Untitled Document

Dynamics and Chemical Evolution of the Earth's Early Mantle

Geoffrey F. Davies 1 , Jinshui Huang 1

1 Research School of Earth Sciences, Australian National University, Canberra ACT 0200, Australia

Reconciliation of geophysical and geochemical observations of the Earth's mantle has been a challenging problem for about three decades. Numerical models are an essential tool in this work because they can simultaneously address the physical and chemical evidence within the framework of the physics governing mantle dynamics and evolution. One challenge has been to explain the persistence of chemical heterogeneity in the mantle for about two billion years despite the mixing effect of mantle convection. Another has been to understand quantitatively how the mantle reached its present condition as it evolved from earlier, hotter states. We have been using two-dimensional (2D) and three-dimensional (3D) numerical models to address these questions.

Last year we reported that the main control on the survival of chemical heterogeneity seems to be the rate at which material fluxes through melting zones near the Earth's surface. Other factors such as the viscosity structure of the mantle and geometry seem to be secondary. These results have been confirmed and extended in spherical 3D models that include the excess density of simulated subducted oceanic crust, which tends to accumulate at the bottom of the models. A simple theory that successfully interpreted the results with neutrally buoyant chemical tracers has been extended to also account for the effect of the excess density. These results provide us with simplified formulas that will help to interpret geochemical observations.

The 3D results are now being extended to higher resolution using another computer code, Citcom, which allows a regional segment of the mantle to be modelled (figure 13). The effects the higher resolution and of more realistic motions of tectonic plates are being investigated.

2D models can be run at much higher resolution than 3D models, and they can therefore be used to explore the dynamics of the mantle early in Earth history when it was hotter and less viscous, plates were thinner and the flow was more complex. Initial results were reported last year showing that a dynamic stratification occurs in hotter mantles, with the denser subducted oceanic crust tending to settle out of the upper mantle, leaving it relatively depleted in that component. Since that component is also the most readily melted, this implies that new oceanic crust, formed by melting under mid-ocean ridges, would be relatively thin, only about 5 km rather than the 20-30 km expected at those temperatures. These results provide a straightforward explanation for evidence from ancient rocks that the early mantle was strongly depleted in the components associated with subducted oceanic crust.

The robustness of these results has by now been explored more thoroughly, and they do not depend significantly on the presence of phase transformation effects, nor on the detailed distribution of internal heat sources. It has been found that the hottest models were not fully resolved initially, and as a result the crustal thickness has been revised upwards to the range 5-8 km.

The ultimate goal of the 2D studies has been to let models evolve from the earliest mantle conditions to the present. This work has now begun, after an extensive exploration of model behaviours at particular temperatures. This will allow more definitive tests of earlier results on simulated isotopic ages, as well as revealing to what degree the early stratification survives to the present.

Figure 13. Cutaway views of three-dimensional models of mantle convection with tracers representing components with basaltic composition. Left: temperature distribution; Right: tracer concentration, relative to the average concentration of 1. The 90-degree cutaway (centre of view) reveals the interior down to the core-mantle boundary. The surface of the view has the top 100 km removed. On the right of the left panel are spreading centres (yellow and red) and on the top and left are subduction zones (deep blue). The red surface area is under a simulated continental region where the hot mantle cannot cool. The remaining blue surface areas are occupied by oceanic plates that cool as the move away from spreading centres. The tracer distribution reveals the complexity of stirring in three dimensions, with high concentrations where simulated oceanic crust has subducted or is subducting (dark blue).