Abstract
The characteristics of transients or peaks in electrochemical noise (EN) data were assessed by inspection from two systems. One was UNS G10100 in Ca(OH)2/NaCl solution and the other was a magnesium alloy (ZA1040) in Mg(OH)2/NaCl solution. Each system exhibited a variety of both sharp and broad peaks. Directly measured quantities include the location in the time record, the current and potential amplitudes, the area under each peak (as coulombs) and the direction (maximum or minimum). The frequencies of transients are readily assessed given their location in the time record. Inferred quantities include the polarization resistance of the responding electrode and the nature of the transient (anodic or cathodic). Progress for computer-based techniques for reliably finding transients within a set of EN data is described. One promising approach is that locations in the time record where the current derivative crosses zero correlates with the apex of simple rounded peaks. However, this is not true of broad or noisy peaks. A promising approach is to apply data smoothing to round broad or “noisy” peaks and permit the derivative to identify the apex. This pre-processing of EN data may enable artificial neural networks to accurately locate peaks. This work also suggested that the sampling frequency influences the number and type of transients detected and thus should be tuned to each particular system. It also suggested that consideration be given to the experimental arrangement to ensure that the current and potential are correlated during transients.
