An overview of adakite, tonalite-trondhjemite-granodiorite (TTG), and sanukitoid: relationships and some implications for crustal evolution

Examination of an extensive adakite geochemical database identifies two distinct compositional groups. One consists of high-SiO2 adakites (HSA) which is considered to represent subducted basaltic slab-melts that have reacted with peridotite during ascent through mantle wedge. The second group consists of low-SiO2 adakites (LSA) which we interpret to have formed by melting of a peridotitic mantle wedge whose composition has been modified by reaction with felsic slab-melts. The chemical composition of less differentiated (primitive) Archaean tonalite-trondhjemite-granodiorite (TTG) magmas evolved from 4.0 to 2.5 Ga. Mg# (molecular Mg/(Mg+Fe2+), Ni, and Cr contents increased over this period of time and we interpret these changes in terms of changes in the degree to which the TTG magmas interacted with mantle peridotite. Over the same period, concentrations of (CaO+Na2O) and Sr also increased, as the amount of plagioclase, residual from basalt melting, decreased in response to increased pressures at the site of slab-melting. In the Early Archaean, it appears that these interactions were very rare or absent thus leading to the conclusion that subduction was typically flat and lacked the development of a mantle wedge. In contrast, the relatively lower heat production by similar to2.5 Ga meant that slab-melting occurred at greater depth, where plagioclase was no longer stable, and where the development of a thick mantle wedge ensured interaction between the slab-melts and mantle peridotite. Close compositional similarities between HSA and Late Archaean TTG (T < similar to3.0 Ga) strongly suggest a petrogenetic analogy. However, an analogy between the older Archaean TTG and HSA is not complete because evidence for mantle wedge interaction is missing in most Early Archaean TTGs. Late Archaean sanukitoids and the compositionally similar Closepet-type granites have compositions significantly different from TTG of all ages. However, they show some affinity with LSA which could be considered as their possible analogue. These magmas are all thought to result from melting of a mantle peridotite whose composition has been modified by reaction with slab-melts. We propose that all these magmas are directly linked to slab-melting. Archaean TTG and HSA represent slab-melts that have interacted with peridotite to varying extent, whereas sanukitoids, Closepet-type granites, and LSA correspond to melts of peridotite previously metasomatised by slab-melt. The changes observed from Early Archaean TTG to Late Archaean TTG and to sanukitoids reflect change in both the nature and efficiency of interaction between slab-melt and mantle wedge peridotite. Comparisons between all of these rocks suggest that ancient styles of subduction that have operated since at least similar to3.3 Ga persist in a limited way today. The secular changes in the degree and style of these interactions is a direct consequence of the cooling of Earth that modified the thermal and dynamic parameters at the subducted slab-mantle wedge interface. (C) 2004 Published by Elsevier B.V.