Thermobarometry Gone Astray

Quantitative thermobarometers require a univariant reaction, consideration of dilutions in all phases, and application to fully buffered systems.  Many empirical thermobarometers, e.g., Al-in-chlorite, Al-in-hornblende, b₀-of-phengite, as well as clay and organic transformations, were not evaluated with buffering phases and solid solutions, and they are not adequately calibrated.  Reactions proposed for Al-in-chlorite involve SiO₂ and H₂O, such that the presence of quartz and the dilution of aqueous fluid by salts are important.  The Al-in-chlorite with epidote-quartz-calcite-fluid should decrease with T, whereas with muscovite/illite-quartz-fluid or epidote-actinolite-quartz-fluid the Al should increase with T for fixed compositions of the other phases.  The Al-in-hornblende barometer is often applied to obtain pressures of granitoid emplacement without consideration of the composition of plagioclase (or of biotite and K-feldspar) or reduced activities of SiO₂/H₂O.  Experimental calibrations were obtained only in the presence of H₂O fluid and did not evaluate the effect of reducing its activity.  Reactions with clays and organic materials are kinetically controlled, depending on heating rates and time at peak T as well as T.  Raman measurements of carbonaceous materials in metamorphic rocks do not obviate these concerns.

Several new thermometers have been proposed recently in CaO-TiO₂-ZrO₂-SiO₂ involving Ti in zircon or quartz, and Zr in rutile or sphene.  Original calibrations assumed that pressure had no influence on the results despite a significant volume change in buffering reactions.  Calibrations and applications included experiments and natural assemblages in the absence of a full buffering reaction, e.g. with rutile.  A high TiO₂ activity (0.5-0.8) was assumed based on garnet-bearing assemblages at high pressures and in application to granitoids lacking both garnet and rutile.  Studies of Ti in zircon from Archean granitoids in support of an early wet/cool earth have large systematic errors and are not recommended without explicit correction for all reduced activities.