Differentiation mechanisms of the early Hadean mantle: Insights from combined 176Hf-142,143Nd signatures of Archean rocks from the Saglek Block

Positive 142Nd anomalies in Eoarchean rocks provide evidence for early (4.40 ± 0.03 Ga; Morino et al., 2017) depletion of Earth's mantle. This model age lies within errors of the Pb-Pb “age of the Earth” (Connelly and Bizzarro, 2016), and is similar to model ages inferred for crystallization of the lunar mantle (McLeod et al., 2014), implying that this large-scale event may reflect crystallization of a magma ocean following the Moon-forming impact. However, differentiation mechanisms responsible for the formation of this early depleted mantle reservoir and the depth at which it formed cannot be constrained from the Sm-Nd isotope system alone, because the magnitude of Sm/Nd fractionation during partial melting or fractional crystallization shows little dependence on pressure-controlled changes in mantle mineralogy. In contrast, the Lu-Hf isotope system is highly dependent on mineralogy, notably the presence or absence of garnet, an upper mantle phase, and thus may be used to constrain the pressure of fractionation. This study provides the first 176Lu-176Hf isotopic results on mafic and ultramafic rocks belonging to the Eoarchean (Nulliak) and Mesoarchean suites of the Saglek Block (northern Labrador, 3.2-3.9 Ga). The 176Lu-176Hf dating confirms the distinction between these two groups of rocks and provides ages consistent with those obtained from 147Sm-143Nd dating. The whole rock 176Lu-176Hf errorchrons yield ages and initial epsilon values of 3766 ± 140 Ma, ε176Hfi = 6.0 ± 2.5 and 3023 ± 390 Ma, ε176Hfi = −0.3 ± 2.5 for the Nulliak suite and the Mesoarchean suite respectively. The time-integrated 176Lu/177Hf for the sources of the Nulliak and the Mesoarchean suites considering a time of differentiation at 4.40 ± 0.03 Ga are estimated to be 0.047 ± 0.005 and 0.033 ± 0.005, respectively. For the Mesoarchean samples, the combined 146,147Sm-142,143Nd and 176Lu-176Hf data are consistent with a near-chondritic mantle source. On the other hand, Nulliak ultramafic rocks were derived from a mantle reservoir with superchondritic Lu/Hf and Sm/Nd. The Nulliak parent reservoir, however, does not plot on the ε176Hf-ε143Nd mantle array defined by modern oceanic basalts. Instead, the Nulliak source more likely belongs to a distinct array defined by Eo- and Meso-Archean komatiites. These results are interpreted in the framework of a simple model of crystallization of a primordial magma ocean. It appears that the fractionation observed in the mantle source of Nulliak was most likely generated by crystallization of a garnet-bearing assemblage in the shallow mantle, above the transition zone rather than by perovskite fractionation in the lower mantle. To preserve this depleted reservoir from the rest of the hot and vigorously convecting mantle, the Nulliak mantle source may have been isolated either at the top of the mantle in a buoyant lithosphere or near the core-mantle boundary, with the latter setting being more consistent with the komatiitic nature of the erupted rocks. Given that the garnet signature argues for differentiation of the Nulliak source at relatively shallow depth (few hundred kilometers), its isolation in the deep mantle would require a cumulate overturn following crystallization of the magma ocean.