Trace element geochemistry of oceanic peridotites and silicate melt inclusions : implications for mantle melting and ocean ridge magmagenesis

Thumbnail Image
Johnson, Kevin T. M.
Linked Authors
Alternative Title
As Published
Date Created
Atlantis II Fracture Zone
American-Antarctic Ridge
Southwest Indian Ridge
Replaced By
Atlantis II (Ship : 1963-) Cruise AII107
Islas Orcadas (Ship) Cruise IO1176
Polarstern (Ship) Cruise PS86
Melville (Ship) Cruise Protea 5
Robert D. Conrad (Ship) Cruise RC29-9
The mantle melting process is fundamental to basalt genesis and crustal accretion at mid-ocean ridges. It is believed that melts ascend more rapidly than the surrounding mantle, implying a process similar to fractional melting may be occurring, but geochemical evidence for this has been lacking. Furthermore, crustal accretion is thought to be episodic at slow spreading ridges, but sampling programs that can test this temporal variability are virtually nonexistent. This dissertation examines the trace element compositions of abyssal peridotites and discusses how they preserve details of the melting process that are not recognizable in mid-ocean ridge basalts. The results support fractional melting as the dominant melting process in the sub-ridge upper mantle. Evidence is also presented supporting non-steady state mantle melting at the Atlantis II Fracture Zone cutting the very slow spreading Southwest Indian Ridge. Trace element compositions of peridotite clinopyroxenes from fracture zones along the American-Antarctic and Southwest Indian Ridges vary as a function of proximity to hotspots. The results presented in Chapter 2 are consistent with higher degrees of melting and greater incompatible element depletion in the upper mantle near hotspots. All peridotites studied are consistent with being residues of fractional melting and inconsistent with batch melting. Some samples recovered near hotspots appear to have begun melting in the garnet stability field, deeper than samples recovered away from hotspots. Most samples show pronounced negative Zr and Ti anomalies, which increase with increasing incompatible element depletion (increased melting), on extended rare earth (spider) diagrams. The results of Chapter 2 indicated the importance of accurately knowing trace element partition coefficients between clinopyroxene and liquid. It was found that existing partitioning studies report either rare earth elements, Ti, or Zr, but not all elements together. Thus, there is ambiguity about relative partition coefficients for these elements. Accurate knowledge of partitioning is important in understanding the formation of negative Zr and Ti anomalies observed in peridotite clinopyroxenes as well as in constructing realistic melting models for peridotites. To that end, Chapter 3 reports the results of a clinopyroxene/basaltic liquid trace element partitioning study carried out on natural dredged basalts and experimental charges of mid-ocean ridge basalts. It was found that there are small negative anomalies in the partiton coefficients of Zr and Ti relative to adjacently plotted rare earth elements on spider diagrams. Fractional melting implies that small parcels of refractory (e.g., high Mg/[Mg+Fe]), incompatible element depleted melts must exist somewhere in the ascending body of melt. Since mixing, wall rock reaction, and fractional crystallization probably alter the compositions of silicate melts extensively on their way from source to surface, representatives of these refractory fractional melts will rarely be erupted as flows on the seafloor. However, some refractory silicate melt inclusions possess compositional characteristics akin to those expected in fractional melts, i.e. low incompatible element concentrations and fractionated trace element ratios. Chapter 4 is a study of refractory melt inclusions from a variety of tectonic settings. The inclusions were obtained from Dr. A. V. Sobolev of the Vernadsky Institute of Geochemistry, Soviet Academy of Sciences, Moscow. They are not ideally suited for studying mid-ocean ridge processes, as only a few of the inclusions are from this environment, but in general, the inclusions show more refractory, incompatible element depleted compositions than their host lavas. Furthermore, the suite of inclusions in different mineral phases contained in a single N-type mid-ocean ridge basalt show variable trace element characteristics indicating unrelated sources for some inclusions. The results of the study do not strongly endorse the fractional melting hypothesis, but some support is suggested by trace element depletions and fractionations warranting a more thorough study of a suite of inclusions. Finally, the along-ridge major and trace element variability in peridotites observed previously and in Chapter 2 is compared to the variability found in a single fracture zone. The high sampling density at the Atlantis IT Fracture Zone on the Southwest Indian Ridge, coupled with its great distance from a hotspot make it a good subject for a baseline study. It was found that the compositional variability observed in peridotites from the Atlantis II Fracture Zone covers nearly the whole range of compositions found along the AmericanAntarctic and Southwest Indian Ridges in Chapter 2. However, there are systematics to this wide range, suggesting different processes may control the depletions. On the eastern side of the transform, a compositional gradient is observed from the center of the eastern wall to the northern ridge-transform intersection. Peridotites on this side have become gradually more depleted in incompatible elements and modal clinopyroxene over at least the last 10-11 million years. Samples from the western side of the transform are, in general, more depleted than those from the eastern side and show some indication of a compositional gradient as well, although sampling is less dense. Basalts from the western side are clearly different in iron composition and degree of rare earth element fractionation. These differences are consistent with higher pressure, higher degrees of melting producing lavas on the western side. It is believed that the long wavelength chemical variations corresponding to hotspot proximity described in Chapter 2 result from regional thermal conditions in the upper mantle imposed, in large part, by the hotspots. On the other hand, the short wavelength variability on a fracture zone or spreading cell scale may result from episodic mantle upwelling and magma production due to non-steady state accretion at very slow spreading ridges.
Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the Massachusetts Institute of Technology and the Woods Hole Oceanographic Institution June 1990
Embargo Date
Cruise ID
Cruise DOI
Vessel Name