*3.1. EBSD*

Figure 1 shows EBSD maps of the two studied alloys shown here to highlight the grain structure of the material. These were obtained by using a Schottky type FE-SEM (JEOL JSM-7001F) at an acceleration voltage of 15 kV. The grains have a similar size (around 100 μm) and a similar amount of twins can be observed in both samples so the microstructure complies with our expectations for these very well known stainless steels. This value of the grain size is rather large compared to the sample diameter, but there are at least ten grains along the diameter so a total number of about 100 grains in a given section of the sample. This is a sufficient number of grains to insure that the mechanical behavior is not strongly influenced by plasticity gradient effects. The assessment of the texture would be interesting but is out of the scope of this paper.

**Figure 1.** EBSD maps of the two different materials (**a**) non-charged AISI316L steel; (**b**) non-charged AISI316 steel; (**c**) inverse pole figure (IPF) coloring.

#### *3.2. Hardening Curves from In Situ X-Ray Computed Tomography*

The sample was mounted vertically in the rig, more or less aligned with the tensile axis, which in turn is more or less parallel to the rotation axis of the rotation stage. This axis is denoted "*z*" in the following. As already performed in [15,24], from the outer shape of the sample measured using X-ray tomography after segmentation at each deformation step, we could measure the section *S* of the sample (*S* being perpendicular to *z*) as a function of the position of this section along *z*. From this list of sections *<sup>S</sup>*(*z*), we could determine the coordinates of the location (center of mass) of the minimal section of the sample *Smin*. Because we recorded *F* at all times, it was then possible to precisely calculate the true stress *σ* inside this minimal section using the expression:

$$
\sigma = \frac{F}{S\_{\min}}.\tag{1}
$$

Assuming no volume dilation of the sample due to damage, it was also possible to precisely calculate the true longitudinal strain, *ε*, in *Smin* using the following standard expression:

$$\varepsilon = \ln(\frac{S\_{min}^0}{S\_{min}}),$$

where *<sup>S</sup>*0*min* is the value of *Smin* in the initial tensile state. Figure 2 shows the true stress—true strain curves (the hardening curves) recorded during our experiments for all the samples tested. It should in principle start at zero true plastic strain, but, in our case, true strain and true stress were measured at each stop during the interrupted test, with a first step at *ε* close to 0.2. We have no precise measurement of the yield stress of these samples during the in situ tensile test.

**Figure 2.** Tensile curve of the different samples (true stress vs. true strain).

The hardening curves are close together with AISI316 hardening a bit more, probably due to its higher C content. We have gathered the ductility values measured from these in situ tensile tests referred to as *f* in Table 3. We also calculated in this table the decrease in ductility induced by hydrogen charging DDH, calculated as:

$$DDH = \frac{(\varepsilon\_{\text{Non-charged}}^f - \varepsilon\_{\text{Hydrogen}-\text{charged}}^f)}{(\varepsilon\_{\text{Non-charged}}^f)} \times 100. \tag{3}$$

**Table 3.** Ductility measured during the in situ tensile tests and calculation of the Decrease in Ductility due to Hydrogen charging (DDH) for the two materials and the two specimen shapes.


DDH is clearly higher for the AISI316 than for the AISI316L.

#### *3.3. Qualitative Damage Evolution*

Figure 3 shows (as a selected representative example) a volume rendering of the evolution of the cavities in the non-charged AISI316L. It shows similar features compared to what was already observed in [17] on the same type of material. Voids, nucleate grow and coalesce during the severe plastic deformation of the sample, especially in the central region of the notch. This typical evolution is also observed in all the different tested samples and will be quantified further in a subsequent section.

**Figure 3.** Damage evolution in the non-charged AISI316L sample observed as a volume rendering. The outer surface of the sample is transparent grey and the cavities are the dark red dots. Damage nucleates then grows and finally coalesce in the bottom right image (**a**) initial state; (**b**) true strain = 0.6; (**c**) = 0.77; (**d**) = 1.17; (**e**) = 1.48; (**f**) = 2.40.

By analyzing carefully volume series like the one shown in Figure 3 for every type of sample, qualitative differences were observed depending on the nature of the material and hydrogen charging. To highlight these differences, Figure 4 compares reconstructed slices of four typical samples in the

last step before fracture. It is very clear from these images that the non-charged samples (left column) exhibit a much higher ductility, the section reduction and necking being much higher. Voids have then nucleated, grown and coalesced (see the big coalescence event observed for the non-charged AISI316 sample). This behavior is very typical of ductile metals. The right column shows the effect of hydrogen-charging on the final deformation stage. As already highlighted by Table 3, ductility was clearly reduced (necking is much less pronounced). Micro-cracks perpendicular to the tensile axis are observed in these reconstructions (as can be seen in Figure 5a) in the AISI316 sample but to a lesser extent in the AISI316L.



**Figure 4.** Reconstruction slices extracted parallel to the tensile axis in a central plane for the four samples in the ultimate state before fracture. AISI316 is clearly less deformed at fracture and contains local cleavage microcracks when hydrogen charged. The deformation in the different images are (**a**) = 2.40 for non-charged AISI316L; (**b**) = 1.36 for hydrogen-charged AISI316L; (**c**) = 1.92 for non-charged AISI316; (**d**) = 0.95 for hydrogen-charged AISI316.

Figure 5 compares 3D renderings of the hydrogen-charged AISI316 and AISI316L in the final stage just before coalescence. The figure clearly show that the AISI316 exhibits many more microcracks than the AISI316L sample in which the voids are elongated along the tensile axis. In the AISI316, cracking from surface can be observed.

**Figure 5.** Volume rendering (**a**) hydrogen-charged AISI316 sample just before coalescence. The amount of penny shaped microcracks is very important. The deformation in the image is = 0.79; (**b**) hydrogen-charged AISI316L sample just before coalescence. The morphology of the cavity is elongated along the tensile direction. The deformation in the image is = 1.15.

#### *3.4. Quantitative Damage Evolution*
