How reaction and permeability develop in dehydrating systems

The triggering of earthquakes at intermediate-depth along subduction zones is often explained by dehydration reactions, releasing free-water and allowing pore-fluid pressure build-up. During dehydration reactions, pore-fluid pressure is increased when permeability is low enough to prevent fluid escape. Permeability is not constant during dehydration reactions but is rather changed by porosity changes (i.e. solid volume reduction or pore compaction). The evolution of permeability during dehydration reactions will thus dictate the pore-fluid pressure evolution that will affect rock strength and earthquake triggering. However, our understanding on the coupling between permeability, pore-fluid pressure, microstructures, deformation and reaction rate is incomplete. In some cases, the development of reactions is distributed uniformly and permeability increases steadily throughout the reaction progress. In other cases, reactions will not proceed uniformly and nature along with previous experiments indicate that "reaction fronts" develop. On the large scale, reaction rate and fluid pressure evolution depend on the movement of these fronts. Experimental results are presented on permeability and reaction front evolution during gypsum dehydration - an analogue for silicate dehydration. Triaxial experiments were conducted using polycrystalline gypsum cores with very low initial porosity. Pore-fluid pressure is controlled at one end of the sample and monitored at the other in order to measure permeability. Gypsum cores were dehydrated at a constant temperature of 115°C. Two parameter spaces were explored: the pore-fluid pressure (20, 40 or 60 MPa) that influences reaction rate, and effective confining pressure (60 or 110 MPa) that influences pore-compaction. The evolution of permeability, porosity, reaction rate and pore-fluid pressure are measured throughout the reaction. SEM observations of post-mortem samples collected at three key stages during the reaction shows how the evolving microstructures and the reaction patterns control permeability, deformation and reaction rate. A theoretical model for reaction front speed is in good agreement with the observations and gives confidence that a quantitative model for reaction front behaviour could be extended to dehydrating silicates.