794
J . Phys. Chem. 1988, 92, 794-796
Electrostatic Potential in Dehydrated Sodtum Zeolite A from Low-Resolution X-ray Diffraction Data Mark A. Spackmant and Hans-Peter Weber*$ Department of Crystallography, University of Pittsburgh, Pittsburgh, Pennsylvania 15260 (Received: April 24, 1987; In Final Form: September 8, 1987)
The difference electron density and the electrostatic potential in dehydrated sodium zeolite A have been mapped by using Fourier summation methods applied to X-ray data of limited resolution published previously. Buildup of electron density is observed in both Si-0 and A1-0 bonds. A shallow trough of low potential value lines the surface of both cavities. The potential difference between this minimum and the center of the 4-ring does not amount to more than 0.3 e A-' (-400 kJ mol-').
Introduction Zeolites are framework aluminosilicates which are distinguished by the variety of their technical applications. As sorbents, they selectively absorb polar and polarizable molecules (fluids like water, olefins, paraffins, and industrial gases). As catalysts, they assist in the cracking of these same hydrocarbons to products containing mostly branched chain paraffins, cycloparaffins, and aromatics. While other compounds (like clays and amorphous silica-alumina) possess some, if not all, of the above properties, zeolites are more attractive to the experimentalist because the structure of the active surface is well-defined and has been described in the past decade in great detail from results of diffraction experiments.' These studies have greatly advanced our understanding of the sorption behavior of zeolites, as this property largely depends on the geometry of the cavities and their connecting channels. Catalytic processes, on the other hand, are more dependent on properties at the microscopic level, the electronic properties of the active surface lining the cavities. In their search for increased resolution, workers in the field have resorted to other techniques (like NMR), which probe these properties at selected sites only. We would like to demonstrate in this work that the same X-ray diffraction data which in the past has been used to reveal only the atomic arrangement within the zeolite cavities can also be used to describe the electrostatic potential within these regions. In principle, X-ray diffraction has always yielded the crystal electron density, but it is only in the past decade that formalisms needed to analyze and parametrize this fundamental electrostatic quantity have been proposed and tested on minerals and organic compounds.2 As a result, it is now feasible to obtain from X-ray data electrostatic properties like the electrostatic potential and its derivatives and to examine the change in the electron density brought about by chemical bonding. For simplicity, we choose a zeolite with an ordered Al/Si distribution, in its dehydrated form. We use data already published for a structural analysis to prove that data sets currently being collected in a more or less routine fashion already contain much of the information of interest. Even though zeolite A is not one of the most technically useful systems, we expect that the conclusions we draw can readily be applied to zeolites with greater industrial application. Method Almost all attempts to obtain reliable electrostatic properties from diffraction data have been compromised to varying degrees by the high degree of crystal perfection common in simple minerals, manifesting itself in large extinction effects. We note here the recent detailed studies on corundum3 (A1203),coesite4 and stishoviteS (SiO,), and rutile6 (TiO,), all of which amply demonstrate the problems. 'Present address: Department of Chemistry, The University of New En land, Armidale, New South Wales, 235 1, Australia. ?Present address: Institut de Cristallographie, UniversitC de Lausanne, B.S.P., CH-1015 Lausanne-Dorigny, Switzerland.
0022-3654/88/2092-0794$01.50/0
On the other hand, zeolite minerals, particularly synthetic ones, occur only as tiny ( 2 4 0 ; space group Fm3c), we performed a spherical atom refinement based on P , with anomalous dispersion corrections from Cromer and Liberman,13 anisotropic thermal parameters, and including an isotropic extinction parameter, g (type 1, Lorentzian), in the Becker-Coppens f0rma1ism.l~ A total of 52 variables were refined; the aluminosilicate framework was constrained to be stoichiometric and ordered as in ref 12. The site populations of the three sodium atoms in the crystal were included as variables, along with all positional and thermal parameters. Final R factors [R(F) = 3.4%, R,(F) = 2.8%, gof = 2.561 are similar to those reported by Pluth and Smith.12 Comparison of our results with those obtained by Pluth and Smith'from a refinement with spherical ions (Si2+,Al+, 0-, and Na+) shows excellent agreement. All position parameters agree within 3a, and most within 1a. Similarly, the maximum difference between Na site populations for the two refinements is 2a. The total Na content of 90.0 (1.4) atoms per unit cell agrees with their value of 9 1.7 (1 .O). Anisotropic thermal parameters are in good agreement. l 5 (1 1) Harris, F. E. In Theoretical Chemistry, Advances and Perspectives; Eyring, H., Henderson, D., Eds.; Academic: New York, 1975;Vol. 1, p 147. (12) Pluth, J. J.; Smith, J. V. J. Am. Chem. SOC.1980, 102,4704-4708. (13) Cromer, D.T.;Liberman, D. J. Chem. Phys. 1970,53, 1891-1898. (14)Becker, P.J.; Coppens, P. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1974, A30, 129-147, 148-153.
Figure 2. Total electrostatic potential of aluminosilicate framework (Le., minus N a atoms/ions). Sections shown are the same as in Figure 1. Contours are drawn at intervals of 0.1 e A-' ( 135 kJ mol-').
-
Although not important for the results discussed by Pluth and Smith, the amount of extinction in the data is critical for our analysis, especially in the mapping of $ ( r ) . The greatest effect is observed for 2,0,0, where the correction (y, defined by ylFo(kinematic)I = IFo(observed)l) is 0.86. Only two other reflections are affected by more than 5% of F (y(2,2,0) = 0.94, y(2,2,2) = 0.95). We conclude that, at least for this data set, extinction is not a serious problem and our results will not be limited by the sort of problems encountered in other mineral studies. Deformation Electron Density. In Figure 1 we map the residual electron density for this refinement, the equivalent of the deformation electron density (Ap), in four sections of interest in the crystal. Figure l a shows essentially no residual density near Na(3) but some large peaks along the left and lower edges of the map. We can correlate these peaks with those in Figure Id, since Figure l a is a section perpendicular to Figure Id, both passing through O( 1). The peak to the left of O( 1) in Figure l a is the bonding electron density, seen more clearly in Figure Id. The remaining peaks nearer (0,'/4,'/4) are associated with the partial occupancy of Na(2) in the four sites inside the 8-ring (see Figure Id) and may be attributed to imperfect modeling of this complex distribution. Figure 1b,c,d displays very nicely the Ap features expected for 0 atoms bonded to A13 or Si.4*5There is a localized buildup of electron density around the oxygen, generally with a larger peak in S i 0 bonds than in A10 bonds. We also note the apparent polarization of these features toward the adjacent Na atoms. However, with data of such limited resolution, we hesitate to place too much emphasis on these features. Electrostatic Potential. In Figure 2 we map $(r) in the same planes as Figure 1, with contours $(r) in intervals of 0.10 e A-1. Typical esd's in the extranuclear regions are between 0.06 and 0.10 e A-' (1 e A-1 = 1389.4 kJ mol-'). We have omitted the partly because we are inNa atoms in our calculation of (15) We note that in Table I1 of ref 12 for Na(3) BIZ= DZ3= -2 (4)X and 813 = 4 (14)X lo4. For this site, the equivalences should be BIZ = -813, and we obtain p12= -2 (8) X lo4 and 823 = -3 (5) X lo4. (16)The residual electron density is not always equivalent to the deformation electron density. For example, in the refinement reported by Pluth and Smith, spherical ions were used, and hence the residual density in that case is the differencebetween the observed density and a model consisting of spherical ions. Ap(r) is strictly the difference between the observed density and a model of spherical atoms (IAM).
796
J . Phys. Chem. 1988, 92, 796-803
terested in the electrostatic potential generated by the aluminosilicate framework and partly because not all Na sites are fully occupied in the "average" structure, but full occupancy must be the case for any specific N a interaction with the framework. The most important result to be learned from these maps is the shape of the aluminosilicate framework. It is characterized by a large positive potential around each of the nuclei and looks very much like a model of the structure constructed with overlapping spheres. It is nothing like the pictures of $(r) constructed by assuming point charges at the atomic sites (e.g., see calculations by Preuss, Linden, and Peuckert"). We believe our mapping of $(r) in Figure 2 is more realistic, except for reservations we have concerning neglect of the N a atoms. For a more detailed study it would be necessary to include the contribution of the Na+ ions. Regions of chemical interest in 4 ( r ) are the areas of minimum potential and high electric field (i.e., steep slope of +(r)). In Figure 2c there are shallow minima inside the 6-ring associated with the O(3) atoms. In the 8-ring (Figure 3d) there are minima associated with all 0 atoms, but it is easy to see that a Na ion (or atom) would more readily occupy the site (2) (close to one O(2) and two 0 ( 1 ) atoms) rather than close to one O(1) and two O(2) atoms. The Na(2) fractional site population of 0.23 (0.01) suggests that almost every 8-ring has a Na atom bound inside it. Perhaps the most interesting section for 4 ( r ) mapping is that through the large cavity (Figure 2a). The Na(3) atom lies in the center of a broad, flat area of low potential in the mapped plane. The minima in this plane are at top left (-0.48 e A-I) and right of center on the top of the map (-0.64 e A-'). The right-hand corner of the map (1/4,1/4,'/4) is a local maximum (-0.19 e A-'). Thus, a point charge would experience a barrier of 0.45 e A?], or 625 kJ mol-', to movement across the cavity. This is not expected to be a realistic estimate of diffusion energies through the structure. Even if $ ( r ) were being mapped reliably, the diffusing particles are not point charges, and their interaction energy with the aluminosilicate framework would be an integral over a substantial region of space. Moreover, a diffusing particle
need not pass directly through the center of the large cavity. Instead, it could move closer to the walls (i.e., the oxygen atoms) where the electrostatic potential is virtually constant. Although we do not give maps of the electric field in the cavities, we can derive E(r) (a vector quantity) from 4 ( r ) readily, and it can be expressed in a similar way to 4 ( r ) (eq 1) as a combination of Fourier and direct space summations (e.g., see ref 11). As shown elsewhere for stishovite," direct mapping of E(r) via this strategy is not likely to be accurate. Rather, we would prefer to derive a pseudoatom model representation of p(r), and from this extract E(r), essentially with no thermal motion. However, as we discuss below, this approach requires a more extensive data set than that analyzed here.
Acknowledgment. We thank Prof. B. Craven for encouragement and criticisms, and we are grateful to Mrs. Joan Klinger for technical assistance. Dr. J. Pluth, University of Chicago, kindly supplied the zeolite A data. This work was supported in part by a grant (HL-20350) from the National Institutes of Health.
(17) Preuss, E.; Linden, G.; Peuckert, M. J . Phys. Chem. 1985, 89, 2955-2961.
41. 721.
Conclusions and Future Prospects This work is intended as a feasibility study, to show how much information on bonding electron densities and electrostatic properties in the zeolite structures can be extracted from lowresolution X-ray data. The results are promising. However, we defer a comparison with results on other mineral systems, or with theoretical calculations, until more extensive X-ray data are available. The data should extend to sin 8/X 3 1.0 A-l, requiring shorter wavelength (Mo or Ag), which in turn would further reduce effects of extinction on the low-angle data. Also, it would be desirable to collect the X-ray data at reduced temperatures. Since the zeolite crystals are generally soft and have large open structures, thermal motion is much higher than in simple minerals. For example, B , values for Si (1.85 (3) A2), AI (1.97 (3) A2), and 0 (3.0 A2) are far higher than observed in quartz [0.49 A* (Si) and 0.99 A* (O)'*]and corundum [0.23 A2 (AI) and 0.27 A2 (oy]at room temperature.
(18) LePage, Y.; Calvert, L. D.; Gabe, E. J. J . Phys. Chem. Solids 1980,
Adsorption and Decomposition of HCOOH on Potassium-Promoted Rh( 111) Surfaces Frigyes Solymosi,* Jiinos Kiss, and Imre Koviics Reaction Kinetics Research Group of the Hungarian Academy of Sciences and Institute of Solid State and Radiochemistry, University of Szeged, P.O. Box 105, H-6701 Szeged, Hungary (Received: May 20, 1987)
Preadsorbed potassium significantly altered the adsorption and the reactions of HCOOH on Rh(1 i 1) surface. A potassium-induced desorption peak of HCOOH was identified, with Tp = 254 K. Preadsorbed potassium enhanced the dissociation of HCOOH and stabilized a formate species characterized by the photoemission peaks at 5.2, 8.9, 10.3, and 12.2 eV in the He 11 spectrum. These peaks were eliminated at 267 K on clean Rh, at 330 K at 8, = 0.1, and above 422 K with a monolayer of potassium (8, = 0.36). Decomposition of the formate species led to the formation of H2,COz, HzO, and CO, which desorbed at significantly higher temperatures than from the K-free surface. In the interpretation of the effects of potassium, an extended charge transfer between HCOOH and the K/Rh( 111) surface (at 6, = 0.1) and a direct chemical interaction between potassium and HCOOH involving the formation of potassium formate like species (8, = 0.36) are assumed.
Introduction In a previous paper we investigated the interaction of HCOOH with clean and oxygen-dosed Rh( 111) surfaces.' As pointed out earlier,'s2 there is a strong evidence that the formate species is an important surface intermediate in the formation of oxygenated carbon compounds over Rh. In the present study we report on *Address correspondence to this author at Reaction Kinetics Research Group, The University of Szeged, H-6701 Szeged, P.O.Box 105, Hungary.
the influence of potassium on the adsorption and decomposition of H C W H on m(l11);this is strongly connected with a Program relating to evaluation of the effects of potassium additive in the hYdrogenations of co and co2 on Rh catalyst. As Part of this Program, the effects of potassium have been examined on the (1) Solymosi, F.; Kiss, J.; KovPcs, I. Surf. Sri. 1987, 192, 47. (2) Deluzarche, A.; Hindermann, J. P.; Kieffer, R.; Kiennemann, A. Reu. Chem. Zntermed. 1985, 6, 625.
0022-3654/88/2092-0796$01 .50/0 0 1988 American Chemical Society
