Reversed experimental calibration of the garnet-clinopyroxene Fe — Mg exchange thermometer

Abstract

The temperature-sensitive Fe,Mg exchange equilibrium,

$$\begin{gathered} = {1 \mathord{\left/ {\vphantom {1 3}} \right. \kern-\nulldelimiterspace} 3}Mg_3 A1_2 Si_3 O_{12} + CaFeSi_2 O_6 \hfill \\ {\text{ pyrope hedenbergite}} \hfill \\ \hfill \\ {\text{ = }}{{\text{1}} \mathord{\left/ {\vphantom {{\text{1}} 3}} \right. \kern-\nulldelimiterspace} 3}Fe_3 A1_2 Si_3 O_{12} + CaMgSi_2 O_6 \hfill \\ {\text{ almandine diopside}} \hfill \\ \end{gathered}$$

has been investigated experimentally at 15 and 29 kbar and 700°–1200° C, with 59 reversals and 6 half reversals of the equilibrium clinopyroxene compositions from synthetic crystalline solid solutions. The effect of Ca in garnet was investigated with grossular mole fractions of 0.2, 0.25, 0.3, 0.4 and 0.5. Garnet compositions in garnet-clinopyroxene reversal mixes were maintained constant during the runs by overwhelming abundance of garnet (90–95%). Reequilibration of rims of high-Fe and low-Fe clinopyroxene seeds gave tight reversal brackets. Ca and Al contents of the clinopyroxenes showed systematic variations with T and bulk compositions but were not reversed. The results calibrate a geothermometer for garnet amphibolites, granulites and eclogites with clinopyroxene low in Na. The garnet-clinopyroxene distribution curves are asymmetric, in contrast to those of the previous experimental studies of Råheim and Green (1974) and Ellis and Green (1979). There is an increase of indicated temperature of about 100° C per 10 mole percent increase of grossular in garnet and a pressure effect of about 5.5° C/kbar. Overall, the present calibration generally gives temperatures 0–60° C lower than Råheim and Green, 60–150° C lower than Ellis and Green, and 100–300° C lower than other semi-empirical and theoretical scales. When applied to many samples from granulite facies terrains, the general result is a spread of calculated temperatures, many in the range 550–650° C, with some in the range 700–800° C. This appears to indicate that retrograde resetting to lower temperatures is common and that the most reliable indication of peak or near-peak temperatures for a given terrain is the upper part of the calculated temperature range. The Fe,Mg distribution curves of this study require asymmetric non-ideal mixing in diopside-hedenbergite. The magnitude of nonideality is not as large as found by modelling Fe,Mg distributions with ilmenite (Bishop 1980) or orthopyroxene (Saxena et al. 1986).