Partitioning of chalcophile and highly siderophile elements (HSEs) between sulfide and carbonated melts – Implications for HSE systematics of kimberlites, carbonatites, and melt metasomatized mantle domains

Abstract

Highly Siderophile Elements (HSEs; Os, Ru, Ir, Rh, Pt, Pd, Au and Re) combined with their isotopic systematics (Re-Os and Pt-Os) are powerful tools for tracking evolution and genesis of mantle derived magmas. Given sulfides (accessory sulfide minerals and/or molten sulfides) are the primary hosts of HSEs in the mantle and low-degree carbonated melts are extracted from large portions of mantle volume, partitioning of HSEs between sulfide and carbonated melt might play a critical role in distributing HSEs between the mantle and crustal reservoirs. Although, partitioning of HSEs and chalcophile elements between sulfide melt and silicate melt has been previously studied, partitioning of these elements between sulfide melt and carbonated melts has not received much attention. Here we use high P-T experiments to determine the partitioning of HSEs and chalcophile elements (Ni, Co, Mo, Os, Ru, Pd, Pt and Re) between (i) sulfide melt and carbonated silicate melt (CO2 ~ 17 wt.%) and (ii) sulfide melt and carbonatitic melt (CO2 ~ >30 wt.%) at a pressure (P) of 3 GPa and temperatures (T) of 1300–1600 °C in graphite capsules. All experiments produced quenched Fe-sulfide melt blobs + carbonated silicate melt matrix. Concentrations of major elements were measured using electron microprobe, and HSEs and chalcophile elements were measured using LA-ICP-MS. We find that all the elements measured are compatible in the sulfide melt to varying degrees and their Dsulfide/carb. melt sequence is Mo < Co < Ni < Re < Pt ≤ Pd < Ru ≤ Os varying from around 10 for Mo to 105 for Os. Comparing the Dsulfide-carb. melt with Dsulfide-silicate from previous studies, we show that the partition coefficients of HSEs between sulfide and carbonated melts are lower than the partition coefficients of these elements between sulfide and silicate melts, indicating greater mobilization of these elements in carbonatites and carbonated silicate melts. Calculating bulk D (D¯) for carbonated peridotite using our experimentally measured D values, we model the HSE contents of mantle derived low-degree partial melts using an aggregate fractional melting equation and compare the primitive mantle normalized HSE patterns of our model with natural kimberlites, carbonatites, ocean island basalts, and alkaline basalts. We also calculate proportions of sub-lithospheric continental mantle (SCLM) xenolith detritus in the natural kimberlite and carbonatite samples from Karelian, Kaapvaal, Canadian shield and North China craton by using mass balance calculations based on Ru concentration in the primary carbonated melt and the SCLM xenoliths. Our calculations show that detritus proportion in natural kimberlites are 2–28% for Karelian, 7–28% for Kaapvaal, and 6–16% for Canadian shield, which are in agreement with previous studies using various other proxies. We also show that the extent of Re/Os fractionation is less for events of carbonate melt metasomatism as compared to similar events of basaltic melt metasomatism.