Dense, non-aqueous phase liquids (DNAPLs) are nonwetting fluids with respect to water when they exist in most geologic materials and therefore, these oily liquids flow preferentially along fractures and high permeability strata in otherwise low permeability deposits of silt and clay or sedimentary rocks. Mathematical models of media with simple fracture geometry and laboratory and field experiments using natural clayey deposits are used in this thesis to determine the importance of aqueous-phase diffusion from immiscible phase organic liquids in fractures. The influence of phase transfer by diffusion is incorporated into a conceptual model where the solvent DNAPL initially present in the fractures completely dissolves away causing a complete change in contaminant phase. Initially, the mass is located in fractures where active groundwater flow occurs and finally it exists primarily in the matrix where there is no significant flow.

Mathematical modelling for diffusion away from stationary, relatively soluble solvent DNAPLs in single planar fractures shows that the time for complete conversion of the immiscible-phase to dissolved and sorbed phase in natural clayey matrix material between fractures ranges from days to several weeks for fractures with apertures of common size (5 to 100 microns). Lower solubility DNAPL takes longer to disappear. Diffusive mass transfer in sedimentary rocks is generally slower due to smaller matrix porosities, lower sorption capacities and larger fractures, which can increase disappearance times to years or decades. For cases of parallel planar fractures or cubic fracture blocks there is a maximum contaminant storage capacity for dissolved and sorbed mass in the matrix between fractures. A mass storage capacity ratio greater than one indicates the number of times the fracture void volume can be replenished with DNAPL and still result in disappearance. Each replenishment consumes part of the storage capacity and the time for each subsequent replenishment to disappear increases. For a specific mass of DNAPL and fracture porosity, more closely spaced and smaller fractures lose the DNAPL in the fracture network quicker than fewer, larger-aperture fractures due to the greater surface area over which diffusion occurs and an increase in the surface area to volume ratio of the immiscible phase liquid.

Laboratory and field evidence for diffusive migration for time periods between 15 days and 6.5 years was obtained for three common chlorinated organic solvents, tetrachloroethene (PCE), trichloroethene (TCE), and dichloromethane (DCM), in two natural clay-rich deposits. The measured mass was used to determine the mass flux per unit surface area, which provides the volume of DNAPL transferred by aqueous-phase diffusion. Concentration versus distance profiles away from the DNAPL sources were evaluated using one-dimensional diffusive transport models with a least squares simplex optimization routine. The best-fit diffusion coefficients were consistent with each other, with literature values and with the calculated mass fluxes. These results substantiate the model predictions of rapid mass flux across planar surfaces separating solvent DNAPL from clayey matrix material and support the conceptual model in which solvent DNAPL disappears rapidly from common types of fracture networks. Rapid disappearance of solvent DNAPL at contaminated industrial sites has major implications with respect to delineation of DNAPL source zones, interpretation of monitoring results and selection of remediation methods.