PHOSPHORUS AND HIGH-FIELD STRENGTH ELEMENT ANOMALIES IN ARCHEAN HIGH-MAGNESIAN MAGMAS AS POSSIBLE INDICATORS OF SOURCE MINERALOGY AND DEPTH

Two suites of late Archean (almost-equal-to 2.7 Ga) high-MgO rocks in the Yilgarn Block, Western Australia and the Superior Province, Canada display negative P anomalies relative to Pr and Nd on primitive mantle normalized diagrams. In contrast, most late Archean high-magnesian basalts and recent oceanic basalts are characterized by smooth primitive mantle normalized patterns through Pr-P-Nd. Negative P anomalies occur in many mafic Phanerozoic arc and intracontinental alkaline magmas, and may be accounted for by a number of processes, including: (1) contamination by continental crust, (2) hydrous metasomatism of depleted mantle, (3) fractionation of a phosphorus-bearing phase from the magma, and (4) retention of a phosphorus-bearing phase in the source region of the magma. In mafic magmas, P, Nb and Ti are all relatively depleted by crustal contamination and hydrous metasomatism, so that negative P anomalies are accompanied by corresponding Nb and Ti anomalies. In oceanic basalts, P abundance is correlated with the LREE. The Archean high-MgO (15.2-29.4 wt%) suites do not show the normalized Nb/ LREE and Ti/MREE fractionations indicative of hydrous mantle metasomatism, and they have high Cr (940-1340 ppm) and Ni (1380-1960 ppm) contents. This collectively rules out crustal contamination of the magma and hydrous metasomatism of the mantle source. Theoretical considerations and experimental data indicate that phosphorus undergoes progressively greater substitution into garnet with increasing pressure. Modelling of P/REE and HFSE/ REE systematics during partial melting and fractional crystallization processes indicates that the trace element geochemistry of the two suites of Archean high-MgO rocks is consistent with retention of garnet (majorite-pyrope solid solution) in the mantle source, a conclusion supported by the extreme Al depletion (Al2O3/TiO2 = 4-5), HREE fractionation (Gd/Yb(n) = 2.0-2.5) and the negative normalized Sc and V anomalies of these rocks. This distinctive geochemical signature is also present in most early Archean and rare late Archean Al-depleted komatiites. Mantle phase relationships and density considerations suggest that this signature can only develop in melts from mantle plumes in which melting commenced between 8 and 24 GPa, where pyrope-majorite garnet is expected to be the dominant fractionating or residual phase. In addition, the melts would have undergone little re-equilibration with shallower asthenosphere and lithosphere. These requisite conditions may explain the restriction of this distinctive geochemical signature to the Archean, where mantle sources melted at deeper levels in a hotter mantle. This study demonstrates that P/REE fractionations, when allied with major element and low-level trace element data, can provide insights into melting depths and the petrogenesis of mafic rocks.