Thin horizons of sulfide minerals enriched in platinum group elements (PGEs), known as PGE “reefs,” are a feature of many layered mafic intrusions, including the Sonju Lake intrusion (Midcontinent rift, Minnesota). In traditional petrologic models, PGE reefs are thought to form during upward accumulation of mineral deposits, either by downward settling of immiscible sulfides through a silicate melt (the orthomagmatic model) or by concentration during upward flow and deposition in a fluid (the hydromagmatic model). While each model explains some of the observations from reefs, neither provides a consistent explanation for both the overall PGE distribution and evidence for offsets in peak concentrations of PGEs and base metals.
If, alternatively, layered intrusions form by a top-down process of sill injection and reaction (thermal migration zone refining) as postulated in Lundstrom et al. (2011a), then PGE reefs could form as a moving sulfide band passes downward through a mineral-melt mush at the particular temperature of sulfide saturation. A simple temperature gradient experiment in a sealed quartz tube illustrates how sulfide moves and aligns parallel to a temperature gradient. Thus, while silicates form in a mush zone and remain immobile, the immiscible sulfide liquid position is controlled by the temperature of sulfide saturation and thus moves down relative to the silicates as new sills underplate previous ones; effectively, all PGE derived from the upper portion of the intrusion (above the reef) partition into this layer resulting in the enriched reef layer, consistent with observed distribution and mass balance in the Sonju Lake intrusion. The higher sulfide-silicate mineral partition coefficients for the PGE relative to Cu as sulfides move relative to the silicates results in Cu being chromatographically separated from PGEs during the process. Thus, the mechanism can account for the two major observations, the overall distribution of PGEs in layered intrusions and the chromatographic separation of PGEs from Cu, perhaps better than existing models. The model provides a sharp alternative to current models, with its predictive aspects potentially improving future exploration for PGE deposits.