Centre Technique des Industries Agroalimentaires

Unexpected corrosion phenomena caused by a priori non-corrosive organic highly resistive solutions have been identified in some industrial petrochemical plants. Due to the extremely low conductivity of this kind of non-polar solution, performing reliable and reproducible electrochemical tests remains a difficult task. In which concerns EIS, Nyquist diagrams are prone to be crippled by a strong bulk-related capacitive behavior imposed by the dielectric nature of the electrolyte. The low frequency processes at the metal interface are then often hidden or at least badly resolved. In spite of these inherent difficulties, some of these phenomena have been successfully reproduced and studied in laboratory with a two-electrode cell in which two identical large surface plates of carbon or stainless steel were assembled facing each other and separated by a gap of 1 mm. EIS was then used to detect the onset and investigate the possible causes of the corrosion phenomena. Results appeared to be related either to water contamination or to the presence of solution additives. In the case of water contamination results showed that, even below the solubility coefficient, droplets of aqueous solution are ejected from the organic matrix at the metal interface creating clusters of aggressive electrolyte yielding localized corrosion. The corresponding Nyquist diagrams after water contamination have been fitted with a two time-constants electrical equivalent circuit in which the low frequency domain was straightforwardly associated to the aqueous droplets formation and was used to monitor the onset of corrosion. In the case of additive-containing organic matrix, however, the Nyquist diagrams exhibited a time-evolving profile that required a dual electrical circuit fitting strategy to be applied. Indeed, the process time evolution was so that diagrams were alternatively fitted with two different circuits for the same system depending on the immersion duration: a 2 time-constants EEC for the first ca 36 hours and a single bulk-related time-constant after that. This fitting methodology, although mandatory from an experimental point of view due to the time evolution profile of the diagrams, brings to the forefront of the problem the question of the validity of shifting between two or more different EEC for the survey of a monotonically evolving single interface. In other words, the shift between the two different circuits must match the physical continuity of characteristic parameters. This means that the fitting results for the different elements introduced in the EEC must not show leaps or discontinuities as the EEC are changed, since they are supposed to represent the same system. This parametric continuity condition has been used throughout the study and is illustrated in Figure 1 for the time evolution of the dielectric constant of the bulk organic phase and the solution resistance as obtained from the numeric values of the fitted EEC. It can be seen that in the initial stages the modified solution showed a higher dielectric constant that was associated with corrosion processes taking place till the reactive species were depleted at the interface or till the surface was blocked by adsorbed species. The bulk phase consistently retrieved his usual dielectric constant values after ca 36 hours in the conditions of Figure 1. Figure 1