Band 87, Heft 1

The electronic structure and energies of Au atoms adsorbed on an atomic cluster of kaolinite were calculated using the self-consistent-field Discrete Variational (DV) method. A hexagonal ring of SiO 4 tetrahedra and three AlO 2 (OH) 4 octahedra, comprising 38 atoms, was used to model the kaolinite crystal flake. An Au atom was used to model the submicrometer Au. Calculations were performed with Au atoms adsorbed at different sites. The systems with lower total energy are those with Au adsorbed on the edge surfaces. The adsorption of Au atoms on the edges is more stable than Au atoms adsorbed on the basal plane. Bond order calculations suggest that significant shifting of atomic charge and overlap of electron clouds between Au and the surface atoms of kaolinite takes place in the systems with Au adsorbed on the edges, especially at sites near Al octahedra.

Scanning probe microscopy (SPM) has become a common tool in mineralogy but distortion of images complicates interpretation and often limits the amount of information one can extract. Image distortion arises from a discrepancy between the intended and actual scan area caused by relative movement between tip and sample that is additional the intended scanning motion. We present a mathematical model to describe distortion in SPM images and provide a simple algebraic correction method. It uses Fourier periodicities for correcting high-resolution images; for micrometer-scale images, it can use any three non-colinear points that define a feature with known geometry. Observed distortion can be accounted for by two components: drift, the vector that quantifies the shape change of the intended scan area from a square into a parallelogram, and scaling, a constant that describes an isotropic change in dimension of the resulting scan area. The correction restores angular relationships and distances. The method was tested on the mineral graphite. In order to define the most important parameters affecting distortion, we made a sensitivity analysis by systematically varying temperature, scan speed, and time lapsed after the microscope was powered on. Neither drift nor scaling were found to be temperature dependent as such. However, both do depend on the time lapse after imaging begins. After the instrument is powered on, an initial 40 minute period of erratic drift is observed, whereafter drift velocity decreases with time while scaling increases slightly. Temperature variations in the range of 23 to 43 °C have negligible influence on distortion whereas scan speed affects scaling.

The microtopography of {001} surfaces on single crystals of graphite from a Neoproterozoic marble of the Swakop group, near Wlotzkas Baken, western Namibia, has been studied using differential interference contrast (DIC) microscopy and atomic force microscopy (AFM). A unique aspect of the observed surface microtopography is the presence of growth spirals and hillocks on three different length scales. The largest spirals are polygonized and can be seen without magnification. Steps on this feature are roughly 4 μm high and 90 μm apart. The second-order features are hexagonal growth hillocks with an average step height of 1.5 nm and total lateral dimensions of 5-40 μm. The apex of these hillocks coincides directly with the apex of reentrants in the macrosteps of the large spiral. Morphology suggests the formation of these polygonized hillocks by some mechanism other than simple spiral growth. We speculate that these features may be due to pinning of the macrostep by impurities and subsequent formation of the second-order hillocks. The third length-scale features are spirals found on terraces forming the vicinal faces of the second-order hillocks. These spirals have steps that are 6.7 Å high (unit-cell length along [001]) and an average step spacing of 900 Å. These double-layer steps also show some regions with partial step separation into 3.3 Å high monolayer steps. The observed microtopographic features give us insight into the conditions in, and mechanisms by which these graphite crystals formed during carbonate metamorphism. Crystal growth, unrestricted by the surrounding calcite, was from a fluid phase at low graphite supersaturation and was dominated by the spiral growth mechanism. Comparison with theoretical and simulation studies suggests a critical radius for two-dimensional nucleation on the (001) surface on the order of 100 Å.

We have studied short-range cation ordering across the diopside (CaMgSi 2 O 6 )-Ca-Tschermak pyroxene (CaAI 2 SiO 6 ) (Di-CaTs) solid solution in samples synthesized at 1400 °C and 2 GPa, for 24 hours. Peak positions in 29 Si MAS NMR spectra are sensitive to A1 substitution, both in the corner­sharing NN tetrahedral sites on the single chain and in one of the three NN octahedral Ml sites. The substitution of A1 for Mg on Ml causes the 29 Si chemical shift to be shielded by about the same magnitude as the deshielding caused by substitution of A1 for Si in NN tetrahedra, causing severe peak overlap among central peaks. Two pairs of the unique local environments have very similar chemical shifts, leaving only four peaks resolved in the spectrum, for which six site assignments have been made. 29 Si MAS NMR peak intensities for end-member CaTs are consistent with 70% of the Si in the tetrahedral chain being locally ordered into alternating [4] Al-O-Si-O- [4] Al linkages, although space group C2/c would suggest that Si and A1 are long-range disordered over the tetrahedral chain. CaTs clearly violates Lowenstein’s (1954) Al-O-Al avoidance principle, with 0.175 Al-O-Al linkages per formula unit (based on six O atoms per formula unit), as calculated from observed 29 Si MAS NMR peak intensities. For intermediate members of the solid solution, site populations have been mod­eled using a random distribution model (RD), an Al-O-Al avoidance model (AA) and a modified AA model which couples the cation distribution between the tetrahedral and octahedral sites (AAOC). The RD model fits the 29 Si MAS NMR peak intensities poorly. The AA and AAOC models fit the 29 Si NMR data well to at least 75% CaTs, suggesting that diopside-rich Di-CaTs clinopyroxenes obey Lowenstein’s Al-O-Al avoidance principle. For compositions >75% CaTs, observed cation distributions deviate considerably from that predicted by Al-O-Al avoidance, indicating formation of Al-O-Al linkages. The AAOC model yields an improved fit to the observed 29 Si MAS NMR intensities, particularly near Di 50 CaTs 50 , implying the presence of [4] Si-O- [6] Mg and [4] Al-O- [4] Al couples. This suggests the presence of short-range ordering on both tetrahedral and octahedral sites in Di-CaTs pyroxenes. The high sensitivity to octahedral A1 substitution observed in these chain silicates is not ob­served by 29 Si MAS NMR of more-polymerized silicates, such as framework and sheet structures. Thus, 29 Si MAS NMR of pyroxenes has potential to reveal tetrahedral-octahedral cation coupling and octahedral cation ordering information, which is potentially important for geothermobarometry of aluminous pyroxenes.

Gd-doped fluorapatite (1.2 ± 0.2 wt% Gd 2 O 3 ), synthesized from CaF 2 -rich melts, has been inves­tigated as single crystals and powder samples by using X-band (~9.4 GHz) electron paramagnetic resonance (EPR) spectroscopy at -295 and 120 K. The well-resolved X-band EPR spectra yielded a previously unreported type of Gd 3+ center “a” (S = 7/2) and also suggested the possible presence of a second and partly resolved type of Gd 3+ center “b.” In particular, the single-crystal X-band EPR spectra of center “a” from three orthogonal-rotation planes allowed determination of the spin-Hamil- tonian parameters, including the spin terms of type BS (matrix g) and S 2 (matrix D) and the param­eters associated with the high-spin terms of type S 4 and S 6 as well as BS 3 and BS 5 . The validity of the parameters has been confirmed by agreement between observed and simulated EPR spectra for both single crystals and powder samples. The principal values of the matrices g and D indicate that the local symmetry of center “a” in the X-band EPR spectra is rhombic. The principal axis directions of the D suggest that this Gd 3+ center arises from a substitution of Gd 3+ ion into the Ca2 type of site. This assignment is supported by the results of a pseudo-symmetry analysis using the S 4 parameters, e.g., the calculated twofold pseudo­symmetry axis coincides with the twofold rotoinversion axis of the Ca2 site. The local structural environment of this Gd 3+ ion suggests that the ion is incorporated via the mechanism Gd 3+ + 0 2- ↔ Ca 2+ + F-.

A W-band (94 GHz) electron paramagnetic resonance (EPR) study of synthetic fluorapatite with 57 ± 4 ppm Gd has been made on single crystals at ~287 K. The spectra disclosed the presence of a previously unreported type of Gd 3+ center denoted by “b” herein (S = 7/2), in addition to the Gd 3+ center “a” assigned to the Ca2 site using the results of a previous X-band (9.5 GHz) EPR study (Chen et al. 2002). In particular, the single-crystal W-band EPR spectra from three orthogonal-rota­tion planes allowed determination of an appropriate spin-Hamiltonian for center “b,” including the spin terms of type BS (matrix g) and S 2 (matrix D) and the parameters associated with the high-spin terms of type S 4 and S 6 as well as BS 3 and BS 5 . Agreement between the observed and simulated single-crystal spectra confirmed the validity of the spin-Hamiltonian analysis.β The principal values of the matrices g and D [e.g.. D/g e β e = 1069.2( 1) G and E/g e β e = 52.4(3) G] suggest a considerably distorted rhombic local environment for the Gd 3+ ions in center “b.” The principal directions of D suggest that “b” corresponds to Gd 3+ at the Ca 1 site. This site assignment is supported by a pseudo-symmetry analysis of the term S 4 , i.e., approximate matching of the direc­tions of the calculated pseudo-symmetry axes to the bond directions and face normals of the coordi­nation polyhedron of the ideal Cal site. The data suggest that the incorporation of Gd 3+ into the Cal site is achieved by a coupled substitution (2Gd 3+ + □ ↔ 3Ca 2+ ) involving a Ca 2+ vacancy □ and that the vacancy is located at a next-nearest-neighbor Ca2 site, resulting in a Gd 3+ - - □ - Gd 3+ arrange­ment, with the cations well separated.

A method of determining the boric acid (H 3 BO 3 ) concentration in fluid inclusions and volatilerich glasses using laser Raman microprobe spectroscopy is described. The H 3 BO 3 content of melt and fluid inclusions is determined with good accuracy and precision in the concentration range of 0 to 20 wt%, and with a minimum detection limit (mdl) of 0.05 wt% H 3 BO 3 . The method is applicable to complex solutions that contain components such as NaCl, KCl, RbCl, or CsCl, which do not form stable hydrated borates with boric acid at room temperature. This Raman study demonstrates that boric acid is an abundant component in volatile-rich melt inclusions and in genetically linked primary and pseudosecondary fluid inclusions, particularly in tourmaline-bearing granites and boron-rich pegmatites. In quenched pegmatite melt boron occurs predominantly as a trigonal coordinated B(OH) 3 complex. The analytical method is applied to a series of melt and fluid inclusions from different localities and geological environments; namely hydrothermal tin and tungsten deposits, granite pegmatites, and rhyolites. Special emphasis is given to the analysis of fluorine-, phosphorus-, and boron-rich pegmatites of the Variscan Ehrenfriedersdorf complex (Central Erzgebirge, Germany). In this pegmatite system, boron in combination with water, fluorine, and phosphorous is responsible for immiscibility of two coexisting silicate melts, a silicate- rich H 2 O-poor melt and a silicate-poor H 2 O-rich melt accompanied by strong element fractionation between the two. Boron fractionates into the water-rich silicate melt by factors between 1.8 at 500 °C and 1.0 at 712 °C. If, at low pressure, a vapor phase is additionally formed, boron is partitioned into the vapor phase. The utility of the developed method was also quantitatively tested by experiments with a hydrothermal diamond anvil cell.

A hydrothermal diamond anvil cell (HDAC) was used to heat natural H 2 O + CO 2 + NaCl fluid inclusions in quartz. Heating in the HDAC allows external pressure (the fluid pressure on the host mineral) to be regulated so the total homogenization of the aqueous and carbonic phases at high temperature can be observed and recorded with minimal risk of decrepitation. The total homogenization temperature, the CO 2 -clathrate melting temperature, and the homogenization temperature of the carbonic phase can be used precisely to calculate the pressure of total homogenization (on the solvus) and the corresponding isochore. The equation of state of Duan et al. (1995) was used to calculate the solvus pressure by an iterative technique of varying the mole fraction CO 2 and the solvus pressure with a given total homogenization temperature and NaCl:H 2 O molar ratio. Bulk molar volume values of the calculated fluids were iterated into a mixing equation for the aqueous and carbonic phases at low temperature until the fluid compositions, predicted by both equations, were identical. Quartz samples from the Adirondack Mountains, New York, were used in the heating experiments. Seven H 2 O + CO 2 + NaCl fluid inclusions from post-metamorphic, hydrothermal cavities near Moose River yielded homogenization temperatures of 185-232 °C and homogenization pressures of 863 to 1870 bars. Two H 2 O + CO 2 + NaCl fluid inclusions from a drill core at the Willsboro wollastonite mine yielded homogenization temperatures of 310 °C and 324 °C and homogenization pressures of 3893 and 5333 bars, respectively. These inclusions are interpreted to have formed along a retrograde path characterized by early isobaric cooling and later isothermal decompression following Ottawan-age metamorphism.

A thermodynamic model has been developed and calibrated for monoclinic and orthorhombic amphiboles compositionally contained within the Ca-Mg-Fe 2+ amphibole quadrilateral. The model incorporates the energetic consequences of cation ordering of Fe 2+ and Mg over nonequivalent sites in the crystal structure, and accounts for the temperature, pressure, and compositional dependence of the orthorhombic-monoclinic phase transition. Calibration is based on previously published work on the thermodynamic properties of Fe 2+ -Mg amphiboles and tremolite, and experimental cation-ordering data, along with solvus width and tieline orientations of natural coexisting amphiboles in the quadrilateral. Among derived parameters is the enthalpy of formation of end-member ferroactinolite (-10534.966 kJ/mol). A calculated FeMg -1 isopotential solvus at mid-composition in the quadrangle agrees well with a revised calibration of experimental data from Cameron (1975). The solvus is not strongly asymmetric and narrows with increasing Fe/Mg ratio and temperature. Phase relations along the Mg and Fe sides of the quadrilateral are compared and contrasted. Cummingtonite-anthophyllite phase relations are shown in the T-X Fe plane, with and without coexisting calcic amphibole. The breakdown reactions of quadrilateral amphiboles to assemblages of two pyroxenes, olivine, quartz, and H 2 O are depicted as functions of temperature, pressure, composition, H 2 O activity, and oxygen fugacity. Some highlights of their relevance to metamorphic and igneous systems are discussed.

In situ synchrotron X-ray experiments of the GeO 2 system were made at pressures of 28-45 GPa and temperatures of 300-2300 K, using a diamond anvil cell combined with a laser heating and a 6- 8 type multianvil high-pressure apparatus. We observed a second-order phase transition between tetragonal rutile-type (P4 2 /mnm) and orthorhombic CaCl 2 -type (Pnnm) phases under high pressure and temperature. The transition kinetics seem to have little effect on the second-order phase transition because the cell constants exhibit no discontinuities between the phases. Therefore, the phase transitions could be observed at low temperature conditions in this study. The phase boundary was determined to be P (GPa) = (34.9 ± 1.2) + (0.0086 ± 0.0024) × (T - 1300) (K).

The nature and stability of end-member illite (I), K 0.88±0.01 /O 10 (OH) 2 , a K-deficient mica, has been a subject of much controversy. Evidence for the metastability of “illite” with respect to ideal muscovite (Ms) + pyrophyllite (Py) has been discussed in the literature but conflicting evidence from studies of natural and synthetic systems point to the stability of K-deficient micas with respect to Ms below ~360 °C. Py coexists with K-deficient micas in natural assemblages thought to have formed between 300 and ~360 °C. Available evidence suggests that end-member illite (I) has an ordered, domain structure. Structural strain due to K-deficiency in micas may be accommodated by the creation of Py domains. The resulting structure could be stable with respect to Ms + Py at low temperatures only if the domains are ordered. Recent FTIR studies have established the presence of local Py domains and molecular water in Al-rich illite. Py stability (~360 °C at 1 kbar) limits the stability of the domain structure; Ms coexists with andalusite + water at 400 °C. The prograde, stepwise transformation of S (smectite) → I-S → I probably leads to a metastable, compositional end-member (I w ) containing “excess,” interlayer water. Dehydration of I w results in the formation of a metastable, disordered K-deficient mica (I d ) that may recrystallize to form a stable, ordered, Py domain structure (I o ). I w may persist metastably but solid-state alteration of Ms can only yield I o . Fine-scale, Py domains have not been observed in K-deficient micas by TEM as yet, but available evidence suggests that end-member illite (I o ) is a distinct, ordered, domain structure, stable below ~360 °C.

Melanterite (FeSO 4 ·7H 2 O)-rozenite (FeSO 4 ·4H 2 O) and chalcanthite (CuSO 4 ·5H 2 O)-bonattite (CuSO 4 ·3H 2 O) equilibria were determined by humidity measurements at 0.1 MPa. Two methods were used; one is the gas-flow-cell method (between 21 and 98 °C), and the other is the humiditybuffer method (between 21 and 70 °C). The first method has a larger temperature uncertainty even though it is more efficient. With the aid of humidity buffers, which correspond to a series of saturated binary salt solutions, the second method yields reliable results as demonstrated by very tight reversals along each humidity buffer. These results are consistent with those obtained by the first method, and also with the solubility data reported in the literature. Thermodynamic analysis of these data yields values of 29.231 ± 0.025 and 22.593 ± 0.040 kJ/mol for standard Gibbs free energy of reaction at 298.15 K and 0.1 MPa for melanterite-rozenite and chalcanthite-bonattite equilibria, respectively. The methods used in this study hold great potential for unraveling the thermodynamic properties of sulfate salts involved in dehydration reactions at near ambient conditions.

Garnets are present in two horizons of the schist of Stevens Spring from the Basin Creek area of the Grouse Creek Mountains in northwest Utah. The two horizons possess different bulk compositions, which resulted in garnet growth by different reactions along the same pressure-temperature (P-T) path. Garnet in the upper horizon grew from the breakdown of chlorite at upper greenschistfacies conditions and garnet in the lower horizon grew from the breakdown of staurolite at upper amphibolite-facies conditions. From the upper horizon, five garnets from three samples were analyzed. All display growth zoning, ragged morphologies associated with secondary rim consumption, and yield garnet-biotite geothermometry temperatures of ~460-490 °C. From the lower horizon, one garnet from each of two samples was analyzed. These also display growth zoning, but differ from garnet in the upper horizon in that they are dominantly idioblastic and yield garnet-biotite geothermometry temperatures of ~635 °C. Garnet-biotite geothermometry calculated for every point along detailed compositional traverses across the garnets revealed localized reequilibration along rims, cracks and inclusions in both generations of garnet. Garnets from the upper horizon display prograde reequilibration and the garnets from the lower horizon display retrograde reequilibration. Numerical simulations of garnet growth using the Gibbs method with Duhem’s theorem were carried out to determine P-T paths. The P-T path defined by the modeling of five garnets from the upper horizon is an isothermal pressure increase of ~1.7 kbar. The P-T path defined by the modeling of two garnets from the lower horizon has a steep P-T trajectory (dP/dT = 32 bars/°C) and a total pressure change of ~0.9 kbar. Both paths are indicative of thrust burial; however, the two paths cannot be reconciled as products of a single thrust episode. These data are interpreted to indicate two different episodes of thrust burial during the Sevier Orogeny, separated by ~150 °C of heating and partial exhumation.

Dark pyroxene crystals found in a calcite vein cutting across marbles and skarns at Andranondambo, southern Madagascar, were investigated with various techniques in order to characterize and identify the mineral phase. Methods applied include powder X-ray diffraction, thinsection microscopy, electron-microprobe analysis, densimetry, and thermogravimetry coupled to mass spectrometry (TGA-MS), FTIR, and Mössbauer spectroscopy. The chemical composition of the mineral phase was determined as {Ca 0.98 ,Na 0.03 }[Mg 0.68 ,Fe 2+ 0.07 Fe 3+ 0.03 ,Al 0.16 ,Ti 0.04 ](Si 1.77 ,Al 0.23 )O 6 . The refined unit-cell parameters are: a = 9.719(2) Å, b = 8.838(2) Å, c = 5.287(1) Å, β = 106.05(2)°, and V = 436.33(6) Å 3 . The refractive indices are: α = 1.690 ± 0.002, β = 1.700 ± 0.002, and γ = 1.712 ± 0.002. The density is 3.370 (55) g/cm 3 . The mean refractive index calculated from the chemical composition yields a Mandarino compatibility index of K = -0.0167. From the various analytical data, we were able to conclude that the mineral corresponds to Fe- and Ti-bearing aluminian diopside. TGA-MS data suggest the presence of calcite solid inclusions and minor H 2 O and CO 2 fluid inclusions. Chemical analyses of the calcite host matrix revealed clear rare earth element enrichment. Mössbauer spectra obtained at varying temperatures show the presence of Fe 2+ and Fe 3+ . Both exhibit broad quadrupole-splitting distributions, indicating widely fluctuating chemical and structural environments for the probe Fe nuclei. The broad temperature range (80-470 K) over which the spectra were collected resulted in a precise determination of the Fe 2+ /Fe tot ratio as 0.68, which is consistent with the chemical composition as determined from the microprobe analyses. For comparison, the Mössbauer spectra of two specimens of the type locality of aluminian diopside (the mineral formerly named “fassaite”) were analyzed as well and found to be consistent with those obtained for the Madagascar crystals. The relation between mineral paragenesis and local geological features is briefly discussed.

To understand compositional variation in igneous biotite, full analyses of a suite of biotites of variable composition from volcanic and xenolith parageneses have been completed. Major and minor elements were determined by electron microprobe analysis, water was determined by manometry and SIMS analysis, and Fe 3+ /Fe 2+ was determined by microXANES and Mössbauer spectroscopy. Our new data, together with previous biotite analyses (total of 52 analyses), reveal correlations between O 2- (2-F-Cl-OH) and the sum of the octahedral cations Al + Ti + Fe 3+ + Cr. This correlation allows estimation of either OH - or Fe 3+ /Fe 2+ as long as one or the other has been determined. The hydroxyl site in most mantle micas contains at least 1.0 O 2- atoms per formula unit (apfu), indicating that the oxy-component cannot be ignored. The large oxy-component in melt inclusion micas from the martian meteorite Chassigny does not necessarily indicate oxidized or hydrous magmatic conditions because dehydrogenation may have occurred and/or because the oxy-component may be stable at low oxygen fugacity. The large variation in Ti, Al, and Fe 3+ in natural igneous micas is most likely dependent upon bulk compositional differences in each specific system such as variation of a TiO₂ and a Al₂O₃ in silicate melts.

The proton positions in leucophoenicite, ideally Mn 7 Si 3 O 12 (OH) 2 , have been determined by neutron powder diffraction under ambient conditions on a natural sample from Franklin, New Jersey. Refinement in the P21/a space group gave R p = 2.0% (wR p = 2.1%), χ 2 = 4.04 for 110 refined parameters. The two non-equivalent protons form a pair of hydroxyl groups and make hydrogen bonds to the same O7 atom of the Si1 tetrahedron at distances of 1.99(1) and 2.09(1) Å. The H····H distance in leucophoenicite is 2.16(1) Å, which is more than twice the Van der Waal’s radius of H (>2 Å) and so no proton positional disorder is expected in leucophoenicite. A comparison of the H environments is made between leucophoenicite and Phase B, Superhydrous B, and Phase A.

Ominelite, (Fe,Mg)Al 3 BSiO 9 , is the Fe 2+ analog of grandidierite. The mineral occurs as elongated and euhedral to equant and anhedral grains in close association with sekaninaite (Fe-dominant analogue of cordierite), garnet, biotite, andalusite, topaz, alkali feldspar, plagioclase, muscovite, quartz, dumortierite, schorl, zircon, ilmenite, apatite, monazite, and pyrite in a porphyritic granite of Miocene age exposed along the Misen River in Tenkawa, Yoshino, Nara Prefecture, Japan (34°12'40''N, 135°53'40''E). Temperatures <700 °C and pressures below 4 kbars are suggested for the formation of ominelite and associated sekaninaite, topaz, andalusite and dumortierite. The Al-rich minerals could be either magmatic or restitic in origin. A representative electron microprobe analysis of ominelite is SiO 2 19.34, TiO 2 <0.01, Al 2 O 3 48.85, FeO 19.37, MnO 0.43, MgO 1.33, CaO <0.01, P 2 O 5 0.13, B 2 O 3 10.91 wt%, total 100.36 wt%, corresponding to Fe 0.85 Mg 0.10 Mn 0.02 Al 3.01 B 0.99 P 0.01 Si 1.01 O 9 . Mohs’ hardness is about 7. No cleavage is observed. Its color is blue, and the streak is pale blue. It is pleochroic X = Z = pale blue-green and Y = colorless. Optically, it is biaxial (-) and, at λ = 589 nm, has α = 1.631 (1), β = 1.654 (1), γ = 1.656 (1), 2V X (meas.) = 31.5 (6)°. Y = c (prism elongation direction). Dispersion is v >> r. Major lines in the powder pattern [d in Å, (I), (hkl)] are 5.57(m)(020), 5.21(vs)(200), 3.73(m)(121), 3.51(m)(130), 2.97(s)(101), 2.79(s)(040), 2.18(s)(150, 421, 312). Space group is Pbnm. Lattice parameters are a = 10.343 (2), b = 11.095 (1), c = 5.7601 (8) Å and V = 661.0(2) Å 3 , Z = 4, D calc = 3.169 g/cm 3 . Refinement of the structure confirms that ominelite is isostructural with grandidierite with no detectable substitution of Al by Fe 3+ .

Crystal size distribution (CSD) measurements are susceptible to the closure problem, just like chemical compositions. In its simplest form this means that the total crystal content of a rock cannot exceed 100%. Where chemical or thermal effects limit the total quantity of a single phase, closure can occur at lower volumetric phase proportions. This means that parts of the CSD diagram [ln(population density) vs. size] are not accessible. If the volumetric phase proportion is constant, then straight CSDs will appear to rotate around a point at small sizes giving a fan of CSDs. These fans are significant and do show changes in crystal sizes that can be interpreted in terms of magmatic processes. However, the correlation between the slopes (or characteristic lengths) and intercepts of individual CSDs in a family is not significant, but just a consequence of the constant phase proportion effect. Many other graphs, such as characteristic length vs. volumetric phase proportion, can give more information on magmatic processes.

A new type of framboidal pyrite, with icosahedral domains, is described in this study. Examining the microcrystals on sections of framboids from various localities using a scanning electron microscope, we found pentagonal and trigonal patterns. These are made up of rectangular and fan-shaped domains, and octahedral microcrystals are regularly linked by sharing of edges in each domain. These symmetrical arrangements are interpreted to be different sections of icosahedrally arranged framboids which are composed of twenty tetrahedral domains. Thus some pyrite framboids are not spherical, but are fundamentally icosahedral both in appearance and internal structure. The formation of the icosahedral framboids might be related to the initial nucleation rate and the number of microcrystals within each framboid.