Abstract

Experimental work since the late 1980s has shown that hydrous basaltic rocks and their metamorphosed equivalents greenstone, amphibolite and eclogite, classically regarded as protoliths for Archean TTD suites, are also important source rocks for a number of Na-rich Phanerozoic granitoids comprising the large Mesozoic Cordilleran plutonic complexes of theWestern Americas, Antarctica and New Zealand. Implicit in petrogenic models developed from these experimental studies is that the source region and the heat source itself is pre-existing or newly accreted mafic underplate. Coupled thermo-chemical models of partial melting due to the periodic intrusion of basaltic magma suggest that the ratio of the period of magma intrusion (ti) to a characteristic timescale for heat loss (td) defines an important variable R that can be used to assess the thermal behaviour of the melting column1, with both melt temperature and average (maximum) melt fraction maximised where R = 1. The ratio R is used as an indicator of composition (expressed through changes in REE content) of a hypothetical partial melt, with smaller numbers of thicker intrusions (R ! 0) resulting in higher modelled La/Yb(N) ratios. Thermo-mechanical models of buoyancy driven segregation of the partial melt fraction2 suggest that the spatial distribution of melt within the source region is governed primarily by the magnitude of a dimensionless parameter termed the "effective thermal diffusivity" (keff ). The value of keff depends upon the characteristic compaction length- and timescales, in addition to the usual thermal parameters. As the melt migrates upwards, it thermodynamically equilibrates with progressively cooler matrix, so the higher temperature fractions freeze out. Consequently, the melt which forms the magma typically has a composition which corresponds to only a small degree of partial melting of the protolith, which compositionally is a small melt fraction. This is important as TTG melt compositions are obtained only for small degrees of partial melting of a mafic protolith. The model predicts that the melt in the porosity wave evolves from tonalite to granodiorite through to trondhjemite, over timescales ranging from 4,000 yrs. to 10 M.y. This sequence of melt extraction is in agreement with the order of intrusion of Archean TTG suites, and recent high-Na suites such as the Miocene Cordillera Blanca Batholith, Peru. Moreover, the predicted restite mineralogy agrees with that observed in many granulite terranes.