Biotitegamet geothermometry in the granulite facies: the influence of Ti and Al in biotite

Temperature estimates obtained with various biotite-garnet thermometers are inconsistent in granulite facies rocks. Although the inconsistency lies partly in the lack of definition of equilibrium criteria, it still remains after rigorous selection of analyzed mineral pairs (isolated biotite and garnet cores for peak equilibrium conditions) and originate from minor element effects in garnet and biotite solid solutions. The departure from ideality due to Ca in garnet can be calculated either with the available thermodynamic data (model A), or with a combination of thermodynamic and empirical data (model B), whereas that due to Mn is probably negligible due to the low Mn content of high-grade garnets. The high Ti content of biotites seems responsible for the abnormally high temperatures obtained with an ideal lnKilT relation (e.g., the experimental calibration of Ferry and Spear). This deviation and that introduced by Al"rsubstitution in biotite can be evaluated using the classical equations of regular solutions provided independent T estimates are available. Ti and Al interaction parameters have thus been evaluated as: m: (l7B'"rr WlirJ : -464 or -1590 and n: (Wqt"rrWli"i: -6767 or -1451 (in cal per mole) depending on the model adopted for the non-ideality in garnet. Starting from the experimental ideal model of Ferry and Spear, a new calibration of the biotite-garnet geothermometer based upon the above correction for Al and Ti and the interaction of Ca in Fe-Mg garnets is proposed here: 12454 -0.057P (bar) + 3 (m Xl' + n Xf) ( Wc"Xx:. + LWM"X#") T " K : Application of this calibration to granulite facies rocks gives results compatible with those obtained using Thompson's calibration. 1969), one experimental (Ferry and Spear, 1978) and two empirical calibrations (Thompson, 1976; Goldman and Albee, 1977) have been used with improved confidence to evaluate temperatures in medium to high-grade pelitic and semipelitic rocks. In rocks of middle amphibolite facies, the latter three calibrations give relatively consistent results. In rocks of granulite facies, however, severe inconsistencies in garnetbiotite temperatures led Bohlen and Essene (1980) to state that existing calibrations of biotite-garnet thermometry are inadequate to evaluate temperature variations in highgrade metamorphic terranes. If one tries to find reasons for such inconsistency, a discrimination should be made between reasons inherent in the calibrations and those resulting from the ambiguity in the determination of equilibrium criteria. In granulite facies rocks, ambiguity in equilibrium criteria is introduced by the presence of significant Ko differences on the scale of the thin section due to local retrograde reequilibration. This point can be resolved by INDIRES AN D M ART IGN O LE : BIOT IT E_GARN ET GEOT H ERM O M ET RY I N GRAN T] LIT E 273 selecting particular locations for analysis within minerals in such a way that calculated Ko can be attributed with some confidence to a specific stage of evolving metamorphic conditions (Indares and Martignole, 1985). It is recognized that in high grade rocks, a pervasive equilibrium is established between garnet and biotite. Although, during cooling, garnet rims and biotite grains adjacent to garnet are affected by late FrMg exchange, garnet cores and matrix biotite isolated from garnet preserve their high grade compositions, and are suitable for calculating'peak temperatures, In Contrast, from the compositions of garnet rims and adjacent biotite, temperature from some stage during cooling is obtained (Indares, 1982). In the present contribution we intend to reevaluate the various bi-gal geothermometers in granulite facies rocks and, from our own observations and analyses, provide an improved calibration which takes into account the affects of Ti and Al in the biotite solid-solution. Biotite-garnet geothermometets An experimental calibration of the biotite-garnet thermometer has been made by Ferry and Spear (1978) using pure Mg-Fe solid solutions. The linear correlation obtained between LnKo and llT is consistent with ideal behavior of both binary solid solutions. Since natural systems also contain significant amounts of Ca, Al, Ti, and Mn, departure from this relationship should be considered in the application of this calibration. Thompson (1976) proposed an empirical linear relationship between LnKo and 1/7. This calibration is based on natural metapelitic assemblages of low and medium grade and on independent temperature estimates obtained from various metamorphic reactions. Although it implicitly integrates minor elements substituting for Fe and Mg, the reliability of this calibration could decrease considerably in the case where the analyzed samples depart markedly in terms of minor constituents from the samples used in the calibration. Moreover, the nature of the calibration precludes the introduction of correction factors for such departures. Natural assemblages were also used in the calibration of Goldman and Albee (1977), and a relationship between Ko and oxygen isotope fractionation between quartz and magnetite was established. In order to take into account compositional effects, a relationship between Ko and the Ca and Mn content of garnet and the Fe, Ti and Al content of biotite was derived, in which Xf'" and Xfi, are the most important factors.2 The solution of Fe and Mg in biotite I Abbreviations are as follows: bi-biotite; phl-phlogopite; ga-garnet; py-pyrope; pl-plagioclase; Ksp-potash feldspar; sil-sillimanite ; qtz--qtJafiz; ilm-ilmenite ; gph-graphite. Xit : =-* --l--. * where i: Fe, Mg, Mn, Al"t, Ti ' F e * M g + M n + A l ' ' + T i X Y : J ' Fe + Mg + Mn * " where j: Fe' Mg' Mn' Ca was considered as non-ideal, and there were uncertainties in the Kp us. composition relationship. The 16O/1EO us. T dependencd was determined at temperatures below 600"C and the extrapolation to higher temperatures may not be valid. Note afuled in proof: Perchuk and Lavrent'eva (1983, Experimental investigation of exchange equilibria in the system cordierite-garnet-biotite, in Kinetics and Equilibrium in Mineral Reactions. Ed Saxena, Springer-Verlag, 199-239) have experimentally derived a new K, vs llT relationship which yields temperature estimates slightly lower than those given by Thompson's calibration. Biotite-garnet geothermometry in the Maniwaki area Samples used to test and refine available bi-ga geothermometers are from an atea of 750 km2 near Maniwaki in the Gatineau valley, 100 km north of Ottawa. This region belongs to the Central Metasedimentary Belt of the Grenville Province (Wynne-Edwards, 1972). Typical lithologies are marbles, quartzites and metapelites interlayered with ortho-amphibolites (Gauthier, 1981). In the metapelites, sillimanite is the only Al2SiOs polymorph, while cordierite occasionally forms rims around garnet or biotite. The ubiguitous presence of orthopyroxene in quartzofeldspathic rocks of suitable composition suggests that the whole area underwent granulite facies metamorphism. Maximum metamorphic temperatures (Indares and Martignole, 1984) calculated with the clinopyroxene-garnet geothermometer range from 740" to 820"C (Ellis and Green, 1981; Ganguly, 1979). Maximum pressures of 6.58.5 Kbar are obtained with the plagioclase-garnetsillimanite-q u ar tz and pyroxene-plagioclase-garnet-quartz barometers (Newton and Haselton, 1981; Perkins and Newton, 1981). Application of biotite-garnet thermometers on the Maniwaki area samples has been carried out with garnet core and matrix biotite compositions (Tables 1 and 2) considered as representing peak equilibrium (Indares, 1982). Temperatures range between 760-860"C when evaluated with Thompson's equation, and are less dispersed than those obtained from the two other calibrations (Table 3). Moreover these values are consistent with independent estimates (Indares and Martignole, 1984). Temperatures estimated according to Goldman and Albee's calibration vary between 800"-980"C; the highest values, in the range 9001065'C, are obtained using Ferry and Spear's calibration. Nowhere in the Maniwaki area do other mineral assemblages reflect such high temperatures (Indares and MarTable 1. Average garnet-core compositions S A H P L E 3 l 0 i 2 5 1 3 23-03 31-02F 29-02 36-04