Constraining the thickness of the crystal mush in layered mafic intrusions

When basaltic magma stalls in the crust, cooling leads to growth of a marginal mushy layer. The thickness of this crystal mush can be constrained using the step-changes in textural maturity (quantified by the median augite(cpx)-plag-plag dihedral angle, Θcpp) caused by the change in fractional latent heat accompanying the arrival of a new liquidus phase. At the instant of its saturation in the bulk magma, the top of the mush is marked by the first appearance of the new primocryst phase. At this moment, the high porosity upper zone of the mush comprises poorly consolidated material, with no cpx-plag-plag junctions: at deeper levels porosity decreases by primocryst overgrowth, growth of interstitial augite, and compaction. In the context of dihedral angle populations, the mush zone can be divided into 3: the upper zone, where melt is adjacent to all plag-plag junctions; the middle zone where augite fills some of the pore corners; and the lower zone where augite fills all pore corners. For our purposes, the base of this lower zone corresponds to the point at which diffusive change of grain boundary orientations has effectively ceased. The change in fractional latent heat accompanying the addition to the liquidus assemblage will be fully recorded within the upper zone since all cpx-plag-plag junctions are created after the change and its consequent decrease in the contribution of sensible heat to the total enthalpy loss: Θcpp will be high. The middle zone will record a mixture of the new and old thermal regime: junctions which were melt-filled at the moment of arrival of the new phase will have less opportunity to increase the cpx-plag-plag angle, while those which were already filled by augite will have higher angles. Θcpp will therefore range from the new higher value at the top of middle zone, to some lower value at the base. Cumulates in the lower zone had no melt-filled junctions: Θcpp will increase from a low value corresponding to the previous value of fractional latent heat recorded at the base of the lower zone to meet the lowest angles recorded at the base of the middle zone. The width and shape of the step-change in Θcpp thus tells us about mush thickness and structure. The arrival of cumulus apatite in the Skaergaard Layered Series coincides with the associated step-change in Θcpp in the centre of the intrusion, constraining mush thickness to be of the order of a few metres. However, the first appearance of cumulus apatite in floor cumulates adjacent to the intrusion walls is offset by 150m from the associated step-change in Θcpp, consistent with a mush some 150m thick. These differences are associated with significant variations in trapped liquid content calculated from bulk rock P2O5 concentration: the cumulates forming in the centre of the intrusion floor contain ~ 5 vol.% trapped liquid, while those near the walls contain ~15 vol.% trapped liquid. Thick mushes in the Skaergaard chamber thus are more orthocumulate in nature than the thin mush. These differences probably reflect variable contributions of eroded material from the vertical chamber walls, compounded by less efficient compaction in the more rapidly cooled marginal regions.