Combined time-lapse magnetic resonance imaging and modeling to investigate colloid deposition and transport in porous media.

Colloidal particles can act as vectors of adsorbed pollutants in the subsurface, or be themselves pollutants. They can reach the aquifer and impair groundwater quality. The mechanisms of colloid transport and deposition are often studied in columns filled with saturated porous media. Time-lapse profiles of colloid concentration inside the columns have occasionally been derived from magnetic resonance imaging (MRI) data recorded in transport experiments. These profiles are valuable, in addition to particle breakthrough curves (BTCs), for testing and improving colloid transport models. We show that concentrations could not be simply computed from MRI data when both deposited and suspended colloids contributed to the signal. We propose a generic method whereby these data can still be used to quantitatively appraise colloid transport models. It uses the modeled suspended and deposited particle concentrations to compute modeled MRI data that are compared to the experimental data. We tested this method by performing transport experiments with sorbing colloids in sand, and assessed for the first time the capacity of the model calibrated from BTCs to reproduce the MRI data. Interestingly, the dispersion coefficient and deposition rate calibrated from the BTC were respectively overestimated and underestimated compared with those calibrated from the MRI data, suggesting that these quantities, when determined from BTCs, need to be interpreted with care. In a broader perspective, we consider that combining MRI and modeling offers great potential for the quantitative analysis of complex MRI data recorded during transport experiments in complex environmentally relevant porous media, and can help improve our understanding of the fate of colloids and solutes, first in these media, and later in soils.