Inhibition of Mg Corrosion by Sulfur Blocking of the Hydrogen Evolution Reaction on Iron Impurities

A Mg foil (rolled, purity 99.9%wt, 60 ppm Fe, Goodfellow, England) with composition reported in Table I was used in this work. The microstructure obtained after chemical etching is presented in Table I. Samples (disks of 15 mm diameter and 1 mm thickness) were mechanically ground on SiC paper from 2400 grit (about 10 μm) to 4000 grit (about 5 μm) then polished on a rotating pad impregnated with alumina suspension (2–3, 1 and 0.3 μm). Between each step, the samples were rinsed with pure ethanol. Then samples were sonicated for 10 min in pure ethanol and finally dried with filtered compressed air.

Table I.  Chemical composition in parts per million (weight ppm) of the Mg sample (99.99 wt%, purchased from Goodfellow) and microstructure obtained after chemical etching.

Element	Al	Cu	Fe	Mn	Ni	Si	Zn	Mg
ppm	50	10	60	180	20	180	30	balance

All the glassware used during this study was cleaned with a Piranha solution (H₂SO₄/H₂O₂).

Electrochemical measurements were performed using a conventional three-electrode cell with a magnesium sample as a working electrode, a platinum wire as a counter electrode and a saturated calomel reference electrode (SCE). Measurements were carried out at room temperature in 0.1 mol.l⁻¹ NaCl (AnalaR Normapur, analytical reagent, VWR) without or with 0.01 mol.l⁻¹ Na₂S·9H₂O (Sigma Aldrich). The pH was adjusted to 6.0 (±0.1) (corresponding to the measured pH of a NaCl solution without H₂S_aq) by addition of concentrated HCl (1 mol.l⁻¹) to make sure that dissolved hydrogen sulfide (H₂S_aq) is the predominant species (pKa (H₂S/HS⁻) = 7).⁴³ Potentiodynamic polarization curves were recorded at a scan rate of 1 mV·s⁻¹ using an EC-Lab SP200 potentiostat (Bio-Logic Science Instruments SAS, France). Anodic and cathodic domains have been scanned separately, starting from the OCP. In parallel, dihydrogen has been collected using a classical setup described in the literature.⁴⁴

The entire surface of Mg samples, before and after electrochemical measurements, was observed by optical microscopy using a digital VHX-5000 Keyence microscope with a magnification from 100 to 1000 (Lens ZR1000). XPS analysis was performed using a Thermo ESCALAB 250 X-ray photoelectron spectrometer with a monochromatic Al-Kα X-ray source (hυ = 1486.6 eV) operating at a pressure of 2.10⁻⁹ mbar. Survey and high-resolution spectra were recorded with a pass energy of 50 and 20 eV, respectively, using Thermo Avantage software version 5.966. The spectrometer was calibrated using Au4f_7/2 at 84.1 eV. During analysis, charging effects were not compensated.

ToF-SIMS analysis was performed using a ToF-SIMS V spectrometer (ION TOF GmbH, Munster, Germany) operating at 10⁻⁹ mbar pressure. Negative ions depth profiles were performed by sequentially analyzing (Bi⁺, 1.2 pA, 100 × 100 μm²) and sputtering (Cs⁺, 500 eV, 20 nA, 500 × 500 μm²) the sample surface. Data acquisition and post-processing analyses were performed using Ion-Spec software (version 4.1). MgO⁻ (39.980 amu), MgOH⁻ (40.988 amu), MgS⁻ (55,957 amu) FeO⁻ (71,930 amu) and FeS⁻ (87,907 amu) negative fragments, corresponding to Mg oxide, Mg hydroxide, Mg sulfide, Fe oxide and/or metal and Fe sulfide, respectively, have been analyzed during the depth profiling.

The combination of these two techniques allows us to obtain a quantitative surface chemical composition with a good sensitivity (<0.50 at%) by XPS (500 scans were used to reduce the signal-to-noise ratio) and obtain qualitative chemical information with a very high sensitivity (a few ppm) and a very good in-depth resolution by ToF-SIMS. XPS and ToF-SIMS are highly complementary techniques.

Figure 1a shows the evolution of the open circuit potential measured for a magnesium electrode immersed in 0.1 mol.l⁻¹ NaCl electrolyte (a, black line) and 0.1 mol.l⁻¹ NaCl electrolyte containing 0.01 mol.l⁻¹ H₂S (b, red line) for short immersion time (30 min). First, a rapid (in the first 5 s) drop of the potential, to −1.75 V_SCE is observed in both electrolytes. Without H₂S, a plateau is reached after 10 min followed by a weak decrease. After the plateau, electrochemical noise is observed and can be correlated to the initiation of the dark corroded area and the formation of H₂. With H₂S, the potential increases slowly and no plateau is observed. After 30 min no H₂ bubble or dark corroded areas are observed. Optical micrographies obtained after 30 min in NaCl electrolyte (Fig. 1C a) show the presence of the well-known dark filiform corrosion areas characteristic of magnesium corrosion in chloride media. These dark corroded areas are accompanied by strong H₂ evolution. In NaCl electrolyte containing H₂S, no dark corroded area (Fig. 1C b) and no hydrogen evolution were observed at the surface of the magnesium electrode. The magnesium sample previously exposed to H₂S containing solution during 30 min was subsequently immersed (without rinsing) in 0.1 mol.l⁻¹ NaCl solution without H₂S (Fig. 1A c). The evolution of the potential presented in Fig. 1A c shows a similar behavior to that of the OCP curve observed for sample immediately immersed in NaCl electrolyte (Fig. 1A a). Optical microscopy images show a strong limitation of the growth of the dark corroded areas and a limited evolution of hydrogen was observed during the immersion (Fig. 1C c). These results show that hydrogen sulfide inhibits the HER and the growth of the dark corroded areas both for Mg sample immersed in chloride solution with H₂S and for Mg sample "conditioned" in chloride containing H₂S and further exposure to chloride solution without H₂S. H₂ has been collected (every 5 min) during immersion at OCP for long time (>3 h) in NaCl electrolyte (Fig. 1B a)) and in NaCl electrolyte containing H₂S (Fig. 1B b)) and H₂ volume measurements are presented in Fig. 1b.

Zoom In Zoom Out Reset image size

After 30 min immersion in NaCl, OCP continues to increase slowly. This increase is not linear and can be split into two domains (30–120 min and 120 min–240 min). This evolution can be correlated to the increase of the collected H₂ evolution which presents also two distinct steps. In the presence of H₂S, OCP goes through a maximum (20 min, −1.53 V SCE⁻¹) before reaching a steady state (−1.56 V SCE⁻¹). In each electrolyte, a final pH of about 11 is measured indicating an alkalization of electrolytes. For each curve, the signal presents a strong noise which appears when H₂ evolution begins as already observed.⁴⁵ Evolution of collected H₂ volume is similar for both electrolytes and it could be split into two domains. In the presence of H₂S, H₂ evolution is always lower (red curve in Fig. 1b). In the first domain (0–100 min), hydrogen evolution is not detectable. A first volume of 0.1 ml (which is the first quantitative value that could be measured with our experimental setup) is collected after an immersion of 40 min in NaCl and 90 min in NaCl + H₂S, showing the mitigating action of H₂S on the H₂ evolution In the second domain (>220 min), the H₂ volume follows an almost linear rise which is smaller in the presence of H₂S (0.114 ± 0.004 ml.cm⁻².min⁻¹ in NaCl solution and 0.034 ± 0.002 ml.cm⁻².min⁻¹ in NaCl with H₂S). In the presence of H₂S, H₂ evolution is reduced by about 70%, showing a significant effect of H₂S on the hydrogen evolution at OCP.

It is to be noted that, for our Mg sample with 60 ppm Fe, in NaCl solution before addition of H₂S the amount of H₂ is high compared to values of the litterature. For example, some papers have reported a small volume of H₂ (1.5 ml.cm⁻² for UHP Mg (Fe: 0.1 ppm)⁴⁶ and 3 ml.cm⁻² for a HP Mg (Fe: 40 ppm)⁴⁴) after 80 h immersion in 0.1 mol.l⁻¹ NaCl solution while other papers reported a volume of 30 ml.cm⁻² in 3.5% NaCl after 10 h immersion (Fe: 45 ppm).²⁷ These different measured H₂ flows are attributed to the difference of purity of samples. Other impurities could modify the rate of H₂ evolution.

After the immersion tests at OCP for short time (30 min), the Mg samples were analyzed by XPS and compared to a freshly polished Mg substrate (Mg 2p in Fig. 2 and O 1s in Fig. 3). The freshly polished Mg sample spectrum (Fig. 2a, blue line) shows two components at 49.8 and 51.6 eV in the Mg 2p core level region, which can be attributed to metallic and oxide/hydroxide layer, respectively, while the third component at the highest energy (60.3 eV) corresponds to the plasmon loss characteristic of metallic magnesium. After immersion in a 0.1 mol.l⁻¹ NaCl solution with or without H₂S, the Mg 2p photopeak (Fig. 2b, black line and 2c, red line) shows only one component at a high binding energy of 52.9 eV. It is assigned to magnesium oxide/hydroxide. The binding energy shift, +1.3 eV with respect to the native oxide, is assigned to a charging effect and indicates a thickening of the insulating Mg oxide/hydroxide layer induced by exposure to the solutions. The thickening of the oxide is also confirmed by the absence of the metallic component at 49.8 eV for samples exposed to NaCl or NaCl + H₂S and indicates an oxide thickness higher than 11 nm (using a dense and continuous magnesium hydroxide layer model; mean free path: ${\lambda }_{Mg2p}^{Mg\left(OH\right)2}\,=\,4.2\,{\rm{nm}}$). The spectra corresponding to the second step in the OCP measurements in 0.1 mol.l⁻¹ NaCl for sample preconditioned 30 min in NaCl + H₂S solution are shown in Fig. 2d (dark green line) and 2e (light green line) for "uncorroded" and dark corroded areas, respectively. The absence of the metallic component at 49.8 eV and the larger energy shift (+3 eV and +5.2 eV for uncorroded and dark corroded areas, respectively) also indicates that the Mg oxide layer thickens both in the "uncorroded" and dark corroded regions. However, the most significant energy shift on dark corroded areas indicates the formation of an oxide or hydroxide thicker than on the so called "uncorroded" areas. These observations have already been reported in our previous paper.²⁴ In addition, the Mg2p spectrum recorded for the dark corroded areas presents a main component characteristic of the Mg oxide/hydroxide and a weak component at low energy (54.5 eV, Fig. 2e, marked by a green arrow). This binding energy already observed for the "uncorroded" areas (Fig. 2d) also corresponds to the binding energy of the MgO/Mg(OH)₂ components. These results, combined with the optical microscopy observations, clearly show that the dark corroded areas are rough and not homogeneous and comprise some "uncorroded" spots.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The O 1s core level spectra were also analyzed (Fig. 3a) and similarly to the Mg 2p spectra, no charge correction was done. Typically, the O 1s spectrum for the freshly polished Mg sample shows three components⁴⁷ (Fig. 3b) corresponding to the oxide Mg–O (531.4 eV), hydroxide Mg–OH (533.3 eV) and oxygen-containing contamination species C_xO_y (535.0 eV) such as carbonate species. It was previously proposed by Marcus and coworkers,⁴⁷ that the native layer observed on the polished Mg surface had a duplex structure composed of an inner MgO and an outer Mg(OH)₂ layer. After immersion in 0.1 mol.l⁻¹ NaCl with (Fig. 3A b) or without H₂S (Fig. 3A c), only one component assigned to hydroxylated species (Mg–OH) and shifted towards higher binding energies is observed. The energy shift is similar to the one previously described⁴⁸ on Mg exposed to pure water and assigned to the thickening of the outer Mg(OH)₂ layer. The thickness of the Mg(OH)₂ layer does not allow us to observe the MgO and Mg peaks by XPS and it is not possible to confirm the constant MgO thickness already observed during growing of hydroxide/oxide layer.⁴⁸ Based on inelastic mean free path values classically used^49,50 and corrected with specific references of our group (${\lambda }_{O1s}^{Mg\left(OH\right)2}\,=\,2.4\,{\rm{nm}}\,and\,{\lambda }_{Mg2p}^{Mg\left(OH\right)2}\,=\,4.2\,{\rm{nm}}$), a minimum Mg(OH)₂ thickness of 12.6 nm was estimated. As MgO is never observed after immersion in NaCl and NaCl + H₂S (component at 531.4 eV after charge effect correction), considering a homogeneous hydroxide layer, it is possible to determine the stoichiometry of the hydroxide layer using the O 1s/Mg 2p area ratio. From measured intensities, the at% of each species can be determined using the Eq. 1, where I is the intensity measured for each species, λ the inelastic mean free path (nm), σ the relative sensitivity factor and T the transmission factor of the spectrometer.

$$\begin{eqnarray}&&X\left(at \% \right)\,=\,\displaystyle \frac{100\tfrac{I(X)}{{\lambda }_{X}{\sigma }_{X}{T}_{X}}}{\displaystyle \sum \tfrac{I(i)}{{\lambda }_{i}{\sigma }_{i}{T}_{i}}}\end{eqnarray} \tag{ 1 }$$

The obtained ratios are close to 2 (1.99 in NaCl solution and NaCl and 1.96 in NaCl + H₂S solution). This confirms the presence of Mg(OH)₂ in the outer part of the oxide film. Thus, it can be concluded that H₂S has no influence on the composition of the outer layer. After "conditioning" in NaCl + H₂S solution and subsequent immersion in NaCl solution, the values of the O1s/Mg2p ratio are similar (1.87 for the uncorroded area and 1.82 for the dark corroded areas) and indicate that H₂S_aq does not modify the outer Mg(OH)₂ layer. The analysis of the Cl 2p core level (Fig. 4a)) reveals that after restarting the OCP measurements in 0.1 mol.l⁻¹ NaCl, a slight chloride enrichment can be observed in the dark corroded areas (∼0.4 at% compared to ∼0.2 at% in the uncorroded areas or on the surface after immersion in NaCl solution containing H₂S). This enrichment of chloride species on dark corroded areas was already observed during anodic polarization in NaCl in our previous work²⁴ but it was stronger (about 5 at%) and it could be explained by a higher surface fraction covered by dark corroded areas and a higher surface area due to the surface roughness. While the exact influence of chlorides on the Mg corrosion mechanisms has not been investigated in detail in the literature so far, the presence of chlorides mainly on the dark corroded areas shows an influence of chlorides on the (dendritic) growth of dark corroded areas.

Zoom In Zoom Out Reset image size

Figure 4b shows the evolution of the S2p spectra obtained for polished Mg (a), and after immersion in 0.1 mol.l⁻¹ NaCl electrolyte for 3 min (b) and in 0.1 mol.l⁻¹ NaCl electrolyte containing 0.01 mol.l⁻¹ H₂S for 30 min (c). On the polished Mg surface (Fig. 4B-a), a main spin–orbit doublet at 168.6 eV (S2p_3/2) and 169.8 eV(S2p_1/2) (after binding energy correction from charging effect) indicates the presence of sulfate species. A very small amount of sulfites with components at 167.4 eV (S2p_3/2) and 168.5 eV (S2p_1/2) is also observed. The small amount of sulfates, in the order of 0.20%, results probably from the contamination during glassware cleaning in Piranha solution (H₂SO₄/H₂O₂ mixture). After immersion in NaCl for 3 min (Fig. 4B-b) a strong decrease of the intensity of the peaks corresponding to sulfate and sulfite species is observed indicating that these species are either removed during the sample immersion (and thus do not influence the magnesium corrosion mechanism) or covered by the corrosion products layer. After immersion in NaCl containing H₂S (Fig. 4B c), similar peaks, assigned to sulfates (168.7 and 169.9 eV) and sulfites (167.4 and 168.5 eV) are also observed. The presence of sulfate and sulfite anions may originate from the chemicals used for electrolyte preparation (Na₂S·9H₂O, SO₃²⁻ contamination <0.5%). However, the presence of Mg sulfites in the corrosion products cannot be excluded. The formation of MgSO₃ species has been already observed as corrosion products in presence of SO₂ traces.⁵¹

In Fig. 4B-c, another peak is found at 162.6 eV, characteristic of a sulfur-metal bond. This small S-Met peak observed only when H₂S is added to the electrolyte can be assigned to S_ads on metal surfaces. At these low levels, and due to the detection limit of XPS, it is difficult to discriminate between a S_ads–Mg or a S_ads–Fe bond. Indeed the binding energies for S⁰-Met bonds (Fe,⁵² Ni,⁵³ Ni based alloys^54,55) vary between 162 and 163 eV. The binding energy measured for the S–Au⁵⁶ in the case of SAM formation from alkanethiol is similar. At the same energies are also observed FeSx⁵⁷ and MgSx⁵⁸ species. The amount of S adsorbed is about 0.08 at% (calculated following Eq. 1). The number of scans of the S2p peak has been drastically increased (500 scans) to reduce the signal-to-noise ratio. This value corresponds to the percentage of S (at%) with respect to all species detected by XPS (except carbonaceous species) averaged over the thickness from which XPS signals are measured This low value can be compared with the expected theoretical value for a complete monolayer on segregated Fe (see calculation in the discussion).

In order to get additional information on the nature of the S species bonded to the metal surface, ToF-SIMS ion depth profiles were performed for a polished Mg sample (Fig. 5a) and for the sample immersed in NaCl electrolyte containing H₂S (Fig. 5b). The ions characteristics of the Mg oxide (MgO⁻ at 39.980 amu), Mg hydroxide (MgOH⁻ at 40.988 amu), Mg sulfide (MgS⁻ at 55.957 amu), Fe (metal or oxide) (FeO⁻ at 71.930 amu) and Fe sulfide (FeS⁻ at 87.907 amu) were used. ToF-SIMS ion depth profiles obtained for the freshly polished Mg sample show high intensity MgO⁻ and MgOH⁻ signals in the whole range of sputtering time (600 s) indicating that the Mg metallic substrate was not reached and only the Mg oxide/hydroxide film covering the Mg substrate was analyzed (in agreement with XPS data).

Zoom In Zoom Out Reset image size

Looking closely at the MgO⁻ and MgOH⁻ signals (Fig. 5a), it can be observed that MgOH⁻ intensity is higher in the outer part of the profile and MgO⁻ is higher in the inner part of the profile suggesting a bilayer structure with an inner Mg rich oxide and an outer Mg rich hydroxide as already observed and discussed above. However, there is no sharp interface. The increase of the MgO⁻ signal and the stability of the MgOH⁻ signal when probing deeper suggest a weak enrichment of MgO oxide in the inner part.

The weak MgS⁻ signal, which does not vary much during sputtering, may originate partly from a low concentration of sulfur initially present in the Mg bulk, although it was not mentioned in the impurities content by the manufacturer. ToF-SIMS has a very high sensitivity and low levels of impurities can be detected. The affinity between sulfur and Mg is well known and Mg is often used in the desulfuration processes. A closer look at the other fragments shows that the intensity of FeO⁻ is very low through the whole depth profile and FeS⁻ is not detected. According to our previous work describing the presence of metallic Fe particles at the grain boundaries,²⁴ it is possible to consider the FeO⁻ signal as a marker of metallic Fe. Indeed, for these very low amounts (not detectable by XPS), the Fe metal signal is too low in the negative polarity and FeO₂⁻ at 85.929 amu, which is the marker of oxidized Fe, is not observed. Thus using the FeO⁻ signal as a marker of metallic Fe in contact with oxidized magnesium is consistent with the fact that the Fe particles segregated at the grain boundaries terminations are electrically connected to the Mg matrix hence at the OCP of magnesium (∼−1.5 V SCE⁻¹), Fe is not oxidized.²⁴

After immersion in H₂S containing NaCl solution, the ToF-SIMS ion depth profiles (Fig. 5b) show again clearly the signals characteristic of MgO and Mg(OH)₂. The MgOH⁻ signal remains high (higher than MgO⁻) through the entire depth profile. This confirms the formation of a thicker hydroxide layer compared to the freshly polished Mg sample, as already observed by XPS.

The FeS⁻ and MgS⁻ signals shown in Fig. 5b reach a maximum intensity after about 20 s of sputtering. The MgS⁻ signal intensity decreases after 20 s of sputtering and is similar to that measured on the freshly polished sample (Fig. 5a). The FeS⁻ signal also decreases and disappears while the FeO⁻ signal, characteristic of the Fe particles, increases and stabilizes at a level similar to the one observed on the freshly polished sample (Fig. 5a). The FeS⁻ and MgS⁻ signals, observed at the beginning of the sputtering time, can be assigned to surface Mg–S and Fe–S corresponding to S_ads resulting from a surface reaction between H₂S in the electrolyte and both the Mg matrix and the Fe particles. The low levels measured for these ions by ToF-SIMS and the small S 2p XPS signal at low binding energy both indicate a small surface coverage by adsorbed sulfur.

The MgS⁻ signal could also be attributed to Mg sulfite in the corrosion products, however this would imply that characteristic fragments of MgSO₃ (MgSO⁻ at 71.952 amu, MgSO₂⁻ at 87.947 amu and MgSO₃⁻⁻ at 103.942 amu) were observed, which is not the case.

Figure 6 presents the potentiodynamic polarization curves obtained for a magnesium electrode in 0.1 mol.l⁻¹ NaCl without and with 0.01 mol.l⁻¹ H₂S. These experiments were performed in a non-deaerated electrolyte and in a single upward scan at a rate of 1 mV s⁻¹ after immersion at OCP for 2 min in NaCl electrolyte and 10 min in NaCl/H₂S electrolyte. Anodic and cathodic domains have been scanned separately and starting near the OCP (OCP + 0.05 V for the cathodic domain and OCP − 0.05 V for the anodic domain). The value of OCP (∼−1.55 V SCE⁻¹) is similar for both immersion times. As described previously,²⁴ after this time of immersion in NaCl, no dark corroded areas and no bubble formation are observed. The shape of the curves is similar for both solutions and consistent with the literature for high purity Mg samples.^20,46 For both media, no passive domain is observed. H₂S has no strong effect on the anodic current and on the corrosion potential. The cathodic current is slightly higher in the presence of H₂S, which may be explained by a blocking effect of sulfur on the formation of Mg surface oxide/hydroxide limiting the cathodic reaction.

Zoom In Zoom Out Reset image size

To understand the behavior in the anodic domain, potentiostatic polarizations have been performed in NaCl and NaCl containing H₂S at low anodic overpotential (OCP + 0.25 V, corresponding to −1.32 V SCE⁻¹) and H₂ gas has been collected simultaneously (Fig. 7). Polarizations have been stopped after a flowed charge of at least 6 C.cm⁻². Micrographies obtained after a charge of 0.25 C.cm⁻² and at the end of the polarization (>6 C.cm⁻²) are also presented in Fig. 7. As at OCP, H₂ gas evolution follows a same trend for both electrolytes but is always lower in the presence of H₂S. In contrast to the OCP, the first domain attributed to the initiation of the H₂ evolution is very short. Then, the collected H₂ volume increases almost linearly. The decrease of the H₂ evolution rate in the presence of H₂S is similar to that observed at OCP (∼75% decrease under anodic polarization and ∼70% decrease at OCP). While the evolution rate is similar, the required time to collect 5 ml of H₂ is longer under anodic polarization (90 min) compared to OCP (300 min). Micrographies of surfaces in the presence of H₂S show a limitation of the dark corroded areas.

Zoom In Zoom Out Reset image size

XPS analyses performed on the Mg samples after the anodic polarizations at OCP + 0.25 V and OCP + 0.5 V for a charge of up to 250 mC.cm⁻² allowed us to determine the O 1s/Mg 2p ratio. The dark corroded areas covering the surface at high charges (6 C.cm⁻²) led to inhomogeneous surfaces and interpretation of data obtained by XPS on these surfaces is quite complex. This O1s/Mp2p ratio (1.9 close to expected 2) does not depend on the nature of the electrolyte or the overpotential, which indicates that the anodic polarization has no effect on the global stoichiometry of the layer which remains mainly composed of hydroxide in the outer part. A significant charging effect is also observed for these two samples indicating an insulating character of the surface explained by a thickening of the hydroxide layer. The corresponding binding energy shift in NaCl solution containing H₂S is estimated to 4.1 eV and 4.2 eV for the samples polarized at OCP + 0.25 V and OCP + 0.5 V, respectively. For comparison, in NaCl solution, this shift is estimated to 1.9 and 2.2 eV for the samples polarized at OCP + 0.25 V and OCP + 0.5 V, respectively. The amount of chlorine is also similar (about 0.5 at%) without or with H₂S and after the two polarizations.

After anodic polarization in NaCl + H₂S solution a sulfur signal is also observed by XPS (Fig. 8) as after OCP measurements (Fig. 4). Again it shows the doublet peaks corresponding to sulfate (168.6 and 169.8 eV), sulfite (167.4 and 168.5 eV) and the peak indicative of a surface S-Met bond (at 162.6 eV). By selecting the best XPS experimental conditions for the analysis of the latter peak, particularly by increasing the number of scans to 500, we could reduce the signal-to-noise ratio and thus enhance the quantitative analysis. The amount of adsorbed sulfur measured by XPS is low and varies from 0.08 at% for the sample immersed at OCP, to 0.05 at% for the sample polarized at OCP + 0.25 V and to 0.03 at% for the sample polarized at OCP + 0.5 V. These at% correspond to the S amount measured in the analyzed volume. These atomic percents are determined considering all species quantified by XPS (using Mg2p, O1s, Cl2p, S2p) except carbonaceous species (C1s peak).

Zoom In Zoom Out Reset image size

As already mentioned in the analysis of Fig. 4, the low binding energy peak is assigned to S_ads on segregated Fe particles and on Mg.

The ToF-SIMS ion depth profiles obtained after anodic polarization in 0.1 mol.l⁻¹ NaCl + 0.01 mol.l⁻¹ H₂S at OCP + 0.25 V and OCP + 0.5 V are presented in Figs. 9a and 9b, respectively. As after immersion at OCP in NaCl +H₂S solution (Fig. 5), the profiles show a strong modification compared to the ToF-SIMS depth profiles obtained on the freshly polished Mg sample (see discussion about Fig. 5). The two ToF-SIMS ion depth profiles in Fig. 9 show intense MgOH⁻ signals, that remain higher than the MgO⁻ signal over the whole analyzed thickness. This indicates an outer Mg-rich hydroxide, as already observed by XPS (see discusion about Fig. 3). Compared to the ToF-SIMS analysis performed after immersion at OCP in NaCl + H₂S electrolyte (Fig. 5), the analyses carried out after anodic polarization show a slightly different behaviour for the FeS⁻ and MgS⁻ fragments. In particular, the maximum of the MgS⁻ signal is slightly shifted (∼20 s for E_ocp + 0.25 V and ∼70 s for E_ocp + 0.5 V) suggesting that the species with Mg–S bond is slighty buried into the hydroxide layer when the potential increases. The FeS⁻ fragment follows the same trend as previously observed (Fig. 5) with a complete and fast decrease of the signal indicating that the S–Fe species is located at the surface. Compared to the results obtained after immersion without anodic polarization (see Fig. 5), the FeO⁻ fragment (characteristic of metallic iron as explained previously) appears simultaneously with the FeS⁻ for the low anodic polarization and in a large domain similar to that observed for the FeS⁻ fragment for the high anodic polarization suggesting the presence, at the same location, of Fe–S and metallic Fe species in the outer layer. The presence of the FeO⁻ signal in the whole profile confirms the existence of metallic Fe segregating at the grain boundary terminations. Under anodic polarization, a porous Mg(OH)₂ layer covers the segregated Fe. However the connectivity of the segregated iron particles and the porosity of the Mg(OH)₂ layer can maintain the catalytic activity of Fe on the HER. The evolution of the MgS⁻ signal is similar to that observed previously (Fig. 5a) with a final intensity slightly higher than for the freshly polished sample. Moreover, the shape of the MgS⁻ signal profile suggests the presence of Mg–S species in the Mg hydroxyde layer.

Zoom In Zoom Out Reset image size

During Mg dissolution, Fe enrichment at the Mg surface occurs by anodic segregation. The Fe particles (shown in red on Fig. 11a) segregated at the grain boundaries surface terminations supply catalytic sites for the HER, explaining the excess H₂ evolution on Mg compared to the HER deduced from extrapolation of the cathodic branch.

In the presence of H₂S_aq, (orange on Fig. 11b) S is adsorbed on the Fe particles segregated at the grain boundaries surface terminations during the Mg corrosion. This S_ads monolayer inhibits the HER and the development of the dark corroded areas. Simultaneously, a porous Mg(OH)₂ forms from Mg²⁺ released during Mg dissolution. Some Mg released by the corrosion process reacts with S and the Mg–S species remain trapped in the Mg(OH)₂ layer.

Under anodic polarization (Fig. 11c), a similar mechanism is proposed. Anodic segregation of metallic Fe at the grain boundaries surface terminations takes place, and the adsorption of S blocks the iron surface sites for HER.