An experimental study of the influence of regolith micro-structure on seismic wave velocities

Introduction:

Determining the seismic wave velocity in a given soil is of great value in the study of its elastic properties, and to establish a coherent sub-surface model. This non-destructive technique is widely used in both geotechnical and geophysical communities, and has also been used in planetary exploration. The InSight seismometer SEIS [Lognonné et al., 2019] operated for 4 terrestrial years and was able to record near-surface events generated by the trials of the HP3 instrument to penetrate the martian soil.[Brinkmann et al., 2022]. This has enabled the seismic wave velocity in the regolith (i.e the entire unconsolidated cover that overlies more coherent bedrock) to be determined. Similarly, the seismic wave velocity in the lunar regolith has been inferred [Tanimoto et al., 2008] thanks to seismometers on board the Apollo missions [Latham et al., 1969].

However, martian and lunar regolith do not have the same characteristics. Due to aeolian processes, martian grains are more rounded than the lunar ones, that are generated by impact processes. Moreover, martian and lunar regolith present different grain size distributions

 The goal of our work is to check if the regolith grain size distribution can have an impact on measured seismic wave velocities. A deep understanding of this effect is of interest for interpreting data from past missions (InSight, Apollo), and also for planning future missions, such as the Farside Seismic Suite for the Moon [Panning et al., 2022] or seismometers for asteroids [Murdoch et al. 2017; Murdoch et al. 2024; Bernauer et al. 2020]. To this end, we perform laboratory experiments to determine seismic wave speeds in samples with different grain size distributions.

Methodology:

The experimental set-up is described in Fig. 1. The main components are the bender elements [Dyvik and Madshus, 1985]. These piezoelectric pieces can bend and generate a shear seismic wave when a voltage is applied, or generate an electric signal when a movement is applied to them. An emitting and a receiver bender element are placed at each end of a cylindrical sample. Different levels of confining pressure are reached by applying vacuum inside the sample with a vacuum pump. Sample density variations are measured when the pressure is changed.

 

The tested samples are constituted of binary glass beads. They are characterized by two parameters: the grain size ratio (GSR), which is the ratio of the small bead diameter to that of the large grains, and the mass fraction (MF), which is the proportion in mass of the small beads. Three values of the GSR are tested (0.3, 0.4, 0.5) and for each grain size ratio, the mass fraction is varied from 0.05 to 0.8. Two levels of confining pressure are studied here: 25 and 50 kPa.

 

Results:

 

For GSR = 0.3, the seismic velocity increases with the mass fraction up to MF = 0.2. We interpret those variations as a marker of the filling of the voids between the large beads by the small beads, leading to the seismic wave path shortening. The velocity decrease for MF>0.2 shows that small beads are pushing apart the large beads, rupturing the contacts between large beads [Choo and Lee, 2021] and lengthening the path of the seismic waves.

 

When GSR=0.5, as small beads are too large to fit into the voids between large beads, the contact rupture mechanism makes the velocity decrease up to MF = 0.4. For higher MF, the path of the seismic wave is mainly composed of small beads. Thus, increasing the MF does not impact the seismic velocity. At GSR = 0.4, we suggest that there is a competition between hole filling and contact rupture mechanism, leading to a slight increase in the velocity with the mass fraction.

 

Velocity variations are similar at both confining pressure levels and they are independent of the frequency of the input signal used.  In addition, the observed velocity variations cannot be explained by density variations.

 

Conclusions:

Our results demonstrate that grain size distribution has a density-independent effect on the seismic velocities. Given the large variety of regolith grain size distributions found on different planetary surfaces, size distribution has to be characterized to deeply understand mechanical macro-parameters differences from one planetary regolith to another.