By Kathryn Hayward

In my last post I wrote about my adventures in Paris working at the École Normale Supérieure (ENS) using state-of-the-art laboratory experiments to explore how fluids influence fault rupture. In this contribution I will tell you about the five weeks that I spent in Rome working at the National Institute of Volcanology and Geophysics (INVG) with Professor Giulio Di Toro, Dr Elena Spagnuolo and Dr Francios Passelegue, focusing on the initiation of earthquake slip.

So what controls whether a fault rupture grows to become a very large, damaging earthquake, or dies out as a small, inconsequential event? Well, this is thought to relate to the material properties of the fault surface as the velocity increases. If the frictional strength of the fault weakens with increasing slip, a large earthquake is more likely to occur. A reduction in fault strength during slip is referred to as dynamic weakening.

But as we know, in continental crust earthquakes typically nucleate at depths between 5 and 20 km, and thus direct observation of faults during an earthquake is impossible. Consequently, what we know about fault weakening processes is obtained through experiments.

To date, most experimental work examining weakening processes during seismic slip has been undertaken using a rotary shear apparatus. This machine rotates two cylindrical rock specimens against each other at high slip velocities. Experiments have led to the identification of a number of different mechanisms that could cause the dynamic weakening. Unfortunately most rotary shear experiments are undertaken unconfined and at low normal stresses (normal stress is the stress that acts perpendicular to a surface, with increasing normal stress having a clamping effect). These experimental conditions are not realistic of estimated crustal faulting conditions and results in uncertainty in the extrapolation of results.

More recently, using Rig 1, the triaxial deformation apparatus here at RSES, I have combined realistic normal stresses with small displacement, rapid slip events to examine processes operating during the initial stages of fault weakening. We have shown that the onset of microstructural changes (such as amorphization and frictional melting, see Fig. 1) occur at slip distances that are up to several orders of magnitude smaller than previously recognised. However, friction is measured indirectly and slip is limited to a maximum of 2 mm, making it difficult to extrapolate our results to the larger slips of damaging earthquakes. To advance our understanding of earthquake nucleation and propagation, we need observations of the processes occurring at the onset of dynamic weakening during the first centimetres of seismic slip at realistic crustal stress conditions.

BIS010 30 oppss 27
Figure 1: Frictional melt formed on the slip surface of a small displacement experiment undertaken in Rig 1, the triaxial apparatus at RSES.

With this in mind I spent 5 weeks in the National Institute of Geophysics and Volcanology (INGV) in Rome undertaking experiments using the SHIVA apparatus (Slow to High Velocity Apparatus, see Fig. 2). This is a rotary shear apparatus but is capable of accelerations of up to 21 ms-2 at normal stresses up to ~30 MPa (this is an order of magnitude higher than many of the classic experiments of this type) and, using a recently developed configuration, is capable of undertaking confined experiments.

20170703_141632
Figure 2: The Slow to High Velocity Apparatus (SHIVA) that I used during my visit to INGV in Rome.

During my stay in Rome I undertook nearly 70 experiments on a number of different quartz based materials. Compared with the jacketed and enclosed triaxial experiments, the unconfined rotary shear experiments gave a wonderful visual understanding of heat generation during slip. The following movie was taken of an experiment slipped for 0.5 m at a peak velocity of 3 ms-1 and with a clamping stress of 20 MPa. During this relatively short slip, hotspots rapidly developed, spreading until the entire fault glowed white hot. Following the experiment we found lots of evidence for the formation of quenched melt (glass).

To me this was compelling confirmation of processes involved in the development of microstructures during our triaxial experiments. Skepticism is often the first reaction when people see microstructures that indicate that the quartz interfaces have melted during slip events that are only a few tens of microns in length. This doubt is born from two factors: first, is the huge temperature increase that is needed to melt quartz (around 1700ºC) and second, that the results of older, slow slip, low normal stress – large displacement experiments suggested that melting was unlikely, instead forming a low friction interface that would inhibit the generation of heat. These experiments showed that mechanical work could destroy the quartz crystal lattice structure resulting in the formation of amorphous silica. It was hypothesized that the highly unstable amorphous material would react with any available water molecules forming a low strength ‘gel’. However, the current experiments have unequivocally shown that melting can and does occur, and that potentially different mechanisms are activated depending on the normal stresses acting on the fault.

In the current era of defining national benefit and justifying research, how is this research relevant more broadly to society? The need for a better understanding of earthquake nucleation and slip processes was made very apparent during my final weekend in Rome. At this time I had the opportunity to join a field trip
held for national and international scientists, to visit the epicentral locations of the three largest earthquakes (Mw > 6.0) to hit the Central Apennines over the past 20 years.

Apennines1
A village destroyed in the 24 August 2016 Mw. 6.0 earthquake located between Amatrice and Norcia. This rupture reactivated nearly 20 km of normal fault segments in approximately 7 seconds. The directivity of the rupture contributed to ground shaking.
Apennines2
Nearly two metres of vertical displacement on a fault scarp following the October 2016 Mw 6.6 earthquake near Colli Alti e Bassi.

The tectonics of this central part of Italy is complex, with various rock types and structures developed during and subsequent to the collision of the African and European plates. Currently it is a region of uplift and extension with numerous NWSE oriented faults. The normal faults that are active, occur in dominantly limestone sequences and have estimated recurrence intervals of 350 years for Mw > 6.5.

The devastation caused by the three earthquakes is very real. More than 300 people were killed and >1500 injured, 125,000 were forced to flee their homes and many of the beautiful historic hill towns have been destroyed. The estimated damage bill from these three earthquakes amounts to more than A$18 billion.

Coming face-to-face with the devastation and impact on people’s lives is certainly a strong motivation for continuing to work towards understanding the processes by which earthquakes nucleate and propagate. During my trip I was able to perform experiments and learn new techniques, to work closely with collaborators and to see first hand the impact of earthquakes on the lives of people. This trip would not have been possible without financial help from the Research School of Earth Sciences and the Mervyn and Katalin Paterson travel grant endowment – many thanks for your support!