Christian Tegner, University of Aarhus (Denmark)
Peter Thy, University of California (United States)
Marian B. Holness, University of Cambridge (United Kingdom)
Jakob K. Jakobsen, Nordic Volcanological Centre (Iceland)
Charles E. Lesher, University of California (United States)
Crystal mush and igneous differentiation processes are constrained from bulk compositions, densities and mineral modes of 131 cumulate samples in a new reference profile of the Layered Series in the Skaergaard intrusion, East Greenland. Using excluded elements P and U of cumulates and residual magma, modeled by Rayleigh fractionation, the trapped liquid contents can be determined. The trapped liquid contents decrease upwards from ∼47% at the base of LZa troctolites to 4-12% at the top of LZb olivine-gabbros and remains low (1-13%; 5.7% on average) in the oxide-gabbros of LZc, MZ and UZ. Local variations in trapped liquid content are associated with modal layering with leucocratic, low-density rocks having higher proportions of trapped melt than adjacent melanocratic, high-density rocks.
These observations are explained by the relative rates of crystal mush compaction and solidification. The local variations in the trapped liquid contents between leucocratic and melanocratic layers occur over few to tens of metres of stratigraphy and are interpreted as typical depths of the compaction zone. We show that theoretical length- and time-scales of compaction are similar to observed compaction depths and rates of crystal mush deposition for effective viscosities of the solid matrix of 1014-1016 Pa.s. The up-section decrease in trapped liquid contents from LZa to LZb is explained by transient rapid heat loss through the intrusion floor, leading to relatively high freezing rates, consistent with the orthocumulate texture and low textural maturity of LZa and the Hidden Zone. The disappearance of this transient rapid heat loss approximately at the level of oxide-in, and efficient compaction resulting in low trapped liquid contents above this level, is explained by the onset of Fe-Ti oxide crystallization. Not only is the density difference between crystal matrix and melt higher, helping to drive compaction, but the increase in textural maturity at the level of oxide-in shows the specific cooling rate at the solidification front slowed down in consequence of increased latent heat of crystallization, leaving more time for compaction to be efficient.
In contrast to earlier conclusions for Skaergaard and other layered intrusions, compaction of the mush on the chamber floor was effective over depths of only a few metres to tens of metres. This evidence, together with stepwise increases in textural maturity and the magnitude of slumped layering associated with the impact of blocks falling from the roof, supports a hard-ground crystallization front. Crystal mush compaction was an efficient means of differentiation in the Skaergaard intrusion.