Geochemical observations and one layer mantle convection

https://doi.org/10.1016/S0012-821X(99)00258-7Get rights and content

Abstract

Rare gas systematics are at the heart of the discrepancy between geophysical and geochemical models. Since more and more robust evidence of whole mantle convection comes from seismic tomography and geoid modeling, the interpretation of high 3He/4He in some oceanic island basalts as being primitive has to be revisited. A time dependent model with five reservoirs (bulk mantle, continental crust, atmosphere, residual deep mantle and D″) is studied for Rb/Sr and U/Pb/He systems. The dynamics of this model correspond to whole mantle convection in which subducted oceanic crust, transformed into dense assemblages, partially segregates to form a D″ layer growing with time as in Christensen and Hofmann [J. Geophys. Res. 99 (1994) 19867–19884]. A complementary cold and depleted harzburgitic lithosphere remains above D″. We assume that hotspots arise from the deep thermal boundary layer and tap, in variable proportions, material from both the residual deep mantle and D″. The difference between HIMU and Hawaiian basalts is attributed to HIMU being mostly from strongly degassed oceanic crust, though enriched in incompatibles (D″), while Hawaii is mostly from MORB source residuals that are variably degassed and depleted. We suggest that a significant part of the Earth’s radioactive elements (∼1/3) is trapped in the D″ layer.

Introduction

Current views of mantle stratification derived from geochemistry are often in contradiction with most recent geophysical observations. The former generally assume the existence of independent reservoirs, including a pristine or more or less primitive lower mantle [2], [3]. In contrast, seismological imaging, interpretation of the large scale gravity field, and numerical simulation all suggest a significant material exchange throughout the entire mantle [4], [5], [6]. It thus seems necessary to re-examine the geochemical observations on mantle stratification to see whether or not they can be interpreted in terms of large mass exchange in the whole mantle following the attempt by Albarède [7].

Rare gas systematics are at the heart of the contradiction between geochemistry and geophysics. Oceanic basalts often show contamination from atmospheric Xe, Ne and Ar. Therefore, we will focus on the interpretation of He, which should be easier as He simply escapes from the mantle before being lost in space (although some reincorporation has been suggested [8]). 4He is the radioactive daughter of U or Th. Mantle rocks also contain the primordial isotope 3He. During melting, the concentrations of an element in the melt Cmelt and in the source Csource are related by a fractionation coefficient K:Cmelt=K (F,D) CsourceThis coefficient is related to the melt fraction F and to the partition coefficient between the solid and the melt, D:Csolid=DCmelt

Although various assumptions can be made to derive K, in the simplest case of batch melting one has:K=1F+D(1−F)For two isotopes, D is the same and therefore isotopic ratios are conserved during melting. Isotopic measurements from mid-ocean ridge basalts (MORBs) give 3He/4He=1.2×10−5 with very little dispersion [9]. Oceanic island basalts (OIBs) show much larger scattering from 0.7×10−5 at Mangaia [10] to 5×10−5 for the Loihi seamount [11], [12].

Changes in the 3He/4He ratio should be reflected by changes in the 3He/U ratio as U decay produces 4He [13]. A high 3He/4He ratio means either a high content of 3He or a low content of 4He. The first hypothesis is commonly accepted: Loihi and Iceland hotspots are tapping a primitive and undegassed source while the MORB source is depleted and degassed (see references in [14]). The second hypothesis means that the source of high 3He/4He ratio is depleted in U and Th. This point of view has been shared by Anderson [8] and Albarède [7], although with different implications.

The degassed or undegassed nature of the source of hotspots should be determined by the direct observation of He abundances. However, these observations are much more ambiguous than isotopic ratios as the amount of escaped gases at the surface is very difficult to estimate. The trace element concentrations of basalts can give information about the depletion of their source. As seen in Fig. 1, 3He/4He of some OIBs is anti-correlated with the U content; the more U depleted the basalt, the higher the 3He/4He. As U is highly incompatible (D=0.005), its fractionation during melting decreases with the melt fraction. When corrected for magma fractionation, the sources of Hawaiian basalts and MORBs have the same U content of 0.008 ppm [15]. This contradicts the interpretation of a primitive origin for the source of high 3He/4He basalts.

Isotopes other than rare gases have been used to infer mantle dynamics and imply that the OIB source is heterogeneous and made by a mixture between different end-members (see [14], [16] for reviews). One of these end-members, called HIMU (high μ means high U/Pb), clearly shows recycling of oceanic crust. The HIMU source is explained by a process that fractionates U with respect to Pb. Such a U enrichment occurs in the oceanic crust by alteration and by dehydration during subduction. Even Hawaii shows evidence of crustal recycling, suggested by O [17], Os and Hf isotopes [18], even though its He ratio is described as primitive. The observed isotopic values (Pb, Os, Hf) require a long isolation period between 1 and 3 Ga.

Section snippets

Radioactive sources

In addition to producing rare gases, U, Th and K are the sources of radioactive heat. It is well known that the budget of radioactive elements in the bulk silicate earth (BSE) cannot be balanced by the content of the continental crust (CC) and that of the MORB source extended to the whole mantle. A hidden reservoir of trace elements must exist (unsampled or poorly sampled). The balance of radioactive heat production rate for the BSE can be written:HBSE=HMORB source+HCC+Hhid

From the

Mantle convection

A number of geophysical results have made the hypothesis of a stratified model (at least stratified at 670 km depth) difficult to sustain. The most visual evidence comes from seismic tomography. Since Grand [26] we know that the Pacific subduction below North America is continuous through the whole mantle. A similar pattern of downgoing slabs is observed below the Tethyan suture, from the MediteSrranean Sea to the North of Australia [4], [27], [28]. On a larger scale, these models together with

Mantle mixing

Since McKenzie [38], many papers have studied the mixing properties of passive [39], [40], [41] and active tracers [1], [42] in mantle convection. However, no general consensus has emerged. By varying by only a small amount the characteristics of the convection, very efficient to highly inefficient mixing properties can be obtained [40]. In a rather smooth 3D flow field, totally unmixed islands can survive within a well mixed background [41]. At 3D, convection models seem to predict much slower

A box model of mantle chemistry

In the previous sections we have discussed various geochemical and geophysical constraints. We now want to show that all these observations are in agreement with a model in which a large mass flux exists throughout the transition zone. Although ultimately chemical exchanges will have to be modeled along with thermo-chemical convection models, this goal is presently out of reach: the convection codes are not able to generate plates self-consistently and to model oceanic and continental crust

Results

Having defined all the parameters entering the five equations of mass conservation Eq. 6 and the 45 equations of element conservation Eq. 7 we only need to choose the initial concentrations in BM and RDM (the other reservoirs are not present at time t=0) which are assumed primitive [19], [61]. The initial isotopic ratios are from Zindler and Hart [16] except for He [61].

To explore the parameter space, a general nonlinear inversion using the least squares criterion [62] has been chosen. This

Discussion and conclusions

Our model does not take into account the time evolution of the various fluxes with larger CC formation rates and shorter residence times during the Archean [55]. However, it gives a reasonable model for the sources of HIMU type and high 3He/4He Hawaiian type basalts, the former containing a rather large quantity of ancient oceanic crust stagnating near the CMB, the later mostly formed from depleted old harzburgite. In the framework of this model, most of the mantle would be of uniform

Acknowledgements

We would like to thank F. Albarède and Ph. Gillet for constructive discussions, S. Sheppard for improving our English, W. White, G. Davies and P. van Keken for their extensive and fruitful reviews. This work has been supported by CNRS-INSU programs. Some of the computations have been performed thanks to the PSMN computing facilities.[AH]

References (65)

  • A.W. Hofmann

    Chemical differentiation of the Earth: the relationship between mantle, continental crust, and oceanic crust

    Earth Planet. Sci. Lett.

    (1988)
  • M.T. McCulloch et al.

    Geochemical and geodynamical constraints on subduction zone magmatism

    Earth Planet. Sci. Lett.

    (1991)
  • G.F. Davies

    Geophysically constrained mantle mass flows and the 40Ar budget: a degassed lower mantle?

    Earth Planet. Sci. Lett.

    (1999)
  • U.R. Christensen et al.

    Segregation of subducted oceanic crust in the convecting mantle

    J. Geophys. Res.

    (1994)
  • R.K. O’Nions et al.

    Heat and helium in the earth

    Nature

    (1983)
  • C.J. Allègre et al.

    Rare gas systematics: formation of the atmosphere, evolution and structure of the mantle

    Earth Planet. Sci. Lett.

    (1986)
  • S.P. Grand et al.

    Global seismic tomography a snapshot of convection in the Earth

    GSA Today

    (1997)
  • Y. Ricard et al.

    A geodynamic model of mantle density heterogeneity

    J. Geophys. Res.

    (1993)
  • G.F. Davies

    Ocean bathymetry and mantle convection 1. large-scale flow and hotspots

    J. Geophys. Res.

    (1988)
  • D.L. Anderson

    Helium-3 from the mantle primordial signal or cosmic dust?

    Science

    (1993)
  • M. Moreira et al.

    Rare gas systematics in popping rocks isotopic and elemental compositions in the upper mantle

    Science

    (1998)
  • T. Hanyu et al.

    The uniform and low 3He/4He ratios of HIMU basalts as evidence for their origin as recycled materials

    Nature

    (1997)
  • M.D. Kurz et al.

    Helium isotopic systematics of oceanic islands and mantle heterogeneity

    Nature

    (1982)
  • D. Graham et al.

    Extreme temporal homogeneity of Helium isotopes at Piton de la Fournaise, Réunion Island

    Nature

    (1990)
  • A.W. Hofmann

    Mantle geochemistry the message from oceanic volcanism

    Nature

    (1997)
  • A. Zindler et al.

    Chemical geodynamics

    Annu. Rev. Earth Planet. Sci.

    (1986)
  • J.M. Eiler et al.

    Oxygen isotope constraints on the sources of Hawaiian volcanism

    Earth Planet. Sci. Lett.

    (1996)
  • J. Blichert-Toft et al.

    Hf isotope evidence for pelagic sediments in the source of Hawaii

    Science

    (1999)
  • R.L. Rudnick et al.

    Nature and composition of the continental crust: a lower crustal perspective

    Rev. Geophys.

    (1995)
  • H.N. Pollack et al.

    Heat flow from the Earth’s interior: analysis of the global data set

    Rev. Geophys.

    (1993)
  • N.H. Sleep

    Hotspots and plumes: some phenomenology

    J. Geophys. Res.

    (1990)
  • H.-P. Bunge et al.

    A sensitivity study of 3-D spherical mantle convection at 108 Rayleigh number: effects of depth-dependent viscosity, heating mode and endothermic phase change

    J. Geophys. Res.

    (1997)
  • Cited by (143)

    • Tectonics is a Hologram

      2023, Dynamics of Plate Tectonics and Mantle Convection
    • Melting in the Earth's Deep Interior

      2018, Magmas Under Pressure: Advances in High-Pressure Experiments on Structure and Properties of Melts
    • Melting in the earth’s deep interior

      2018, Magmas Under Pressure: Advances in High-Pressure Experiments on Structure and Properties of Melts
    • Extreme incompatibility of helium during mantle melting: Evidence from undegassed mid-ocean ridge basalts

      2016, Earth and Planetary Science Letters
      Citation Excerpt :

      Class and Goldstein (2005) pointed out that ocean islands having higher 3He/4He have depleted (high) time-integrated Sm/Nd, and show elevated 143Nd/144Nd compared to expected bulk silicate Earth, suggesting an association between source depletion by partial melting and elevated 3He/4He. Albarède (1998) and Coltice and Ricard (1999) even suggested that bulk solid/melt distribution coefficients for He and U were ∼0.01 and ∼0.005, respectively, so that mantle residues of partial melting would have lower U/3He ratios, and over time would tend to show elevated 3He/4He compared to less depleted mantle regions. They proposed that accumulations of recycled oceanic lithosphere could be the source of high 3He/4He ratios in basalts derived from deep hotspot sources.

    View all citing articles on Scopus
    View full text