Structural deformation mechanisms of zeolites under pressure

Abstract The HP behavior of the natural zeolite yugawaralite and of the synthetic zeolite Na-A was studied by in situ synchrotron X-ray powder diffraction, using a non-penetrating P-transmitting medium. The unit-cell parameters of yugawaralite were refined up to the pressure of 10 GPa, at which reductions were found of about 7, 2.4, 7, 1.3, and 15% for a, b, c, β, and V, respectively. Contractions of 6.5 and 18.4% were found for a and V, respectively, for zeolite Na-A in the range 10-4 to 6.8 GPa. Diffraction patterns collected during decompression show that the effects induced by high pressure on both samples are almost completely reversible. These results are compared with those obtained under similar experimental conditions for other natural zeolites, with the aim of rationalizing the deformation mechanisms of these porous materials and comparing their flexibility under high-pressure and high-temperature conditions.

The XRPD experiments were performed using the ID09 beamline at ESRF (European Synchrotron Radiation Facility, Grenoble), at a fixed wavelength of 0.41782 Å, using a diamond anvil cell and silicon oil as non-penetrating pressuretransmitting medium (Gillet et al. 1996). The choice of non-penetrating pressure-transmitting medium used in the highpressure studies of porous materials is always a compromise between the molecule dimensions, which must be larger than those of the pores, and the degree of hydrostaticity. It is well known that these media can suffer from a lack of hydrostaticity at high pressure. The hydrostaticity of silicon oil can be considered satisfactory at least up to about 5 GPa, on the basis of the close agreement between the results obtained from single crystals of heulandite and bikitaite using glycerol (Comodi et al. 2001(Comodi et al. , 2003 and silicon oil on powders (Vezzalini et al. 2001;Ferro et al. 2002). Moreover, Le Bihan et al. (1996) studied several intermetallic uranium compounds under high pressure with different pressure-transmitting media (silicon oil, ethanol-methanol-water, and liquid argon) and observed a stress due to the solidification of silicon oil at about 9 GPa. However, at lower pressure no significant difference was observed with respect to the other pressure transmitting media. Above 5 GPa we cannot, however, rule out a contribution of deviatoric stress to the peak broadening of our powder diffraction spectra.
The pressures were measured by the ruby fluorescence method (Forman et al. 1972) on the non-linear hydrostatic pressure scale (Mao et al. 1986). The estimated error in the pressure values is 0.1 GPa. The experiments were performed from 10 -4 up to 11 and 6.8 GPa, for YUGA and ZEOA, respectively, with pressure increments of 0.5-1 GPa. Some other patterns were collected during decompression. A MAR345 detector (pixel dimension 100 mm) was used at a fixed distance of 446.72 mm from the samples; the exposure time was 34 s for each pressure. The samples were rocked by ±3 in w to reduce texture in the diffraction images. One-dimensional diffraction patterns were obtained in the range 0-22 2q (YUGA) by integrating the two dimensional images with the program FIT2D (Hammersley et al. 1996) and are reported in Figures 1 and 2 for YUGA and ZEOA, respectively.
The unit-cell refinements were carried out in the 2q range 2-20 up to 10 and 6.8 GPa for YUGA and ZEOA, respectively, by means of the GSAS program (Larson and Von Dreele 1994) using the Rietveld method for YUGA and the Le Bail method for ZEOA. The atomic coordinates of the structural models are from Kvick et al. (1986) and Gramlich and Meier (1971) for YUGA and ZEOA, respectively. The background curves were fitted by a Chebyshev polynomial with 22 and 20 coefficients for YUGA and ZEOA, respectively. The pseudo-Voight profile function proposed by Thomson et al. (1987) was used with refined Gaussian (GW) and Lorentzian (LX) terms for YUGA, while for ZEOA LY was also refined above 3.1 GPa. An asymmetry correction from Finger et al. (1994) was used for YUGA. The refined cell parameters as a function of pressure are reported in Tables 1 and 2, and in Figures  3 and 4, for YUGA and ZEOA, respectively. The isothermal bulk modulus of all the zeolites studied here were determined with the EOS-FIT program (Angel 2001) using a second-or third-order Birch-Murnaghan equation of state (Birch 1952).

RESULTS
Figures 1 and 2 show powder patterns for YUGA and ZEOA, respectively, as a function of pressure. For both samples, the peak intensities decrease and the peak profiles become broader with increasing pressure. These effects can be due to a number of factors: an increase in the long-range structural disorder, a reduction of the average crystallite size, or the presence of microstrain caused by deviatoric stress in the quasi-hydrostatic pressure-transmitting medium silicon oil (Yamanaka et al. 1997;Weidner et al. 1998;Fei and Wang 2000;Lee et al. 2002). Moreover, we did not observe the appearance of peaks not due to the YUGA and ZEOA space groups, thus ruling out any pressureinduced change of the original symmetry.
The images of all the X-ray powder diffraction spectra collected on the MAR345 image plate detector are characterized by the presence of numerous spots caused by the poor statistics of the diffraction data, resulting from the low mosaicity of the crystallites and the small volume hit by the beam. These factors, together with the low peak-to-background ratio, make it impossible to refine the crystal structure of the zeolites. Hence, only the unit-cell parameters were extracted from the powder patterns.
As reported above, the contraction of the a and c parameters is three times greater than that of b. The interpretation of this different compressibility can be based on the geometry and position of the extra-framework cation polyhedron. At room pressure the Ca atom is located on one side of the channels (Figs. 5a and 5b) and is asymmetrically bonded to four framework O atoms and to four water molecules (Kvick et al. 1986). In reality, the coordination of Ca is complicated by the presence of two water molecules located at two partially and alternatively occupied positions (not shown in Fig. 5). It is evident from the two projections along a and c that the compression of the framework along b is hindered by the presence of Ca-O frame bonds, which are mainly oriented in this direction. These bonds, pointing toward opposite channel walls, brace the channel apertures, whereas the mobile water molecules allow a greater compressibility along a and c. To assess this proposed mecha-   3.75 (3) 10 -3 2.1 (4) 10 -3 2.7 (1) 10 -3 b c 1.07 (7) 10 -2 9.1 (6) 10 -3 3.4 (1) 10 -3 b b 1.92 (1) 10 -3 1.1 (2) 10 -3 9 (1) 10 -4 nism, molecular dynamics computations, based on the experimentally determined cell parameters of YUGA under pressure, are in progress.

Zeolite Na-A
The structure of the synthetic hydrated zeolite Na-A (Na 12 Al 12 Si 12 O 48 ·27H 2 O, Z = 8, Fm3 c; Gramlich and Meier 1971) can be described in terms of cuboctahedral cages (b cages) that are connected by double four-membered rings to form larger a cages. The silica and alumina tetrahedra alternate in an ordered arrangement. The cation sites, refined by Pluth and Smith (1980) from dehydrated Na-A, are: Na1, near the center of six-rings; Na2 in the plane of the eight-membered rings; and Na3 in the large cavity opposite a four-membered ring. The compressibility of ZEOA was first studied by Hazen (1983) and Hazen and Finger (1984) up to 4 GPa by singlecrystal X-ray diffraction, using different penetrating and nonpenetrating pressure-transmitting media. They found that the compressibility was significantly higher when non-penetrat-a b FIGURE 3. Experimental unit-cell parameters of yugawaralite, normalized to room condition values vs. pressure. The data reported in (a) are relative to the complete investigated pressure range and those reported in (b) to the partial range10 -4 to 4 GPa. The errors associated with the cell parameters are smaller than the symbols used.  ing pressure-transmitting media were used. The bulk modulus obtained on data up to 2 GPa using glycerol and FC-75 was 21 GPa. More recent work on ZEOA, in different exchanged forms, has mainly focused on pressure-induced amorphization and its reversibility (Huang 1998;Secco and Huang 1999;Huang and Havenga 2001;Rutter et al. 2001). Particular attention was devoted to the study of pressure effects on ionic conduction in these phases (Rutter et al. 2000), since an anomalous increase in ionic conductivity was observed up to 1.8 GPa, as a result of the increased structural disorder. Upon decompression, conductivity maintained a value higher than that at room pressure.
We now discuss new high-pressure data for hydrated ZEOA collected in the pressure range 10 -4 to 6.8 GPa by means of in situ synchrotron XRPD. The powder patterns shown in Figure  2 indicate that no X-ray amorphization occurs up to about 7 GPa. However, a residual peak broadening and a low peak-tobackground ratio characterize the patterns collected after decompression. This partial reversibility is in agreement with the data of Secco and Huang (1999) and Rutter et al. (2000) and could be the origin of the enhanced ionic conductivity retained by the material after decompression. Table 2 and Figure 4 show the variation of cell parameters as a function of pressure. In the investigated pressure range, the reductions of a and V parameters (6.5 and 18.4%, respectively) are not linear. The linear regression line equation for the normalized P-V dependence is shown in Figure 4. The volume compression data, the isothermal bulk modulus K 0 , and its pressure derivative K 0 ' were fitted up to 6.8 GPa with a thirdorder Birch-Murnaghan equation of state. The refined values are V 0 = 14899(20) Å 3 , K 0 = 19.6(6) GPa, and K 0 ' = 5.1(3). If the data are fitted to a second-order equation, K 0 becomes 22.1(3) GPa. Both bulk modulus values are in close agreement with the K 0 = 21 GPa obtained by Hazen and Finger (1984).
The same authors, in their study of ZEOA using single crystals and different non-penetrating pressure-transmitting media, reported a volume discontinuity at about 1 GPa. Limiting our data to the pressure range 10 -4 to 2 GPa and using a linear fit, we also observed a very slightly decreasing slope for the volume P-dependence at a similar pressure value.
The data extracted from three XRPD patterns collected during the decompression of the sample are reported in Table 2 and Figure 4. The values corresponding to room pressure indicate that the original unit-cell parameters are completely recovered after decompression; however the reverse path shows a hysteresis effect with respect to the forward one.

COMPARATIVE DISCUSSION
Below we compare the high-pressure behavior of the zeolites studied in this paper with that of other natural zeolites studied under similar experimental conditions (heulandite: HEUL, Vezzalini et al. 2001;scolecite: SCOL, Ballone et al. 2002;bikitaite: BIKI, Ferro et al. 2002) and, in general, with the available literature data obtained with non-penetrating Ptransmitting media. The deformation mechanisms induced by HP and HT in these porous materials are also compared.

Compressibility
In Figure 6 we compare the pressure dependence of the lattice volumes of YUGA and ZEOA with that of the other zeo-FIGURE 6. Unit-cell volumes, normalized to room condition values, vs. pressure for YUGA and ZEOA (this work), heulandite (Vezzalini et al. 2001), scolecite (Ballone et al. 2002), and bikitaite . The bulk modulus was calculated up to 6.8, 3.5, 4.0, 5.7, and 5.0 GPa for ZEOA, heulandite, YUGA, bikitaite, and scolecite, respectively. Mineralogist 2003Mineralogist , 88, 1416Mineralogist -1422Mineralogist https://doi.org/10.2138Mineralogist /am-2003Mineralogist -1004 DISCUSSION lites studied under the same experimental conditions. We can single out three different trends in compressibility: with increasing pressure HEUL, YUGA, and ZEOA show a compressibility decrease, and SCOL an increase, while the compressibility of BIKI does not vary with pressure. The unusual behavior of SCOL, rarely occurring in silicate minerals, was, however, also observed for gillespite-type phases (Miletich 2000) and was interpreted by Ballone et al. (2002), on the basis of ab initio molecular dynamics calculations, as due to a phase transition involving structural modifications in the extra-framework water system. All the bulk modulus values available for zeolites compressed with non-penetrating media are plotted in Figure 7 against the so called framework density (FD) (Baerlocher et al. 2001), i.e., the number of tetrahedrally coordinated atoms per 1000 Å 3 . These K 0 cover a large variability range from the minimum for ZEOA (22.1 GPa) to the maximum for SCOL (61 GPa); however, in general, zeolites turn out to be easily collapsible under pressure, as expected for this type of very open structures. The triangles represent the samples studied by synchrotron XRPD, while the squares are K 0 values reported in the literature for zeolites studied by single-crystal X-ray diffraction. The close agreement between the results of the two techniques should be noted.

American
From Figure 7 no clear correlation can be singled out between FD and K 0 for all the topologies. This could be ascribed to the fact that the compressibility of these complex structures is certainly strongly influenced not only by the framework density, but also by the type, amount, and location of the extraframework cations and water molecules.

Comparison between the deformation mechanisms induced by high pressure or high temperature
As the high-temperature behavior of the zeolites discussed in this work is known in details, we can compare the deformation mechanisms of these different frameworks under high-pressure and high-temperature conditions. Contrary to the behavior observed under compression, dehydration processes in zeolites involve water loss and extra-framework cation rearrangements, and therefore we will focus our attention on the framework distortions.
The flexibility of zeolites was rationalized on the basis of the Rigid Unit Mode model (Dove et al. 1995;Hammonds et al. 1997Hammonds et al. , 1998, which described the zeolites as frameworks of corner-linked rigid tetrahedra, able to distort by modifying their internal T-O-T angles. Following Baur (1992aBaur ( ,b, 1995, zeolite frameworks can be defined as "flexible" or "inflexible" upon changes of the physico-chemical conditions. The flexible structures are classified as collapsible or non-collapsible; in collapsible frameworks the T-O-T hinges co-rotate, while in the noncollapsible ones the hinges anti-rotate, preventing cell volume changes. On the basis of the observed cell parameter variations, YUGA is collapsible to quite a similar extent under both HT (Alberti et al. 1996) and HP. In particular, the cell volume contraction of YUGA at HP (15%) is slightly higher than that at HT (12%). Moreover, the extent of contraction of each unitcell parameter under the two conditions is significantly different, suggesting different deformation mechanisms.
In contrast, ZEOA is not collapsible at high temperature (Pluth and Smith 1980), but is highly compressible under high pressure. A similar behavior was also observed for bikitaite Quartieri et al. 2002). The anomalously low collapsibility of ZEOA and bikitaite under high temperature was interpreted on the basis of the model proposed by Baur (1992a,b), i.e., anti-rotation of the T-O-T angles. Concerning bikitaite, the analysis of the computational results for its deformation mechanism at high pressure revealed that, contrary to that observed at high temperature, all T-O-T angles co-rotate and decrease as a function of pressure. As a consequence the volume decrease under these conditions is about one order of magnitude higher than that observed at high temperature.

ACKNOWLEDGMENTS
This paper is dedicated to Orazio Ferro, who died prematurely at age 33 years. He was deeply involved in this research and made a large contribution to the present paper.
The European Synchrotron Radiation Facility is kindly acknowledged for allocation of beam-time under proposal nos. CH-831 and CH-1019. This work was supported by the Italian MIUR (COFIN2001 "Zeolites, materials of interest for industry and environment: synthesis, crystal structure, stability, and applications") and CNR. The authors warmly thank M. Hanfland for his careful advice during high-pressure experiments and M. Welch and two anonymous referees for their useful suggestions, which improved the manuscript.