CHALLENGES IN PHASE DIAGRAM MODELING, FOCUSING ON METAPELITES

Phase diagrams can now be calculated with great speed and can predict detailed modal and chemical information for various rock compositions, chemical systems and pressure-temperature conditions. The question is whether these computational advances have outstripped the experimental and natural data used to constrain the thermodynamic data needed for these increasingly sophisticated calculations, and the petrographic and chemical analysis of natural samples to evaluate whether the predictions reproduce what is seen in nature. The latter includes the problem of evolving effective bulk composition during metamorphism, and recognition of the possibility of kinetic impediments to the continuous equilibrium assumption implied by these diagrams. The following are some examples of possible problem areas in metapelites: (1) too-large stability for staurolite? Whereas there are many examples of staurolite-to-andalusite and staurolite-to-kyanite piezothermal arrays (~ metamorphic field gradients), there seem to be few examples of kyanite- and andalusite-free staurolite-to-sillimanite sequences, contrary to what is expected from current phase diagrams. (2) slope of cordierite-to-andalusite reaction? This continues to be a problem in modeling contact aureoles and other low-pressure settings. (3) predicted compositional changes in micas and chlorite greater than observed in nature? Because these phases may be modally more abundant than porphyroblast phases, they can exert a significant influence on predicted phase equilibria in mass-balance constrained calculations like phase diagram sections (pseudosections). (4) Fe2+/Fe3+? Most recognize this as a potentially important factor, yet our current and past inability to measure this routinely, in either experiment or nature, is a major hurdle. (5) predicted pressures derived from phase equilibrium modelling higher than predicted from overburden analysis. This seems to be a problem especially in lower pressure metamorphic settings.