Knowledge of the P-T location and stoichiometry of the fluid-absent reaction Phl + Qtz = En + Sa + M is required for a better understanding of crustal anatexis and crystallization and the stability of biotite in granitoid magmas. This equilibrium has been experimentally investigated by Bohlen et al. (1983), Montana and Brearley (1989), and Peterson and Newton (1988, 1989). However, there has been consensus neither on the chemographic relationships among the possible phases involved in this equilibrium nor on its precise P-I location. We have carried out experiments, at pressures between 100 and 1500 MPa and temperatures between 800 and 920 °C, in order to remove or explain these uncertainties. The following brackets were obtained for the reaction: 100 MPa, 805–815 °C; 150 Mpa, 798–811 °C; 200 MPa, 808–821 °C;300 MPa, 810–821 °C; 500 MPa, >838 °C; 800 MPa, 860-865 °C; 1000 MPa, 849-876 °C, 1500 MPa, 909-921 °C. In addition, we reversed the equilibrium at 1000 MPa. Our experiments confirm the presence of sanidine as a reaction product over the pressure range investigated.
Because the reaction Sa + Qtz + Fl = M nearly coincides with Phl + Sa + Qtz + Fl = M and En + Sa + Qtz + Fl = M, it can be considered as an excellent analogue of the solidus in the KMASH system. It is possible to calculate tlire P-T location of this solidus as a function of aH2O using models for the interactions between H2O and aluminosilicate melts. Using available thermodynamic data, we can also calculate the shift of the subsolidus reaction Phl + Qtz = En + Sa + Fl as a function of decreasing aH2O. The intersections of these curves, at the same aH2O, provide a theoretical location for the fluid-absent melting reaction. This calculated curve has a slope very similar to, and only 10–15 °C above, our experimental curve. This modeling, together with the composition of the phases, allows us to calculate that the fluid-absent reaction will produce up to 50 ° 15 wt% melt in the pressure range 100–1500 MPa.
On the basis of the experimentally determined locations of the solidus reactions (Phl + Qtz + Sa + Fl = M and En + Sa + Qtz + Fl = M) and the subsolidus equilibrium (Phl + Qtz = En + Sa + Fl), Grant (1986) concluded that there should be a thermal divide on the liquidus below 500 MPa. Our experiments and modeling of the solubility of H2O in the melts suggest that a thermal divide may well exist but that it must be restricted to a very narrow domain between 65 and 100 MPa, at about 810 °C. Additional components such as Ti and F should have the effect of significantly extending such a thermal barrier.