A comprehensive evaluation of the transport of U, Th, Ra, and Rn nuclides of the U-238- and Th-232-decay series in an unconfined sandy aquifer (Long Island, NY) was conducted. Groundwater data are compared with results of a theoretical transport model of weathering of aquifer grains and interaction with surface coatings to establish relationships between the concentrations of the radionuclide activities in the water and flow line distance. The data provide estimates for geochemical parameters including weathering rates and chemical reactivities in both the vadose zone and the aquifer. A theoretical treatment of the transport is presented that considers the reaction between the water and a reactive surface layer. It is found that a model with chemical exchange between the surface layer and the water for all species is not valid, and that the effects of saturation and "irreversible" precipitation of Th is required. The water table shows a relatively wide range in U activities, the only element in the U-Th series for which vadose zone input is significant in the aquifer. High weathering of U and recoil inputs of U-234 to the water occur in the upper 3 m of the vadose zone, while lower weathering and removal of U from the water occur below. The deeper aquifer has variable U-238 activities that can be accounted for by input from the vadose zone and is nor a result of non-conservative behavior. The isotopic composition of U is shown to be directly related to the recoil rate relative to the weathering rate. The wide range of U-238 in the aquifer waters is a reflection of diverse vadose zone inputs, showing that dispersive mixing is not a dominant effect. The higher values of delta U-234 in the aquifer reflect the recoil/weathering input ratios from within the aquifer where the weathering rate is lower than the vadose zone. Both high U activities and high delta U-234 values cannot be obtained in the vadose zone or within reasonable flow distances in the aquifer. Radium isotopes are found to be in exchange equilibrium with the surface layer. Ra-224, Ra-228, and Ra-226 have comparable activities throughout the aquifer. In the vadose zone. the dominant input of Ra to groundwater is weathering and recoil. As found elsewhere, the Rn-222 in the water is a large fraction (similar to5%) of the Rn produced in the aquifer rock. This cannot be due to Ra precipitation onto surface coatings in the aquifer as supported by present weathering with Th in exchange equilibrium with the surface layer. It is found that Th is saturated in the waters under oxidizing conditions so that the weathering input is irreversibly precipitating onto surfaces. However, it is shown that under somewhat reducing conditions, Th activities are much higher and the Th/U ratio in the solution is approximately that of the rock, We propose that under oxidizing conditions the source of Rn is a surface coating enriched in Th-232 and Th-230. This Th was precipitated in an earlier phase during rapid dissolution of readily weathered phases that contain similar to 10% of the U-Th inventory of the rock, with the associated U carried away in solution. Therefore, the previously precipitated and Th-232 produce daughter nuclides in the surface coating which are the dominant contributors of Ra and Rn to the ground water. In particular, Rn is provided by very efficient losses (by diffusion or recoil) from the surface coating, This then does not require recent, large recoil losses from the parent rock or the presence of nanopores in the rock. The first data of both long-lived Th-232 and short-lived Th-234 and Th-228 in ground water is reported. The Th isotope activities indicate that desorption kinetics are slow and provide the first estimate, based on field data, of the Th desorption rate from an aquifer surface. The mean residence time of Th in the surface coating is similar to 3000 y while in the water it is similar to1 h. Ra is in partition equilibrium with the aquifer surface layer. However, the strong fixation of Th on surface coatings is very susceptible to changes in oxidation state as is shown by a comparison of two adjacent aquifers. This makes it difficult to define with certainty the retentive characteristics in natural systems. In general, it is shown that the distributions of naturally occurring nuclides can be used to calculate values for transport parameters that are applicable to the transport of anthropogenic nuclides. Copyright (C) 2001 Elsevier Science Ltd.
