Extended and improved Ti-in-quartz solubility model

Experiments addressing Ti solubility in quartz were conducted at conditions not previously explored to extend and improve the Ti-in-quartz solubility model for thermobarometric applications. Experiments contained a variety of starting materials including silica glass, anatase, rutile, Ti-enriched silica gel, ZrO₂ and water. We performed experiments in cold-seal devices at pressures up to 3.5 kbar and temperatures 700 to 900°C, and in piston-cylinder devices from 5 to 32 kbar and 575 to 1050°C. Quartz crystals contained abundant rutile, zircon and aqueous fluid inclusions, demonstrating that quartz crystallized from SiO₂-, TiO₂- and ZrSiO₄-saturated hydrothermal fluids. Time-series experiments indicated that anatase in the starting material transformed to rutile within the first three minutes at run conditions. Titanium K-edge XANES measurements showed that Ti⁴⁺ substitutes for Si⁴⁺ on fourfold tetrahedral sites in quartz grown at 5 to 32 kbar, and the same is expected for quartz crystallized at lower pressures. EPMA measurements and CL images show that quartz crystals are relatively homogenous. Importantly, Ti concentrations of quartz crystals grown at the same conditions using several sources of TiO₂ and multiple growth media are identical. Titanium concentrations of quartz grown in piston-cylinder devices systematically vary with P and T, but results from cold-seal experiments at P<2 kbar are highly variable. High Ti variability in quartz grown in cold-seal devices may be attributed to non-equilibrium conditions resulting from low TiO₂ solubility in aqueous fluids at low P-T conditions. Inclusion of the new piston-cylinder results in the Ti-in-quartz solubility model has a small but significant effect on predicted Ti concentrations of quartz crystallized at both low and high pressures. Titanium and Al concentrations of quartz crystals and Zr concentrations of rutile crystals grown in our piston-cylinder experiments yield isopleths that cross at the experimental run conditions, cross validating the various solubility models and verifying our experimental approach.