Computer simulations of corrosion: Selective dissolution of binary alloys

Abstract
In this paper we present a new framework, based upon percolation theory, within which various aspects of corrosion in alloy systems may be understood. We have developed a new model of selective dissolution (de-alloying) which is able to account for all the known features of the phenomenon. Monte Carlo simulations of de-alloying in the two-dimensional (2D) square and three-dimensional (3D) simple cubic lattices are presented, employing simple rules for dissolution of the reactive element and surface diffusion of the more noble element. The simulation results reproduce many of the features usually associated with de-alloying in real binary noble-metal alloy systems, including a porous morphology of the de-alloyed residue, coarsening of this porosity, sharp de-alloying thresholds or parting limits, and the development of intermediate compositions. When dissolution of the reactive element is permitted from highly coordinated (terrace) sites, the de-alloying threshold is found to be close to the conventional site percolation threshold for the structure. For the 3D simple cubic lattice, if dissolution is allowed to occur only from ledge or kink sites, then the problem adopts an essentially 2D character and the de-alloying threshold is found to be close to the relevant 2D site percolation threshold. The roughness of the de-alloying front has been shown to vary with the particular rule employed in the simulation and the initial composition of the alloy and is describable in terms of a fractal dimension. The self-similar nature of the interfacial structure and its dependence on composition are a direct result of the cluster structure as the percolation threshold is approached. The functional forms of the current decays below the de-alloying thresholds in the simulations have been analysed and are shown to result from the nature of the initial cluster structure. Other aspects of de-alloying such as mass transport effects and the critical potential are discussed within the context of the model.