The link between Microstructural Heterogeneity, Diffusivity, and Hydrogen Embrittlement

TL;DR

This work tackles how microstructure governs hydrogen redistribution and embrittlement under multiaxial loading for hydrogen infrastructure. It introduces a stochastic, multiscale finite element framework that couples crystal-scale plasticity with continuum diffusion and hydrogen trapping, enabling mm-scale analyses to reflect grain-scale effects. Key findings show that, in elastically anisotropic cubic metals, hydrostatic-stress fluctuations follow Gaussian statistics and plasticity broadens microstructural hydrogen gradients; for metals with very low diffusivity like 316L, microstructure dominates diffusion, while nickel exhibits significant interaction between microstructure and mm-scale diffusion across strain rates. The framework provides a quantitative, scalable tool to predict hydrogen embrittlement and informs material design strategies for hydrogen-resistant alloys.

Abstract

Green hydrogen is likely to play a major role in decarbonising the aviation industry. It is crucial to understand the effects of microstructure on hydrogen redistribution, which may be implicated in the embrittlement of candidate fuel system metals. We have developed a stochastic multiscale finite element modelling framework that integrates micromechanical and hydrogen transport models, such that the dominant microstructural effects can be efficiently accounted for at millimetre length scales. Our results show that microstructure has a significant effect on hydrogen localisation in elastically anisotropic materials, which exhibit an interesting interplay between microstructure and millimetre-scale hydrogen redistribution at various loading rates. Considering 316L stainless steel and nickel, a direct comparison of model predictions against experimental hydrogen embrittlement data reveals that the reported sensitivity to loading rate is strongly linked with rate-dependent grain scale diffusion. These findings highlight the need to incorporate microstructural characteristics in the design of hydrogen resistant materials.

Figures (10)

• Figure 1: Overview of the CPFE model. (a) Mesh and boundary conditions imposed on a 100 $\times$ 100 $\times$ 100 $\upmu$m$^3$ representative polycrystal model. (b) Comparison of bulk mechanical response of polycrystal model (using 316L properties) with 316L experimental data KANG2010. The 10 $\upmu$m and 20 $\upmu$m grain diameter models yielded identical mechanical responses. A subfigure of the 291 grain polycrystal model is shown, in which each colour represents a grain with unique crystallographic orientation.
• Figure 2: The relative effects of elastic and plastic anisotropy on the redistribution of hydrogen at saturation, i.e., as $t\to\infty$ within polycrystalline 316L. (a) Mechanical response of 316L with and without plasticity up to 330 MPa (equivalent to 2% strain with plasticity). (b) Typical saturated distribution of hydrogen in untextured polycrystalline 316L after deformation; note that high and low values correspond to high and low values of hydrostatic stress, the locations of which are mostly near GBs and other stress concentration features. (c) As the distribution of hydrogen throughout the microstructure is found to conform well to a normal distribution, this subfigure shows a typical isosurface plot of regions within the microstructure which contain hydrogen concentrations beyond 3 standard deviations of the mean. (d) - (f) Frequency distributions of the relative change in lattice occupancy ($\theta_{\mathrm{L}}/\theta_{\mathrm{L}}^{\mathrm{ini}}$) at stresses 250, 300, and 330 MPa, respectively, with and without plasticity.
• Figure 3: The relationship between the mean von Mises stress across an elastically deformed polycrystal with cubic crystallography and the standard deviation of the normally distributed hydrostatic stress. Extreme examples of pure hydrostatic, pure shear, and uniaxial loading are given. The linear relationship is characterised by the stress ratio, $\omega$.
• Figure 4: (a) Comparison of the predicted stress ratio for randomly textured cubic polycrystals (for unique combinations of elastic constants, $C_{11}$, $C_{12}$, and $C_{44}$) with CPFE-predicted values. Values specific to nickel and 316L are also included. (b) Comparison of analytical and CPFE models in capturing the evolution of the hydrostatic stress distribution with increasing mean von Mises stress in fully elastic, transitionary, and plasticity-dominated regimes.
• Figure 5: Capturing the strain-rate sensitivity of relative lattice occupancy distributions for a material with arbitrary mechanical and hydrogen transport properties using analytical modelling. Results for polycrystals with mean grain sizes of (a) 10 $\upmu$m and (b) 20 $\upmu$m are shown. Legend labels 'CP' and 'Bulk' correspond to detailed micromechanical (crystal plasticity) modelling results (considered ground truth) and microstructure-enriched continuum modelling results, respectively.