IN SITU LA-ICP-MS TRACE ELEMENT DATA ON PEROVSKITE: IMPLICATIONS FOR ORIGINAL MELT COMPOSITIONS FROM KIMBERLITES IN KANSAS

This study presents new microprobe and LA-ICP-MS data for perovskites from Cretaceous-age kimberlites in Kansas in order to better understand their petrogenesis. Kimberlites are heterogeneous rocks that typically entrain abundant xenoliths during ascent. Primary phases are also commonly altered to secondary serpentine and calcite, with modal proportions of secondary minerals sometimes >90%. These characteristics make it difficult to determine the original kimberlite melt composition. This has led to an increasing interest in the geochemical and isotopic analysis of groundmass minerals as an alternative approach to characterizing these rocks. Perovskite is a minor primary phase (>10%) in some kimberlites that is also relatively resistant to serpentinization and weathering. It is therefore an alternative way to constrain kimberlite melt composition prior to alteration. Two different perovskite textural types have been identified in the Kansas kimberlites. In the Bala kimberlite, the perovskites occur as large, octahedral crystals, whereas in the Tuttle kimberlite, they occur as small, square shapes ringing serpentine pseudomorphs after olivine. At least three different trace element patterns are observed, although two predominate and neither correlates with textural type. One pattern shows high concentrations of all trace elements except Rb (i.e. > 10% primitive mantle), with negative anomalies for Rb, Ba, Sr and Zr; Th is also strongly enriched relative to U. The second pattern is similar to the first in having a strong depletion in Zr, but concentrations of most trace elements are lower. It is also more enriched in Rb and exhibits large positive Y, Yb and Lu anomalies, and has a low ratio of Th to U. Both patterns are observed in the Bala kimberlite, but only the second pattern is observed in Tuttle. The results suggest that the perovskites from Bala crystallized from a more evolved melt composition and that this kimberlite underwent a more complicated evolution, possibly involving multiple injections of melt within the same conduit. Trace-element modelling of melts in equilibrium with the different perovskite types will be presented and compared with published data for bulk rock kimberlite compositions to evaluate this approach to estimating original kimberlite melt compositions.