Mechanism for Propagation of Intergranular Corrosion in Pipeline Steel

Abstract

"High-pH" (pH 8 to 10.5) stress corrosion cracking (SCC) and intergranular corrosion (IGC) of low-carbon pipeline steels both occur in the active-passive transition potential region, suggesting that both processes are driven by metal dissolution (1). In non-sensitized pipeline steels, there are no grain boundary (GB) precipitates and no consensus on the mechanism of IGC. IGC in non-sensitized steels forms triangular wedges of corrosion product around GBs (2,3). The wedges display characteristic angles determined by enhanced rates of dissolution at the GB compared to that of the grain surface (2,3). Recently we reported the formation of such corrosion product wedges around GB triple junctions during high-pH IGC of X70 pipeline steel in 1 M sodium bicarbonate solution (4). Nanoindentation measurements detected reduced hardness in the metal around corroded GBs, which was suggested to arise from softening by vacancies produced by anodic oxidation of reactive Si solute atoms (5). In the present work, a mathematical model was developed for evolution of GB morphology during IGC of pipeline steel. The model incorporates generation of vacancies at the dissolving steel interface and vacancy diffusion to GBs. The vacancy concentration at the corroding surface is taken to be the same as that of Si solute, 0.77 at. %. Vacancies reaching GBs are assumed to diffuse rapidly to the steel surface and contribute to GB recession, thus accounting for enhanced penetration of corrosion along the boundary. Finite-element simulations were performed to predict GB shape evolution. The predicted shapes were compared with measurements from cross sections of X70 steel samples corroded in 1 M NaHCO3at -0.521 V vs Ag/AgCl. Short corrosion exposures of 2 hr produced GB wedges with flat triangular sides having angles of 120o- 140o(5). Calculations with the vacancy diffusion model showed that these angles are consistent with a vacancy diffusivity of 3 x 10-18m2/s, one order of magnitude smaller than the diffusivity of 6 x 10-17m2/s obtained by extrapolation from high-temperature measurements in bcc Fe (6). The lower diffusivity in ferritic steel is considered reasonable in view of likely vacancy trapping in the alloy. For longer corrosion exposures of 5 hr, the current density decreased by two orders of magnitude due to the formation of a precipitated corrosion product on the outer steel surface. Pronounced sharpening of the GB wedges was observed to angles of 20o- 30o. For comparison, calculations with the vacancy diffusion model incorporated the diffusivity from short-time experiments and included the transient decay of the corrosion rate. The calculations demonstrated dramatic sharpening of the GB wedges in quantitative agreement with experiment (Fig. 1). Sharpening is caused by an enhanced contribution to GB morphology evolution from vacancy diffusion at low current densities, relative to that from grain surface corrosion. Further, sample cross sections after long corrosion exposures showed evidence of GB crack initiation. It is suggested that wedge angle sharpening leads to intensification of tensile wedging stress due to GB corrosion product (5), which eventually initiates cracks. Therefore, the vacancy diffusion model explains IGC morphology evolution and also provides an explanation for the transition of IGC to early-stage SCC. REFERENCES R. N. Parkins, S. Zhou, Corros. Sci., 52, 347 (1996). B. Gwinner et al., Corros. Sci., 107, 60 (2016). L. Beaunier, M. Froment, C. Vigneaud, Electrochim. Acta, 25, 1239 (1980). D. Yavas et al., Electrochem. Commun., 88, 88 (2018). D. Yavas et al., Electrochim. Acta, 285, 336 (2018). M. I. Mendelev, Y. Mishin. Phys. Rev. B, 80, 144111 (2009). ACKNOWLEDGMENTS Support was provided by the USDoT PHMSA under Competitive Academic Agreement Program No. DTPH5614HCAP03 and DTPH5614HCAP01. Figure 1. Grain boundary wedge apex angle variations over time. Curves (a), (b) and (c) are calculated from the FE simulation using time-dependent corrosion rates from experimental current transients. Data points (a), (b) and (c) are experimental wedge angles from the same corrosion experiments. Error bars represent standard deviations of measured wedge angles in each experiment. Figure 1