Imaging the precursors to rupture in rocks and their role in earthquake nucleation


The dynamics of deformation in rocks is a critical element in earthquakes. Direct high-resolution information on this process is obtained by imaging microfractures as they evolve in laboratory fracturing experiments. The results demonstrate the existence of precursors prior to catastrophic earthquake slip.

  • Share

The dynamics of rock brittle deformation prior to macroscopic instability controls the nucleation phase of an earthquake and the accumulation of faults in the Earth’s crust. Prior to failure, a rock may deform by the formation of microfractures that grow and coalesce together (Figure 1). When a large enough microfracture cluster has formed, a subtle change in stress conditions leads to a rupture instability propagating through the entire sample and the formation of an earthquake. The route to failure has now been imaged in centimetre-scale crystalline rock specimens using in situ dynamic X-ray microtomography. The evolution of microfracture opening, closing, and shearing, leading to system-size failure, has been documented with unprecedented resolution. How brittle failure results from the coupling between microfracture growth, opening, coalescence and closing has been quantified, as well as left-lateral shear displacement and right-lateral shear along microfractures.

Failure in rock

Figure 1. Failure in rock occurs by the formation of microfractures that grow, interact elastically and evolve toward a shear fault where an earthquake may be nucleated. This process of earthquake nucleation has been imaged using dynamic X-ray microtomography and the dynamics of microfracture formation has been characterised. Rupture is characterised by a power-law evolution of the microfractures that show similar statistical behaviour to what is observed in various physical systems characterised by a phase transition.

The Hades deformation rig used in this research permits a centimetre-scale rock sample to be loaded at conditions of pressure and temperature similar to those at several kilometres depth below the Earth’s surface (Figure 2, top). The samples were imaged in situ using X-ray microtomography with 6.5 micrometres spatial resolution and 1.5 minutes time resolution [1]. This facility is installed on beamline ID19 and it is available to the scientific community. The unique dynamic four-dimensional (space + time) data acquired with the Hades rig allows rocks to be imaged as they undergo geodynamic transformations such as ductile and brittle failure [2] and/or metamorphic transformations [3]. Data processing techniques enable the visualisation of voids in the rock, including the formation of microfractures, and the growth or dissolution of new minerals by using the adsorption and phase contrast properties of X-rays. It is also possible to extract relevant in situ incremental strain quantification by using digital volume correlation techniques [2].

The Hades rig and evolution of microfractures in a crystalline rock sample

Figure 2. Top: The Hades rig can reproduce conditions of pressure and temperature at 8 kilometres depth in the Earth and be used to image rocks at in situ conditions by using dynamic X-ray microtomography [1]. Bottom: Evolution of microfractures in a crystalline rock sample when stress is increased. The rock is rendered transparent and the microfractures are coloured in blue. The upper and lower pistons used to apply the axial load are represented in grey. The confining pressure around the sample was 20 MPa. The microfracture dynamics shed new light on the precursors to earthquake rupture.

Two core samples of monzonite, a rock similar to granite that composes most of the continental crust, were used in this study. The samples were imaged while the stress applied on them was increased in small increments. After an initial elastic deformation phase, some microfractures started to nucleate in the rock volume. Initially randomly oriented, these fractures merged into larger clusters until a fracture spanned the entire volume of the sample and produced a shear fault (Figure 2, bottom). The volume of these microfractures was extracted from the dynamic three-dimensional images and their statistics of size and orientation distributions were quantified. Results show that the total damage (i.e. the volume of all the microfractures) as well as the largest microfracture cluster increase as a power law of loading stress. When approaching failure, these quantities increase fast and diverge at the failure point. The microfractures, initially oriented parallel to the direction of the main compressive stress, rotate to produce the shear fault.

The rock evolves from a phase where it is unbroken to a phase where it is broken over a gradual transition through the formation of damage. Theoretical models of such a transition have already been proposed, while the data collected have provided the first visual and experimental demonstration of the relevance of these models. An important observation was that the dynamics of microfractures, precursors to the main earthquake, display predictable properties of power-law divergence with increasing stress. Such observations at a small scale in these experiments provide new understanding of the nucleation phase of earthquakes, which is usually challenging to observe when hidden several kilometres underneath the Earth’s surface.

Analysis of the data collected sheds new light on the nucleation of faults in rocks, which at the onset of failure displayed a complete three-dimensional connection of most microfractures, generating a fracture network that spanned the entire volume and evolved into a geometrically complex three-dimensional fault zone. In addition, the amount of damage (i.e. microfracture volume) exhibited power law divergence as the loading stress approached the stress at failure, and the distribution of microfracture volume increments driven by the increasing loading stress exhibited power-law statistics similar to the Gutenberg-Richter law for earthquakes.


Principal publication and authors
Critical evolution of damage towards system-size failure in crystalline rock, F. Renard (a, b), J. Weiss (b), J. Mathiesen (c), Y. Ben Zion (d), N. Kandula (a), B. Cordonnier (a, e), Journal of Geophysical Research, 123, 1969-1986 (2018); doi: 10.1002/2017JB014964.
(a) The Njord Centre, University of Oslo (Norway)
(b) ISTerre, University Grenoble Alpes & CNRS, Grenoble (France)
(c) Niels Bohr Institute, University of Copenhagen (Denmark)
(d) University of Southern California, Los Angeles (USA)
(e) ESRF


[1] F. Renard et al., J. Synchrotron Rad., 23, 1013-1034 (2016); doi : 10.1107/S1600577516008730.
[2] J. McBeck et al., J. Geophys. Res. Solid Earth, 123 (2018); doi: 10.1029/2018JB015676.
[3] X. Zheng et al., Geochemistry, Geophysics, Geosystems, 19 (2018); doi: 10.1029/2017GC007322.


Top image: Development of microfractures into a centimetre-scale crystalline rock at the onset of earthquake failure. Red: microfractures. Green: Largest microfracture cluster that leads to system-size failure.