PSI - Issue 14

Rakesh Kumar et al. / Procedia Structural Integrity 14 (2019) 668–675 Author name / Structural Integrity Procedia 00 (2018) 000–000

669

2

1. Introduction Premature failure of metals in the presence of hydrogen was reported for the first time in year 1875 (Johnson 1875). After almost one and half century, all mechanisms linked with hydrogen-based failure of metals (Hydrogen Embrittlement) are still elusive, primarily due to the non-traceability of hydrogen within metallic structures caused by its small size and high diffusivity (Robertson et al. 2015). Some of the main mechanisms identified for Hydrogen Embrittlement (HE) include hydrogen enhanced localized plasticity (HELP) (Beachem 1972; , Birnbaum and Sofronis 1994), hydrogen enhanced decohesion (HEDE) (Oriani R.A. 1972; John and Gerberich 1973), adsorption induced dislocation emission (AIDE) (Lynch 2011), Hydride formation and cleavage (Nagumo and Nakamura 2001; Nagumo 2007) etc. These mechanisms may not function independently but are often interlinked with each other making hydrogen embrittlement a multi-faceted complex problem. The already complex scenario becomes incomprehensible when metals under hydrogen environment are exposed to cyclic loading. Such scenarios exist for several critical applications such as nuclear reactors, hydrogen gas storage and transportation infrastructure. Fatigue life of a material is sum of the number of load cycles taken for crack initiation and propagation. Crack nucleation in metallic microstructures generally takes place at the sites of deformation inhomogeneities caused by localized elastic and plastic anisotropy due to different orientation of the grains e.g. grain boundaries (Krupp 2007). The reduced fatigue life of metals under hydrogen environment is a consequence of reduced number of load cycles for crack initiation. Thus understanding the effect of hydrogen on plastic deformation/dislocation motion (HELP) and grain boundary strength (HEDE) is fundamental problem linked with HE. In addition, the grain boundary orientation with respect to loading direction may also play a significant role due to the normal stresses developed at such interfaces. Palumbo et al. (1991) recognized the grain boundary normal stress as one of the many reasons responsible for inter-granular stress corrosion cracking (IGSCC). Grain boundary normal stress is of great importance for low cycle fatigue loading, as this value combined with local plasticity and local hydrogen concentration lead to early crack nucleation. The other factor responsible for IGSCC is hydrogen concentration which reduces grain boundary cohesion energy (Robertson et al. 2015). Furthermore, Novak et al. (2010) with four point bend test have also found a decrease in the nominal stress leading to fracture as a function of increasing hydrogen concentration in steel. To improve our understanding of fatigue failure of metals in the presence of hydrogen, a computation model is desired to elucidate the effect of various material parameters on hydrogen related damage mechanics. This paper presents our work in this direction with an aim of identifying crack nucleation sites in metals considering the interaction of hydrogen with dislocations and grain boundaries. Despite of few available computational models in literature to illustrate the hydrogen embrittlement at crystal level (Rimoli and Ortiz 2010; Castelluccio et al. 2018; Schebler 2011), the present computational model provides an unique framework of nonlocal dislocation density based crystal plasticity model and upgraded hydrogen transport model. The nonlocal crystal plasticity model considers dislocation density as a natural variable during the inelastic deformation of the material. The contribution of statistically stored dislocations (SSDs) and geometric necessary dislocations is considered towards total dislocation density at any material point. The SSDs evolve due to the random trapping of mobile dislocation which can form dipoles and increase their density or can annihilate leading to reduction in the dislocation density. The geometrically necessary dislocations arise to accommodate the dimensional inhomogeneity at the triple points, grain boundaries etc. and can be recognized by the gradient of plastic strain. Hydrogen transport model, on the other hand, accounts for diffusion and trapping behavior of hydrogen in metal due to concentration gradient, pressure gradient and plastic slip-rate. For the sake of simplicity, dislocations are considered as the sole trapping sites in the material, but it is recognized that there may be other trapping sites in the material e.g. vacancies, micropores, grain boundaries, particle-matrix interface etc. The model is used to simulate the fatigue crack initiation behavior observed during low cycle fatigue experiments of hydrogen-charged polycrystalline nickel specimens (Arora et al. 2019). However, the crystal plasticity model parameters are first calibrated using the results of single crystal tensile experiments available in literature for hydrogen-charged and uncharged specimens of nickel, as shown in Fig. 1(a) (Yagodzinskyy et.al. 2008). It is reported that due to hydrogen charging of the material, an additional force is required for dislocations to overcome the pinning due to hydrogen which causes rise in stage 1 hardening of the material. This rise is about 20% in single crystals of nickel. Similarly, fatigue experiments performed in our group using shallow notch specimens showed