The study of the upper-thermal stability of ferro-actinolite (Ca2Fe5Si8O22(OH)2=Fact) by Ernst (1966, Am. J. Sci.) remains to this day the primary experimental data from which the thermodynamic properties of Fact have been derived. The petrological importance of this boundary and the recent recognition of inconsistencies in the thermodynamic data for Fact has prompted the first author to re-investigate this boundary in a study that has lasted about 5 years. Advances in analytical technology, recognition of the role of grunerite, and an enhanced thermodynamic database for minerals allows us to make some improvements on the earlier study by Ernst (1966).

Hydrothermal experiments were performed on the reaction Fact=2 Hed + 1.5 Fay + 2.5 Qtz + H2O at 1-5 kbar using the starting mixtures Ca1.8Fe5.2Si8O22(OH)2 and Ca1.6Fe5.4Si8O22(OH)2 treated at Co-CoO and Fe°-Mt/Mt-FeO, respectively, for durations of 200-2100 hours. Grunerite was present in the latter mixture. The reaction boundary was reversed between 490 - 510°C at 1 kbar, 478 - 520° at 2 kbar, 500 - 520° at 3 kbar, and 512 - 551 at 5 kbar for the first mixture, and between 491 - 500° at 1 kbar, 501 - 515° at 2 kbar, and 508 - 530° at 5 kbar. Both boundaries are essentially coincident in P-T space and both have a steeper dP/dT slope than was reported by Ernst (1966). The weak reaction signal and complications from grunerite growth/nucleation may account for the discrepancies.

Characterization of the actinolite using the analytical TEM (AEM) along with powder XRD Rietveld refinement to obtain phase proportions, unit-cell volumes, and the approximate Ca/Fe ratio on the M4 site, yielded an average Ca content of 1.66(6) apfu. Very few (<2%) chain-width defects are present. Grunerite has approximately 0.23 Ca apfu. The unit cell volume of hedenbergite (450.8ų) agrees with pure hedenbergite, as assumed here, though AEM analyses suggest up to 10 mol% Fs component. Fayalite and quartz are assumed to be pure. With the above experimental brackets, simple regular-solution theory (W=15.3 kJ) to model the Ca-Fe solid solution in actinolite, and the thermodynamic data of Holland and Powell (1998) for all other phases, one can derive an enthalpy of formation for Fact=-10,503.1(11.0) kJ and an entropy of 714.8(14.0) J/K that are consistent with the reversals.