A reevaluation of the stability relations of andalusite: Thermochemical data and phase diagram for the aluminum silicates

Until recently, little effort has been made to evaluate critically the numerous experimental studies on andalusite stability relations. In this contribution, we carefully examine most studies done since 1966. Experimental problems include (1) difficulty in measuring small amounts of reaction, (2) Ihe effect of fibrous sillimanite and intergrown quartz, and (3) the effect of elevated Fe₂O₃ in sillimanite. By evaluating the effects of these and other factors on previous experiments, we have determined valid, self-consistent reversals and half-reversals for the kyanite = andalusite and andalusite = sillimanite equilibria. Almost all of the experimental studies have provided reversal information. We have determined the best possible molar volumes of the pure Al₂SiO₅ end-members. By combining our molar volumes and the thermochemical data of Hemingway et al. (1991) with the acceptable reversals for the above reactions and for kyanite = sillimanite, we have obtained (1) entropy and enthalpy for kyanite, andalusite, and sillimanite, and (2) a phase diagram that has a triple point at 504 ± 20°C, 3.75 ± 0.25 kbar. The andalusite = sillimanite boundary may be up to 20°C above that for pure Al₂SiO₅ because of the efects of Fe₂O₃ on sillimanite and andalusite. We emphasize the importance of careful evaluation of experiments and the secondary role of comparison of experimental results with natural systems. Determination of phase diagrams from the positions of equilibria in natural systems may produce significant errors resulting from the experimental calibration of the natural system.