Look at the depth distribution of earthquakes on Earth (Fig. 1):
In general, earthquakes are located at the boundaries between tectonic plates. Shallow earthquakes (< 60 km) happen at all plate boundary types, but intermediate (60-300 km) and deep (> 300 km) earthquakes mainly occur in subduction zones, where one plate moves beneath another. Because these earthquakes are located either within the subducting plate or between the two plates, they get deeper and deeper the further they are from the surface trace of the plate boundary. Because the plate located west of South-America moves towards the east and is subducted under South-America (Fig. 2), the earthquakes on the west coast of South-America get deeper from west to east (Figs 1, 2).
They get so deep, actually, that it is surprising that they exist at all: as rock surfaces are never perfectly flat, they also have to move apart a little bit in order to move against each other and produce earthquakes. This is not a problem for most earthquakes, which happen relatively close to the Earth’s surface. Deep down, however, the rocks are squeezed together so tightly by all the others rocks surrounding them that they should be unable to produce earthquakes. Nevertheless, earthquakes at these depths are detected by seismographs. Why do they happen?
The deepest earthquakes (below 300 km) are generally thought to be related to a change in the crystal structure of olivine, the most abundant mineral in the Earth’s mantle, which becomes denser at a certain pressure. While there is still some mystery about how exactly this works, I want to focus on intermediate-depth earthquakes (purple in figures 1 and 2) here. There are two main theories aiming to explain how they form: dehydration embrittlement and thermal shear instabilities.
- Dehydration embrittlement can happen when a mineral that contains water in its crystal structure loses this water at higher temperatures. Serpentine, for example, a water-bearing mineral, breaks down to olivine and water (a process called deserpentinisation). Olivine and water have a larger volume than serpentine. Therefore, the fluid will exert a pressure on the rocks surrounding it. If the fluids press against the rock from the inside, this cancels out some of the pressure coming from the outside. And this, in turn, enables the rocks to move so that they can produce an earthquake. Many researchers think that serpentine is too weak for this to work; it will deform so easily that earthquakes will not happen. For an earthquake to happen, it is necessary that stress can build up in the rock, which is then suddenly released in an earthquake. If you apply pressure on one end of a matchstick, as if you were trying to bend it, nothing will happen at first. But if you increase the pressure, the matchstick will break. If you do the same with a strip of chewing gum, the gum will bend immediately and never break. Pure serpentinite often seems to behave like the strip of gum.
- Thermal shear instability means that a slow movement in a fine-grained zone of a rock produces enough heat to reduce the strength of this zone, which leads to a faster movement and ultimately an earthquake. A problem with this theory is that some fault movement is necessary to initiate this process, and how that starts is unclear.
There are many arguments for and against both theories, and both of them have been investigated experimentally, but the results are not conclusive. So what we really need are natural examples. Unfortunately, there are not very many that we know of, but now we have one more: we found evidence of deserpentinisation-induced earthquakes on the island of Leka in Norway. The whole island is an ophiolite; it used to be part of the oceanic crust and mantle and has moved onto the continent some 450 million years ago.
Dunites, rocks consisting almost exclusively of olivine, are very common in some parts of the island. They are cut by veins, which also consist of olivine (Fig. 3). From the composition of the olivine in the veins we concluded that it formed by deserpentinisation. This implies that water initially infiltrated the original dunite along fractures, and reacted with the olivine around the fractures to form serpentine (Figs 4.1-2). Later, the temperatures increased, serpentine became unstable and reacted back to olivine and water (Fig. 4.3). This in itself is not very exciting, but what is exciting is that faults are located in many of these veins (Fig. 3). These faults are extremely sharp (Fig. 5), which does not necessarily mean that they were related to earthquakes, but it does make it likely.
A much better argument for earthquakes is the size distribution of the faults. We measured the displacement of chromite layers as in figure 3 on more than 500 faults. The size distribution of the displacement, that is how many small displacements there are compared to large displacements, looks just like the ones produced by earthquakes.
Additionally, the olivine next to the faults is very fine-grained, but has not moved: it was pulverised during the faulting (Figs 4.4, 5), which is also characteristic for earthquakes.
These rocks from Leka show that deserpentinisation can cause embrittlement and earthquakes. What is special about them is that small veins of serpentinite within a dunite were dehydrated, instead of a pure serpentinite, which is typically the case in experiments. Dunite is much stronger than serpentinite, and can withstand higher stress before it breaks (remember the matchstick and the gum). Therefore, we think that the amount and distribution of serpentine are very important for the question whether or not the dehydration of serpentinite can cause intermediate-depth earthquakes.
If you are interested in the full story, read our new article about the deserpentinisation and deformation on Leka.