Back diffusion of chlorinated solvent contaminants from a natural aquitard to a remediated aquifer under well-controlled field conditions: predictions and measurements.

Vertical profiles of tetrachloroethene (or perchloroethylene, PCE) and trichloroethene (TCE) were used to validate a diffusion process in a natural aquitard at Dover Air Force Base, Delaware. PCE and TCE distributions in the aquitard underlying an unconfined aquifer were sampled from core tubes obtained at four times over the course of a 35-month field investigation within "test cells" that were isolated from the surrounding ground water by means of grout-sealed steel sheetpile barriers (Mackay et al. 2000). For the final 23 months of this period, boundary conditions at the aquifer/aquitard interface were such that a "back diffusion" of contaminants from the aquitard was induced. Modeling predictions of concentration changes were made on the basis of the earliest coring results and an assumption of sorption-retarded diffusion and using laboratory information about sorption and diffusion characteristics of the media. The predictive modeling was complicated by the fact that "initial" and "final" PCE and TCE distributions in the aquitard were measured at different (albeit proximate) coring locations, such that results reflect spatial variations in aquitard characteristics. This problem was solved by means of an inverse interpretation that involved spatial "translation" of observed profiles on the basis of the laboratory characterizations and assuming a common aquifer-side contaminant history. Predictions indicated substantial change in PCE and TCE concentrations within the upper aquitard (near the aquifer/aquitard interface) and the development of a back-diffusion profile up into the aquifer. Modeling also predicted comparatively minor profile changes in the deeper aquitard, and especially in the deep layer where sorption was strongest. All of these predicted effects were observed in the coring results. Although not exact, the agreement between predictions and observations was sufficiently good to justify the basic tenets of the diffusion model and to support a conclusion that major processes of advection and/or transformation were unimportant within the 35-month time scale of this work.