A simple predictive model of quartz solubility in water-salt-CO2 systems at temperatures up to 1000 °C and pressures up to 1000 MPa

Knowledge of the solubility of quartz over a broad spectrum of aqueous fluid compositions and T-P conditions is essential to our understanding of water-rock interaction in the Earth's crust. We propose an equation to compute the molality of aqueous silica, mSio(2(aq)), mol.(kg H2O)(-1), in equilibrium with quartz and water-salt-CO2 fluids, as follows: log m(SiO2) - A(T) + B(T) . log 18.0152/V-H2O(+) + 2logx(H2O) Here A (T) and B(T) are polynomials from Manning's (GCA 58 (1994), 4831) equation for quartz solubility in pure water, and x(H2O) and V-H2O(+) stand for the mole fraction and effective partial molar volume of H2O in the fluid, respectively. The value of V-H2O(+) is computed from the relation V-mix = x(H2O)V(H2O)(+), + Sigma x(s)V(s), where V-mix is the molar volume of the fluid mixture (in cm(3) mol(-1)), and x(s) and V-s denote the mole fraction and the intrinsic volume of the solute, s, respectively. Values of V-mix may be obtained from experimental data on the fluid mixture or from a reliable equation of state for the mixture. Adoption of the V-s values V-Nacl = 30.8 cm(3) mol(-1) and V-CO2 = 29.9 cm(3) mol(-1) permits satisfactory prediction of quartz solubility both in binary and ternary aqueous systems. In lieu of experimental data V-s can be estimated from pure substance properties: the intrinsic volumes of molten salts yield V-s for the electrolyte components, whereas the excluded volumes of gas species in Redlich-Kwong-Soave-type equations of state yield V-s for the volatiles. The accuracy of our density model is only slightly inferior to the empirical regressions that experimentalists have used to interpolate their measurements of quartz solubility. The strength of our model lies in its ability to predict trends in quartz solubility in fluid mixtures over an extremely wide range of T-P-x(s) conditions relevant to the Earth's crust, including conditions hitherto unexplored experimentally. This success is attributable to our model having only one adjustable parameter per solute. (C) 2008 Elsevier Ltd. All rights reserved.