Serpentine rheology and subduction zone earthquakes

There’s something about subduction zones. Maybe it’s the potential for magnitude 9+ earthquakes and major tsunamis... and for silent earthquakes, and slow slip events, and postseismic deformation. Maybe it’s the explosive volcanoes. Maybe it’s the accretion of stuff scraped off the downgoing plate. Maybe it’s my old fondness for minerals that look purple under a microscope. Maybe I’ve just been seduced by the name.

Anyway, although I’ve worked in rocks that have been down in subduction zones, I don’t have a very good idea about the nature of the faults that generate the immense earthquakes. I work on land, and on rocks that have made it to the earth’s surface; the subducting plate starts on the ocean floor, and goes down from there. So I don’t have a mental image of what’s involved – not like I have of, say, the Wasatch normal fault, or the Snake Range decollement.

An article in last week’s Science by Nadege Hilairet and co-authors, about experimental deformation of serpentine, gives me a better idea.

Minerals can deform in a number of different ways, depending on the temperature, the stress they experience, and the rate of deformation. At low temperatures, things deform elastically - they change shape in response to the stress until either the stress goes away, or they break. At higher temperatures, chemical bonds break and reform within the crystals, allowing the crystals to change shape without physically breaking. This can lead to some major differences in behavior when the rocks are stressed – at low temperatures, the release of elastic energy when rocks break and slip gives us earthquakes. At high temperatures, rocks behave more like Silly Putty – they’re elastic at short time scales (and can transmit seismic waves), but over longer times, they change shape at a rate that’s proportional to the stress they feel.

Different minerals deform by different mechanisms. Some change shape ridiculously easily – rock salt, for instance, flows so easily that it is used to equalize pressures in rock deformation experiments. Others, such as olivine, are strong even at quite high temperatures – that is one reason why the uppermost part of the mantle, the mantle lithosphere, is the strongest part of most tectonic plates.

And then there’s serpentine, the hydrous magnesium-iron-aluminum silicate that forms from adding water to mantle rock. It’s the mineral that dominates the state rock of California (and the rock found in the core across the San Andreas Fault). And it is, presumably, a mineral that forms in the mantle wedge above the subducting plate, as the downgoing plate loses its water. (Some subduction zones also have evidence that it really is there – seismic waves traveling through the overriding plate are slowed by a material that has the right seismic velocity and other elastic properties to be serpentinite.)

Serpentine is soft and slippery at surface conditions, so one might expect it to be weak in subduction zones, as well. But serpentine is brittle in low-pressure experiments (below 0.7 GPa, which would be equivalent to a depth in the neighborhood of 20 to 25 km) – near the surface, it breaks. The authors, however, managed to do deformation experiments at pressures of 1 to 4 GPa (equivalent to around 33 to 120 km – for a subduction zone beneath a continent, we’re talking about depths from the average base of continental crust to the base of the mantle lithosphere – though at a subduction zone, I’m not sure what is typically found at those depths).

They found that serpentine behaves differently at different pressures, presumably because different deformation mechanisms are at work. At both pressures, the behavior could be modeled as power-law creep. Power-law creep means that the strain rate is related to stress taken to some power n – that is, the higher the stress, the faster the strain rate, and the higher the value of n, the more sensitive strain rate is to increases in stress. Power-law creep is typical of deformation mechanisms involving movement of dislocations (defects in a crystal lattice). But at lower pressures, the stress exponent was higher – increasing stress really made deformation happen more quickly – which is typical of sliding along grain boundaries. At higher pressures, the behavior was closer to that of a viscous material (though still with viscosities that change with stress).

Ok, nice. So why, exactly, is this interesting? Well, for one thing, I usually think about the effect of temperature on deformation mechanisms, but I don’t usually see discussions of the role of pressure. Temperature’s important because intracrystalline deformation involves breaking chemical bonds, and higher temperatures generally make it easier to break and re-form bonds. In fact, I had always thought about the great depths of subduction-zone earthquakes as a function of the cold temperatures in the subducting slab. (The rock descends faster than heat can flow into it – among other things, that makes high pressure/low temperature metamorphic rocks possible.) So it’s interesting to see rheology controlled by pressure as well.

For another, the experiments show that, at depths in the range of approximately 30 to 100 km, serpentine is much weaker than most other likely minerals (pyroxene, plagioclase, and wet and dry olivine – the likely constitutents of oceanic crust and mantle). Is the likelihood of large earthquakes on subduction zones governed by the presence or absence of serpentine, rather than age of subducting crust or rate of plate movement? This would seem difficult to test, but if the seismic velocity and Poisson ratio of serpentinites are unusual, maybe tests are possible.

And what about those weird not-really-earthquake events that take place? Silent earthquakes, like the ones along the Cascadia subduction zone beneath Vancouver Island? Or afterslip – slow movement after a major earthquake, like occurred after the 2004 Boxing Day earthquake off Sumatra? (GPS data suggests that, during the month after the earthquake, slow slip equivalent to a M 8.7 earthquake occurred (Subarya et al., 2006).) The authors calculate that stresses in serpentine should relax on the same time scales described for slow earthquakes and afterslip, and imply that serpentine could explain some of the things that have been observed. There are some problems with that explanation, though. Modeling of afterslip following the Sumatra earthquake (Subarya et al., 2006), placed the slip in the same area as the initial earthquake, at depths shallower than 25 to 50 km, depending on the model they used to invert their ground deformation data. And Hilairet et al. cite studies saying that Sumatra has no evidence of serpentinization. So...the pieces don’t all fit together. Not perfectly.

This study, plus the recent study that found talc in the San Andreas Fault, are interesting because they tie together mineral behavior, rock mechanics, and seismology. The combination seems fruitful. I’ll be interested to see how the modelers who study triggered earthquakes work with these data.

References:

Hilairet, N., Reynard, B., Wang, Y., Daniel, I., Merkel, S., Nishiyama, N., and Petitgirard, S., 2007, High-pressure creep of serpentine, interseismic deformation, and initiation of subduction: Science, v. 318, p. 1910-1913.

Subarya, C., Chlieh, M., Prawirodirdjo, L., Avouac, J-P., Bock, Y., Sieh, K., Meltzner, A.J., Natawidjaja, D.H., and McCaffrey, R., 2006, Plate-boundary deformation associated with the great Sumatra-Andaman earthquake: Nature, v. 440, p. 46-51, doi:10.1038.