Carbon dioxide is one of the most abundant volatile components in magmas after H2O along with S and Cl, which are of great importance to the extraction of trace metals into magmatic-hydrothermal fluids and ore minerals. Yet the effect of CO2 on the partition coefficients of chlorine between water-rich fluid and melt (i.e. the ratio of mass Cl concentrations in the corresponding phase; [D-Cl(fluid/melt)]) is still poorly constrained. We conducted a set of experiments to constrain the effect of CO2 on D-Cl(fluid/melt) by equilibrating felsic silicate melts with aqueous NaCl-bearing fluids while varying the concentration of CO2 at pressures between 120 and 300 MPa and temperatures of 850 and 1000 degrees C. The starting melt was synthetized as a glass in the ternary albite-quartz-Al2O3 system to ensure the only significant metal chloride species in the equilibrium fluid phase was NaCl. The results demonstrate that D(Cl)(fluid/melt)values increase with the concentration of Cl in the fluid phase and the silicate melt. The addition of CO2 into aqueous metal chloride-bearing fluids induces a pronounced drop in D-Cl(fluid/melt) the extent of which is only weakly affected by pressure, temperature and fluid salinity, at least at relatively low Cl concentrations (<12 wt% NaCl equivalent). The values of D-Cl(fluid/melt) drop by approximately a factor of 3 in response to the addition of 20-25 mol% CO2 to the fluid phase due to the decreased ability of the fluid to hydrate NaCl ion pairs. A new empirical equation describing wt%based D-Cl(fluid/melt) is derived: In[D-Cl(flui)d/melt] = 1.419(+/- 0.048) *In[P]+0.912(+/- 0.031) *ln[C-Cl(fluid) +1.434(+/- 0.260)] + 4547(T +/- 443)/T -4.026(+/- 0.155)* X-CO2- 9.790(+/- 0.440) which expresses D-Cl(flui)d/melt as a function of pressure (in MPa), temperature (in Kelvins), and the equilibrium concentration of Cl (in wt%) and CO2 (as molar fraction) in the fluid. The average absolute percentage error of the model predictions relative to the experimental data is 7.3%. The equation contains only monotonous functions to allow moderate extrapolation outside the range of the calibration dataset as long as the fluid remains in the single-phase field (e.g. pressure above 100 MPa and up to 500 MPa or even higher). Our model equation represents a reasonable approximation for granitoid magma evolution in the upper and middle crust. The presence of CO2 suppresses fluid salinity at a given chloride content of a granitoid melt, which will also hinder the extraction of chloride-complexed ore metals (e.g. Cu, Pb, Zn, Mo, and Ag) into magmatic fluids, reducing the likelihood of base-metal ore formation (e.g. porphyry copper deposits) from such fluids in the uppermost crust. On the other hand, the highly volatile components, CO2, H2S and SO2, are enriched in magmatic fluids exsolving early during the ascent of hydrous magmas. Chloride suppression by CO2 and enrichment of fluids in sulfur species near the SO2/H2S predominance boundary favors extraction of sulfide-complexed metals, notably Au as Au(HS)(0), Au(HS)2(-), NaAu(HS)(2)(0) or Au(HS)S-3(-) into low-salinity magmatic fluids. Such fluids resemble those forming mid-crustal lode gold deposits that are typically poor in base metals. Our experimental results may therefore be taken as indirect support for a magmatic component in fluids forming orogenic lode gold deposits. (C) 2019 Elsevier Ltd. All rights reserved.
