The Effect of Moisture Adsorption on the Nonlinear Elastic Response of Granular Media

Nonlinear elastic effects are ubiquitous in the Earths crust. They are thought to explain at least partly the occurrence of small co-seismic drops in wave velocities that are measured along fault zones. Lab friction experiments suggest that small drops in wave velocity also occur prior to the seismic event, which could be of great interest for earthquake prediction, although such observations in nature are seldom. Our study focuses on unraveling the microphysical origins of the nonlinear elastic response in granular media. Previous work has shown that the nonlinear elastic response of consolidated granular media like rocks likely arises from two distinct mechanisms, one that might be related to the opening/closing of grain contacts, and the other one related to shearing, as well as contact rejuvenation and healing. To gain further insights on the mechanisms taking place at the microscopic scale, we investigate how relative humidity (RH) affects the nonlinear elastic response of granular media. We use Dynamic acousto-elastic testing (DAET) at a constant uniaxial stress of 4 MPa on samples of glass beads under dry (~10%), ambient (~60%) and humid (~100%) conditions at room temperature. The pump frequency at which stress oscillates is kept constant at 10 Hz for all measurements, and the dynamic stress peak amplitude ranges from 0.01 to 0.3 MPa. The samples are probed throughout the experiments with a pair of 1 MHz transducers. DAET allows us to retrieve the full nonlinear elastic response, including transient softening and hysteretic effects. We find that the elastic nonlinearity of humid samples is an order of magnitude larger than for dry samples. Moreover, we find that all extracted nonlinear parameters increase with RH. This overall increase in nonlinearity is consistent with findings from previous studies and with the idea that water adsorption on the grains makes the contact junctions weaker and prone to greater disturbances when subjected to dynamic stressing. Our results also suggest that, if indeed both mechanisms exist, they are affected in a similar fashion in these glass bead samples and cannot be distinguished using changes in RH.