On the hydrogen embrittlement of an additively manufactured CoCrFeMnNi high entropy alloy

Abstract

The field of study on the hydrogen embrittlement of CoCrFeMnNi high entropy alloy manufactured by laser powder bed fusion are wrought with issues related to short electrochemical charging durations leading to low hydrogen ingress into the sample. This cascades to the subsequent analysis of the hydrogen trapping and diffusion parameters which are essential in characterising hydrogen behaviour in successfully designing hydrogenresistant alloys. That is, trapping and diffusion analysis are mostly done with simplifying assumptions that are not validated such as whether the diffusion is lattice dominated or dissociation -controlled. Moreover, there is a lack of understanding on the role of additive manufacturing microstructure in hydrogen trapping and embrittlement and lack of reports that identify the challenges in mapping hydrogen-alloy interactions using atom probe tomography. To address these issues, prolonged electrochemical hydrogen charging to ensure hydrogen saturation of specimens were conducted. Moreover, thermal desorption analysis was utilised as a microstructure sensitive technique to relate microstructural conditions to hydrogen trapping as well as to calculate hydrogen diffusivity for the alloy in the as-printed, annealed and wrought conditions for comparison. Atom probe tomography was employed for direct detection and mapping of hydrogen on microstructural defects and challenges in the successful utilisation of the technique were identified. The prolonged hydrogen charging at elevated temperature resulted to 0.5 mm-thick specimens to be saturated with hydrogen with approximately 100% brittle fracture zone in the gauge fracture surface, and increased hydrogen penetration depth in 1 mm thick samples. The hydrogen saturation entails maximum trap occupancy allowing for the validity of the assumption of local equilibrium necessary for the correct application of the McNabb and Foster and Oriani models of hydrogen trapping. With these conditions satisfied, it was shown that the alloy in both as-printed and annealed conditions contains mostly weak, reversible traps of similar binding energies (~20 kJ/mol). However, the annealed samples have a trap density an order of magnitude less than the as-printed specimen. These weak traps were associated with dislocations and grain boundaries and possibly coherent secondary particles such as nano-oxides or nano-precipitates. This correlates to the presence of numerous dislocations arranged in a network of interconnected cells in the as-printed sample which were reduced and disintegrated after annealing. The annealed samples were shown to have significantly less hydrogen content than the as-printed sample with calculated diffusivities of 2.52 x 10^-14 m2/s for the as-printed sample and 1.24 x 10^-15 m2/s for the annealed sample. For reference, the diffusivity for the wrought sample was calculated to be 8.05 x 10^-16 m2/s which is the lowest among the three specimens. The presence of a relatively higher reversible trap density and hydrogen diffusivity manifested in severe hydrogen embrittlement of as-printed samples during slow strain rate testing. Annealed samples showed a significant recovery of ductility, measured in terms of reduction in area, of up to 30%. The decrease in the hydrogen embrittlement susceptibility of annealed samples point to the role of dislocations and dislocation network in hydrogen-assisted cracking. For as-printed samples, the intact interconnected dislocation network with dense dislocations on cell walls resulted to high local hydrogen concentration, facilitating hydrogen mobility and thus, hydrogen-assisted cracking along the cell boundaries and interdendritic regions in addition to grain boundary decohesion due to the high hydrogen concentration. On the other hand, reduction of dislocation density and disintegration of the dislocation network resulted to relatively lower localised hydrogen concentration with hydrogen facilitating the mobility of dislocations instead. This resulted to increased plasticity which can explain the recovery of ductility in the tensile tests. Moreover, the hydrogen concentration may have been reduced to within a range that is enough to lower the stacking fault energy of the atoms and trigger nanotwinning in the annealed samples. Fractography revealed a typical transgranular ductile fracture due to microvoid coalescence in both as-printed and annealed samples without hydrogen. In contrast, hydrogen pre-charged samples exhibited a brittle fracture zone that is mainly intergranular and interdendritic with regions of quasi-cleavage for as-printed samples and a similar fracture surface for annealed samples albeit with the increased presence of ductile features such as microvoids in certain regions and slip traces. As expected, the brittlefracture zone was greater in the as-printed samples. These observations, in conjunction with the analyses of fractographs point to hydrogen enhanced decohesion (HEDE) mechanism as the dominant hydrogen embrittlement mechanism in the as-printed sample whereas for the annealed sample, a synergy of hydrogen enhanced localised plasticity with hydrogen enhanced decohesion, or plasticity-mediated HEDE. Direct detection of hydrogen in microstructural traps by atom probe tomography were met with challenges in sample preparation and deuterium gas charging, with deuterium unable to penetrate into additively manufactured samples and focused ion beam and energetic electrons from Transmission Kikuchi Diffraction inducing damage to the specimen making it difficult to identify inherent traps in the microstructure.