Some Constraints on Material Properties Relevant to the Earth's Inner Core

Andrew P. Jephcoat (andrew@earth.ox.ac.uk)

Department of Earth Sciences, University of Oxford, Oxford OX1 3PR, UK

The range of alloying components responsible for lowering the Earth's core density below that of pure iron is large, with perhaps no dominant element responsible for the overall density deficit. Carbon is one possible candidate and may have been alloyed with liquid iron under conditions of Earth's early formation. Thermodynamic calculations also suggest possible precipitation of a carbide phase from sulphur-rich Fe melts to form a solid Fe-C phase within the inner core. Compression data for phases in the iron-carbon system are sparse and we report measurements of the equation of state (EOS) and phase-transition behaviour of Fe₃C (cohenite, orthorhombic, S.G. Pnma) to 50 GPa performed at the European Synchrotron Radiation Facility (ESRF) in a diamond-anvil cell. Cohenite occurs as one of the intermetallic phases in quenched iron melts and the synthetic phase (cementite) was obtained by electrochemical extraction from an annealed high-carbon steel (0.5wt% C) containing some Mn. The diffraction data showed no evidence for a structural phase transition and a Vinet EOS fit gave a bulk modulus K₀ = 162 GPa with K'=6.4.

Models of the seismologically-observed elastic anisotropy of the inner core have depended so far on experimental measurements and theoretical predictions of the elastic moduli of hcp iron under pressure. The transverse-optical, doubly-degenerate (E₂ g) phonon mode in hcp metals is Raman active and can provide information on their dynamics under compression. Here we report new optical measurements of this phonon in rhenium to 60 GPa and other hcp metals that can place a direct constraint on the pressure-dependence of the elastic constant C₄₄. The calculated pressure dependence of this constant agrees well with first-principles theoretical predictions in contrast with the measurements derived from recent nonhydrostatic, lattice-strain, x-ray techniques.

Viscosity of Silicate Melts

Christophe Journeau (cjourneau@cea.fr), Muriel Ramcciotti & Gérard Cognet (cognet@cea.fr)

CEA Cadrache, DTP 6 SMET, 13108 St Paul lez Durance, France

Silicate melts have a high viscosity due to the formation of networks of tetrahedric SiO₄. Urabin's model [1,2] considers that a silicate melt is made of glass former (SiO₂, GeO₂,...), network modifiers ( MgO, CaO,...) which contribute to decrease melt viscosity and amphoteric materials (Al₂O₃...) which can act either as glassformers or modifiers. This model has been extended to incorporate other materials such as UO₂, ZrO₂ and FeO as modifiers and Fe₂O₃ as an amphoteric. The viscosity is supposed to follow a Weyman law µ = AR exp(B/T) in which a relationship links the coefficients A and B. It has been found that the relationship proposed by Urbain has a wider range of validity than originally thought. The relationship between B and the relative fractions of modifiers and amphoteric in a melt has been extended to a large range of published data. Satisfactory fits are found as long as the concentration in iron is lower than 10-15%mol. During crystallization, this approach coupled to a methodology [3,4] taking into account the volume fraction of crystals, is useful to estimate the viscosity of a partially solidified silicate.

Urbain G, Geochim. Cosmochim. Acta, 46, 1061-1072, (1982).

Urbain G, Steel Res, 58, 111-116, (1987).

Ryerson, FJ, Weeb HC & Piwinski, AJ, J. Geophys Res, 93, 3421-3436, (1988).

Seiler, JM & Ganzhorn, J, Nucl. Eng. Des, 178, 259-268, (1998).

An Experimental Approach Designed to Determine the Liquid Line of Descent of Calc-alkaline Magmas

Ralf Kaegi (kaegi@erdw.ethz.ch), Peter Ulmer (pulmer@erdw.ethz.ch) & Alan Thompson (alan@erdw.ethz.ch)

Institute of Mineralogy and Petrology, ETH-Zurich, Sonneggstrasse 5, 8092 Zurich, Switzerland

The goal of this study was to experimentally determine the liquid line of descent of hydrous, calc-alkaline magmas typical of 'island arc' environments. We focussed on one of the most simple models, namely, fractional crystallization. A natural picrobasalt (2.6 wt% H₂O) from the Adamello-Batholith (Italy) was used as starting material. In our model we assume primary magmas ascend from deeper in the mantle, loosing their positive buoyancy and differentiating by fractional crystallization at the crust mantle boundary. Accordingly, we run our experiments at a pressure of 10 kbar, corresponding to 35 km depth. The first set of experiments used a double capsule technique with graphite as inner and platinum as outer capsule material. This method was chosen for two reasons: i) no iron loss to the graphite capsules occurs and ii) platinum capsules provide fluid conservative conditions. However, graphite capsules fix oxygen fugacity (fO₂) at QFM - 1.5, significantly lower than under natural conditions (QFM +0.5). The liquidus minerals under these conditions were: 1230°C: olv, spl 1200°C: olv, cpx, spl 1170 C: cpx, (spl) 1140°C: cpx, opx, spl 1110°C: cpx, opx, plg 1080°C: plg, cpx, opx

Late plagioclase crystallization causes an iron enrichment (tholeiitic) trend. Experiments with higher water content resulted in suppression of plagioclase crystallization, but not enough to prevent iron enrichment. Another series of experiments were conducted at higher fO₂ by replacing the graphite inner capsules with iron presaturated AuPd capsules. An outer platinum capsule filled with the same starting material was used to achieve fluid conservation. Liquidus minerals at 1230°C also are olivine and spinel, but the olivine is more forsteritic and spinel is more abundant.

In situ Measurements of Viscosities and Densities of Silicate Melts with Dynamic Neutron Imaging

Andreas Kahle (ank@min.uni-kiel.de)¹, Bjoern Winkler (bjoern@min.uni-kiel.de)¹, Bernard Hennion², Philippe Boutrouille² & Guy Bayon²

¹ Institut fuer Geowissenschaften, Mineralogie/Kristallographie, Olshausenstr.40, 24098 Kiel, Germany

² CEA, LLB, Centre d`Etudes de Saclay, Gif-sur-Yvette, France

We are developing a new tool for in situ measurements of viscosities and densities of silicate melts at high temperatures and moderate pressures. This method is based on dynamic neutron imaging and the conventional falling sphere technique (Kanzaki et al.,1987).

Dynamic neutron imaging is the time and position dependent measurement of the attenuation of a neutron beam due to the absorption of the neutrons. Due to the different absorption cross sections of furnace, crucible, melt and spheres one can clearly follow the path of the falling sphere and therefore exactly locate the sphere as a function of time. The sample holder is a graphite crucible which has a height of about 120 mm and an inner diameter of about 30 mm. Spheres (Hf, Er, ...) which differ in density and radius are used. The data is analysed by using Stoke's equation: = 2 C_F g r² ((rho)_sphere - (rho)_melt) / 9 v. The main advantage of using neutrons compared with synchrotron experiments is the possibility to monitor the fall of the sphere over a longer distance (about 80 mm in our experiments versus 1-4 mm in synchrotron experiments (Dobson, 1996)). This allows in principle very accurate measurements of melt viscosities and densities.

All experiments were carried out at the neutron radiography facility of the Orphee reactor, CEN, Saclay, France. Our first in situ measurements were performed with an alkali-silicate-lead melt in the temperature range from 1000-1400K at ambient pressure. The measuring times ranged from 5-60 minutes. The results were in very good agreement with other data based on rotational viscometry. This method will now be extended to measurements at higher temperatures and of melts containing volatiles.

Kanzaki M, Kurita T, Fujii T, Kato T, Shimomura O & Akimoto S, Geophysical Monographs, 39, 195-200, (1987).

Dobson DP, Earth and Planetary Science Letters, 143, 207-215, (1996).