On a model for the codiffusion of isotopes