Basement Sill, Antarctica: Constraints from its PGE Abundance Patterns and Isotopic Compositions on Magma Source Characteristics and Crystallization Processes

The Basement Sill in the Dry Valleys of southern McMurdo Sound and the Dufek layered mafic intrusion in the Pensacola Mts. are part of the Ferrar large igneous province (FLIP), emplaced in a Paleozoic mobile belt developed adjacent to the East Antarctic Craton during the Mid Jurassic. Zircon, baddeleyite and rutile U-Pb dating we performed prior to this award shows that emplacement occurred in a short amount of time at 184 ± 1 Ma, contemporaneously with fragmentation of the supercontinent Gondwanaland. We therefore considered understanding the production of copious amounts of magma in a short time to be fundamental to understanding LIPs in general. Mantle plumes provide an attractive mechanism for generating short-duration, voluminous magmas in LIPs while at the same time providing a mechanism for the often concurrent break-up of supercontinents. For the FLIP, however, we have generated PGE and isotopic data that challenge the plume interpretation.

A number of studies have demonstrated that Pt and Pd behave incompatibly during partial melting of the mantle. These two elements particularly have an affinity for S-undersaturated melts, whereas the Ir, Ru, and Rh behave compatibly, possibly ending up in spinel or some PGE alloys. A survey of PGE abundance patterns in several tectonic settings shows that basalts from plume activity exhibit one or two order of magnitude depletions in Ir and Ru compared to Pt and Pd. Mid-ocean ridge basalts (MORB) exhibit abundance patterns with only a slight positive slope and no dramatic depletions or enrichments for any of the PGEs. This positive slope probably reflects preference of Pt and Pd for silicate liquids during the melting of the asthenosphere. Komatiite lavas representing large volume melting of the asthenosphere show that this region of the mantle has flat PGE abundance patterns. The mantle lithosphere, often invoked in some models as the source of FMP magmas, exhibits a wide range of PGE abundance patterns commensurate with the degree of prior melt extractions. Significant Pt and Pd depletions will be evident in the more refractory mantle lithosphere; there will be less Pt and Pd depletion in lithospheric mantle domains that have not undergone severe partial melting. Lithospheric mantle materials with a history of being in the suprasubduction environment have a pattern that deviates from the records of melt extraction described above. One study of arc peridotite xenoliths showed significant Os enrichment, interpreted to be due to the partitioning of this element into oxidized and chlorine-rich slab-derived fluids that go on to fertilize the lithospheric portion of the mantle wedge, leaving the rest of the mantle column depleted in this element.

We have analyzed a total of 37 samples for their PGE abundance patterns, the large majority of them in duplicates to assess reproducibility. Most of the samples from the Basement Sill and Dais Intrusion lobe (or opx tongue) display coherent patterns, and most striking about the opx tongue - here represented by the Dais Intrusion - are the remarkable depletions in Os and Ir relative to primitive mantle. Os/Ir ratios are in the range of 0.1 to 0.3 for these samples compared to the more commonly observed values close to 1. The coherent data for the Dais Intrusion agrees very well with the data from other internal layers of the Basement Sill. Extreme depletions in Os and Ir compared to the Ru, Pt and Pd abundances are atypical of plume-derived magmas and are more consistent with the alternative view that FLIP resulted from the decompression of a fossil subduction zone along the Proto-Pacific margin of Gondwanaland, disaggregated by the rifting related to plate rearrangements during supercontinent break up.

Close to the upper contact of the Basement Sill, however, we observe wide differences and great complexity between samples in both concentration and abundance patterns, which seem to be related to fluid flow associated with hydrothermal cells established in the contact zones. This interpretation is supported by Alan Boudreau's recent studies of fluid inclusions in orthopyroxene and plagioclase and halogens in apatite from samples similar to ours. Boudreau's studies of Cl and F in interstitial apatite have shown that Cl/(Cl+F) is constant or increases modestly with height until dropping off in the upper parts of the intrusion. This drop-off parallels a decrease in bulk MgO and the discombobulation of PGE abundance patterns, no doubt because of selective element migrations. The mostly likely cause of the low Cl in the upper parts of the intrusion is loss of a Cl-bearing fluid, as there is no mineral phase that would prefer Cl over F. If so, this process has implications for the transport of S and PGE's. Therefore PGE concentrations near the roof zone of the Dais Intrusion, in the follow-up study to this one, are being evaluated for consistence with the volatile saturation model.

Could some of the details we see in the abundance patterns be attributed to fractionation effects or other secondary processes? For the Dais Intrusion, we do not observe any variation in Pd/Ir ratios with MgO which would otherwise implicate fractionation effects on the PGE ratios. However, we observe variation in the rocks from Bull Pass. Curiously, it is the rocks with the most similar MgO compositions that show the greatest variation in Pd/Ir. As a working model a small amount of upper crustal contamination might be able to explain the observations at Bull Pass, which may not be terribly surprising considering that these samples came from near the top of the Basement Sill.