Roadmap | The review process | Home
   Gravitational instability

Continental magmatism caused by lithospheric gravitational instability

Linda T. Elkins-Tanton

Department of Geological Sciences, Brown University, Providence, RI, USA

There are three fundamental ways to cause the mantle to melt and produce magma:

  1. bring the material closer to the surface and allow it to melt through depressurization;
  2. change the composition of the material such that its melting point is lower; and
  3. raise the temperature of the material.

Process one, melting through adiabatic decompression, is the main process that operates at mid-ocean ridges. Beneath ridges, there are upwelling zones in which melting begins at about 50 km depth.

Process two, changing the composition of the material, is the main process that operates in subduction zones. Material in the mantle wedge is brought to higher pressures as it goes through the corner flow and begins to descend with the slab, so there is only a small region in which decompression can occur, before the corner of the wedge is reached. In subduction zones, melt is predominantly created by the addition of incompatible chemical species that rise off the slab, mainly water, sodium, and potassium. The addition of these oxides, which do not fit into common mantle materials, acts thermodynamically to reduce the temperature of melting of the mantle, sometimes by as much as 100°C or more (Hirschmann et al., 1999). In this situation, mantle material can melt without changing its pressure or temperature.

Process three, raising the temperature of a piece of mantle material, is a slow, low-volume process that can only operate in specialized circumstances where very hot material is brought into contact with colder material, and the colder material heats up conductively. This is an inefficient process that appears to have little bearing on the production of major volumes of primary volcanic magma on Earth.

Because depressurization and the addition of volatiles are the major processes that produce mantle melts on Earth, the production of large-volume magmatism on continents is puzzling. In general, continental lithosphere is thicker than oceanic lithosphere, and so upwelling material beneath a continent cannot depressurize to the same extent it can beneath oceanic lithosphere: its ascent is blocked by the bottom of the thick continental lithosphere (Figure 1). Researchers who study mantle plumes under thick lithosphere find that very high temperatures are required to allow an upwelling plume to cross its solidus before it reaches the bottom of the thick lithosphere and cannot rise further. However, there is little petrologic evidence for very high mantle temperatures. Kinzler & Grove (1992) found that ambient mantle temperatures between 1315°C and 1475°C can explain all geochemical variation at mid-ocean ridges. Kojitano & Akaogi (1997) found, based on calorimetric melting studies, that MORBs erupting at 1200°C come from mantle at 1260°C, even lower than the temperature concluded by Kinzler & Grove (1992).

Figure 1. Thermal model for melting under continents. Approximate wet and dry solidii are shown for both granite (a proxy for crust) and peridotite (a proxy for both lower lithosphere and mantle). An adiabatic geotherm is shown in the mantle, changing to a conductive geotherm in the lithosphere. a) With a mantle potential temperature of 1300°C no melt is produced under the lithosphere. b) the lithosphere has been thinned by half, and in both the upper mantle and the lower crust ambient temperatures exceed the solidii, creating melt. a) & b) demonstrate the difficulty of producing mantle melt under thick lithosphere. Click on images for enlargement.

Though some researchers have proposed that the mantle is hotter beneath continents, there is presently no compelling evidence for this. Mantle with a potential temperature of 1475°C will melt through decompression if it is raised to a depth of about 115 km. Below 115 km, the hydrostatic pressure of the mantle prevents melting. Thus we have the problem of how to depressurize mantle beneath continents, where the lithosphere is thought to be 150 – 250 km thick. It should also be noted that numerical modeling indicates that even very hot mantle plumes do not erode the bottom of lithosphere (Moore et al., 1999), and so cannot produce a longer melting column by removing part of the lid. In addition to the problem of depressurizing mantle beneath continents, there is no ready source there of volatiles that could be added to the mantle. Other processes must therefore be sought to produce continental magmatism in the absence of subduction.

The removal of the lower lithosphere by gravitational instability, is a process that can produce melt under continents. By removing the lower lithosphere, mantle can rise higher and depressurize more, producing more melt. Why, though, would the lower lithosphere delaminate? There are two conditions that must be met before instability can occur:

  1. the lower lithosphere has to be unstable gravitationally: that is, it has to be denser than the material underlying it, and thus be inclined to sink. The lower lithosphere can be denser because of an intrinsic compositional difference, such as having buoyant mantle residuum beneath it, or it can become denser through cooling, or by consisting of denser minerals;
  2. viscosity must be lowered. For lithosphere to detach and begin to flow downward under the influence of gravity, its viscosity must be low enough to allow flow. Though some researchers have suggested brittle delamination, there is little understanding of that proposed process (Bird, 1979; 1988; Ed: Click here for a webpage explaining these mechanisms). Here, the assumed process of lithosphere thninng is Rayleigh-Taylor instability: fluid flow under the influence of a density instability. The flowing material is still completely solid, but it behaves according to the rules of fluid dynamics on a geologic time scale.

The control of viscosity on gravitational instability has been studied by a number of researchers, for problems concerning both the Earth and the Moon (e.g., Hess & Parmentier, 1995; Conrad & Molnar, 1997; Houseman & Molnar, 1997; Morency et al., 2002; Elkins-Tanton et al., 2002). In the simplest case, following the results of Hess & Parmentier (1995), if fluid density decreases linearly with depth (a density inversion is required for a Rayleigh-Taylor instability to form) and the viscosities of all the materials are equal and constant, the time scale for a Rayleigh-Taylor instability to form can be expressed as:

The initiation of an instability is accelerated by lower viscosity and by a thicker unstable layer. While d is the height of the instability, the width of the instability can be given as:


In the case of gravitational instability of continental lithosphere over hot mantle, the viscosities of the two materials will be different. In this case the time for onset of an instability can be given as:

(Hess & Parmentier, 1995; Whitehead, 1988). Though the viscosity of the mantle contributes more to the time of instability than the viscosity of the lower lithosphere does, the difference in their viscosities may be as much as six orders of magnitude, and so a stiff lower lithosphere significantly inhibits the formation of an instability.

In our research, we find that the viscosity of the lower lithosphere is the limiting factor in gravitational instability. One simple scenario for gravitational instability occurred in the Sierra Nevada about 3 million years ago, when we think the thick root of the Sierra Nevada batholith delaminated, causing a pulse of high-potassium magmatism (Elkins-Tanton & Grove, 2003). A plate had been subducting beneath the Sierras for tens of millions of years, injecting subduction fluids and hot silicate melts into the lower lithosphere. This had the dual effect of heating the lower lithosphere and hydrating it, both of which lower its viscosity (Hirth & Kohlstedt, 1996). These are ideal conditions for gravitational instability, and in fact gravitational instability is supported both by studies of mantle xenoliths (e.g., Lee et al., 2001; Ducea & Saleeby, 1998) and by seismic studies of mantle structure under the Sierras (Jones & Phinney, 1998).

A second case where continental lithospheric gravitational instability may have played a part is in the eruption of the Siberian flood basalts. Though the standard model for the Siberian flood basalts is the rise of a huge, superheated plume from the lower mantle, some important geological evidence contradicts the pure plume model (Czamanske et al., 1988; see also Siberia page). In the plume model, the plume head is highly buoyant because of its heat, and when it reaches the bottom of the lithosphere it lifts it up into a dome while it melts (Richards et al., 1989). In Siberia, evidence of terrestrial fossils combined with pillow basalts show that the surface of Siberia was actually subsiding, keeping the eruptions just beneath sea level, for the first kilometer of the flood basalt stack. This prompted us to suggest a scenario in which a weak plume rises beneath Siberia, melts only a small amount, and injects melt into the lower lithosphere (Elkins-Tanton & Hager, 2000). This melt freezes as eclogite in the lower mantle, increasing its density, and the latent heat of freezing heats the lower lithosphere, lowering its viscosity and allowing gravitational instability. When the lower lithosphere delaminates, its traction pulls the lithosphere down and creates topographic subsidence, consistent with the geological evidence. Our more recent results, as yet unpublished, show that at least part of the Siberian flood basalt magmas originated from mantle melting at about 100 km depth, requiring a lithosphere thinner than this. It might have been expected that the Siberian craton was thicker than 90 km because of its great age, and so our results are consistent with gravitational instability.

How does gravitational instability cause magmatism? First, when the lower lithosphere delaminates, the mantle is immediately sucked into the void, and may melt adiabatically. The dome left in the lithosphere has a horizontal temperature gradient across its edges, from the hot asthenosphere now inhabiting the center, to the cool adjacent lithosphere that did not delaminate. Horizontal temperature gradients drive convection, and the dome is soon filled with an umbrella-shaped convection cell, pulling hot material up from depth. This material may also melt, since it can now rise to higher levels where the pressure is lower (Figure 2).

Figure 2. Schematic model for loss of the lithosphere by a Rayleigh-Taylor instability. As the instability falls, asthenosphere is sucked into the resulting lithospheric dome, and may melt adiabatically. A convection cell forms in the lithospheric dome due to horizontal temperature gradients, and may also cause adiabatic melting. As the instability falls, if it is hydrous, it may dewater, triggering melting of the mantle or of itself.

Second, the delaminating piece itself may induce magmatism, by dewatering as it falls. This is the subject of our current research. If the delaminating piece is hydrous, then as it falls through the mantle and heats conductively, the water will be forced out of it, just as hydrous fluids rise from subducting slabs as slabs move out of the stability regimes of various low-pressure and low-temperature hydrous minerals. It is easy to demonstrate that heating is fast and efficient enough to dewater a falling instability. This hydrous fluid can have two effects: it can enter the overlying hot mantle and allow it to melt, or it can allow parts of the delaminating material itself to melt (Figure 2). This provides a number of geochemically different source regions that can produce melt. Numerical models indicate that, over a range of lithospheric thicknesses and mantle potential temperatures, these processes can create a wide range of melt volumes, even up to the size of a flood basalt province (Figure 3).

Figure 3: Numerical experimental results for delaminating hot, heavy lower lithosphere in the Siberian flood basalt province (see also Siberia page). The image shows a cross-section from the top of the crust down into the mantle. The left boundary is an axis of symmetry, passing through the center of the Rayleigh-Taylor instability. Length scales on the vertical and horizontal axes are non-dimensionalized, but the scaled model box represents 200 km in width and height. Arrows represent the velocity vector field, colors represent temperature, with white being the mantle potential temperature of 1480°C. Solid contours mark degree of melt infiltration into the lower lithosphere, where it causes both a density instability and lower viscosity. The falling instability is already heated almost to ambient mantle temperatures, indicating the likelihood of dewatering if hydrous phases are present.

Any process that causes a density instability and lowers the viscosity in the lower lithosphere, including subduction or the existence of any upwelling creating melt beneath the lithosphere, thus can trigger gravitational instability. Gravitational instability itself causes three processes that may lead to mantle melting under continental lithosphere:

  1. convection in the resulting lithospheric dome,
  2. hydrous fluids entering the asthenosphere from the falling lithosphere, and
  3. melting of the falling piece of lithosphere.

These processes can operate on small or large scales, controlled in part by the thickness of the unstable layer in the lithosphere, and modeling indicates that a wide range of mantle melt volumes can be produced.


  • Bird P. (1979) Continental delamination and the Colorado Plateau. J. Geophys. Res., 84, 7561-7571.
  • Bird P. (1988) Formation of the Rocky Mountains, western United States: A continuum computer model. Science, 239, 1501-1507.
  • Conrad, C.P. and P. Molnar (1997) The growth of Rayleigh-Taylor-type instabilities in the lithosphere for various rheological and density structures. Geophys. J. Int., 129, 95-112.
  • Czamanske, G.K., A.B. Gurevitch, V. Fedorenko, and O. Simonov (1988) Demise of the Siberian plume: Paleogeographic and paleotectonic reconstruction from the prevolcanic and volcanic record, North-central Siberia, Int. Geology Rev., 40, 95-115.
  • Czamanske, G.K. and V.A. Fedorenko, Siberia.html
  • Ducea M. and J. Saleeby (1998) A case for delamination of the deep batholithic crust beneath the Sierra Nevada, California, Int. Geology Rev., 40, 78-93.
  • Elkins-Tanton L.T. and B.H. Hager (2000) Melt intrusion as a trigger for lithospheric foundering and the eruption of the Siberian flood basalt, Geophys. Res. Lett., 27, 3937-3940.
  • Elkins-Tanton L.T., J. A. Van Orman, B. H. Hager, and T. L. Grove (2002) Reexamination of the lunar magma ocean cumulate overturn hypothesis: Melting or mixing is required, Earth Planet. Sci. Lett., 196, 249-259.
  • Elkins-Tanton L.T. and T.L. Grove (2003) Evidence for deep melting of hydrous, metasomatized mantle: Pliocene high potassium magmas from the Sierra Nevadas, J. Geophys. Res., 108, 2350, DOI 10.1029/2002JB002168.
  • Hess P.C. and E.M. Parmentier (1995) A model for the thermal and chemical evolution of the moon’s interior: Implications for the onset of mare volcanism. Earth Planet. Sci. Lett., 134, 501-514.
  • Hirschmannn, M.M., P.D. Asimow, M.S. Ghiorso, and E.M. Stolper (1999) Calculation of peridotite partial melting from thermodynamic models of minerals and melts. III. Controls on isobaric melt production and the effect of water on melt production, J. Pet., 40, 831-851.
  • Hirth, G. and D.L. Kohlstedt (1996) Water in the oceanic upper mantle: implications for rheology, melt extraction and the evolution of the lithosphere, Earth Planet. Sci. Lett., 144, 93-108.
  • Houseman G.A. and P.A. Molnar (1997) Gravitational (Rayleigh-Taylor) instability of a layer with non-linear viscosity and convective thinning of continental lithosphere. Geophys. J. Int., 128, 125-150.
  • Jones C.H. and R.A. Phinney (1998) Seismic structure of the lithosphere from teleseismic converted arrivals observed at small arrays in the southern Sierra Nevada and vicinity, California. J. Geophys. Res., 103, 10,065 - 10,090.
  • Kinzler R.J. and T.L. Grove (1992) Primary magmas of mid-ocean ridge basalts 2. Applications. J. Geophys. Res., 97, 6907 - 6926.
  • Kojiaki H. and M. Akoagi (1997) Melting enthalpies of mantle peridotite; calorimetric determinations in the system, CaO-MgO-Al2O3-SiO2 and application to magma generation. Earth Planet. Sci. Lett., 153, 209 - 222.
  • Lee C.-T., R.L. Rudnick, and G.H. Brimhall Jr. (2001) Deep lithospheric dynamics beneath the Sierra Nevada during the Mesozoic and Cenozoic as inferred from xenolith petrology. Geochem. Geophys. Geosys., 2.
  • Moore, W.B, G.Schubert, and P.J. Tackley (1999) The role of rheology in lithospheric thinning by mantle plumes, Geophys. Res. Lett., 26, 1073-1076.
  • Morency C., M.-P. Doin and C. Dumoulin (2002) Convective destabilization of a thickened continental lithosphere. Earth Planet. Sci. Lett., 202, 303-320.
  • Richards, M.A., R.A. Duncan, and V.E. Courtillot (1989) Flood basalts and hot-spot tracks: Plume heads and tails, Science, 246, 103-107.
  • Whitehead J.A. (1988) Fluid models of geological hotspots. Ann. Rev. Fluid Mech., 20, 61-87.

last updated 22nd November, 2006