Nanoseismic sources made in the laboratory: source kinematics and time history

When studying seismic signals in the field, the analysis of source mechanisms is always obscured by propagation effects such as scattering and reflections due to the inhomogeneous nature of the earth. To get around this complication, we measure seismic waves (wavelengths from 2 mm to 300 mm) in laboratory-sized specimens of extremely homogeneous isotropic materials. We are able to study the focal mechanism and time history of nanoseismic sources produced by fracture, impact, and sliding friction, roughly six orders of magnitude smaller and more rapid than typical earthquakes. Using very sensitive broadband conical piezoelectric sensors, we are able to measure surface normal displacements down to a few pm (10^-12 m) in amplitude. Thick plate specimens of homogeneous materials such as glass, steel, gypsum, and polymethylmethacrylate (PMMA) are used as propagation media in the experiments. Recorded signals are in excellent agreement with theoretically determined Green’s functions obtained from a generalized ray theory code for an infinite plate geometry. Extremely precise estimates of the source time history are made via full waveform inversion from the displacement time histories recorded by an array of at least ten sensors. Each channel is sampled at a rate of 5 MHz. The system is absolutely calibrated using the normal impact of a tiny (~1 mm) ball on the surface of the specimen. The ball impact induces a force pulse into the specimen a few ms in duration. The amplitude, duration, and shape of the force pulse were found to be well approximated by Hertzian-derived impact theory, while the total change in momentum of the ball is independently measured from its incoming and rebound velocities. Another calibration source, the sudden fracture of a thin-walled glass capillary tube laid on its side and loaded against the surface of the specimen produces a similar point force, this time with a source function very nearly a step in time with rise time of less than 500 ns. The force at which the capillary breaks is recorded using a force sensor and is used for absolute calibration. A third set of nanoseismic sources were generated from frictional sliding. In this case, the location and spatial extent of the source along the cm-scale fault is not precisely known and must be determined. These sources are much more representative of earthquakes and the determination of their focal mechanisms is the subject of ongoing research. Sources of this type have been observed on a great range of time scales with rise times ranging from 500 ns to hundreds of ms. This study tests the generality of the seismic source representation theory. The unconventional scale, geometry, and experimental arrangement facilitates the discussion of issues such as the point source approximation, the origin of uncertainty in moment tensor inversions, the applicability of magnitude calculations for non-double-couple sources, and the relationship between momentum and seismic moment.