Abstract
Many projects involving nuclear fuel rest on a quantitative understanding of the co-existing phases at various stages of burnup. Since the many fission products have considerably different abilities to chemically associate with oxygen, and the oxygen-to-metal molar ratio is slowly changing, the chemical potential of oxygen is a function of burnup. Concurrently, well-recognized small fractions of new phases such as inert gas, noble metals, zirconates, etc. also develop. To further complicate matters, the dominant UO2 fuel phase may be non-stoichiometric and most of the minor phases themselves have a variable composition dependent on temperature and possible contact with the coolant in the event of a sheathing breach.
A thermodynamic fuel model to predict the phases in partially burned CANDU (CANada Deuterium Uranium) nuclear fuel containing many major fission products has been under development. The building blocks of the model are the standard Gibbs energies of formation of the many possible compounds expressed as a function of temperature. To these data are added mixing terms associated with the appearance of the component species in particular phases. In operational terms, the treatment rests on the ability to minimize the Gibbs energy in a multicomponent system, in our case using the algorithms developed by Eriksson. The model is capable of handling non-stoichiometry in the UO2 fluorite phase, dilute solution behaviour of significant solute oxides, noble metal inclusions, a second metal solid solution U(Pd-Rh-Ru)3, zirconate, molybdate, and uranate solutions as well as other minor solid phases, and volatile gaseous species.
