Background: A Primer on Laboratory Friction Studies

Why study the friction of rocks? Many crustal earthquakes are caused by the sudden movement on preexisting faults. Therefore, an understanding of the friction between rock surfaces is important for understanding earthquake mechanisms. The simplest model of this process is shown in a schematic diagram of a block and spring experiment: A mass (m) rests on a flat surface, and a tangential force is applied to it by pulling the end of the spring slowly.

A plot of the force in the spring as a function of displacement at the end of the spring would look like this:

At first there is a linear response as the spring stretches. Then the curve departs from a linear response as the block begins to slide. At some maximum force, the block may suddenly slip forward so that the force in the spring drops. The cycle resumes until the block suddenly slips again. This is called stick-slip behavior, and is the laboratory equivalent of the earthquake process. As in our simple model, the earthquake cycle begins with a period of loading (interseismic period, i.e., between earthquakes). In the next phase, the fault material begins to yield and premonitory creep is observed in some cases. Finally, the fault ruptures and the stress on the fault drops.

Frictional sliding between surfaces does not always proceed in this jerky, stick-slip motion. Alternatively, in our block and spring model, the block may slide stably, such that the motion between the block and the flat surface is smooth and the force in the spring is constant. Stable sliding of the block is equivalent to creep along an active fault. Both types of deformation can occur in the block and spring model depending on characteristics of the sliding surface and the spring. Indeed, both types of deformation occur along faults such as the San Andreas in California. The northern and southern portions of the San Andreas Fault are locked for long periods of time as stresses build up in the earth due to the relative motion of the Pacific and North American Plates. The sudden release of energy causes slip of the fault surface and an ensuing earthquake. In Central California, the fault slowly creeps, causing numerous microearthquakes but apparently no large events.

The simple block and spring model is not fully adequate to simulate the complexities of fault behavior at depth in the earth, so several types of experimental apparatus have been devised for more realistic laboratory experiments.

In a biaxial experiment (biaxial = two directions of force), rectangular blocks of rock are squeezed together laterally, and at the same time forced to slide past one another due to an applied vertical force. The third face of the block is open to the air with no applied pressure, much the same as on the surface of the earth. Relative displacement of the blocks and the force necessary to cause slip between the blocks is recorded. From this, the coefficient of friction of the rock, or of granular fault gouge placed between the rocks is determined. The coefficient of friction, µ, defined as the ratio of shear stress, , (the tangential stress in the direction of sliding) to normal stress, n, (the stress pushing the blocks together) is an important physical parameter used in the modeling of earthquake behavior.

In a triaxial experiment (triaxial = three directions of force), a cylindrical sample of rock is sealed in a leak-proof jacket and then placed inside a pressure vessel. The vessel is pressurized equally in three dimensions by a hydraulic fluid to simulate the pressure acting on rock at depth in the earth. The rock can then be squeezed along the axis of the cylinder (simulating tectonic forces in the earth) by the action of a piston that pushes against the end of the sample. In a triaxial apparatus, the failure strength of intact rock (force required to break the specimen) can be determined, as well as the frictional strength (force required to slide preexisting fractures, or "artificial faults" past one another).

The total amount of slip between rock surfaces in both biaxial and triaxial experiments is limited by the geometric constraints of the apparatus. When greater fault displacements are desired, a third type of apparatus called a rotary shear machine is used. In this type of experiment, two rings of rock rotate past one another so that the relative displacement of the two "fault" surfaces can be very large. Here, analysis of the stresses and displacement of the rock is more complex because the stress distribution and relative velocity of the rock assembly varies between the inner and outer parts of the rock ring.

Other important parameters that affect the behavior of faults in the earth are pore fluid pressure and temperature. These conditions are both reproduced in laboratory triaxial experiments by injecting water into the rock and heating the sample during testing. In this way, all of the important conditions controlling earthquakes on natural faults are reproduced on test specimens in the laboratory.