THE ORIGIN OF CONTINENTS AND EPISODIC EVOLUTION OF THE EARTH

The principal reason for proposing that plate tectonics operated through the Archean comes from models for the formation of granitoids that do not invoke subduction. These models are not very plausible. Melting at the base of an oceanic plateau is not credible because this site lacks the water needed to produce large volumes of granitic magma. In the Archean, as now, most granites probably formed through reprocessing of hydrous basaltic magma in the crust overlying a subduction zone. Compilations of U-Pb ages of zircons from granitoids and major rivers show large peaks at 2.7, 2.5, 2.1, 1.9 and 1.1 Ga. These peaks had previously been equated with periods of accelerated growth of continental crust, but recent theory relates them to times of supercontinent assembly and enhanced preservation of the continental crust. We reject the latter interpretation for several reasons: (1) the age peaks do not correlate well with periods of supercontinent formation; (2) many zircon peaks directly follow periods of accelerated mantle convection, as evidenced by the eruption of komatiite and other plume-derived magmas. Using the results of fluid mechanics experiments that simulate destabilization of a dense layer at the base of the mantle, we propose that the crustal growth peaks are related to major thermochemical plumes. The plumes displaced cold material from the upper mantle, accelerating subduction and promoting a period of accelerated formation of granitic magma; i.e. a crustal growth peak. The continental crust thus formed episodically, the relict plumes becoming subcontinental lithosphere and flanking subduction zones becoming the sites of crust formation.