THE FORMATION OF THE EARLIEST CONTINENTAL CRUST – A TRACE ELEMENT PERSPECTIVE

The early continental crust is dominated by TTG gneisses which show trace element characteristics reminiscent of modern subduction environments, including troughs in the trace element pattern at Nb and Ta, and low Nb/Ta ratios, whereas Zr/Sm ratios are higher than in modern island arcs basalts. Melting of garnet amphibolites can explain these features as long as the amphiboles are of low to intermediate Mg/(Mg+Fe). On the contrary, the melting of rutile eclogite would lead to high Nb/Ta in melts, in contrast to the patterns seen in TTG gneisses (Foley et al., 2002). Rapp et al. (2003) favoured partial melting of rutile eclogite, criticising the choice of rutile partitioning data used by Foley et al. (2002), and noting that major elements require high degrees of melting. They argued that the source already had low Nb/Ta due to stacking of subduction-related terranes, in which case the involvement of rutile may still be possible. However, the garnet amphibolite melting calculations presented by Foley et al. (2002) already included high degrees of melting, and recent rutile partitioning data confirm the problem of its opposite fractionation of Nb from Ta to that required in TTG genesis.

Nevertheless, low Nb/Ta in the source remains a possibility, but the petrological mechanism by which Nb and Ta are fractionated from each other and from other elements such as La and U, remains unexplained. The analogy of the modern subduction signature is weak, as subduction may not have operated in Archaean times. There are two main possibilities, both of which carry implications for the tectonic regime in operation. The first is partial melting which leaves amphibole in the source: although the experiments interpreted by Foley et al. (2002) contained SiO2-poorer melts than those relevant to TTG, the conclusions for Nb and Ta should apply also in SiO2-richer melts as long as amphiboles are dehydrogenated and have high DTi (Tiepolo et al. 2000). The second option is that the upper mantle was ultradepleted in the early Archaean, as implied by very low Nb/Ta in Archaean mafic rocks and eclogites. Both these mechanisms may not require the operation of plate tectonics.

S. Foley, M. Tiepolo & R. Vannucci (2002) Nature 417, 837-840; R. Rapp, N. Shimizu & M. Norman (2003) Nature 425, 605-609; M. Tiepolo et al. (2000) Earth Planet. Sci. Lett. 176, 185-201.