Solubility of enstatite plus forsterite in H2O at deep crust/upper mantle conditions:: 4 to 15 kbar and 700 to 900°C

Ten reversed brackets of aqueous silica concentration in equilibrium with enstatite (Mg2Si2O6) and forsterite (Mg2SiO4) were determined at 700 to 900degreesC and 4.35 to 14 kbar. Silica solubilities were measured by a weight loss-weight gain method and by approaching equilibrium from higher and lower silica concentrations. Both single-crystal and fine-grained synthetic enstatite and forsterite starting material were used, as well as low-Fe natural minerals. Results for all three kinds of starting material were concordant. Silica concentrations buffered by enstatite + forsterite at 10 kbar increase from 0.16 moles/kg H2O (in) at 700degreesC to 0.50 m at 900degreesC, but show only small variation with pressure in the range 7 to 14 kbar. The ratio X-S(Q)/X-S(FF), where X-S(Q) is the mote fraction of aqueous SiO2 (S) in equilibrium with quartz (Q) from Manning (1994), and X-S(FE) is the corresponding mole fraction in equilibrium with forsterite (F) and enstatite (E) at the same P and T, from 1 to 15 kbar and 700 to 900degreesC is described by X-S(Q) / X-S(FE) = 2.61 - 0.000930T + 0.0236 + 290P / T where P is in kbar and T in Kelvins. All of our measured silica solubilities are significantly lower than predicted with the conventional assumption of unit activity of silica for concentrations from zero to quartz saturation. Aqueous silica therefore exhibits negative, composition-dependent departures from ideality, in agreement with the deduction of Zhang and Frantz (2000) based on measurements at higher temperatures. Zhang and Frantz (2000) suggested that the formation of solute silica dimers accounts for the lowered activity. The equilibrium constant, K, for the proposed dimer formation reaction, 2H(4)SiO(4) = H6Si2O7 + H2O is K = X(dimers)a(H2O)/X-monomers(2), where the activities (a) of solute species are equated with their mole fractions, and a(H2O) approximate to 1 at the low SiO2 concentrations considered. Our solubility data at 700degreesC and 10 kbar give logK = 2.2, which predicts that solute silica in equilibrium with quartz at these conditions is about 40% polymerized, assuming dimers are the only polymer species. This estimate is in good agreement with Raman spectroscopic results of Zotov and Keppler (2002). However, if higher polymers, including rings and chains, are also present, the estimates of K based on either the solubility measurements or the spectroscopic observations would be substantially smaller. Our solubility measurements show that the activity of aqueous silica is substantially less than the molar concentration at deep crust/upper mantle PT conditions. Copyright (C) 2002 Elsevier Science Ltd.