Higher-Order and Time–Frequency Analysis of Electrochemical Noise for Critical Pitting Temperature Assessment of Stainless Steels in Chloride Media: A Review Linking Bispectrum and Wavelet Signatures to Localized Corrosion

Abstract

The most important degradation mechanism for the service life of stainless steels in chloride-bearing environments such as concentrated brine systems, desalination trains and seawater cooling circuits is still localized pitting corrosion. Unlike a pitting potential measured at a single temperature, the critical pitting temperature (CPT) is the temperature threshold between self-healing and self-sustaining corrosion and is the most objective measure of an alloy's resistance to localized attack of chloride. CPT is the lowest temperature at which a metastable pit can become a stable pitting pit. In this paper, we offer a unified perspective on how an alloy approaches and crosses the CPT using electrochemical noise (EN) and its higher-order and time-frequency transformations, the bispectrum and the discrete wavelet transform. We focus on EN, power spectral density, higher order spectrum analysis, and wavelet decomposition as signal processing techniques in the first section of the review. We also give their statistical basis, sensitivity to transient events, and blind spots regarding any one corrosion parameter. In the second section, we relate all of these methods to the physical metallurgy and electrochemistry of pitting by showing why the CPT differs by tens of degrees between potentiostatic tests for sodium chloride (ASTM G150) and ferric chloride immersion (ASTM G48), as well as how the threshold is formed by passive film chemistry. The key point is that the metastable-to-stable pitting transition is nonlinear in nature, with the quadratic phase coupling as a discrete peak in the bispectrum as the wavelet energy distribution changes from fine to coarse crystals, and as the temperature increases towards the CPT, the individual passive-film rupture events stop being statistically independent and come to be the result of shared diffusion fields and local acidification. Here is the case study for hyper-duplex UNS S32707 and super-austenitic 904L. We show that EN-based temperature (58-65 °C) predictions for CPT are the same as electrochemical methods but also provide early non-perturbative warning based on potentiodynamics, potentiostatic, impedance, and electrochemical noise measurements of S32707 in sodium chloride. To provide a more data-driven model of corrosion monitoring and digital twin applications, we conclude our study by presenting a data-driven acquisition-prediction process that combines higher-order, wavelet, and statistical descriptors.