Volume 82, Issue 3-4

A lattice-dynamical treatment of displacive phase transitions leads naturally to the softmode model, in which the phase-transition mechanism involves a phonon frequency that falls to zero at the transition temperature. The basic ideas of this approach are reviewed in relation to displacive phase transitions in silicates. A simple free-energy model is used to demonstrate that Landau theory gives a good approximation to the free energy of the transition, provided that the entropy is primarily produced by the phonons rather than any configurational disorder. The ‘‘rigid unit mode’’ model provides a physical link between the theory and the chemical bonds in silicates and this allows us to understand the origin of the transition temperature and also validates the application of the soft-mode model. The model is also used to reappraise the nature of the structures of high-temperature phases. Several issues that remain open, such as the origin of first-order phase transitions and the thermodynamics of pressure-induced phase transitions, are discussed.

The hydrostatic compression of synthetic single crystals of diopside, CaMgSi 2 O 6 , and hedenbergite, CaFeSi 2 O 6 , was studied at 33 pressures up to 10 GPa by X-ray diffraction. In addition, intensity data for hedenbergite were collected at 12 pressures up to 10 GPa. For determination of the elasticity two crystals were loaded together in a diamond cell. The axial compressibilities β a , β b , and β c of diopside and hedenbergite are 2.36(4), 3.17(4), and 2.50(4) × 10 -3 GPa -1 , and 1.93(5), 3.38(6), and 2.42(8) × 10 -3 GPa -1 , respectively. The bulk moduli (K T₀ ) and their pressure derivatives (K′ T₀ ) were determined simultaneously from a weighted linear fit of a third order Birch-Murnaghan equation of state to the volume data at elevated pressures. K T₀ and K′ T₀ are 104.1(9) GPa and 6.2(3) for diopside and 117(1) GPa and 4.3(4) for hedenbergite, respectively. The unit-cell parameters decrease continuously with pressure. The larger polyhedra show more compression than the smaller ones. Between 0.1 MPa and 10 GPa the polyhedral volumes of CaO 8 , FeO 6 , and SiO 4 decrease by 8.4, 6.6, and 2.9%, respectively. The longest bonds of CaO 8 and FeO 6 show most compression. Significant compression in the two shortest Si-O1 and Si-O2 bond lengths of the SiO 4 tetrahedra was observed at relatively low pressures, resulting in a tetrahedral volume compression of 1.6% between 0.1 GPa and 4 GPa and 1.3% between 4 and 10 GPa. The compression of the unit cell can be described by the volume compression of the individual CaO 8 and FeO 6 polyhedra, with the SiO 4 tetrahedron playing a minor role. Diopside is more compressible than hedenbergite as shown by their axial and volume compressibilities because the FeO 6 octahedron is significantly more rigid than MgO 6 at high pressures. This observation implies that octahedrally coordinated Fe 2+ behaves differently from Mg at high pressures, in contrast to their behavior at ambient conditions.

Clinopyroxene (Cpx) is a principal upper-mantle phase for concentrating large cations but has not been viewed as a major crystal-chemical reservoir for K because K + is considered too large to enter the largest site, M2, in the pyroxene structure. Accumulating data from high-pressure conditions indicate this inference is incorrect, so multianvil experiments have been performed to evaluate maximal K solubility in Cpx at high pressure. End members and mixtures of diopside, jadeite, and kosmochlor have been mixed with K 2 CO 3 , KHCO 3 or both in welded platinum capsules and heated typically for 24 h in the range of 5 to 14 GPa and 1200 to 1700 °C. These experiments produced K-rich Cpx in solid solutions by means of a fictive Kcpx component (KCrSi 2 O 6 or KAlSi 2 O 6 ). The maximum K 2 O content obtained is 4.7 wt% in a Cpx (Di 38 Ko 39 Kcpx 22 En 1 ) formed from a 50:50 Di 1 Ko mixture at 10 GPa, 1400 °C. K uptake and partitioning is dependent positively on P, complexly on Cpx composition but not demonstrably on T. Cpx/liq D K₂O is in the range of 0.03-0.1 and Cpx/liq D Na₂O varies from 0.5 to 5, although the variations for each with Cpx composition are different. In diopside, Kcpx increases are always accompanied by increases in Nacpx, and cooperative Nacpx solution is necessary for Kcpx solution in the compositional systems examined. K appears to be accommodated in the M2 site of the Cpx structure by two types of spatial averaging: a large average M2 site, as in the case of Di, ameliorates the fit, but local accommodation by size averaging with a smaller M2 occupant, presumably Na, appears necessary, suggesting that the polyhedral compressibility of Na and K are large in comparison with Ca. In application to Cpx inclusions in diamond, the data here imply that a chromium diopside with ∼1 wt% K 2 O forms in the presence of a C-rich melt with 15-28 wt% K 2 O.

Wadsleyite [b-(Mg,Fe) 2 SiO 4 ] is known to accept variable amounts of H and may be an important reservoir of H in the transition zone of the mantle. The crystal structure of Fo 94.6 hydrous wadsleyite (Mg 1.730 Fe 0.098 Al 0.008 Si 0.991 H 0.355 O 4 ) synthesized at 1400 °C and 17 GPa was refined in space group I2/m from 1784 measured intensities of which 830 were unique, and, of these, 650 were of intensity greater than 4s. Unit cell parameters are a 5 5.6715(7), b 5 11.582(2), c 5 8.258(1) Å , and β = 90.397(9)°. The final R(F) for all reflections was 0.026 (R w = 0.024); goodness of fit was 1.68. An H position was located on the nonsilicate O atom (O1). Partial occupancy of a normally vacant tetrahedral site adjacent to M3 was observed. This is postulated to be the result of Si moving from M3 to the adjacent tetrahedral void on decompression. Deviation from orthorhombic symmetry appears to result from ordering of H, (Mg, Fe), and possibly Si, within the two nonequivalent M3 sites.

A new KAlSiO 4 polymorph was found in a granulite facies gneiss from the Punalur district, southern India. The structure was solved and refined on a twinned crystal to an R index value of 1.98% for 265 independent reflections. Metamorphic kalsilite is trigonal, space group P31c with a 5.157 (1) Å , c = 8.706 (3) Å , V = 200.52 (9) Å 3 , Z = 2. The overall diffraction symmetry 6/mmm exhibited from all the crystals examined arises from a {0001} twinning, related to a mistake in the ordered Al-Si-Al-Si sequence along the c axis. The crystal structure is a stuffed derivate of tridymite, and is characterized by sixmembered tetrahedral rings with ditrigonal shape. Individual layers of this structure are the same as those of P6 3 kalsilite, but are stacked along the c axis in an eclipsed manner rather than in the staggered manner of the P6 3 structure.

A series of aluminous tremolites was synthesized at 800-850 °C and 6-13 kbar along the join Ca 1.8 Mg 5.2 Si 8 O 22 (OH) 2 -Ca 1.8 (Mg 4.2 Al)(AlSi 7 )O 22 (OH) 2 by means of the Mg-Tschermaks substitution. Although the amphibole yields were high (97-99 wt%), traces of orthopyroxene, clinopyroxene, amorphous silica, and, for the most aluminous mixture, corundum, were identified in some synthesis products. The samples were characterized by electron microprobe (EMP), X-ray diffraction Rietveld structure refinement (XRD), Fourier- transform infrared (FTIR) spectroscopy, and magic-angle-spinning nuclear magnetic resonance (MAS NMR) spectroscopy for the purpose of identifying the extent of Al, Mg, and Si order-disorder in the octahedral and tetrahedral sites. Both EMP and 27 Al NMR verify that the Mg-Tschermaks substitution, with even distribution of Al between tetrahedral and octahedral sites, is obeyed in these amphiboles up to a limit of 1.9 total Al atoms per formula unit (apfu). 29 Si MAS NMR spectra indicate that Al substitutes essentially equally into both the T1 and T2 sites. For octahedral sites, the FTIR spectra indicate the presence of Al on the M1, M3, or both sites (presently indistinguishable) by the presence of an (MgMgAl)-OH band at 3652 cm -1 , which increases steadily with increasing Al content. The FTIR and 27 Al MAS NMR spectra show that Al is always present in the M1-M3 sites as well as the M2 site but partitioned among these sites in ways that are not clearly resolved at present. It is clear in this study that Al is more widely distributed over the octahedral and tetrahedral sites in synthetic aluminous tremolite with only a moderate Al content (< 2.0 Al apfu) than has been traditionally thought.

Amphiboles were synthesized at 750 °C, 1 kbar (H 2 O) on the binary joins (nickel, magnesium)-richterite and (magnesium, cobalt)-richterite. Structural variations and site occupancies were characterized by Rietveld structure refinement, with final R Bragg indices in the range 4-9%, and by powder infrared spectroscopy in the principal OH-stretching region. Site-occupancy refinement of Ni-Mg and Mg-Co distributions give the partition coefficients over M1,3 and M2 where K M2+ = (M 2+ /Mg) M1,3 /(M 2+ /Mg) M2 , and M 2+ = Ni 2+ or Co 2+ , K d Ni = 2.98 ± 0.37 and K d Co = 1.34 ± 0.31. Both K d values are greater than 1.0, whereas [6] r(Ni 2+ ) < [6] r(Mg) < [6] r(Co 2+ ); this indicates that cation size is not the primary factor affecting the ordering of Ni-Mg and Mg-Co over the octahedral sites. The infrared spectra of intermediate binary compositions show fine structure caused by ordering of Ni- Mg or Mg-Co over the M1,3 sites and by ordering of Na and ⃞ (vacancy) at the A site; thus intermediate compositions show an eight-band spectrum in the principal OH stretching region. Precise band intensities were derived by nonlinear least-squares fitting of Gaussian band shapes to the observed spectra. The relative observed intensities of the combinations of bands 3 I₀ A + 2 I₀ B + I₀ C and I₀ B + 2 I₀ C + 3 I₀ D are in accord with the equations of Burns and Strens (1966), indicating that there is no significant variation in molar absorptivity with frequency (energy) for individual bands within a single sample (spectrum). Combined with the results of Skogby and Rossman (1991) on polarized single-crystal infrared spectra of amphiboles, this result suggests that different local configurations of M1,3 cations in amphiboles couple such that the transition probabilities of the associated OH groups are equal.

Synthetic mackinawite (tetragonal FeS) has been found to transform rapidly to greigite (Fe 3 S 4 ) above ∼373 K during heating experiments, as observed by in situ X-ray diffraction. Using monochromatic synchrotron radiation (λ = 0.60233 Å), we measured the unit-cell parameters of both synthetic mackinawite between 293 and 453 K and of greigite formed from this mackinawite between 293 and 593 K. The coefficients of thermal expansion for mackinawite are α 1 = α 2 = (1.36 ± 6 0.11) × 10 -5 , α 3 = (2.98 ± 0.12) × 10 -5 , and α vol = (5.67 ± 0.19) × 10 -5 between 293 and 453 K. The coefficients of thermal expansion for greigite are α 1 = α 2 = α 3 = (1.63 ± 0.15) × 10 -5 , and α vol = (4.86 ± 0.25) x 10 -5 between 293 and 593 K. On further heating in situ, we observed the reaction greigite → pyrrhotite + magnetite. Partial transformation of mackinawite to greigite was also observed using transmission electron microscopy (TEM) following in situ heating. Electron diffraction patterns show that (001) of mackinawite is parallel to (001) of greigite, and [110] of mackinawite is parallel to [100] of greigite. This orientation relationship confirms that the cubic closepacked S array in mackinawite is retained in greigite and implies that oxidation of some Fe 2+ in mackinawite drives rearrangement of Fe to form the new phase. Small regions of the crystallites show Moiré fringes resulting from the lattice mismatch between mackinawite and greigite. Electron diffraction patterns of mackinawite subjected to prolonged exposure to the atmosphere also show faint spots corresponding to greigite. We propose that in these experiments surplus Fe is accommodated by reaction with either adsorbed O 2 or H 2 O to form amorphous nanophase Fe-O(H). Because greigite is so easily formed by oxidation from mackinawite, greigite should be an important precursor for pyrite nucleation, although any orientation relationship between greigite and pyrite remains to be determined.

Cathodoluminescence (CL) emission spectra of forsterite from the Allende (CV3) and Murchison (C2) meteorites and from Cr-doped and pure synthetic forsterite samples have been obtained at room (296 K) and liquid nitrogen (77 K) temperatures using a transmission electron microscope. At room temperature, three broad CL emissions centered near 420, 640, and 800 nm occur in the meteoritic forsterite, and one peak at 800 nm occurs in one Cr 3+ -doped forsterite sample. Only the 420 nm peak is present in pure synthetic forsterite. Relative to room temperature, at liquid-nitrogen temperature there is a general increase in overall CL intensity, whereas a broad peak located between 700 and 800 nm changes to a series of sharp emissions centered near 700 nm and a narrower but still broad peak centered near 800 nm. The portion of the broad peak near 700 nm at room temperature and equivalent sharp peaks at low temperature are assigned to Cr 3+ in octahedral coordination, whereas the broad peak near 800 nm is attributed to Cr associated with structural defects. The peak at 640 nm is consistent with a similar peak in Mn 2+ -doped forsterite. The peak at 420 nm results from an unknown structural effect that can be eliminated in synthetic forsterite by mechanical deformation. The variation of relative intensities at room temperature for the two broad peaks at 700 and 800 nm in meteoritic forsterite can be correlated with meteorite type and may reflect different processes in forsterite formation or subsequent processing.

Amphibole from alkali granites in the Ramnes cauldron in the Oslo rift was altered hydrothermally by corrosion and growth through multiple events of fluid circulation. The alteration developed successive zones of amphibole at the crystal margins, resulting in a (1) brownish ferro-edenitic hornblende core (FE), (2) deep bluish-green hastingsitic hornblende zone (HH), and (3) light-greenish Fe-rich actinolite rim (FA). The edenitic core preserves the original igneous amphibole composition. The Cl content of amphibole strongly increases from the FE (0.78-0.82 wt%) to the HH zone (2.07-2.96 wt%) and abruptly decreases in the FA rim (0.01-0.36 wt%). In the Cl-rich HH amphibole zone, amphibole has characteristically high Cl content [2.96 wt%, 0.82 atoms per formula unit (apfu)] and high concentrations in both [4] Al (1.73 apfu) and A-site occupancy (0.86 apfu) with a large K/(Na + K) value of 0.47. Both [4] Al and A-site occupancy increase systematically with positive correlation with Cl content throughout the three amphibole zones. On the other hand, Fe 2+ content is not so simply correlated to the Cl content. Based on crystal structure considerations on Cl-rich amphiboles, the cation substitutions are illustrated by structural (geometrical) constraints for [4] Al and by a chemical constraint for Fe 2+ . These contributions for Cl incorporation are expressed empirically by ln(Cl/OH) amp = ln(Cl/OH) fluid + A· [4] Al·Fe 2+ /RT + B/RT, where A and B are constant. [4] Al·Fe 2+ vs. ln(Cl/OH) plots of the three distinct amphibole zones suggest different fluid conditions in chemistry and temperature for the three zones. The zoning was developed through two stages of hydrothermal alteration. In the early hydrothermal event, a saline and high-temperature fluid altered the original hornblende (FE) to the Cl-rich HH zone. Late stage alteration by a high Fe/Cl and relatively low-temperature fluid partially over-printed the FA zone at the crystal margin.

Many crystalline solids have multiple nonequivalent sites among which different atoms show substitutional long-range order-disorder phenomena. The order-disorder kinetics of an atom among any n nonequivalent sites in a crystal can be described by the equation where x i is the site occupancy of the atom at site s i , n is the number of nonequivalent sites, λ j (λ 1 = 0) is constant at a given temperature, pressure, and total composition of the crystal, and c ij (t) is constant or polynomial in t. Four theorems governing a multi-site order-disorder process have been proved, requiring that λ j must be either zero (only λ 1 = 0), a negative real number, or a complex-valued quantity with the real part being a nonpositive number. The kinetic model becomes constrained and naturally complies with crystal-chemical conditions when the mole number per formula unit is chosen as the unit of all site-occupancy variables, or site multiplicities are explicitly incorporated into the model. When the mole fraction is directly used as the unit, the model becomes unconstrained, but it is a valid treatment that is as equally applicable to the multi-site order-disorder kinetics as the constrained model.

Piston-cylinder experiments with C-H-O fluids in equilibrium with graphite as redox sensor were performed at 900 °C and 10 kbar. With this technique, it is possible to investigate the hydrogen permeability of the material surrounding the samples and to determine the hydrogen fugacity. The furnace assembly consisted of boron nitride or unfired pyrophyllite. Solid organic compounds (C 4 H 4 O 4 , C 9 H 10 O 2 , and C 14 H 22 O) were used as starting materials. The different H/O ratios of these compounds (1:1, 5:1, 22:1) result in different initial hydrogen fugacities. The fluid formed during the experiments was analyzed by gas chromatography. The experimental data demonstrate that boron nitride is nearly impermeable to hydrogen at the investigated conditions. The hydrogen fugacity observed in the gold capsules is controlled by the composition of the C-H-O starting compound. After approximately 1 h it is 340 bar for experiments with C 4 H 4 O 4 , 2400 bar for experiments with C 9 H 10 O 2 , and 3200 bar for experiments with C 14 H 22 O and remains constant for at least 8 d. The situation is different with unfired pyrophyllite as the pressure-transmitting material. In these experiments, the same hydrogen fugacity was determined irrespective of the compositions of the organic compounds. The hydrogen fugacity seems to be controlled only by the furnace assembly. It is approximately 470 bar after 3 d. The hydrogen fugacity in two samples placed side-by-side in the boron nitride assembly and containing different starting compounds (H/O = 1 and 5) becomes identical during experimental durations of less than 2 d. This is because of the exchange of hydrogen through the adjacent capsule walls. This observation can be used to measure the hydrogen fugacity of any piston-cylinder experiment. In boron nitride assemblies, a second capsule with a C-H-O fluid and graphite can also be applied to control the hydrogen fugacity within given limits.

Reversed phase-equilibrium data, collected over the P-T range 12-20 kbar at 850-1100 °C, define the solubility of Al 2 O 3 in ferrosilite (Fs) in equilibrium with almandine garnet (Alm). The new data indicate significantly lower Al 2 O 3 solubility in Opx in the Fe-bearing system compared with the Mg-system and extrapolate well to merge with the higher pressure bracket of Kawasaki and Matsui (1983). Orthopyroxene-garnet thermometry in crustal rocks, based on the equilibrium studied here (Alm = 3 Fs + Al 2 O 3 ), is considerably more robust than previous calibrations based on the equivalent equilibrium in the Mg system. Results for several granulite terrains show that Al-Opx temperatures based on the new experimental data are generally higher than results based on Fe-Mg exchange thermometry, consistent with suggestions of previous workers. For many samples, the difference in apparent closure temperature between these equilibria (generally 50-130 °C) is within the combined uncertainty of their calibration (~75 °C) and is not as extreme as differences calculated on the basis of Harley’s (1984) unreversed experimental data (Bégin and Pattison 1994). The lack of sensitivity of this new thermometer to late Fe-Mg exchange makes it a powerful tool for deciphering near-peak P-T conditions for high-grade rocks.

The results of multi-anvil melting experiments are reported for a wide range of komatiite analog mixes with compositions in the system CaO-MgO-FeO-Fe 2 O 3 ± Fe 0 -Al 2 O 3 -SiO 2 at 5 GPa (CMAS and CMFAS ± Fe 0 ). The liquidus crystallization fields for olivine, orthopyroxene, clinopyroxene, and garnet were mapped out, as were their intersections at various cotectics and invariant points. For the assemblage L + Ol + Opx + Cpx + Gt, the compositions of liquids at the invariant point in CMAS and at several pseudoinvariant points in CMFAS were determined to within ± 0.50 wt% (1σ). The effect of FeO is to expand the liquidus crystallization fields in the following relative ways: garnet at the expense of all other crystallizing phases, pyroxenes at the expense of olivine, and clinopyroxene at the expense of orthopyroxene. The results reported here are in excellent agreement with previous determinations (Fujii et al. 1989; Herzberg 1992; Trønnes et al. 1992), but the uncertainties are much lower. It is demonstrated that the multi-anvil apparatus is capable of yielding crystal-liquid phase-equilibrium information at 5 GPa with an accuracy that is comparable to those at lower pressures.

In order to model quantitatively exsolution of volatiles over the range of basaltic melt compositions found on oceanic islands, I present compositional parameterizations of H 2 O and CO 2 solubilities and use these parameterizations to develop vapor saturation and degassing models for alkalic basaltic liquids. Vapor-saturation diagrams generated as a function of melt composition are used to determine the pressure at which the melt was last in equilibrium with a vapor and the composition of the vapor phase based on measured H 2 O and CO 2 contents in basaltic glasses. These models allow the calculation of the pressure at which a magma of known initial volatile content reaches vapor saturation and begins to exsolve a vapor phase. The higher solubility of CO 2 in alkalic magmas causes vapor saturation in CO 2 -bearing alkalic magmas to be reached at lower pressures than in CO 2 - bearing tholeiitic magmas having identical volatile contents. However, if variations in major element and volatile concentrations were linked by variations in the extent of melting, then volatile-rich, strongly alkalic magmas would begin to exsolve a vapor at slightly higher pressures than volatile-poor alkali olivine basalts or tholeiites. Partitioning of H 2 O and CO 2 into the vapor during volatile exsolution is controlled by the difference between H 2 O and CO 2 solubilities. As melts become more alkalic, the relative difference between H 2 O and CO 2 solubilities decreases, thus diminishing the preferential partitioning of CO 2 into the vapor. Exsolution of volatiles from tholeiites is characterized by strong partitioning of CO 2 into the vapor such that most or all CO 2 is lost before any significant loss of H 2 O. In contrast, the combination of higher CO 2 solubility and higher volatile contents (and perhaps higher CO 2 /H 2 O ratio) in alkalic melts results in less fractionation between CO 2 and H 2 O during volatile exsolution.

A variety of smectitic and illitic clays were studied by TEM and AEM, following expansion by L.R. White resin, to define phase relations for clay minerals undergoing digenesis and low-grade metamorphism. Samples included a prograde shale sequence from the Gulf Coast, a hydrothermal bentonite from Zempleni, Hungary, and shales from the Nankai Trough, Japan, Michigan, and the Welsh sedimentary basin. All samples were dominated by various proportions of only three kinds of clay minerals: smectite having no 10 Å layers, R1 illite-smectite (I/S) (50% illite), and illite with only small proportions of smectite-like interlayers; mixed-layer phases with intermediate ratios of I/S were observed only as minor components of other layer sequences. Lattice fringe images show that the common layer spacing of R1 I/S is 21 Å and that it has 0.7-0.8 K pfu; its properties are not an average of those of smectite and illite, which is consistent with the uniqueness of the R1 I/S structure. A prograde sequence of clay mineral transitions in the studied samples can be characterized by five stages with different combinations of the three major phases (i.e., smectite, R1 I/S, and illite) corresponding to different grades. The sequence from low to high grade is (1) pure smectite, (2) smectite with small proportions of discrete R1 I/S and illite, (3) R1 I/S with small proportions of smectite and illite, (4) illite with some R1 I/S and smectite, and (5) illite. Common exceptions to this scheme are inferred to be caused by the inherent metastability of all phases, in occurrences that are rate or path dependent.

Mineralogically zoned, decimeter-diameter calc-silicate boudins enclosed within paragneisses from Little Italy Island, Rauer Group, east Antarctica underwent granulite-facies metamorphism (P = 700-900 MPa, T = 840 ± 40 °C) followed by near-isothermal decompression to 200-400 MPa. During decompression several mineral reactions occurred in the presence of a pore fluid (X CO₂ ≈ 0.4). The calc-silicate boudins show a general increase in calcite δ 13 C values from core (as low as - 17‰) to rim (-9 to -10‰); by contrast, δ 18 O values show little variation across the boudins. The δ 13 C profiles are similar to those predicted to result from diffusion within a sphere that is surrounded by a homogeneous reservoir. Diffusion of carbon isotopes probably occurred synchronous with the post-peak metamorphic mineral reactions. At that time, centimeter-scale diffusion would have been facilitated by the presence of the fluid, while isotopic exchange between the minerals and the fluid would have been promoted by recrystallization. For metamorphic porosities of 10 -5 - 10 -3 , the δ 13 C profiles could have formed in a few thousand to hundreds of thousands of years. Small (millimeter to centimeter) scale variations in δ 13 C values that may have been initially present within the boudins would have been homogenized on much shorter timescales than those required to form the profiles. The calculated timescales may reflect the time over which metamorphic recrystallization occurred and a reactionenhanced porosity was present. Graphite formed locally in the paragneisses only at margins of the boudins probably reflects the local escape of CO 2 -bearing fluids from the boudins into relatively low f O₂ rocks.

Tsaregorodtsevite [N(CH 3 ) 4 ][Si 2 (Si 0.5 Al 0.5 )O 6 ] 2 is a unique feldspathoid with a tetramethylammonium (TMA + ) organic cation in an ordered, sodalite-like framework of orthorhombic symmetry (I222) with a = 8.984(3), b = 8.937(2), and c = 8.927(2) Å . Here, 13 C cross polarization MAS and 1 H MAS NMR spectra give direct evidence that the organic material in the cavities in tsaregorodtsevite is TMA + . The 27 Al and 29 Si MAS NMR spectra, and XRD data show that the framework is well ordered with Si in the T1 and T2 sites and both Si and Al in the T3 site. Upon heating, the colorless tsaregorodtsevite crystals change color becoming yellow, then brown, and finally black. In this study 1 H, 13 C, 27 Al, and 29 Si MAS NMR, powder XRD, and electron microprobe analyses are used to investigate the structural changes in tsaregorodtsevite on heating. The cell dimensions and T-O bond lengths decrease, and there are two phase changes. At 690 °C, the TMA + molecule breaks down to form aromatic rings and acidic groups, possibly including benzene and the pyridinium ion, with ammonia and other gases produced. The Al-Si framework changes slightly to give a less well-ordered tetragonal structure (I422), with a = 8.925(1), c = 8.908(1) Å , and a small amount of Al enters a separate phase. After heating at 970 °C, the organic material has further broken down with little C and H visible by NMR spectroscopy. The Al-Si framework forms at least two new phases: one with a cubic crystalline structure (I432) with a = 8.817(3) Å , and an amorphous aluminosilicate phase. There is a small residue of the initial structure even after heating for one hour at over 900 °C.

Surite has been reported to be a mineral that intercalates cerussite-like materials in a smectite interlayer. However, the chemical analysis and infrared absorption spectra indicate that the interlayer material has the composition of Ca 0.5 OH·2PbCO 3 . A one-dimensional Fourier synthesis of electron density for acid-treated surite shows that the 2:1 layer of surite has the structure of a general smectite (the distance between the Si plane and the apical O atom plane is about 1.6 Å ). Moreover, the one-dimensional Fourier synthesis method showed that the crystal structure of surite is a 2:1 dioctahedral smectite interlayered with basic lead calcium carbonate.

A new high-silica zeolite, terranovaite, was recently found in cavities of Ferrar dolerites at Mt. Adamson (Northern Victoria Land, Antarctica). The mineral [(Na 4.2 K 0.2 Mg 0.2 Ca 3.7 ) Σ8.3 (Al 12.3 Si 67.7 ) Σ80.0 O 160 · > 29 H 2 O] occurs as globular masses that flake off in transparent lamellae; it has a vitreous luster, white streak, {010} perfect cleavage, and {001} distinct parting. The observed density is 2.13 ± 0.02 g/cm 3 . Optically, it is biaxial positive, with 2V = 65°, α = 1.476, β = 1.478, γ = 1.483 (all ± 0.002). The orientation is X = c, Y = a, and Z = b. Terranovaite is orthorhombic with a = 9.747(1), b = 23.880(2), c = 20.068(2) Å and topological symmetry Cmcm. The strongest powder X-ray diffraction lines are (d (Å), I, hkl ): 11.94,40,020; 10.16,65,021,002; 9.04,33,110; 3.79,100,025,240; 3.61,40,153. Terranovaite topology, hitherto unknown in either natural or synthetic zeolites, is characterized by the presence of pentasil chains and of a twodimensional ten-membered ring channel system. The mineral was named terranovaite after the Italian Antarctic Station at Terranova Bay, Antarctica.