The Low-δ¹⁸O Late-Stage Ferrodiorite Magmas in the Skaergaard Intrusion: Result of Liquid Immiscibility, Thermal Metamorphism, or Meteoric Water Incorporation into Magma?

Abstract

We report new laser fluorination oxygen isotope analyses of selected samples throughout the Skaergaard intrusion in East Greenland, particularly relying on ∼1-mg separates of the refractory, alteration-resistant minerals zircon, sphene, olivine, and ferroamphibole. We also reexamine published oxygen isotope data on bulk mineral separates of plagioclase and clinopyroxene. Our results show that the latest-stage, strongly differentiated magmas represented by ∼3 to 6 km³ of ferrodiorites around the Sandwich Horizon (SH), where the upper and lower solidification fronts met, became depleted in ¹⁸O by about 1.5‰–2‰ relative to the original Skaergaard magma and the normal mantle-derived mid-ocean ridge basalt. Earlier studies did not recognize these low-δ¹⁸O ferrodiorite magmas (δ18O = ∼ 3‰–4‰) because after the intrusion solidified, much of the intrusion and its overlying roof rocks were heavily overprinted by low-δ¹⁸O meteoric-hydrothermal fluids. We consider three possible ways of producing these low-δ¹⁸O ferrodiorite magmas. (1) At isotopic equilibrium, liquid immiscibility may cause separation of a higher-δ¹⁸O, higher-SiO₂ granophyric melt, thereby depleting the residual Fe-rich ferrodiorite magma in 18O. However, such a model would require removal of many cubic kilometers of coeval granophyre, a greater proportion than is observed anywhere in the intrusion; there is no evidence that any such magmas erupted to the surface and were eroded. (2) While direct migration of low-δ¹⁸O water seems implausible, we consider a model of “self-fertilization,” whereby oxygen from meteoric waters entered the SH magma by devolatilization and exchange with hydrated, low-δ¹⁸O stoped blocks of the upper border series. Such reactive exchange between residual melt and adjacent hydrothermally altered, water-saturated rocks contributed low-δ¹⁸O crystalline components and low-δ¹⁸O pore water to the residual melt. The low-δ¹⁸O zircon and sphene may have crystallized directly from this contaminated low-δ¹⁸O melt, even though the entire mineral assemblage did not, simplifying the mass balance problem. (3) Finally, after SH crystallization, fracturing, and subsequent subsolidus meteoric-hydrothermal alteration depleted these rocks in ¹⁸O, intrusion of the 660-m-thick Basistoppen sill, emplaced 150–200 m above the still hot SH, may have reheated and partially melted these late-stage differentiates. In this scenario, zircon and sphene could have crystallized from a low-δ¹⁸O partial melt, while other minerals may have simply reequilibrated. We favor models 2 and 3 and discuss their strengths and weaknesses.