Molybdenum, tungsten and manganese partitioning in the system pyrrhotite–Fe–S–O melt–rhyolite melt: Impact of sulfide segregation on arc magma evolution

Abstract

Experiments were performed to determine the partitioning of molybdenum, tungsten and manganese among a rhyolitic melt (melt), pyrrhotite (po), and an immiscible Fe–S–O melt (Fe–S–O) . Sulfide phases such as these may be isolated from a silicate melt along with other crystallizing phases during the evolution of arc magma, and partition coefficients are required to model the effect of this process on molybdenum and tungsten budgets. We developed an experimental design to take advantage of properties of the phases under study. Careful control of temperature allowed pyrrhotite and magnetite to be stable along with an Fe–S–O melt, and this phase assemblage allowed the composition of run-product pyrrhotite to be used to calculate both fS 2 and fO 2 for the experiments. At run temperature, (1042 ± 2 °C), a rhyolitic melt can be formed at low pressure, under nominally dry conditions, which removed the need for confining pressure as well as externally imposed fugacities. The silica-saturated melt allowed the charges to be contained in sealed evacuated silica tubes without danger of reaction, and with closed system behavior for molybdenum and tungsten. Experiments were run for durations up to 2000 min. Molybdenite (mb) and wolframite (wo) were added to the experiments as sources for molybdenum and tungsten, respectively. Manganese was added to the system as both a component of the starting rhyolitic pumice, and of Mn-bearing wolframite. Oxygen fugacity in these experiments was fixed at the Ni–NiO oxygen fugacity buffer. Sulfur fugacity was 10 −1 bar. Run products were analyzed by EPMA and LA-ICP-MS. Analysis of the run products yielded ( D i a / b ± 1 σ X ¯ standard deviation of the mean): D Mo Fe – S – O / Si-melt = 90 ± 10 , D W Fe – S – O / Si-melt = 9 ± 3 , D Mo po/Si-melt = 35 ± 3 , and D W po/Si-melt = 0.0012 ± 0.0006 . The partition coefficients for manganese in this system are D Mn Fe – S – O / Si-melt 2.9 ± 0.3 and D Mn po/Si-melt = 1.1 ± 0.1 . Simple Rayleigh fractionation modeling suggests that oxidized felsic melts produced through fractional crystallization may have lost as much as 14% of their initial molybdenum, but only 2% of their initial tungsten, through the removal of an Fe–S–O melt along with crystalline phases. Modeling consistent with conditions of oxygen and sulfur fugacity influenced by assimilation of sulfide (with low concentrations of molybdenum and tungsten) from, for example, sedimentary rock, results in evolved magmas significantly depleted in molybdenum, but only moderately depleted in tungsten. The molybdenum:tungsten ratio can vary by two orders of magnitude. These systematics may help to explain some of the variability in metal ratios of intrusion-related hydrothermal ore deposits.