Abstract

Most published studies of olivine-hosted melt inclusions from mid-ocean ridges have been based on a single sample. Here we present a comprehensive melt inclusion study of major and trace elements from a single ocean ridge segment, the FAMOUS segment of the Mid-Atlantic Ridge. The melt inclusion dataset includes 312 olivine-hosted (Mg-number 85^92) melt inclusions from 14 samples distributed along the segment.This permits a more comprehensive assessment of the variability within melt inclusions from a single region, and of the relationship between melt inclusion and lava compositions. One recent question has been the extent to which melt inclusions truly preserve the original melt compositions, or instead are modified by late-stage processes occurring at shallow levels. In the FAMOUS inclusions, major elements have been affected by post-entrapment processes, but trace elements show no evidence of such processes, suggesting that diffusion coefficients for incompatible elements are small. Melt inclusions can be divided into three groups. (1) High-Mg inclusions are the most primitive and may potentially constrain the composition of the parental magmas that contribute to other melt inclusion and lava compositions. Although their trace element contents range from highly depleted to almost as enriched as the FAMOUS segment lavas, they are on average more depleted and the melts appear to be derived by greater extents of melting than the lavas. (2) Low-Al inclusions occur in the lower Mg-number olivines, and their major and trace element characteristics reflect mixing between high-Mg melt inclusion and lava compositions. (3) High-Al melt inclusions display Al2O3 contents as high as 18·4 wt %, SiO2 as low as 46·6 wt %, a strong depletion in the most incompatible elements and distinctively low middle to heavy rare earth element (MREE/HREE) ratios.The high Al2O3 and low SiO2 contents, as well as positive Sr anomalies in some of the high-Al melt inclusions, are best explained by assimilation of plagioclase-bearing cumulates. The trace element variability in the high-Mg melt inclusions is not consistent with a simple continuous melting column and requires pooling of near-fractional melts within the melting regime and a variable mantle source composition. Because the mean composition of these melt inclusions reflects greater extents of melting than the lavas, we propose that the melt inclusions come from the upper portions of the melting regime. Lavas, in contrast, sample the entire melting regime, including low-degree melts from the wings of the regime that are transported more directly to the surface along high-porosity channels. The high-Al, trace element ultra-depleted, low MREE/HREE melt inclusions derive from melting of a residual mantle source formed by previous melt extraction in the garnet stability field.There is a marked lack of correspondence between major and trace element variations in the melt inclusions. This may reflect a combination of processes, such as cumulate assimilation and re-equilibration of the magmas during ascent, which can reset major elements while having little effect on the trace element variations. The melt inclusions are not simply representative unpooled melts from the melting regime and they do not fully reflect the range of melt compositions contributing to the lavas.Their compositions reflect source heterogeneity as well as melting processes, and major and trace element indicators of depth of origin do not correspond. Combined comprehensive studies of lavas and melt inclusions have much more to reveal than studies based on either data source alone.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call