Mantle sources and magma evolution in Europe's largest rare earth element belt (Gardar Province, SW Greenland): new insights from sulfur isotopes

William Hutchison, Adrian A. Finch, Anouk M. Borst, Michael A.W. Marks, Brian G.J. Upton, Aubrey L. Zerkle, Eva E. Stüeken, Adrian J. Boyce

Research output: Contribution to journalArticlepeer-review

18 Citations (Scopus)
11 Downloads (Pure)


Alkaline igneous complexes are often rich in rare earth elements (REE) and other metals essential for modern technologies. Although a variety of magmatic and hydrothermal processes explain the occurrence of individual deposits, one common feature identified in almost all studies, is a REE-enriched parental melt sourced from the lithospheric mantle. Fundamental questions remain about the origin and importance of the mantle source in the genesis of REE-rich magmas. In particular, it is often unclear whether localized enrichments within an alkaline province reflect heterogeneity in the mantle source lithology (caused by prior subduction or plume activity) or variations in the degree of partial melting and differentiation of a largely homogeneous source. Sulfur isotopes offer a means of testing these hypotheses because they are unaffected by high temperature partial melting processes and can fingerprint different mantle sources. Although one must be careful to rule out subsequent isotope fractionation during magma ascent, degassing and crustal interactions.

Here, we present new S concentration and isotope (δ34S) measurements, as well as a compilation of major and trace element data, for a suite of alkaline magmatic units and crustal lithologies from the Mesoproterozoic Gardar Province. Samples span all phases of Gardar magmatism (1330–1140 Ma) and include regional dykes, rift lavas and the alkaline complexes Motzfeldt and Ilímaussaq, which represent two of Europe's largest REE deposits. We show that the vast majority of our 115 samples have S contents >100 ppm and δ34S of −1 to 5‰. Only 8 samples (with low S contents, <100 ppm) show evidence for crustal interactions, implying that the vast majority of Gardar melts preserve the S isotopic signature of their magma source. Importantly, samples from across the Gardar Province δ34S have above the canonical mantle range (≤−1.4‰) and therefore require recycled surface S in their mantle source. Elevated values are explained by a period of Andean-style subduction and mantle metasomatism which took place ∼500 Ma before rift onset and are also supported by trace elements signatures (e.g. Ba/La) which match modern subduction zones.

Comparing the various generations of Gardar magmas, we find that δ34S  values, large ion lithophile elements (K, Ba, P) and selective incompatible elements (Nb and light REE) are particularly enriched in the Late Gardar dykes, alkaline complexes and clusters of silica-undersaturated dykes spatially associated with the alkaline complexes. These data indicate that subduction-related metasomatism of the Gardar mantle was spatially heterogeneous, and that alkaline complexes are sourced from localized mantle domains highly enriched in 34S, REE, alkalis and volatiles (particularly, F). Since alkalis and volatiles play an essential role in driving extreme differentiation of alkaline melts and fluids, we suggest the co-location of these species plus incompatible metals at high concentrations in the lithospheric mantle is a critical first-step in the genesis of a world-class alkaline REE deposit. S isotopes are powerful tools for identifying enriched mantle domains and the sources of mineralized alkaline igneous bodies.
Original languageEnglish
Article number117034
Number of pages16
JournalEarth and Planetary Science Letters
Early online date17 Jun 2021
Publication statusPublished - 15 Aug 2021


  • Rift
  • Magmatism
  • Sulfur
  • Geochemistry
  • REE
  • Volatiles


Dive into the research topics of 'Mantle sources and magma evolution in Europe's largest rare earth element belt (Gardar Province, SW Greenland): new insights from sulfur isotopes'. Together they form a unique fingerprint.

Cite this