High-field-strength elements in carbonatitic rocks: Geochemistry, crystal chemistry and significance for constraining the sources of carbonatites

Carbonatites and related rocks exhibit a significant variation in their content of high-field-strength elements (HFSE), i.e. Ti, Nb, Ta, Zr and Hf. The average abundances of these elements (calculated for 119 localities worldwide) decrease from phoscorites and silicocarbonatites to calciocarbonatites to magnesiocarbonatites. The analytical data currently available for ferrocarbonatites are insufficient to establish if this trend persists through the entire carbonatitic series. In comparison with the primitive mantle, the average carbonatite has significantly higher Nb/Ta and Zr/Hf ratios, and lower Zr/Nb and Zr/Ta ratios (35, 60, 1 and 29, respectively). Alkali-ultramafic rocks associated with carbonatites exhibit, on average, higher Zr/Nb ratios and no (or little) depletion in heavy HFSE (Ta and Hf). Implications of these data for constraining the source(s) of intracratonic carbonatites are discussed using the Kola Alkaline Province (NW Russia and Finland) as an example. The bulk of incompatible HFSE (> 85%) is estimated to reside in Ti, Nb and Zr oxide minerals (in particular, perovskite, pyrochlore, ilmenite, baddeleyite and zirconolite) and zircon, whereas a significant proportion of Ti is contained in magnetite and ferromagnesian silicates. The lowest recorded Nb/Ta and Zr/Hf ratios in the primary HFSE oxides and titanite (but not zircon) are significantly below the primitive-mantle values, which makes these phases an effective vehicle of heavy-HFSE fractionation at crustal levels. The HFSE geochemistry of carbonatites en masse is not consistent with their derivation from a hypothetical carbonate-rich alkali-ultramafic (e.g., nephelinitic) magma by either liquid immiscibility or crystal fractionation. The bulk of carbonatites are interpreted to originate by low-degree partial melting of a metasomatized HFSE5+-enriched mantle source with elevated Nb/Ta and Zr/Hf ratios. The trace-element signature of carbonatites cannot be adequately explained by clinopyroxene-controlled decoupling of HFSE5+ and HFSE4+ during partial melting, and requires the presence of amphibole and, possibly, refractory phlogopite in their mantle source. The HFSE geochemistry, field relations and relative timing of Kola carbonatites and associated silicate rocks are consistent with their derivation by low-degree melting of a vertically heterogeneous amphibole-bearing mantle reservoir, followed by differentiation of the primitive magmas. The previously established volume, tectonic and geochemical constraints suggest that the metasomatic enrichment of the mantle may have resulted from reaction of garnet lherzolite with carbonate melts of asthenospheric provenance, which is in agreement with the published experimental and mantle-xenolith studies. There is no evidence for the presence of refractory rutile (i.e., subducted crust) in the postulated mantle source.