Pressure, sulfur, and metal-silicate partitioning: The effect of sulfur species on the parameterization of experimental results

Abstract

11Performing well-controlled metal-silicate partitioning experiments at conditions directly simulating 12 those of a deep magma ocean is difficult. It is therefore common to perform experiments at lower 13 pressures and temperatures, which are used to determine the effects of salient variables. Often, these 14 effects are determined by multiple linear regression of a dataset covering a large range of P-T-15 composition space. In particular, these datasets often contain the results of experiments performed both 16 with and without sulfur in the system. Data are often regressed, however, using a relationship based 17 only upon the formation of oxide species in the silicate melt. Several studies have suggested that, when 18 sulfur is present in the system, siderophile trace metals may also dissolve into silicate melt as S-bearing 19 species. We have derived a relationship for regressing experimental metal-silicate partitioning data that 20 considers the formation of both oxide and sulfide species in the silicate melt. Using model datasets, we 21have assessed the ability of this relationship, and the more typical single-species relationship, to 22 accurately parameterize data in which the formation of S-bearing species is important. We have also 23 applied this new relationship to experimental results on the metal-silicate partitioning of gold and find 24 This is a preprint, the final version is subject to change, of the American Mineralogist (MSA) from the resulting solubility data. Most metal-silicate partitioning data have been produced at pressures 36 below ~27 GPa due to the difficulty of achieving higher pressures using large-volume press techniques. 37Although these conditions overlap with lower estimates for the maximum depth of core-segregation 38 from a magma ocean (e.g., Li & Agee, 1996;, it is significantly below the 39 maximum pressures invoked in some studies (e.g., 40-80 GPa;Corgne et al., 2009; Rubie et al., 2011; 40 Siebert et al., 2013). Considerable extrapolation of the experimentally determined metal-silicate 41 partition coefficients to relevant conditions is therefore required. This is typically achieved by 42 parameterizing the available data using multiple linear regression, as a function of pressure (P), 43 temperature (T), composition and in some cases oxygen fugacity (fO 2 ). These parameterizations are 44 based on reactions where the element of interest dissolves in silicate melt as an oxide species. For 45 systems containing sulfur, however, it has been suggested for several elements that the formation of 46 sulfur-bearing complexes is important (e.g. Botcharnikov et al. 2010;Laurenz et al. 2013; Mungall & 47 Brenan, 2014;). Here, we consider whether these elements require a different 48 This is a preprint, the final version is subject to change, of the American Mineralogist (MSA) Cite as Authors (Year) Title. American Mineralogist, in press. DOI: https://doi.org/10.2138/am-2018-6282Always consult and cite the final, published document. See http:/www.minsocam.org or Geosci...

Supplemental Material: Pyrrhotite–silicate melt partitioning of rhenium and the deep rhenium cycle in subduction zones

“…then 𝑙𝑙𝑃𝑃𝑙𝑙𝐷𝐷 𝑅𝑅𝑅𝑅 𝑝𝑝𝑝𝑝/𝑠𝑠𝑠𝑠 obtained at various fO2s depend not only on the FeO content in silicate melt but also on fO2.Li et al (2021) found that the partition coefficients of Re, Mo, As, Sb, and Bi between different sulfide phases (sulfide liquid and monosulfide solid solution) and silicate melt, which elements have variable valence states, at various P-T-fO2 can be empirically described as: 𝑙𝑙𝑃𝑃𝑙𝑙𝐷𝐷 𝑠𝑠𝑠𝑠𝑠𝑠𝑓𝑓𝑠𝑠𝑠𝑠𝑅𝑅/𝑠𝑠𝑠𝑠 = 𝑎𝑎 𝑇𝑇(𝐾𝐾) + 𝑏𝑏 • 𝑃𝑃(𝐺𝐺𝑃𝑃𝑎𝑎)/𝑇𝑇(𝐾𝐾) − 𝑛𝑛/2𝑙𝑙𝑃𝑃𝑙𝑙(𝑋𝑋𝑎𝑎𝑅𝑅𝑂𝑂) + 𝑐𝑐 • ∆𝑎𝑎𝑀𝑀𝐹𝐹 + 𝐶𝐶 (11) Li and Audetat (2015), Feng and Li (2019), and Li et al (2021) pointed that if an element M forms not only M-O 2but also M-S 2species in silicate melt, to what extent the 𝑙𝑙𝑃𝑃𝑙𝑙𝐷𝐷 𝑀𝑀 𝑠𝑠𝑠𝑠𝑠𝑠𝑓𝑓𝑠𝑠𝑠𝑠𝑅𝑅/𝑠𝑠𝑠𝑠 is correlated with the FeO content in silicate melt depend significantly on the fraction of the element M that forms M-O 2species. Significant fractions of Pt, Pd, Au, Cu, Ag, and even Mn and W in silicate melt may form sulfidespecies, as discussed in recent studies (e.g.,Zajacz et al, 2013;Bennett and Fei, 2018;Li et al, 2019; Evan et al, 2009;Laurenz et al, 2013).𝑎𝑎 𝑇𝑇(𝐾𝐾)…”