The solubility of rhenium in silicate melts: Implications for the geochemical properties of rhenium at high temperatures

https://doi.org/10.1016/S0016-7037(01)00582-8Get rights and content

Abstract

The solubility of rhenium (Re) in a haplobasaltic melt (anorthite-diopside eutectic composition) has been experimentally determined using the mechanically assisted equilibration technique at 1400°C as a function of oxygen fugacity (10−12 < fO2 ≤ 10−7 bar), imposed by CO-CO2 gas mixtures. Samples were analysed by laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS). This is a true microanalytical technique, which allows small-scale sample heterogeneity to be detected, while providing a limit of detection of 2 ppb Re. Time-resolved LA-ICP-MS spectra revealed the presence of suboptically sized micronuggets of Re in all samples, which, because they are present at the 0.5 to 10 ppm level, dominate the true solubilities of Re (<1 ppm at the conditions of the experiment) in bulk analyses of the samples. Nevertheless, the micronuggets could be filtered out from the time-resolved spectra to reveal accurate values of the true Re solubility. A number of time series of samples were taken at constant fO2 to demonstrate that the solubilities converge to a constant value. In addition, solubilities were measured after increasing and decreasing the imposed fO2. The results show that Re dissolves in the silicate melt as ReO2 (Re4+) and ReO3 (Re6+) species, with the latter predominating at typical terrestrial upper-mantle oxygen fugacities. The total solubility of Re is described by the following expression (fO2 in bars): [Re/ppb] = 9.7(±1.9) × 109 (fO2) + 4.2 (±0.3) × 1014 (fO2)1.5Assuming an activity coefficient for Re in Fe-rich metal of 1, this gives a value of DRemet/sil of 5 × 1010 at log fO2 = IW-2, appropriate for metal-silicate partitioning in an homogenously accreting Earth. Thus, Re is indeed very highly siderophile, and the mantle’s abundance cannot be explained by homogenous accretion.

Introduction

Rhenium (Re), together with the six platinum group elements (Ru, Rh, Pd, Ir, Os, and Pt) and Au, comprise a group of elements known as the highly siderophile elements (HSEs), the defining geochemical property of which is that they have metal-silicate distribution coefficients DMmet/sil in excess of 104. Their high metal-silicate distribution coefficients have resulted in their partitioning into the metal of the Earth’s core. This, together with the fact that the HSEs are heavy elements with intrinsically low solar abundances, means that their abundances in the Earth’s mantle are extremely low. In fact Re, with a bulk silicate Earth abundance of 0.2 ppb, is the rarest of all the naturally occurring elements, apart from the noble gases (e.g., O’Neill and Palme, 1999). Nevertheless, advances in geochemical analytical capabilities now enable the HSEs to be determined in a wide variety of igneous rocks. Interpretation of these analytical data requires some knowledge of the high-temperature geochemical properties of the HSEs, in particular quantification of their metal-silicate and sulfide-silicate distribution coefficients.

In addition, identification of the valence states of Re in the high-temperature geological environment is a useful first step towards understanding the geochemical behaviour of this element. The common valence states of Re are 4+, 6+, and 7+, but previous experimental work (e.g., O’Neill et al 1995, Righter and Drake 1997) has suggested that Re dissolves in silicate melts as species with unusually low valence states (1+ or 2+). These low valence states are so rare that they are effectively unknown in solid state chemistry, and consequently, there is no indication of even the most fundamental chemical properties of Re in such states. However, these earlier experimental studies used bulk analytical methods. Some recent parallel studies on other HSEs raise the possibility that these earlier studies may have been compromised by the “micronugget effect” O’Neill et al 1995, Ertel et al 1999, whereby the small amounts of Re chemically dissolved in the silicate melts may have been masked by larger amounts present as micronuggets. Micronuggets are submicroscopic particles, which seem to form something akin to a colloidal suspension. In this study we use a microanalytical technique, laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS), to see through the micronugget contamination and determine the true solubilities of Re in a haplobasaltic silicate melt (Anorthite-Diopside 1-atm eutectic composition). We demonstrate that the previous work was indeed in error from micronugget contamination. The valence states of dissolved Re in silicate melts are in fact the familiar 4+ and 6+. Micronuggets also probably affected previous experimental studies of Re metal-silicate partitioning, which are consequently biased to low values; the present results demonstrate unambiguously that Re is indeed an HSE with an enormous metal-silicate distribution coefficient at likely core-forming conditions.

Section snippets

Sample equilibration

Experiments were performed in a vertical tube furnace equipped for gas mixing, using the mechanically assisted equilibration technique of Dingwell et al. (1994). Oxygen fugacities were imposed using CO-CO2 gas mixtures.

Starting silicate melt compositions were prepared by a gelling process (Hamilton and Henderson 1968) from Al metal powder, CaCO3, MgO, and tetraethylorthosilicate (TEOS), to assure maximum homogeneity. The gel was fused in a Naber box furnace for 25 min in air in an Al2O3

Re solubilities

The experiment reported here lasted 7227 h (∼10 months), during which 93 samples were taken at oxygen fugacities from 10−12 to 10−7.3 bars at 1400°C. All samples were analysed initially by INAA. Re concentrations were generally found to be in the range 0.5 to 10 ppm, with only a vague correlation with fO2. Subsequently, the same samples were analysed by LA-ICP-MS. The latter method is a true microanalytical technique, since each ablation pulse removes material with diameter 70 μm to a depth of

Comparison with previous work

Previously we have reported preliminary results from an earlier experiment at similar conditions of melt composition, temperature, and fO2, which used the same experimental approach as here, except that the samples were analysed only by INAA (O’Neill et al., 1995). Because only a bulk analytical method was used, these earlier results are compromised by the micronugget effect and, therefore, do not represent true solubilities. The total Re contents are similar to those obtained by INAA in this

Acknowledgements

We thank John Morgan and Ed Mathez for helpful and prompt reviews.

Associate editor: H. E. Newsom

References (24)

  • Barin I. (1989) Thermodynamic Data of Pure Substances. VCH Verlagsgesellschaft, Weinheim,...
  • A. Borisov et al.

    An evaluation of Re, as an alternative to Pt, for the 1 bar loop techniqueAn experimental study at 1400°C

    Am. Mineral.

    (1999)
  • Cited by (79)

    View all citing articles on Scopus

    Present address: Lunar and Planetary Laboratory, University of Arizona, Tucson, AZ 85721, USA.

    Present address: Institut für Mineralogie, Petrologie und Geochemie, Universität München, D 80333 München, Germany.

    View full text