Baron, H. P.; Weis, P.; Horn, H. Theor. Chim. Acta 1992, 83, 455. (19) Ditchfield, R. Mol. Phys. 1974, 27, 789. (20) Schifer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992,97,2571. (21) Gauss, J. Chem. Phys. Left. 1992, 191, 614. Gauss, J. J. Chem. Phys. 1993, 99, 3629. (22) Gauss, J. Personal communication. (23) Bleiber, A.; Sauer, J. Manuscript in preparation. (24) Pfeifer, H.; Freude, D.; Hunger, M. Zeolifes 1985, 5, 274. (25) Vega, A. J.; Luz, Z . J . Phys. Chem. 1987, 91, 365. (26) Luz, 2.;Vega, A. J. J . Phys. Chem. 1987, 91, 374. (27) Vega, A. J. J. Am. Chem. SOC.1988, 110, 1049. (28) Chauvel, J. P., Jr.; True, N. S.Chem. Phys. 1985. 95,435. (29) Schneider, W.G.; Bernstein, H. J.; Pople, J. A. J. Phys. Chem. 1958, 28, 60 1. (30) Harris, R. K.; Mann, B. E. NMR and the Periodic Table; Academic Press: London, 1978; p 89. (31) Krossner, M.; Haase, F.;Sauer, J. Manuscript in preparation.
