J. Phys. Chem. 1994,98, 13151-13156
13151
29Si MAS NMR Spectroscopy of Tectozincosilicates Miguel A. Camblort and Mark E. Davis* Chemical Engineering, Califomia Institute of Technology, Pasadena, Califomia 91125 Received: August 8, 1994; In Final Form: September 27, 1994@
29SiMAS NMR spectroscopic data from dense (NazZnSiO4, NazZnSi206, K2ZnSi4010) and microporous (VPI7, VPI-9) tectozincosilicates are presented. These results allow for the first time the assignment of the chemical shift ranges for silicon environments (Si:nSi,4-nZn; n = 0-4) in tectozincosilicates. A comparison between the chemical shift ranges of tectozincosilicates and tectoaluminosilicates (zeolites) is provided, and the differences are rationalized in terms of the Zn-0-Si bonding. Unlike aluminosilicates, a high degree of Zn, Si ordering is observed in all the materials investigated. Predictions of structural details for the tectozincosilicates VPI-7 and VPI-9 are made, and several general aspects of the crystal chemistry of tectozincosilicates are discussed in view of the NMR data.
Introduction Tectozincosilicates are crystalline materials possessing a three-dimensional framework of TO4 (T = Si or Zn) tetrahedra that comer share oxygens with four other tetrahedra through single TOT bridges. To preserve charge neutrality, the presence of extraframework cations is necessary in order to compensate for the charge created by inserting Zn042- tetrahedra in a silicate framework. Thus, the general formula for tectozincosilicates is MunZn,Si(l-,)02:wH20, where M denotes one or several cations of charge n and w is the number of occluded water molecules. These materials are frequently dense. However, very recently, microporous zincosilicates have been purposely The interest in tectozincosilicates is to prepare new materials for catalysis and for adsorption, separation, and cation-exchange processes. One of the goals in these syntheses is the preparation of new microporous materials with low framework densities (FD: number of T atoms per nm3), since it is speculated that there is a relationship between the minimum attainable FD and the MINR parameter (MINR: the size of the smallest ring of tetrahedra to which all the T atoms belong) in zeolite^.^ This relationship predicts that a lower FD requires a smaller MINR (below 4), so the preparation of materials with 3-membered rings (3MR: contain 3 T atoms and 3 oxygen atoms in the ring) (MINR = 3 or 3+: the plus denotes that not all the T atoms are in 3MR) is desirable. The 3MR is more frequently encountered in zinco- and beryllosilicates than in aluminosilicates, and the zincosilicate hydrothermal chemistry recently provided a new and interesting MINR = 3 f microporous material.' One of the problems in dealing with zincosilicate or aluminosilicate microporous materials is that they are synthesized by hydrothermal methods at low temperatures (typically below 473 K) that generally lead to microcrystalline powders unsuitable for single-crystal X-ray analysis. Thus, the solution of the framework structure is frequently accomplished by the timeconsuming model-building method, and the framework topology is determined by comparison of the experimental X-ray powder diffraction (XRD) pattern to the computer-simulated patterns of the models. Here, we show that for tectozincosilicates, a good deal of structural information can be obtained from 29Si MAS NMR, much more than normally achievable with zeolites
* To whom correspondence should be addressed. T @
On leave from Instituto de Tecnologia Quimica, Valencia, Spain. Abstract published in Advance ACS Abstracts, November 15, 1994.
0022-365419412098-13 151$04.50/0
since the zinc has a greater tendency to order than aluminum (vide infra). The NMR information can be used in combination with XRD data to limit the range of possible model structures that can be used during the structure solution process (as illustrated below for the microporous zincosilicate VPI-9). Moreover, 29SiMAS NMR can provide additional information that can assist in checking the validity of a proposed model or its proper symmetry (as shown below with the microporous zincosilicate VPI-7 (MINR= 3f)). From this work, it becomes apparent that the crystal chemistry of framework alumino- and zincosilicates (especially in terms of T ordering and TOT angles) is much more different and that 29SiMAS NMR spectroscopy may even be more useful for the study of microporous zincosilicates than it is for zeolites (a consequence of the greater tendency of zinc to order in zincosilicates). Experimental Section
Hydrothermal syntheses were carried out in Teflon-lined, stainless steel, 25-mL Parr bombs heated in a convection oven under static conditions and autogenous pressure. Chemical reagents included TEAOH (40 wt % aqueous solution, Johnson Matthey), TMAOH (25 wt % aqueous solution, Johnson Matthey), TMACl(97%, Aldrich), amorphous Si02 (Cab-0-si1 M-5), silica sol (HS-40, DuPont), zinc acetate dihydrate (Fisher), ZnO (zincite, Aldrich), NaOH (97%, EM), NaCl (EM), KOH (Aldrich), RbOH (50 wt % aqueous solution, Aldrich), and distilled water. NazZnSi04 was synthesized from a starting mixture of composition 4.86NaOH: 1.~~TEAOH:O.~~Z~(CH~C~O)Z:S~ 28.57H20 that was heated at 473 K for 4 days. A yield of over 86% (moles of Zn in the product over moles of Zn in the initial mixture x 100%) was achieved. NazZnSi206 (synthetic ckalovite) was synthesized from a mixture of composition ~.~~N~OH:O.~~TEAOH:O.~~Z~(CH~COO)Z:S~OZ: 14.29H20that was heated at 473 K for 4 days. A yield of 94% (defined as above) was achieved. VPI-7 (with an ideal composition in the full Na form of NQZn&Olg:wH20) was synthesized by methods I and I11 described by Annen? VPI-7(1) was synthesized from a reaction mixture of composition 0.88NaOH:0.30KOH:0.08TEAOH: 0.039ZnO:Si02:22HzO that was heated at 473 K for 3 days. VP1-7(111) can be crystallized from a reaction mixture with composition 0.46NaOH:O. 15NaC1:0.23ZnO:Si02:7lH2O prepared by evaporating to dryness at 373 K for 30 h a mixture 0 1994 American Chemical Society
13152 J. Phys. Chem., Vol. 98, No. 50, 1994
Camblor and Davis
TABLE 1: Crvstallographic Data and Chemical Composition of the Phases Investigated by 29SiMAS NMR name
formula
sodium zinc silicate synthetic ckalovite potassium zinc tetrasilicate VPI-7 VPI-9
NazZnSiO, NaZnSi206 KzZnSL010 Na&12Si70ls:wHz@ Rbd;n&Ozo:wHzO'
cryst syst and space group monoclinic, Pc orthorhombic, Fdd2 orthorhombic, P212121 tetragonal,I - 4m2b tetragonald
unit cell params
ref
a = 7.02, b = 5.48, c = 5.33, /3 = 90'20' a = 21.54, b = 7.139, c = 7.413
6 7 8 1,2 2,3
a = 10.068, b = 14.047, c = 7.067 a = 7.179, b = 40.62 a = 9.888, c = 36.88
Ideal composition. Maximum topological symmetry. Estimated from chemical analysis and NMR data. Unknown structure and space group.
I
with half the amount of water, grinding the solid into a fine I powder, and then adding the full required amount of water to the solid and heating the mixture for 55 days at 423 K. Rb-VPI-9(1) was synthesized from a reaction mixture of composition O.9RbOH:0.08TEAOH:0.039Zn(CH~C00)~:SiO~: 22H20 that was heated at 473 K for 4 days. A bulk molar composition of Rb1.gZnSi4.2 was obtained from chemical analysis. VPI-9(zI) was synthesized from a reaction mixture of composition O.60RbOH:0.08TEAOH:0.30Zn0:Si0~:22H~0 that was heated at 473 K for 1 month (this sample contains some amorphous material and a phase that shows a peak at -8.5" 20 that does not appear in VPI-9(1) or -(HI)). VPI-9(III) was synthesized from a reaction mixture of composition 1.18RbOH:0.08TEAOH:O.04Zn0:SiO~:22H~0 that was heated at 473 K for 3 days. A bulk molar composition of Rb1.7ZnSi3.6 was obtained from chemical analysis. The material K2ZnS4010was supplied by Prof. A. Kawahara at Okayama University, Okayama, Japan. -60 -80 -100 -120 The phase purity of the solids was determined by X-ray 6 (ppm from TMS) powder diffraction (XRD) using a Scintag XDS-2000 diffractometer (Cu K a radiation). 29Si MAS NMR spectra were Figure 1. 29SiMAS NMR spectra of the dense tectozincosilicates NazZnSiO4 (bottom), NazZnSiz06 (middle), and KZnSbOlo (top). The collected on a Bruker AM-300 spectrometer at a 4-kH2 spinning resonance marked with a star is due to a willemite impurity. rate, referenced to tetrakis(trimethylsily1)silane (downfield resonance at -10.05 ppm) and are reported referenced to TMS. Deconvolutions of the spectra were performed by using the Bruker LLNESIM Gaussian-fitting software. TOT angles were calculated from the reported structural data using the program KRIBER.5 Results A number of tectozincosilicates ranging from dense to microporous and from Si/Zn ratios of 1 to 6.5 were synthesized by low-temperature hydrothermal methods (except for K2ZnS4010). The XRD pattems reveal that the samples have high crystallinity (except for VPI-9(II) and possibly VPI-9(III)). For all the phases studied here that are of known structure, the framework consists of a three-dimensional framework of TO4 tetrahedra (T = Si or Zn) sharing all comers through twocoordinated 0 bridges with four other TO4 tetrahedra, and the Si-0 and Zn-0 distances agree with reported covalent bond distances for tetrahedral coordination (around 1.62 and 1.95 A, respectively). Other zincosilicate phases where tricoordinated 0 (willemite) or octahedral Zn (clinohedrite) exist are not considered here. This is not a severe limitation since zinc has a high propensity to be tetrahedrally coordinated in zincosilicates. For the materials studied here, the crystallographic data and composition of the phases are listed in Table 1. The 29SiMAS NMR spectra of the dense zincosilicates and VPI-7 are shown in Figures 1 and 2, respectively, and the results from spectral deconvolution are presented in Table 2. By analogy to the Loewenstein rule (avoidance of A1-0-A1 pairings in aluminosilicates) in zeolites, we will not consider that Zn-0-Zn pairing can occur in framework in zincosilicates, as both the charge repulsion and bond order of the 0 will make this environment even more unfavorable than the A1-0-A1 pairing in aluminosilicates. In this paper, we will call this
-40
-80
- 1 20
6 (ppm from TMS) Figure 2. 29SiMAS NMR spectra of the microporous tectozincosilicate VPI-7 synthesized by methods I (bottom) and III (top) of ref 2.
exclusion principle the extended Loewenstein rule. This rule allows us to calculate the Si/Zn ratios from the intensities, I , of every Si(nZn) resonance (where n refers to the number of Zn atoms 0-bridged to a given Si atom; other 4 - n 0 bridges are to Si atoms) in much the same way as is done when obtaining SUA1 ratios for zeolites and other framework alumino~ilicates:~
J. Phys. Chem., Vol. 98, No. 50, 1994 13153
Spectroscopy of Tectozincosilicates TABLE 2: Chemical Shifts,Intensities, and Assignments of the 29SiMAS NMR Resonances in Tectozincosilicates with Known Structure 29Sichem re1 phase shiff intensityb assignment ( S j n n ) m c Si(4Zn) 1 NazZnSi04 -66.5 100 Si(2Si,2Zn) 2 NazZnSiz06 -83.8 100 (ckalovite) 50.4 Si(3SiJZn) KzZnSLO1Od -88.8 Si(3Si,lZn) 4 -92.8 25.5 24.2 (Si(3Si,lZn) -94.2 -80.2 VPI-7 14.8 Si(2Si,2Zn) -88.5 5.7 Si(3Si,lZn) -91.8 36.1 Si(3SiJZn) 3.5 -93.3 17.0 Si(3Si,lZn) -95.3 26.4 Si(3Si,lZn) VP1-7(111) -77.9 7.5 Si(2Si,2Zn) -81.0 6.3 Si(2Si,2Zn) Si(3Si,lZn) 3.5 -91.8 35.5 43.0 -93.8 Si(3Si.lZn) 7.8 -95.6 Si(3Si,lZn) In ppm from TMS. Ratio of the intensity of a given resonance to the sum of intensities of all the lines for a given phase. Calculated from the intensities and assignments of the 29SiNMR resonances by using eq I (extended Loewenstein rule). A peak at -71.3 ppm, due to a willemite impurity, is not included. (2
This calculation will be used to check the validity of the 29Si NMR assignments of the resonances by comparison with the known chemical composition. Additionally, it is possible to calculate the population of every Si(nZn) species assuming a random distribution in a “Loewensteinian” tectosilicate by applying the equation
P, = 24p”(l - ~)~-“/(n!(4 - n ) ! ) , p-’ = Si/Zn
(11)
When populations calculated by using eq I1 are markedly different from the observed intensities, it can be concluded that a nonrandom distribution exists (the opposite is not necessarily true).g NazZnSiO4 shows a single resonance at -66.5 ppm, in agreement with the fact that the Si/Zn ratio is 1 and in following the extended Loewenstein rule. Obviously, this resonance can only be assigned to Si(4Zn), and there is a strict alternation of Zn and Si throughout the framework. This result completely agrees with the structure reported for this material.6 The structure is such that the two crystallographic positions are occupied by Si and Zn, respectively, with every Zn surrounded by 4Si and every Si surrounded by 4Zn. NazZnSi206 (synthetic ckalovite) reveals also a single, wellresolved resonance at -83.8 ppm, which in view of the SiEn ratio of 2 can only be assigned to Si(2Si,2Zn) according to eq I. This result implies that there is complete Zn, Si ordering throughout the framework; a random distribution would give resonances for n = 1,2, and 3 with intensities 0.25,0.375, and 0.25, respectively ( n = 0 and 4 would have intensities below 0.07). This result agrees well with the reported structure for synthetic ckalovite’ which can be described as a P-cristobalitelike framework of Si04 and ZnO4 tetrahedra with 5-coordinated Na+ occluded in framework voids (a “stuffed cristobalite”). There are two crystallographic sites occupied by Si and Zn, respectively. Si is surrounded by 2Si and 2Zn, and Zn is surrounded by 4Si (again obeying the extended Loewenstein rule). Amazingly, this material is a highly ordered framework that is isostructural with the high-temperature polymorph of cristobalite even though it is synthesized at low temperature (473 K) under hydrothermal conditions. The spectrum of K2ZnSi4010 (Figure 1, the downfield resonance is due to a willemite impurity) clearly shows three
resonances at around -88.5, -92.5, and -94.0 ppm with relative intensities 2.1: 1.1:1. One possible assignment of these resonances that is consistent with the Si/Zn ratio and the extended Loewenstein rule is to assign the two lines at upper field to Si(4Si) and the one at lower field to Si(2Si,2Zn). However, these upper field lines are far from the chemical shift region expected for Si(4Si) species (Si(4Si) in aluminosilicates -100 to - 118 ppm). Even though zincosilicates are expected to have small TOT angles and it is well-known that the 29Si NMR chemical shift is strongly dependent on the TOT angle, a value of -92.5 ppm seems to be too far downfield for a Si(4Si) environment. Therefore, a more likely assignment of the resonances in the K2ZnS4010 phase is that all peaks are from Si(3Si,lZn) environments. Otherwise, a Si/Zn ratio of 4 could not be achieved. If this is true, then Zn, Si ordering exists also in this plane, as a random distribution would give resonances for n = 0, 1, and 2 with relative intensities of 0.32, 0.42, and 0.21, respectively. The presence of three peaks for Si(1Zn) must be due to the resolution of crystallographic sites, and the relative intensities provide information about the multiplicities of these sites (could be three sites in a ratio 2: 1: 1 or four sites in a ratio 1:1:1:l with two overlapping resonances). Inspection of the structure of K2ZnSi4010 reported by Kohara and Kawahara8 confirms that the latter assignment is correct: four unique crystallographic sites (in a 1:1:1:1 ratio) are found to be occupied by Si (each one surrounded by 3Si and 1Zn) and a fifth site is occupied by Zn that is surrounded by 4Si (again in compliance with the extended Loewenstein rule). The good agreement between the crystallographic data and the 29Si MAS NMR spectrum supports both the reported structure and the analysis presented here. The 29Si MAS NMR spectrum of VPI-7(III) (Figure 2 and Table 2) shows four rather well-resolved lines at -77.9, -81.0, -91.8, and -93.8 ppm and a shoulder at -9.56 ppm with relative intensities in the ratios of 1.2:1:5.6:6.8:1.2. The chemical shift data from the dense phases described above can be used to assist in making the assignments of the resonances in WI-7. The lines at low field (-77.9 and -81.0 ppm) should be assigned to Si(2Si,2Zn), while all other resonances are from Si(3Si,lZn) environments. These assignments give a ( S ~ / Z ~ ) N M R ratio of 3.5, which is the value consistently found in the chemical analyses of VPI-7 synthesized by different methods. For VPI7(I), the spectrum shows a poorer resolution; i.e., only one broad band, rather than two well-resolved lines, appears at low field. However, the assignments should be similar (Table 2) for both samples. These assignments imply that there is a Zn, Si ordering in both materials, as a random distribution would require the presence of intense Si(4Si) resonances: for a random distribution in a Loewensteinian tectozincosilicate with Si/Zn = 3.5, four lines with n = 0, 1, 2, and 3 would appear with relative intensities of 0.26, 0.42, 0.25, and 0.06. Also, the splitting of lines assigned to the same environment for Si is due to resolution of crystallographically unique sites (this effect seems to depend on the specific synthesis procedure). Annen and Davis already proposed that the downfield resonances in the 29SiMAS NMR spectra of lovdarite and VPI-7 must be ascribed to the central Si atom in the spiro-5 unit (see below) present in both struct~res;~ the complete assignment of the spectra as to Si(nZn) or Si(nBe) species was not presented. The assignment of the downfield lines to the central Si atom in the spiro-5 unit was based on a comparison between these two materials and other zinco- and beryllosilicates (non-tectosilicates), and the downfeld shift of this line was rationalized in terms of the small TOT angles in all four bonds of that Si. Although we now assign the downfield lines to Si(2Si,2Zn), we believe that both
-
Camblor and Davis
13154 J. Phys. Chem., Vol. 98, No. 50, 1994
.. ....
........... ........... ........... l y ............ 11 ............ TI----.------. m.. .......... m............ m.. .......... n............ m. m. T(.
.
..
.I...
+
............ .......... T4. -. - - -.- - -..
Spiro-5 Unit
T.2
m..
60
70
.80
80
100
110
.120
Chemlcal Shift, ppm
.
(
.......... ............ TI ............ B.. ..........
E
Si or Zn
)
m.. m
Figure 3. Ranges of chemical shifts for Si(nZn) species in tectozincosilicates.
assignments are correct, thus giving Si(2Si,2Zn) in the center of the spiro-5 units. The ranges of chemical shifts for Si(nZn) species in the tectozincosilicates are shown in Figure 3. The results for two zincosilicates (Si/Zn = 6.5 and 12.8) with the sodalite structure whose synthesis and characterization we describe elsewherelo and VPI-8 (high-silica tectozinc~silicate)~have also been included in this figure. Unfortunately, we have not been able to make a phase showing a Si(3Zn) resonance. However, this resonance appears at around -77.5 to -78.2 ppm in K2-bZnl-,Sil+,04 materials with the “stuffed” cristobalite structure (solid solutions ranging from x = 0 to about 0.25), and these chemical shifts are included in Figure 3.” The data shown in Figure 3 reveal that Si(nZn) resonances appear at lower field than Si(nA1) resonances (for example, -75 Si(2Si,2Zn) I -85, -90 < Si(2Si,2A1) -100; see ref 12 for complete 29Si shift ranges for Si(nAl), although this effect is only apparent for n 0 and becomes greater as n increases). This behavior is at least in part due to the smaller TOT angles for Zn-0-Si vs AI-0-Si, although the charge differences between Zn2+ and A13+ will also have an effect on the 29Si NMR chemical shift. Despite some degree of overlap in the various environments shown in Figure 3 (especially for Si(2Zn) and Si(3Zn) due to the downfield shift of Si(2Zn) in spiro-5 units), fairly direct assignments of Si(nZn) resonances in tectozincosilicates will be possible in general by using the information provided in this figure. The NMR data when used in combination with the extended Loewenstein rule and the chemical composition will allow one to extract structural information from tectozincosilicates with unknown structures and/or to assist in the structure refinement of known phases. Data on the presence of Zn, Si ordering, specific structural features (spiro-5 units), and even the number of crystallographic sites appear possible from the analysis of the 29SiNMR spectra. To illustrate how 29SiNMR data can be used to check the validity of a structural model and its symmetry constraints, we investigate the tectozincosilicate VPI-7. The most remarkable topological feature of VPI-7 is the presence of the so-called spiro-5 unit, where two 3MR share a central Si atom. This unit is found also in lovdarite and in the natural beryllophosphate weinebeneite (the only known microporous material having a MINR = 3).13 An inspection of the proposed topology for VPI-7 with ideal maximum symmetry (Table 3 and Figure 4) shows that there are two symmetrically nonequivalent spiro-5 units: one formed by sites T4 (central atom) and T3 and another one formed by sites T6 (central atom) and T5.l The remaining two sites (T1 and T2) are not located in the spiro-5 units and are in 5MR and 4MR. Actually, it can be shown that every T atom that lies at the same z atomic coordinate belongs to the same T site (Figure 4) and that the structure can be described
*
Figure 4. Schematic drawing of the VPI-7 ideal topology along the (010) direction.’ Only tetrahedral atoms are shown. T sites are indicated at the left (all the T atoms with the same z coordinate are symmetrically equivalent in the topology with maximum symmetry). Projection down the (100) direction gives the identical scheme.
TABLE 3: Crystallographic Sites and Relative Ratios of Si Sites in the Topology of VPI-7 with Ideal Maximum Symmetry’ multisite
descripn of the site
plicity
T1 T2 T3
5MR+4MR 5MR+4MR 3MR not in the center of spiro-5
8 8
T4
center of spuo-5
2
8
T5 3MR not in the center of spiro-5
8
T6
2
center of spiro-5
Si, Zn re1 occupancyo ratio“ 8Si 8Si 4Si 4Zn 2Si 4Si 4Zn 2Si
4
4 2 2
1 2 2 1
Derived from the 29SiNMR analysis (see text).
as being formed by two types of “layers” of spiro-5 units with alternating layers comprised of 5MR or 8MR. Since the downfield lines must be assigned to Si(2Si,2Zn) in the central position of the spiro-5 units, we must conclude that every 3MR in VPI-7 has one and only one Zn atom (assuming that the extended Loewenstein rule holds). Because T3 and T5 are in different spiro-5 units, each one of these sites must be occupied by alternating Zn and Si. This statement implies that the relative ratios of Si in the different T positions are 44:2:2:1:1 (Tl, T2, T3, T5, T4, T6) rather than 4:4:4:4:1:1 (Table 3) and that the actual symmetry of VPI-7 is lower than that reported previously ( h e n et al. reported only the maximal topological symmetry1). The 29Si MAS NMR results indicate that there is a complete Zn, Si ordering within any “layer” of 3MR. Once a Zn atom is assigned to a given T3 (or T5) site, the NMR spectra indicate that the distribution of Zn in the remainder of the T3 (or T5) sites within the same layer of 3MR is always the same: the T2 (or T1) site next to the Zn is a Si(SSi,lZn) species, so it is linked to the next 3MR by a Si atom, and this determines the Zn occupancy in that 3MR (see Figure 5 ) . However, this distribution could be different in a different spiro-5 layer. Thus, the interesting situation of full order in layers and disorder among layers is possible and is compatible with the NMR results presented here. Very recently, Rohrig et al.14 completed a Rietveld refinement for VPI-7 (R, = 16.2%). These authors report a lower symmetry (orthorhombic, space group Fdd2) for VPI-7 than was suggested by Amen et al. which contains five crystallographic unique positions: one occupied by Zn (in 3MR and out of the center of the spiro-5 unit) and the other four by Si (in the ratio 1:2:2:2). Additionally, there is Zn in every 3MR, in agreement with our conclusions from the NMR data. This
Spectroscopy of Tectozincosilicates
J. Phys. Chem., Vol. 98, No. 50, 1994 13155
-80
-40
-120
S (ppm from TMS)
Figure 5. Portion of the VPI-7 topology showing the Zn, Si ordering scheme according to the 29Si MAS NMR spectra. Open circles represent Zn atoms, while the filled circles are Si atoms. The center of the spiro-5 unit is marked with a * and is a Si(ZSi,ZZn). The other Si atoms are Si(3Si,lZn). When a position is chosen to be occupied by Zn (for example, the one with the arrow), this determines which side of every 3MR in the whole 3MR layer is occupied by Zn. However, the ordering scheme does not necessarily propagate to the other 3MR in the same spko-5 unit or to the next spiro-5 unit. In each 3MR layer, Zn always occupies the same side of the 3MR, but in the next 3MR layer, either the same or the other side of the 3MR's is occupied by Zn. atomic arrangement suggests that there is complete Zn ordering in VPI-7. The NMR data shown in Figure 2 are not consistent with the structure proposed by Rohrig et al. The reason for the discrepancy could be due to the disordering in layers as described above or to other factors unknown to us at this time. However, clearly the 29SiNMR data are providing further clues to the precise structure of VPI-7. Ercit and van Velthuizen recently reported the crystal structure (determined by single-crystal Patterson methods) of the natural zincosilicate gaultite, and it has the topology of VPI-7.I5 (The results from Ercit and van Velthuizen are in good agreement with those of Rohrig et al.) Although gaultite and VPI-7 are topologically identical, Ercit and van Velthuizen did not strictly consider them as natural and synthetic analogues since the ordering in VPI-7 remains unknown. Our results show that Zn, Si ordering does exist in VPI-7, although more work is needed to account for all of the observations in the 29SiMAS NMR spectra. A 29Si NMR spectrum of gaultite would be very informative and greatly assist in resolving the issues concerning the relationships between gaultite and VPI-7. To show how the 29SiNMR data can be used to assist the structure solution of new material, we investigated VPI-9, another of the microporous zincosilicates first synthesized by AnnenS2This material reversibly adsorbs 0.1 1 glg of water at 298 K and P H ~= O 0.026 atm and is stable up to at least 923 K. The 29SiMAS NMR spectra of samples synthesized by several different methods are shown in Figure 6, and the spectral deconvolutions are given in Table 4. Three striking features are apparent from these data. First, it is clear that there are at least two crystallographically unique silicon sites in VPI-9 that are of composition Si(3Si,lZn). Second, although the SUZn
Figure 6. 29SiMAS NMR spectra of the microporous tectozincosilicate VPI-9 synthesized by methods I, 11, and III (from bottom to top) (see Experimental Section). TABLE 4: 29Si MAS NMR Chemical Shifts and Tentative Assignments for WI-9 6, PPm re1 sample from TMS intensities assignment (Si/Zn)-93.6 65.4 Si(3Si,lZn) VPI-9(1) -96.7 34.6 Si(3Si,lZn) 4.0 -93.2 58.6 Si(3Si,lZn) VPI-9(II) -96.4 41.4 Si(3Si,lZn) 4.0 -94.1 79.8 Si(fSi,lZn) VPI-9(III) -97.5 20.2 Si(3Si,lZn) 4.0
-
ratio is near that for VPI-7, there are no spiro-5 units in VPI-9 (lack of resonances at -80 ppm). Third, the zinc is again showing some degree of order. For a material with Si/Zn = 4 and a random distribution of Zn and Si, there would be observable resonances for environments other than Si(1Zn) (using eq 11). Thus, these points can be used when developing structural models for VPI-9. Finally, the results presented here show that a high degree of Zn, Si ordering is frequently found in the frameworks of tectozincosilicates,even in samples synthesized by hydrothermal methods at relatively low temperatures (1473 K). It is observed that Zn shows a strong preference for sites with low mean TOT angles: in NazZnSiO4, the mean ZnOSi is 129.8', and in Na2ZnSi206, the mean ZnOSi angle is 125.5' (whereas for Si, the mean TOT angle considering all four angles is 133.4', while the mean SiOSi angle (only two TOT angles) is 141.3'). In KzZnSi4010, the mean TOT angle for Zn is 123.0 and for each Si in different crystallographic positions is 137.1, 133.6, 131.6, and 137.5'. For gaultite, Zn is present in every 3MR with a mean ZnOSi of 126.9', and the mean TOT angle for each Si is 139.4, 139.1, 134.1, and 121.2' (for the central Si in the spiro-5 unit). The strong preference of Zn for positions with rather small angles can be rationalized by taking into account the ratio L/NBR of Zn and Si, where L is the T - 0 distance and NBR is the nonbonded radius of element T.16 This ratio is 0.93 for Si and 0.83 for Zn and is a convenient way to show the preference that Zn has for occupying positions with small TOT angles. Be has a LNBR ratio of 0.82, and Be, Si ordering is found in beryllosilicates like lovdarite (with Be in 3MR") and beryl.18 In contrast, A1 has a L/NBR ratio of 0.92 and is much more like Si. Thus, the disorder in aluminosilicates is easy to
Camblor and Davis
13156 J. Phys. Chem., Vol. 98, No. 50, 1994 rationalize (in low-silica aluminosilicates like K-feldspars, for example, specific but not well-known conditions (like crystallization temperatures and cooling rates) are necessary to achieve some degree of Si, A1 ~rdering).'~The properties of Zn can account for both the Zn, Si ordering found in tectozincosilicates and for the effect of Zn in the synthesis of microporous materials. Substitution of Zn for A1 in aluminosilicate reaction mixtures has been shown to strongly direct the crystallization to materials that usually have no counterpart among aluminosilicate zeolites (like VPI-7, VPI-8, VPI-9, and VPI-10).2 As Zn is much different from A1 and Si, it promotes the formation of topologies with very small angles and with Zn in the crystallographic sites with small TOT angles. However, this does not happen in the central Si atom in the spiro-5 units of VPI-7 (this site has a very small TOT angle but is occupied by Si). The most likely reason for this is that if Zn enters the central position of the spiro-5 unit, then that unit will not be able to accommodate another Zn atom (due to the extended Loewenstein rule). Clearly two Zn atoms per spiro-5 unit release strain in that unit more efficiently than one unique Zn atom (even if this is in the central position). An exception to the Zn, Si ordering in tectozincosilicates is zinc sodalite, which according to synchrotron X-ray powder diffraction experiments shows a random distribution of Zn and Si throughout the framework.'O This may be due to the fact that, ideally, there is only one T site in the sodalite topology?0 or in other words, there is not a position for which Zn can show any preference. Conclusions The strong tendency for Zn to occupy specific positions (with small TOT angles) in the framework of tectozincosilicates has been shown by a combination of 29SiMAS NMR analysis and X-ray diffraction. This tendency promotes Zn, Si ordering in these materials, and 29Si MAS NMR spectroscopy can give insights complementary to the structure characterization by diffraction experiments. Thus, this technique will provide clues to the structure solution of microporous zincosilicates, which are usually microcrystalline and unsuitable for single-crystal diffraction. 29SiNMR can also give information on the actual symmetry of known zincosilicates frameworks, as resolution of crystallographic sites frequently appears. The large differ-
ences between Si, Al, and Zn, especially in terms of T-0 distances and T-T nonbonded radii explain why microporous zincosilicates usually display framework topologies much different from (and with higher T ordering than) aluminosilicate zeolites. Thus, new topologies can be expected to appear as zincosilicate, low-temperature, hydrothermal chemistry continues to be explored and understood. Acknowledgment. Support of this work was provided by the NSF Alan T. Waterman Award to M.E.D. M.A.C. gratefully acknowledges the Spanish Ministry of Science and Education and the Fulbright Commission for a postdoctoral fellowship and the Instituto de Tecnologia Quimica (Valencia, Spain) for permission to leave. References and Notes (1) Annen, M. J.; Davis, M. E.; Higgins, J. B.; Schlenker,J. L. J. Chem. SOC., Chem. Commun. 1991, 1175. (2) Annen, M. J. Ph.D. Thesis, Virginia Polytechnic Institute and State University, Blacksburg, 1992. (3) Annen, M. J.; Davis, M. E. Microporous Mater. 1993, 1 , 57. (4) Brunner, G. 0.;Meier, W. M. Nature (London) 1989, 337, 146. (5) Bialek, R. Institut fur Kristallographie und Petrografie, ETH-Z, KRIBER 1.0, 1991. (6) Struct. Rep. 1966, 31A, 218 and references cited therein. (7) Struct. Rep. 1981, &A, 345 and references cited therein. (8) Kohara, S . ; Kawahara, A. Acta Crystallogr. 1990, C46, 13731376. (9) Engelhardt, G.; Michel, D. High Resolution Solid-state NMR of Silicates and Zeolites; John Wiley & Sons: Chichester, 1987. (10) Camblor, M. A.; Lobo, R. F.; Koller, H.; Davis, M. E. Chem. Mater., in press. (11) Dollase, W. A. Personal communication, UCLA, June 1994. (12) Engelhardt, G.; Koller, H. NMR Basic Principles Prog. 1994, 31, 1. (13) Walter, F. Eur. J . Mineral. 1992, 4 , 1275. (14) Rohrig, C.; Gies, H.; Marler, B. In Book of Abstracts, 6th German Zeolite Conference; Gies, H., Ed.; Ruhr-Universiat: Bochum, March 7-8, 1994. (15) Ercit, T. S.; van Velthuizen, J. Can. J . Mineral., in press. (16) O'Keeffe, M.; Hyde, B. G. In Structure and Bonding in Crystals; O'Keeffe, M., Navrotsky, A., Eds.; Academic: New York, 1981; Vol. I, p 227. (17) Merlino, S. Eur. J . Mineral. 1990, 2, 809. (18) Struct. Rep. 1968, 33A, 480 and references cited therein. (19) Griffen, D. T. Silicate Crystal Chemistry; Oxford University: New York, 1992. (20) Meier, W. M.; Olson, D. H. Atlas of Zeolite Structure Types, 3rd ed.; Buttenvorth-Heinemann: London, 1992.