Considerable geochemical evidence supports initiation of plate tectonics on Earth shortly after the end of the Hadean. Nb/Th and Th/U of mafic-ultramafic rocks from the depleted upper mantle began to change from 7 to 18.2 and 4.2 to 2.6 ( respectively) at 3.6 Ga. This signals the appearance of subduction-altered slabs in general mantle circulation from subduction initiated by 3.9 Ga. Juvenile crustal rocks began to show derivation from progressively depleted mantle with typical igneous epsilon(Nd):epsilon(Hf) = 1: 2 after 3.6 Ga. Cratons with stable mantle keels that have subduction imprints began to appear by at least 3.5 Ga. These changes all suggest that extraction of continental crust by plate tectonic processes was progressively depleting the mantle from 3.6 Ga onwards. Neo-archean subduction appears largely analogous to present subduction except in being able to produce large cratons with thick mantle keels. The earliest Eoarchean juvenile rocks and Hadean zircons have isotopic compositions that reflect the integrated effects of separation of an early enriched reservoir and fractionation of Ca-silicate and Mg-silicate perovskite from the terrestrial magma oceans associated with Earth accretion and Moon formation, superposed on subsequent crustal processes. Hadean zircons most likely were derived from a continent-absent, mafic to ultramafic protocrust that was multiply remelted between 4.4 and 4.0 Ga under wet conditions to produce evolved felsic rocks. If the protocrust was produced by global mantle overturn at ca. 4.4 Ga, then the transition to plate tectonics resulted from radioactive decay-driven mantle heating. Alternatively, if the protocrust was produced by typical mantle convection, then the transition to plate tectonics resulted from cooling to the extent that large lithospheric plates stabilized.

The lithium isotope compositions of 30 well-characterized samples of glassy lavas from the three major mid-ocean ridge segments of the world, spanning a wide range in radiogenic isotope and elemental content and sea floor physical parameters, have been measured. The overall data set shows a significant range in delta(7) Li (+1.6 to +5.6), with no global correlation between Li isotopes and other geochemical or tectonic parameters. The samples with the greatest lithophile element depletion (NMORB: K2O/TiO2 < 0.09) display an isotopic range consistent with the extant database. Samples with greater trace element enrichment display a greater degree of isotopic variability and trend toward heavier compositions (delta(7) Li = +2.4 to +5.6), but are not distinct on average from N-MORB. Together with published data, N-MORB is estimated to have mean delta Li-7 = +3.4 +/- 1.4 parts per thousand (2 sigma), consistent with the estimate for an uncontaminated MORB source based on pristine peridotite xenoliths. Locally, where sampling density permits, sources of Li isotope heterogeneity may be evaluated. Sample sets from the East Pacific Rise show correlation of delta Li-7 with halogen concentration ratios. This is interpreted at 15.5 degrees N latitude to represent incorporation of <5 weight percent recycled subduction-modified mantle in the MORB source. At 9.5 degrees N latitude the data are more consistent with shallow level magma chamber contamination by seawater-derived components (<0.5 wt.%). (C) 2008 Elsevier Ltd. All rights reserved.

The emplacement of the 2.05- billion- year- old Bushveld complex, the world's largest layered intrusion and platinum- group element ( PGE) repository(1), is a singular event in the history of the Kaapvaal craton of southern Africa(2-4), one of Earth's earliest surviving continental nuclei. In the prevailing model for the complex's mineralization, the radiogenic strontium and osmium isotope signatures of Bushveld PGE ores are attributed to continental crustal contamination of the host magmas(5-11). The scale of the intrusion and lateral homogeneity of the PGE- enriched layers(1), however, have long been problematical for the crustal contamination model, given the typically heterogeneous nature of continental crust. Furthermore, the distribution of Bushveld magmatism matches that of seismically anomalous underlying mantle(3,12), implying significant interaction before emplacement in the crust. Mineral samples of the ancient 200- km- deep craton keel, encapsulated in macrodiamonds and entrained by proximal kimberlites, reveal the nature of continental mantle potentially incorporated by Bushveld magmas(13,14). Here we show that sulphide inclusions in 2- billion- year- old diamonds from the 0.5- billion- year- old Venetia and 1.2- billion- year- old Premier kimberlites ( on opposite sides of the complex) have initial osmium isotope ratios even more radiogenic than those of Bushveld sulphide ore minerals(6,15). Sulphide Re - Os and silicate Sm - Nd and Rb - Sr isotope compositions indicate that continental mantle harzburgite and eclogite components, in addition to the original convecting mantle magma, most probably contributed to the genesis of both the diamonds and the Bushveld complex. Coeval diamonds provide key evidence that the main source of Bushveld PGEs is the mantle rather than the crust.