Nature of the depleted upper mantle beneath the Atlantic:: evidence from Hf isotopes in normal mid-ocean ridge basalts from 79°N to 55°S

Light rare earth depleted Mid-Atlantic Ridge (MAR) basalts unaffected by hotspots (i.e. NMORB) show a very regular decrease in Hf-176/Hf-177 from 79degreesN to 55degreesS. The north-south nearly linear gradient ranges from 0.283448 (epsilon(Hf)=23.9) to 0.283137 (epsilon(Hf)= 12.9). This remarkable gradient in NMORB suggests a similar gradient in the depleted upper mantle whose probable origin is explored here. Comparison with previously collected isotope data shows that this regular north to south epsilon(Hf) gradient is accompanied by a less pronounced decrease in epsilon(Nd) and an increase in Pb-207/(204) Pb. In addition to this survey of Atlantic NMORB, Hf-176/Hf-177 are reported for 11 plume-affected enriched MORB from the MAR. Each EMORB is the most LREE enriched basalt from a region of plume-ridge interaction. Hf-176/Hf-177 of these EMORB vary between 0.282891 (epsilon(Hf)= 4.2) and 0.283198 (epsilon(Hf)= 15.1). The epsilon(Hf) variation of these EMORB is likely due to simple plume-ridge interactions. Isotope-isotope correlations for the NMORB suggest that the regular gradient in eFff is probably not caused by mixing with recycled oceanic crust, subcontinental lithosphere or plume-head restite. Rather it is likely caused by mixing with a depleted mantle component that is the residue of ancient mantle melting in the presence of gamet. The correlation between isotope ratios and MAR-to-continent distance suggests that this gamet bearing restite was sequestered in the asthenosphere under the continents until the opening of the Atlantic. An alternative to mixing models is that differences in the evolution of the upper mantle may have given rise to the NMORB isotope gradients. Simple two-stage models would suggest that even small differences in how much melt was extracted from the mantle or when melt extraction began could have caused the observed gradient. In both cases, however, these small differences in evolution must be longstanding, on the order of 1 Ga or more and would have to be gradual on a 15,000-km scale. (C) 2004 Elsevier B.V. All rights reserved.