"W

henever you have a novel structure you usually get novel properties." Edith M. Flanigen [1] (First winner of the International Zeolite Association lifetime contribution and achievement award).

Zeolites and related microporous materials contain regularly spaced molecule-sized pores and are fascinating examples of molecular engineering. More than just aesthetically inspiring, as can be seen in Fig. 1, these materials represent billions of dollars in commerce, as they have widespread industrial applications: Materials of this type are used as catalysts to produce gasoline and pharmaceuticals. For medical and industrial purposes, they are employed to separate N2, O2, and other gases. Formulated in household detergents, they remove the calcium ions that make water "hard," replacing environmentally unfriendly phosphates. Zeolites and related microporous materials are also employed to sequester radioactive ions for bioremediation. Many new applications are being investigated in areas such as selective membranes, batteries, and fuel cells.

Properly used, the term

zeolite should be restricted to naturally occurring aluminosilicates. However, related microporous materials (herein called zeolitic) have been synthesized, where elements such as B, Be, Cr, Fe, Ga, Ge, Mn, P, Ti, and Zn substitute for tetrahedrally bonded Si or Al. These atoms have roughly the same bonding configurations, so they tend to adopt the same framework connectivities as aluminosilicates and thus have the same pore structures. Further, since each tetrahedral (T) atom is linked by four oxygen atoms, the frame

work must have a generalized chemical formula [702],.,"[T’O2]," where the charge on each tetrahedral unit (m or n) is typically zero or -1. Additional extra-framework cations, such as H+, Na+, Lit, or Ca2+, are then needed to balance changes. The number and nature of these cations is significant as cations are often the chemically active site for interactions with guest species, or their presence may affect the framework properties. Thus, they may dictate the catalytic or sorptive properties of the material. Empirically, the most negatively charged zeolitic frameworks that have been found, e.g., those that have the largest number of extra-framework cations, are materials with formula M+[SiO2][AIO2]'. Pauling's rules, restated in the zeolite field as Lowenstein's rule, dicate that the ratio of Al:Si cannot exceed unity, limiting the cation fraction to 1/7 the total number of atoms.

A new class of materials that have Li incorporated as a tetrahedral framework species has recently been discovered [2,3].

These materials, called microporous lithosilicates, are novel for two reasons. One is that [LIO,] tetrahedra appear more flexible than other [70] tetrahedra. This means that lithosilicates can bond in configurations that are too strained to exist for other silicates. Thus, lithosilicates offer the promise of new families of pore structures. Second, the general formula for these materials is M1[SiO2][LiO2],3, so that lithosilicates have the potential to be more negatively charged than aluminosilicates, if the Li:Si ratio exceeds 1:4.

3x

1-x

A team of NCNR scientists and collaborators has recently completed the first complete structural characterization of a microporous lithosilicate, RUB-29. To determine the framework geometry, synchrotron diffraction measurements were performed at the NSLS using an extraordinarily small single crystal - with dimensions 10 μm x 10 μm x 2 μum. (For comparison, human hair is typically 50 μm to 100 μm thick.) With 35 symmetry-unique atoms comprising the framework, RUB-29 is one of the most complex zeolitic structures. Powder neutron diffraction at NIST was then used to better determine the siting of the framework Li atoms, as well as four additional extra-framework Li atoms and seven other extraframework species. The RUB-29 framework is shown in Fig. 1.

The structural studies of RUB-29 demonstrate two novel structural building blocks, a Li,Si-spiro-3,5 and a Li,Si-spiro-5 unit (see Fig. 2). It should be noted that both these building units contain

Brian H. Toby

NIST Center for Neutron Research
National Institute of Standards and Technology
Gaithersburg, MD 20899-8562

"three-rings" where three T atoms, in this case one Li and two Si atoms, are bonded in a cyclic structure. This three-ring structural entity is highly strained in silicates; only one silicate example has ever been found.

In RUB-29, only 1 in 5 T atoms are Li, so the total framework charge is comparable to 1:1 aluminosilicates. However, RUB-29 appears to be stable under conditions where these high-aluminum zeolites tend to degrade. Further, there is promise that new lithosilicate materials can be synthesized with even higher Li:Si ratios.

Another interesting property exhibited by RUB-29 is that the Li atoms, both in framework and non-framework sites, appear to move on an NMR timescale at temperatures as low as 250 °C. Much more work is needed to learn about conduction in this material, but it may hold promise for ionic conduction applications, such as in batteries.

REFERENCES

[1] S. Borman, R. Dagani, R. L. Rawls, and P. S. Zurer, Chemical & Engineering News, January 12, 1998.

[2] S.-H. Park, P. Daniels, and H. Gies, Microporous Mater. 37, 129 (2000).
[3] S.-H. Park, et al. Journal of the American Chemical Society (2000), in press.

W

hile many methods can be used to determine the elemental composition of a material, diffraction is one of the few techniques that is also sensitive to the physical arrangement of atoms and molecules in the solid state. Even phases with the same chemical composition, such as graphite and diamond, have different diffraction patterns, dictated by the particular structure of each substance. The diffraction pattern from a mixture is the weighted sum of the patterns from each phase that is present, making it possible for quantitative phase composition to be determined. Prior to the application of Rietveld refinement techniques to the problem of quantitative phase analysis, these measurements required difficult calibrations and were often imprecise. However, Rietveld analysis involves the fitting of the entire diffraction pattern of each component phase based upon a structural model of the material and no standards or prior calibration is required. It should be noted though that even Rietveld analysis could not be used directly to quantify materials with unknown structures or amorphous phases, since such materials cannot be modeled crystallographically. An example where quantitative phase analysis explains the inevitable failure of ceramic thermal barrier coatings was presented in the 1998 NCNR Annual Report.

NIST participation in a round robin on determination of quantitative phase abundance, sponsored by the International Union of Crystallography Commission on Powder Diffraction, provided an excellent opportunity to demonstrate the high quality of data obtained using the 32-detector NCNR high-resolution powder diffractometer at BT-1. The NCNR implementation of the Rietveld technique for phase quantification was found to give excellent agreement with the nominal compositions. In addition, a new method for determining the amorphous phase content of a mixed-phase sample without sample adulteration was validated.

Data were collected using a Cu(311) monochromator (λ= 1.5402 Å) and 15' incident collimation, and were then processed in the usual procedure to obtain a pseudo-single detector data set. The phase fractions were determined using standard Rietveld refinement techniques, including full refinement of crystallographic and instrumental parameters, as implemented by the GSAS suite of programs. Results for a sample consisting of nearly equal mass fractions of Al2O3, CaF2, and ZnO, along with values reported by the other round-robin participants, are given in Table 1; the results obtained at NIST for all other samples are reported in Table 2 along with the nominal phase content. It can be seen from Table

1 that neutron data give significantly more accurate results than synchrotron or laboratory x-ray data, and that the results obtained at NIST are exceedingly good. The high accuracy of these results can be attributed to the intrinsic Gaussian line shape of the reactor neutron source, as well at to the lack of microabsorption and preferred orientation effects that frequently plague x-ray data but are normally negligible with neutron data. The data presented in Table 2 further confirm this conclusion, in that the results for sample 2 (preferred orientation), sample 3 (amorphous content) and sample 4 (microabsorption) all agree well with the nominal phase content.

The determination of amorphous content in a crystalline sample has traditionally involved integration of the area under the broad amorphous hump, giving the relative intensity compared with that of the Bragg scattering. However, in multiphase samples this technique is impractical. An alternative approach is to add a known quantity of a material as an internal intensity standard.

The unique properties of neutron diffraction suggested an alternative approach. Since absorption is negligible for most elements, and since the entire sample is irradiated in the neutron beam, a strategy based on absolute scattering intensities using an external standard was devised.

In the Rietveld technique, the mass of each crystalline phase, w1, is proportional to the product of the scale factor for that phase, S1, and the molecular weight of the unit cell contents (Z,M) where Z, is the number of formula units per unit cell and M, is the molecular weight. Thus for a crystalline multiphase sample of mass wc (wc = Σw) the relation we ΣSZM, is true. The proportionality constant can be determined using mass wst of a completely crystalline known standard under identical data collection conditions, so the relation w/wsd = (ΣS,Z,M,)/(Std Zsid Ms) can be used. If the sample mass, w's,

complete sample irradiation, and identical data collection parameters

were employed. In this way, data on any number of unknowns could be compared to a single standard sample since no changes were made to the experimental conditions. An additional benefit was that no adulteration or mixing of the samples was necessary. This technique was used with several single-phase samples in order to compare the crystallinity of potential standards as well as of the unknown samples. In fact, the round robin organizers subsequently sent samples of each of the unmixed phases to NIST for analysis. The results obtained for the round robin sample with amorphous

content gave excellent agreement with the mass fractions determined by weighing (see Table 2); the slightly higher amorphous content obtained using the neutron Rietveld technique is explained by the presence of a small amount (1 % to 2 %) of amorphous material in the component crystalline phases.

This external standard technique to determine amorphous content could also be used to determine the mass fraction of a crystalline phase with an unknown structure. For both applications, however, obtaining the best results depends upon obtaining the best diffraction data. The unique capabilities of the NCNR high-resolution diffractometer at BT-1 make this possible; these are summarized below.

A

ccurate determination of residual stresses by means of diffraction relies on the knowledge of the elastic constants that translate lattice strain into macro-stress. Because of the difference between the elastic behavior of the aggregate and that of a single crystallite, the straightforward relationship between strain and stress as mediated by single crystal or polycrystal elastic constants no longer holds. Instead, the relationship between lattice strain and macro-stress is mediated by diffraction elastic constants (DEC).

Very recently, we proposed a theory that allows a transparent calculation of DEC. This theory applies for almost the entire range of polycrystalline elasticity, including that for aggregates of arbitrary phase composition and arbitrary symmetry, both of the aggregate

and of the constituents [1]. Results show that for a particular crystallographic plane (hkl) and an arbitrary anisotropic material there are usually six independent DEC. These DEC depend on the orientation of the scattering vector, on the grain shape, and on the elastic constants both of the crystallites and of the aggregate. Figure 1 shows Young's modulus vs. the orientation parameter for different crystallographic planes (hkl) of two plasma-sprayed coatings with different types of anisotropy. Calculated DEC for comparison to measurement were not previously available.

Figure 1 also illustrates the difference between the anisotropy of the aggregate and that of the crystallites that comprise the aggregate. The slope of Ek vs. I depends mostly on the ratio

hkl