Messages from the core-mantle boundary? The role of D'' in layered-mantle evolution Igor Tolstikhin, Ian Kramers & Albrecht W. Hofmann

Prologue Message from cosmochemistry: Please upgrade your pyrolite composition

Ringwood‘s method: Mantle peridotites have been depleted by the removal of melt. Original mantle composition can be reconstituted by recombining the melt with the depleted peritotite. Idea basically sound, but afflicted by the uncertainty of the exact compositions and the required proportion of peridotite and melt Ringwood‘s pyrolite is obsolete

Cosmochemical Estimates of Mantle Composition The silicate Earth consists of carbonaceous chondrites Minus the elements now residing in the core Minus the elements lost by volatilization

Method of Palme & O‘Neill, 2003, Treatise on Geochemistry FeO in mantle peridotite nearly constant: FeO = 8.1% Because FeO is not significantly affected by melt extraction

MgO in Silicate Earth

The five major oxides

Pyrolite versus Bulk Silicate Earth Recommended pyrolite upgrades: McDonough, W. F. and S.-S. Sun (1995). The composition of the Earth. Chem. Geol. 120, Palme, H. and O‘Neill, H. St.C. (2003) Cosmochemical estimates of mantle composition. In Treatise on Geochemistry. Vol

FUNDAMENTAL UNSOLVED PROBLEM OF EARTH SCIENCE

Role of mineral physics

Standard 2-layer model preferred by geochemists (Allègre version)

Smith &Sandwell Map of the Oceans Geochemical contrasts between ridges and plumes Old problem:

Geochemical Evidence Favoring 2-Layer Mantle

The geochemical constraints are not new, but they have not gone away, even though mineral physics no longer requires separate mantle reservoirs.

Nd-Sr Isotope Comparison: MORB versus OIB MORB data: PetDB (Su, Y.J. Dissertation Columbia Univ., 2002 OIB data: Albarède, Introduction to Geochemical Modeling, 1995.

Tolstikhin 3He/4He frequency 4 He/ 3 He distributions in Plumes and MORB MORB-source low in 3 He: „relatively degassed“ OIB-source high in 3 He: „relatively undegassed“ 3He = primordial, solar 4He = radiogenic,  particles Tolstikhin & Hofmann, (2005) PEPI, 148,

Neon isotopes in MORB and OIB Graham, D.W. 2003, in Porcelli et al. Noble Gases in Geochemistry and Cosmochemistry. Reviews in Mineralogy and Geochemistry. Min. Soc. Am. 47:

Xenon isotope production from extinct radioactivity

129 Xe/ 130 Xe Xe/ 130 Xe in MORB and OIB The 129 Xe excess requires very early mantle degassing and preservation of a very ancient noble gas reservoir The 129 Xe excess requires very early mantle degassing and preservation of a very ancient noble gas reservoir

Contradictory closure ages for 2-stage Xe degassing models Closure age for 129 Xe production ≤ 150 Ma Closure age for 136 Xe production ≥ 550 Ma Closure age for 129 Xe production ≤ 150 Ma Closure age for 136 Xe production ≥ 550 Ma Assume Complete Xe loss until T(clo) Complete retention after T(clo) Assume Complete Xe loss until T(clo) Complete retention after T(clo) This discrepancy requires existence of two Xe reservoirs

Three-Reservoir Mass Balance

Two-layer noble gas degassing evolution model Variations of this model developed by several authors, e.g. Tolstikhin, Kellogg, Porcelli, Allègre et al. Almost all of the action takes place in the upper mantle. Tiny whiffs of primordial noble gas from the nearly undegassed lower mantle are entrained in mantle plumes, and they leak into the upper mantle.

O‘Nions & Tolstikhin, 1996, Results Result of the two-layer degassing model: Mass transfer from the lower mantle to the upper mantle, described by the coefficient , is less than 1% of the mass of the lower mantle per Ga. This is at least 50 times less than the present-day subduction flux. Results provide strong support for two-layer convection with little mass exchange between layers But they are based on the assumption that the deep reservoir represents the entire lower mantle below 670km Result of the two-layer degassing model: Mass transfer from the lower mantle to the upper mantle, described by the coefficient , is less than 1% of the mass of the lower mantle per Ga. This is at least 50 times less than the present-day subduction flux. Results provide strong support for two-layer convection with little mass exchange between layers But they are based on the assumption that the deep reservoir represents the entire lower mantle below 670km

EVIDENCE FAVORING WHOLE-MANTLE CONVECTION

Seismic Tomography: Deep subduction of Farallon Slab Grand & van der Hilst (1997) Seismic tomography: Strong evidence for whole-mantle Circulation. Conventional two-layer convection is no longer tenable Figure 1. Cross sections of mantle P-wave (A) and S-wave (B) velocity variations along a section through the southern United States. The endpoints of the section are 30.1°N, 117.1°W and 30.2°N, 56.4°W. The images show variations in seismic velocity relative to the global mean at depths from the surface to the core-mantle boundary. Blues indicate faster than average and reds slower than average seismic velocity. The large tabular blue anomaly that crosses the entire lower mantle is probably the descending Farallon plate that subducted over the past ~100 m.y. Differences in structure between the two models in the transition zone (400 to 660 km depth) and at the base of the mantle are probably due to different data sampling in the two studies.

Finite Frequency Tomographic Images of Deep-Mantle Plumes Montelli et al., 2004, Science 303,

Larson: God‘s kitchen

How can we reconcile the contradictory evidence for whole-mantle and layered-mantle convection?

Kellogg‘s Stealth Layer Kellogg et al. 1999, Science, 283, : A smaller, seismologically invisible „stealth layer“

VLVP Seismic Characteristics A very broad region. A large negative shear velocity gradient (-2% %). Steeply dipping edges. Rapidly varying thickness and geometry over small distances. The maximum P velocity reduction is -3%.

BUT WE NEED A VERY ANCIENT RESERVOIR THAT PRESERVES PRIMORDIAL (≥4.5 Ga) COMPOSITION

Rationale for Model of Tolstikhin & Hofmann (PEPI, 148, , 2005)

Tolstikhin model-2 New Model: Noble gases stored in primitive D‘‘ layer

Creation of Ancient Chondritic Reservoir in D‘‘

Evolution of Nd Isotopes in the Mantle

New Evidence from 142 Nd, the decay product of extinct 146 Sm (t 1/2 = 100 Ma) Measurable differences in the abundance of 142 Nd can only be produced in the first 0.5 Ga of Earth evolution.

Non-chondritic 142 Nd Isotopes in Crust-Mantle Science, in press Requires ancient hidden reservoir (D‘‘ ?) with low Sm/Nd ratio. 142 Nd produced by extinct 146 Sm

Consequence for 143 Nd isotope mantle evolution Boyet & Carlson, Submitted to Science Hidden reservoir (D‘‘ ?) with low Sm/Nd ratio must be ancient and small The rest of the mantle is depleted (=high Sm/Nd) and MORB-like

Conclusions - 1 D‘‘ formed about 4.5 Ga ago, containing: (1) Subducted basaltic crust (80 %) containing heat-producing Th, U, K, plus other incompatible elements (2) Fe-rich chondritic material (20 %) causing high density (3) Implanted solar noble gases trapped in this layer

Dense Dregs at Base of the Mantle Christensen & Hofmann, 1994, JGR

Subsequent evolution Very slow loss by entrainment in plumes 80% remains in D‘‘ 20 % of original D‘‘ reservoir has been –entrained by uprising mantle flow –mixed with mantle

Evolution of Mass Fluxes

D‘‘- Mantle - Crust - Atmosphere Mass Balance

Role of melting at the base? Hotter early mantle + superheated core leads to partial melting in lower mantle. If melt is denser than solid, it will migrate downward and create a chemically dense layer at the base of mantle.

D‘‘ Eutectic (Boehler) T Dense Melt Magnesiowüstite (Mg,Fe)O Perovskite MgSiO 3

Conclusions - 3 Partially molten lowermost mantle High density because of melt composition Melt must segregate downward to create denser bulk composition in D‘‘ May have solidified or still be molten in part

Three Wise Men (Jacoby) Wolfgang Jacoby