In situ determination of the eutectic melting temperature of Fe-FeS-Fe3C between 4.5 and 24.5 GPa and implications for Mercury;s core

Audrey M. Martin,J. Orman,S. Hauck,,Bin Chen,N. Sun,R. D. Moore,Jian Han
2014-08-19
Abstract:Introduction: Recent data from the MESSENGER mission have shown that Mercury’s surface has a low iron content compared to other terrestrial planets like the Earth or Mars (<4 wt% Fe [1], with a maximum amount as FeO <2-3 wt% [2]). This low Fe content and the large surface S contents (up to 4 wt% [3]) have led to the inference that Mercury’s differentiation occurred at particularly reducing conditions, so that most iron was in metal form and was incorporated into the core. The S content of the core could be large [4-6]. Moreover, up to ~4 wt% C could be partitioned into the core from chondritic precursor materials under suitably reducing conditions during the planet formation. Previous experiments have demonstrated that S strongly influences the melting conditions of iron, potentially creating an iron snow process in Mercury’s core [4,7], wherein iron precipitates at shallow levels and then sinks through the core. This is due to the decrease of eutectic melting temperatures (TE) and S contents with increasing pressure. In order to constrain the role of C, the melting of Fe-S-C systems at 2-20 GPa has been investigated using piston-cylinder and multi-anvil experiments, by analyzing the recovered quenched samples [8,9]. [8] found that eutectic melting occurs below 1050°C at 2 GPa. In addition, they showed that immiscibility between a S-rich melt and a C-rich melt occurs at low pressure, up to 4 GPa in the Fe-5wt% S-5wt% C system and up to 5 GPa in the Fe-5wt% S-15wt% C system. [9] found that TE ~1100°C at 10 GPa, i.e., a much higher temperature than in the simpler Fe-S system (~855°C [10]). [10] also found that increasing pressure decreases TE in the Fe-S system up to 14 GPa. Above this pressure, TE increases again, reaching 1150°C at 25 GPa [11]. According to [9], the Fe-S-C system melts around 1225°C at 20 GPa. By extrapolating their curve, TE>1250°C at 25 GPa. One of the main difficulties for the determination of solidus temperatures in metallic systems is the interpretation of the textures of quenched samples at low melt fraction. In situ spectroscopic methods allow to determine the phase changes – in particular, the appearance of melt – during the experiments. In order to constrain TE in the Fe-S-C system, we therefore made in situ multi-anvil experiments on a synchrotron. Experimental method: The experiments were performed using the 1000-ton mutli-anvil apparatus at Advanced Photon Source, Argonne (13-IDD). A 10/5 (octahedron edge length/WC cubes TEL) assembly was used for the experiments between 4.5 and 15.5 GPa, and an 8/3 assembly between 18 and 24.5 GPa. The sample was placed in a MgO capsule plugged at each end by MgO powder. Temperature was measured using a W5-Re26 thermocouple crossed at one end of the sample. Pressure calibrations were made during each experiment using a MgO pressure marker. A rolled rhenium foil was used as furnace. It was separated from the MgO octahedron using a LaCrO3 insulating sleeve. In addition, graphite windows were put into the Re furnace and LaCrO3 sleeve on the X-ray beam path. During the experiments, the pressure was first increased to the desired value. Then, the temperature was increased by increments in order to regularly take X-ray images and diffraction spectra of the sample. In most cases, temperature was quenched soon after the melting was observed, so that eutectic melting textures would be preserved. Starting material: In order to distinguish metallic melts and solids, we physically separated the Fe phase from the FeS+Fe3C mixture by inserting a Fe rod inside the capsule, and put the mixture below it. At the interface between the rod and mixture, the three phases Fe, FeS and F3C were in contact. In order to avoid any kinetic issues with C diffusion, we pre-saturated the Fe rod with C using high temperature furnace experiments.
Materials Science
What problem does this paper attempt to address?