Identifying stable pitting pathways in 316 L stainless steel via fractal-inspired PCA-based clustering

This study introduces a fractal-inspired PCA-based framework to distinguish stable pitting pathways in 316 L stainless steel in chloride media. By transforming potentiodynamic polarisation data into a Mandelbrot space, the approach reveals two distinct scenarios of stable pitting growth: directly following passivity breakdown (case I) and preceded by metastable activity (case II). Clustering identifies critical pitting potentials (E_pit and E_sp) with high accuracy, effectively capturing rare metastability-driven events often overlooked in traditional analyses. The method demonstrates robust performance across varying chloride concentrations, with classification metrics highlighting its ability to detect low-frequency E_sp events. Results show that metastability-driven stable pitting (case II) occurs at higher activity levels and potentially at lower potentials. This work advances the understanding of the probabilistic nature of pitting and provides a scalable, data-driven strategy for predicting stable growth regimes.