The effect of sulfide dissolved in silicate melts on enhancing the solubility of the Highly Siderophile Elements

Abstract:

There are large inconsistencies among experimental studies of Highly Siderophile Element (HSE) partitioning relations between silicates and metal or sulfide phases, which has usually been attributed to “micronuggets”, a general term for sub-optical (approximately < 0.2 µm) particles or blobs of metal or sulfide in the silicate phase. But there have then been differences of opinion as to whether these micronuggets (and, sometimes, associated optical-sized metal or sulfide particles) should be viewed as contamination, or are produced by precipitation on quenching the experiment, which need to be included to recover true partitioning relations. Although quench precipitation is perhaps inevitable in experiments at extremely high temperatures and/or pressures, or with high loads of dissolved sulfide, it may be superimposed on the usual micronugget contamination. The problem is particularly acute in HSE sulfide-melt/silicate-melt partition coefficients, where results range over several orders of magnitude. Moreover, nearly all the reported results of directly determined sulfide-melt/silicate-melt partition coefficients are considerably lower than values calculated by combining metal/silicate-melt with metal/sulfide-melt partition coefficients. This discrepancy has been attributed to large effects of S dissolved as sulfide in the silicate melts on HSE solubilities. As such large effects are not expected from the thermodynamic modeling of sulfide solubilities in silicate melts, it has been proposed that HSEs dissolve in sulfide-containing silicate melts by forming HSE-S complexes. This idea has been tested by experiments that compare the solubilities of Ir, Re and Ru in a high-TiO₂ silicate melt both with and without dissolved sulfide at 1400 to 1600ºC at atmospheric pressure. The high TiO₂ suppresses micronuggets. Experiments were analysed by LA-ICP-MS, with detection limits approaching 2 ppb. For Ir, the results show that at fO₂ low enough to enable measurable sulfide in the melt, the presence of the sulfide just raises the level of dissolved Ir above detection limits. These results suggest a fairly large influence of the sulfide, but are not quantitative. By contrast, the experiments on Ru and Re clearly show only a modest effect of dissolved sulfide.