Tag Archives: Mantle convection

How does plate tectonics work?

Read about a new computer model that charts the co-evolution of the mantle and lithosphere, i.e. the linkages between deep convection and plate tectonics.

Still from a movie of simulated breakup of a supercontinent, in bland blue-grey, showing what happens at the surface (left) and, at the same time, in the mantle (right): note the influence of rising plumes (credit: Nicolas Coltice)

Tectonics of the early Earth

Tectonics on any rocky planet is an expression of the way heat is transferred from its deep interior to the surface to be lost by radiation to outer space. Radiative heat loss is vastly more efficient than either conduction or convection since the power emitted by a body is proportion to the fourth power of its absolute temperature. Unless it is superheated from outside by its star, a planet cannot stay molten at its surface for long because cooling by radiation releases all of the heat that makes its way to the surface.  Any football supporter who has rushed to get a microwaved pie at half time will have learned this quickly: a cool crust can hide a damagingly hot centre.

Thermal power is delivered to a planet’s surface by convection deep down and conduction nearer the surface because rocks, both solid and molten, are almost opaque to radiation. The vigour of the outward flow of heat might seem to be related mainly to the amount of internal heat but it is also governed by limits imposed by temperature on the form of convection. Of the Inner Planets only Earth shows surface signs of deep convection in the form of plate tectonics driven mainly by the pull exerted by steep subduction of cool, dense slabs of old oceanic lithosphere. Only Jupiter’s moon Io shows comparable surface signs of inner dynamics, but in the form of immense volcanoes rather than lateral movements of slabs. Io has about 40 times the surface heat flow of Earth, thanks largely to huge tidal forces imposed by Jupiter. So it seems that a different mode of convection is needed to shift the tidal heat production; similar in many ways to Earth’s relatively puny and isolated hot spots and mantle plumes.

Most of the yellow and orange hues of Io are d...

An analogy for the early Earth, Jupiter’s moon Io is speckled with large active volcanoes; signs of vigorous internal heat transport but not of plate tectonics. Its colour is dominated by various forms of sulfur rather than mafic igneous rocks. (credit: Wikipedia)

Shortly after Earth’s accretion it would have contained far more heat than now: gravitational energy of accretion itself; greater tidal heating from a close Moon and up to five times more from internal radioactive decay. The time at which plate tectonics can be deduced from evidence in ancient rocks has been disputed since the 1970s, but now an approach inspired by Io’s behaviour approaches the issue from the opposite direction: what might have been the mode of Earth’s heat transport shortly after accretion (Moore, W.B. & Webb, A.A.G. 2013. Heat-pipe Earth. Nature, v.  501, p. 501-505). The two American geophysicists modelled Rayleigh-Bénard convection – multicelled convection akin to that of the ‘heat pipes’ inside Io – for a range of possible thermal conditions in the Hadean. The modelled planet, dominated by volcanic centres turned out to have some surprising properties.

The sheer efficiency of heat-pipe dominated heat transfer and radiative heat lost results in development of a thick cold lithosphere between the pipes, that advects surface material downwards. Decreasing the heat sources results in a ‘flip’ to convection very like plate tectonics. In itself, this notion of sudden shift from Rayleigh-Bénard convection to plate tectonics is not new – several Archaean specialists, including me, debated this in the late 1970s – but the convincing modelling is. The authors also assemble a plausible list of evidence for it from the Archaean geological record: the presence in pre- 3.2 Ga greenstone belts of abundant ultramafic lavas marking high fractions of mantle melting; the dome-trough structure of granite-greenstone terrains; granitic magmas formed by melting of wet mafic rocks at around 45 km depth, extending back to second-hand evidence from Hadean zircons preserved in much younger rocks. They dwell on the oldest sizeable terranes in West Greenland (the Itsaq gneiss complex), South Africa and Western Australia (Barberton and the Pilbara) as a plausible and tangible products of ‘heat-pipe’ tectonics. They suggest that the transition to plate-tectonic dominance was around 3.2 Ga, yet ‘heat pipes’ remain to the present in the form of plumes so nicely defined in the preceding item Mantle structures beneath the central Pacific.

In the mantle wet may not imply soft

For half a century the Earth’s planetary dynamism – plate movements, mantle convection and so on – has been ascribed to its abundance of water. Experiments on the ductility of quartz seemed to show that it became much weaker under hydrous conditions, and that was assumed to hold for all common silicates, a view backed up by experiments that deformed minerals under varying conditions. It was widely believed that even a few parts per million in a rock at depth would weaken it by orders of magnitude, a view that increasingly dominated theoretical tectonics on scales up to the whole lithosphere and at different mantle depths. Strangely, the founding assertion was not followed up with more detailed and sophisticated work until the last year or so. Though rarely seen in bulk, the dominant mineral in the mantle is olivine and that is likely to be a major control over ductility at depth, in plumes and other kinds of convection.

Peridotite xenoliths in basalt—olivines are li...

Peridotite xenoliths —olivines are light green crystals, pyroxenes are darker. (credit: Wikipedia)

Experimental work at the temperatures and pressures of the mantle has never been easy, and that becomes worse the more realistic the mineral composition of the materials being investigated. High-T, high-P research tends to focus on as few variables as possible: one mineral and one variable other than P and T is the norm. This applies to the latest research (Fei, H. et al. 2013. Small effect of water on upper mantle rheology based on silicon self-diffusion coefficients. Nature, v. 498, p. 213-215) but the measurements are of the rate at which silicon atoms diffuse through olivine molecules rather than direct measurements of strain. The justification for this approach is that one of the dominant processes involved in plastic deformation is a form of structural creep in which atoms diffuse through molecules in response to stress – the other is ‘dislocation creep’ achieved by the migration of structural defects in the atomic lattice.

Contrary to all expectations, changing the availability of water by 4 to 5 orders of magnitude changed silicon diffusion by no more than one order. If confirmed this presents major puzzles concerning Earth’s mantle and lithosphere dynamics. For instance, the weak zone of the asthenosphere cannot be a response to water and nor can the relative immobility of hotspots. Confirmation is absolutely central, in the sense of repeating Fei et al.’s experiments and also extending the methods to other olivine compositions – magnesium-rich forsterite was used, whereas natural olivines are solid solutions of Mg- and Fe-rich end members – and to materials more representative of the mantle, e.g.  olivine plus pyroxene as a minimum (Brodholt, J. 2013. Water may be a damp squib. Nature, v. 498, p. 18-182)