Composition, Mixing, and Melting of the Mantle II: Beneath Hot Spots
Presiding: K Lee, California Institute of Technology; G Ito, Department of Geology and Geophysics, University of Hawaii
V42A-01 INVITED 10:30h
Constraints on Thermochemical Convection of the Mantle from Plume-related Observations
Although geochemical observations have long suggested a layered mantle with more enriched mantle material in the bottom layer to provide a significant amount of heat to the top layer, the nature of such a layering remains unclear. An important observation that has been used to argue against the conventional layered mantle model (i.e., the layering at the 670 km depth) was the plume heat flux [Davies, 1999]. Plume heat flux is estimated as ~ 3.5 TW, or 10% of the surface heat flux [Davies, 1988; Sleep, 1990]. In this study, we demonstrate with 3-D spherical models of mantle convection with depth- and temperature-dependent viscosity that observed plume heat flux, plume excess temperature (<350°C), and upper mantle temperature (~ 1300°C) can pose important constraints on the layered mantle convection. We show that for a purely thermal convection model (i.e., a whole mantle convection), the observations of plume heat flux, plume excess temperature, and upper mantle temperature can be simultaneously explained only when internal heating rate is about 65%. For smaller internal heating rate, plume heat flux and plume excess temperature would be too large, and upper mantle temperature would be too small, compared with the observed. This suggests that for a whole mantle convection the CMB heat flux needs to be > 10 TW. For a core with no significant heat producing elements, such large CMB heat flux may lead to too rapid cooling of the core or a too young inner core. A layered mantle convection may help reduce the CMB heat flux. For layered convection models, we found that the top layer needs to be ~70% internally heated to explain the upper mantle temperature and plume-related observations, and this required internal heating ratio is insensitive to the layer thickness for the bottom layer (we used ~600 km and 1100 km thicknesses). This result suggests that heat generation rate for the bottom layer cannot be significantly larger (< a factor of 2) than that for the top layer. thus challenging the conventional geochemical inference for an significantly enriched bottom layer. However, this is more consistent with recent estimate of the MORB source composition that increases heat producing element concentration by a factor of three compared with the previously proposed.
V42A-02 10:50h
Phase assemblage and stability of pyroxenite at Lower-Mantle conditions
We investigated a mixture of natural pyroxenites (MIX1G [Hirschmann et al., Geology, 2003]) in the laser-heated diamond-anvil cell at Lower-Mantle conditions (pressures between 25 and 60 GPa and temperatures above ~2000 K) in order to determine the phase assemblage and stability of this composition and to evaluate its suitability as a candidate deep ocean island basalt (OIB) source or widespread Lower Mantle component. Quenched samples were characterized by in-situ high-pressure synchrotron-based x-ray diffraction. Recovered samples were also analyzed for composition by energy-dispersive x-ray spectroscopy on a scanning electron microscrope in thick-section and thin-film modes. At these Lower-Mantle conditions, pyrolite-like compositions yield a three-phase assemblage (Mg, Fe, Al)SiO3 orthorhombic perovskite, (Mg, Fe)O magnesiowüstite and CaSiO3 calcium perovskite. This pyroxenite, on the other hand, shows an excess of silica, evident with the identification of stishovite, as well as a sodium, aluminum-rich phase in addition to (Mg, Fe, Al)SiO3 orthorhombic perovskite and CaSiO3 calcium perovskite phases. Whereas pyrolite compositions yield a density ~2% lower than Preliminary Reference Earth Model (PREM), our estimates suggests MIX1G pyroxenite is ~1% denser than PREM. Present mineral physics data therefore suggest a Lower Mantle composition different from the upper mantle but not so enriched as pure pyroxenite.
V42A-03 11:05h
Geochemistry of Ocean Island Basalts: A Deep Mantle Reservoir or Deep Upper Mantle Melting?
Geochemical observations of ocean island basalts (OIBs) and mid-ocean ridge basalts (MORBs) indicate that the mantle is heterogeneous on scales much smaller than the sizes of the mantle melting zones. Materials with higher incompatible-element and volatile contents will begin melting at greater depths than depleted materials, and thus will tend to be expressed more heavily in OIBs compared to MORBs for two main reasons. First, melting beneath hotspots, which generates most OIBs, is likely to occur at greater depths, owing to the presence of thick lithosphere (intraplate) and/or thick crust (near-ridge). Second, if hotspots are caused by narrow, buoyant mantle upwellings (e.g., mantle plumes), calculations predict a larger flux of mantle will pass through the deepest portion of the melting zone than through the shallowest portion. Such "plume flow" contrasts with flow beneath mid-ocean ridges, where mantle flux is likely to be more uniform through all depths of the melting zone. Assuming the isotopic end-member DMM begins melting shallowest and that other isotopic components, broadly similar to EM1, EM1, HIMU and C (or FOZO), begin melting deeper, calculations of flow and melting of a heterogeneous but well-mixed mantle can explain many differences between OIBs and MORBs in terms of Sr, Nd, and Pb isotope ratios. Models can also explain the relatively high values and large variability of 3He/4He of OIBs relative to MORBs if C is the source of highest 3He/4He. One test of the importance of melting depth is provided by the Hawaiian and Hawaiian-Emperor volcano chain, in which the Hawaiian hotspot erupted lavas onto seafloor of progressively increasing age. He, Sr, Nd, and Pb isotope ratios become increasingly DMM-like towards the oldest parts of the chain [Regelous et al. 2002; Keller et al., 2004], which can be explained by melting of a heterogeneous mantle plume beneath lithosphere of decreasing thickness. The main requirement is that an EM-like source begin melting deepest, a high 3He/4He, C-like source begin melting at intermediate depths, and a refractory, DMM-like source begin melting shallowest. Thus, rather than reflecting a deep reservoir in a compositionally layered mantle, the isotopic characteristics of OIBs could reflect deep melting of a heterogeneous, possibly non-layered mantle.
http://www.soest.hawaii.edu/GG/FACULTY/ITO
V42A-04 11:20h
Recycled Oceanic Mantle Lithosphere in Hawaii: The Samples and the Models
Subduction of basaltic oceanic crust is a key mechanism for introducing chemical heterogeneities into the Earth's mantle. Together with the crust however, the associated depleted lithosphere is also subducted. In the mantle, basaltic crust has a lower solidus while the depleted lithosphere has a higher solidus temperature than the ambient mantle. Thus, if a recycled depleted lithosphere is part of an upwelling plume, it could survive melting and be more easily recognized in mantle xenoliths found in plume volcanism, than its basaltic counterpart. Some spinel peridotite xenoliths from Salt Lake Crater, Oahu, Hawaii, have coupled highly radiogenic Hf (εHf= 3 to 114) and unradiogenic Os isotopes (187Os/186Os = 0.1291-0.1134). Os and Hf isotopes correlate with indices of depletion (e.g. Mg&35;) suggesting that their extreme values are related to melting and not to an exotic mantle component never before sampled. Both Lu-Hf systematics and Re-depletion ages in these samples give ages between 500 million to 2 billion years; those are hard to reconcile with the 90 million year old Pacific lithosphere in Oahu. We suggest that these peridotites represent ancient recycled depleted lithosphere brought to the surface by the Hawaiian plume, and represent a plume component previously not recognized in lavas. However, the lack of extreme Hf-Os isotope compositions in the erupted lavas suggests that although this recycled lithosphere is part of the plume, it does not significantly contribute to plume volcanism. It has been suggested that recycled metasomatized lithosphere can be the source of OIB volcanism. Therefore, we can use the trace element compositions of peridotite and pyroxenite xenoliths from Oahu, representing the metasomatized oceanic lithosphere, to test this hypothesis. Assuming an origin from a MORB or primitive mantle source and an age of 1.5 to 2.5 Ga, this lithosphere develops highly variable compositions in any combination of Hf-Nd-Sr-Pb isotopes that hardly overlap or extent towards any proposed mantle endmembers. Pyroxenites (representing fertile veins in the mantle) develop compositions that fall below the terrestrial array in Hf-Nd space, while the peridotite compositions fall both above and below the array. In Sr-Nd space, only some pyroxenites with phlogopites and carbonates will develop compositions falling within or at the extension of the OIB field, while peridotites evolve with unradiogenic Sr, for a given Nd. In any Pb isotope space, peridotites will develop radiogenic 206Pb/204Pb while pyroxenites are highly variable and can develop both radiogenic and unradiogenic Pb values. We show that recycled metasomatized lithosphere is too heterogeneous to develop the well-defined OIB arrays in the different isotope correlation diagrams. We suggest that some other mechanism in the mantle is needed that either homogenizes the heterogeneous recycled lithosphere, or only a small portion of that lithosphere participates in OIB melting. The survival of recycled lithospheric peridotites with extreme isotopes at Salt Lake Crater however, suggests that long storage and stirring in the mantle alone does not erase such heterogeneities.
V42A-05 11:35h
Transition metal evidence for coherent sections of recycled oceanic lithosphere in hotspot source regions: implications for the origins of hotspot magmatism
The debate over the origin of "hotspots" has recently been reinvigorated. In the traditional view, "hotspots" represent the surface manifestations of thermal plumes arising from deep thermal boundary layers. Alternatively, "hotspots" may derive from preferential melting of more fertile and hence more fusible zones in the mantle without calling upon anomalously high temperatures and hence no need for thermal plumes. This view has been fueled by the emerging acceptance that eclogite, characterized by a low temperature solidus, may be widespread in hotspot source regions. These eclogite zones could represent fragments of oceanic crust that have been recycled into the convecting mantle through subduction zones. Eclogite pods could also exist in the thermal plume scenario, but in that case, the eclogite pods are simply entrained in hot upwelling plumes. Both scenarios, however, come with their own paradoxes. In the thermal plume scenario, the hotter temperatures would result in higher degrees of melting. Because hotspot magmas are already highly enriched in incompatible trace elements compared to MORBs, the high degrees of melting in turn imply source regions with somewhat unreasonable enrichments in incompatible elements. In the alternative view, the problem is that eclogite is much denser than normal mantle (e.g., pyrolite) so that once eclogitized oceanic crust sinks to the deep mantle, it is unlikely to return on its own accord unless it is aided by a thermal upwelling itself! Both paradoxes might be easily reconciled if previously melt-depleted mantle, such as the harzburgitic residuum of oceanic crust generation, is involved in some manner. Harzburgite has a higher melting temperature than normal mantle, eliminating the requirement for high degrees of melting in the plume model and relaxing the requirement for highly enriched source regions. In addition, it has a lower density than normal mantle and eclogite, perhaps allowing it to return to the uppermost mantle by virtue of its own chemical buoyancy and in so doing entrain eclogite pods along the way. Given the evidence for recycled oceanic crust in hotspot source regions, the presence of harzburgite seems almost inescapable since the proportion of complementary harzburgite in oceanic lithosphere is much greater than that of oceanic crust itself. The question is whether this harzburgitic residuum really exists in hotspot source regions. Most trace element and isotopic systems are based on incompatible element systems and hence are sensitive to metasomatic or crustal components but not residual mantle. Only mildly incompatible to compatible elements can be used to track residual mantle. Such elements include the first series transition metals, V, Sc, Cr, Mn, Fe, Co and Ni. We show that many hotspot magmas are characterized by high Ni, Cr, Fe, and Co, low Sc and variable V contents. While any one of these features can be explained by a number of hypothetical scenarios, only small degree melting of harzburgite can successfully explain the group systematics. Hotspot magmas may represent a mixture of high degree melts of eclogite with small degree melts of harzburgite. These observations suggest that entire sections of oceanic lithosphere may be present in some hotspot source regions. Our geodynamic models show that large sections of oceanic lithosphere, if sufficiently dry, can be preserved in the convecting mantle and sink to the lower mantle by their negative thermal buoyancy. However, upon thermal re-equilibration with the ambient mantle, these coherent sections can return to the uppermost mantle by entrainment in thermal upwellings or possibly by intrinsic chemical buoyancy alone.
V42A-06 11:50h
What does hotspot-ridge interaction tell us about mantle plumes?
While the scientific community is currently contending the origin and nature of 'plumes', there is no doubt that close proximity to hotspots affects the structural, volcanic and geochemical character of spreading ridges. We can consider two end-member processes that may account for the 'plume' phenomenon: the classic and current paradigm of a column of rising mantle, made buoyant by an excess of temperature and originating deep within the mantle, or a non-dynamic mechanism involving a geochemical anomaly, residing passively in the shallow mantle, that melts spontaneously as a result of intraplate stresses or asthenospheric motion. While a priori, both mechanisms may hold true, they can be distinguished for individual 'plumes' where they interact with a spreading ridge that leaves a lithospheric and crustal trail recording the history of that interaction. The Reykjanes Ridge, southwest of Iceland, has all the attributes associated with a ridge-centred 'plume'. These include: thickening of the crust, shallowing of the ridge axis, an increase in segment length, and enrichment in geochemical tracers of the 'plume' mantle. Evidence for dynamic interaction between ridge and plume comes from southward closing V-shaped ridges, centred on the plate boundary, that indicate southward advection of plume mantle away from Iceland. Geochemical tracers include incompatible trace element enrichment and isotope ratios (e.g. Sr, Nd, Pb and He) that show multi-component mixing between several depleted and enriched sources. These sources are resident in the mantle beneath the Reykjanes Ridge and show modification by melting processes, consistent with a history of advection away from Iceland. Further evidence for interaction between a dynamic 'plume' and spreading ridge comes from the Reunion-Central Indian Ridge couplet, which comprises an off-axis plume and medium-rate spreading centre. Here, the ridge also exhibits attributes associated with 'plume' influence such as shallowing depth, increase in segment length, and multi-component geochemical mixing that varies with distance along the plate boundary. Both the Icelandic and Reunion 'plume'-ridge couplets provide compelling evidence for advection and lateral flow of 'plume' mantle away from a central source and subsequent mingling with depleted, sub-ridge mantle characteristic of MORB genesis. As such, these studies support the current paradigm for mantle 'plumes' as phenomena associated with actively upwelling material that has resided deep within the mantle and whose influence upon spreading ridges is profound.