Geochemical variations in mid-ocean ridge basalts have been attributed to differing proportions of compositionally distinct mantle components in their sources, some of which may be recycled crust. Oxygen isotopes are strongly fractionated by near-surface interactions of rocks with the hydrosphere, and thus provide a tracer of near-surface materials that have been recycled into the mantle. We present here oxygen isotope analyses of basaltic glasses from the mid-Atlantic ridge south of and across the Azores platform. Variations in δ18O in these samples are subtle (range of 0.47‰) and may partly reflect shallow fractional crystallization; we present a method to correct for these effects. Relatively high fractionation-corrected δ18O in these samples is associated with geochemical indices of enrichment, including high La/Sm, Ce/Pb, and 87Sr/86Sr and low 143Nd/144Nd. Our results suggest two first-order conclusions about these enriched materials: (1) they are derived (directly or indirectly) from recycled upper oceanic crustal rocks and/or sediments; and (2) these materials are present in the north Atlantic MORB sources in abundances of less than 10% (average 2–5%). Modeling of variations of δ18O with other geochemical variables further indicates that the enriched component is not derived from incorporation of sediment or bulk altered oceanic crust, from metasomatism of the mantle by hydrous or carbonate-rich fluids, or from partial melting of subducted sediment. Instead, the data appear to require a model in which the enriched component is depleted mantle that has been metasomatized by small-degree partial melts of subducted, dehydrated, altered oceanic crust. The age of this partial melting is broadly constrained to ∼250 Ma. Reconstructed plate motions suggest that the enriched component in the north Atlantic mantle may have originated by subduction along the western margin of Pangea.
The magmatic record of the easternmost part of the Trans-Mexican Volcanic Belt elucidates how temporal changes in subduction parameters influence convergent margin volcanism. In the Palma Sola massif, three phases of magmatic rocks with distinct chemical characteristics were emplaced in a relatively short time span (∼17 Ma): Miocene calc-alkaline plutons, latest Miocene-Pleistocene alkaline plateau basalts, and Quaternary calc-alkaline cinder cones. Plutons have arc-like trace element patterns (Ba/Nb = 16–101), and their Sr, Nd, and Pb isotopic compositions become more “depleted” with increasing SiO2 contents. Their Pb isotopes are bracketed by the subducted sediments and Pacific mid-ocean ridge basalts (MORB), requiring the participation of an unradiogenic component that mixes with a sediment contribution. High Sr/Y and Gd/Yb ratios in the least radiogenic pluton might indicate a melt coming from the subducted MORB. Trace element patterns of the plateau basalts show moderate or negligible subduction contributions (Ba/Nb = 6–31). Rocks without subduction signatures are similar to ocean island basalts, indicating melting of an enriched mantle wedge. The plateau basalts form an array in 206Pb/204Pb-207Pb/204Pb space that trends toward the composition of the subducted sediment. The sediment component is also indicated by the inverse correlations between Pb isotopes and subduction signals. This component has high Th/Nd coupled with low 143Nd/144Nd, but lower Pb/Nd and Sr/Nd ratios than the bulk sediment. These suggest melting of a sediment that has lost fluid mobile elements prior to melting. The Quaternary cinder cones have moderate subduction signals (Ba/Nb = 16–41), and their isotopic compositions correlate with differentiation indices. Contamination with the local Paleozoic basement can explain the petrogenesis of the youngest rock suite. The geochemical differences among the suites indicate temporal modifications in the chemical characteristics of the slab input. These variations can be associated with modifications in the Pacific subduction regime. We suggest the Miocene magmatic phase was formed by an essentially flat subduction angle that favored melting of the subducted oceanic crust. Slab rollback in the Pliocene allowed melting of deeper portions of the wedge by the injection of dehydrated sediment melts. In the Quaternary, an even steeper subduction angle provided negligible slab contributions to the Palma Sola region, and upper crustal contamination largely controls the petrogenesis.
Trace element abundances in melt inclusions are commonly used to interpret melting and melt extraction processes. These interpretations, however, often assume that the chemical compositions of melt inclusions are identical to the liquid from which the host crystal grew, even though driving forces for postentrapment diffusion and modification are demonstrable. This paper begins to quantify the effects of diffusion on melt inclusions using a numerical model. The model calculates the compositional evolution of a spherical inclusion which initially is in equilibrium with a crystal host out to some distance rjump and out of equilibrium beyond. In particular we consider the end-member scenario, whereby the trapped melt is initially out of equlibrium with the neighboring crystal as this sets the minimum time for reequilibration. A package of numerical codes is provided that allows the user to explore other initial conditions. The model calculates the change in inclusion composition and also the structure of diffusion halos that grow around the inclusion as it reequilibrates with the surrounding crystal. The detection of these profiles in naturally occurring inclusions may allow the time since entrapment and the initial inclusion concentration to be estimated. The extent of reequilibration is most strongly influenced by the partition coefficient, diffusivity, and the inclusion radius. Fast-diffusing elements with high mineral/melt partition coefficients are modified rapidly, particularly in small inclusions. Because minerals have very different Dmineral/melt for the various elements, the effects of diffusive reequilibration differ substantially from one mineral to another. For example, the higher partition coefficients of the heavy rare earth elements (HREE) in olivine make HREE concentrations easier to modify than light rare earth elements (LREE) concentrations. In contrast, Sr, Eu, and Ba in plagioclase hosted inclusions equilibrate more rapidly than the other trace elements. Examination of published trace element concentrations of olivine hosted inclusions show little evidence for reequilibration, at least for the light REE and other highly incompatible elements. It is difficult, however, to provide firm constraints due to the uncertainties in olivine diffusivities and the initial condition. In contrast, trace element diffusivities in plagioclase have been determined experimentally [Cherniak, 2001], and the trace element concentrations of published plagioclase hosted inclusions show evidence for extensive diffusive exchange with the host in a manner consistent with model predictions. Postentrapment modification therefore is likely an important factor in the interpretation of some melt inclusion data.