Volume 81, Issue 3-4

Assignment of the peaks near 485 cm -1 in vibrational spectra of SiO 2 -M 1/n n+ AlO 2 (where M = an alkali or alkaline earth element) glasses to v s (T-O-T) modes in four-membered rings was tested with molecular orbital calculations. Potential energy minima of cyclic (H 8 Si m Al n O 12 ) n- (where m + n = 4), Li 2 H 8 Si 2 Al 2 O 12 ,Mg 2 H 8 Al 4 O 12 ,(H 6 Si 2 AlO 9 ) 1- , and LiH 6 Si 2 AlO 9 molecules were calculated. Force-constant analyses of the potential-energyminimized structures predict v s (T-O-T) modes between 490 and 430 cm -1 depending on the number of Al atoms in the negatively charged four-membered rings and v s (T-O-T) modes at 525 and 545 cm -1 in Li 2 H 8 Si 2 Al 2 O 12 and Mg 2 H 8 Al 4 O 12 ,respectively. The theoretical frequency shift of the v s (T-O- T) mode calculated for the (H 8 Si m Al n O 12 ) n- rings is consistent with the experimentally observed shifts of the peaks at 495-485 and 495-515 cm -1 in glasses along the SiO 2 -NaAlSiO 4 and SiO 2 -CaAl 2 Si 2 O 8 joins, respectively (Mc- Millan et al. 1982; Seifert et al. 1982). Force-constant analyses of (H 6 Si 2 AlO 9 ) 1- and LiH 6 Si 2 AlO 9 molecules predict a frequency of 573 cm -1 for the v s (T-O-T) mode in both molecules, consistent with the assignment of peaks near 600 cm -1 in glasses along these joins to v s (T-O-T) modes in three-membered rings (McKeown et al. 1984; Kubicki and Sykes 1993a). Intermediate-range order in SiO 2 -M 1/n n+ AlO 2 glasses is interpreted in terms of changes in ring statistics. In SiO 2 , the Raman peaks at 606 (D 2 ), 495 (D 1 ), and 430 cm -1 correspond to three-, four-, and >four-membered rings, respectively. The predominant low-frequency peaks in the vibrational spectra of glasses along the SiO 2 -NaAlSiO 4 and SiO 2 -CaAl 2 Si 2 O 8 joins at 540-606 and 485-515 cm -1 are ascribed to three- and fourmembered rings, respectively.

Raman spectra of Na 2 Si 2 O 5 glass and Na 2 Si 2 O 5 liquid have been collected in situ to pressures of 16 GPa at 300 K and 10 GPa at 885 K. In glass compressed at 300 K, bands associated with intertetrahedral bridging 0 atoms decrease in intensity and ultimately become unresolvable above 8 GPa. Over this same pressure range, new peaks appear in the glass between 600 and 800 cm -1 ; these new features are consistent with the formation of SiO 5 or SiO 6 polyhedra through the destruction of nonbridging O atoms. Between ∼10 and 16 GPa, band intensities shift, the total integrated Raman scattering intensity of the glass decreases by more than a factor of ten, and the pressure dependence of mode shifts changes. These changes are consistent with a new densification mechanism initiated in the glass above ∼ GPa, probably from the formation of linkages between two or more highly coordinated Si atoms. No vibrations attributed to bridged silica tetrahedra are observed in the liquid above 8 GPa, indicating that at least 50% of the Si in the liquid at 8 GPa is highly coordinated. Upon decompression of thermally quenched samples at 300 K, bands associated with bridged tetrahedra reappear at pressures below 1 GPa. Similarly, glasses that have been compressed only at 300 K and then decompressed have bands associated with bridged tetrahedra. The formation of bands associated with bridged tetrahedra on decompression both in glasses quenched from liquids formed at high-pressure and in glasses compressed at 300 K demonstrates that these species arise from structural reorganizations during decompression. Such reorganizations plausibly involve the breakdown of highly coordinated Si polyhedra and the resultant formation of bridged tetrahedra. The nearly complete decomposition ofthe high-pressure structure in glasses on decompression documents that in situ high-pressure measurements are crucial in deriving accurate constraints on the structure of silicate liquids at high pressure.

Thermodynamic and transport properties of ten compositions in the binary system NaAlSiO 4 -SiO 2 have been determined by the method of molecular dynamics simulation. The computations were performed on collections of 1300 particles at temperatures between 2500 and 4500 K and pressures of 4 ± 1.5 GPa using a pairwise-additive potential energy function with Coulombic and Born-Mayer interactions. Results are presented for both thermodynamic and transport properties; compositional patterns in the computed properties are considered in relation to structural information, such as nearest-neighbor and second-nearest-neighbor coordination and intertetrahedral angle distributions (see Stein and Spera 1995). Heat-capacity values extrapolate to within 0.25R of those found in laboratory measurements in the range 1000-1800 K, and estimates of compressibility and expansivity agree with available data within a factor of two to four. Transport properties are computed by time-correlation statistics from configuration histories between 0.04 and 0.09 ns in duration, and results are presented for ionic conductivity and self-diffusivity. For the charge-balancing cation Na, diffusivity displays Arrhenian variation from 2 × 10 -5 cm 2 /s to 1 × 10 -4 cm 2 /s. These results extrapolate with fair success into the thermal regime of experimental measurements. For network formers O, Si, and Al, diffusivity varies from about 2 × 10 -8 cm 2 /s for SiO 2 -rich compositions at 2500 K to 2 X 10 -6 cm 2 /s for NaAlSiO 4 at 4500 K. For all ions, diffusivity tends to increase along isotherms with addition of NaAlO 2 to SiO 2 , but activation energy shows no systematic compositional variation. Diffusive activation energy for network formers (averaging 128, 118, and 140 kJ/mol for Al, Si, and O, respectively) displays neither the magnitude nor compositional variability anticipated from application of the Stokes-Einstein or Eyring relations to extensive viscometric data for this compositional series. Analysis shows that in the ranges of temperature, density, and duration of the present series of experiments, a critical transition occurs in the O diffusion mechanism, from continuous network reorganization at high temperatures to discrete hops by a few mobile particles between static sites at low temperatures. Simulation temperatures lie mostly below a critical transition region postulated in mode-coupling theory for condensed matter, and bulk network diffusivity under these conditions is relatively uninformative regarding processes of viscous flow. Aspects of the transition are explored by using the statistics of individual particle displacements, using the van Hove density-autocorrelation function, and studying the trajectories of individually tagged particles. Analysis of the frequency of site exchange among a subpopulation of mobile O ions gives results more consistent with extrapolated values of viscosity at these temperatures. The apparent Arrhenian activation energy of the site-exchange process is similar to that measured for viscous flow and shows the requisite compositional variation identified in viscometric studies. Studies in which the dependence of the critical temperature of the ergodic-to-nonergodic transition is studied as a function of pressure and composition have the potential to shed additional light on the relation between silicate melt structure and dynamics on the atomic scale.

Melting of NaCl and MgO has been simulated with a two-phase molecular dynamics method at constant pressure using newly developed interaction potentials. Equations of state for NaCl and MgO simulated by molecular dynamics are in good agreement with available experimental data. Equations of state for MgO and NaCl were obtained by fitting simulated volumetric properties at pressures and temperatures up to 300 kbar and 3000 K (for NaCl) and 2000 kbar and 7000 K (for MgO). The pressure dependence of the melting temperature was predicted up to 1000 and 1400 kbar for NaCl and MgO, respectively. Crystallization and melting were observed without hysteresis. The simulated melting curve of NaCl is fully consistent with experimental measurements. The pressure dependence of the melting temperature of MgO is consistent with experimental data at 1 bar and previous theoretical estimations by Jackson (1977) and Ohtani (1983). The melting temperature of MgO is substantially higher than that determined by Zerr and Boehler (1994) (by 1000 K at 300 kbar) and substantially lower than that predicted by Cohen and Gong (1994) (by 1500 Kat 300 kbar). The melting temperature of MgO at the pressure of the core-mantle boundary is calculated to be 6900 ± 200 K. The procedures for simulation of melting of NaCl and MgO, starting from a calculation of the interatomic potential and ending with analysis of results, are identical.

Specimens of dense high-pressure hydrous magnesium silicates, notably phase A, phase B, and superhydrous phase B, have been synthesized for use in calorimetric and spectroscopic experiments. X-ray diffraction and electron microprobe analysis were used to study both major and minor phases occurring in the experimental products. We observed the formation of phase A at 16 GPa, 1172 °C; phase B at 15 GPa, 1158 °C; chondrodite at 12 GPa, 1158 °C; and clinohumite at 12 GPa, 1170 °C. These pressures and temperatures are higher than those reported in previous studies. We also observed the formation of phases E and superhydrous B at conditions at which they have been previously observed. However, we propose that there is a high-low transition in superhydrous B. We have not identified phase C in the pressure-temperature-bulk compositional space where it has been observed in previous studies. We conclude that this phase may be identical with superhydrous B, as previously suggested, or that in some studies a humite may have been misidentified as phase C.

Single-crystal X-ray diffraction data have been obtained at several pressures to 5.6 GPa for synthetic LiScSiO 4 olivine. The bulk modulus is 118 ± 1 GPa, assuming K′ = 4. This value is smaller than that of forsterite because compressibility of Li-O bonds in the M1 octahedral site is approximately twice that of the M1 2+ -O bonds in other isomorphs. Compressibilities of a, b, and c orthorhombic axes are 2.70, 2.80, and 2.61 (all × 10 -3 GPa -1 ), respectively. This nearly isotropic compression (axial compression ratios = 1.00: 1.04:0.97) contrasts with that offorsterite (1.00: 1.99:1.55), fayalite (1.00:2.83: 1.22), monticellite (1.00: 1.85:1.10), and chrysoberyl (1.00: 1.30:1.17). These differences arise from the distinctive distribution of cations of different valences and consequent differences in M1-O and M2-O bond compressibilities. The Li M1 octahedron displays a significant decrease in polyhedral distortions with pressure, a behavior not observed in other olivines.

The temperature dependence of the lattice parameters of synthetic lawsonite [Ca- Al 2 Si 2 O 7 (OH) 2 . H 2 O], natural zoisite [Ca 2 Al 3 Si 3 O 12 (OH)], natural clinozoisite [Ca 2 Al 3 Si 3 O 12 (OH)], and synthetic diaspore [AlO(OH)] have been measured at ambient pressure. The volume thermal expansion coefficients for lawsonite, zoisite, and clinozoisite are approximately constant over the measured temperature ranges (25-590 °C for lawsonite, 25-750 °C for zoisite, and 25-900 °C for clinozoisite), whereas the thermal expansion of diaspore increases slightly over the range 25-300 °C. Interestingly, the room-temperature volume of clinozoisite is greater than that of zoisite, but this situation is reversed above -300 °C.The experimental results may be summarized as follows: lawsonite: V/V 0 = 1 + 3.16 (± 0.05) × 10 -5 (T - 298), V 0 = 101.51 (± 0.01) cm 3 /mol; zoisite: V/V 0 = 1 + 3.86 (± 0.05) × 10 -5 (T - 298), Vo = 136.10 (± 0.02) cm 3 /mol; clinozoisite: V/V o = 1 + 2.94 (± 0.05) × 10 -5 (T - 298), V 0 = 136.42 (± 0.05) cm 3 /mol; diaspore: V/V 0 = 1 + 7.96 (± 0.28) × 10 -5 [T - 298 - 20 (√T - √298)], V 0 = 17.74 (± 0.01) cm 3 /mol.

The pressure dependence of the lattice parameters of natural zoisite [Ca 2 Al 3 Si 3 O 12 (OH)], clinozoisite [Ca 2 Al 3 Si 3 O 12 (OH)], and epidote [Ca 2 Al 2 FeSi 3 O 12 (OH)]as well as synthetic lawsonite [CaAl 2 Si 3 O 7 (OH) 2 ·H 2 O] have been measured at ambient temperatures by energy- dispersive X-ray diffraction in a diamond-anvil cell. The experimental results for each phase may be summarized concisely in terms ofthe ambient-temperature isothermal bulk modulus K 298 (using the Murnaghan equation of state and assuming Kʹ = 4): Lawsonite: K 298 = 191 ± 5 GPa; zoisite: K 298 = 279 ± 9 GPa; clinozoisite: K 298 = 154 ± 6 GPa; epidote: K 298 = 162 ± 4 GPa. These new measurements, together with the new thermal expansion data in the companion paper (Pawley et al. 1996), were used to calculate some phase equilibria for lawsonite dehydration to high pressures for comparison with experimental brackets. Important discrepancies between calculated and experimentally determined reactions become evident above 3 GPa.

The enthalpy of ordering of Al and Si between tetrahedral sites in framework structures is now known from several studies to be in the range 0.3-1.4 eV per two Al-O-Si linkages, depending on the structure and type of linkage between tetrahedra. On the basis of this value a simple Bragg-Williams model predicts an ordering phase transition at a temperature, T c , on the order of a few thousand degrees after adjusting the estimate to account for fluctuations in short-range order. However, there are several materials with T c below 1800 °C (a typical melting point), with some having T c so low that it cannot be observed because of the slowness of the ordering kinetics, e.g., gehlenite and leucite. We discuss two mechanisms that can reduce T c substantially: low effective dimensionality and low Al concentration. The cases of sillimanite, gehlenite, cordierite, and leucite, and some simple demonstration systems, are treated quantitatively with the aid of Monte Carlo computer simulations. Other materials (kalsilite, anorthite, albite, and nepheline) are discussed qualitatively in terms of the same principles.

We report a structure refinement of a P2 1 /m ferromagnesian cummingtonite crystal with Fe/(Mg + Fe) = 0.37 at 140 K; this crystal was inverted from the C2/m structure, which is stable at room temperature. The phase-transition temperature is ∼240 K. The most significant structural changes associated with the C2/m-to-P2 1 /m phase transition are as follows: (1) one crystallographically independent O-rotated silicate chain becomes two, with the A chain S rotated and the B chain 0 rotated; (2) the M4 site collapses, with a marked decrease (0.020 Å) in the average M4-06 bond distance; and (3) the largest decrease in the isotropic displacement factor is for the M4 cation. From 295 to 140 K, total changes of 14 and 11° occur in the O5-O6-O5 kinking angles of the A and B chains, respectively. The volume of the M1 octahedron decreases more than that of the other three MO 6 polyhedra, whereas the volume of the M3 octahedron is essentially unchanged. All four MO 6 polyhedra become more regular as a result of the C2/m-to-P2 1 /m transformation. These results provide new evidence that the effective size of the M4 cation controls the relative stability of the C2/m and P2 1 /m structures. The results also provide a new explanation for the unresolved broadening of the cummingtonite Raman spectra measured at low temperatures by Wang et al. (1988).

The high-temperature cell parameters of K 2 MgSi 5 O 12 have been measured by X-ray powder diffraction between room temperature and 842 K. A first-order ferroelastic phase transition between the room-temperature P2 1 /c structure and a high-temperature Pbca structure occurs at 622 K. The spontaneous strain behaves as the order parameter for the transition, following a simple Landau model. Simple lattice dynamics simulation of the magnesiosilicate framework reveals that acoustic mode softening drives the elastic instability. The first-order character of the transition is attributed to a large excess volume, which arises from collapse of the framework around the alkali site. At higher temperatures the orthorhombic cell edges converge toward a metrically cubic structure, but this transformation requires Mg-Si disorder, which was not observed on the time scale of these experiments.

Curie temperatures (T c ) of the (Fe 3 O 4 ) x (MgAl 2 O 4 ) 1-x solid solution have been determined from measurements of magnetic susceptibility (χ) vs. temperature. The trend in T c vs. composition extrapolates to 0 K at x = 0.27. This behavior is rationalized in terms of the trend in cation distribution vs. composition suggested by Nell and Wood (1989), with Fe occurring predominantly on tetrahedral sites for x < 0.27. High-temperature χ-T curves are nonreversible because of the processes of cation ordering and exsolution, which occur in the temperature range 400-650 °C. The Curie temperature of single-phase material is shown to be sensitive to the state of nonconvergent cation order, with a difference in Te of more than 70 °C being observed between a sample quenched from 1400 °C and the same sample after heating to 650 °C. This interaction between magnetic and chemical ordering leads to thermal hysteresis behavior such that Te measured during heating experiments is approximately 10 °C higher than that measured during cooling. The hysteresis is due to a reversible difference in the state of cation order during heating and cooling caused by a kinetic lag in the cation-ordering behavior. Samples with compositions in the range 0.55 < x < 0.7 undergo exsolution to a mixture of ferrimagnetic and paramagnetic phases after heating to 650 °C. Room-temperature hysteresis loops of the starting material and the high-temperature experiment products are compared. All starting materials are multidomain with coercivities H c < 1.26 mT and M rs /M s < 0.051. Samples that exsolved during the experiments have coercivities up to 20 mT and M rs /M s up to 0.36. This change in hysteresis properties is caused by grain subdivision during exsolution and implies a transition in the magnetic domain state from multi- to single-domain.

Two structurally related magnesium phosphates, phosphoellenbergerite (P6 3 mc) and holtedahlite (P6 3 ), have been investigated using a combination of high-frequency IR and 1 H MAS, 31 P MAS, and 31 P CP MAS NMR spectroscopy on synthetic material. For phosphoellenbergerite, the two proton sites, H1 and H2, detected in single-crystal XRD studies, are confirmed. The protons on the first site, H1, delocalized around the threefold axis, form OH groups with the unshared apical O atom of PO 4 tetrahedra. Those on the second site, H2, belong to Mg 2 O 8 OH dimers of face-sharing Mg octahedra, which form double chains. The spectroscopic data reveal additional protons (H3) in phosphoellenbergerite. By combining both crystallographic and spectroscopic data, it can be inferred that the H3 protons are associated with Mg vacancies in the single chains off ace-sharing Mg octahedra. Furthermore, a quantitative inspection of signal intensities indicates that the proton sites in synthetic phosphoellenbergerite are not all saturated. Therefore, the substitution P 5+ = Si 4+ + H + might account for part of the Si-P replacement in the silicate-to-phosphate ellenbergerite series. The results obtained for the proton distribution in holtedahlite, Mg 2 PO 4 OH, were used as a reference to facilitate the interpretation of the IR and NMR spectra of phosphoellenbergerite.

Spectroscopic (micro FTIR, Raman, MAS NMR) and mass spectroscopic techniques have been used to examine, on both single-crystal and powder samples, the behavior of H 2 O and CO 2 molecules in the structural channels of an alkali-poor, volatile-rich beryl from Brazil. Polarized single-crystal FTIR and Raman spectra were obtained on oriented wafers of isolated crystals. Location and orientation of H 2 O and CO 2 molecules were determined from IR spectra. The proton-proton vectors of type-I and type-II H 2 O are parallel and perpendicular to the c axis, respectively; the molecular axis of the CO 2 molecule is perpendicular to the c axis. Relative proportions of both types of H 2 O were determined from their respective IR absorption-band intensities and were found to be nearly equivalent. There is, contrary to what is generally claimed in the literature, no relation between alkali and type-II H 2 O contents. Absorptivity coefficients for H 2 O and CO 2 were computed for every specific orientation. A 1 H MAS NMR spectrum resolved as a “Pake” doublet seems to confirm the absence of any anisotropic movement of the H 2 O molecules from one orientation to the other on the NMR time scale. Mass discrimination of volatiles released by pyrolysis under vacuum confirms the weak mobility of H 2 O and CO 2 molecules in the channel sites, although the plugging effect of alkalis in the channels can be neglected for such an alkali-poor beryl. The volatile vs. alkali content ratio in beryl could possibly be used as an environmental indicator.

Retrograde hydration of anthophyllite found in ultramafic lenses of the Lepontine Alps, southern Switzerland, resulted in the formation of talc and additional ordered and disordered biopyribole polysomes. Among the ordered sequences, 13 new short-range biopyribole polysomes were discovered using high-resolution transmission electron microscopy (HRTEM). The most significant new polysomes are ordered sequences of quadruple-, triple-, and double-chains: (42), (432), (4323), and the first pyribole polysome with a composition between pyroxene and amphibole, (2111). Up to 40 unfaulted repetitions of these sequences are present in several grains. More complex polysomes like (423232) occur as intermediate steps in the formation of shorter polysomes, e.g., the polysome (332). The polysome (42) is the first short-range polysome, for which each step of the formation mechanism is documented. The precursors are disordered sequences countering quadruple- and double-chains, which are formed through coherent zipper transformations from anthophyllite. The disordered sequences are subsequently transformed to the ordered polysome (42). The multiple-step reaction path strongly suggests that energy minimization is partly responsible for the appearance of some of the short-range polysomes. The major element chemistry of the short-range polysomes, determined by analytical electron microscopy (AEM), is identical to that of the macroscopically occurring biopyriboles, excluding, therefore, a stabilization of the short-range polysomes by additional elements. The ordering of (010) stacking faults in enstatite connecting partial dislocations, each with a 1/4 [100] ± 1/3 [001]displacement component, led to the formation of the (2111) polysome. Along the fault plane, pairs of pyroxene I beams are juxtaposed and transform subsequently to amphibole chains. The composition of the (2111) polysome is colinear with the coexisting enstatite and anthophyllite.

Mg 2+ - Fe 2+ substitution in the dodecahedral interstice of eight natural single crystals and two synthetic polycrystalline garnets was investigated by measurement of mid- and farinfrared (IR) reflectance spectra across the pyrope-almandine [Py-Al = (Mg x - Fe 1-x ) 3 Al 2 Si 3 O 12 ] binary. The effect of minor Mn and Ca substitution was investigated by comparison with spectra of four additional natural garnets, Py 8 Al 82 SP 10 and Py 10 Al 76 SP 14 , where Sp denotes spessartine (Mn 3 Al 2 Si 3 O 12 ) and Py 72 Al 18 Gr 10 and Py 54 Al 36 Gr 10 where Gr denotes grossular (Ca 3 Al 2 Si 3 O 12 . Data obtained from the two synthetic polycrystals are in excellent agreement with data from the single crystals. Frequencies of all 17 IR active fundamental modes were observed in all specimens except pyrope. Of the 17 modes, 16 show linear correlations between frequency and composition along the pyrope-almandine join. The remaining mode, assigned to translations of the divalent cations against the O framework, exhibits two-mode behavior. The trends show that two modes in pyrope are accidentally degenerate. Frequencies of garnets with up to 10 mol% spessartine and grossular contents lie on the trends defined by the pyrope-almandine samples with negligible impurities. However, one to three additional modes are seen for some intermediate compositions, apparently resulting from either resonances among combination modes or two-mode behavior due to high Ca contents. The nearly linear behavior of the modes indicates that mixing should be close to ideal for the pyrope-almandine series because the lattice contribution dominates heat capacity, entropy, and compressibility.

We present results of static lattice energy and vibrational energy calculations for the 12 garnet end-members in the system (Ca,Mg,Mn,Fe) 3 (Al,Cr,Fe) 2 Si 3 O 12 . The structure of the end-member phases was first simulated with the aid of expressions of the Novak-Gibbs type followed by a distance least-squares treatment (DLS), with an appropriate choice of the ionic radii of the cations in the crystal structure. The high-P garnet structure was simulated with the assumption that cation-to-O distances obey the generalizations of Hazen and Finger (1979, 1982). Polyhedral compressibilities were modified to account for the P dependency of bulk modulus. The resulting bulk moduli are in satisfactory agreement with experimental observations and are internally consistent. The thermal expansion of the various end-members was derived from linear polyhedral expansivities, in a fashion analogous to that used to determine compressibility, by structural simulation and DLS refinements. Compressibilities of garnet end-members are shown to be consistent with the usual exponential form of short-range pair potentials, with a hardness factor ranging from 0.45 to 0.51 Å and averaging approximately 0.48 Å. Adopting the Huggins-Mayer formulation of repulsive terms (constant hardness factor p = 0.48 for the family of isostructural compounds) and assuming initially repulsive radii to be equal to the crystal radii, it is shown that the preexponential b factors closely obey the Born-Mayer generalization (repulsive factor, b, constant in the same family of compounds). Static energies of the common compounds were then solved with the assumption of full ideality (i.e., p and b constant in the family ofisostructural salts) and for the appropriate repulsive radii. Because the static energies of the 12 garnet end-members are rigorously coplanar in the chemical space of interest, in light of the Born-Haber-Fayans thermochemical cycle, the repulsive energies of the six uncommon end-members (hence the bulk static energy and the corresponding enthalpy at reference state) were readily obtained by application of the combined Huggins- Mayer and Born-Mayer approach. Heat capacities and entropies for all end-members were determined by following the guidelines of the Kieffer model, adjusting the lower cut-off frequency of the optical continuum ω 1,Kmax such that the calorimetric third-law entropy (after correction for anharmonicity and magnetic spin) is reproduced and at the same time conforming to the low-T C p calorimetric data. A complete set of thermodynamic parameters is given for all 12 garnet end-member components.

Use of a thin-film technique and an ion microprobe make it possible to conduct cation self-diffusion experiments of 44 Ca in near end-member grossular garnet at temperatures of 800-1000 °C. The experiments were conducted at 1 atm and under quartz + fayalite + magnetlte f O2 conditions. The resulting activation energy (E a = 155 ± 10 kJ/mol) and the frequency factor (D o = 7.2 × 10 -16 m 2 /s; log D o = -15.1 ± 0.4) were obtained from the temperature dependence of the diffusion data. Comparison of these data with a comparable study of 25 Mg self-diffusion coefficients in pyrope confirmed that Ca diffuses more slowly through garnet than Mg under identical conditions.

The composition-dependence of the solubility of H 2 O in silicate melts along the binary joins NaAlSi 3 O 8 -KAlSi 3 O 8 , NaAlSi 3 O 8 -LiAlSi 3 O 8 , and KAlSi 3 O 8 -LiAlSi 3 O 8 has been determined at 2 kbar and 1040 °C. The study involved 1 atm dry oxide fusion, hydrothermal saturation, isobaric rapid quench and macroscopic analysis for H 2 O using Karl Fischer titration (KFT) and microscopic investigation of homogeneity using infrared absorption spectroscopy, respectively. The solubility of H 2 O in these melts increases in the order KAlSi 3 O 8 (5.12 wt%) < NaAlSi 3 O 8 (6.03 wt%) < LiAlSi 3 O 8 (7.32 wt%). The total relative weight percent variation in solubility is 43%, which corresponds to a relative mole percent variation of 23%. Along the joins, small but systematic deviations from additivity are observed at the 1σ uncertainty level. These deviations are positive on the NaAlSi 3 O 8 - KAlSi 3 O 8 join but negative on the LiAlSi 3 O 8 -bearing joins. The present results confirm the early suggestion of Voigt et al. (1981) that solubility variation along the NaAlSi 3 O 8 -KAlSi 3 O 8 join is nonlinear and disagree in detail with the equimolal approximation of Burnham (1975, 1981) and Burnham and Davis (1974). Models of H 2 O solubility must take into account not only the nonequimolal solubility of H 2 O in feldspathic melts but also nonlinear binary variations as well. We interpret the relative solubilities of the end-member compositions in terms of the relative stability of the tectosilicate melt structure and the nonlinearities along the joins in terms of the next-nearest neighbor distributions of Al and Si in these melts.

Experiments on tonalite and granodiorite were performed at conditions ranging from 2.1 to 18 kbar and 550 to 850 °C to establish the magmatic stability field of epidote as a function of P, T, and f O₂ . At water-saturated conditions and f O₂ buffered by NNO, epidote has a wide magmatic stability field in tonalite. At 13 kbar, this field extends from the wet solidus at 630 to 790 °C. The low-pressure intersection of the magmatic epidote crystallization reaction and the H 2 O-saturated tonalite solidus, and hence the minimum pressure for magmatic epidote, occurs at about 5 kbar. The Clapeyron slopes of epidote melting reactions are moderately positive in P-T space at pressures below the intersection of the epidote melting and the gamet-in reactions at 13 kbar, 790 °C. At this intersection, epidote reaches its maximal thermal stability (790 °C) in tonalite-H 2 O. In the presence of garnet, that is above 14 kbar, epidote melting reactions have steep negative Clapeyron slopes in P-T space. At P-T conditions near the low-pressure intersection of the epidote melting reaction with the water-saturated solidus, experiments were also performed with f O₂ buffered by HM. At these oxidizing conditions the magmatic stability field of epidote is enlarged down to a pressure of about 3 kbar. In granodiorite, the stability of epidote has been investigated at pressures from 7.5 to 18 kbar. With respect to the tonalite, the epidote-out reaction is located almost at the same P-T conditions at 10 kbar, but the maximum thermal stability is about 50 °C lower toward higher pressures. Although few experiments indicate that epidote stability does not increase dramatically with decreasing water activity, at present state it is not possible to predict melting and crystallization behavior of epidote at waterundersaturated conditions. Modal abundances of epidote are low (≤ 5 vol%) in the presence of major (>40 vol%) melt fractions. Thus, epidote in calc-alkaline magmatic rocks is not likely to affect major elements strongly by fractional crystallization.

The decomposition of the iron mica, annite, to sanidine plus magnetite and vapor, KFe 3 AlSi 3 O 10 (OH) 2 = KAlSi 3 O 8 + Fe 3 O 4 + H 2 , has been reexamined experimentally with the use of a variety of bufferscoupledwith f H2 sensorsat 2 kbar and between 400 and 840 °C. Various capsule configurations were used in this study to delineate the equilibrium constant for this reaction in conjunction with 57 Fe Mössbauer spectroscopy measurements to monitor the oxy-annite component in mica in selected experiments. Results at the most reducing and highest temperature conditions of this study extend the annite stability field to higher temperature and fa, values than those defined in previous work. Lower temperature results indicate that the annite-sanidine-magnetite stability boundary does not intersect the hematite + magnetite + H 2 O buffer at 400 °C as previously reported but rather is subparallel to the buffer curve at lower f H2 values. The equilibrium f H2 (in bars) for the assemblage annite + sanidine + magnetite + vapor at 2 kbar and between 400 and 840 °C can be described by the relation log f H2 (± 0.08) = 13.644 - (17368/T) + (5.168 x 10 6 )/T 2 , where T is temperature in kelvins.

Isograds were mapped in siliceous dolomites from the Ballachulish aureole on the basis of the formation of geikielite (MgTiO 3 ), baddeleyite (ZrO 2 ), and qandilite (Mg 2 TiO 4 ) by the following model reactions: rutile + dolomite = geikielite + calcite + CO 2, zircon + 2 dolomite = baddeleyite + forsterite + 2 calcite + 2 CO 2 and geikielite + periclase = qandilite. The (T, X CO2 ) conditions of reaction inferred from (1) mineral equilibria in pelitic rocks, (2) calcite + dolomite thermometry, and (3) the diopside + dolomite + forsterite + calcite + tremolite and dolomite + periclase + calcite equilibria are: geikielite isograd (640-655 °C, 0.76-0.80); baddeleyite isograd (660-710 °C, 0.76-0.95); qandilite isograd (725-755 °C, <0.08). These T-X CO2 conditions for the geikielite and baddeleyite isograds are the same within error of those independently estimated from rutile + dolomite + geikielite + calcite and zircon + dolomite + baddeleyite + forsterite + calcite equilibria using Berman’s thermodynamic data base. Rutile, zircon, geikielite, and baddeleyite are common in dolomites from at least two other contact aureoles in Scotland and Montana. Although the minerals occur in concentrations <0.01 %, they appear to have been in local equilibrium during metamorphism both with each other and with coexisting carbonates and silicates and are potentially useful but currently unexploited records of the physical conditions of metamorphism.

Biotite and amphibole containing up to 4.6 and 5.1 wt% Cl, respectively, are found in gneisses from Black Rock Forest, located within the Hudson Highlands of New York. The area was metamorphosed to granulite facies during the Grenvillian orogeny at ∼ 1100 Ma. Small, subeconomic magnetite deposits coexist with the gneisses. Cl contents in biotite range from 0.1 to 4.6 wt%. Some individual biotite grains are spectacularly zoned, with Cl contents varying from 1.0 to 4.6 wt% within < 20 μm. In zoned biotite, an increase in Cl content correlates with increasing Fe and decreasing Mg, Ti, and F. When data from five specimens are compared, however, Cl contents in biotite show only two strong correlations: Cl contents correlate positively with Fe (r 2 = 0.75) and negatively with Ti (r 2 = 0.84). Cl and Mg are uncorrelated for the complete dataset but show strong correlation within individually zoned grains (r 2 = 0.74). F contents show only one correlation, a positive one with Mg (r 2 = 0.62). The structural formula for the most Cl-rich biotite analyzed is (K 2.00 Na 0.02 )(Mg 1.94 Fe 3.90 Ti 0.07 Al 0.04 )(Si 5.85 Al 2.15 )O 20 (OH 2.46 F 2.46 Cl 1.29 ), as normalized to 24 (O + OH + F + Cl). This biotite is one of the most Cl rich ever reported. Cl contents in amphibole range from 0.2 to 5.1 wt%. Cl contents correlate positively with Fe 2+ /(Fe 2+ + Mg) (r2 = 0.84) and K (r 2 = 0.70) and negatively with Si/(Si + [4] Al) (r 2 = 0.49) and Ti (r 2 = 0.76). F contents are uncorrelated with any of the other elements analyzed. The structural formula for the most Cl-rich amphibole, a hastingsite, is (K 0.57 Na 0.31 )(Ca 1.90 Na 0.10 )(Fe 2.69 2+ Mg 0.90 Mn 0.03 Ti 0.06 Fe 1.02 3+ Al 0.29 )(Si 5.80 Al 2.20 )O 22 (OH 0.55 F 0.07 Cl 1.38 ), as normalized to 13 octahedral and tetrahedral cations. In rocks with coexisting amphibole and biotite, Cl partitions preferentially into the amphibole and F partitions preferentially into the biotite. Infiltration of the gneisses by fluids rich in Fe and Cl (perhaps as an FeCl 2 compound) could explain the formation of magnetite bodies as well as the Fe- and Cl-enriched phases found in and around Black Rock Forest.

Medenbachite occurs in association with mixite, preisingerite, alunite, goethite, reichenbachite, and malachite in a silicified barite vein from Reichenbach near Bensheim, Odenwald, Hesse, Germany. It forms aggregates of small tabular crystals. No cleavage was observed; fracture is conchoidal and Mohs hardness is 4.5. The color varies from yellow to brown-yellow; the luster is vitreous to adamantine. Medenbachite is biaxial negative, 2 V = 43(3)°, dispersion r > v, α = 2.03(2), β = 2.09 (calc.), and γ = 2.10(2)°. Observed crystal forms are {001}, {1̅01}, {101}, {1̅10}, {01̅1}, and {131̅}. Medenbachite is triclinic, space group P1̅, with a = 4.570(1), b = 6.162(1), c = 8.993(1) Å, α = 94.56(1), β = 99.69(1), γ = 94.28(1)°, V = 247.9 Å 3 , Z = 1, D calc = 5.90 g/cm 3 . The six strongest lines of the powder diffraction pattern are [d(Å),(I/I o ),(hkl)]: 8.823(62)(001), 3.749(100)(01̅2,101), 3.596 (77)(1̅11), 3.468(58)(110, 111̅), 2.903(69)(102), and 2.810(51)(021). Electron microprobe analyses gave Bi 2 O 3 53.36, Fe 2 O 3 8.99, FeO 1.00 (calc.), CuO 7.85, As 2 O 5 25.32, H 2 O (calc.) 3.07%, sum 99.59%. Microchemical tests indicate the presence of both Fe 3+ and Fe 2+ ions. On the basis of crystal-chemical considerations, the Fe 3+ -Fe 2+ distribution was calculated from the total iron content to achieve equal numbers of Fe 3+ and (Cu 2+ + Fe 2+ ) ions, for two crystallographically different positions. This yields the empirical formula Bi 2.04 Fe 1.00 3+ Cu 0.88 Fe 0.12 2+ ) Σ1.00 O 1.11 (OH) 3.03 (AsO 4 ) 1.96 ,or ideally Bi 2 Fe 3+ (Cu,Fe 2+ )(O,OH) 2 (OH) 2 - (AsO 4 ) 2 . The structure refinement converged at R = 0.079 for 1124 unique reflections. Edge-sharing octahedra with an ordered occupation of alternating Fe 3+ and (Cu 2+ ,Fe 2+ ) ions form columns parallel to [010], which are linked parallel to (001) by arsenate tetrahedra. The Bi 3+ is probably distributed over two sites, which are 0.463(5) Å apart. Refinement of the site-occupancy factors indicates 50% probability for the location of Bi 3+ on each position.