Phase transition evolution and convection style in the martian mantle

(AGU FM07, P41A-0191, modified and extended version)

T. Ruedas P. J. Tackley
Department of Terrestrial Magnetism
Carnegie Institution of Washington, Washington DC, USA
email: ruedas at dtm dot ciw dot edu
Institut für Geophysik
ETH Zürich, Switzerland

Cite as: Ruedas, T., Tackley, P. J. (2007): Phase transition evolution and convection style in the martian mantle. Eos Trans. AGU, 88(46), Fall Meet. Suppl., Abstract P41A-0191, 2007.

Model design

We present preliminary models of mantle convection in Mars over a timespan of several billions of years. We have combined the mantle convection program STAG3D (e.g. Tackley 1998) with a parameterized thermodynamic model of martian mantle mineralogy (fig.1a,b) and carried out calculations of convection in a 2D compressible model of the planet's mantle. The model fully takes into account the thermoelastic properties of the mantle, including p,T-dependent density, expansivity, heat capacity, thermal conductivity and phase transitions of olivine and non-olivine phases; mineral endmember data are mostly taken from Saxena et al. (1993). All four models are heated from below by a cooling core, and two of them are also heated from within by the radioactive decay of 40K, 235U, 238U, and 232Th, whose concentrations were taken from Wänke and Dreibus (1994). The viscosity of the mantle is temperature and pressure-dependent. The radius of the martian core, and hence the depth of the mantle, is only known to within a few hundred kilometers; we ran models with mantle depths of 1700 and 2000 km, which essentially cover the range of possible values.

The higher iron content of the martian mantle as compared to Earth's mantle results in a mineralogy and phase transition pattern which is somewhat different to that of the Earth (fig.1a); for instance, ringwoodite appears already at lower pressures than wadsleyite. Moreover, the two-phase loops in the Mg2SiO4-Fe2SiO4 system are broader and their position and width seem to be more sensitive to changes in temperature. We have parameterized phase diagrams for this system (Katsura and Ito 1989, Ito and Takahashi 1989, Katsura et al. 2004) in terms of p, T, and Mg#; for the non-olivine phases, which are more complicated and less well investigated, and for the general phase proportions we have used the results by Bertka and Fei (1997), supplemented by data on endmembers for the Clapeyron slopes of phase transitions (e.g. Gasparik 2003, Fei et al. 2004). Water and melting have not been considered in these models.

The core-mantle boundary in Mars happens to lie at about the depth of the ringwoodite-(perovskite+ferropericlase) transition. Previous studies (e.g. Harder and Christensen 1996) have already demonstrated the possibility that Mars had a perovskite layer at the base of its mantle which may have disappeared during the history of the planet as a consequence of secular cooling. Here we show the evolution of the phase transition patterns, i.e. the changes in position and width of transitions and the mineralogical changes, as a consequence of secular cooling and discuss their effect on the long-term convection style of the martian mantle.


Fig.1: a) Isothermal phase diagram, Mg2SiO4-Fe2SiO4 system; b) Mantle mineralogy along the initial adiabat for h=2000 km; c) Initial adiabat, averaged areotherms; d) Initial averaged density and gravity. Click on images for larger versions.

Results

Fig.2 shows the temperature and olivine system phase stability fields of models with h=1700 km (LC) and 2000 km (SC), respectively. Model LC has no pv+fp layer and develops small plumes in the lower part of the mantle. During the first ca. 1.43 Ga, convection is mostly separate in two layers defined by the ol-rw transition, with only sporadic cold downwellings crossing the transition; the upper layer cools much more than the lower. After that, there is a transition of a few hundred Ma, in which cold material from the upper layer massively breaks into the lower. Eventually, the lower layer has cooled, and whole-mantle convection prevails.

Model SC initially has a pv+fp layer, which seems to suppress the development of plumes in the beginning. Convection initially occurs mostly in two layers, although the separation is not as strong as in Model LC. As the pv layer shrinks, more cold material from above enters the lower layer, and eventually small, weak plumes form; there are intermittent appearances of rw+2fp+st at cold places at the bottom of the mantle, and pv and 2fp+st disappear at around 2.78 and 3 Ga, respectively.

In all models, the ringwoodite layer expands at the expense of the olivine and perovskite layers as cooling proceeds.

Fig.2: Models with h=1700 km and internal heating (upper two rows) and h=2000 km and no internal heating (lower two rows): temperature T, stability regions of olivine phases system phiol.
Movie (.avi, 4.57 MB) Movie (.avi, 2.42 MB)
Movie (.avi, 4.68 MB) Movie (.avi, 2.54 MB)

Acknowledgments

TR is supported by the NASA Planetary Geology and Geophysics Program through grant NNG04GI64G.

References

Bertka, C. M., and Y. Fei (1997), J. Geophys. Res., 102(B3), 5251-5264.
Fei, Y. et al. (2004), J. Geophys. Res., 109(B2), B02305, doi:10.1029/2003JB002562.
Gasparik, T. (2003), Phase Diagrams for Geoscientists, Springer.
Harder, H., and U. R. Christensen (1996), Nature, 380, 507-509.
Ito, E., and E. Takahashi (1989), J. Geophys. Res., 94(B8), 10637-10646.
Katsura, T., and E. Ito (1989), J. Geophys. Res., 94(B11), 15663-15670.
Katsura, T. et al. (2004), J. Geophys. Res., 109(B2), B02209, doi:10.1029/2003JB002438.
Saxena, S. K., N. Chatterjee, Y. Fei, and G. Shen (1993), Thermodynamic Data on Oxides and Silicates, Springer, Berlin/Heidelberg.
Tackley, P. J. (1998), in The Core-Mantle Boundary Region, edited by M. Gurnis, M. E. Wysession, E. Knittle, and B. A. Buffett, no.28 in Geodynamics Series, pp. 231-253, AGU, Washington.
Wänke, H., and G. Dreibus (1994), Phil. Trans. R. Soc. Lond., A 349, 285-293.
Back to my home page
Thomas Ruedas
Last modified: Mon Dec 17 16:05:07 EST 2007