FERROCOLUMBITE-MANGANOTANTALITE TRENDS IN GRANITES AND PEGMATITES: EXPERIMENTAL AND NATURAL CONSTRAINTS

It has long been recognized that the chemical compositions of columbite-tantalite record fractionation trends in granites and granitic pegmatites, from more primitive ferrocolumbite to highly fractionated manganotantalite (e.g. Cerny et al., 1986). The solubility experiments of Linnen and Keppler (1997) explained why Nb/Ta decreases with fractionation, but Mn/Fe trends in natural columbites are more complex, and remain poorly understood. To clarify this, the solubilities of FeNb₂O₆ and FeTa₂O₆ have been determined at 800°C, 200 MPa and fO₂ near Ni-NiO, for the 200 MPa haplogranitic minimum melt composition (ASI=1.0). These values are ~1x10^-3 and 2x10^-3 mol²/kg², respectively, which are roughly an order of magnitude greater than those of the Mn end-members at the same conditions. The lower solubilities of the Mn end-members indicate that columbite crystallization should result in a decrease of the Mn/Fe ratio of the residual melt. This is the opposite of most natural trends, indicating that columbite-tantalite crystallization does not control the Mn/Fe ratios in highly evolved granitic melts. The Mn-Fe compositions of columbite-tantalite reflect, in part, the control of Mn-Fe concentration in the melt by the crystallization of other phases. Crystallization of garnet should result in a decrease of the Mn/Fe ratio in the residual melt whereas the crystallization of tourmaline and muscovite should result in an increase of the Mn/Fe ratio in the residual melt. Inflections in columbite-tantalite composition trends are likely related changes of phases on the liquidus.