Subducted carbonates not required: Deep mantle melting explains stable Ca isotopes in kimberlite magmas

Abstract

Ca isotope geochemistry has great potential for improving our understanding of magmatic systems and for tracing the deep Earth carbon cycle. There are still many open questions, however, regarding the proper application of this relatively novel proxy to the study of mantle-derived magmas, including (i) the possible effects of pressure on mineral-melt Ca isotope fractionation factors, and (ii) the potential for Ca isotopes to be used as tracers of recycled marine carbonates in mantle-derived magmas. Kimberlites are mantle-derived melts that are highly enriched in CO2 and are the deepest-sourced magmas (>200 km depth) known to erupt at Earth’s surface, providing an excellent opportunity to explore these questions. We present Ca isotope data combined with detailed petrographic observations, bulk-carbonate C-O isotope data, and bulk-rock major element analyses, for a suite of 23 well-characterized kimberlite samples from their type-locality (Kimberley, South Africa). These kimberlites have abundant previous evidence for recycled surface materials in their mantle source, including low S isotope and moderately radiogenic Sr isotope compositions, yet display only limited variations in their Ca isotope compositions (δ44CaBSE of −0.08‰ to −0.27‰), with an average of −0.17 ± 0.02‰ (2SE, n = 21). This composition is indistinguishable from average carbonatites [−0.19 ± 0.03‰ (2SE, n = 106)] and OIB from recent studies [−0.16 ± 0.01‰ (2SE, n = 41)], and slightly lower than average MORB [−0.11 ± 0.02‰ (2SE, n = 31)]. Although our samples display a wide range of emplacement styles, alteration conditions, extents of magmatic differentiation, and degrees of mantle-cargo entrainment (i.e., xenocryst accumulation), we find no correlations between Ca isotopes and any of these factors. Instead, we find that low-degree partial melting of the likely kimberlite source lithology (i.e., carbon-bearing garnet lherzolite) yields modelled melt δ44CaBSE values ranging between −0.12‰ and −0.16‰ (at 1400–1500 °C), in agreement with the measured Ca isotope compositions of the Kimberley kimberlites. This observation, and the lack of heavy carbon isotope signatures in the examined samples, indicates that kimberlites do not require subducted carbonates in their mantle sources, despite their very high CO2 contents. Although several recent studies have suggested that equilibrium mineral-melt Ca isotope fractionation factors (e.g., 1000lnαgrt-melt) could be significantly different at higher pressures (i.e., due to pressure-induced changes in CaO bond lengths and coordinations), our models successfully reproduce the kimberlite data using pressure-independent predictions for mineral-melt fractionations. It remains possible, however, that differences in isotopic fractionation due to the peculiar composition of kimberlite melts (e.g., high CO2, low SiO2) are effectively cancelled out by competing pressure effects, and future work independently targeting these factors will be especially important for our understanding of Ca isotope fractionation in mantle-derived melts and the Ca isotope systematics of Earth’s mantle.