New structures—new insights: Progress in structure analysis of nanoporous materials

ELSEVIER Microporous and Mesoporous Materials 21 ( 1998) 183-197

New structures -new insights: Progress in structure analysis of nanoporous materials

Hermann Gies ‘,*, Bernd Marler ‘, Silke Vortmann a, Uwe Oberhagemann a, Parwis Bayat a, Kirsten Krink a, Jordi Rius b, Ingo Wolf a, Colin Fyfe ’

a Inst. fiir Mineralogie, Ruhr-Universitdt Bochum, 44780 Bochum, Germany b Institut de Ci2ncia de Materials de Barcelona, CSIC, Bellaterra, Catalunya, Spain

’ Depr. qf Chemisfry, UBC Vancouver, Vancouver, Canada

Received 13 September 1997; received in revised form 10 November 1997; accepted 17 December 1997

Abstract

In the recent past structure determination of microporous materials has experienced considerable developments in methodology. The FOCUS method: high resolution powder diffraction data used for direct method structure solution, in combination with crystal chemistry based modelling. The models are retrieved from electron densit)r maps calculated in direct method runs, energy minimized and checked through for realistic angles and distances values. The SUM-TF method: diffraction patterns at moderate resolution analysed with direct methods using a modified tangent formula which includes Patterson information for the structure solving. In this way the atomic resoiution criterion for direct methods is bypassed. This overview gives a summary of the structures successfully solved using these new techniques. 0 1998 Elsevier Science B.V. All rights reserved.

Keywords: Powder diffraction; Structure solution; Rietveld analysis; Zeolites; Layer silicates; VNI; RUB-15; RUB-18; Kanemite

1. Introduction

Nanoporous materials can be either crystalline or amorphous solids which are widely used as ion exchangers, heterogeneous catalysts, supports, and sorbents [I]. Their chemical and physical proper- ties are closely related to the specific crystal struc- ture of the material, rendering structure characterization an indispensable task [2].

* Corresponding author. Fax: +49 234 7094 433; e-mail: [email protected]

Researchers working in the fields of synthesis, modification, and application of nanoporous mate- rials extensively use detailed structural knowledge in order to achieve meaningful progress in their respective fields. Many different techniques are used to describe the structural properties of a material; however, all information gained is restricted to the particular probe/material inter- action. Whereas diffraction studies probe the long range periodicity of crystalline materials, spectro- scopic techniques are most sensitive to changes in the local atomic environments in crystalline and amorphous materials. Sorption and diffusion

1387..181 l/98/%19.00 0 1998 Elsevier Science B.V. All rights reserved. PII: St387-1811(98)00003-l

184

Table 1 Flow diagram of the structure determination process from powder data including the FOCUS routine

The FOCUS Environment

Powder Data Collection

1 Intwlsily Exlnlction

1 >I F, Scale Facta Detamitmion

1 Partitioning of Ovuiapping Intensities

7

Atttomatic Fourk Recycling and Topology Search F

1 : u Classiticadon and Sating of Topologies S

I 7

Distance Least Squares

1 - Rietvveld Retinement and Difknmx Fourier Analysis

experiments with probe molecules, thermal studies, catalytic test reactions are other sources of struc- tural information in a less direct approach.

For crystalline materials, diffraction experiments are the most powerful tools for structural charac- terization. Because of their wide availability, X-ray diffraction experiments are most important for structure analysis whereas electron and neutron diffraction experiments are still reserved for the specialist. Since crystalline nanoporous materials, i.e. zeolites and layer silicates, are synthesized and used as microcrystalline powders, X-ray powder

Fig. 1. The 4.s2 layer of the structure of VPI-9 (LLBU B). The same layer is also found in the structures of the zincosilicates VPI-7 and RLJB-17.

diffraction experiments have to be employed. However, the structural information in powder diffraction data is reduced from three-dimensional to one-dimensional space and, therefore, it is most important to have additional information available from other, complementary characterization tech- niques such as chemical analysis, electron diffrac- tion. IR and NMR spectroscopy and sorption studies. This leads to the most complete knowledge of the physical and chemical properties of the material.

In the recent past structure determination and Rietveld analysis have become almost routine for highly resolved diffraction data sets from materials with small unit cells and low systematic overlap of diffraction signals [3,4]. The latest developments in the area have focused on the extension of the limits, in particular the solving and refining of more complicated structures from powder data and also the application of the technique to notori- ously poorly crystalline and defect-loaded materi- als. This article will present the concepts of two new X-ray structure analysis experiments which were specifically developed for the structure solu- tion of nanoporous host frameworks. The different approaches have been successfully tested on a

H. Girs et al. : Microporous and Mesoporous Materials 21 ( 1998) 183-197 la5

Fig. 2. The second layer-like building unit (LLBU A) of the structure of VPI-9 together with a model of the silicon skeleton of the framework structure.

number of previously unknown structures, some of which will be presented in this article.

2. New zeolite structures solved from high resolution X-ray powder diffraction data

2.1. Structure solution trchniqw.~

In pioneering studies Baerlocher and McCusker successfully showed that high resohrtion powder diffraction data can be used quite routinely for the solution of simple crystal structures. They used essentially the standard techniques developed for single crystal structure determinations adapted to the specific necessities of the powder experiment. The breakthrough was achieved with the solution of the structure of zeolite sigma-2 (SGT) [5] and a summary of the development is given in ref. [3]. This was followed by the consequent development of concepts in order to overcome the inherent problems of signal overlap [6,7] and lack of reso- lution in powder diffraction together with the massive use of computer power for the trial and error testing of potential model structures [8]. This is the state of the art and was introduced recently as the FOCUS method (Table 1). In the following, a brief outline of the basic principle will be given.

The superposition of all diffracted intensity in one dimension in the powder diffraction experi- ment creates severe accidental and systematic over-

Table 2 Summary of structural data of the zincosilicate VPI-9

Chemical composition per unit cell Space group symmetry

Lattice parameters +I (A, b,(A) co (A,

Wavelength (A) 28 range (- ZR) Step width (” 2N) Number of reflections Number of soft restraints Number of structural variables R w R -P

Rb,&[S&,Zn,,O,.,,,j .48H,O I: P4,/ncm (averaged symmetry for structure solution) (No. 138) II: P4,2,2 (true SC symmetry for Rietveld analysis) (No. 92)

I: 9.8946 I: 9.8946 I: 36.8715 I: 0.94734 I: 2 55 I: 0.01

11: 9.8837 II: 9.8837 II: 73.6505 II: 0.99995 II: 2.5-60 II: 0.01 II: 1906 180 170 0.147 0.099

186 H. Gies et (II. i Microporous and Mesoporous Materials 21 (1998) 183-197

Table 3 Summary of structural data of the tetramethylammonium silicate RUB- 15

Chemical composition per unit cell Space group symmetry Lattice parameters 0, c“k ho (A) (‘. (Aj

Wavelength (A) 20 range (’ 20) Step width (” 20) Number of reflections

[N(CH,),I,[S~,,O,,(OH),I~~~H~O Iba2 (No. 45)

27.905(6) 8.408(l) 11.518(2) 1.5406 S-65 0.00776 260

Number of soft restraints Number of structural variables FWHM at 20.0” (” 28) R w R CXP

45 57 0.115 0.114 0.089

SCHAKAL

Fig. 3. View parallel to the layer of the TMA silicate The silicate layer terminates the clip. The hydrogen bond network connecting the neighbouring layers via hydrate water is shown in broken lines. For clarity, the templating TMA molecule is shown as a large sphere, oxygen atoms representing water molecules and the silanol group are slightly smaller, bridging oxygen atoms are even smaller, silicon atoms are smallest and hydrogen atoms are omitted.

laps of diffraction peaks. Whereas the choice of wavelength and resolution enhancing techniques

obvious remedy for symmetry related systematic overlap. In a high resolution data set 01’ the

can tackle accidental peak overlap, there is no hexagonal zeolite EMT crystallizing in space

If. Gies ei al. / Microporous and Mesoporous Materiuls 21 (1998) 183-197 18’7

Fig. 4. Space-filling model of the TMA cation within the cage-like environment in RUB-l 5

group P6,/mmc recorded with I= 1.39 A to a reso- lution of d= 1.10 A, 76% of 1158 reflections overlap, precluding routine structure solution with direct methods. Here a reflection of the type Ml= 123 coincides with 15 other, symmetrically equivalent reflections. In the early stages of structure solution from powder data the measured intensity was equ- ally partitioned amongst the contributing reflec- tions, introducing a considerable uncertainty in the calculations. However, in many cases the informa- tion was very helpful in the solution process although it was only a crude estimate. The experi- mental veritication of the model structure is finally realized with the Rietveld refinement of the intensity data set leading to the real structure model.

As a first major improvement, in 1991 Estermann et al. introduced the FIPS method (fast iterative Patterson squaring) into the structure solving process from powder diffraction data 171. Here the intensity partitioning of overlapping reflections based on the calculation of the Patterson function using all reflections leads to an improved data set. This was for the first time

successfully proven with the then unknown struc- ture of SAPO-40 [9].

So far, for the solution of zeolite crystal struc- tures, crystal chemical information has only been exploited in classical model-building attempts. Apart from building physical models derived from existing structures [lo], a simulated annealing rou- tine was developed which created possible tetrahe- dral networks using the symmetry and metric information from the diffraction experiment [ 111. Even an attempt to determine rigorously all pos- sible tetrahedral structures subject to particular symmetry conditions has been made for a symme- try subset in the hexagonal system [12]. However, in all cases the researcher himself had to choose the appropriate model and had to evaluate it, against the experimental data. Recently, a new approach has been put into reality, combining experimental, direct method based structure solu- tion techniques with computer assisted model building in a recursive approach [ 131. Based on the FIPS assisted intensity partitioning, the so-called FOCUS technique generates an electron

I t a

Fig. 5. Structure of RUB-18 showing the skeletal model of the silicate layer with the sodium/hydrate- water octahedra intercalated.

density map from a starting set of phased diffrac- tion intensities, usually generated by a direct method structure solution program. In a peak search a well defined set of maxima is selected and geometrically optimized. Combining the space group symmetry and crystal chemical information for zeolite tetrahedral networks, a model structure is deduced and unambiguously characterized in its coordination sequence. From this model a new phase set is calculated and tested for its con- vergence against the experimental data set. In this

way a series of model structures is generatIed by the multi-solution direct methods approach, all of which are geometrically tested and classified. In the analysis of all models created by using different sets of starting reflections and phase assignments, the correct model usually shows up many times as a solution and by far outnumbers ‘other’ propos- als. FOCUS has been very successfully tested with a number of zeolite structures and so far has solved a series of complex new zeolite crystal structures with unexpected new topologies.

It should also be mentioned that computer mod- elling of crystal structures has become an irnpor- tant technique for structural studies, so far used mainly as an additional tool for structure elucida- tion [ 141. The important contribution to the struc- ture solving process is the verification of a proposed model. Here, for complicated structures only well energy minimized starting models should be used for the Rietveld refinement, which essen- tially proves the correctness of the proposed model.

2.2. New zeolite crystal structures

An excellent example of the power of the FOCUS method is the structure solution of the zincosilicate zeolite VPI-9, structure code VN I [ 151 (Table 2). Early attempts to solve the structure all failed because the indexing of the powder pattern was ambiguous. Several small peaks indicated a doubling of the c-parameter from -37 .& to - 74 A and, in addition. a few peaks could not be indexed and were attributed to impurity phases, Only after optimizing the synthesis conditiorrs was a pure and highly crystalline sample obtained whose high resolution synchrotron powder data set was indexed unambiguously. After the extrac- tion of integral intensities from the powder data, a large number of electron density maps were generated from random starting phases. These were then subjected to a Fourier recycling pro- cedure combined with a specialized topology search routine optimized for three-dimensional four-connected frameworks consistent with typical angles and distances for zeolite structures. A detailed description of the structure solving prlocess is given in ref. [ 1.51. More details on the structure solution are given in Table I.

Table 4 Summary of structural data of the sodium silicate RUB-18

Chemical composition per unit ceil Space group symmetry Lattice parameters a, ih h, (A, ‘h(A)

Wavelength (A) 28 range ( 28) Step width (’ 20) Number of reflections Number of soft restramts Number of structural variables FWHM at 18.5‘ (’ 20) R v’ R CXF

Na,[Si,,O,(OH )s] .32H,O; Z= 8 Iil,/urnrl (choice 2) (No. 141 )

7.3276( 1 ) 7.3276( 1) 44.3191(61 I .5406 6 75 0.00776 18X 5 I9 0.109 0.167 0.146

Applying FOCUS the framework of the topol- ogy of VNI emerged as the most probable one. A flow diagram of the general routine for the struc- ture solution from X-ray powder data including FOCUS is given in Table 1. The structure can be composed from two building blocks which stack in an alternate fashion. The 4.8’ (UUDDUUDD sequence of tetrahedra [ 151) layer is well known and has already been found in the zincosilicate zeolites VPI-7 (VSV) [16] and RUB-17 (RSN) [ 171 (Fig. 1). The second layer is unique for VNI. Here pairs of [S33] polyhedra share common 3-rings and are joined to adjacent pairs via edges to form sinusoidal chains (Fig. 2). Similar to other zincosilicate zeolites such as VSV and RSN, the Rietveld refinement of the powder data set revealed that zinc is perfectly ordered on specific T-sites which are all part of 3MR. There is also a detailed analysis of the non-framework constituents Rb’, Kf, and water which are coherently ordered in the channel voids. Although the hydrogen bond network could not be fully established as in the more simple zincosilicates VSV and RSN, the quality of the Rietveld analysis of 59 atoms with 170 structural parameters in the asymmetric unit within a unit cell of a 7189 A” in its details and precision is remarkable. The combimttion of MAS 29Si NMR experiments with the structure retine- ment further strengthens the analysis, in particular the perfect ordering of zinc on T-sites in the

framework and the lowering of symmetry through a doubled c-parameter.

FOCUS has been used successfully for a number of test structures and, in addition, also applied to less complicated unknown zeolite frameworks. So far, the method is restricted to three-dimensional four-connected framework structures for which the framework search routine was developed. Due to the limited computing time available there is also a limit in the complexity of structures which can be treated so far. This limit will certainly be extended with every new generation of computers. All technical details are discussed at length in ref. [8]. A complete list of the successful applica- tions can be obtained from the author on web site http:/,latb.csb.yale.edu/-rwgk/focus-structures.ps.

3. New microporous crystal structures solved from low to moderate resolution X-ray powder diffraction data

3.1. Structure solution technique

In many cases the poor crystal quality of the material is the resolution-limiting factor in the powder X-ray experiment. Broad diffraction peaks are, therefore, an inherent property of the material unless new synthesis strategies are successful in

190 H. Gies et al. / Microporous and Mesoporous Materials 21 (1998) 183-l 97

Fig. 6. ‘H spectrum of RUB-18 showing two proton resonance signals. The low field signal (- 16 ppm) is indicative of strong hydrogen bonding and is clearly separated from the hydrate water proton signal.

improving the crystallinity. Even with high inten- sity and high resolution synchrotron X-ray diffractometers, diffraction peaks can only be indexed unambiguously to d-values of 2-2.5 A which is far above atomic resolution. Diffracted X-ray intensity from higher angles can still be recorded in the powder X-ray diffraction experi- ment; however, the peak assignment is no longer unambiguous because of severe overlap. All diffraction information collected in a powder X-ray experiment should, however, be used for the Rietveld analysis even if only broad and featureless signals are recorded at high 20 angles. Thus, the structure refinement leads to almost atomic reso- lution in many cases.

In the preparation of the crystal structure solu- tion the indexing and lattice parameter refinement has to be performed. Automatic routines very often fail with low resolution powder data sets since the precision of the cl-values is rather low and much information in the pattern is lost through peak overlap. The use of complementary

techniques is, therefore, required of which electron diffraction on rotation/tilt stages is the most valuable.

According to Sheldrick’s rule, at least 50% of the diffracted intensities in the range between 1 and 1.1 A should be observed with statistical signi- ficance in order to solve crystal structures with direct methods [ 181. The rule was derived for organic crystal structures where peak separation in the electron density map between neighbouring carbon atoms with d(C-C) = 1.54 A defines the limit. In the case of inorganic tetrahedral frame- works, the [TO,] tetrahedron might be regarded as a rigid and invariant subunit of the crystal structure. Once the centre of mass of such a unit is known the complete network can be derived. With a d(T-T) of more than 3 A for silicates, phosphates, aluminates, etc. the resolution require- ment for the experimental diffraction data is con- siderably lowered to 2.5-2 A. In standard diffraction experiments with Cu Ka, radiation (;L = 1.5406 A) unambiguous indexing of the X-ray

H. Gies et al. / Microporous and Mesoporow Materials 21 (1998) 383-197 101

I ' 1 - I ' 9 8 I - 1. I - I., I.. , WI -90 -100 -iI0 -120

Fig. 7. 2D ‘H-*“Si CP MAS spectrum of RUB-I 8 showing the magnetization transfer to the silicon nuclei from the low field proton involved in strong hydrogen bridge bonding between the two Si-0 groups.

powder pattern up to approximately 40” 2% is required which can be achieved in many poorly crystalline materials.

In recent years for the direct method structure solution a robust and effective new tangent for- mula, the SUM-TF, has been derived which takes advantage of structural information gained from Patterson arguments [ 191. In electron density maps calculated from low or moderate resolution diffraction data, i.e. in the resolution range d= 2- 1 .S A, it is interesting to notice that [TO,] units appear as isolated peaks thus allowing for the derivation of the T-T bonding network. In many cases the interpretation of the electron density maps is far from trivial and is assisted by comple- mentary structural information including NMR, electron diffraction, and computer modelling. It should be noted that the approach is not limited to zeolites but holds for all structures with similar rigid subunits.

3.2. New structures of microporous materials

In a short review three new crystal structures of hydrous layer silicates will be presented as typical cases for materials with diffraction properties giving only moderate to low resolution in the X-ray pattern. Since these materials dehydrate easily and undergo a series of phase transitions at moderate temperatures and also decompose at elevated temperatures, carefully controlled diffrac- tion experiments had to be carried out. The exam- ples chosen also underline that the strategy to solve structures from X-ray powder data at moder- ate solution is not restricted to zeolites but is of general use for crystal structure analysis. Once the framework topology of the silicate structure is known approximately, the model can be optimized with distance least squares (DLS) [20] or energy minimizing techniques (e.g. METAPOCS) [21 J and the details of the structure can be derived from a

192 H. Gie.5 ei al. i Microporous and Mesoparous Mareriuls 21 i 1998) 1X3- I97

,T,“‘~-,-“rrrrn~,r-n-r-mmrrrm-~ w -40 -60 -80 ---x “““--‘;‘;d”“““‘Illllrr7rmmnmmrm -11(0

Fig. 8. !H-‘%i CP MAS ‘%i NMR spectrum of the H-form of RLJB-IX showing the integrity of the silicate layer

full pattern Rietveld refinement using Fourier analysis and investigations with other complemen- tary techniques such as electron diffraction and solid state NMR. All information gathered about the material together with the diffraction analysis leads to a most complete picture of the structure with atomic resolution.

3.2. I. The crystal structure of‘the neM‘ tetrumethylammonium luger silicate RUB-15 [22/

A new tetramethylammonium silicate RUB- 15. [ N(CH,),],[ SiS,O,,(OH ),I .20H,O, has been syn- thesized from aqueous silicate solutions under autogenous pressure at 120 to 140°C. The perfect cleavage perpendicular to [ 1001 and the 2gSi MAS NMR spectrum indicated a layered structure with Q”/Q4 ratio of 2:l. Th e material was not only a microcrystalline powder but also of poor crystall- inity. Since there is no need for very high resolution diffraction using synchrotron instrumentation, the powder data set for the structure solution was collected on a laboratory SIEMENS D-5000 powder diffractometer with transmission geometry and a 6” OED detector system. The powder pattern was unambiguously indexed up to -45 20 corre-. sponding to 2 w d-spacing. This was sufficient to

resolve neighbouring [SiO,] tetrahedra in the electron density map which was calculated with the direct method program XLENS [23] using the SUM-TF for the phase refinement. The topology of the silicate layer was completed with the oxygen atoms connecting the T-centres and the terminat- ing silanol groups. The silicate layer network was optimized with the distance least squares program DLS [20] and used as such in the course of the Rietveld analysis. From Fourier synthesis the tem- plating tetramethylammonium cation was located and the oxygen atoms of the hydrate water also determined. Using diffraction data up to 65’ 20 (- 1.43 A d-spacing) in the full pattern Rietveld analysis, the structure model was refined to x2 = 1.64 (R,, = 11.4%). More details on the :;truc- ture analysis are summarized in Table 3.

The structure shows a remarkable bonding net- work. The silicate layer is a slice of the sodalite structure with open P-cages alternating on either side of the layer. Neighbouring silicate layers are connected through the hydrogen bonding network involving the silanol groups and the hydrate r.vater creating cage-like cavities for the tetramethylam- monium template (Figs. 3 and 4). It is interesting to note that the silicate/hydrate-water network

H. Gies et ui. i Micropurous und Mesoporous Materials 21 i 19981 183 -~I97 lY3

resembles the sodahte structure with a single T-site as defect site (see also Fig. 4). All attempts to introduce a scatterer on the defect site in order to complete the sodalite network failed throughout the Rietveld analysis of the powder X-ray data set leading to significantly lower R-factors. The oxygen-oxygen distances in the water network including the silicate surfafe Q” silanol groups range between 2.5 and 2.9 A, indicating classical hydrogen bridge bonding and the spatial distribu- tion of the negative charge.

Since the scattering contribution of the proton to the X-ray diffraction process is only very weak. ‘H and 2D MAS NMR experiments were carried out. In particular the location of the proton, the H/D exchange and the nature of the hydrogen bridge bond were interesting details inaccessible from X-ray Rietveld analysis. However, these important physical properties of the solid are closely related to typical applications of layer silicates. In a “H MAS NMR spectrum a small but significant signal at very low field, y 16 ppm,

is observed, indicating strong hydrogen bridge bonding. It is surprising that the low field signal is distinct and exchange between different proton’ species is considerably suppressed. Comparison with the results from the Rietveld structure refinement indicates that the protons of the silanol group should be involved in strong hydrogen bridge bonds. The distance between the neighbour- ing silanol oxygen atoms has been determined to be la(O-0) =2.6 A.

3.2.2. The structure of the sodium layer silicate RUB-18 [24]

There is a large family of crystalline, natural and synthetic hydrous sodium layer silicates [25]. Knowledge of the crystal structures of these mate- rials is poor and based only on models which were derived from the indexing of the X-ray powder pattern and 29Si MAS NMR experiments. So far only the structure of makatite is well understood and this served as the basis for the structure models of the other phases.

Fig. 9. ‘“Si-~MAS NMR spectrum of the water-free, high temperature form of H-RUB-18 showing a noticible shift of the Q3 silicon signal to higher field upon dehydration.

194 H. Gies et al. / Microporous and Mesoporous Materials 21 (1998) 183-197

The sodium silicate RUB-18, Nas[Si,,O,, (OH )s] s 32H,O, showed remarkably good crystall- inity and the X-ray powder pattern was indexed unambiguously up to 28=65” (Cu Ka, radiation) with lattice parameters of a0 = bo= 7.328 A and co=44.319A in the tetragonal system (Table4). After extraction of 138 integrated intensities, a structure model for the silicate layer was derived from the electron density map calculated with the SUM-TF routine of the direct method program XLENS. The model was in full agreement with the results from 29Si MAS NMR spectroscopy and had Q” and Q4 silicon atoms in the ratio 1: 1. The completion of the crystal structure was performed as described above using a DLS optimized silicate layer for the search of the missing intercalated species with Fourier analysis in the course of the Rietveld refinement.

Again, the silicate layer showed a surprising resemblance to a previously well known feature in zeolite structures. [54] cages known as building units in zeolite structures as MFI, MEL, and beta are linked via oxygen bridges to a two-dimensional array (Fig. 5). Neighbouring layers are rotated by 90” and house one-dimensional chains of sodium cations octahedrally coordinated by hydrate water. The terminating silanol groups of the silicate layer interact with the intercalate building hydrogen bridge bonds. The details of the crystal structure and in particular of the arrangement of the interca- lated species have all been derived from careful Rietveld analysis of the laboratory X-ray powder data set. However, the hydrogen bridge bonding network has been derived from only the oxygen positions representing the silanol group and water molecules and still requires experimental verification.

In a series of ‘H and 29Si MAS NMR experi- ments, attempts have been made to identify the different proton species which are observed in the regular ‘H MAS spectrum (Fig. 6-8). From two- dimensional ‘H-Z9Si cross-polarization (CP) spectra [26], the A 16 ppm low field signal can be assigned to the silanol group of the sihcate layer (Fig. 7). The unusual proton shift indicates strong hydrogen bridge bonding between the neighbour- ing Si-0 groups which is also confirmed in the X-ray analysis where the distance between the

Fig. IO. Structure of kanemite showing the boat-shaped 6MR of the silicate layer and the distorted sodium/hydrat8e-water octahedra.

silanol oxygen atoms 04 was determined to be d(04-04) = 2.3 A. In order to confirm the result from the CP experiment, partially D,O exchanged samples were investigated using ‘H-NOESY experiments. The spectra show that for the temper- ature at which the experiment was carried out, i.e. ambient temperature, spin diffusion was mmimal and chemical exchange between the bridging hydrogen and the water molecules was suppressed to almost zero.

An interesting feature of RUB-18 is its ion exchange capability. in particular its conversion to the H-form. The removal of the sodium ion essen- tially leads to proton terminated neutral silicate layers. The ion exchange is accompanied by a reduction of the c-parameter from 44.3 A to 29.7 A, i.e. about 30%. whereas the a- and b-

H. Gies et al. 1 Microporous and Mesoporous Materials 21 (199Sj 183-197 195

Table 5 Summary of structural data of the sodium silicate kanemite

Chemical composition per unit cell Space group symmetry Lattice parameters a0 (A) bo (A) co (4

Wavelength (A) 20 range (” 28) Step width (” 28) Number of reflections Number of soft restraints Number of structural variables FWHM at 20.0” (” 20) R VP R CXP

NaH[Si,O,].3H,O; 2=4 Phcn (No. 60)

4.946( 1) 20.509( 1) 7.276( 1) 1.5406 8-98 0.007778 369 0 23 0.11 0.102 0.070

.-

parameters remain unchanged. The crystallinity of the exchanged product decreases drastically, making structure solution almost impossible. However, the 29Si NMR spectrum clearly shows that the silicate layer has preserved its structure (Fig. 9). However, the clear ‘H NMR spectrum of the sodium form is lost and replaced by a broad and featureless signal with a maximum at about 7.3 ppm indicating strong interactions between the different proton sites. Upon heating, the 29Si MAS NMR spectrum of the sample changes signifi- cantly, leading to an upfield shift of the Q3 signal from -99 ppm to - 102 ppm at 100°C while still maintaining the Q3/Q4 ratio of 1: 1 (Fig. 9). Rehydration only partially converts the sample back to its room temperature form. Attempts to condense neighbouring layers to a three-dimen- sional silica framework are currently being carried out. Extensive heating of the sample could perhaps lead to a phase transformation to a three-dimen- sional network with so far unknown framework topology.

Finally, it should be mentioned that the X-ray powder diagram of RUB-18 closely resembles those of so-called Ilerite [27] and octosilicate [28]. Since a structure model for Ilerite and octosilicate has been proposed and is completely different from the structure of RUB- 18 presented here, a detailed investigation of the relationships between these materials is required.

32.3. The crystul structure of the sodium layer silicate kanemite

Kanemite, NaH [Si,Os] + 3H20, was first synthe- sized by Kalt and Wey in 1968 [29] and later discovered in evaporites of lake Kanem as a natu- ral mineral [ 301. The material belongs to the family of sodium disilicates of which the water-free sodium derivative, F-Na&O,, is used as a builder in washing powders in combination with, or as an alternative to, zeolitic materials as a substitute for the classical phosphates [ 3 11. Recently, intercalates of kanemite have been converted to ordered meso- porous materials FSM-16 [ 321, which are relateid to the MCM-41 family of ordered mesopororus structures. Although specific properties related to the various applications have been extensively studied, there is no crystal structure analysis of the material so far and, therefore, most of the understanding of the physical and chemical proper- ties is model-based. From a pure sample of kanetn- ite of moderate crystallinity, 169 integral intensities were extracted from the X-ray powder diagra:m and used for the direct method structure solution as described above. The details are summarized ,in Table 5. Again, the electron density map clearly revealed the topology of the silicate layer and a crude understanding of the structure of the interca- late. The subsequent full pattern Rietveld analysis led to a complete picture of the crystal structure including its hydrogen bridge bonding network.

196 H. Gies et al. i Microporous and Mesoporous Materials 21 (19981 183-197

The structure of kanemite is built from 6MR silicate layers of Q3 species with the 6-rings in a boat conformation. Neighbouring layers are sepa- rated by layers of corner and edge sharing [NaO] octahedra (Fig. IO). The detailed analysis of angles and distances confirms the distortion of the sodium octahedra predicted from 23Na MAS NMR experi- ments. It is also interesting to notice that the intercalate sodium hydrate layer is centrosymmet- ric between neighbouring silicate layers and vice versa. Although the structure analysis confirmed the proposed structure of the silicate anion as being isostructural with water-free Ba-silicate sanbornite or with KHSi,Q,, no details were known about the sodium/water intercalate. The Rietveld refinement clearly shows the coordination and location of the sodium cations and indicates the hydrogen bond network between the water and silanol hydrogen atoms. In order to establish the full details of the structure of the intercalate. similar NMR experiments as described for RUB-18 are presently being carried out on different hydration states of kanemite. A detailed report on the structure solution of kanemite is in preparation [ 331.

4. Conclusion

Most of the new crystal structures in zeolite science in the past five years have come from X-ray powder diffraction data analysis. These also include less well crystallized materials giving only moderate resolution in the diffraction experiment. However, in combination with other, complemen- tary techniques a more complete picture of the crystal structure can be gained. NMR spectroscopy is a powerful tool, in particular when NMR sensi- tive nuclei are involved as in the case of the hydrous layer silicates presented here. Although the analysis of diffraction experiments is highly automated for single crystals. the interpretation of powder data, especially those with low resolution. is far from trivial. Structural studies still require expert crystallographers.

References

[I] J.C. Jansen, M. Stocker. H.G. Karge, J. Weitkamp (Eds.). Studies in Surface Science and Catalysis. vol. 85. Advanced Zeolite Science and Applications, Elsevier. Amsterdam, 1994.

[2] P.B. Venuto. in: H Chon. S.-K. Ihm, Y.S. Uh (Eds.). Studies in Surface Science and Catalysis, vol. 105, Progress in Zeolite and Microporous Materials. Elsevier. Amsterdam, 1997, p. Xl I.

[3] L.B. McCusker. Mater. Sci. For. 133.. 136 ( 1993) 423. [4] K.D.M. Harris M. Tremayne, Chem. Mater. 8 (1996)

2554. [5] L.B. McCusker, J. Appl. Cryst. 21 (1988) 305. [6] W.J.F. David. J. .4ppl. Cryst. 20 (1987) 316. 171 M.A. Estermann, V. Grdmhch. J. Appl. Cryst. 26 (1993)

396. [8] R.W. Grosse-Kunstleve. LB. McCusker, Ch. Baerlocher.

J. Appl. Cryst., in press. [9] M.A. Estermann. L.B. McCusker. Ch. Baerlocher,

J. Appl. Cryst. 25 (1992) 539. [ 101 G.T. Kokotailo. CA. Fyfe, H. Gies, D.A. Cox. in: Studies

in Surface Science and Catalysis, vol. 49. Zeolitcx Facts, Figures, Future, Elsevier. Amsterdam, 1989, p. 7 15.

[I I ] M.W. Deem, J.M. Newsam. Nature 342 (1989) 260. [I?] M.M. Treaty, S. Rao. J. Rivin, in: R. van Ballmoos. J.B.

Higgins, M.M.J. Treaty (Eds.), Proc. 9th Int. Zeolite C’onf.. Butterworth-Heinemann, Stoneham, 1993. p, 381.

[ 131 R.W. Grosse-Kunstlevc, Zeolite structure determination from powder data: C’ompuler-based incorporation ofcrys- tal chemical information. Ph.D. Thesis, ETH Zurich. 1996.

[14] C.R.A. Catlow (Ed.), Modelling of Structure and Reactivity in Zeolites. Academic Press, London. 1992.

[ 151 L.B. McCusker, R.W. Grosse-Kunstleve. Ch. Baerlocher. M. Yoshikawa. M.E. I)avis, Microporous Mater. 6 (1996) 295.

1161 C. Rohrig, H. Gies. B. Marler. Zeolites 14 ( 1994) 498. il7j

[IsI

[‘Yl [N]

[?I1

[22]

[?3]

[24]

[2S] [26]

C. Rdhrig H. Gies, Angeu. Chem.. Int. Ed. Engl. 34 (1995) 63. J. Rius, J. Sane, c‘. Miravitlles, H. Gies, B. Marler, U. Oberhagemann, Acta Cryatallogr. A51 ( 1995) 840. J. Rius. Acta Crystallogr. A49 ( 1993) 406. Ch. Bderlocher, A. Hepp. W.M. Meier. DLS-76 4 pro- gram for the simulation of crystal structures by geometric refnement. Institute of C’rystallography and Petrography. ETH Zurich, Switzerland, 1977. S.C. Parker, A.N. C’ormack. C.R.A. Catlow. .4cta C‘rystallogr. B40 ( 1984) 200. U. Oberhagemann. I’. Bayat, B. Marler, H. Gies, J. Rius, Angew. Chem.. Int. Ed. Engl. 35 (1996) 2869. J. Rius. XLENS A program for crystal structure deter- mination. ICMAB-CSIC. Catalunya. Spain. 1994. S. Vortmann. J. Rius, S. Siegmann. H. Gies. J. Phys. Chem. RIOI (1997) 1292. H.-P. Rieck, Nachr. Chem. Tech. Lab. 44 (1996) 699. P. Caravatti, G. Bodenhausen. R.R. Ernst. Chem. Phys. Lett. 89 (1982) 363.

H. Gies et ~11. i Microporous and Mmqmous Materials 21 (1998) 183.-197 197

[27] K.H. Bergk, W. Schwieger. M. Porsch. Chem. Tech. 36 (1987) 459.

[28] R.K. Iler, J. Colloid Sci. 19 (1964) 648. [29] A. Kalt, R. Wey, Bull. Sot. Fr. Mineral. Cristallogr. 94

(1968) 1.5. [30] 2. Johan, G.F. Maglione, Bull. Groupe Fr. Argiles 20

(1968) 205.

[31] F. Wolf, W. Schwieger. 2. Anorg. Allg. Chem. 457 (1979) 1779.

[32] S. Inagaki, A. Koviai, N. Suzuki, Y. Fukushima, K. Kuroda. Bull. Chem. Sot. Jpn. 69 (1996) 1449 and refer- ences cited therein

[33] S. Vortmann. J. Rim, B. Marler, H. Gies, submitted to Eur. J. Mineral.