LAUSR

Because of the strong preference of fluorine (F) for “hydrous” magnesian silicate minerals, the temperature T(0F) given by the reaction (Bi–Gt) of Fe–Mg exchange between biotite and garnet [phlogopite (Phl) + almandine (Alm) = annite (Ann) + pyrope (Py)] can be dramatically underestimated when the reciprocal effects between octahedral and hydroxyl site occupancies in biotite are ignored. In the granulites of southeastern Madagascar, widespread Mg metasomatism was associated with F-rich fluids. It is mainly manifested by the regional occurrence of diopside skarns hosting F-phlogopite and thorianite mineralisation. At first, biotite- and garnet-bearing granites and leucogneisses appear unaffected by this event. However, their biotites are rich in F [up to 6 wt%; XF = XF/(F + OH) = 0.69) and in Mg (XMg = Mg/(Mg + Fe) up to 0.8], which shows the tendency of equilibrium with the regional fluids. This equilibration leads to low values of Fe/Mg, which in turn cause strong underestimation of the temperature: T(0F) as low as 500 °C, whereas the temperature T(Ti) given by the TiO2 content of biotite is approximately 750 °C. The slightly modified thermodynamic model of Zhu and Sverjensky (Geochim Cosmochim Acta 56:3435–3467, 1992) for reciprocal solid solution is applied to a set of 19 samples to correct this underestimation and to quantify the effect of F on thermometry. The reciprocal activity coefficients for phlogopite (λ(Phl)) and annite (λ(Ann)) calculated by the model are added to the winTWQ software to obtain corrected values T(F) of the temperature. Similar values of ΔT = T(F) − T(0F) are obtained by (1) the linear fit of the whole set of the samples and (2) RTlnKλ = RTlnλ(Ann) − RTlnλ(Phl) = 33.417XF − 0.1092 (r2 = 0.9994) equation applied to a median sample: ΔT = 30–40, 60–70, 95–100, 160, and 200 °C for XF = 0.1, 0.2, 0.3, 0.5, and 0.7, respectively. The same calculations based on the Thermocalc thermodynamic data set lead to values of ΔT 1.5 times higher than those given by winTWQ. Thus, the results of these calculations confirm (for the first time from a set of natural samples) the validity of the thermodynamic model of reciprocal solid solution and the necessity of accounting for F in thermometry based on Fe–Mg exchanges involving biotite.