Volume 81, Issue 9-10

Order-disorder phenomena, which play a key role in the energetics, crystal chemistry, and physical properties of numerous solids, may be strongly affected by pressure. The effect of pressure on the state of cation order depends fundamentally on the volume of disordering, Δ V dis = V disordered - V ordered . Although ΔV dis may be positive or negative, it is usually positive for cation ordering in rock-forming minerals; pressure, therefore, tends to increase order in geological systems. Three structural mechanisms that contribute to ΔV dis are (1) positive ΔV dis (typically ≤ 1%) associated with ordering of similar cations over two or more similar sites, (2) variable ΔV dis (-1 % ≤ ΔV dis ≤ 1%) associated with Al-Si tetrahedral ordering in framework minerals, and (3) large ΔV dis (up to 5%) associated with cation ordering involving mixed valence, mixed coordination, or both. Pressure alters order-disorder reaction rates by restricting cation motions or changing cation diffusion mechanisms. Influences of pressure on ordering in olivines, orthopyroxenes, high-pressure ferromagnesian silicates, spinels, garnets, perovskites, and rhombohedral carbonates are discussed. Several experiments are proposed to elucidate the phenomenon of high-pressure ordering in systems of mineralogical interest.

X-ray and ultraviolet photoelectron spectroscopy (XPS and UPS, respectively) and scanning tunneling microscopy (STM) were used to observe the initial oxidation of pyrite surfaces in air. The results show the growth of oxide-like oxidation products, with minor contributions from sulfate. UPS shows a decrease in the density of electronic states in the uppermost valence band of pyrite, corresponding to oxidation of surface Fe 2+ . This allows reliable interpretation of STM images, which show that initial surface oxidation of FeH proceeds by growth of oxidized patches. The borders of oxidized patches contain small segments oriented in the (110) and (100) directions. STM of as-received pyrite cube surfaces, oxidized in air for years, also show the importance of the (110) crystallographic directions, on the surface, in controlling reaction progress. A model in which oxidation probabilities for FeH surface sites are proportional to the number of nearest-neighbor oxidized (FeH) sites was tested using a Monte Carlo approach and reproduces the surface patterns observed in STM. An oxidation mechanism consistent with the XPS, UPS, STM, and Monte Carlo results is proposed. The rate constant for electron transfer from surface-exposed pyrite FeH to O 2 is small. Electron transfer is more rapid from pyrite FeH to FeH present on the surface as an oxidation product, such as in the patches we observed. FeH in oxide is a better reductant than FeH in pyrite, so electron transfer to O 2 from the oxide is also fast. However, this two-step mechanism is faster overall only if electron transfer to the surface oxide patches is irreversible (e.g., because of S 2 oxidation or electron hopping within the surface oxide patches). Cycling of Fe between the FeH and FeH forms, particularly along borders between oxidized and unoxidized areas, is thus a key feature of the pyrite oxidation mechanism. An understanding of the surface electronic and band structure aids definition of the redox potentials of electrons in various surface states. Rates of electron transfer from these states to O 2 are estimated using a kinetic theory of elementary heterogeneous electron transfer.

The rigid-unit mode model provides many new insights into the stability and physical properties of framework silicates. In this model the SiO 4 and AlO 4 tetrahedra are treated as very stiff, to a first approximation as completely rigid, in comparison with intertetrahedral forces. In this paper we apply the model to several important examples. The model is reviewed by a detailed study of quartz, and it is shown that the α-β phase transition involves a rigid-unit mode that preserves the Si-O-Si bond angle. The model is used to explain the phase transitions in cristobalite and the different feldspar, sodalite, and leucite structures. We also use the model to explain the nature of the high-temperature disordered phases of cristobalite and tridymite, to interpret the observations of streaks of diffuse scattering in electron diffraction patterns, to interpret the structures in the kalsilite-nepheline solid solution, to explain volume anomalies in the cubic leucite structures, and to explain qualitatively the negative linear thermal expansion in cordierite. The results for the highest symmetry sodalite structure show that there is a rigid-unit mode at every wave vector, a finding with significant implications for the understanding of the sorption and catalytic behavior of zeolites.

Lawsonite single crystals were investigated by polarized FTIR spectroscopy at wave-numbers between 8000 and 1000 cm -1 and temperatures between 82 and 325 K. This temperature range contains three lawsonite phases-Cmcm > 273 K, 273 K > Pmcn > 150 K, P2 1 cn < 150 K-which are characterized by different rotations of hydroxyl groups and H 2 O molecules. Unlike previous studies of H 2 O in minerals, which assumed weakly bonded, symmetric H 2 O molecules, the highly asymmetric H 2 O molecule in lawsonite required a modified approach that uses the single, uncoupled O-H stretching frequencies and orientations of the individual OH groups in the H 2 O molecule. The formation of a strong hydrogen-bond system with decreasing temperature is characterized by a shift of O-H stretching bands from 2968 and 3252 cm -1 at 325 K to 2817 and 3175 cm -1 at 82 K. These frequencies are in good agreement with the corresponding hydrogen-bond lengths (H···O = 1.66 and 1.74 Å, O-H···O = 2.60 and 2.66 A) at low temperatures. The orientations of the O-H vectors determined from polarized IR measurements also confirm the H-atom positions refined from previous X-ray structure determinations at low temperatures. However, the disagreement between spectroscopically determined distances (and orientations) and those from X-ray refinements at ambient conditions indicates that the room-temperature Cmcm structure of lawsonite contains dynamically disordered hydroxyl groups and H 2 O molecules. The smooth changes of stretching and bending frequencies across the phase boundaries at 273 and 150 K also suggest that the lawsonite phase transitions are of a dynamic order-disorder type rather than a displacive type. Deuteration experiments on differently oriented, single-crystal lawsonite slabs at 350°C and 1.2-2.5 kbar showed that lawsonite has a preferred H-diffusion direction parallel to [001]. This is consistent with the crystal structure showing channels parallel to [001], which are solely occupied by H atoms. The spectra of isotopically diluted samples, which are almost identical to those of natural lawsonite, indicate that band-coupling effects are generally weak. The FTIR powder spectra of the lawsonite-type mineral hennomartinite, SrMn 2 [Si 2 O 7 ](OH) 2 ·H 2 O, are similar to the lawsonite Z spectra and confirm the existence of both strong and weak hydrogen bonds in its structure.

High-pressure and high-temperature Raman spectra of ilmenite-type MgSiO 3 have been collected up to 7 GPa and 1030 K, respectively. The observed Raman frequency shifts with pressure and temperature were used to calculate the isothermal, isobaric, and isochoric (or intrinsic) anharmonic mode parameters. The low values of the intrinsic mode parameters indicate that ilmenite-type MgSiO 3 has a nearly quasi-harmonic behavior, and intrinsic anharmonic corrections to the entropy and relative enthalpy of ilmenite-type MgSiO 3 do not exceed 4(3) J/(mol·K) and 4(3) kJ/mol at 2000 K, respectively. These corrections are lower than those inferred for other mantle minerals. Above 900 K, a backtransformation to an MgSiO 3 glass was first observed, prior to recrystallization to enstatite near 1000 K. Back-transformation systematics for high-pressure silicate phases are discussed.

The crystal structure of synthetic, deuterated katoite [Ca 3 Al 2 (O 4 D 4 ) 3 ] has been refined at 0.0001,0.78,3.6,6.1,7.9, and 9.0 GPa using the opposed-anvil Paris-Edinburgh cell and the POLARIS powder diffractometer at the U.K. pulsed spallation source, ISIS. A constrained Birch-Murnaghan fit to the unit-cell volume yields K 0 = 52(1) GPa for K 0 = 4.0. The lower bulk modulus of katoite relative to anhydrous Ca-bearing garnets (K 0 = 159-179 GPa) is due to the greater compressibility of the Ca dodecahedron and O 4 D 4 tetrahedron. At high pressure, comer-sharing tetrahedra and octahedra undergo a rigidbody rotation, which causes a shift in the a position. This rotation, coupled with the large compression of the two tetrahedral edges shared with the dodecahedron, changes the relative lengths of the Ca-O bonds at high pressure, such that Ca2-O < Ca1-O for P > 6 GPa. With increasing pressure, the O-D···O angles widen slightly, as the O-D vector rotates toward the tetrahedral face. Both O-D and O···D bond lengths decrease as a function of pressure, which was unexpected in view of the results of recent high-pressure studies. Comparison of crystallographic and spectroscopic data collected at high pressure for hydrogarnets suggests that empirical relationships derived for hydrogen-bond systems at ambient conditions, where a relaxed, equilibrium state is attained, may not always apply to O-D···O hydrogen bonds under compression.

Crystals of sodium tetrasilicate (Na 2 Si 4 O 9 ) have been grown at 6 GPa and 1000 °C to 9 GPa and 1500 °C using the MA6/8 superpress at Edmonton, and the X-ray structure was determined at room pressure (R = 5.0%). High-pressure sodium tetrasilicate is monoclinic, with a = 10.875(2), b = 9.326(1), c = 19.224(7) Å, β= 90.18(2)°, space group P2 1 /n, and D x = 3.090 g/cm 3 . Si occurs in both tetrahedral and octahedral coordination, with [6] Si: [4] Si= 1:3; the structural formula is Na 6 Si 3 [Si 9 O 27 ], Z= 4. Nine-memberedrings of SiO 4 tetrahedra are collapsed around and interconnected by SiO 6 octahedra, at shared comers, giving a framework structure that is analogous to but distinct from those of K 2 Si[Si 3 O 9 J (wadeite-type) and K 2 Ge[Ge 3 O 9 J(A 2 Ge 4 O 9 -type). Although all nine independent SiO 4 tetrahedra have similar nearest-neighbor stereochemistries, [4] Si-O- [4] Si bond angles vary markedly (130.6-172.1°). The Na cations are displaced to one side of framework cavities in irregular polyhedra, as in other framework structures, with six to eight bonds extending to 2.9 A. It appears that even transition-zone pressures do not dominate the stereochemical requirements of the large cations in determining the structures of the alkali and alkaline-earth aluminosilicates.

The structure of lizardite-1T from Monte Fico, Elba, was refined in space group P31m using neutron diffraction data, measured at 8, 150, and 294 K, and full-profile Rietveld refinement techniques. The lattice parameters at 8 K [a = 5.3267(2), c = 7.2539(6) Å], 150 K [a = 5.3260(2), c = 7.2574(6) Å], and 294 K [a = 5.3332(2), c = 7.2718(6) Å] show nonlinear expansion, with nearly all volume change above 150 K. H positions were precisely refined at 8 K. The inner H4 atom deviates from the idealized O,O,z positions and is disordered over three symmetry-related positions 0.24 Å away from the ternary axis. The outer H3 atom location is consistent with the previous single-crystal X-ray structure refinement. On the basis of the present thermal expansion data and previous compressibility measurements, the following equation of state for lizardite-1T is proposed: V P,T = V 0 [1 + 32.8 × 10 -6 (T - 294) - 15.5 × 10 -4 (P - 0.001)]. Accordingly, the constant volume condition requires geothermal gradients on the order of 15 °C/km.

near-end-member natural tremolite, Na 0.01 Ca 1.97 Mg 4.98 Fe 0.03 Al 0.01 Si 8.00 O 22 (OH) 2 , studied by single-crystal X-ray diffraction at 140 and 295 K to seek a possible crystalchemical explanation for the typically low Ca/Σ M ratios, relative to the ideal ratio of 2/5, observed in both natural and synthetic tremolite samples. Difference-Fourier maps revealed the presence of a residual electron density close to the M4 site along the diad axis toward the octahedral strip. Structure refinements indicated that the M4 and M4ʹ sites are occupied by Ca + Na and M(Fe + Mg), respectively. In comparison with the configuration of the M2 coordination polyhedron in diopside, the degree of distortion and the volume of the M4 coordination polyhedron in tremolite are relatively large and the M4 cation is slightly underbonded. These two factors contribute to an energetic drive toward M-enriched tremolite. The average unit-cell volume of 906.6(2) Å 3 determined at 295 K for nearly pure tremolite in this study suggests an end-member reference-state volume for tremolite of 907 Å 3 . This indicates that cell volumes of synthetic tremolite of 904.2(4) Å 3 reflect 8-10% cummingtonite solid solution, as previous authors have claimed. Similarly, β angles of 104.53-104.58° for synthetic tremolite suggest similar amounts of cummingtonite, when compared to β = 104.76(2)° for near-end-member tremolite and 104.75(5)° for many natural tremolite samples from calc-silicate parageneses. The compositions of natural and synthetic tremolite samples are consistent with theoretical constraints on the Ca/Σ M ratio of tremolite. Decreasing temperature increases the Ca/Σ M ratio of tremolite in both Mg-saturated (+ talc, anthophyllite, cummingtonite, or enstatite) and Ca-saturated (+ diopside, quartz, H 2 O) parageneses. Stabilization of cummingtonite- tremolite solid solution caused by M4-site splitting (M on M4ʹ) is small in comparison with the enthalpic consequence of size mismatch. The configurational entropy term (mixing of Ca and Mg) accounts for the coexistence of cummingtonite-bearing tremolite with diopside, quartz, and H 2 O, and for the problem of achieving 100% synthesis of tremolite of ideal composition.

In situ-heating transmission electron microscopy and high-resolution transmission electron microscopy studies of the thermal decomposition of tremolite show that transformed pyroxenes have C2/c and P2 1 /c structures and retain axial orientation of the tremolite structure with I2/m orientation. Amorphous silica forms only on the surface of the heated crystals. Thermal decomposition of tremolite, which takes place at 740 °C, is characterized by the formation of (010) clinopyroxene slabs with thicknesses of 18 and 36 Å along the b axis. The (010) slabs with C2/c symmetry are coherent with the tremolite matrix. Heated tremolite with (010) slabs has the same composition as “anhydrous tremolite.” Clinopyroxene domains formed at 780 °C are composed of C2/c and P2 1 /c clinopyroxenes with an exsolution lamella-like texture. The products characterized by clinopyroxene domains are isolated by tremolite with (010) clinopyroxene slabs. The bulk composition of transformed clinopyroxenes is almost the same as that of anhydrous tremolite. The structure of the clinopyroxenes with stoichiometry (Ca,Mg) 7 Si 9 O 23 can be considered to be the pyroxene structure with a random distribution of (Ca,Mg)O vacancies in the octahedral bands. The clinopyroxenes with (Ca,Mg)O vacancies are thermodynamically unstable. The texture consisting of pyroxene chains in multiples of four within the amphibole matrix may indicate a dehydration reaction of amphibole, rather than hydration of pyroxene. Further annealing of the heated specimen at higher temperature increases the concentration of Ca cations in C2/c clinopyroxene and results in coarser clinopyroxene lamellae. Transformed clinopyroxene with only P2 1 /c symmetry, described by Freeman and Taylor (1960), is an artifact of overlapping single-crystal X-ray diffraction patterns of C2/c and P2 1 /c structures. The transformation product of one diopside-like clinopyroxene, described by Wang and Zhang (1991), results from weak diffraction spots violating the C-centered lattice structure that did not appear in X-ray powder diffraction patterns.

Microcline + albite and albite + quartz standard mixtures were quantitatively analyzed using Rietveld refinement to investigate the reliability of the method. The method was also applied to the determination of weight percentages of component minerals in a perthite sample. The analysis resulted in a standard deviation of approximately I wt% for the microcline + albite mixtures and 4 wt% for the albite + quartz mixtures, indicating that the accuracy of quantitative Rietveld analysis is far better than that of conventional XRD methods in which only a few select peaks are used. The inclusion of the Dollase-March preferred-orientation correction function slightly improved the accuracy of the results. That function also improved the rate of convergence and the accuracy of unit-cell parameters. However, the Rietveld-Toraya correction function did not work well with the quantitative analyses. The result of the quantitative analysis of the albite + quartz mixtures was poorer than that of the microcline + albite mixtures, and the weight fraction of albite tended to be overestimated. The analytical result of the microcline + albite mixtures implies that the Rietveld method is more accurate when component minerals have similar preferred-orientation characteristics. However, if component minerals have different crystal habits, as represented by the albite + quartz mixtures, the deviation is much greater and systematic errors may occur. Quantitative phase analysis of a perthite specimen using the Rietveld method resulted in very accurate values for the mineral proportion and unit-cell parameters of the constituent alkali feldspars, demonstrating the power of the Rietveld method to analyze multiphase samples.

Samples along the dolomite-ankerite join were synthesized using a piston-cylinder apparatus and the double-capsule method. Some of the ankerite samples may be disordered. Thermal analysis and X-ray diffraction showed that all samples can be completely decomposed to uniquely defined products under calorimetric conditions (770°C, O 2 ), and a well-constrained thermodynamic cycle was developed to determine the enthalpy of formation. The energetics of ordered and disordered ankerite solid solutions were estimated using data from calorimetry, lattice-energy calculations, and phase equilibria. The enthalpies of formation of ordered dolomite and disordered end-member ankerite from binary carbonates, determined by calorimetry, are -9.29 ± 1.97 and 6.98 ± 2.08 kJ/mol, respectively. The enthalpy of formation of ordered ankerite appears to become more endothermic with increasing Fe content, whereas the enthalpy of formation of disordered ankerite becomes more exothermic with increasing Fe content. The enthalpy of disordering in dolomite (approximately 25 kJ/mol) is much larger than that in pure ankerite, CaFe(CO 3 ) 2 (approximately 10 kJ/mol), which may explain the nonexistence of ordered CaFe(CO 3 ) 2 .

We have measured the pressures of decrepitation of vesicles in synthetic glasses of feldspar compositions (NaAlSi 3 O 8 -KAlSi 3 O 8 ). Vesicles filled with Xe do not decrepitate at internal pressures of 160 MPa, indicating that the unflawed surface of the vesicle wall has an intrinsic strength > 80 MPa. Vesicles containing CO 2 escaped decrepitation and displayed ductile deformation when the Tg was reached at the maximum P of 200 MPa (indicating an intrinsic strength higher than 100 MPa). Vesicles containing H 2 O showed dramatically reduced strength, decrepitating at internal pressures on the order of 1-5 MPa. The H 2 O-filled vesicles leaked slowly over periods of several weeks or months. The relative stability of the inclusions is strongly dependent on the quench rate, with rapidly quenched inclusions showing greater stability over long periods of time. Microscopic examination revealed the presence of radial microfractures in the walls of H 2 O-filled vesicles. We account for the microfracturing with reference to recent studies of chemical-gradient stress. Our observations may account for a variety of phenomena, which occur wherever hydrous vesicular glasses are formed, including explosive decompression of vesicular glassy rock in near-surface volcanic environments, spontaneous decrepitation of vesicular basaltic glass dredged from the seafloor (“popping rocks”), and rapid loss of H 2 O from synthetic vesicular glasses produced in laboratory experiments investigating fluid-melt phase equilibria.

The viscosity η of dry and hydrous haplogranitic melts (anhydrous normative composition: QZ 28 Ab 34 Or 38 ) has been determined between 3 and 10 kbar and 800 and 1400 °C using the falling-sphere method. The H 2 O content of the melt ranged from 0.03 to 8.21 wt%. Experiments were performed in internally heated pressure vessels (T = 900-1400 °C)and cold-seal pressure vessels (T = 800 °C). The viscosity decreases with increasing H 2 O content of the melt. The strongest decrease is observed at low H 2 O concentrations. The effect of H 2 O is smaller at high H 2 O concentrations in the melt, with an almost linear behavior between log η and H 2 O content expressed as weight percent H 2 O (decrease of 0.26 log units per weight percent H 2 O for H 2 O contents ≥4 wt% H 2 O). No pressure effect could be observed between 3 and 10 kbar at 900 °C for a melt containing 5.90 wt% H 2 O. In the investigated range the activation energy of the viscous flow decreases from 195 to 133 kJ/mol for melts with 1.05 to 8.21 wt% H 2 O. The effect of T and H 2 O content of the melt on viscosity can be calculated with a precision of ±2 log units with the use of the following expression: log η= -1.57 + [23.398 - 13.197(c H₂O ) 0.11 ] × 10 3 (1/T). Viscosities calculated using the model of Shaw (1972) show that, for the investigated composition, the model underestimates the influence of H 2 O for low H 2 O concentrations (≤4 wt% H 2 O, difference up to two orders of magnitude at 800 °C) and overestimates slightly the influence of H 2 O for high H 2 O concentrations (≥5 wt% H 2 O). In comparison with the model of Persikov et al. (1990), which takes into account the OH- and molecular H 2 O proportions, the experimental data at 800°C are in good agreement with the calculated viscosities (deviation ≤1 log unit) assuming that the proportions of OH - groups and molecular H 2 O are those found in an in situ spectroscopic investigation of the melt. However, at higher temperatures (1000-1300 °C)the viscosity is overestimated for the OH - and H 2 O proportions recalculated for the appropriate temperatures.

Current models of lunar crustal stratigraphy are based primarily on models of impact cratering coupled with what is known about the distribution of rock types in the ejecta blankets associated with the large multiring basins. These models postulate an upper crust rich in the ferroan anorthosite component underlain by a more mafic (noritic) lower crust (Spudis 1993). We attempted to test these models by determining the depth at which various lunar plutonic rocks formed. Despite intense brecciation and shock, many lunar highland samples retain vestiges of a prior magmatic history. For example, pyroxene samples preserve a record of their thermal history in such phenomena as cation ordering and exsolution textures. By computer simulation of the growth of exsolution lamellae in augite and pigeonite, we deduced cooling rates and from these estimated a depth of burial on the basis of a simple model for the thermal evolution of the lunar crust. Despite the fairly large uncertainties in the calculations, it is clear that the ferroan anorthosite samples we studied formed at a depth between 10 and 20 km, whereas some members of highland magnesian and alkali suites formed as cumulates in magma chambers intruded into the uppermost layers of the crust. As yet there is no evidence to indicate that crystalline rocks from the lower half of the lunar crust are present in the samples returned by the Apollo missions. Most samples derived from the lower crust appear to be impact melts of low-K Fra Mauro (noritic) composition.

The time-dependent loss of NaKα X-ray intensity during electron-beam irradiation of hydrous alkali alumino silicate glasses is apparently more significant during the initial few seconds of beam exposure than it is for anhydrous glasses, and it is pronounced for incident beam currents > 2-5 nA (using 15-20 μm beam diameters). Exponential fits of NaKα intensity vs. time show a progressive decrease in the apparent zero-time intercepts for incident beams from 2 to 20 nA, and thus methods for correcting Na concentrations solely on the basis of curve fitting and extrapolation to zero-time values may underestimate Na contents by almost 10% (relative) for higher beam currents. Similar exponential fits to the intensity-time data for AlKα and SiKα show that “grow-in” is greater for Al than for Si. For incident currents ≥ 5 nA, the magnitudes of all intensity changes also increase with total H 2 O content of glass. On the basis of these observations, the optimal conditions for analysis of hydrous alkali aluminosilicate glasses include a 2 nA beam with 20 μm diameter and counting times of 20-40 s for the analysis of alkali aluminosilicate components, with Na and Al analyzed first (simultaneously, if possible). These methods minimize Na loss and grow-in for Al and Si to the point that little or no correction is needed, provide good statistical accuracy, and work with a wide variety of standard materials (i.e., glass standards with compositions and H 2 O contents comparable to the unknowns are not needed). For complete analysis of more complex multicomponent systems, two beam conditions are recommended: an initial 2 nA, 20 μm diameter beam for analysis of alkali alumino silicate components, followed by a 20 nA, 20 μm diameter beam for analysis of all other components. With the use of these methods, the H 2 O contents of hydrous glasses (H 2 O as the only unknown) can be determined by difference with uncertainties mostly <5% (relative to FTIR values) for glasses containing up to 10 wt% H 2 O. At beam currents > 5 nA, corrections for Na loss ignoring Al (and Si) grow-in underestimate H 2 O contents by about 10-50% of concentration.

The availability of a synthetic multilayer crystal and accurately calibrated oxide and silicate standards make it possible to use the electron microprobe for precise a analysis of spinel. A requirement of the a measurement routine described is the use of repetitive statistical analyses of the a standards and subsequent corrections and recalibration. A representative set of a analyses for each spinel population studied is essential to obtain reliable data, and the danger of using single datum is emphasized. Magnesiochromite spinel grains, having broad compositional similarities but different petrogenetic and cooling histories, were analyzed for a and their stoichiometry was assessed. Diamond-indicator spinel from the Aries kimberlite and Argyle lamproite is stoichiometric. Spinel inclusions in olivine phenocrysts from Ti-poor tholeiite from the Hunter Fraction Zone and Ca-rich boninite from the Tonga Trench show a range of nonstoichiometry. High Fe 2+ / Fe 3+ values calculated assuming stoichiometry for such spinel are invalid. Spinel samples from metamorphosed volcanics from the Peak Hill-Glengarry Basin and the Heazlewood River Ultramafic Complex are also nonstoichiometric, having significant Fe 83 O 4 -Cr 83 O 4 components. Our results demonstrate that nonstoichiometry is a common feature of Cr-rich spinel. This has important implications for the use of Fe 3+ and Fe 2+ concentrations to estimate the oxidation state or temperature of formation.

Trace element (REE, Cr, Ti, Y, Y, and Zr) analysis of garnet from the garnet, staurolite, and lower sillimanite zones of an aluminous schist of the Black Hills, South Dakota, indicates that REE zoning varies as a function of grade. Garnet-zone garnet has high concentrations of REEs, Cr, Ti, Y, Y, and Zr in the cores and low concentrations in the rims. Profiles of heavy REEs contain inflections between the cores and rims, which are approximately symmetric about the cores. Staurolite-zone garnet contains cores enriched with Y and heavy REEs, which decrease toward the rim and increase again at the rim edges but to lower concentrations than in the cores. Cr, Y, Ti, Zr, and light REE zoning is less pronounced than heavy REE zoning and is less symmetric about the garnet cores. Almandine-rich garnet of the lower sillimanite zone displays no major element zonation. Trace element (Ti, Cr, Y, and Zr) concentrations are minimal, and the zoning is irregular and not symmetric about the garnet cores. Garnet from all three zones has core-to-rim Fe/(Fe + Mg) profiles that suggest garnet growth was uninterrupted with respect to major element components and that Mn zoning formed by a fractionation process. Analysis of trace element zoning in this garnet reveals that the major element zoning was relatively unaffected by volume-diffusion reequilibration. Trace element zonation of all samples of garnet is best explained by a fractionation mechanism in conjunction with limited intergranular diffusion and changing partition coefficients during garnet growth. Heavy REE partitioning is especially dependent on the major element composition of garnet. This research complements previous research by others on the use of trace elements as metamorphic petrogenetic indicators, which demonstrated the importance of bulk-rock composition and phase assemblage on trace element partitioning.

We have determined the concentration profiles across the interface of a natural garnetgarnet couple in a biotite-grade rock from eastern Vermont. The couple consists of a grossular-spessartine garnet that had formed during regional metamorphism associated with the Acadian orogeny on an almandine core, which had crystallized during an earlier episode of metamorphism related to the intrusion of Fairlee granite at 411 ± 5 Ma. The concentration profiles were measured by both electron microprobe and analytical transmission electron microscope, and they were modeled to retrieve the value of ∫D(t)dt through the time that diffusion was effective but without recourse to any diffusion data. The length of the concentration profiles measured in microprobe is barely resolvable from that resulting from a convolution effect from the spatial averaging in the spot analyses. Deconvolution of the microprobe profiles yields a value of ∫D(t)dt = 0-3.4 × 10 -11 cm 2 , suggesting very little or no diffusion. TEM analyses of the concentration profiles, which are not subject to any significant convolution effect, show a very small but definitive diffusion zone across the interface of the garnet-garnet couple, which yields a value of ∫D(t)dt = 7.6 × 10 -12 cm 2 . Because D is a function of time through its dependence on temperature, this value of ∫D(t)dt provides an important constraint on the thermal history during the regional metamorphism. As an example, we used it in conjunction with the available diffusion data for garnet to derive ∼40-50 Ma as the probable time scale for the biotite-grade metamorphism, taking into account the effects of the off-diagonal terms and thermodynamic nonideality on the diffusion process.

Several coesite grains along garnet-omphacite grain boundaries were discovered in eclogites from Yangkou Beach of the Sulu region, eastern China. These grains exhibit variable extents of conversion of coesite to palisade- through mosaic- to granoblastic-quartz aggregates. The rare occurrence of intergranular and matrix coesite in ultrahigh-pressure (UHP) crustal rocks is related to the fast rate of exhumation and the low amount of fluid infiltration required during retrograde recrystallization. Sluggish reactions due to the lack of fluid in both UHP and retrograde metamorphism resulted in the occurrence of intergranular coesite in these eclogites, the preservation of gabbroic texture and relict igneous minerals, and the metastable persistence of low-P assemblages in the coesite-stability field.

The compositional evolution of tourmaline from diagenetic to lower amphibolite-facies conditions was investigated in two metasedimentary units (redbed and black shale formations) along a traverse across the Central Alps. With increasing metamorphism, three distinct rim zones grew around detrital tourmaline cores, simultaneously with the development of prismatic neoblasts. Detrital cores and all successive rim zones are preserved in the amphibolite-facies rocks. As indicated by whole-rock data, the B required for tourmaline growth was released from illite and muscovite and was not introduced from external sources. The studied samples of tourmaline exhibit compositional polarity, i.e., their compositions are different at the opposite ends of the c axis. The compositional difference between the two poles can be expressed by the substitutions Na + Mg ⇌ ⃞ + Al and 2Al ⇌ Ti + Mg, with the positive pole always richer in Al and having more X-site vacancies. Both chemical and optical polar effects are most pronounced in the internal rim zones and become less prominent toward the external zones. The systematic prograde compositional trends are depicted in a vector space for Li-poor tourmaline. At low-grade conditions, the tourmaline composition is clearly controlled by the host-rock composition, but with increasing metamorphic grade the compositions of tourmaline from a variety of rocks converge. The data show that in the Central Alps, increasing metamorphic grade is reflected by increases in Ca/(Ca + Na) and Mg/(Mg + Fe 2+ ), a decrease in Fe 3+ , and an increase in the occupancy of the X site of tourmaline.

Relatively low-temperature, secondary alteration is common in samples of metamict betafite and initially proceeds by the substitution mechanisms A Na Y F → A ⃞ Y ⃞, A Ca Y O→ A ⃞ Y ⃞,and A Ca X O→ A ⃞ Y ⃞. Alteration is usually accompanied by hydration (∼10-15 wt% H 2 O) together with minor increases in Al, K, Mn, Fe, Sr, and Ba. At this stage, U and Th remain relatively unaffected by the alteration process. Once Na and F are removed and the Ca content drops below about 0.2-0.3 atoms per formula unit (∼2.5-3.5 wt% CaO), betafite bulk compositions fall within the stability field of liandratite + uranpyrochlore + rutile (or anatase), thus promoting major element mobility (including Th, U, Pb, and B-site cations), incipient recrystallization, and partial dehydration. New phase assemblages occur as a function of decreasing bulk U content in the order liandratite + rutile, liandratite + uranpyrochlore + rutile, and uranpyrochlore + rutile. The same phase assemblages also occur in laboratory heating experiments performed in an inert atmosphere at 1000°C. Up to 20-30% of the original amount of U may be lost during severe secondary alteration and recrystallization of betafite. Part of this U is retained by liandratite crystallized in the adjacent host rock. Loss of radiogenic Pb results from both long-term diffusion and secondary alteration, aided by radiation damage-induced volume expansion and microfracturing.

A high-precision electron microprobe (EMP) technique has been developed that is capable of analyzing major, minor, and trace element abundances (Si, Ti, AI, Fe, Mn, Mg, Ca, Na, K, Cl) in hydrous rhyolitic glasses. The technique was developed to characterize the chemical compositions of rhyolitic glass inclusions in phenocrysts that occur in layers of Paleozoic altered volcanic ash. The compositions of these inclusions serve as excellent chemical "fingerprints" of the altered volcanic ash layers for use in stratigraphic correlation. The precision and reproducibility of the analyses is sufficient not only to distinguish one altered volcanic ash layer from another on the basis of inclusion compositions, but also to discern differences in the compositions of different inclusions from the same layer. A high-precision instrumental neutron activation analytical (INAA) technique was also developed that is capable of measuring an additional suite of trace elements (e.g., Sc, Co, Rb, Cs, Sr, Ba, La, Ce, Sm, Eu, Tb, Yb, Lu, Zr, Hf, Ta) in rhyolitic glass inclusions in quartz phenocrysts with excellent accuracy and precision after correcting for the presence of the host quartz. The abundances of elements measured by the EMP technique and the corrected abundances determined using the INAA technique are identical within analytical uncertainty, thus demonstrating the internal consistency of the results.

Florencite-(La) (La/Ce = 1.09) with fissiogenic REEs and florencite-(Ce) (La/Ce = 0.62) have been identified in illite from the clay mantle surrounding a natural, 2 Ga fission reactor at Bangombe and in sandstone beneath the reactor zone, respectively. Florencite- (Ce) is apparently unrelated to nuclear processes and occurs with monazite-(Ce), apatite, Ti02 (probably anatase), zircon, and illite. Grains of florencite-(Ce) contain inclusions of thorite, chalcopyrite, and galena. Florencite-(La) was found 5 cm from the “core” of the reactor and contains inclusions of galena and U-Ti-bearing phases. Secondary uraninite and coffinite have precipitated on some of the florencite grains. The chemical composition of florencite-(La) as determined by electron microprobe analysis is (La 0.38 Ce 0.35 Nd 0.06 Sm 0.01 Ca 0.03 Sr 0.17 )(Al 2,98 Fe 3+ 0.02 )(PO 4 )[PO 3.80 (OH) 0.20 ](OH) 6. Secondary ion mass spectrometry revealed that between 27 and 30% of Nd and 67 and 71% of Sm in florencite-(La) is fissiogenic. The presence of fissiogenic REEs in “florencite” from the reactor zone in Bangombe and their preferential concentration in florencite relative to the bulk sample of clay demonstrate that aluminous phosphates may have played a more significant role in the fixation of fissiogenic REEs released from uraninite after the sustained fission reactions than sorption onto clays.

Fianelite occurs in iron and manganese ores at Fianel, a small mine in Val Ferrera, Graubunden, Switzerland. It is associated with quartz, aegirine, and limonite in thin open fractures crosscutting quartz + rhodonite + kutnohorite + palenzonaite ± aegirine ± parsettensite ± saneroite ± pyrobelonite veinlets. Only a small amount of very minute, tabular, (010)-flattened fianelite crystals have been collected. Fianelite, with ideal formula Mn 2 V(V,As)O 7 ·2H 2 O, crystallizes in the P2 1 /n space group with a = 7.809(2), b = 14.554(4), c = 6.705(4) Å and β = 93.27(3)°. The strongest lines in the X-ray powder pattern are d 022,230 = 3.039, d 120 = 5.32, d 231 = 2.721, and d 163,362 = 1.593 Å. The orange-red mineral is transparent; its optical sign is unknown, and 2V is small. Fianelite shows a strong pleochroism from yellow to red. Microprobe analyses revealed a partial substitution of As for V, leading to the empirical formula Mn 1.98 (V 1.57 As 0.44 Si 0.01 )O 7 ·2H 2 O. The crystal structure of fianelite, refined to R = 0.058, consists of Mn in octahedral chains bound together by pairs of V in tetrahedra. Among the two V positions, only one hosts As. H is present in H 2 O groups belonging to the octahedra and occupying channels extending through the structure along [001]. No mineral having a chemical formula with the same atomic proportions as fianelite shows similar structural features.

Laurelite, Pb 7 F 12 Cl 2 , from the Grand Reef mine, Graham County, Arizona, is hexagonal, P6̄, with a = 10.267(1) and c = 3.9844(4) Å and Z = 1. The crystal structure was solved by direct methods and refined to R = 0.035 and R W2 = 0.089 for 693 measured reflections (Fo > 9σ Fo ). The structure is related to that of α-PbF 2 . Both are based upon ninefold-coordinated Pb as tricapped trigonal prisms (TCTPs), which share edges and faces. The two structures can be described with respect to the face-sharing linkages of their TCTPs. The structure of α-PbF 2 consists of corrugated sheets of face-sharing TCTPs that interlock by edge-sharing perpendicular to the c axis. In laurelite, the Pb2 TCTPs form three-membered face-sharing clusters about the threefold axis that are propagated into trigonal cylinders by sharing faces in the direction of the c axis. The Pb 1 and Pb3 TCTPs are linked by face-sharing into a three-dimensional framework with corresponding cylindrical voids. Asymmetric coordinations about Pb1 and Pb2 are attributed to the stereoactive lone-pair effect. Although the coordinations about the anions appear to disallow substitution of OH for F, stacking defects along the c axis provide a mechanism for accommodating limited OH or H 2 O for F substitution. A new density determination yielded 7.65(5) g/cm 3 , in reasonable agreement with the density of 7.77 g/cm 3 calculated on the basis of the empirical formula Pb 0.97 [F 1.68 Cl 0.25 (H 2 O) 0.07 ], Z = 7.

Raman spectra of two synthetic, noncubic garnet samples rich in MgSiO 3 component were measured in a diamond-anvil cell to 200 kbar at room temperature. Sample 1 was pure MgSiO 3 majorite; sample 2 was a solid solution with 93.4 mol% MgSiO 3 and 6.6 mol% Mg 3 Al 2 (SiO 4 ) 3 . Although the X-ray patterns of both samples can be indexed assuming tetragonal symmetry, it cannot be completely ruled out that sample 2 is orthorhombic. High-pressure Raman spectra of sample 1 show a continuous shift of all bands to higher frequency with increasing pressure; there are no indications of phase transitions. Sample 2, however, undergoes a phase transition between 51 and 85 kbar on compression; this phase transition is completely reversible between 54 and 38 kbar on decompression. In the frequency range between 300 and 800 cm -1 , several bands that are present in the low-pressure phase disappear or merge in the high-pressure spectra, whereas some of the remaining bands intensify. The high-pressure spectrum of sample 2 is very similar to the room-pressure spectrum of sample 1. The phase transition could possibly be the result of a symmetry change from an orthorhombic to a tetragonal space group.

First-principles linear-response calculations were used to investigate the lattice dynamics of what is thought to be the third most abundant phase in the lower mantle, CaSiO 3 perovskite. The commonly assumed cubic structure (Pm3m) was found to be dynamically unstable at all pressures, exhibiting unstable modes along the Brillouin zone edge from the M point to the R point. On the basis of these results, we predict that the ground-state structure of CaSiO 3 perovskite is a distorted phase with lower than cubic symmetry. We predict that a phase transition occurs in CaSiO 3 perovskite within the Earth's lower mantle from the low-temperature distorted phase to the high-temperature cubic phase. The predicted phase transition possibly explains some of the seismological observations of reflective features within the lower mantle.

An equation is presented for the calculation of the shear viscosity of hydrous (0-12.5 wt% H 2 O) leucogranitic melts from 10 2 to 10 13 Pa·s. The equation is a multiple nonlinear least-squares regression of a data set of III viscosity determinations in the literature. It is based on the Vogel-Fulcher-Tammann (VFT) form and thus accounts for the very important non-Arrhenian temperature dependence of the viscosity. This is possible because of the inclusion of data obtained recently in the high-viscosity region. The equation includes a logarithmic dependence of the three adjustable parameters of the VFT equation on the H 2 O content of the melt. The root-mean-standard deviation for the III data points included is 0.46log 10 units. In comparison with earlier methods based on the Arrhenian approximation of the temperature dependence of viscosity, the present method provides significant improvement in the accurate prediction of viscosity. We recommend its use in all petrologic calculations involving hydrous, metaluminous, leucogranitic melts. It is an especially powerful method for calculating the high viscosities thought to be relevant to erupting silicic volcanic systems and magmatic hydrothermal granitic and pegmatitic systems at the brittle-ductile transition.