Rethinking aqueous phase diffusion related isotope fractionation: Contrasting theoretical effects with observations at the field scale.

Aqueous phase diffusion-related isotope fractionation (DRIF) was investigated through modelling to determine under what subsurface conditions carbon isotope DRIF effects would be observable using typical sampling approaches. A dispersive enrichment factor was defined based on heavy and light isotopologue dispersion coefficients. For a given ratio of source concentration (C) to method detection limit (MDL), the maximum DRIF in a system increased linearly with transverse dispersive enrichment factor. Using this linear relationship, the critical dispersion enrichment factor for which DRIF would not be observable was quantified. Dispersive enrichment factors for various contaminants (benzene, toluene, chlorinated compounds) were estimated using field scale transverse dispersion coefficients upscaled from compound specific or non-compound specific local scale dispersivity. All predicted dispersive enrichment factors with non-compound specific dispersivity are smaller than critical values even for high C/MDL ratios (e.g. 25,000), indicating DRIF would generally not be observable in systems where soil dispersivity is non-compound specific. To date, this finding has not been clearly articulated in the DRIF literature. While the calculated dispersive enrichment factors for some compounds with compound specific transverse dispersivity exceeded the critical values at which DRIF might become significant, the zones in which DRIF could potentially be observable were limited to bands below 0.5m wide on lateral plume edges. In aquifer-aquitard systems, DRIF was theoretically detectable only in thin aquifers (e.g. 0.5m) bounded by thick (e.g. meters) aquitards. DRIF due to back diffusion from aquitards would not be observable regardless of aquifer thickness. Simulations addressing the mixing effect in wells demonstrated that DRIF effects would be difficult to identify in the field without a sampling strategy (including smaller than industry norm well screen lengths, and fine sampling scales) expressly targeted towards that goal. The results of this study help identify what the required characteristics of such a field strategy might be.