Mechanism and Prediction of Hydrogen Embrittlement in fcc Stainless Steels and High Entropy Alloys

The urgent need for clean energy coupled with the exceptional promise of hydrogen (H) as a clean fuel is driving development of new metals resistant to hydrogen embrittlement. Experiments on new fcc high entropy alloys present a paradox: these alloys absorb more H than Ni or SS304 (austenitic 304 stainless steel) while being more resistant to embrittlement. Here, a new theory of embrittlement in fcc metals is presented based on the role of H in driving an intrinsic ductile-to-brittle transition at a crack tip. The theory quantitatively predicts the H concentration at which a transition to embrittlement occurs in good agreement with experiments for SS304, SS316L, CoCrNi, CoNiV, CoCrFeNi, and CoCrFeMnNi. The theory rationalizes why CoNiV is the alloy most resistant to embrittlement and why SS316L is more resistant than the high entropy alloys CoCrFeNi and CoCrFeMnNi, which opens a path for the computationally guided discovery of new embrittlement-resistant alloys.

Figure 1

Mechanism of HE in complex alloys. (a) Schematic of embrittlement process at the crack tip (small spheres: “columns” of H atoms with colors indicating the projected H concentration). Embrittlement occurs because H in sites 4 and 7 “nano” diffuses to the newly created crack surface, increasing the surface H concentration and decreasing the fracture energy, while H in sites 10 and 11 blocks dislocation emission. (b) Schematic of the crack tip H concentrations $C_{\text{cleave}}$ and $C_{\mathrm{emit}}$ that control embrittlement as a function of $K_I^{\mathrm{app}}$ (computed for SS304 at bulk H concentration $C_b=2600\text{at.p}\mathrm{pm}$ , $\mathrm{T}=300\mathrm{K}$). (c) Predicted embrittlement (in SS304, $\mathrm{T}=300\mathrm{K}$) due to nanodiffusion and blocking of dislocation emission: $K_I^{\mathrm{app}}$ (black) reaches $K_{\mathrm{Ic}}$ (red) prior to reaching $K_{\mathrm{Ie}}$ (blue). (d),(e) Embrittlement is not predicted in the absence of blocking emission or in the absence of nanodiffusion [in both cases, $K_I^{\mathrm{app}}$ (black) reaches $K_{\mathrm{Ie}}$ (blue) prior to reaching $K_{\mathrm{Ic}}$ (red)].

Figure 2

(a) Alloy elastic constants, anisotropic elastic coefficients, and mean $\overline{E}$ and standard deviation $\sigma$ of the bulk H absorption energy distribution. (b) ${\gamma }^F$ at surface coverage $C_S=0\%$, 50%, 100% on the (111) surface for $C_b=5800\text{at.}\mathrm{ppm}$. (c) ${\gamma }^{\mathrm{usf}}$ at fault coverage $C_{\mathrm{usf}}=0\%$, 50%, 100% on the (111) slip plane on an area of $108{\mathrm{\AA }}^2$; the mean values ${\overline{\gamma }}^{\mathrm{usf}}$ and standard deviations ${\sigma }_{\mathrm{usf}}^{\mathrm{DFT}}$ are shown.

Figure 3

Embrittlement predictions for fcc alloys. Critical stress intensities $K_{\mathrm{Ic}}$, $K_{\mathrm{Ie}}$ (upper, lower limits) as a function of $K_I^{\mathrm{app}}$ for all alloys studied, as predicted using the experimentally reported bulk H concentration $C_b^{\mathrm{expt}}$. Theory predictions are determined by the black dot—embrittled if $K_I^{\mathrm{app}}=K_{\mathrm{Ic}}$ first, not embrittled if $K_I^{\mathrm{app}}=K_{\mathrm{Ie}}$ first—and uncertain (three dots shown) if the intersection of $K_I^{\mathrm{app}}=K_{\mathrm{Ic}}$ lies in between the upper and lower $K_{\mathrm{Ie}}$. Experimental observations of embrittlement are stated, and cases where theory and experiment agree are highlighted by shading of the legend.

Figure 4

Predicted embrittled (red) and unembrittled (blue) domains of H concentration for six alloys and Ni along with experimental findings. The transition region corresponding to the upper and lower limits of $K_{\mathrm{Ie}}$ is indicated by the thick black lines.