C–O–H–S fluids and granitic magma: how S partitions and modifies CO2 concentrations of fluid-saturated felsic melt at 200 MPa

Hydrothermal volatile-solubility and partitioning experiments were conducted with fluid-saturated haplogranitic melt, H2O, CO2, and S in an internally heated pressure vessel at 900°C and 200 MPa; three additional experiments were conducted with iron-bearing melt. The run-product glasses were analyzed by electron microprobe, FTIR, and SIMS; and they contain ≤0.12 wt% S, ≤0.097 wt% CO2, and ≤6.4 wt% H2O. Apparent values of log fO2 for the experiments at run conditions were computed from the [(S6+)/(S6++S2−)] ratio of the glasses, and they range from NNO −0.4 to NNO + 1.4. The C–O–H–S fluid compositions at run conditions were computed by mass balance, and they contained 22–99 mol% H2O, 0–78 mol% CO2, 0–12 mol% S, and <3 wt% alkalis. Eight S-free experiments were conducted to determine the H2O and CO2 concentrations of melt and fluid compositions and to compare them with prior experimental results for C–O–H fluid-saturated rhyolite melt, and the agreement is excellent. Sulfur partitions very strongly in favor of fluid in all experiments, and the presence of S modifies the fluid compositions, and hence, the CO2 solubilities in coexisting felsic melt. The square of the mole fraction of H2O in melt increases in a linear fashion, from 0.05 to 0.25, with the H2O concentration of the fluid. The mole fraction of CO2 in melt increases linearly, from 0.0003 to 0.0045, with the CO2 concentration of C–O–H–S fluids. Interestingly, the CO2 concentration in melts, involving relatively reduced runs (log fO2 ≤ NNO + 0.3) that contain 2.5–7 mol% S in the fluid, decreases significantly with increasing S in the system. This response to the changing fluid composition causes the H2O and CO2 solubility curve for C–O–H–S fluid-saturated haplogranitic melts at 200 MPa to shift to values near that modeled for C–O–H fluid-saturated, S-free rhyolite melt at 150 MPa. The concentration of S in haplogranitic melt increases in a linear fashion with increasing S in C–O–H–S fluids, but these data show significant dispersion that likely reflects the strong influence of fO2 on S speciation in melt and fluid. Importantly, the partitioning of S between fluid and melt does not vary with the (H2O/H2O + CO2) ratio of the fluid. The fluid-melt partition coefficients for H2O, CO2, and S and the atomic (C/S) ratios of the run-product fluids are virtually identical to thermodynamic constraints on volatile partitioning and the H, S, and C contents of pre-eruptive magmatic fluids and volcanic gases for subduction-related magmatic systems thus confirming our experiments are relevant to natural eruptive systems.