Crystallization of Heterogeneous Pelitic Migmatites: Insights from Thermodynamic Modelling

Pelitic migmatites are texturally and mineralogically heterogeneous owing to variable proportions of light and dark coloured domains (leucosome and melanosome), the degree to which the two domains are segregated from one another and the variable composition of the leucosome (tonalite-trondhjemite to alkali feldspar granite). We use thermodynamic modelling to (1) provide insights into the variation in leucosome composition and solidification conditions for different melting and crystallization processes and (2) address the practical problem of how to sample heterogeneous migmatites for the purpose of constraining the pressure-temperature conditions of their formation. The latter is challenging because the appropriate bulk composition is affected by the above heterogeneity, as well as by the possibility of melt loss. Both dehydration melting and excess-water melting (for 2.0wt % excess water) are simulated for both equilibrium and fractional melting endmember processes. Three end-member processes are considered during the crystallization stage: (1) fractional crystallization of the melt; (2) equilibrium crystallization of just the melt in isolation from the solid phases; (3) crystallization during which melt and solids maintain chemical communication. Loss of melt during heating and cooling are also considered. Some of the key results of our simulations of pelitic migmatites are as follows. (1) Leucosome composition is primarily a reflection of the melt composition and to lesser degrees the crystallization process and the temperature of melt loss during cooling. Granite (sensu stricto) is the most common leucosome type, arising from many melting and crystallization processes, whereas tonalite or trondhjemite leucosome is generally indicative of excess-water melting. (2) Melts lost from partially molten regions, which have the potential to coalesce and form granites at shallower crustal levels, show less compositional variation than leucosomes in migmatites. Extracted melts are monzogranitic, or rarely granodioritic at low temperatures. (3) Comparing plagioclase compositions between leucosome and melanosome is potentially an effective means of distinguishing between crystallization processes, as well as the degree of retrograde melt-leucosome-melanosome chemical interaction. Fractional crystallization produces zoned plagioclase in the leucosome, isolated equilibrium crystallization produces plagioclase of different composition in the leucosome and melanosome, whereas chemical interaction causes the leucosome and melanosome plagioclase compositions to remain similar during crystallization. (4) The peak temperature of heterogeneous migmatites is most reliably constrained from an equilibrium phase assemblage diagram calculated using just the melanosome composition. Addition of leucosome to the melanosome composition can lead to peaktemperature estimates that differ from actual peak conditions by -25 to +50 degrees C. In extreme cases, such as in rocks containing high proportions of leucosome to melanosome, or in which K-feldspar was absent as a solid phase at peak-temperature conditions, the assemblage developed in the rock at peak temperature may not appear in the phase diagram at all. (5) The solidification temperature (the temperature at which a partially molten rock changes to a completely solid migmatite, owing to either crystallization of the last melt or physical loss of the melt) can be accurately constrained if the melt and solids chemically communicated during cooling. If the melt crystallized in isolation from the melanosome, the solidification temperature is the temperature at which the last melt was physically lost or that of a water-saturated granitoid solidus (625- 645 degrees C). In the latter case, the solidification temperature may be much higher, but can only be estimated with a precision of tens to hundreds of degrees Celsius.