In the present work, electrochemical noise (EN) was measured in three different types of experimental set-ups in order to obtain and compare various types of corrosion. Simultaneously with EN measurements, a parallel technique was used, regarding the type of experimental set-up: a computer visualization system combined with optical microscopy, or measurements of mechanical changes. In order to distinguish between these corrosion types and to assess corrosion rate, the measured EN signals were analysed by two different techniques: spectral and statistical. On the basis of estimated significant spectral parameters (power spectral densities of current and voltage noise, spectral noise resistance) and statistical parameters (standard deviation, localisation index and noise resistance), the passive state, localized corrosion types and uniform corrosion can be recognized. However, by these parameters it is not possible to distinguish between certain localized corrosion processes: metastable pitting, initiation and growth of stable pits, stress-corrosion cracking. It was ascertained that these corrosion processes exhibit unstable nature, and consequently generated EN signals are usually non-stationary. Since stationarity is required for spectral and statistical analysis, the unstable nature of these processes was recognized as the main reason for unreliable results. It is believed that these types of signal analysis give poor information about corrosion processes which can be easily identified from EN signal itself: transient corrosion processes and transitions between different corrosion processes.
展开▼
