MECHANISTIC MODELING OF NUCLEAR WASTE FORM LEACHING BY AQUEOUS SOLUTIONS.

摘要

Predictive modeling of nuclear waste form leaching requires a mechanistic representation of the processes and factors controlling the rates at which species are released from the solid waste form into the contacting aqueous solution. A basic distinction is made among bulk, surface, and leachant species, as well as between major (i.e., network formers) and minor (i.e., network modifiers) constituents of the solid structure. Major constituents exhibit no mobility within the solid bulk. Minor constituents may migrate within the solid matrix according to an electrically-assisted diffusion process. When the major constituents of the solid network are released from the outer surface layer into the solution, a breakdown of this layer ensues. This corrosion of the solid matrix occurs at a rate limited by the solubility of the network formers in the aqueous medium. Forward and backward reactions of the solid-liquid interface are modeled through two phenomenological rate processes. After showing the relevance of the semi-infinite medium approximation for describing leaching, all identified, potentially rate-controlling leaching mechanisms are integrated into a mathematical formulation which also incorporates two important system parameters: the specimen surface area-to-solution volume ratio and the leachant renewal frequency. Asymptotic analysis of a linearized version of the model shows that predicted short- and long-term leaching behaviors are physically correct. Surface processes tend to be initially rate-controlling, while at longer times dilution conditions of the leachant determine the rate-determining mechanism. Under dynamic leaching conditions, network dissolution eventually controls the leaching process whereas under static conditions bulk diffusion eventually prevails. The non-linear version of the model has been adapted to sodium and silicon leaching from borosilicate glass in deionized water, and fully implemented on computer by using a numerical strategy. The resulting code has been named LIX. This more complex version upholds the conclusions reached by the asymptotic analysis. Moreover, when tested against actual PNL 76-68 glass leaching data, LIX shows excellent capabilities in reproducing the experimental evidence, in particular the effects of the surface area-to-solution volume ratio.