Spinel-peridotite xenoliths from the Sidamo region (SE Ethiopia, East African Rift) range in composition from fertile, cpx-rich, lherzolites to refractory harzburgites, and show variable degree of recrystallization between two end-members. The less recrystallized samples (deformed xenoliths) are apatite-bearing porphyroclastic peridotites with low cpx/opx ratios. On the other hand, granular xenoliths consist of apatite-free peridotites with coarse-grained recrystallized textures and relatively high cpx/opx ratios.
25 samples were analyzed by ICP-MS for incompatible trace elements. Coexisting clinopyroxene, orthopyroxene, olivine and spinel were separated from representative samples and analyzed after acid leaching. Acid leachates, as well as apatite separated from one sample, were also analyzed. As preliminary results pointed to the importance of spinel-related micro-phases, a complex leaching procedure involving different acids was applied to three separates of this mineral. Furthemore, spinel surfaces were investigated by SEM and electron microprobe to determine the composition of the attached micro-phases.
A striking feature of trace-element distribution in whole rocks is the existence of an overall correlation with textures. Deformed peridotites have high concentrations of Large Ion Lithophile Elements (LILE: Ba, Th, U, Sr and LREE), coupled with High-Field Strength Elements (HFSE: Nb, Ta, Zr, Hf and Ti) negative anomalies, and variable mineral compositions according to peridotite fertility. In contrast, granular peridotites are depleted or only slightly enriched in LILE, without significant anomalies of HFSE. Moreover, they are characterized by relatively constant mineral compositions in various rock types, in spite of a wide range of modal compositions (lherzolites to olivine-rich harzburgites).
In spite of their highly variable trace-element signatures, the different rock types show similar trace-element distribution between peridotite constituents. Clinopyroxene display systematic, though variable, depletion of HFSE relative to REE, both in HFSE-depleted and undepleted peridotites. Our results indicate that this mineral is not the dominant host for the most incompatible elements. Olivine and orthopyroxene have enhanced HFSE abundances relative to REE, which, nevertheless, do not balance clinopyroxene depletion. Mass balance calculations show that in the apatite-free peridotites, the silicate minerals account for total HREE abundance in whole rocks and 60-90% of LREE, Sr and Zr-Hf. However, their contribution to whole-rock budget for the most incompatible elements is much less important, these elements being mainly hosted in other peridotite constituents. In particular, micro-inclusions in silicates play a significant role for Rb (20-25%), and to a lesser degree for Ba, Th and U. When present, apatite, largely predominates the budget of Th, U, Sr and LREE (40-85%). Leaching experiments also revealed the importance of a pervasive grain-boundary component, which contributes 20-40% of the whole rock budget for Ba, Th, U and Sr, and 10-25% for Nb and LREE, in apatite-free samples. Unexpectedly, our results point to the importance of spinel as a major repository for highly incompatible elements, such as Nb, Ba, Rb and Ba. "Spinel-free" peridotite compositions reconstructed mineral separates, match the trace-element abundances of whole rocks, except for systematic deficiencies of Nb-Ta (50-90%), Rb-Ba (50-80%), and to a lesser degree Zr-Hf (¾ 25%). This is due to the existence of thin reactional layers (< 10 µm thick) coating the surfaces of spinel grains and concentrating these elements. A detailed study of spinel surfaces by SEM and electron microprobe, has revealed that they are mainly composed of Ti-oxides and phlogopite hardly detectable by optical methods. In spite of their extremely low abundance (e.g. ¾ 0.015% Nb-rutile analyzed by electron microprobe), Ti-oxides are the predominant repository of Nb-Ta in the xenoliths.
Our results indicate that all studied peridotites were affected by interactions with mantle melts. Correlation between mineralogical/textural variations and trace-element signatures of whole rocks and minerals leads to the distinction of two contrasting types of metasomatism. As confirmed by porous-flow modelling, the composition of deformed xenoliths may be explained by interaction with LILE-enriched melts at low melt/rock ratios. In contrast, the granular xenoliths have been extensively re-equilibrated with basaltic melts at higher melt/rock ratios. Our data and recent findings in orogenic peridotites suggest that these two processes were coeval and associated with pervasive infiltration of asthenospheric magmas in the lower lithosphere, above a mantle plume. Small melt fractions enriched in volatiles would evolve from infiltrated basalts, upon reaction with lithospheric peridotites at decreasing melt volume (granular peridotites). Because of their low viscosity and solification temperature, such melts may migrate upwards in low-porosity peridotites (deformed xenoliths). Upon thinning of the lithospheric mantle, upward migration of the percolation front would be associated with repeatedly superimposed crystallization of small melt fractions, and further melt infiltration. This mechanism can produce extreme LILE enrichment in melts, while the HFSE remain depleted due to the crystallization of very small fractions of Nb-rutile after spinel. This process may account for the origin of the LILE-enriched peridotites commonly ascribed to carbonate-melt metasomatism.