The partitioning behavior of As and Au in S-free and S-bearing magmatic assemblages

The partitioning of As and Au between rhyolite melt and low-salinity vapor (2 wt% NaCl eq.) in a melt–vapor–Au metal ± magnetite ± pyrrhotite assemblage has been quantified at 800 °C, 120 MPa and fO2=NNOfO2=NNO. The S-bearing runs have calculated values for the fugacities of H2S, SO2 and S2 of logfH2S=1.1logfH2S=1.1, logfSO2=-1.5logfSO2=-1.5, and logfS2=-3.0logfS2=-3.0. The ratio of H2S to SO2 is on the order of 400. The experiments constrain the effect of S on the partitioning behavior of As and Au at magmatic conditions. Calculated average Nernst-type partition coefficients (±1σ  ) for As between vapor and melt, DAsv/m, are 1.0 ± 0.1 and 2.5 ± 0.3 in the S-free and S-bearing assemblages, respectively. These results suggest that sulfur has a small, but statistically meaningful, effect on the mass transfer of As between silicate melt and low-salinity vapor at the experimental conditions. Efficiencies of removal, calculated following Candela and Holland (1986), suggest that the S-free and S-bearing low-salinity vapor can scavenge approximately 41% and 63% As from water-saturated rhyolite melt, respectively, during devolatilization assuming that As is partitioned into magnetite and pyrrhotite during second boiling. The S-free data are consistent with the presence of arsenous acid, As(OH)3 in the vapor phase. However, the S-bearing data suggest the presence of both arsenous acid and a As–S complex in S-bearing magmatic vapor. Apparent equilibrium constants, logKAs′(±1σ), describing the partitioning of As between melt and vapor are −1.3 (0.1) and −1.1 (0.1) for the S-free and S-bearing runs, respectively. The increase in the value of KAs′ with the addition of S suggests a role for S in complexing and scavenging As from the melt during degassing.The calculated vapor/melt partition coefficients (±1σ  ) for Au between vapor and melt, DAuv/m, in S-free and S-bearing assemblages are 15 ± 2.5 and 12 ± 0.3, respectively. Efficiencies of removal (Candela and Holland, 1986) for the S-free melt, calculated assuming that magnetite is the dominant Au-sequestering solid phase during crystallization (Simon et al., 2003), suggest that magmatic vapor may scavenge on the order of 72% Au from a water-saturated melt. Efficiencies of removal calculated for the S-bearing assemblage, assuming pyrrhotite and magnetite are the dominant Au-sequestering solid phases, indicate that vapor may scavenge on the order of 60% Au from the melt. These model calculations suggest that the loss of pyrrhotite and magnetite from a melt, owing to punctuated differentiation during ascent and emplacement, does not prohibit the ability of a rhyolite melt to generate a large-tonnage Au deposit. Apparent equilibrium constants describing the partitioning of Au between melt and vapor were calculated using the mean DAuv/m values for the S-free and S-bearing assemblages; only S-bearing data from runs longer than 400 h were used as shorter runs may not have reached equilibrium with respect only to vapor/melt partitioning of Au. The values for logKAu′(±1σ) are −4.4 (0.1) and −4.2 (0.2) for the S-free and S-bearing runs, respectively. These data suggest that the presence of S does not affect the mass transfer of Au from degassing silicate melt to an exsolved, low-salinity vapor in a low-fS2fS2 assemblage (i.e., pyrrhotite–magnetite at NNO) at the experimental conditions reported here. Efficiencies of removal are calculated and used to model the mass transfer of Au from a crystallizing silicate melt to an exsolved, low-salinity vapor phase. The calculations suggest that the model, absolute tonnage of Au scavenged and transported by S-free and S-bearing vapors, from a crystallizing melt, would be comparable and that the time-integrated flux of low-salinity vapor could be responsible for a significant quantity of the Au in magmatic-hydrothermal ore deposits.