Multiscale computational analysis of crack initiation at the grain boundaries in hydrogen-charged bi-crystalline alpha-iron

This paper presents a mesoscale concurrent atomistic–continuum (CAC) simulation of crack initiation at the atomically structured grain boundaries (GBs) in bi-crystalline BCC iron ($\alpha$-Fe) charged with hydrogen (H). By retaining the atomistic GB structure evolution together with the long-range dislocation-mediated plastic flow away from the GB in one model at a fraction of the cost of full molecular dynamics (MD), CAC enables us to probe the interplay between the atomic-level H diffusion, the nanoscale GB cavitation, crack initiation, growth, as well as the dislocation activities far away from the GB. Our several main findings are: (i) a tensile strain normal to the GB plane largely promotes the H diffusion towards the GB. (ii) the plasticity-induced clustering of H atoms (PICH) is identified as an intermediate process in between the H-enhanced localized plasticity (HELP) and H-enhanced de-cohesion (HEDE). (iii) PICH significantly amplifies the local stress concentration at the GB and decreases its cohesive strengths, and (iv) the GBs with different atomic structures fail differently. In detail, the H-charged $\Sigma 3$ GB fails through micro-twinning assisted void nucleation and coalescence, while the H-charged $\Sigma 9$ GB fails through crack initiation and growth accompanied by dislocation emission. Compared with nanoscale molecular dynamics (MD) simulations, the mesoscale CAC models get one step closer to the experimentally measurable length scales and thus predict reasonably lower GB cohesive strengths. This research addresses one key aspect of how H impacts the GB cohesive strengths in $\alpha$-Fe. It offers insights into the multiscale processes of hydrogen embrittlement (HE). Our findings highlight the importance of using concurrent multiscale models, such as a combination of CAC, crystal plasticity finite element (CPFE), and cohesive zone finite element method (CZFEM), to understand HE. This will, in turn, support the development of new strategies for mitigating HE in a variety of engineering infrastructures.