Figure 2 presents the ToF-SIMS analysis of the elemental in-depth distribution for the native oxide-covered sample. In Figure 2a, the intensities of selected secondary ions characteristic of the oxide film (¹⁸O⁻, CrO2-, FeO2-, NiO2-, and MoO2-) and substrate (Cr2-, Fe2-, and Ni2-) are plotted in logarithmic scale vs. sputtering time. The position of the “modified alloy” region between the oxide film and metallic bulk substrate regions, where Ni is found enriched in agreement with previous studies on austenitic stainless steels (Olefjord and Elfström, 1982; De Vito and Marcus, 1992; Olsson and Hörnström, 1994; Hakiki et al., 1998; Maurice et al., 1998, 2015; Bojinov et al., 2001; Yamamoto et al., 2009; Wang et al., 2019), was defined using the Ni2- ions intensity profile and placed between 70 and 130 s of sputtering. This region is in principle metallic and does not contain any oxide species in a stratified model of the surface composition. The residual oxide species observed in this region in the ToF-SIMS profile may, at least partially, result from a profiling artifact caused by surface roughness due to the fact that the primary ions and sputtering ions impinge the surface at different angles. In the oxide film region, the most intense profile of the alloying elements are those of the CrO2- and FeO2- ions followed by the MoO2- ions. The NiO2- ions are the least intense.

Figure 2b shows that the FeO2- and CrO2- ions profiles peak at different positions, at 20 and 40 s, respectively. This is consistent with the native oxide film having a bilayer structure with iron and chromium oxides more concentrated in the outer and inner layers, respectively, as commonly reported for oxide films on stainless steels (Marcus and Olefjord, 1988; Mischler et al., 1991; Maurice et al., 2015; Wang et al., 2019). The interface between outer and inner layer was positioned at 30 s which is the median sputtering position between the two intensity maxima (Wang et al., 2019). Figure 2c shows that the MoO2- ions profile peaks at the same position as the FeO2- profile, meaning that Fe oxide and Mo oxide are concentrated in the outer layer and confirming previous data for native oxide film on 316L SS (Wang et al., 2019). Figure 2d shows that, starting from the extreme surface, the CrO2-/FeO2- intensity ratio continuously increases in the outer and inner parts of the oxide film before reaching saturation in the inner layer.

For the sample treated at U_OCP (Figure 3), the “modified alloy” region was positioned between 45 to 110 s using the same method, suggesting a significant thickness decrease of the oxide film. The FeO2- and CrO2- intensity maxima are at 12 and 20 s of sputtering, respectively, with the median position at 16 s for the interface between outer and inner layers of the oxide film (Figure 3b). Compared to the native oxide film, the outer layer is much thinner (16 instead of 30 s) and the inner layer less altered (33 instead of 40 s) after treatment at U_OCP. The maxima of the MoO2- and FeO2- ions profiles are still observed at the same sputtering time in the outer layer, however with a marked decrease in intensity for the FeO2- ions compared to the MoO2- ions suggesting the marked preferential loss of iron oxide (Figure 3c). Consistently, the profiles of the CrO2-/FeO2- intensity ratio, still increasing from extreme surface to inner part in the oxide film region, is overall higher (~6 vs. ~3 at saturation in the inner layer), suggesting an increase of the Cr enrichment of the oxide film due to iron oxide dissolution induced by immersion in sulfuric acid at open circuit potential.

For the sample treated at U_PASS (Figure 4), the “modified alloy” region is observed between 18 and 36 s using the same positioning method. However, due to the smaller sputtered area during analysis (300 × 300 μm² instead of 600 × 600 μm²), the sputtering rate was ~4 times higher in this experiment. After correction, this “modified alloy” region is between 72 and 144 s, which is similar to that of the native oxide-covered sample (70–130 s). Figures 4b,c shows that the bilayer structure persists after passivation with molybdenum concentrated in the outer layer. The interface between outer and inner layer was positioned at 28 s (after correction of the sputtering rate), meaning that the thickness of passive film is quite similar to that of the native oxide film and larger than after treatment at U_OCP. The CrO2-/FeO2- intensity ratio shows a similar increasing profile than for the other two samples and reaches the value of ~6 at saturation in the inner layer, like for the film after treatment at U_OCP and suggesting similar enrichment in Cr in the inner layer of the passive film.

These data confirm that the in-depth chemical structure of the surface oxide films formed on 316L SS is bi-layered. This duplex structure is already formed in the initial native oxide film formed in air and persists after treatment at U_OCP and U_Pass. Chromium oxide accumulates in the inner part to form the barrier layer of the oxide film whereas iron and molybdenum oxide accumulates in the outer part of the film that can be described as the exchange layer in contact with the environment. In sulfuric acid, the observed changes are the positions of the interfaces, reflecting the thickness variations of the oxide films, and the intensity ratios of the secondary ions characteristic of the oxidized alloying elements present in the films and reflecting the variations of the Cr and Mo enrichments induced by immersion and anodic passivation. Iron oxide preferentially dissolves but this is partially counter-balanced by anodic oxidation. The XPS data presented hereafter allow us to quantitatively discuss these electrochemically induced thickness and compositional changes.

The XPS Cr 2p_3/2, Fe 2p_3/2, Mo 3d, and O 1s core level spectra recorded for the native oxide-covered sample are shown in Figure 5 and those measured after treatment at U_OCP and U_Pass are shown in Figures 6, 7, respectively. Table 1 compiles the values of the binding energies (BE), full widths at half maximum (FWHM) and relative intensities of the component peaks used for reconstruction of the experimental curves by curve fitting. It also includes the measured Ni 2p3/2, S 2s, and S 2p peak components.

Following a previously proposed procedure (Yamashita and Hayes, 2008; Biesinger et al., 2011; Wang et al., 2019), the Fe 2p_3/2 spectra were fitted with seven peaks, as exemplified in Figure 5 and detailed in Table 1. The first peak (Fe1) corresponds to metallic Fe⁰ in the substrate. The next five peaks (Fe2-Fe6) form a series at fixed BE intervals, FWHMs, and relative intensities and correspond to Fe^III oxide in the surface oxide layers as previously determined from reference samples (Marcus et al., 1980; Haupt et al., 1985; Yamashita and Hayes, 2008; Biesinger et al., 2011). The additional peak (Fe7), needed at higher BE to optimize the fit, corresponds to Fe^III hydroxide also in the oxide films. No series of five peaks, at lower BEs and associated to Fe^II oxide (Biesinger et al., 2011), was needed for reconstruction. After treatment at U_OCP (Figure 6, Table 1), the Fe^III to Fe⁰ intensity ratio markedly decreases, reflecting the loss of oxidized iron in the surface oxide film in agreement with the ToF-SIMS data. Within the range of ±0.1 eV allowed for fitting, one observes no significant changes of the BE and FWHM values of the metallic and oxidized iron components. After treatment at U_Pass (Figure 7, Table 1), the Fe^III to Fe⁰ intensity ratio increases, indicating re-oxidation of iron. The BE and FWHM values remain unchanged.

Consistently, and also as proposed before (Biesinger et al., 2011; Payne et al., 2011; Wang et al., 2019), seven peaks were also used for reconstructing the Cr 2p spectra (see e.g., Figure 5, Table 1). The first one (Cr1) is associated to metallic Cr⁰ in the substrate, the next five peaks (Cr2-Cr6), also forming a well-defined series as previously determined from reference samples (Yamashita and Hayes, 2008; Biesinger et al., 2011; Payne et al., 2011), to Cr^III oxide in the surface oxide layers, and the additional peak (Cr7) to Cr^III hydroxide in the oxide film. No Cr^VI peak expected at a BE of ~579.5 eV (Biesinger et al., 2011) was needed for curve fitting. After treatment at U_OCP (Figure 6, Table 1), the Cr^III(ox) to Cr⁰ intensity ratio slightly decreases but the Cr^III(hyd) to Cr⁰ intensity ratio markedly increases, indicating significant hydroxylation of chromium in the surface oxide. After treatment at U_Pass (Figure 7, Table 1), the Cr^III(ox) to Cr⁰ intensity ratio increases markedly but the Cr^III(hyd) to Cr⁰ intensity ratio only slightly, indicating re-oxidation and dehydroxylation of chromium. BE and FWHM values show no significant variations after treatment at U_OCP and U_Pass.

The Mo 3d spectra were fitted with three 5/2–3/2 spin-orbit doublets (see e.g., Figure 5, Table 1; Yang et al., 1984; Brooks et al., 1986; Clayton and Lu, 1989; De Vito and Marcus, 1992; Di Castro and Ciampi, 1995; Biesinger et al., 2011; Maurice et al., 2015; Wang et al., 2019). One (Mo1/Mo1') corresponds to metallic Mo⁰ in the substrate and the other two, (Mo2/Mo2′) and (Mo3/Mo3′), to Mo^IV and Mo^VI in the surface oxide layers, respectively. The Mo^VI to Mo^IV intensity ratio is 6.4 in the native oxide film, indicating that Mo(VI) species are mostly present. After treatment at U_OCP (Figure 6, Table 1), the Mo^IV+VI to Mo⁰ intensity ratio slightly increases in agreement with the Mo enrichment in the oxide outer layer observed by ToF-SIMS. It further increases after treatment at U_Pass (Figure 7, Table 1). The Mo^VI to Mo^IV ratio also increases after treatment at U_OCP (7.1) and further after treatment at U_Pass (7.4), indicating further enrichment in Mo(VI) species after electrochemical treatment. An additional peak S1, assigned to the S 2s core level, was necessary for reconstruction of the Mo 3d spectra measured after the electrochemical treatments (Figures 6, 7, Table 1). The presence of sulfur originating from the sulfuric acid electrolyte was confirmed by the presence of a S 2p doublet assigned to sulfate species (Table 1), most likely adsorbed at the surface of the oxide films (Table 1).

The Ni 2p_3/2 spectra were fitted using a single peak corresponding to metallic Ni⁰ in the substrate (Table 1; Haupt and Strehblow, 1995; Maurice et al., 2015; Wang et al., 2019), evidencing that the Ni oxide species measured by ToF-SIMS were at trace level below the detection limit of XPS in all three cases (< 0.5 at%).

The O 1s spectra were fitted with three components O1, O2, and O3 assigned to oxide (O²⁻), hydroxide (OH⁻) and water (H₂O) ligands in the surface oxide films, respectively (see e.g., Figure 5, Table 1; Clayton and Lu, 1986; Castle and Qiu, 1989b; Biesinger et al., 2011; Maurice et al., 2015; Wang et al., 2019). The OH⁻/O²⁻ intensity ratio increases from 0.58 for the native oxide film to 1.72 after treatment at U_OCP and then decreases to 1.24 after treatment at U_Pass. Similarly, the H₂O/O²⁻ intensity ratio increases from 0.12 to 0.68 and then decreases to 0.39. Clearly, these variations reflect the markedly increased hydroxylation of the initial air-formed oxide film after immersion in the aqueous solution at OCP, showing that the oxide film not only decreases in thickness and gets enriched in Cr and Mo due to preferential iron oxide dissolution but also incorporates much more hydroxyl and water ligands. Anodic passivation causes competitive re-oxidation of the metallic elements leading to the oxide film growth in thickness, with the preferential formation of oxide ligands since the fractions of hydroxyl and water ligands decrease.

Based on the ToF-SIMS results, the 2-layer model, previously proposed to calculate the thickness and composition of the outer and inner layers of the oxide films as well as the composition of the modified alloy underneath the oxide films (Wang et al., 2019), was used to process the XPS intensity data. This model assumes a mixed iron-chromium hydroxide outer layer and a mixed iron-chromium oxide inner layer. Molybdenum oxide is included in the outer layer and neglected in the inner layer. Assignment of the intensities of the different components was as follows: Cr2-Cr6 and Fe2-Fe4 components to the oxide film inner layer, Cr7, Fe5-Fe7, and Mo2-Mo3 components to the oxide film outer layer and Cr1, Fe1, Mo1, and Ni1 components to the modified alloy region underneath the oxide film. The results are presented in Table 2. The overall compositions of the oxide films were obtained by weighting the cation concentration value of each element by the fractional thickness of the inner and outer layers.

Let us first discuss the initial native oxide film. The total thickness is found to be 2.2 nm with outer and inner layers of 0.7 and 1.5 nm thickness, respectively. This total value is very close to that (2 nm) found following the same procedure on the same as-prepared surface of polycrystalline 316L samples (Wang et al., 2019). The overall composition of the film shows that chromium is already enriched in the air-formed native oxide film with Cr^III ions representing 41% of the metal cations in the film. Consistently, Cr⁰ is found depleted (17 at% instead of 20 at% in the bulk alloy) in the modified alloy underneath the oxide film. Nickel, not in measurable amount in the oxide film, is confirmed to be enriched in the alloy underneath the oxide film (26 at% instead of 12 at% in the bulk alloy) in agreement with the ToF-SIMS profile. Concerning molybdenum, the calculated composition shows that it is enriched in the native oxide film (4 at% globally and 13 at% in the oxide outer layer instead of 1.6 at% in the bulk alloy). It is also found enriched in the alloy underneath the oxide film (4 at% instead of 1.6 at% in the bulk alloy). Compared to the more hydroxylated native oxide film formed on the same as-prepared surface of polycrystalline 316L samples (OH⁻/O²⁻ intensity ratio of 0.93 instead of 0.58 in the present case; Wang et al., 2019), the present data show lower Cr^III (41 vs. 55% of the metal cations) and Mo^IV/VI (4 vs. 6%) enrichments in the oxide film. This might be due to an aging effect, some studies having reported a decreasing Cr/Fe balance in the native oxide with aging in air (Yang et al., 1994; Maurice et al., 1996, 1998).

After treatment in sulfuric acid at U_OCP, the results reported in Table 2 confirm the overall thickness decrease of the oxide film observed by ToF-SIMS. However, this decrease affects the inner layer only, in contrast with the ToF-SIMS data. The overall composition of the film shows the loss of Fe and the related increases of the Cr and Mo enrichments suggested by ToF-SIMS but the hierarchy of the Cr/Fe ratio between outer and inner layers is not respected. These discrepancies originate from the assignment of the Cr7 component, markedly increased in intensity. Since immersion at OCP causes iron oxide dissolution and pronounced hydroxylation of the film, it can be reasonably assumed that not only the outer layer but also the inner layer is hydroxylated, and thus that a fraction of the intensity of the Cr7 component can be assigned to the inner layer. The values reported in brackets in Table 2 have been calculated assigning 50% of the Cr7 intensity to the oxide inner layer. The hierarchy of the Cr/Fe ratio between outer and inner layers is now reproduced as well as the decrease in thickness of both the outer and inner layers. Based on these assumptions, the composition in the outer and inner layers are calculated as 26Fe^III-48Cr^III-26Mo^IV/VI and 23Fe^III-77Cr^III, respectively, vs. 82Fe^III-5Cr^III-13Mo^IV/VI and 42Fe^III-58Cr^III for the native oxide. The overall composition of the film also reflects the compositional changes, independently of the balance in the assignment of the Cr7 peak. The Cr^III and Mo^IV/VI concentrations increase up to 68.5 and 7.6% in the oxide film, respectively, after treatment at U_OCP, as a result of the preferential loss of iron oxide by dissolution. In the modified alloy underneath the oxide film, the calculated values confirm the Ni enrichment and suggest no preferential consumption at U_OCP of one of the alloying elements if one considers a ±1 at% accuracy.

Clearly, these data show that the loss of oxide is caused by the preferential dissolution of iron oxide at open circuit potential. It is also shown that oxide growth is too slow to compensate the loss of oxide due to dissolution and, if occurring, does not preferentially consumes one of the alloying elements. To our knowledge, this is observed for the first time on an oxide pre-covered SS surface thanks to the transfer from liquid to UHV environments avoiding the contact with ambient air. Such a procedure was previously applied to ferritic Fe-17Cr samples but starting from an oxide-free surface, not an oxide pre-covered surface, and showed the preferential growth of Cr(III) oxide at anodic potential (Keller and Strehblow, 2004). The present data highlight that the alterations of the surface oxide film caused by immersion at open circuit potential results from the preferential dissolution of iron oxide.

After passivation at U_PASS and compared to the treatment at U_OCP, the overall thickness of the oxide film increases with both partitions growing (Table 2), which confirms the ToF-SIMS observation. The steady state thickness of the passive film increases due to faster growth of the surface oxide film. There are no significant variations of the overall composition of the oxide film and of the composition of the modified alloy region if one considers a ±1 at% accuracy of the calculated values, indicating that all alloying element are consumed at U_PASS but with iron oxide preferentially dissolving in order to maintain the Cr enrichment. Cr remains enriched in both layers of the passive film and more predominantly in the inner layer. In the hydroxide outer layer, Mo would be less enriched than before re-growth of the oxide film under anodic polarization.

Compared to the native oxide film, there is no significant thickness variation after passivation at U_PASS. However, the composition of the oxide film is markedly modified with further Cr(III) enrichment in both layers and Mo(IV-VI) enrichment in the outer layer, in agreement with the previous studies on the same polycrystalline 316L surface (Wang et al., 2019) and on a single-crystalline model Fe-Cr-Ni-Mo surface (Maurice et al., 2015). Thanks to the transfer procedure adopted here and avoiding the contact with ambient air, it is shown here that the preferential dissolution of Fe(III) is at the origin of these enrichments and essentially takes place at open circuit potential in the absence of any applied anodic polarization. Anodic passivation causes competitive re-growth of the oxide film and dehydroxylation of Fe(III), Cr(III) and Mo(IV-VI) oxide species It increases the steady state thickness of the passive film despite dissolution still occurring and counteracting the formation of the oxide species. The preferential dissolution of Fe(III) oxide species persists to maintain the Cr and Mo enrichments.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.