Abstract
For 9–12Cr ferritic–martensitic steels in supercritical water, the dependencies of the thicknesses of three oxide layers (diffusion, inner, and outer layers) on each of seven principal independent variables were separately investigated using three supervised artificial neural networks (ANN) and fuzzy curve analyses. The latter were employed to evaluate the relative significances of independent variables, indicating that on the whole, temperature and exposure time are the most important variables, while the oxide dispersion strengthening (ODS) comes in the first place for the thickness of the diffusion layer. A thicker diffusion layer occurs easily at intermediate temperature approximately 600 °C and/or on ODS steels, due to higher growth rate of the barrier layer than that of the outer layer. The periodic growth of the diffusion layer, including the “shrinking”/“thickening” stages, was revealed by ANN prediction, which may be the basic cause of periodic pore-assembled layers within the inner layer. Finally, the physicochemical basis of classical point defect model was extended to describe the growth of oxide scales for explaining the ANN predictions at the atomic level. The growth of the inner layer into the metal is attributed to the inward transport of oxide ions (actually via outward transport of oxygen vacancies), while the outward transport of cations through the inner layer via a cation vacancy mechanism or as interstitials results in the thickening of the outer layer. The periodic variation in oxygen potentials at the diffusion/inner layer interface is responsible for the periodic growth of the diffusion layer. The iron caves at the inner–diffusion interface left behind by the generation reaction of cation interstitials may be the intrinsic sources of pores within the inner layer.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have