Volume 84, Issue 5-6

The George Ashley Block (GAB) is a rockslide block located in the Pala pegmatite district of Southern California. It is layered, asymmetric, pocket containing, and peraluminous. The GAB con­sists of quartz (42 vol%), Na-rich plagioclase (27%), potassium feldspar (24%), muscovite (7%), Mn-rich garnet (2%), biotite (1%), and a trace of tourmaline and gahnite. It contains only small amounts of the incompatible elements that characterize differentiated pegmatite bodies. P 2 O 5 , MnO, and F are present in amounts of <1 wt% each; B, Be, Ce, Li, Nb, Nd, and Th are <100 ppm each. More than 90% of the garnet grains in the GAB are zoned toward Mn-rich rims, and a symmetrical change in garnet-core composition occurs across the body. The mean X site contents for garnet (in at%) are 57% Fe, 40% Mn, 3.1% Mg, and 0.4% Ca. The Mn contents of garnet range from 30 to 55 at%; Fe contents vary inversely with Mn and range from about 66 to 43 at%. It is concluded that the bulk chemistry yields little information about fractionation, but the garnet, muscovite, and biotite mineral chemistry is more useful. There may have been two separate injections of magma to form the GAB.

Pegmatites of the Pala and Mesa Grande Pegmatite Districts, San Diego County, California are typically thin, sheet-like composite pegmatite-aplite dikes. Aplitic portions of many dikes display pronounced mineralogical layering referred to as “line rock,” characterized by fine-grained, gamet-rich bands alternating with albite- and quartz-rich bands. Thermal modeling was performed for four dikes in San Diego County including the 1 m thick Himalaya dike, the 2 m thick Mission dike, the 8 m thick George Ashley dike, and the 25 m thick Stewart dike. Calculations were based on conductive cooling equations accounting for latent heat of crystallization, a melt emplacement temperature of 650 ℃ into 150 ℃ fractured, gabbroic country rock at a depth of 5 km, and an estimated 3 wt% initial H 2 O content in the melt. Cooling to <550 ℃ at the center of each dike occurred in ~9 years for the Stewart dike, ~340 days for the George Ashley dike, ~22 days for the Mission dike, and ~5 days for the Himalaya dike. Based on these calculations, growth rates for large pegmatitic minerals such as the 10 cm long Himalaya hanging wall tourmaline crystals may have been on the order of 10 -5 cm/s. Crystal size distribution (CSD) studies of garnet from layered aplites suggest growth rates of about 10 -6 cm/s. These results indicate that the dikes cooled and crystallized rapidly, with variable nucleation rates but high overall crystal-growth rates. Initial high nucleation rates coincident with emplacement and strong un­dercooling can account for the millimeter-size aplite grains. Lower nucleation rates coupled with high growth rates can explain the decimeter-size minerals in the hanging walls, cores, and miarolitic cavities of the pegmatites. The presence of tourmaline and/or lepidolite throughout these dikes suggests that although the melts were initially H 2 O-Undersaturated, high melt concentrations of incompatible (or fluxing) components such as B, F, and Li (±H 2 O), aided in the development of large pegmatitic crystals that grew rapidly in the short times suggested by the conductive cooling models.

The Crystal Peak area of the Pikes Peak batholith, near Lake George in central Colorado, is world-renowned for its crystals of amazonite (the blue-green variety of microcline) and smoky quartz. Such crystals, collected from individual miarolitic pegmatites, have a remarkably small variation in crystal size within each pegmatite, and the shapes of plots of their crystal size distributions (CSDs) are invariably lognormal or close to lognormal in all cases. These observations are explained by a crystal growth mechanism that was governed initially by surface-controlled kinetics, during which crystals tended to grow larger in proportion to their size, thereby establishing lognormal CSDs. Surface-controlled growth was followed by longer periods of supply controlled growth, during which growth rate was predominantly size-independent, consequently preserving the lognormal shapes of the CSDs and the small size variation. The change from surface- to supply controlled growth kinetics may have resulted from an increasing demand for nutrients that exceeded diffusion limitations of the system. The proposed model for crystal growth in this locality appears to be common in the geologic record, and can be used with other information, such as isotopic data, to deduce physico-chemical conditions during crystal formation.

Simulations of crystal growth in hydrous albite-quartz and albite-orthoclase systems were performed with the Jackson, Gilmer, and Temkin formulation of the Ising model. These simulations demonstrate that at 873 K (approximately 100 K undercooling), comb textures are produced when growth and diffu­sion probabilities are equivalent. If the growth probability is decreased to 0.1 times that of diffusion, discrete alternating zones of albite and quartz are produced; however, increasing the growth probability to 10 times that of diffusion results in an intergrowth of small domains of albite set in a quartz matrix. At near-liquidus temperatures (973 K), textures similar to those at 100 K undercooling are produced although the crystal-melt interfaces of the comb texture are smoother and the absolute value of the diffusion probability exerts a strong control on the texture, unlike results at 873 K. The textures produced in the simulations with a scale of 10 -7 m are similar to experimentally produced textures with scales of 10 -6 to 10 -4 . The fractal dimension of a comb-textured simulation has been measured and shown to be similar to the fractal dimension of a natural pegmatite with a meter-scale texture. This scaling similarity suggests that these simulations may provide insight into the formation of natural pegmatites despite the 5-7 orders of magnitude difference in scale.

Beryllium contents of granitic melts at saturation in beryl (± chrysoberyl, phenakite, or quartz) have been determined by forward- and reverse-direction experiments at 650-850 ℃ and 200 MPa using natural beryl or the compositionally equivalent mixture of phenakite + alumina + quartz added to metaluminous haplogranite, peraluminous haplogranite, and macusanite (a peraluminous Li, B. F, and P enriched rhyolite obsidian from Macusani, Peru). Chrysoberyl coexists with beryl ± quartz in moderately to strongly peraluminous systems (ASI approximately ≥ 1.05); the assemblage phenakite + beryl + chrysoberyl is stable in silica-undersaturated bulk compositions. The BeO content of the melts, which ranged between 0.03-0.40 wt%, varies principally with temperature (the solubility of beryl falls with decreasing T) and with the activity product of beryl, (a Beo ) 3 (a Al₂O₃ )(a SiO₂ ) 6 . Beryllia contents are lowest in strongly peraluminous and quartz-saturated haplogranite melts; however, the highest BeO content measured in glass products occurs for the strongly peraluminous macusanite at 850 ℃ (4016 ppm); we infer that the greater solubility of beryl in macusanite melt results from melt speciation reactions involving Li, F, B, and P. For all bulk compositions studied, the BeO content of melt projects to a narrow range of similar values at low T, near the solidus of haplogranite. Thus, metaluminous to peraluminous granitic magmas with and without common volatile and fluxing components will, if cooled to subsolidus temperatures, have similarly low BeO requirements for beryl saturation. The small BeO content of melt near the solidus of haplogranite results in a minor (10 ℃) depression of the freezing point. Compared to Be-free haplogranite melt, the mean melt composition at the beryl-saturated granite minimum shifts slightly toward quartz. Beryl is common in peraluminous granitic rocks in part because lower BeO contents are required to saturate these melts in beryl, and possibly because these melts acquire higher BeO contents by mica melting reactions at their sources. Beryl is also a common constituent of border-facies assemblages in granitic pegmatite dikes, regard­less of their ASI values. This early crystallization of beryl results from the low BeO content required to saturate any granitic melt in beryl at the low-T environments in which pegmatite dikes solidify.

Over 90% of the fluid inclusions in quartz, topaz, and beryl from the Morefield pegmatite, Amelia Co., Virginia, contain mixed CO 2 -H 2 O fluids. In order of abundance, there are four inclusion types: (1) three- phase, 1-1-v; (2) two-phase, H 2 O-rich; (3) three- or four-phase, 1-v-s and 1-1-v-s; and (4) two-phase, CO 2 - rich. CO 2 -rich inclusions formed by necking of inclusions containing mixed fluids. Topaz and quartz contain all four inclusion types, and all types occur throughout the pegmatite, although no H 2 O-rich inclusions occur in beryl. Quartz from gneiss that hosts the pegmatite contains dominantly H 2 O-rich inclusions. Most inclusions are secondary and occur along healed fractures. Possible primary inclu­sions occur only among those that contain solids. All inclusions homogenize into the liquid phase; a few homogenize to liquid CO 2 . H 2 O-rich inclusions with <2 wt% NaCl (eq.) and homogenization tem­peratures <200 ℃ occur in the host gneiss and throughout the pegmatite and are interpreted to contain meteoric waters. H 2 O-rich and mixed fluid inclusions with 3-7 wt% NaCl (eq.) and homogenization temperatures >290 ℃ contain fluids interpreted to derive primarily from the pegmatite. Inclusions with salinities between 3-13 wt% NaCl (eq.) and intermediate homogenization temperatures are in part cooled equivalents of the pegmatite fluids, but the origin of those with salinity >7 wt% NaCl (eq.) is uncertain. Homogenization temperatures for CO 2 indicate average CO 2 densities near 0.65 g/cm 3 . For mixed fluids homogenizing near 300 ℃ and this CO 2 density, the trapping pressure is estimated to be 2.5 kb. Neither age relations nor physical proximity tie the Morefield pegmatite to known granitic intrusions. Fluid compositions preclude a genetic relation between the pegmatite and its host gneisses.

Coarse crystals of niobian rutile occur in the hydrothermally altered core-margin zone of the McGuire granitic pegmatite, Park County, Colorado, associated with potassium feldspar, quartz, biotite, ilmenite, and monazite-(Ce). Primary homogeneous niobian rutile, with Fe 3+ ≅ Fe 2+ and a small excess of (Fe,Mn) over the amount required to compensate the incorporation of (Nb,Ta,W), under­went three stages of exsolution. Primary homogeneous niobian rutile exsolved a fine trellis-like pat­tern of minor lamellar Nb-bearing pseudorutile I. Most of this phase was broken down to pseudomorphs consisting of microgranular Nb-rich pseudorutile II imbedded in niobian-ferrian “ferropseudobrookite.” Continued exsolution in niobian rutile and reconstitution of the early exsolution products generated (Fe,Nb)-depleted, microgranular niobian rutile, titanian ferrocolumbite, and mi­nor ilmenite. These three phases did not attain chemical equilibrium but may represent a stable phase assemblage. All these processes seem to have maintained charge balance, suggesting a closed sys­tem. Subsequent to the three stages of exsolution, extensive oxidation converted the mineral assem­blages to anatase + hematite + titanian-tungstenian ixiolite; primary ilmenite was oxidized into an anatase + hematite intergrowth. In both cases, the hematite component was almost completely leached out, leaving highly porous aggregates of the other phases. The exsolution products in niobian rutile are controlled by the (Fe+Mn+Sc)/(Nb+Ta+W) ratio of the primary phase and by its (Fe 3+ +Sc)/ (Fe 2+ +Mn) ratio. Dominance of divalent A-cations facilitates exsolution of titanian ferro- to manganocolumbite or titanian ixiolite, whereas dominant trivalent cations lead to exsolution of titanian (Fe,Sc) 3+ NbO 4 phases. Excess (Fe,Mn) over the columbite-type Fe 2+ Nb 2 stoichiometry causes exsolution of (Fe,Mn)-rich Nb,Ta-oxide minerals or complementary (Fe,Mn,Ti) phases.

The crystal structure of sogdianite, hexagonal, a = 10.053(1), c = 14.211(2) Å, V= 1243.8(4) Å 3 , space group = P6/mcc, Z = 2, has been refined to an R index of 2.0% using 607 observed (|F o | > 5σ|F|) reflections collected with an automated four-circle diffractometer using MoΚα X-radiation. Site­scattering refinement and electron-microprobe analysis indicate the composition (Zr 0.76 Ti 4+ 0.38 Fe 3+ 0.73 Al 0.13 ) ∑ = 2 (❐ 1.15 Na 0 . 85 ) ∑ = 2 K[Li 3 Si 12 O 30 ]. The T2 tetrahedron is completely occupied by Li, and all Al occurs at the octahedrally coordinated A site. The <A-O> bond length is compatible with all Fe being in the trivalent state. There is no significant positional disorder of the A-group cations, in accord with the absence of H 2 O in sogdianite. Single-crystal X-ray precession photographs of sogdianite show the presence of diffractions that are incommensurate with the sogdianite diffraction pattern. The relative intensity of the incommensurate diffractions increases with the degree of apparent alteration of sogdianite (clear → turbid → murky), suggesting that sogdianite is altering to, or possible exsolving, a second phase, probably zektzerite.

Simmonsite, Na 2 LiAlF 6 , a new mineral of pegmatitic-hydrothermal origin, occurs in a late-stage brec­cia pipe structure that cuts the Zapot amazonite-topaz-zinnwaldite pegmatite located in the Gillis Range, Mineral Co., Nevada, U.S.A. The mineral is intimately intergrown with cryolite, cryolithionite and trace elpasolite. A secondary assemblage of other alumino-fluoride minerals and a second generation of cryolithionite has formed from the primary assemblage. The mineral is monoclinic, P2 1 or P2 1 /m, a = 7.5006(6) Å ,b= 7.474(1) Å, c = 7.503(1) Å, β = 90.847(9)°, V= 420.6(1) Å 3 , Z= 4. The four strongest diffraction maxima [d (Å), hkl, I/I 100 ] are (4.33, 111 and 111̅, 100); (1.877, 400 and 004, 90); (2.25, 131̅, 113, 131 and 311, 70); and (2.65, 220,202, 022, 60). Simmonsite is pale buff cream with white streak, somewhat greasy, translucent to transparent, Mohs hardness of 2.5-3, no distinct cleavage, subconchoidal fracture, no parting, not extremely brittle, D m is 3.05(2) g/cm 3 , and D c is 3.06(1) g/cm 3 . The mineral is biaxial, very nearly isotropic, N is 1.359(1) for λ = 589 nm, and birefringence is 0.0009. Electron micro­probe analyses gave (wt%) Na = 23.4, Al = 13.9, F = 58.6, Li = 3.56 (calculated), with a total of 99.46. The empirical formula (based on 6 F atoms) is Na 1.98 Li 1 . 00 Al 1.00 F 6 . The crystal structure was not solved, presumably because of unit-cell scale twinning, but similarities to the perovskite-type structure exist. The mineral is named for William B. Simmons, Professor of Mineralogy and Petrology, University of New Orleans, New Orleans.

Ferrotitanowodginite is a new mineral of the wodginite group, which is found in and defined from two closely adjacent pegmatites in the San Luis province, Argentina, but previously encountered in the Tanco (Manitoba) and Marko’s (Ontario) pegmatites. In the San Elias pegmatite, ferrotitanowodginite occurs aggregated with ferrowodginite and associated with microlite, ferrotapiolite, cleavelandite, and quartz. In La Viquita pegmatite, ferrotitanowodginite occurs in a replacement unit associated with wodginite, ferrowodginite, titanowodginite and ferrotapiolite in yellow muscovite and quartz. Ferrotitanowodginite is very dark brown to black, with a dark brown streak and submetallic luster. Mohs hardness is 5½ and D calc = 7.368 g/cm 3 . In reflected light, it is creamy white and gray in air and oil, respectively; anisotropy is distinct, bireflectance and pleochroism moderate. Ferrotitanowodginite is monoclinic, space group C2/c, Z = 4. Unit-cell dimensions for San Elías’material are a = 9.403(4) Å b = 11.384(3) Å, c = 5.075(1) Å, β = 90.55° (2), V= 543.24(22) Å 3 . The strongest lines in the X-ray powder diffraction are d = 2.963 Å, I = 100%, (hkl = 221̅); d = 2.939 Å, I = 90% (hkl = 221); d = 3.626 Å I = 70% {hkl = 220); and d= 1.715 Å I = 50% (hkl = 402). Eighteen analyses by electron microprobe gave the following mean composition: WO 3 0.02, Nb 2 O 5 6.52, Ta 2 O 5 70.68, TiO 2 7.10, SnO 2 1.25, ThO 2 0.01, UO 2 0.02, As 2 O 3 0.03, Sb 2 O 3 0.02, Bi 2 O 3 0.03, Fe 2 O 3 2.18, MgO 0.01, CaO 0.01, MnO 1.05, FeO 10.27, PbO 0.05, total 99.25 wt%. The La Viquita sample is much richer in Mn and Sn. The simplified formula for ferrotitanowodginite is: (Fe 2+ ,Mn 2+ ) (Ti,Sn 4+ ,Ta,Fe 3+ ) (Ta,Nb) 2 O 8 , ideally Fe 2+ TiTa 2 O 8 .

The crystal structure of a rubidian cesian phlogopite-1M from pegmatite exocontacts at Red Cross Lake, Manitoba, monoclinic, a = 5.343(1),6 = 9.247(2),c = 10.397(3) Å, β = 100.04(2)°, V= 505.8(2) Å 3 , has been refined to an R index of 4.5% based on 519 observed reflections measured with graphite-monochromated MoΚα X-radiation on an automated four-circle diffractometer. The crystal used in the collection of the X-ray intensity data was also analyzed by electron microprobe, giving the unit formula (K 0.46 Rb 0.28 Cs 0.23 )(Mg 1.20 Fe 1.00 Al 0.38 Li 0.34 Mn 0.04 Ti 0.04 )(Si 2.91 Al 1.09 )O 10 [(OH) 1 . 55 F 0.45 ]. The interlayer site, X, contains large amounts of Rb and Cs, and cell dimensions and the <X-O> distance are in accord with data from synthetic Rb and Cs phlogopites and plutonic phlogopites. The interlayer coordination is much more regular in rubidian cesian phologopite than in other trioctahedral micas, defining a trend of increasing regularity with increasing interlayer-cation size.

A chemical and paragenetic study was performed on gadolinite-group minerals occurring in miarolitic pink granite and granophyric leucogranite of the subvolcanic Hercynian plutons at Baveno and Cuasso al Monte, Southern Alps, Italy. In the localities investigated, gadolinite-group minerals are hosted in massive pegmatite, in aplite, and in miarolitic cavities having different degrees of evo­lution. The petrological relations indicate that progressive crystallization has occurred from mag­matic through to hydrothermal conditions. At Baveno, Ce-rich gadolinite-(Y) (with ∑REE >Y) formed during the primitive stages of pegmatite crystallization. Gadolinite-(Y) (with SREE <Y) formed in pegmatites and granophyric aplites during primitive to moderately evolved stages of these dikes. Gadolinite-(Y) (with ∑REE <Y) and hingganite-(Y), which contains a significant amount of the datolite component, occur in miarolitic cavities together with several rare-element accessory phases. During the latest stages, datolite formed with zeolites. At Cuasso al Monte, gadolinite is found only in primitive to highly evolved miarolitic cavities. The cores of these crystals consist of Nd-rich gado- linite-(Y) (with ∑REE >Y). Gadolinite-(Y) (with ∑REE <Y) formed during intermediate stages of evolution, and hingganite-(Y) is dominant in highly evolved miarolitic cavities together with several rare-element phases. The chemical differences observed in the gadolinites from the two localities may indicate a different parental magma composition and reflect a difference in the crystallization processes. In contrast to Baveno, the crystallization at Cuasso al Monte occurred under open-system conditions, which prevented the formation of a zeolite (datolite-bearing) stage and generated a typical medium- to low-temperature hydrothermal mineral assemblage consisting of quartz, fluorite, barite, sulfides, and carbonates. The large variations in the Y/Dy ratio observed in the studied samples may be due to a change in the fluorine abundance in hydrothermal fluids related to paragenetic effects and mixing processes.

Textural features combined with mineral chemistry preserved in metamorphic rocks provide in­sights into metamorphic reaction mechanisms as well as open vs. closed system processes. Prograde tourmaline-rich muscovite pseudomorphs after staurolite develop in sillimanite zone metapelites adjacent to peraluminous granitoid intrusives in NW Maine. Tounnalines occur in discrete domains restricted to central regions of muscovite-rich, quartz-poor pseudomorphs with biotite-rich margins. These tourmaline grains are relatively large (≤ 1.0 nun), lack detrital cores and exhibit only minor compositional zoning, in sharp contrast to matrix tourmaline from other samples. These features suggest fluid-infiltration as the causative mechanism for the fonnation of these tounnaline-rich mica pseudomorphs after staurolite. Irreversible thermodynamic models of local reactions and material transport in combination with mineral chemistry allow evaluation of reaction mechanisms that produced these pseudomorphs. Ther­modynamic models in the NKCMTFASHOB system mimic the observed textural features if a three- stage process is used. Stage 1: Stamolite replacement is initiated by infiltration of an aqueous phase that adds K-Na-H 2 O to the rock with the concomitant removal of Al+Fe. Because the system is initially undersaturated with respect to tourmaline, a pseudomorph containing muscovite with minor biotite develops at the expense of staurolite. Stage 2: With continued infiltration, concentration of B increases, tourmaline saturation is exceeded, tourmaline nucleates and grows. Local material trans­port constiaints mandate that tourmaline precipitation be spatially restricted to regions of staurolite dissolution. Consequently, tourmaline forms in clusters at sites containing the last vestiges of stauro­lite in the pseudomorph core, also evidenced by staurolite inclusions within several tourmaline grains. Resultant domains of staurolite replacement during this stage contain about equal amounts of musco­vite and tourmaline. Typical staurolite poikiloblast pseudomorphing reactions require silica trans­port. matrix quartz dissolves from the surrounding host resulting in a local enrichment of biotite and plagioclase at the pseudomorph margin. Stage 3: Small amounts of sillimanite nucleate and grow throughout the rock. Late-stage aqueous fluids from the adjacent monzonitic intrusive are likely to be the primary B source. Theoretical, textural, and compositional modeling combined with observational data indicate that boron must have been derived externally from the rock, that the modal amount of tourmaline is very sensitive to the B content of the fluid, that tourmaline is stable throughout the sillimanite zone depending on other cation activities and pH of the fluid, and that these pseudomorphs provide insight into B contents of metamorphic fluids and the timing of the B influx. The outer geographic extent of the tourmaline-bearing pseudomorphs marks the boundary of a reactive geochemical front, and thus defines an advective isograd. Inteipretation of subtle textural features preserved in the rock in con­junction with irreversible textural modeling provides a powerful tool with which to understand the chemical evolution of metamorphic rocks and the fluids involved.

Alkali suite anorthosites and norites are the second most common plutonic rock association in the western lunar highlands (after the magnesian suite), but then origin poses an enigma for most petrologic models of lunar crustal evolution. Some models suggest that the alkali and magnesian suites formed from distinct, unrelated parental magmas, whereas other models propose that both suites formed from the same parental magma. The contrast in major element chemistry of the cumulus phases in each suite is difficult to reconcile with their similar incompatible element chemistry. We present herein a detailed ion microprobe (SIMS) and electron microprobe study of seven alkali suite rocks. Our data show that most alkali suite anorthosites preserve major and trace element characteristics acquired during their formation as igneous cumulate rocks, and that these characteris­tics can be used to reconstruct the parental-magma composition. The data indicate that cumulates of the alkali suite crystallized from magmas with rare-earth element (REE) contents ~0.6-2.0 × high-K KREEP, and small but consistently positive Eu anomalies (Eu/Eu* ~2) relative to KREEP. Snyder and others (1995a) have proposed that the alkali suite parental magma is similar to Apollo 15 pristine KREEP basalt. Our model suggests that the major element composition of cumulus plagioclase in most alkali suite rocks is too sodic for the calculated crystal line-of-descent of pristine KREEP basalt, and that assimilation of pre-existing calcic anorthosite is required. This conclusion is supported by the REE patterns of th alkali-suite parental magma determined here. We propose that alkali suite anorthosites formed as flotation cumulates in KREEPy plutons that may have formed norites as complementary bottom cumulates. The alkali flotation cumulates reflect fractional crystallization of their parental phiton. local equilibrium crystallization, assimilation of plagioclase-rich roof rock, and episodic magma-mixing during convective overturn of the crystalliz­ing magma bodies. Texturally pristine alkali anorthosites exhibit petrographic characteristics that are consistent with their origin as cumulates in a KREEPy pluton. including abundant modal plagio­clase. post-cumulus pyroxenes (both augite and pigeonite) that generally lack exsolution lamellae and that have equilibration temperatures of 950-1100 ℃. relict igneous textures and. in some cases, igneous lamination. The lack of cumulus mafic phases in rocks that should be pyroxene-saturated suggests separation of the plagioclase by flotation, not sinking. Assimilation of plagioclase from older, anorthositic highlands crust is indicated by the high Eu contents of the cumulates and by the positive Eu anomalies in their calculated parental melts relative to high-K KREEP. Mixing of the evolved alkali-suite parental magma with primitive melt occured episodically, as shown by reverse zoning profiles hi some cumulus plagioclase. Injection of this primitive, hot magma into the crystal­lizing pluton may have induced convective overturn of the magma chamber.

We report new findings related to sample 14161,7373, the first analyzed limar intrusive rock to show textural evidence for the formation of granitic material by silicate liquid immiscibility (SLI) and geochemi­cal evidence for separation of that material, at least on a small scale, from mafic residua. Ion micro­probe analyses of rare-earth element (REE) concentrations in silicates record the compositional evolution of this late-stage assemblage. For a sample whose bulk assemblage, that of whitlockite monzogabbro, has REE concentrations at ~5-6× KREEP levels, the silicates have only moderately high REE concen­trations. The calculated REE concentrations of melt in equilibrium with pigeonite, after re-integration of pigeonite host and augite lamellae, is approximately 2 × high-K KREEP. The high bulk REE concen­trations of the assemblage are the result of a high proportion of whitlockite, probably an excess, i.e., the assemblage is a whitlockite cumulate. Variations in silicate REE concentrations reflect the co-crystallization of whitlockite, which we interpret to have begun crystallizing prior to the onset of late-stage immiscibility of the felsic melt fraction. Calculated cooling rates, based on a quantitative model of pyroxene exsolution, reveal a two-stage thermal history, which involved initial crystallization at a mini­mum depth of 700-800 m during which the primary augite and pigeonite grains unmixed to form the observed host-lamella pairs. Unusual compositional profiles close to the host-lamella interface are consistent with a second stage of mild reheating followed by rapid cooling at a depth of ~30 m, most likely in an ejecta blanket. The shallow depth of origin indicated for 14161,7373 is consistent with other studies of evolved lunar intrusive rocks, supporting the interpretation that lunar QMD and granite, as found among the Apollo samples, are late-stage differentiates of high-level intrusions and are not related to differentiation of deep crustal KREEP reservoirs.

We performed experiments at 3.0 GPa and 1530-1565 ℃ to investigate the effects of crystal composition on trace element partitioning between garnet and anhydrous silicate melt. Bulk compo­sitions along the pyrope (Py: Mg3Al 2 Si30 12 )-grossular (Gr: Ca 3 Al 2 Si 3 O 12 ) join, doped with a suite of trace elements (Li, B, K, Sc, Ti, Sr, Y, Zr, Nb, Cd, In, REE, Hf, Ta, Th, and U) produced homoge­neous garnets, ranging in composition from Py 84 Gri 6 to Py 9 Gr 91 , in equilibrium with melt. Trace element partition coefficients (D-values), measured by SIMS, depend greatly on the Mg/(Mg + Ca) of garnet. For example, from Py 84 to Py 9 , D La increases from 0.004 to 0.2, whereas D U increases from 0.029 to 0.42. These variations can be explained by the lattice strain model of Blundy and Wood (1994), which describes trace element partitioning of an element i in terms of the ionic radius of i (r i ), the size of the lattice site on which i partitions (r 0 ), the Young’s modulus of the site (E), and the (theoretical) partition coefficient D 0 for an ion of radius r 0 . For trivalent cations substituting in the garnet X-site (Y, REE, Sc, and In), apparent values of r 0 fitted to our data vary systematically from 0.935 ± 0.004 Å (Py 84 ) to 0.99 ± 0.01 Å (Py 9 ), a trend consistent with variations in the size of the X-site. Values of D 0 show an increase from Py 9 (D 0 = 2.8 ± 0.1) to Py 84 (4.8 ± 0.1) and Young’s modulus E varies from 257 ± 20 GPa for Py 60 to 590 ± 40 GPa for Py 84 . These results allow a quantitative assessment of the influence of crystal chemistry on gar- net-melt D-values, thereby forming the basis for a predictive model similar to that recently developed for clinopyroxene-melt partitioning by Wood and Blundy (1997). Our new data emphasize the importance of taking into account crystal composition when modeling trace element behavior in natural systems.

The room-temperature static compression of magnesite (MgCO 3 ) has been measured by X-ray powder energy-dispersive diffraction in a diamond-anvil cell. Different pressure transmitting media and an internal pressure standard (gold) were used in order to differentiate the effects of stresses or pressure measurement errors in determining the isothermal bulk modulus K 0 and first pressure deriva­tive Kʹ 0 . A third order Eulerian equation of state accounts well for the sets of data obtained in different configurations and yields a bulk modulus K 0 = 115(1) GPa and an ambient unit-cell volume V 0 = 278.9(1) Å 3 with a pressure derivative Kʹ 0 fixed to 4, consistent with recent measurements of Zhang et al. (1997) and Ross (1997). Measurements collected to 72 GPa allow us to have a reliable constraint on the first pressure derivative K 0 . The best fit to the data yields K 0 = 108(2) GPa, K 0 = 4.6(2), and an ambient unit-cell volume V 0 = 279.1(1) Å 3 . The first pressure derivative K 0 would be 4.5(1) with a bulk modulus fixed to the value of 110 GPa derived from the ultrasonic measurement of Christensen (1972).

Room-temperature volume measurements of the complete set of calcite-structure carbonates in the pressure range 0-8.1 GPa revealed systematic differences in compressibilities that depend on cation type, resulting in significant deviations from empirical relations of an inverse linear correla­tion between K 0 and V 0 . The bulk modulus for MgCO 3 lies approximately 18 GPa below the trend for the 3d transition metal carbonates, which show an expected inverse linear correlation of bulk modu­lus with ambient cell volume (and M-O bond length). The bulk modulus of CdCO 3 , whose M-O bond length and cell volume are only slightly smaller than those of CaCO 3 , lies up to 10 GPa above the trend of the 3d transition metal carbonates and about 30 GPa above that of calcite. These deviations in compressibility trends as a function of cell volume (or M-O bond length) are expressed as differ­ences in a axis compressibility but not in c axis compressibility, which shows a nearly linear increase with M-O distance. Hence, systematic behavior is apparently limited to subsets of carbonates in which metal ions share the same valence electron character (i.e., s vs. 3d vs. 4d) and is primarily attributed to unexpected compressibility differences along the a axis. Crystal-field effects, beyond those reflected in the M-O distances, cannot account for the ob­served compressibilities. Nonbonded interactions also fail to explain the deviations from predicted trends. Variations of electronegativity and vibrational frequency with ionic radius for the relevant metal ions show differences that are qualitatively similar to the observed trends of bulk modulus, suggesting that differences in bonding character may contribute to the different behaviors among the subsets of calcite-structure carbonates. However, it is most likely that a combination of factors is necessary to account fully for the observed behavior.

In nanocrystalline anatase coarsened under hydrothermal conditions (250 ℃, P sat ), the anatase- to-rutile phase transformation is nucleated at anatase {112} twin boundaries formed by oriented attachment. The anatase twin boundary is constructed from structural elements common to rutile. Specifically, rutile nucleation involves displacement of only one half the titanium cations within the twin slab. Subsequent transformation of bulk anatase involves rupture of 7 of the 24 Ti-O bonds per unit cell and cooperative displacement of Ti and O. As the transformation advances into the bulk material, adjacent slabs of anatase octahedra are destabilized, resulting in rapid progression of the transformation of bulk anatase to rutile. The implied chain reaction, scarcity of partly reacted crys­tals, absence of multiply twinned rutile, and the importance of nucleation at anatase twins indicate a rate law based on slow nucleation and rapid growth. The displacements are comparable to those proposed previously for macroscopic anatase at much higher temperatures, indicating the atomic mechanism is not modified by particle size or temperature, despite the rapid kinetics in finely crystal­line aggregates. In addition to the formation of twins, clusters with rutile-like character may occur at some fraction of random anatase-anatase particle contacts. Such interfaces should result in decreased activation barriers for rutile nucleation and, thus, contribute significantly to the observed faster trans­formation rates in nanocrystalline compared to coarsely crystalline materials.

Expanding upon our previous studies of the properties of Au complexes, we present calculations for several Hg 2+ species in aqueous solution and for molecular models for cinnabar. Hydration effects are treated with a combination of “supermolecule” calculations containing several explicit water molecules and polarizable continuum calculations. We focus upon the following problems: (1) calcu­lation of the stabilities of HgL 2 , L = F - , Cl - , OH - , SH - , and CN - and HgCl 2-n n n = 1-4; (2) development of a molecular model for cinnabar of the form Hg 3 S 2 (SH) 2 ; and (3) dissolution or adsorption reac­tions using this cinnabar model. The absolute and relative formation enthalpies of the HgL 2 species can be satisfactorily reproduced at the Hartree-Fock plus Moller-Plesset second order correlation correction level using relativistic effective core potential basis sets if the hydration of neutral HgL 2 is explicitly taken into account. Evaluating the energetics for the series of complexes HgCl 2-n n is more difficult, because great accuracy is needed in the large hydration energies and some of the species are highly nonspherical. The Hg 3 S 2 (SH) 2 species shows an equilibrium structure very much like that in cinnabar. The relative energetics for dissolution of cinnabar by H 2 O, H 2 S, SH - , and SH - + elemental S are correctly reproduced using this model molecule. Calculations on Hg 3 S 2 ClI provide a model for understanding the adsorption of I - ions on cinnabar surfaces in the presence of Cl - .

Goethite (010), (110), and (111) growth faces and (010) cleavage surfaces of large, natural single crystals, as well as a high surface area synthetic sample were characterized using various surface sensi­tive microscopies and spectroscopies. Differential interference contrast and atomic force microscopy characterization of the natural single crystal faces showed microtopography indicative of growth, disso­lution, and cleavage. Low energy electron diffraction patterns of the goethite (010) surface exhibit sharp, intense diffraction spots, indicating long-range order on this important surface. These patterns have two-dimensional point group symmetry 2mm, consistent with an undistorted surface structure and unit­cell parameters a = 4.62 ± 0.14 Å and c = 2.99 ± 0.08 Å. These parameters equal the equivalent bulk cell dimensions given the uncertainties. Ultra-high vacuum scanning tunneling microscopy was performed on (010) cleavage faces, although the tunneling properties of the surface were very heterogeneous. Atomic resolution was not obtained; however, microtopographic images are identical to those collected with AFM. XPS spectra from the (010) faces of two natural samples as well as the synthetic powder all have peak maxima for Fe (2p 3/2 ) at 711.5 ± 0.1 eV. The O(1 s) line originating from the goethite can be fit with two peaks with a chemical shift of 1.3 eV. The peak at higher binding energy (531.3 eV ± 0.1 eV) represents the protonated oxygen in the structure, and the peak at lower binding energy (530.0 eV ± 0.1 eV) represents the proton-free oxygen in the structure. Ab initio and semi-empirical models of the (010) surface suggest that cleavage occurs through the hydroxide plane at 1/4 b in the structure. This is contrary to cleavage through the oxide plane at 1/2 b, which has been assumed in several previous studies.

The temperature induced goethite-hematite phase transformation that occurs at about 250 ℃ was studied using in situ synchrotron X-ray powder diffraction with a capillary Debye-Scherrer geometry and a translating image plate system (TIPS). To our knowledge, this is the first time the goethite- hematite transformation has been investigated in real time. The sample was a pure, synthetic, sto­ichiometric goethite with 1 μm long needle-shaped crystals. The microstructural characterization showed that the sample was well crystallized. The Rietveld refinement of 30 powder patterns extracted from the image in the range 25-800 ℃ demonstrates that an intermediate phase with non- stoichiometric composition (“protohematite”) forms after the decomposition of goethite. The cell parameter b of goethite dramatically decreased during the phase transformation while a and c instead continued to increase. Protohematite is iron-deficient and retains residual hydroxyls for charge bal­ance. With temperature protohematite progressively transforms into hematite. Empty layers (pores) are consequently formed about the hematite clusters. The distribution of iron vacancies was modeled in the powder patterns with stacking faults that were simulated using anisotropic broadening coeffi­cients of the pseudo-Voigt profile function. Its disappearance with temperature was effectively fol­lowed with a decrease of the density of stacking faults.

Powder neutron diffraction [ISIS Facility (U.K.), POLARIS diffractometer] was used to investigate the effect of elevated pressure on cation partitioning in synthetic Mg 0.94 Al 2.04 O 4 spinel. The distributions of Mg, Al, and vacancies were studied as a function of pressure, by refinement of the T- and M-site scattering lengths, and determination of the cation partitioning through numerical mini­mization methods. The partially disordered Mg/Al distribution, which results from the synthesis process, show an increase in ordering between 6 and 18 kbar, where Mg and Al order to the T- and M- sites, respectively. Pressure effectively tends to stabilize MgAl-spinels with a “normal structure,” and this behavior is supported by numerical simulations based on classical electrostatic calculations.

The crystal structure of a Th-U-rich, Y-poor hellandite-(Ce) occurring in a volcanic ejectum from Capranica (Vico volcanic complex, Latium, Italy) was refined to an R index of 1.5% for 2226 ob­served reflections [I > 3 σ(I)]. Hellandite from Capranica is monoclinic P2/a with a = 19.068 (8), b = 4.745 (2), c = 10.289 (3) Å, p = 111.18 (3)°. The extra-framework (distorted) tetrahedral cavity, usually occupied by H in hellandite, is partially (35-40%) occupied by Be and/or Li. When this is the case, the [B 4 Si 4 O 22 ] tetrahedral chains of hellandite are locally cross-linked to form a sheet, which is built up by the same types of rings as in semenovite but in a different arrangement. The structure refinement and the electron- and ion-microprobe analyses provide constraints on hellandite-group crystal chemistry. The presence of [4] (Be, Li) at the Be site is coupled with the substitution of F and/ or O for OH at the 05 site; the amount of extra-framework cations thus constrains the OH content in the general formula. X-ray data exclude the presence of other OH sites in the structure, in agreement with direct H determination by ion probe (1.02 H apfu + 0.98 F apfu = 2.0 apfu of monovalent anions at O5 vs. 4.5-8 apfu proposed previously). The complete chemical analysis also eliminates the need for a vacancy at the REE sites. The general formula of hellandite-group minerals is thus (Ca,REE) 8 (Th,U,Y,REE) 4 (Ti,Fe 3+ ,Al)2(O,F,OH x ) 4 (Be,Li) 4-x Si 8 B 8 O 44 , where electroneutrality is assured by the appropriate mixing of heterovalent substituents at the M sites.

Nine tourmaline crystals for which major and minor element composition data are available have been examined by single-crystal X-ray structure refinement. The single crystals were then analyzed for major elements (by electron microprobe methods), Fe 3 /Fe 2+ (by synchrotron micro- X-ray absorption near-edge spectroscopy). B and Li (by secondary ion mass spectrometry), and bulk H content (by uranium extraction). Despite recent claims based on chemical analyses, struc­ture analysis suggests that no B exists in tetrahedral coordination in these samples. Analysis of cation ordering between the Y and Z octahedral sites suggests that the occurrence of an Fe 2+ atom on a Y octahedral site may be locally associated with the absence of Mg at both of the neighboring Z sites, as substitutions of Fe 2+ on Y and Mg on Z require antithetic shifts of the O6 anion.

A single-crystal X-ray diffraction study of synthetic pseuowollastonite (CaSiO 3 ) shows that this crystal has monoclinic C2/c symmetry, with a = 6.8394(5), b = 11.8704(9), c = 19.6313(7) Å, P = 90.667(6)°, and V= 1593.7(2) Å 3 . Basic features of the C2/c structure are similar to those previously determined with a C1̄ symmetry but fewer sites exist for the monoclinic space group: five symmetri­cally nonequaivalent Ca sites, three Si sites, and nine O sites.

Synthetic epidotes were produced using a Tuttle type hydrothermal vessel in the temperature range from 500 to 700 ℃ and at 4.5 to 5.0 kbars pressure. Single-crystal X-ray diffraction structure refinements yielded intracrystalline cation distributions for two crystals grown at 600 ℃ (0.68 and 0.73 Fe apfu) and five crystals grown at 700 ℃ (0.88 to 1.08 Fe apfu). The resulting Fe occupancies were compared with those calculated according to a thermodynamic model: The samples formed at 700 ℃ display a Fe distribution between M3 and Ml sites that match the calculation; nevertheless, we found the presence of a small but appreciable amount of Fe in the M2 sites that increases with the total Fe content (up to 0.08 Fe apfu). The crystals formed at 600 ℃ show a much higher disorder than expected, but have no Fe in the M2 site. Crystal-chemical features of the samples studied were compared with those of a previously published set of natural epidotes displaying a much lower Fe disorder: In particular, the unit-cell volume vs. Fe content relationship is not affected by the intracrystalline Fe distribution. This relation suggests that the Al-Fe intracrystalline distribution is independent of pressure.

For two series of aluminosilicate glasses on the SiO 2 -NaAlO 2 and SiO 2 -CaAl 2 O 4 joins, 29 Si magic- angle-spinning (MAS) NMR spectra were measured. Systematic variations in peak positions and widths with composition are closely related to the extent of ordering of Si and Al cations. A statistical thermodynamic model based on the quasi-chemical approximation was formulated to calculate the proportions of SiO 4 groups with varying numbers of Al neighbors and thus to quantify the extent of ordering. Multiple spectra in each compositional series were fitted simultaneously with several peaks representing each of these structural species and with area constraints generated by the model. The extent of aluminum avoidance (Q), which was defined using the relative lattice energy differences among the linkages Si-O-Si, Si-O-Al, and Al-O-Al, was optimized for each series. For the calcium aluminosilicates, the best fit is with 0.8 ≤ Q ≤ 0.875, where Q = I represents perfect Al-avoidance. For the sodium series, Q was found to be larger (0.93 ≤ Q ≤ 0.99), as expected from energetic consid­erations and from known variations in ordering in minerals. The contributions to the overall configurational entropy and heat capacity from Si-Al disorder can be calculated, and are significant fractions of experimentally estimated values. However, major contributions must also come from other sources of disorder, such as “topological” disorder of bond angles and length.

The Si L 2,3 core-loss edge can be used to probe the crystal chemistry around Si, providing information on the s- and d-like partial density of unoccupied states of the Si-O bonds. We present Si L 2,3 edges from 59 silicates, glasses, and amorphous materials acquired by parallel electron energy-loss spectroscopy (PEELS) with a transmission electron microscope (TEM) at an energy resolution of 0.7 eV. The Si L 2,3 edge Spectmm of α-quartz is interpreted using the results of a recent pseudopotential band-structure cal­culation. A combination of Si s- and d-like partial density of states derived from this calculation resembles the Si L 2,3 energy-loss near-edge structure (ELNES) of α-quartz. The Si L 2,3 ELNES of the silicates are interpreted using the results of the band-structure calculation of α-quartz. The Si L 2,3 edges of Q 4 , Q 3 , Q 2 , some Q 1 silicates, and amorphous materials have ELNES similar to that of α-quartz, and the Q 0 and some Q 1 silicates have ELNES different from that of α-quartz. A “coordination fingerprint” is defined for Q 4 , Q 3 . and Q 2 Si L 2,3 ELNES because of their similarity to the α-quartz spectrum. The similarities between the L 2,3 core-loss edge shapes of the third-row XO n- 4 (X = Al, Si, S, and P) series attests to a common molecular-orbital picture of their bonding. For Q 0 and some Q 1 spectra a “structure fingerprint” is defined because the Si L 2,3 -edge shapes are indicative of the number, distribution, and nature of the non-nearest- neighbor atoms. Spectra of olivine glasses and metamict zircon more closely resemble the α-quartz spec­trum than their crystalline analogs. In contrast to previous studies, we show that distortion of the SiO 4 tetrahedron is of secondary importance as an ELNES-modifying parameter. Polyhedral distortions be­come less important with increase in polymerization. There is a positive linear correlation between the energies of the Si L 2,3 -edge onsets and polymerization. Si 2p and 2s binding energies, and the 29 Si NMR isotropic chemical shifts. The shift to higher energies of the edge onsets with polymerization corresponds to an increase in effective charge on the Si atom with higher Q n . For silicates with isolated SiO 4 tetrahedra. increases in L 2,3 -energy onsets correlate with increases in polarizing power of the next-nearest-neighbor cations. The Si L 2,3 -edge shapes are affected by the types and coordinations of the next-nearest-neighbor cations. For example, andradite, ilvaite, fayalite, and γ-Fe 2 SiO 4 have FeO 6 bonded to SiO 4 and exhibit similar ELNES. Topaz, diunortierite, staiuolite, and kyanite have similar Si L 2,3 ELNES. with AlO 6 bonded to the SiO 4 . Their edge shapes are distinct from those of silicates with SiO 4 bonded to AlO 4 , as in the feldspars. A comparison of the Al and Si L 2,3 and Al, Si, 0, and F K core-loss edges of topaz illustrates the influence of neighbor effects and mixing of unoccupied states. This mixing illustrates the limitations of ab initio methods that model core-loss edges that neglect non-nearest-neighbor interactions.

The use of a direct crystallographic technique is reported for locating Cr atomic sites in a mullite containing 11.5 wt% Cr 2 O 3 by monitoring variations in characteristic X-ray emission rates as a func­tion of fast electron beam orientation. Systematic examination of two dimensional incoherent chan­neling patterns (ICP), formed from characteristic X-ray emissions from Al, Si, and Cr, and recorded near low index zone axis orientations, has enabled the preferred lattice position of Cr in mullite to be identified as the interstitial site 0, 0.25, 0. Although the method of atom location by channeling en­hanced microanalysis (or ALCHEMI) generally has been applied in situations where introduced mi­nority atom species are accommodated in substitutional atomic positions, this study illustrates the identification of an interstitial site of an introduced dopant species. This result does not coincide with that derived from X-ray Rietveld refinement. The ICP method is analytically robust and, unlike Rietveld refinement, does not require a highly accurate model of the host lattice framework and composition. ICP analysis therefore may be more appropriate for this particular application.

Raman spectra were measured and normal mode calculations were performed for phlogopite. Calculated fundamental mode frequencies were fit to observed frequencies assigned to features in the two polarized Raman spectra collected, so that unambiguous assignment of eigenmodes could be made to nearly all Raman spectral features. Average force constant values determined for Si-O and O-Si-O environments as well as for the non-silicate environments in cyclosilicates and gillespite provided an excellent starting point for the phlogopite calculations that quickly converged to a best fit of theoretical mode frequencies to the observed Raman fundamental mode frequencies. Final values of the force constants for T-O stretch and O-T-O bend are within the range of force constant values determined for Si-O and O-Si-O environments in cyclosilicates and gillespite. The force con­stant similarities extend to the K-O, M(1,2)-O bond stretch and M(1,2)-O-T bond bend environ­ments. Raman-active modes at frequencies greater than 850 cm -1 are localized intra-sheet modes dominated by T-O stretch and O-T-O bend motions. Raman-active modes between 350 and 850 cm -1 are dominated by mixtures of internal sheet modes with displacements from the M2 sites. Raman- active lattice modes are calculated at frequencies less than 350 cm -1 . Calculated frequencies for the IR-active A u and B u , modes generally fall within the frequency ranges of broad bands in the IR spectra for phlogopite presented elsewhere. Eigenmodes calculated for phlogopite are generally unique to that mica structure and differ from those determined for the cyclosilicates and gillespite. The few eigenmodes calculated for phlogopite that resemble those calculated for the cyclosilicates are the most localized T-O stretch motions and are assigned to the highest frequency modes.

We report paragenesis and chemistry of a new occurrence of the rare trioctahedral Al-rich sodium mica preiswerkite. The preiswerkite occurs in a tourmaline-biotite-scapolite rock in the contact zone of a gabbroic boudin surrounded by Proterozoic metasediments near the Blengsvatn, Bamble sector, southern Norway. The preiswerkite occurs as subhedral crystals or is intergrown with biotite in a polygonal fabric together with Cl-rich scapolite + tourmaline ± ilmenite ±plagioclase ±corundum. Accessory minerals are hematite, högbomite, spinel, allanite, apatite, and zircon, with relic calcite. Preiswerkite has the compositional range: (Na 1.84-2.02 K 0 . 02-0.10 Ca ≤0.04 )(Mg 3.13-3 . 42 Fe 0.63-0.77 VI Al 1.87-2.07 )( IV Al 3.58-3.96 Si 4.04-4.29 S ≤0.02 )O 20 [Cl ≤0.03 (OH) ≥3.97 ] and coexists with Na-Al-rich biotite, with the composition: (K 1.38-1.61 Na 0 . 18-0.45 Ca ≤0.03 )(Mg 3.72-3.88 Fe 1.38-1.43 Ti 0.10-0.16 VI Al 0.63-0.85 )( IV Al 2.71-2.93 Si 5.07-5.29 )O 20 [Cl 0.02 (OH) ≥3.98 ]. We suggest that the assemblage preiswerkite + biotite + tourmaline + scapolite ± ilmenite ± plagioclase ± corundum was formed during prograde or peak metamorphism in the area, at ~700 ℃ and 7 kb.

We report 17 O magic-angle spinning (MAS) NMR data for crystalline NaAlO 2 and CaAl 2 O 4 at external magnetic fields of 9.4 and 14.1 T, as model compounds for Al-O-Al sites in tetrahedral networks. The former contains one peak with isotropic chemical shift (δ iso ) = 30.9 ppm and quadrupolar coupling constant (C Q ) ≈ 1.8 MHz. The latter contains several peaks with δ iso ranging from 39 to 87 ppm and C Q ≈ 1.5 to 2.4 MHz. Triple-quantum MAS (3QMAS) spectra of sodium and calcium aluminosilicate glasses with Si/Al < 1 show clearly resolved peaks for Al-O-Al sites, removing ambiguity in the use of such spectra to explore the extent of aluminum avoidance in both glassy and crystalline materials. We also report 23 Na and 27 Al NMR data for the crystalline phases.

Crystals of BaSi 4 O 9 synthesized at 4 GPa and 1000 ℃ were determined to be isostructural with barium tetragermanate [trigonal, space group P3, a =11.2469(11) and c = 4.4851(6) Å, V= 491.3(1) Å 3 ]. The structure (R = 2.4%) features a corner-linked framework of three-member silicate tetrahedral rings, which are cross-linked by isolated silicate octahedra. Ba cations oc­cupy tenfold-coordinated sites in channels defined by the silicate framework. This structure is 4.2% denser than the topologically similar benitoite form of high-pressure BaSi 4 O 9 , which was produced by grinding the barium tetragermanate-type crystals described in this report.