Experimental study of montmorillonite erosion in low ionic strength water under stagnant and flow conditions in artificial fractures

Abstract

Compacted bentonite has been proposed as a buffer material in several geological repository concepts for high-level radioactive waste. However, in dilute groundwater, bentonite, especially when dominated by Na-montmorillonite, has been found to be susceptible to erosion and excessive mass loss may compromise the safety of the repository. Herein, results from twelve erosion tests with either flowing or stagnant aqueous solutions, performed with custom-designed artificial fractures made of Poly(methyl methacrylate) are reported. The fracture aperture was 0.1 mm in all tests. Na-montmorillonite or an equal mix by weight of Na- and Ca-montmorillonite were employed. Eleven of the tests were performed with the fracture being vertical, and in one test the fracture was horizontal. The transparent artificial fracture setup allowed monitoring of the expansion of the montmorillonite into the fracture. In all but one test, the diameter reached a limit value, normally within one month of the start of the test. The final diameters varied between 5.9 and 8.7 cm and no clear correlation to flow rate, type of montmorillonite or ionic strength could be discerned. The initial diameter of the emplaced clay was 3.5 cm. The erosion was monitored regularly by turbidity measurements. In general, erosion rates decreased with time and were ultimately low enough that an extrapolation or upscaling of the results to repository scale and operating time would imply a total erosion below the critical value for repository safety. This was the case for all vertical fractures, regardless of montmorillonite type and ionic strength of the aqueous solution. The observed decrease in erosion rates was not due to depletion of montmorillonite at the source because at the end of each test 90% or more of the montmorillonite remained. Further, the ionic strengths were below the critical coagulation concentration, hence gelation cannot provide an explanation to the reduction in erosion rates. Finally, the diameters of the extruded montmorillonite reached the limit while the erosion rates continued to decrease. Consequently, the observed limited expansion was not due to a steady state between erosion and swelling. Rather, the expansion behaviour suggests the presence of frictional forces between the montmorillonite and the fracture walls. The results from this study are beyond the current understanding of montmorillonite colloid behaviour and cannot be explained by existing models.