The initial stages of cave formation: Beyond the one-dimensional paradigm

The solutional origin of limestone caves was recognized over a century ago, but the short penetration length of an undersaturated solution made it seem impossible for long conduits to develop. This is contradicted by field observations, where extended conduits, sometimes several kilometers long, are found in karst environments. However, a sharp drop in the dissolution rate of CaCO_3 near saturation provides a mechanism for much deeper penetration of reactant. The notion of a kinetic trigger - a sudden change in rate constant over a narrow concentration range - has become a widely accepted paradigm in speleogenesis modeling. However, it is based on one-dimensional models for the fluid and solute transport inside the fracture, assuming that the dissolution front is planar in the direction perpendicular to the flow. Here we show that this assumption is incorrect; a planar dissolution front in an entirely uniform fracture is unstable to infinitesimal perturbations and inevitably breaks up into highly localized regions of dissolution. This provides an alternative mechanism for cave formation, even in the absence of a kinetic trigger. Our results suggest that there is an inherent wavelength to the erosion pattern in dissolving fractures, which depends on the reaction rate and flow rate, but is independent of the initial roughness. In contrast to one-dimensional models, two-dimensional simulations indicate that there is only a weak dependence of the breakthrough time on kinetic order; localization of the flow tends to keep the undersaturation in the dissolution front above the threshold for non-linear kinetics.