Description of new dry granular materials of variable cohesion and friction coefficient: Implications for laboratory modeling of the brittle crust

International audienceCohesion and friction coefficient are fundamental parameters for scaling brittle deformation in laboratorymodelsof geological processes. However, they are commonly not experimental variable, whereas (1) rocks range fromcohesion-less to strongly cohesive and from low friction to high friction and (2) strata exhibit substantialcohesion and friction contrasts. This brittle paradox implies that the effects of brittle properties on processesinvolving brittle deformation cannot be tested in laboratory models. Solving this paradox requires the use ofdry granular materials of tunable and controllable brittle properties. In this paper, we describe dry mixtures offine-grained cohesive, high friction silica powder (SP) and low-cohesion, low friction glass microspheres (GM)that fulfill this requirement. We systematically estimated the cohesions and friction coefficients of mixtures ofvariable proportions using two independent methods: (1) a classic Hubbert-type shear box to determine theextrapolated cohesion (C) and friction coefficient (μ), and (2) direct measurements of the tensile strength (T0)and the height (H) of open fractures to calculate the true cohesion (C0). Themeasured values of cohesion increasefrom 100 Pa for pure GM to 600 Pa for pure SP, with a sub-linear trend of the cohesion with the mixture GMcontent. The two independent cohesion measurement methods, from shear tests and tension/extensional tests,yield very similar results of extrapolated cohesion (C) and show that both are robust and can be used independently.The measured values of friction coefficients increase from 0.5 for pure GM to 1.05 for pure SP. The useof these granular material mixtures now allows testing (1) the effects of cohesion and friction coefficient inhomogeneous laboratory models and (2) testing the effect of brittle layering on brittle deformation, as demonstratedby preliminary experiments. Therefore, the brittle properties become, at last, experimental variables