TOWARD AN EXPERIMENTALLY CALIBRATED GARNET-TOURMALINE GEOTHERMOMETER

We are experimentally calibrating the exchange of Mg, Fe, and Mn between garnet and tourmaline toward a geothermometer that is applicable to peraluminous granites and medium-grade metapelites. Initial experiments involving reactions among sand-sized grains of andalusite, rhodonite, Mn-fayalite and/or forsterite in hydrous boron-bearing, peraluminous granitic melt were conducted at 200 MPa (H₂O) in the temperature range 650°C-750°C. Products from these initial experiments include garnet, tourmaline, (Mn,Fe)-Al-spinel, magnetite, alkali feldspar, plagioclase, quartz, corundum, and pyroxene and/or amphibole. The f(O₂) lies below MH and above FMQ as indicated by the complete oxidation of fayalite to magnetite plus quartz. Graphite was added to one series of experiments to lower f(O₂) (to near the CCO buffer); in these, magnetite rimmed fayalite, (Mn,Fe)-Al-spinel formed, and graphite but no excess water was present on quench, indicating that f(O₂) was reduced in dominantly carbonic vapor. EMPA of garnet-tourmaline pairs from fayalite- and graphite-bearing experiments yield values of K_D^Mn/Fe_Grt/Tur and K_D^Fe/Mg_Grt/Tur that have large standard deviations and show only minor apparent fractionation as a function of temperature. The variability in K_D values stems from heterogeneity among individual garnet and tourmaline crystals due to sluggish diffusion of Mg, Fe, and Mn through melt. Experiments are currently focused on reducing heterogeneity of products by creating a nearly crystal-free, homogenous glass at 850°C and 200 MPa, which is quenched and then ran forward to the temperature of interest. Graphite and fayalite are not used to buffer f(O₂) in current experimentation, but the presence of (Mn,Fe)-Al-spinel, cordierite, (Mg,Fe,Mn)-pyroxene, and lack of magnetite suggests that iron is likely present as Fe²⁺. Garnet and tourmaline grown in this fashion are more homogenous throughout the charge though they exhibit compositional zoning.

Uncertainty in the garnet-tourmaline partitioning arises from the saturation of Mn in tourmaline at low Mn concentrations, by the exceedingly low Mg content in garnet, and by the uncertainties in existing K­_D values. In addition to a thermometer based elemental partitioning, we are also pursuing calibration of Δ¹⁸O between garnet and tourmaline pairs.