The Local Environment of Fe in Dry and Water-bearing Silicate Glasses

Max Wilke (max.wilke@mineralogie.uni-hannover.de)¹, Francois Farges, Harald Behrens², Dorothee Burkhard³ & Pierre-Emanuel Petit⁴

¹ Laboratoire des Géomateriaux, Université de Marne-la-Vallée, F-77454 Marne la Vallée Cedex 2, France

² Institut für Mineralogie, Universität Hannover, D-30167 Hannover, Germany

³ Institut für Mineralogie, Universität Marburg, D-35032 Marburg, Germany

⁴ European Synchrotron Facility, BP 220, F-38042 Grenoble Cedex, France

We present a combination of Mössbauer and high-resolution Fe-K edge XANES spectra collected on Fe-bearing (0.4-1.0 wt.% Fe) glasses of the haplotonalitic ternary SiO₂(Qz)-NaAlSi₃O₈(Ab)-CaAl₂Si₂O₈(An), on Fe-bearing sodium trisilicate glass and several natural glasses (trachyte and pantellerite from Italy, high-silica rhyolite from Mexico). Three dry glass compositions in the haplotonalitic ternary were investigated: Ab, An and Qz₄₈Ab₂₂An₃₀. Hydrous glasses of Ab and Qz₄₈Ab₂₂An₃₀were synthesized in an internally heated pressure vessel at 200 - 500 MPa and 850 - 1000°C. The Fe²⁺/Fe³⁺ ratios measured for most glasses (ranging from 1 to *) were obtained from the ⁵⁷Fe-Mössbauer spectra. Fe-K edge XANES spectra were collected at the European Synchrotron Radiation Facility (Grenoble, France) on beamline ID26 using Si(220) double-crystal monochromator (energy resolution = 1.5 eV). In addition XANES spectra of a series of Fe-model compounds for ferric and ferrous iron. In these compounds the ferrous iron is found in tetrahedral (staurolite, hercynite), square planar (gillespite), trigonal bipyramid (grandidierite), octahedral (fayalite, hedenbergite, cummingtonite,...) and hexahedral (almandine) site geometry. For the ferric iron the studied geometries included tetrahedral (FePO₄, Fe:LiAlO₂, Fe:orthoclase), trigonal bipyramid (yoderite) and octahedral (goethite, hematite, andradite,...) sites. The spectra of the pre-edge in those compounds very often show one peak (^[4]Fe(III), ^[5]Fe(III),), 2 peaks (^[4]Fe(II), ^[5]Fe(II), ^[6]Fe(III)), 3 peaks (^[6]Fe(II)), sometimes 5 (hematite). In any case, the centroid of the pre-edge between Fe³⁺and Fe²⁺ is separated by about 1.4 eV. All glasses show a split pre-edge showing a major peak around 7114.5 eV and a shoulder at ~ 7112.5 eV. Relative integrated intensities decrease with decreasing redox ratio. Therefore, the high energy contribution may be attributed to ferric iron and the one at lower energy to ferrous iron. This is consistent with the observations made on model compounds. Alkali-rich glasses (Ab) show significantly higher intensity preedges than the calcic ones (An), suggesting compensation effects around Fe(III) which promotes lower Fe(III) coordinations (6) for the latter, as previously observed for Ti and Ni (Farges et al., 1996, Farges et al. in prep. On one hand, the observed changes of the pre-edges upon incorporation of water certainly reflect changes in the redox state owing to different redox conditions in the synthesis. Comparison with calculated mixtures of ferric and ferrous model compounds, however, reveals that there are also changes in the coordination environment of Fe. Fe³⁺ is in tetrahedral coordination in most glasses which is also consistent with observed isomer shifts for Fe³⁺ in Mössbauer spectra. Site assignment for Fe²⁺ is not straightforward, due to its relatively lower pre-edge contribution. However, the data analysis suggests five-fold site geometry and/or mixtures of four and six-fold geometries or five and six-fold geometries. Differences are observed in the XANES region between the dry and the hydrous glasses. Two additional features at the crest of the edge are observed for the hydrous glasses which are not resolved in the dry counterparts. The changes are stronger for the tonalitic than for the albitic glass. Towards more reducing conditions one of these features increases considerably. These changes may be attributed to changes in the mesoscopic environment of Fe. If the quench rate is too low, the local structure around Fe is much more ordered on a mesoscopic scale (clinopyroxene-type nanocrystals?), with strong Fe-Fe interactions which is shown by a spectrum of a sample quenched slowly through Tg. The Mössbauer spectrum collected on this sample show also significant changes compared to the normally quenched glass. In contrast, no crystals were detected by XRD and polarization microscopy. Similar interpretations can be done for natural hydrous glasses, the pantellerite showing well defined clusters around Fe(II) (O_h symmetry), related to highly variable thermal history in lava generation.

Farges F, Brown GE jr., Navrotsky A, Gan H., Rehr JJ, Geochim. Cosmochim Acta, 60, 3039-3053, (1996).

Trace Element Distribution During Eclogite Formation from Plagioclase Cumulates

Antje Wittenberg (wittenebrg@mineralogie.uni.hannover.de)¹ & Cliff Shaw (cliff.shaw@uni-bayreuth.de)²

¹ Institut für Mineralogie / Universität Hannover, Welfengarten 1, D-30167 Hannover, Germany

² Bayerisches Geoinstitut / Universität Bayreuth, Universitätsstraße 30, D-95447 Bayreuth, Germany

In order to investigate the distribution of trace elements including REE in Ca-, Al-rich eclogites (grospydites) we have performed high-pressure piston cylinder experiments (Bayerisches Geoinstitut) on glasses of natural plagioclase-rich cumulates doped with 500 ppm of Rb, Sr, Zr, Hf, REE, Th and U. Experimental conditions were 1100°C at 3.0 to 3.5 GPa for times of 86, 168 and 184 hours in Ag₂₅Pd₇₅ capsules. The total iron loss was < 0.5 wt.%. The aim of these experiments is to understand the phase transformations and element partitioning that occurs in delaminated continental crust.

Anhydrous experiments produced garnet, kyanite, clinopyroxene, and glass as well as very fine-grained trace element-rich phases (allanite-type, oxides). The small grain size (< 15 µm), the impurity of crystals (inclusions, zoning) and the heterogeneous distribution of the trace element-rich phases prohibits in-situ trace element measurements.

Addition of 2 wt.% H₂O yields fully crystallized assemblages of garnet, kyanite, clinopyroxene; in these experiments the fine-grained trace element-rich phases are absent. These run products differ from natural grospydites by the presence of abundant zoisite. Their grain size of up to 30 µm allows in-situ trace element partitioning measurements.

Small amounts of water seem to be essential in the formation of eclogite. The appearance of fine-grained trace element enriched phases under dry conditions points to difficulties in forming homogeneous mineral compositions on a hundred-ppm trace element level and may related to the very slow diffusion rate. The production of relatively large amounts of zoisite is an unforeseen effect in the hydrous experiments however, addition of water yields a homogeneous trace element distribution among the mineral phases probably due to the faster distribution rate. Comparison of EMP mineral data from natural grospydites and the experimental assemblages will allow us to better understand the partitioning of trace elements in natural grospydites.