The chemical behaviour of chlorine in silicate melts

We have performed experiments at 0.5–2 GPa and 1200–1500 °C to investigate the dissolution behaviour of chlorine in silicate melts. The experiments were performed with chlorine fugacities controlled by mixtures of Ag, AgCl and AgI and oxygen fugacity buffered at C-CO-CO2 (CCO) and Re-ReO2. The results demonstrate that the initial chlorine dissolution mechanism involves the replacement of O2− in the silicate melt by two dissociated Cl− ions according to the reaction: Cl2 + [O2−]melt = 2[Cl−]melt + 0.5O2. The same dissolution mechanism applies to hydrous, fluid-saturated basalt at 100–200 MPa/1050 °C. Experiments using both an Fe-free haplobasaltic composition (An50Di28Fo22) and an Icelandic basalt followed the predicted dependence of Cl concentration on f(Cl2)0.5 and f(O2)0.25. This Henrian behaviour extends from 0 to at least 2.6 wt% Cl dissolved in the haplobasaltic composition, 1.6 wt% Cl in anhydrous basalt and ∼1.5 wt% Cl in fluid-saturated basalt. Deviations from Henry’s Law behaviour at higher concentrations are consistent with progressive association of Cl− ions. In the Henry’s Law region Cl concentration in the An50Di28Fo22 composition is given by (wt%): logClmelt=1.20632-94040PT-0.25logfO2+0.5logfCl2 P is in GPa, T in kelvin, values in brackets are 1 standard error, and f(Cl2) and f(O2) refer to standard states of pure gas at 0.1 MPa and the temperature of interest. For the natural anhydrous basalt we obtain: logClmelt=0.98464-93070PT-0.25logfO2+0.5logfCl2 By considering the P-T dependences of the Cl contents of melts we find that the concentrations observed in nature are extremely stable in basalt to very low pressures. Basalts containing the typical concentration range of 0.05–0.5 wt% Cl should, for example, only begin to degas their chlorine significantly, as HCl, at pressures in the range 0–5 MPa. Data on hydrous, fluid-saturated basalt at 100–200 MPa are, when corrected for dissolution of Ca, Na and K in the fluid, broadly consistent with our results for anhydrous basalt. Finally, we use recently evaluated thermodynamic data for sodalite (Na4Al3Si3O12Cl) to calculate the conditions under which this phase would stabilise in trachytes and phonolites. We find that the appearance of sodalite as a liquidus phase reflects a combination of low liquidus temperature and high Na2O activity rather than unusually high chlorine fugacity.