Experimental validation of a probabilistic model based on Markov chains for the assessment of pitting corrosion of stainless steels

Localized corrosion of stainless steels is the main concern in industrial applications. Pitting, crevice and stress corrosion cracking in chloride-containing solutions are common corrosion forms which must be accurately avoided. Pitting corrosion consists of two phases: corrosion initiation, which ends with the breakdown of the passive film due to the presence of chlorides, and corrosion propagation, which occurs in the presence of oxygen with high corrosion rate (in the order of mm/year) due to the high cathodic-to-anodic surface ratio. Pitting corrosion initiation depends on a variety of factors associated with the metal (chemical composition, presence of heterogeneities), the environment (chlorides content, pH, temperature) and the geometry of the system (presence of crevices, discontinuities). In this work, pitting initiation is described by means of a probabilistic approach, based on Markov chains theory. The proposed model calculates pitting initiation probability for a specific application. The Markov chain consists of five steps: starting from an initial metastable condition, the system will evolve necessarily to an absorbing state (pitting and passivity). Two transient states, called metapitting and metapassivity, are defined to take into account the metastable pits formation and repassivation. The probability from the metastable to the metapassive condition is called m. Vice versa, (1-m) is the probability to move from the metastable to the metapitting state. The metapassive state shows a probability r to be absorbed in the passive state and (1-r) to go back; at the same way, the metapitting state has probability p to move to the pitting state. The aim of the thesis is to validate the mathematical equation of m-probability, which depends in first analysis on temperature, pH, metal chemical composition and presence of crevice. To pursue this objective, potentiostatic and potentiodynamic tests were carried out changing operative conditions, as solution pH and temperature. The obtained results have been compared with data collected from literature and the experimental trends have been elaborated to revise the original Markov model.