The presence of uranium-series disequilibria in both mid-ocean ridge basalts (MORB) and ocean island basalts (OIB) has provided new insights into the melt generation and magma transport processes occurring beneath volcanic centers. By looking at the isotopes in the U-series decay chains, I attempt to better constrain how mantle rocks melt to produce basaltic magmas, through a series of experimental and analytical approaches.
230Th is a short-lived (half-life of 75,000 years) daughter nuclide from the decay of 238U, a very long-lived isotope (half-life of 4.5 billion years). Unless perturbed by a chemical reaction such as melting, these isotopes exist in a state of secular equilibrium where their activities are the same. (The activity of an isotope is its abundance multiplied by its decay constant, which is inversely proportional to its half-life.) Most MORB and OIB have isotopic U-Th signatures such that the ratio of their activities, (230Th/238U) > 1, indicating that there has been a recent perturbation (the melting event), and that this occurred in the presence of the mineral garnet (or possibly clinopyroxene with small M2 site radii). While a number of upper mantle rocks can contain Gt, the stability of Gt in peridotite, the most important upper mantle lithology, depends on pressure: Gt only occurs in peridotites deeper than ~70 km.
Using experiments, I have explored how U and Th elemental partitioning provides important constraints on melting models, depth of melting, and specifically generation of alkaline OIB. What if a minor mantle lithology (not peridotite) is important during melting, as some scientists have suggested? To fully evaluate the possibility of melting alternate, Gt-bearing lithologies instead of peridotite, I have measured U and Th partition coefficients in silica-poor garnet pyroxenite, a minor mantle lithology for which such measurements have been lacking. Forward-melting calculations using these experimental results to model time-dependent uranium-series isotopes do not support the presence of a fixed quantity of garnet pyroxenite in the source of OIB. Thus if the host of Gt is peridotite, measured (230Th/238U) does provide a depth control for melting (i.e. melting must occur deep enough for garnet to be stable in peridotite rocks).
I am also interested in melting at mid-ocean ridges (MOR), where new crust is forming at the surface of the Earth. I have focused on slow- and ultraslow-spreading ridges in the Arctic, which provide an interesting set of case studies. These ridges are end members of the global ridge system in a number of ways. The Kolbeinsey Ridge north of Iceland is one of the shallowest ridges in the world, probably because it is very close to the hot plume upwelling beneath Iceland, and the additional heat has produced thick crust at this ridge. My U-series measurements have confirmed that the lavas there are the result of large degrees of melting occurring in a deep melting column, extending well into the garnet stability zone. The Mohns and Knipovich Ridges further north have increasingly oblique spreading directions and grow more complex and experience slower and slower spreading towards the north. Their chemistry - long-lived radiogenic isotopes, trace elements, and U-series isotopes - suggests that there is a heterogeneous mantle source underlying this part of the Arctic ridge system. It has been suggested in prior studies that this source resulted from smearing of trapped sub-continental material beneath this relatively young basin.
Finally, the Gakkel Ridge beneath the Arctic ice cap is the slowest-spreading ridge on Earth. As the ridge in the Eastern Volcanic Zone (EVZ) approaches continental Asia, half-spreading rates slow to less than 6 mm/year. Volcanism is expressed as discrete volcanic centers, and in many locations there is mantle rock exposed right on the ocean floor. Such an ultraslow-spreading ridge is expected to have a thick, cold lithospheric cap, and the thin- to nonexistent crust must be the result of only small degrees of melting. My U-series measurements support these predictions: I observe only very small amounts of U-Th disequilibria on the EVZ (from small 238U excesses to small 230Th excesses). I also measured large excesses of 226Ra, a very short-lived isotope, which suggests shallow melting or melt-rock interaction in the thick lithosphere that underlies the ridge.
The work on the Kolbeinsey Ridge has revealed interesting relationships between isotopic ratios in rocks that have experience varying amounts of seawater interaction and alteration. I am currently working to measure more trace elements in these samples, to better constrain the alteration processes involved. U isotopes should be in equilibrium with each other in mantle melts, but the isotope 234U is enriched in seawater. Thus, (234U/238U) activity ratios higher than one reflect seawater contamination of the lava. For the Kolbeinsey, (234U/238U) and (230Th/238U) vary systematically, suggesting that the basalts have experienced variable extents of alteration by interaction with seawater. Preliminary results of this new study suggest that U isotopes, which vary widely in Kolbeinsey lavas, do not record seawater interaction in the same way as other commonly used indicators of seawater contamination in marine basalts, like Cl abundance or Cl/K ratios. Cl is likely sensitive to seawater interaction with lava or erupted glass in all shallow crustal environments, while U is more redox sensitive. Further study will explore this system further, to better understand how basalts are contaminated in the crust and how they are altered after erupting onto the seafloor.