Trace element variations in clinopyroxene and amphibole from alkaline to peralkaline syenites and granites: implications for mineral–melt trace-element partitioning

Abstract

Trace element compositions of clinopyroxene and amphibole from three rift-related alkaline to peralkaline igneous complexes (syenites and granites) in South Greenland reflect evolving magma composition as well as crystal-chemical control on partitioning. Clinopyroxenes and amphiboles evolve from Ca–Mg-dominated members via intermediate to Na–Fe 3+-dominated members. Most trace elements are highly enriched compared to primitive mantle values, consistent with the highly fractionated character of the host rocks. High field strength element (HFSE; Ti, Zr, Hf, Sn, Nb, Ta) abundances appear to be mainly controlled by the major element composition of the host crystal, which in turn determines the crystal site parameters. A crystal-chemical control is also indicated for the REE, since clinopyroxenes and amphiboles show continuous change from LREE-enriched patterns in the calcic minerals via wave-shaped pattern in the Ca–Na minerals towards a more pronounced HREE enrichment in the most Na-rich minerals. The low absolute abundances of large ion lithophile elements (Ba, Sr, Pb, Eu 2+) are interpreted to reflect both a crystal-chemical aversion to incorporate these elements and the effects of prolonged feldspar fractionation on the melt composition. Eu and Pb abundances are also affected by oxygen fugacity and crustal assimilation, respectively. The partitioning of most trace elements between clinopyroxene and co-genetic amphibole is independent of melt composition or major element composition of the crystals. Most incompatible trace elements (particularly Nb, Ta, U, Th, Rb, Ba and Li) show a slight preference for amphibole. Exceptions to this general trend occur in rocks affected by late-stage fluid circulation resulting in the redistribution of some mobile elements. Calculation of trace-element compositions of coexisting melts using published partition coefficients for alkaline systems shows that melt compositions similar to whole-rock compositions are obtained for clinopyroxenes with broadly augitic to diopsidic compositions. There is a considerable mismatch between calculated melt and whole-rock data for various elements, i.e., Zr, Hf and the HREE, when melt compositions are calculated from NaFe 3+-rich aegirine–augites and aegirines. Melt compositions from amphiboles also show only limited overlap with whole-rock data. The reasons for these differences may include the following: (1) The crystallization history of the rocks is very complex so that trace element partitioning cannot be expressed with a single partition coefficient. (2) Published mineral–melt partition coefficients cannot generally be applied due to compositional differences between the alkaline to peralkaline systems of this study and the previous studies. (3) Whole-rocks do not reflect melt compositions. Using a theoretical model of clinopyroxene–melt trace element partitioning based on the crystal chemistry alone, we show that absolute values of D REE decrease but the relative preference for HREE increases as the crystal becomes more aegirine-rich, which is in qualitative accordance with the observed REE patterns. Melts calculated from the theoretically determined partition coefficients show a good overlap with whole-rock data for relatively Fe 3+-poor clinopyroxene compositions. Melts calculated from aegirines do not agree with whole-rock compositions, suggesting that the theoretical model needs refinements for the previously not considered incorporation of the NaFe 3+Si 2O 6 component.