Rhenium volatility in subaerial lavas: constraints from subaerial and submarine portions of the HSDP-2 Mauna Kea drillcore

High Re abundances in mid-ocean ridge basalts (MORB) relative to primitive mantle (ave. MORB[Re] approximate to 930 ppt; PM[Re] approximate to 260 ppt) indicate that Re behaves as an incompatible element during MORB generation. However, contrary to expectations for an incompatible element, average Re abundances in subaerial ocean island and arc lavas ( similar to 330 ppt and similar to 190 ppt, respectively) are much lower than in MORB. Previous studies have argued that the low Re abundances in ocean island and arc lavas reflect greater Re compatibility during melt generation, caused by higher modal abundances of garnet and sulfide. However, available partitioning data for Re suggest that reasonable variations in the modal abundances of these phases cannot fully account for the observed differences in Re abundance in different tectonic settings. Higher modal garnet in the sources of ocean island and arc lavas cannot produce the observed low Re/Yb ratios (relative to MORB), because available data suggest that Re is less compatible in garnet than Yb. High Sulfide abundances can in principle produce the observed Re depletions, but would require extremely high sulfur contents in the sources of ocean island and arc lavas ( > 1000 ppm). Large differences in modal sulfide abundance in different tectonic settings are also contradicted by the similarity of MgO-Os trends in ocean island, arc, and mid-ocean ridge samples. An alternative explanation for the low Re abundances in ocean island and arc settings is that much of the Re initially contained in these (largely subaerial) lavas escaped as volatile Re-oxide or Re-chloride species during magma degassing prior to or during subaerial eruption. The similar to 3 km deep HSDP-2 Mauna Kea drillcore on Hawaii provides a unique opportunity to examine the effects of magma degassing on Re abundances. The upper similar to 800 m of the Mauna Kea portion of the drillcore is composed of subaerial lavas, whereas the lower 2 km is submarine. Os-isotopes are nearly constant throughout the core (Os-187/Os-188=0.128-0.130), so large variations in source composition (e.g, Re content) over the period sampled by the core are unlikely. Rhenium abundances in the subaerial lavas are consistently low (ave. Re approximate to 180 ppt), In contrast, Re abundances in the submarine lavas increase systematically with increasing depth, ranging from an average of similar to 300 ppt in samples emplaced at < 1000 m below sea level (mbsl) up to similar to 755 ppt in lavas from > 1000 mbsl. Rhenium/ytterbium ratios also increase with core depth, ranging to values significantly exceeding those observed in MORB. These trends are best explained by progressive Re loss in the subaerial and shallow submarine lavas during melt degassing. Assuming that the deepest submarine lavas are unaffected by Re loss during degassing, the subaerial Mauna Kea lavas appear to have lost on average similar to 80% of their initial Re. The systematic differences in Re abundance in ocean island, arc, and mid-ocean ridge lavas may therefore reflect the fact that most analyzed ocean island and arc samples are subaerial and therefore degassed, whereas MORB retain a greater fraction of their volatile inventory. (C) 2003 Published by Elsevier B.V.