Experimental Constraints on Fluid Absent Melting in Deeply Subducted Crust

Jörg Hermann (joerg.hermann@anu.edu.au) & David H. Green

Research School of Earth Sciences, The Australian National University, Canberra 0200, Australia

Dehydration and partial melting of subducted crust is an important process for mass transfer from the slab to the mantle wedge. Various rock types of deeply subducted continental and oceanic crust display the paragenesis coesite+kyanite+garnet+clinopyroxene+phengite. The simplest chemical system to produce this paragenesis consist of K₂O-CaO-MgO-Al₂O₃-SiO₂-H₂O (KCMASH). Synthesis piston cylinder experiments were carried out in the range 700-1150°C and 2.0-4.5 GPa in the KCMASH + trace elements system, saturated in kyanite and quartz/coesite. Fluid absent melting and fluid release is determined by the stability of hydrous phases such as amphibole (stable at T<750°C, P<3.0 GPa) and clinozoisite (stable at T<850°C, P<3.0 GPa). Talc was observed at T<730°C, up to 3.5 GPa. Below 3.0 GPa biotite has the highest thermal stability up to temperatures of about 900-950°C and disappears during the melting reaction (I) bt+ky+qtz -> opx+melt. At higher pressures biotite reacts to phengite. Phengite melting occurs along the reaction (II) phe+cs+cpx -> grt+ky+melt, which runs from 900°C, 3.0 GPa to 1050°C, 4.5 GPa. Both fluid absent melting reactions produced peraluminous granitic melts. Trace element partitioning between bulk rock and melt was determined by Laser Ablation ICP-MS. Rb, Sr, Ba preferentially enter the melt (D rock/melt ~0.1) whereas MREE (~3), Y (~8) and HREE (~15) are enriched in the residue. Surprisingly, the LREE (D rock/melt ~1.3-1.7) are compatible in the residue. This is most probably caused by the presence of allanite, which is present as accessory mineral. The partitioning of LREE between allanite and melt is in the order of ~300, demonstrating that the LREE are highly compatible in allanite. The transfer of LREE from the subducted slab to the overlaying mantle wedge by melts/fluids originating from dehydration of subducted crust is thus restricted as long as allanite is stable.

Viscosities of Granitic (sensu lato) Melts: Influence of the Anorthite Component

Kai-Uwe Hess (kai-uwe.hess@uni-bayreuth.de)¹, Donald B. Dingwell¹ & Claudia Romano²

¹ Bayerisches Geoinstitut, Uni Bayreuth, D-95502 Bayreuth, Germany

² Uni Roma III, Lgo San Murialdo, Roma, Italy

The viscosities of a series of granitic (sensu lato) melts have been determined in the range of 10³ to 10¹² Pas. The anhydrous melt compositions are based on the addition of 10, 20, 50 and 75 wt% of the anorthite component to a haplogranitic melt (HPG8) whose composition lies near the 2 kbar water-saturated minimum melt composition in the Ab-Or-Qz system. Melts with 10 and 20 wt% normative anorthite were subjected to high pressure hydration syntheses using a piston cylinder apparatus to generate water contents up to 2 wt%. Viscosities were determined for anhydrous melts using a concentric cylinder apparatus in the low viscosity range and for for all melts using the micropenetration method in the high viscosity range. The results indicate that the influence of the anorthite component on the viscosity of a haplogranitic melt is strongly temperature dependent. At high temperatures and low viscosities, normative anorthite strongly reduces melt viscosity. At low temperatures and high viscosities, the addition of normative anorthite results in a rotation of the viscosity-temperature relationships to steeper slopes, with the net result that the viscosity changes little. The melts become increasingly non-Arrhenian with added normative anorthite. The addition of water to melts with 10 and 20 wt% normative anorthite are adequately reproduced using the model of Hess and Dingwell (1996) whereas the viscosities of melts with higher anorthite contents are higher than predicted. At the temperatures anticipated for intermediate granitic magmatism, the calc-alkaline model can adequately reproduce the viscosity-temperature-water content relationships for melts with up to 15 wt% anorthite in the norm.

Exchange Mechanism of Zinc in Grossular-rich Garnets

Soraya Heuss-Aßbichler (soraya@petro1.min.uni-muenchen.de)

Institut für Mineralogie, Petrologie & Geochemie der LM-Universität, Theresienstr. 41, 80333 Munich, Germany

The distribution of Zn between grandite and clinopyroxene was determined for calcsilicate rocks and skarns of several localities. Two different trends were observed: generally zinc is enriched in clinopyroxene, with D_gt-cpx (=wt% Zn in Gt / wt% Zn in Cpx) < 0.1; higher Zn-concentrations in some grandites coexisting with clinopyroxene result in D_gt-cpx up to 1.4. The amount of an element in a phase depends on the bulk-composition of the reacting system. But it depends also on the ability to include this element in the crystal structure of this phase. There are a few experimental studies of the location of Zn in the crystal structure of clinopyroxene, but no experimental data exist for grandite. Theoretically Zn may take place in all three sites of the crystal structure of garnet: 1) the dodecahedral sites according to the exchange reaction [Ca²⁺_dod: Zn²⁺_dod], 2) the octahedral sites with [Al³⁺_okt: (Zn²⁺_okt + H⁺)], 3) the tetrahedral sites with [Si⁴⁺_tetr: (Zn²⁺_tetr + 2H⁺)]. Hydrogen is included due to charge balance. Experiments according to the three different exchange reactions were conducted in the temperature range of 600°- 900°C and the pressure range of 0.2 - 2 GPa. Based on the measured changes in cation proportions a different exchange mechanism was deviated involving both, the dodecahedral and the tetrahedral site in the garnet structure. The theoretical end-member of the exchange reaction [(Ca²⁺_dod +Si⁴⁺_tetr): (Zn²⁺_dod +Al³⁺_{tetr dod}+H⁺)] is Zn₃Al₅O₉(OH)₃. This exchange mechanism corresponds with the hydrogrossular-substitution [Si⁴⁺_tetr : 4H⁺] in grandites. According to the experimental results high Zn-contents in grandites may be expected at Si-under-saturated and Al-rich reaction conditions. Low temperature conditions enhance Zn-rich grandite solid solutions, in accordance with the stability conditions of hydrogrossular. A shift of the stability limits of Zn-grandites to higher pressures are expected.

The Iron Oxidation State and Systematic Variations of the Fe L X-ray Emission Lines in Synthetic Garnets Measured with the Electron Microprobe

Heidi Eva Höfer (hoefer@em.uni-frankfurt.de) & Gerhard Peter Brey (brey@em.uni-frankfurt.de)

Institut für Mineralogie, Senckenberganlage 28, D-60054 Frankfurt, Germany

The knowledge of the iron oxidation state in garnets is petrologically important for the correct application of thermobarometers based on the exchange of Fe²⁺ and Mg²⁺. Variation of the Fe³⁺/Fe²⁺ ratio yields a change of both the intensity and wavelength of the Fe L(alpha) and Fe Lß X-ray emission lines which may be determined with the electron microprobe. These low-energy emission lines are also sensitive to chemical bonding, crystal structure, and possibly coordination polyhedra, so that isostructural materials need to be considered individually. We investigated synthetic garnet solid solutions which may be subdivided into four groups with respect to their ferric iron content and crystallographic properties concerning cation substitution on the dodecahedral or octahedral sites (ds, os): (I) Fe³⁺/(sum)Fe = 0: almandine - pyrope - grossular - spessartine - knorringite, i.e., replacement of Fe²⁺ by Mg²⁺,Ca²⁺,Mn²⁺ on ds, and also of Fe³⁺ by Cr³⁺ on os; (II) Fe³⁺/(sum)Fe variable: almandine - skiagite, Al³⁺ - Fe³⁺ on os; (III) Fe³⁺/(sum)Fe variable: andradite - skiagite, Ca²⁺ - Fe²⁺ on ds; (IV) Fe³⁺/(sum)Fe = 1: andradite - grossular, Fe³⁺ - Al³⁺ on os. We measured these garnets with the electron microprobe using the "flank method" (Höfer et al., 1994, 2000), and interpreted the resulting Lß/L(alpha) intensity ratios in conjunction with Mössbauer and crystallographic data (Woodland & Ross II, 1994). The identical slopes of group I and III garnets in an intensity ratio/total iron diagram indicate that these ratios are independent of cation substitutions in the garnets. Exploiting the slope to correct for self absorption yields a linear relationship between these corrected intensity ratios and the ferric iron content with all the garnet solid solutions plotting on this line. Thus Lß/L(alpha) intensity ratios can be used unambiguously to determine the ferric iron content in garnets. The results also demonstrate that the self absorption effect consists of two components which have to be treated separately: one accounting for total iron and the other for ferric iron implying a different mass absorption behaviour for ferric and ferrous iron.

Höfer HE, Brey GP, Schulz-Dobrick B & Oberhänsli R, Eur. J. Mineral, 6, 407-418, (1994).

Höfer HE, Weinbruch S, McCammon CA & Brey GP, Eur. J. Mineral, 12, in press, (2000).

Woodland AB & Ross II CR, Phys. Chem. Minerals, 21, 117-132, (1994).