Abstract:
Laser ablation inductively coupled plasma mass spectrometry was used to measure in situ the major, moderately and highly siderophile (MSE and HSE) and refractory lithophile element (RLE) abundances of chondrules, calcium–aluminium-rich inclusions (CAIs), matrix and metals from five chondrites. The analysed chondrites are unequilibrated and included three carbonaceous (Allende, NWA 2364 and NWA 763), one ordinary (Bovedy) and one enstatite (SAH 97096) chondrite. The elemental data were ratioed to Si or Ca and then normalised to CI (Ivuna-like carbonaceous) chondritic element/Ca or Si compositions to circumvent the fact that internal normalisation of the data cannot be carried out due to each analysis ablating mineralogically heterogeneous material. MSE and HSE data reveal ca. 1–2 order of magnitude depletions in chondrules relative to bulk CI chondrites. Measured relative MSE and HSE abundances are largely inconsistent with a volatility control that would be predicted by nebular chondrule formation models dominated by condensation– evaporation processes. The pattern of MSE and HSE depletions are more consistent with elemental metal–silicate partitioning and explainable by metal–silicate segregation. Expulsion of metal droplets from spinning molten chondrules under low gravity is shown to be an unlikely physical mechanism for generating these features. Therefore, a model in which chondrules are derived from the mantles of molten planetesimals, perhaps by impact “splashing” or volcanism, subsequent to metal extraction from planetesimal mantles during core formation is a plausible mechanism that can account for the observed chondrule MSE and HSE depletions. This model is strengthened by metal MSE and HSE data from Bovedy that have broad similarities with magmatic and primitive iron meteorites, suggesting that at least some chondritic metal grains may have originated in the cores of differentiated planetesimals during their near complete disruption induced by impacts. Significant variability in HSE and RLE compositions of chondritic solids have important implications for the utilisation of the putative bulk chondrite composition as a framework for models of Earth’s accretion, differentiation and geochemical reservoirs. This arises as it cannot be assumed with confidence that Earth accreted from an identical proportion of chondrules, calcium–aluminium-rich inclusions (CAIs), matrix and metals as present in bulk chondrites. For example, given the marked contrast between chondrule and matrix HSE abundances, the average HSE composition and, therefore, estimated total mass of the late veneer added to Earth after core formation is strongly dependent on the relative matrix–chondrule proportions in the late veneer. Similarly, RLE abundances in individual chondrite components display significant variability in Sm/Nd and Lu/Hf ratios. A correspondingly wide range of present-day Nd- and Hf-isotopic compositions (ε143Nd = –24 to +32 and ε176Hf = –34 to +168) can be achieved by different mixtures of these components, suggesting that the initial Sm/Nd and Lu/Hf of Earth’s primitive mantle would be highly sensitive to the relative proportions of chondrules, matrix and CAIs that Earth accreted from. This calls into question the long held assumption that the Earth accreted from material that had a bulk chondritic composition. Only a small difference in the relative abundances of these accreted chondritic components as compared with bulk chondrites could have resulted in a Bulk Silicate Earth with slightly superchondritic Sm/Nd ratios. As such some ocean island and continental and ocean flood basalts conventionally viewed as being “depleted” and having positive ε143Nd values could potentially be derived from near-primordial mantle. Moreover, this could also be consistent with the ubiquitous positive offset in 142Nd anomalies observed between all (non-Archean) terrestrial samples and chondrites.