Postcumulus processes recorded in whole-rock geochemistry: a case study from the Mirabela layered intrusion, Brazil

According to the “trapped liquid” paradigm, whole-rock chemical compositions of cumulate rocks in layered intrusions can be represented as closed-system multicomponent mixtures of the cumulus phases and the liquids from which they crystallised. Alternative open-system models for cumulate solidification assert that intercumulus liquid is continuously mobile as a result of compaction of crystal mushes. In the closed-system models, all excluded elements, i.e. those incompatible in the cumulus phases, should be correlated with one another, whereas in open systems the more incompatible elements should be decoupled from the more compatible ones and correlations should be poor. These alternative hypotheses are tested using a database of more than 63,000 whole-rock analyses of mostly ultramafic cumulates from a single package of layers across the entire width of the Mirabela layered intrusion, a 2.5 km wide by >2 km thick intrusion with Great Dyke-like stratigraphy. Variably compatible elements Al, Na, Ti and P are found to show strong correlations with one other in olivine-pyroxene cumulates across the spectrum from mesocumulate to orthocumulate rocks, where drill-core sample intervals are composited over 3m; weaker correlations are found over 1m sample intervals. The closed-system trapped liquid model is robust for this suite of rocks at the scale of 3m lengths of drill core (approx. 1 kg of sample). Furthermore, a very smooth and symmetrical variation is observed from dominantly ad- to mesocumulate rocks in the geometrical centre of the intrusion towards orthocumulates at both margins, attributable to increase in cooling rate towards the margins. The most adcumulate rocks are gabbronorites at the base of the mafic sequence in the centre of the intrusion. There is no corresponding decrease in layer thickness towards the centre that would be predicted by a compaction model. The only incompatible element showing partial decoupling is K, which appears to have been mobile at a very late stage, probably in an aqueous fluid phase. There is no evidence for mobilisation of PGEs or other chalcophile elements by this fluid despite the presence of disseminated sulfide throughout the sampled interval.