The influence of olivine metastability on the dynamics of subduction

What follows is the abstract, some figures and the conclusions of a recent paper by Harro Schmeling, Ralf Monz and David C. Rubie (see Schmeling's publication list). But see also the pages of M. Tetzlaff and R. Monz.

Both seismological observations and mineral physical experiments suggest that old and cold subducting slabs might contain a wedge of metastable olivine ("MO"). Dynamically consistent models of subducting slabs in which the kinetics of the olivine-spinel-transition is treated in a simplified way, are carried out to evaluate the buoyancy effect of MO on the subduction velocity. Assuming slab thicknesses according to the cooling half-space model, we find that MO forms in slabs with ages greater than approximately 70 Myr, and that a significant slowing down of subduction velocity  occurs for ages greater than 100 Myr. Because a decrease in the subduction velocity for old lithospheric ages is not observed, our models suggest that the amount of MO in slabs is less than approximately 5000 km2 in cross section, and that for old lithosphere the cooling half-space model is not applicable.

We came to the following conclusions:

For the first time dynamically consistent models of subduction of lithospheric slabs have been calculated, in which buoyancy effects of the olivine-spinel phase change and of metastable olivine (MO) have been included. Assuming a cooling half-space model for the lithosphere, we have tested the subduction behaviour of slabs of different thicknesses or ages. Old and thick slabs develop large amounts of buoyant MO which retard subduction due to the so-called parachute effect [2]. From our models we are able to quantify this effect and observe a significant reduction of subduction velocities for lithospheric ages greater than 90 Myr.  This reduction is associated with relatively large amounts of MO (> 5000 km2  in cross section). A further effect of this reduction is to restrict the thermal parameter of subducting plates to values of or below roughly 10,000 km. We note that if plates do not continue to cool beyond ages greater than 90 Myr (i.e. if the cooling plate model is applicable) no significant buoyancy effect of MO is expected.
   Because a significant reduction in subduction velocity is not observed for increasing plate ages in the western Pacific, we conclude from our models that MO may not be present in large volumes and the parachute effect is probably not very important. Smaller amounts (< 5000 km2  in cross section) may well be present and may be responsible for deep focus earthquakes but they do not lead to a first order observable effect in subduction dynamics. To fully resolve this question high resolution models are necessary which also take into account the spinel-perovskite transition and the penetration of slabs into the lower mantle.

Here some Figures from our paper:

Simplified phase diagram:

Figure 1.Simplified disequilibrium (kinetic) phase diagram of the olivine   spinel transition as used in the models (c.f. [4], Fig. 9). Lines of 1% and 99% spinel fraction are shown and indicate the start and completion of the olivine  spinel phase transformation along depth-temperature paths characteristic of subduction zones [4]. The vertical segments of the lines show the conditions required for transformation at low temperature and are consistent with metastable olivine persisting at T<600°C. At higher temperatures (>700°C), transformation occurs relatively close to the equilibrium phase boundary.  The spinel fraction is assumed to increase linearly between the 1% and 99% lines.

Self consistent subduction of a slab with a metastable olivine wedge:

Figure 2. Evolution of a subducting slab with time from top to bottom. Initial thickness of slab was chosen corresponding to a 111 Myr old lithosphere. Colors and a few contour lines represent temperature. The dense set of contour lines between 400 and 500 km depth represent the transition from olivine to spinel, and give the fraction of spinel.

Subduction velocity

Figure 3. Absolute value of  velocity of a tracer situated within the subducting slab. Initial position of the tracer is 740 km left of the front of the slab. The full curve gives the velocity of the slab in which MO is present, the dashed curve represents a slab of same age and thickness, but without MO.

Metastable olivine wedge for slabs of different thicknesses

Figure 4. Close up´s of subducting slabs of different initial ages during the second velocity peak (c.f. Fig. 3). The initial ages  and the model times of the snap shots are 33  and 30.1 Myr (top), 70 and  30.3 Myr (middle), and 111 and 34.3 Myr (bottom), respectively.

Amount of metastable olivine

Figure 5. The amount of metastable olivine in km2 for the different models as a function of time. The ages beside the curves give the initial age of the lithosphere before subduction.

Slab velocities with and without metastable olivine

Figure 6. Slab velocity during subduction of slabs with different ages at the trench. Squares represent models in which metastable olivine may be formed, diamonds show models in which metastable olivine is not allowed to form. The velocity is given as the absolute value of velocity of a tracer within the downgoing part of the slab. The numbers beside the symbols give the thermal parameter of the slab (vertical velocity times age at trench) in units of 1000 km. The age at trench is taken as  the initial slab age plus the model time of the second velocity peak (c.f. Fig. 3). The initial slab ages of the models are (corresponding to the symbols from left to right): 5, 33, 49,70, 111, 131 Myr.

Back to H. Schmeling's home page

Last modified on: October, 14, 1998