Volume 80, Issue 7-8

The hydroxyl end-member ofhumite, Mg 7 Si 3 O 12 (OH) 2 , was synthesized at 30 kbar and 940°C in the system MgO-SiO 2 -H 2 O (MSH). Optical properties are n x = 1.6348(10), n y = 1.6397(10), n z = 1.6612(10), 2V z = 56.1(2)°, -Δ n = 0.026. Results of single-crystal X-ray diffraction gave a = 4.78(2), b = 10.22(3), c = 21.02(9) Å, V = 1027(7) Å 3 , space group = Pbnm. The morphology of the synthetic humite crystal is comparable to that of natural F-rich humite. Microprobe analyses confirmed a nearly ideal formula, Mg 6.97(3) Si 3.02(1) - O 12 (OH) 2.00(2) .

It is often assumed that fluid inclusions in quartz do not maintain their density if they experience internal pressures 2 kbar in excess of confining pressure. However, results of a long-term (180 d) laboratory experiment simulating a nearly isothermal decompression PT path showed that two of 160 inclusions recorded the original density. These two inclusions thus maintained about 2 kbar internal overpressure for the duration of the experiment without modification. Comparison with results reported in the literature reveals that a small percentage of fluid inclusions in quartz from high-grade metamorphic rocks maintain peak densities during retrograde uplift. These inclusions represent outliers in homogenization temperature histograms and are often ignored during data interpretation.

The pressure dependence of the anorthite Raman spectrum has been investigated to 11.4 GPa at room temperature using a single crystal in a diamond-anvil cell. We observe a first-order phase transition on increasing pressure at 2.6 GPa, which is consistent with the p1̅-I1̅ transition reported in previous X-ray diffraction experiments. Changes in the Raman spectrum also reveal an additional displacive reversible phase transition to a new phase at 10 GPa. Still not fully characterized, the new phase contains much shorter T-O bond lengths, indicating a major change in the compression mechanism above 10 GPa.

Natural cummingtonite encompasses virtually the entire compositional range from Mg to Fe end-member. It occurs in several metamorphic rock types, but predominantly in amphibolites and metaironstones. Most examples were produced under amphibolite-facies conditions, although the full range for cummingtonite is at least 400-800 °C and < 1-15 kbar. Cummingtonite is also a critical indicator of T, a H₂O ,and f O₂ in silicic volcanic rocks from shallow magma chambers. The solution properties of cummingtonite reflect the entropic and enthalpic consequences of temperature-dependent, long-range ordering of Mg and Fe 2+ on the M1 + M3, M2, and M4 sites, plus modest site-mismatch energies. Solutions show small positive departure from ideal at 700-800 °C and both positive and negative behavior at lower temperatures depending on composition. The quenched state of M-site ordering has an influence on the symmetry of the unit cell (C2/m vs. P2 1 /m) and on the optical indicatrix. The kinetics of diffusion of Fe 2+ and Mg among sites appears to be site dependent. Magnesio-cummingtonite from 0 to about 10% grunerite is less stable than magnesian anthophyllite at metamorphic temperatures of 600-700 °C. The inversion loop in the system FMSH, with anthophyllite as the low-temperature form, probably has a minimum temperature close to the Fe end. The paragenesis of cummingtonite with olivine and quartz (found in metamorphosed iron formation) is confined to pressures below 10 kbar, temperatures below 740 °C, and ƒ O₂ no more than 0.5 log unit above QFM. The paragenesis of cummingtonite with orthopyroxene and quartz typifies more Mg-rich compositions and generally signifies higher temperatures and in many instances higher pressures, and a temperature maximum (extremum) exists in the isobaric T-X Fe diagram. The composition of cummingtonite together with magnetite, quartz, and H 2 O fluid is a sensitive indicator of ƒ O₂ . Cummingtonite must be Mg rich to coexist with hematite. Temperature and ƒ O₂ determined from iron titanium oxide phenocrysts and measured cummingtonite compositions in rhyolites from the Taupo Zone, New Zealand, agree well with predicted relations. The agreement is less good for dacites from Saint Helens and Pinatubo. With increasing pressure, reactions between cummingtonite and the components of feldspar produce biotite and hornblende and restrict the occurrence of cummingtonite in amphibolites and systems of granitic composition.

Maghemite in a tuffite deposit from Patos de Minas, Brazil, has been isolated by highgradient magnetic separation and analyzed by X-ray diffraction, electron microprobe analysis, Mössbauer spectroscopy, and magnetization measurements. The maghemite constitutes 5 wt% of the tuffite, where it is present as separate grains in the 1-400-μm-size range, some with associated anatase inclusions or intergrowths. The lattice parameter is a 0 = 0.8380(2) nm. No Fe 2+ is detected, but the composition of the grains shows Fe ≫ Mg ≥ Ti. The averaged composition and proposed cation distribution is [Fe 0.88 Si 0.01 Mg 0.11 ]- {Fe 0.96 Mg 0.30 Ti 0.32 Al 0.07 Cr 0.03 Mn 0.02 ⃞ 0.30 }O 4 , where [ ] and { } denote A and B sites, respectively, of the spinel lattice, and ⃞ denotes cation vacancies. Magnetization, σ S , is 18-31 J/(T · kg) at room temperature, and the Curie temperature, T c , is 320-360 °C. The Mg-rich maghemite is inherited by magnetic soils forming on the tuffite.

Neutral atomic radii of dissolved noble gas elements in condensed phases are obtained by treating the neutral atoms as “ions” of zero oxidation state and by interpolation from a plot of radius vs. oxidation state for isoelectronic ions. The major assumption is that the radius of an ion or a neutral atom having an electronic configuration of a noble gas element depends primarily on the interaction between the nucleus and the surrounding electrons and not on the interaction between the ion and its neighbors. As expected, the values of the new set of radii are slightly greater than the univalent radii and smaller than the radii of noble gas atoms in their crystals (in which the coordination number is 12). The neutral atomic radii (in ångströms) of noble gas elements are where CN = coordination number. Errors are given at the 2σ level. Values in parentheses were determined by extrapolation and have greater errors. The larger error for the He radius is due to large uncertainty in the H - radius.

The high-pressure phases denoted ANH-B, hydrous B, and SHY-B (Finger et al., 1991; Pacalo and Parise, 1992), containing both [4] Si and [6] Si, are members of a homologous (polysomatic) series in which the [4] Si-containing portions are composed of olivine-like (O) slabs alternating with thinner goethite-like (G) slabs in the ordered arrangements O ∞ , OOOG, and OG. Leucophoenicite, OOOG, containing the same kind of slabs, represents the only macroscopic member of the leucophoenicite series (White and Hyde, 1983). It has the same slab sequence as the hydrous B phase; lamellar micromembers were found with O 4-6 G. The principal difference between the two series is that the leucophoenicite series lacks the purely octahedral layers with [6] Si that are intercalated with the [4] Si-containing ones in the B phases. These relationships allow one to predict further members for both series. They also suggest possible homologues that are intermediate between the two series, e.g., between leucophoenicite and hydrous B and between the hypothetical OG leucophoenicite and SHY-B, and contain a smaller proportion of [6] Si-based layers than these B phases. Mn might play a stabilizing role in the B series. Hypothetical B phases based on O slabs alternating with norbergite-like (N) slabs are much richer in Mg and have lower H 2 O-storing capacity.

The atomic displacement parameters (ADPs) obtained from published refinements of silicate crystals measured at or below room temperature are examined to determine if TO 4 (T = Al,Si) tetrahedra display rigid-body thermal motion. In many cases, the correlation found among the ADPs is consistent with the TLS model of rigid-body motion. For these data, the translational motion is described by the ADPs of the central T atom, whereas both librational and translational motion are displayed in those of the surrounding 0 atoms. The libration angle for rigid tetrahedra is quadratically dependent on the difference between the isotropic equivalent displacement parameter of the T and 0 atoms, B eq (T) and B eq (O), respectively. The value of B eq (O) is on average twice that of B eq (T), with an observed maximum value of ~ 2.0 Å 2 . Variation in the observed Si-O bond lengths of rigid tetrahedra in the silica polymorphs is related only to the fractional s-character of O, f S (O). ADPs that do not indicate rigid-body motion for a given tetrahedron may be used to identify crystals containing disorder or suggest problems with the refinement.

The 29 Si MAS NMR spectra of nine pyrope-grossular garnet solid solutions were recorded along with their mid and far infrared spectra. The 29 Si MAS NMR spectra of pyrope and grossular measured previously show one single resonance at -72.0 and -83.9 ppm, respectively. A maximum of up to 13 resonances were observed in composition Py 50 Gr 50 indicating that 29 Si MAS NMR spectroscopy can record both first and second shell dodecahedral interactions in garnet. Complete resonance assignments were done by qualitative crystal chemical considerations combined with the comparison of fitted peak intensities with the calculated probabilities for the various possible localized chemical environments around the Si tetrahedra. The NMR spectra of garnets with compositions Py 90 Gr 10 , Py 10 Gr 90 ,and Py 05 Gr 95 are consistent with a random distribution of the divalent cations Ca and Mg on the triangular dodecahedral sites. Garnets with compositions between 0.15 ≤ X Ca ≤ 0.75 show evidence for deviations from random mixing. No overt evidence is found to support a model of local cation order in the most pyrope-rich compositions, which had been proposed on the basis of the asymmetric volumes of mixing along the binary. Powder IR spectra show no additional bands in the solid solutions compared to the end-members, and hence all solid solution compositions are consistent with space group Ia3̅d.

The Si-Al ordering in two contrasting samples of leucite was studied using MAS NMR. One sample was synthesized by performing an ion exchange, in molten KCl, on a natural analcite specimen that had a small distribution of Al next-nearest neighbors (NNN). The other leucite sample was a typical natural specimen from the Alban Hills. In the specimen derived by ion exchange, 45-50% of the Al is on T2 sites with the remainder distributed approximately equally between the T1 and T3 sites. The 29 Si peak positions in this sample resolve a debate in the literature over assignment of the more complex 29 Si spectra of natural leucite specimens. It is possible to simulate the 29 Si spectrum of Alban Hills leucite with a wide variety of Al distributions, including either Al ordering onto T2 or Al ordering onto T1 + T3. By analogy with the analcite-derived sample and considering previous neutron and X-ray data, Al ordering onto T2 is considered somewhat more probable than the alternative schemes. Site-ordering information obtained from 27 Al NMR in an earlier study is not considered reliable because of complications arising from the differing (and uncertain) electric-field gradients at these sites.

Practical methods for identification of the complex mica polytypes have been developed by introducing a special function characteristic of their layer stacking sequences that is displayed in their X-ray diffraction patterns. This periodic intensity distribution (PID) function is the Fourier transform of a stacking sequence. The X-ray diffraction patterns of a polytype are expressed by the PID function modified by the Fourier transform of the unit mica layer. The PID function is used to determine the stacking sequences of several polytypes, and the mechanism by which the most frequently observed polytypes form is discussed. The axial settings of mica polytypes are defined in order to compute the PID functions in terms of the layer stacking sequences. Practical methods of obtaining observed PID functions are given together with tables of the PID functions of the three basic polytype series. The common polytypes have the basic sequences such as 1M, 3T, and 2M 1 and are modified by a stacking fault, but the subsequent sequences are so arranged that the original direction of stacking is recovered as readily as possible.

The electrochemical oxidation of chalcopyrite (CuFeS 2 ) and its metal-enriched derivatives, haycockite (Cu 4 Fe 5 S 8 ), mooihoekite (Cu 9 Fe 9 S 16 ), and talnakhite (Cu 9 Fe 8 S 16 ), have been investigated using electrochemical techniques and X-ray photoelectron spectroscopy. The results show that when chalcopyrite, haycockite, mooihoekite, and talnakhite are oxidized in 1 M HClO 4 in the electrode potential region of 0.2-0.6 V vs. S.C.E. at 298 K, the oxidation products are Cu x S* z , Fe 2+ and, at electrode potentials > 0.53 V vs. S.C.E., Fe 3+ ions; Cu x S* z represents a metastable phase at the electrode surface. The relative rates of oxidation are in the order haycockite > mooihoekite > talnakhite > chalcopyrite. After oxidation of Fe surface sites, the rate-determining step apparently involves transport of Fe ions from the bulk to the solid-electrolyte interface. When chalcopyrite, haycockite, mooihoekite, and talnakhite are oxidized in 0.1 M, borax in the electrode potential range of -0.6-0.3 V vs. S.C.E. at 298 K, the primary oxidation products are Cu x S* z and Fe 2 O 3 . The initial oxidation rates of haycockite, mooihoekite, and talnakhite are higher than chalcopyrite, but this initially more facile oxidation of haycockite, mooihoekite, and talnakhite results in metal oxide (mainly iron oxide) surface films, which retard further oxidation to a greater extent than on chalcopyrite. When chalcopyrite, haycockite, mooihoekite, and talnakhite are oxidized in 0.3 MNaCl + HCl solution of pH 0.1, the oxidation rates remain essentially in the order haycockite > mooihoekite > talnakhite > chalcopyrite. However, probably because chloride ions form complexes with Cu + ions and may act as a bridge in electron transfer, the oxidation rate is significantly enhanced compared to that in 1 M perchloric acid and 0.1 M borax aqueous solutions.

Primary alteration of uranpyrochlore from granitic pegmatites is characterized by the substitutions A ⃞ Y ⃞→ A Ca Y O → A Ca Y O, A Na Y F → A Ca Y O, and A Na Y OH → A Ca Y O. Alteration occurred at ~450-650 °C and 2-4 kbar with fluid-phase compositions characterized by relatively low a Na +, high a Ca²⁺ , and high pH. In contrast, primary alteration of pyrochlore from nepheline syenites and carbonatites follows a different trend represented by the substitutions A Na Y F → A ⃞ Y ⃞ and A Ca Y O and A Ca Y O → A ⃞ Y ⃞ . In carbonatites, primary alteration of pyrochlore probably took place during and after replacement of diopside + forsterite + calcite by tremolite + dolomite ± ankerite at ~300-550 °C and 0-2 kbar under conditions of relatively low a HF , low a Na⁺ ,low a Ca²⁺ , low pH, and elevated activities of Fe and Sr. Microscopic observations suggest that some altered pyrochlores are transitional between primary and secondary alteration. Alteration paths for these specimens scatter around the trend A Na Y F → A ⃞ Y ⃞. Alteration probably occurred at 200-350 °C in the presence of a fluid phase similar in composition to the fluid present during primary alteration but with elevated activities of Ba and REEs. Mineral reactions in the system Na-Ca-Fe-Nb-O-H indicate that replacement of pyrochlore by fersmite and columbite occurred at similar conditions with fluid conpositions having relatively low a Na⁺ ,moderate a Ca²⁺ , and moderate to high a Fe²⁺ . Secondary alteration (< 150 °C) is charactlerized by the substitutions A Na Y F→ A ⃞ Y ⃞, A Ca Y O → A ⃞ Y ⃞, and A Ca X O → A ⃞ X ⃞ together with moderate to extreme hydration (10-15 wt% H 2 O or 2-3 molecules per formula unit). Minor variations in the amounts of Mg, Al, K, Mn, Fe, Sr, Ba, and REEs are commonly observed as a result of secondary alteration. Major cation exchange for K, Sr, and Ba is a feature of samples from laterite horizons overlying carbonatites. In most cases U, Th, and B-site cations remain relatively constant. Radiogenic Pb is typically lost via long-term diffusion, but in some grains of uranpyrochlore 25-90% of the Pb is lost as a result of alteration.

Equilibrium compositions of coexisting Ca-rich carbonate scapolite and plagioclase in the presence of calcite have been investigated experimentally at 7.0 kbar in the temperature range 775-850 °C. The CaO-Al 2 O 3 -SiO 2 -CO 2 univariant reaction is with NaSi(CaAl) -1 substitution in both plagioclase and scapolite in the Na 2 O-CaO-Al 2 O 3 -SiO 2 -CO 2 divariant assemblage. The experiments were carried out in a 3/4 in. diameter NaCl pressure cell with a hematite + magnetite buffer and a hydrous oxalic acid flux. Starting materials were a natural Cl and S-free mizzonite (Me 63 ), Baker reagent precipitated calcite, and synthetic meionite and plagioclases of composition An 100 and An 92 . Changes of compositions of the reactants were determined by microprobe analysis. The experimental results confirm the inferences based on observations in natural assemblages made by earlier workers that Na substitution greatly stabilizes scapolite relative to plagioclase. The marked stabilization of mizzonite to lower temperatures is largely a consequence of the large amounts of atomic mixing possible in the scapolite structure. An activity model for the meionite component in terms of its mole fraction, X Me , which involves complete disordering of Ca and Na and, independently, Si and Al, has been derived from the experimental brackets. This formula may lose validity in a lower temperature range if scapolite develops greater atomic ordering near the mizzonite composition. The disordered scapolite model is used to calculate apparent CO 2 activities for recrystallization of some Grenville granulites studied by Moecher and Essene (1991). We find close agreement with their predicted C02 activities based on scapolite + garnet + plagioclase + quartz equilibrium at 800 °C and 10 kbar, despite the different scapolite activity models used in the two sets of calculations. The simplest explanation of this coincidence is that their empirical scapolite model and the present experimentally based model are both valid, although the latter is restricted to high temperatures (above 750°C).

We have determined the phase relationships of melting of synthetic granite (two ternary feldspars + quartz) in the presence of an H 2 O-CO 2 fluid. Synthesis and reversed experiments were conducted in a piston-cylinder apparatus over the range 650-900 °C and 6-15 kbar. At X H₂O = 0.75, melting occurred between 670 and 680 °C (15 kbar), 700 and 710 °C (10 kbar), and 710 and 720 °C (7.4 kbar). At X H₂O = 0.5, melting occurred between 760 and 770 °C (15 kbar), 780 and 790 °C (10 kbar), and 800 and 820 °C (6-7.4 kbar). At X H₂O = 0.25, melting occurred between 830 and 840 °C (15 kbar), 830 and 840 °C (10 kbar), and 860 and 870 °C (6-7.4 kbar). These results provide important constraints on the maximum temperatures of regional metamorphism attainable in vapor-saturated metapelitic and quartzofeldspathic rocks that escaped widespread melting. At pressures below 10 kbar, a fluid phase of X H₂O = 0.75, 0.5, and 0.25 limits temperatures to below ~700-725, ~800-825, and ~850-875 °C, respectively. As a consequence, the formation of granulite does not require CO 2 concentrations in a coexisting fluid to exceed an X CO₂ of 0.25-0.5, a range that may include dilution of the H 2 O component of the fluid through internal buffering by devolatilization reactions. Therefore, the formation of granulites by the influx of CO 2 may be a less common phenomenon than previously thought.

The distribution of O isotope ratios in a slowly cooled metamorphic rock is investigated by a combination of ion microprobe and laser probe techniques. The sample has unusually simple geometry: centimeter-size domains of feldspar and garnet separated by a one-millimeter to one-centimeter thick layer of millimeter-size magnetite grains. Isotopic zonation at two different scales is documented. A sharp decrease in δ 18 O in the outer 10-100 μm of individual magnetite grains is in agreement with modeling of isotopic reequilibration during cooling by interdiffusion between feldspar and magnetite. The presence of this zonation in the outer portions of magnetite grains that are adjacent to garnet indicates extensive transport along grain boundaries during diffusive exchange. Millimeter-scale zonation in the polygranular layer of magnetite is caused by the interaction of volume diffusion and grain boundary transport, the effects of which depend upon local textures. Grain boundaries enhance bulk diffusion in polygranular aggregates and allow exchange between nontouching grains. The results document a complex textural control to millimeter-scale O isotope zonation and geothermometry in slowly cooled rocks and document the importance of fast grain boundary transport in controlling diffusive exchange.

The relative solubilities of monazite (Mnz), xenotime (Xno), and apatite (Ap: REE- and Sr-rich, and REE- and Sr-poor) have been studied in peraluminous granitic liquids in month-long experiments at 750 °C and 200 MPa (P H₂O ).In contrast with the high solubility of the apatite (0.7 wt%P 2 O 5 ) in strongly peraluminous liquids, monazite and xenotime have much lower solubilities (< 0.05 wt% P 2 O 5 ). In mildly to strongly peraluminous compositions, P 2 O 5 in the liquid is 0.03-0.04 wt% at xenotime saturation and 0.02-0.05 wt% at monazite saturation; in keeping with the low P in the glasses, RE 2 O 3 contents are below EMP detection thresholds (≤ 0.08 wt%) for all conditions of experiments (saturation of liquid at equilibrium and local saturation around apatite). Apatite dissolves incongruently, crystallizing REE-rich monazite on its surface (1-4 μm-long grains), resulting in similar low REE contents in liquids. Monazite precipitation occurs along the margins of dissolving apatite crystals, even though the bulk liquid is not monazite-saturated. The abundance of monazite microcrystals increases with the REE content of the apatite and the degree of apatite dissolution. The reaction relationship (Ap + Liq 1 → Mnz + Liq 2 ), stemming from differences in relative solubilities (greater than an order of magnitude) between apatite and monazite, results in the dissolution of much smaller amounts of REE into peraluminous liquids than expected by simple evaluation of apatite REE contents. The amount of REE contributed from apatite directly to peraluminous granitic liquid is related to the amount of apatite dissolved by simple mass balance only if the total REE content of the apatite is sufficiently low that monazite saturation in liquid (50-100 ppm RE 2 O 3 at 750 °C) is avoided. During dissolution of Sr-rich apatite, the Sr partitions into the liquid, and, at 750 °C and 200 MPa (P H₂O ),the diffusion coefficient of Sr in liquid is ~ 2 × 10 -10 cm 2 /s (R 2 = 0.659). The reaction relationships described above may have application to some textural features observed in natural igneous rocks. For example, clusters of monazite microcrystals might be indicators of dissolved apatite; monazite morphology can be used to distinguish the source of its REE and the general petrological process (rock anatexis or magma crystallization) under which the monazite formed. In addition, monazite microcrystals could serve as nucleation sites for other minerals, which might explain their common inclusion in biotite and amphibole within granitoids.

Concentration profiles for Ca, Na, Al, K, Fe, Ti, Mg, Sr, and Ba obtained by electron microprobe and secondary ion mass spectrometry from plagioclase crystals, together with textural observations from interference contrast microscopy, are consistent with contrasting magma dynamics in two subvolcanic reservoirs from which silicic lavas erupted at the Tatara-San Pedro volcanic complex, Chilean Andes. The 1 km 3 late Pleistocene (68 ka) Tatara dacite is chemically homogeneous, phenocryst-poor, and contains crystals of opacitized hornblende, orthopyroxene, and titanomagnetite, plus 2 mm euhedral plagioclase phenocrysts with simple zoning patterns. Except at their rims, analyzed phenocrysts show nearly monotonic core-to-rim changes from An 52 to An 31 , including slight decreases in MgO and FeO' and slight increases in Ba, Sr, and K concentrations. Abrupt, but small, chemical shifts are associated with many subtle dissolution surfaces. In contrast, the 0.1 km 3 of chemically and texturally heterogeneous Holocene San Pedro dacite contains crystals of clinopyroxene, orthopyroxene, hornblende, biotite, titanomagnetite, ilmenite, and abundant 2 mm euhedral plagioclase phenocrysts, plus plagioclase and olivine xenocrysts derived from fragmented quenched basaltic inclusions. Analyzed phenocrysts span the range from An 72 to An 32 . Abrupt increases of 15 mol% An, large increases in MgO, FeO', and Sr, and decreases in Ba and K 2 O concentrations are associated with a few major dissolution surfaces. Plagioclase xenocrysts of An 74-86 have higher Sr, MgO, and FeO' and lower Ba and K concentrations than the phenocrysts, reflecting crystallization from a basaltic melt. In both dacites, phenocryst rims decrease in Sr and increase in K and Ti relatiye to equilibrium values as a consequence of rapid crystal growth at high undercooling during magma ascent and eruption. Modeling of variations in apparent D plag/melt Sr implies growth rates for the ≤ 100 μm phenocryst rims of 10 -9 cm/s, suggesting that magma ascent may have taken several months. Provided these kinetic effects are identified, the remainder of the concentration profiles for Sr, Ba, Mg, and K in plagioclase phenocrysts can be inverted using partition coefficient expressions to monitor the temporal evolution of melts in the preeruptive magma chambers. Monotonic melt trends and periodic thermal dissolution of Tatara dacite phenocrysts imply a magma chamber closed to inputs of new magma and heat. Repeated cycles of crystal growth and weak dissolution occurred during phenocryst retention in thermally driven convection cells characterized by steady-state laminar flow. In contrast, dissolution events associated with large, abrupt shifts in melt Sr, Mg, and K/Ba toward basaltic values indicate that in the Holocene magma chamber heating of silicic melt adjacent to basaltic inclusions during magma mingling caused plagioclase dissolution. Renewed plagioclase growth from the resulting small volumes of hybrid melt recorded highly localized chemical mixing by diffusion and accompanying cycles of fractionation.

Data are presented for the O isotopic composition of clinopyroxene and garnet, the C isotopic composition of diamond, and the OH - content of garnet from eclogite xenoliths brought to the surface by the Udachnaya kimberlite pipe, Yakutia, Siberia. Radiogenic isotopic data suggest that the eclogites could have been derived from an ultradepleted mantle at approximately 2.9 Ga (Pearson et al., 1995; Snyder et al., in preparation). O isotopic compositions generally show equilibration between the eclogitic minerals (Δ Cpx-Grt = 0.11-0.41‰) and have δ 18 O SMOW for both garnet and clinopyroxene that lie near the range of accepted mantle values of 5.7 ± 0.7‰. However, several eclogites indicate significant deviations from this range, at higher values of 6.8- 7.0‰. Also, two eclogites lie at the lower end of the mantle range, at values of 4.8 and 5.0‰ (all in clinopyroxene). C isotopic compositions of diamonds all have δ 13 C PDB in the range of -1 to -7‰ and are centered at approximately - 5‰, also within the range of accepted mantle values. The OH - contents of the garnet are generally between 0 and 22 ppm (as H 2 O), although two samples exceed 70 ppm. This range of OH - is similar to eclogitic garnet from the Kaapvaal craton of southern Africa. The mantle C isotopic values of associated diamonds, the majority of O isotopic data, and the low OH - content of the minerals, although suggesting a general lack of crustal participation in the formation of the Udachnaya eclogites, do not rule out the participation of some ancient crustal material. That these eclogites include both 18 O-enriched and 18 O-depleted types suggests that the protoliths may have been overprinted by both low- and high-temperature hydrothermal events (cf. Jacob et al., 1994). A positive correlation between δ 18 O and 87 Sr/ 86 Sr allows the interpretation of these eclogites as representing a cross section of an Archean ophiolite. However, the lack of a single coherent grouping on a plot of δ 18 O and 87 Sr/ 86 Sr suggests that, if the Udachnaya eclogites were derived from oceanic crust, they cannot be cogenetic and must represent at least two separate ophiolite sequences. Conversely, if the eclogites are found to be cogenetic, then a totally different process affected their formation and a probable metasomatic mechanism was operative. Because of the total lack of correlation of δ 18 O with other geochemical parameters, we find no compelling evidence that all eclogites are derived ultimately from oceanic crust.

Aluminous armalcolite has been found in two sillimanite-bearing xenoliths that were recently exhumed from the lower crust of central Mexico. The ranges of compositions are (Fe 2+ 0.58-0.66 Mg 0.18-0.28 Al 0.15-0.18 V 3+ 0.06-0.10 Fe 3+ 0.00-0.06 Ti 4+ 1.84-1.86 )O 5 and (Fe 2+ 0.20-0.51 Mg 0.18-0.29 Al 0.16-0.19 -V 3+ 0.02-0.06 Fe 3+ 0.30-0.81 Ti 4+ 1.49-1.71 )O 5 . The occurrence of armalcolite is unusual in crustal paragneisses because most terrestrial armalcolite occurs in volcanic rocks that are derived from partial melting in the Earth's mantle. Textures suggest that armalcolite is a late product formed by the reaction rutile + ilmenite ss = armalcolite ss during rapid transport of the xenoliths to the surface. Phase equilibria in the system MgO-FeO-Fe 2 O 3 - TiO 2 , which indicate that armalcolite is stable in the crust at 900-1200 °C, are consistent with this interpretation. Thermodynamic properties are estimated for oxides in the system MgO-FeO-Fe 2 O 3 - TiO 2 to constrain activity-composition relations for armalcolite and conditions of formation. Activity coefficients calculated for armalcolite range from 0.27 to 1.36, depending on the ilmenite model used, at temperatures between 1000 and 1300 °C. Depth of formation of armalcolite in the crust is not well constrained. Thermodynamic calculations at 800-1200°C for the compositions observed indicate that the armalcolite in one xenolith would have been in equilibrium with rutile at values of f O₂ between the hematite + magnetite buffer (HM) and the fayalite + magnetite + quartz (FMQ) buffer, and that armalcolite in the other xenolith would have been in equilibrium with rutile and ilmenite at values of f O₂ between FMQ and two log units below the FMQ buffer.

Some isolated forsterite grains in Allende (C3V) and ALHA 76004 (LL3) show oscillatory zoning when viewed with cathodoluminescence. Concentration profiles for these grains show that the oscillations correspond to cyclic variations in Al and Ti and, by inference with known correlations, V and Sc. Other elements, including Mg, Fe, Ca, and Cr, show only monotonic changes across the oscillatory-zoned area. Under reducing conditions assumed present during forsterite growth, the varying elements should be trivalent or higher in charge, whereas the Mg-Fe-Ca-Cr group would be divalent. Substitution into the olivine structure is considered to be two trivalent elements, dominated by Al, for Mg + Si, which represents a small substitution of the spinel component into forsterite. This same type of substitution is proposed for similar terrestrial oscillatory-zoned olivines in which no Fe or Mg variation is found. By analogy with terrestrial and lunar oscillatory-zoned minerals, crystallization of the forsterite from a liquid is probable. Currently, no such source material is known, although crystallization of porphyritic chondrules may produce grains of equivalent size with zoning. Although no oscillatory-zoned forsterite has been observed in chondrules, their recognition may be confirmation of the source of isolated forsterite grains; alternatively, the source of these forsterite grains has not yet been recognized.

Micas with variable Ba content in the interlayer sites occur in silicate-rich bands within the high-grade, metamorphic banded iron formation enclosing massive sulfide bodies in the Broken Hill deposit, Namaqualand Metamorphic Complex, South Africa. In those layers that contain neither potassium feldspar nor muscovite, a green, barian (up to 17 wt% BaO), trioctahedral mica occurs that is similar to kinoshitalite. In contrast to the originally described kinoshitalite, however, this mica is Fe-rich (average X Fe = 0.65), suggesting the existence of an Fe analogue of kinoshitalite not described to date. Ion chromatographic analysis indicates that about 8% of the total Fe in this mica is present as Fe 3+ , and a typical formula is (Ba 1.2 K 0.4 )(Fe 2+ 3.2 Mn 0.2 Mg 2.0 Al 0.2 Fe 3+ 0.1 Ti 0.2 )(Al 3.0 Si 5.0 )O 20 (OH 1.8 F 2.2 ). The unit-cell parameters were determined as follows: a = 5.383(2), b = 9.328(8), c = 10.055(8) Å, β = 100.44(5)°, with V = 496.5 Å 3 , and Z = 2. The space group is C2/m. The Ba-rich mica formed at the peak of metamorphism (T = 670 ± 20°C, P = 4.5 ± 1.0 kbar) at a pH below the muscovite + potassium feldspar buffer, f O₂ buffered by quartz + fayalite + magnetite and f s₂ between 10 -5 and 10 -7 .

The crystal structure of hillebrandite, Ca 2 SiO 3 (OH) 2 , was solved and refined in space group Cmc2 1 , a = 3.6389, b = 16.311, c = 11.829 Å, to R = 0.041 using single-crystal X-ray data. The structure consists of a three-dimensional network of Ca-O polyhedra that accommodates wollastonite-type Si-O tetrahedral chains. Each of the wollastonite-type chains is an average of two symmetrically equivalent chains related by the mirror plane perpendicular to a. In a given structural channel of the Ca-O polyhedral network, only one chain orientation can be occupied to give reasonable Si-O distances. The O3 and O4 sites corresponding to each vacant Si2 site are occupied by OH groups to achieve charge balance. The wollastonite-type Si-O tetrahedral chains in the hillebrandite structure resemble those reported for many calcium silicate hydrate (CSH) phases.

High-resolution, magic-angle-spinning (MAS), 29 Si NMR spectra have been obtained for a sodium silicate glass and liquid (60 mol% SiO 2 ) to temperatures over 600 °C. Crystallization was monitored in situ, and motional averaging of the peaks for distinct silicate anionic species (Q 2 and Q 3 species) in the liquid was observed. Modeling of the spectra yielded species exchange frequencies. The inverses of these exchange frequencies are slightly longer than shear relaxation times calculated from bulk viscosity using the Maxwell relation; viscosities calculated from the NMR data using an Eyring model are close to, but slightly less than, measured viscosities. These results are similar to those obtained for Q 3 - Q 4 exchange in K 2 Si 4 O 9 liquid, indicating that local Si-O bond breaking occurs on a time scale consistent with it being the primary control on viscous flow even in lower silica liquids.

Eutectic melting temperatures for the assemblage Ca(OH) 2 + CaCO 3 in the presence of excess H 2 O were determined along six isochores of H 2 O in a hydrothermal diamond-anvil cell. These temperatures range from 610 °C at 1204 bar to 583 °C at 7203 bar; they agree very well (within 3 °C) with data obtained from high-pressure differential thermal analysis (HP-DTA) and are about 20 °C lower than those obtained from quench experiments. The additional DTA peaks near 630 °C for the melting experiments of pure portlandite reported by Harker (1964) and Bai et al. (1994) may be the result of a small amount of melting, caused by CaCO 3 contamination, and not the result of a solid phase transition between portlandite and "portlandite II."

X-ray diffraction patterns of opal-CT were modeled using a recursive calculation technique that allows for ordered and disordered intergrowths of planar units. The opal-CT structure was modeled as interstratifications of cristobalite and tridymite layers that were based on an idealized trigonal silica sheet, as found in the high-temperature structures. Modeled patterns matched observed patterns well, including the overall shapes of the broad peaks at ~19.5-24.5° 2θ and 35.9° 2θ (CuKα). The calculations allow several important interpretations of opal-CT: XRD patterns of opal-CT can be modeled by an interstratification of cristobalite and tridymite; both disordered and ordered interstratifications occur; particle size can be inferred from the width of the 35.9° 2θ peak; significant information regarding the structural state of opal is contained in weak, often-ignored bands (e.g., one at ~41-46° 2θ); and H 2 O is not present in an ordered fashion within the cristobalite and tridymite sheets.