Introduction

Carbon and low alloy steels are widely used for marine structures and infrastructures including sheet piles, ship hulls, off-shore platforms, and marine renewable energy devices. Steel surfaces are thus exposed worldwide to very diverse marine environments, in various exposure conditions, including the so-called atmospheric zone, splash zone, tidal zone, immersion zone, and sediment zone. The marine corrosion mechanisms are then complex, a situation aggravated by the biological activity in natural seawater that generates microbiologically influenced corrosion (MIC) processes. Even though considerable efforts were made to reach a better scientific understanding of these mechanisms, some questions remain and reference to literature data sometimes reveals apparent inconsistencies. A typical example is the influence of small compositional changes on the resistance to corrosion of low alloy steels1.

Using the phenomenological model developed for marine corrosion of carbon and low alloy steel in the immersion zone2,3,4, R. Melchers clarified some points, in particular about the role of Cr in low alloy steels1. In this model, two phases are distinguished, the aerobic phase, corresponding to the first years of immersion, when the corrosion rate is controlled by the diffusion of O2, and the anaerobic phase when the corrosion process is mainly MIC, i.e., controlled by bacterial activity4. For low Cr contents (typically <3 wt%), the addition of Cr has a beneficial effect on resistance to corrosion during the aerobic phase, in particular if associated with Mo or Al1. In contrast, it has a detrimental effect during the anaerobic phase, i.e., at long term1.

In a similar way, a recent study based on big data technology addressed the influence of Cr on atmospheric corrosion of weathering steels5. For low Cr contents (up to 2.56 wt%), the influence of Cr is both beneficial and detrimental. It is beneficial with respect to Cr-free steel as it increases the resistance to uniform corrosion5. It is detrimental because the increase in Cr content promotes pitting corrosion5.

Numerous studies were actually devoted to the influence of Cr on the corrosion resistance of low alloy steels. In most cases, the beneficial effect, when observed, was linked to modifications of the corrosion product layer composition and morphology6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22. A previous study dealing with the corrosion behavior of low alloy steel in seawater confirmed that the addition of Cr (1 wt%) and Al (0.5 wt%) increased significantly the corrosion resistance of the metal in both artificial and natural seawater, at least during the aerobic phase (immersion ≤6 months)6. The corrosion product layer covering the steel surface at the end of the experiments was composed of two strata that could be distinguished visually, i.e., a black inner layer in contact with the steel surface, and an orange outer layer, in contact with seawater. Raman spectroscopy analysis revealed the presence, in the black inner layer, of a corrosion product characterized by a spectral signature intermediate between that of Fe3O4 (magnetite) and Cr2O3. This compound could then be a Cr-rich (FeII-III,CrIII)-oxide. Present in the inner part of the corrosion product layer, close to the steel surface, it may play a role in the improved corrosion resistance, increasing the adhesion between the outer layer, mainly made of Fe(III)-oxyhydroxides, and the inner layer mainly composed of magnetite, green rust compounds, and Cr-containing oxide6.

Such Fe-Cr oxides were repeatedly identified in the inner part of the corrosion product layers of Cr-containing low alloy steels immersed in saline solutions7,8,9,10. Other Cr-containing corrosion products were also reported, always in the inner part of the corrosion product layer, in agreement with the generally observed Cr-enrichment of the steel/corrosion product layer interface7,11,12,13,14,15,16. In the particular case of atmospheric corrosion, Cr-containing goethite α-Fe1-xCrxOOH was often reported12,17,18,19,22. The beneficial effect of Cr on the resistance to atmospheric corrosion was often linked to the decrease in crystallinity and particle size of the FeOOH phase, which would lead to a denser and more compact, hence more protective, corrosion product layer11,12,19,20,21,22. The decrease in crystallinity was attributed to the substitution of Fe(III) by Cr(III) in the crystal structure of α-FeOOH20,21,22,23. Cr also favors the formation of goethite at the detriment of lepidocrocite18,21, thus increasing the so-called α/γ ratio, i.e., the ratio between inactive, hence protective, phases and active, hence detrimental, phases. This effect was also reported for Cr-containing weathering steels but not necessarily associated with the presence of Cr24. Lepidocrocite γ-FeOOH is the typical active phase, which can be reduced to Fe(II)-bearing phases and thus participates as an oxidant to the atmospheric corrosion process, while goethite α-FeOOH is the typical inactive phase, more stable and less prone to reduction11,25,26. Finally, Cr(OH)3 is the other Cr-containing corrosion product reported to have a beneficial effect on the resistance to corrosion of Cr-containing low alloy steels8,16,18,20,21. It was assumed to be the core of the nucleation of FeOOH phases and thus assumed to have a role on the decrease in FeOOH particles size20.

However, even if the effects of Cr on composition and morphology of low alloy steel corrosion product layers were demonstrated, the fundamental mechanisms of the influence of Cr on the formation and transformation of the Fe oxides and (oxy)hydroxides were more rarely addressed19,23,27,28. To obtain such fundamental information, the Fe-compounds of interest are prepared in aqueous suspension and the influence of Cr on their formation and possible transformation over time is studied by replacing partially Fe by Cr. The effects of Cr on the crystallinity and local structure of α-FeOOH, or more precisely α-Fe1-xCrxOOH, could be elucidated using this approach19,23. The influence of Cr(III) on the oxidation of the Fe(II-III) hydroxysulfate, i.e., the sulfate green rust GRSO4, was assessed in a similar way27,28. It was observed first that Cr(III) favored the oxidation of GRSO4 to α-FeOOH and hindered the formation of magnetite Fe3O427. Because GRSO4 is one of the main corrosion products of carbon and low alloy steels in seawater29,30,31,32, mainly present at the steel/corrosion product layer interface, the influence of Cr(III) on both precipitation and transformation of GR(SO42-) was studied28. The reported enrichment in Cr at this interface7,11,12,13,14,15,16 should indeed influence the formation of the corrosion products formed directly from the Fe2+ ions produced by the corrosion reaction, i.e., the Fe(II)-bearing phases. It was observed that Cr(III) species decreased the crystallinity of the obtained GRSO4, favored its transformation to magnetite during ageing in anoxic conditions but had no effect on its oxidation in aerated conditions28. This last result shows that adding Cr(III) before28 or after27 the formation of GRSO4 modifies the way Cr(III) species influence the oxidation process of the green rust.

To investigate further the modifications Cr(III) species could induce to the corrosion product layers of low alloy steel in seawater, a fundamental approach based on the study of the oxidation processes of Fe(OH)2 precipitates was considered. Aqueous suspensions of Fe(OH)2 were prepared, without (100 at.% Fe) or with Cr(III) (98 at.% Fe and 2 at.% Cr or 92 at.% Fe and 8 at.% Cr). With the lower Cr content, i.e., 2 at.% as in many low alloy steels, the results showed some interesting trends that were, however, not significant, difficult to interpret, and the Cr content was then increased up to 8 at.%. Therefore, the article focuses on the results obtained with 8 at.% Cr. This preliminary finding is in agreement with most previous studies showing that the influence of Cr was associated with an enrichment in Cr at the steel/corrosion product layer interface and in the inner part of this layer7,11,12,13,14,15,16.

Two methodologic approaches were considered. First, the oxidation of the precipitate in aqueous suspension was monitored over time. The chloride and sulfate concentrations were adjusted, by adding NaCl and Na2SO4 ∙ 10H2O, to those corresponding to seawater. The final products of the oxidation process were then characterized by X-ray diffraction (XRD) and µ-Raman spectroscopy (µ-RS). The initial precipitates were characterized by attenuated total reflection (ATR) Fourier Transform Infrared spectroscopy (FT-IR) in nitrogen atmosphere to prevent their oxidation in air. This first approach, focused on the oxidation process of aqueous suspensions of Fe(OH)2, was already used previously, in particular to study the influence of chloride species33,34, sulfate species35,36, or both chloride and sulfate species37 on the process. The actual corrosion processes, which take place in a thin compact layer that covers the steel surface, may, however, differ from those taking place in a stirred aerated aqueous suspension where the particles of Fe-compounds are homogeneously distributed in a large volume of liquid (200 mL in the present study).

A second approach was then considered. The initial Fe(OH)2 precipitate, obtained with or without Cr(III), was filtered, and the obtained wet paste was mixed with glycerol so as to slow down, but not prevent completely, the oxidation of the Fe(OH)2 particles. The resulting mixture was set in the sample holder of the XRD system, pressed to form a compact thin (~280 µm thick) layer. The evolution of the XRD pattern was monitored over time, which revealed the oxidation processes taking place inside the thin layer of Fe(OH)2 particles, and the influence of Cr(III) species on these processes could be determined.

Results

Oxidation in air of aqueous suspensions of Fe(OH)2

The redox potential E vs time curves recorded during the oxidation of the aqueous suspensions of Fe(OH)2 are displayed in Fig. 1. In any case, E initially decreases rapidly as the two solutions are mixed and Fe(OH)2 precipitates. It reaches a minimal value that depends on both Q ratio and Cr proportion, but is always found approximately between −0.7 and −0.8 V vs Ag-AgCl-3M. In the absence of Cr (Fig. 1a), for Q = 0.88 and Q = 1, the potential remains in this range for about 1.1–1.2 h and increases rapidly, indicating that Fe(OH)2 has been entirely oxidized to Fe(III)-based compounds, i.e., Fe3O4 or FeOOH compounds33,34,35,36,37. The potential then remains stable at a final value that depends on the Q ratio. The E vs time curve obtained for Q = 1.136 has a different shape. At the beginning, the potential stabilizes at a higher value, around −0.6 V vs Ag-AgCl-3M, and decreases after 1.3 h to reach its minimum (−0.7 V vs Ag-AgCl-3M) at time t1. It increases slowly afterward, reaching −0.35 V vs Ag-AgCl-3M after 2.5 h, and finally increases sharply around time t2. This shows that the oxidation process involves at least two steps, the first one corresponding to the oxidation of Fe(OH)2 to a green rust (GR) compound and the second one to the oxidation of the GR compound to Fe(III)-based compounds. Actually, the formation of GR does require a Q ratio higher than 133,34,35,36,37, which explains the specific shape of the E vs time curve obtained here for Q = 1.136.

Fig. 1: Typical E vs. time curves obtained during the oxidation by air of Fe(OH)2 aqueous suspensions, for Q ratios equal to 1, 1.136, 0.88.
figure 1

a Curves obtained in the absence of Cr(III) species. b Curves obtained with Cr(III) = 8 at.%. The circles point out in each curve the time tf corresponding to the end of the oxidation reaction.

Full size image

In a solution containing Cl- and SO42- ions, both chloride-GR (GRCl, i.e., FeII3FeIII(OH)8Cl ∙ 2H2O33,34) and sulfate-GR (GRSO4, i.e., FeII4FeIII2(OH)12SO4 ∙ 8H2O36,38) are, however, likely to form. According to previous work37, GRCl would form first, as it has the smallest average oxidation number for Fe of +2.25, and would be subsequently oxidized to GRSO4 that has a higher average oxidation number of Fe, i.e., +2.33. This was checked with a XRD analysis of the precipitate 15 min before and 15 min after t1 (Supplementary Fig. 1), which demonstrated that GRCl was predominant before t1 and GRSO4 predominant after t1. For Q = 1.136, in the absence of Cr, the oxidation of Fe(OH)2 is then more precisely a three-step process:

$${{\rm{Fe}}\left({\rm{OH}}\right)}_{2}\to {{\rm{GR}}}_{{\rm{Cl}}}\to {{\rm{GR}}}_{{\rm{SO}}4}\to {\rm{Fe}}\left({\rm{III}}\right)-{{\rm{based}}\; {\rm{compounds}}}\,({{\rm{Fe}}}_{3}{{\rm{O}}}_{4}\,{{\rm{or}}\; {\rm{FeOOH}}})$$

With 8% Cr (Fig. 1b), the three E vs time curves have similar shape and are rather characteristic of a one-step oxidation process. The final increase of the potential for Q = 1, however, involves three different slopes, as E increases slowly from 0.7 to 1.0 h, more rapidly between 1.0 h and 1.4 h, and abruptly between 1.4 and 1.6 h. Various oxidation processes may then take place simultaneously and end at different times. Anyway, the main change induced by Cr is observed for Q = 1.136 and relates to the formation of the GR compounds that Cr seems to hinder.

The overall oxidation time, tf, was measured at the inflection point of the final sharp potential increase, as displayed in Fig. 1. For 0% Cr and Q = 1.136, tf corresponds to t2, end of the second reaction stage. For Q = 0.88 and Q = 1, tf is only slightly increased in the presence of Cr, from 1.14 h to 1.19 h for Q = 0.88 and from 1.26 h to 1.28 h for Q = 1. Actually, the accuracy of the measured tf values is about 10%, as determined from the scattering of values observed between 3 experiments performed in each condition. Thus, the slight variations of tf with Cr content may not be significant. In contrast, tf decreases from 2.58 h for 0% Cr to 1.35 h for 8% Cr, which in this case is indeed significant and clearly demonstrates that the mechanisms of oxidation and transformation of Fe(OH)2 is clearly different with 8% Cr.

The final products of the oxidation process were analyzed by XRD (Figs. 2 and 3) and µRS (Fig. 4). The XRD patterns of Fig. 2 relate to the oxidation without Cr. For each Q value, only one phase is observed, and the nature of this phase depends on Q. These final oxidation products are magnetite Fe3O4 for Q = 1, goethite α-FeOOH for Q = 0.88, and lepidocrocite γ-FeOOH for Q = 1.136. This result is consistent with previous works33,35 and illustrates the importance of the Q ratio, which was interpreted as follows: Goethite is favored for Q < 1, i.e., when an excess of OH- ions remains in solution after the precipitation of Fe(OH)2, because its formation is favored in alkaline conditions (pH measured at 11.5–12, once stabilized a few minutes after precipitation). Lepidocrocite is favored for Q > 1 because it is the phase that forms preferentially from GR compounds, and magnetite is favored when neither OH- nor Fe2+ ions remain in excess after the precipitation of Fe(OH)2, i.e., when Q = 1. Note that the pH of the suspension decreases when Q increases. It was measured at 9.5–10.0 for Q = 1 and at 7.5–8.0 for Q = 1.136, once stabilized after precipitation.

Fig. 2
figure 2

XRD patterns (Co-Kα wavelength) of the final products of the oxidation by air of Fe(OH)2 aqueous suspensions obtained in the absence of Cr(III) species for Q ratios equal to 1, 1.136, and 0.88. G are the diffraction peaks of goethite, L those of lepidocrocite and M those of magnetite, denoted with the corresponding Miller index.

Full size image
Fig. 3
figure 3

XRD patterns (Co-Kα wavelength) of the final products of the oxidation by air of Fe(OH)2 aqueous suspensions obtained with Cr(III) = 8 at.% for Q ratios equal to 1, 1.136, and 0.88. G are the diffraction peaks of goethite, L those of lepidocrocite and M those of magnetite, denoted with the corresponding Miller index.

Full size image
Fig. 4
figure 4

Typical Raman spectra of the final products of the oxidation by air of Fe(OH)2 aqueous suspensions obtained with Cr(III) = 8 at.% for Q ratios equal to 1 and 1.136. G are the Raman peaks of goethite, L those of lepidocrocite and M those of magnetite.

Full size image

The XRD patterns obtained with 8% Cr are displayed in Fig. 3. Significant differences are observed with the previous patterns, which demonstrate that this Cr proportion can have an important effect on the formation of Fe-corrosion products. First, for Q = 1, the main oxidation product is now goethite, and magnetite is observed as a minor phase, as only its two main diffraction peaks, M311 and M440, can be seen. This result shows that Cr favors the formation of goethite, as observed previously via the characterization of the rust layer covering Cr-bearing low alloy steel after atmospheric corrosion18. For Q = 1.136, the three phases, i.e., lepidocrocite, magnetite and goethite are formed, and none of them predominates. If Cr hinders the formation of GR compounds, as suggested by the corresponding E vs time curve, then it subsequently hinders the formation of lepidocrocite, which is favored when GR is formed as a transient oxidation product. Consequently, the formation of other compounds, i.e., magnetite and goethite, is favored. Finally, for Q = 0.88, goethite is, as in the absence of Cr, the only oxidation product of Fe(OH)2. Comparing closely both diffractograms, it can be seen that the diffraction lines of goethite are broader in the absence of Cr. The determination of the full width at half maximum (FWHM) of the diffraction lines by computer fitting (using Gauss functions) confirmed it. For the intense G110 peak, for instance, FWHM is 1.65° without Cr and 1.12° with 8% Cr. This indicates that, in our experimental conditions, Cr not only favor the formation of goethite with respect to magnetite and lepidocrocite, but it also favors the growth and/or increase the crystallinity of goethite particles. Various previous works dealing with atmospheric corrosion indicated that, conversely, FeOOH crystallinity and particle size were decreased for Cr-bearing low alloy steel11,12,19,20,21,22. However, the increase in crystallinity and/or particle size observed here for Q = 0.88 correspond to particular pH conditions, i.e., alkaline conditions (pH about 11.5–12), which do not correspond to pH conditions typical of atmospheric corrosion.

µ-RS analysis confirmed the results given by XRD. As an example, a selection of the Raman spectra obtained with Cr = 8% at Q = 1 and Q = 1.136 is displayed in Fig. 4. All the spectra obtained for Q = 1 mainly show the Raman bands of goethite39,40, at (with decreasing intensity) 390, 300, 480, 555, and 249 cm−1, and the main Raman peak of magnetite at 670 cm−1 39,40. The peak at 700–705 cm−1 may be attributed to ferrihydrite41, a poorly ordered and poorly crystallized Fe(III)-oxyhydroxide phase difficult to detected by XRD. For Q = 1.136, the main peaks of goethite and magnetite are seen again, now with those of lepidocrocite, at 252, 380, and 528 cm−1 39,40.

To illustrate the results obtained with 2% Cr, the XRD pattern and three typical Raman spectra of the end products of the oxidation for Q = 1 are displayed in Fig. 5. The XRD pattern reveals, as for 8% Cr, that both magnetite and goethite were formed, but goethite as a minor component in this case. This is confirmed by Raman analysis. The three displayed spectra, obtained with different zones of the sample, are similar. The main Raman peak of magnetite at 670 cm−1 is clearly predominant while the two other minor peaks, at 300 and 550 cm−1, are also visible39,40. Only the main Raman peak of goethite is visible, at 390 cm−1 39,40 and its intensity is similar to that of the smallest peaks of magnetite. In conclusion, the influence of 2% Cr remains the same as that observed with 8 at.% Cr, but the effects are much smaller.

Fig. 5: Characterization of the final products of the oxidation by air of Fe(OH)2 aqueous suspensions obtained with Cr(III) = 2 at.% for Q equal to 1.
figure 5

a XRD pattern. b Typical Raman spectra. G are the peaks of goethite and M those of magnetite, denoted with the corresponding Miller index in the XRD pattern.

Full size image

FT-IR spectroscopy of 1-week aged Fe(OH)2 precipitates

The FT-IR spectra of the 6 considered aqueous suspensions are gathered in Fig. 6. First, it can be noted that even if the procedure involves a drying of the samples in a N2 flow, water molecules cannot be entirely removed, as some molecules are bound to the solid particles surfaces. For the samples prepared without Cr, the persistence of a small amount of water molecules is revealed by a flat broad band around 3400 cm−1, corresponding to the O-H stretching mode, and a very small peak at 1630 cm−1 (O-H deformation mode)42. The spectral signature of H2O molecules is accompanied by that of sulfate species, that corresponds to a doublet of peaks at 1120 cm−1 and 1150 cm−1 43. However, for the samples prepared with 8% Cr, the vibration bands of H2O are much more intense, indicating that the precipitate remained largely hydrated, i.e., that particles of the solid phases formed a gel-like matrix, retaining part of the electrolyte, as revealed by the concomitant increase of the signal of both water molecules and sulfate species. The solid phase precipitating in the absence of Fe, i.e., for 100% Cr and Q = 1, was prepared in the same conditions and analyzed by XRD (Supplementary Fig. 2). It revealed mostly amorphous and/or nanocrystalline matter, but some diffraction peaks were, however, clearly seen. They corresponded to Cr(OH)3 ∙ 6H2O, i.e., a hydrated Cr(III) hydroxide. The OH- ions associated with Cr(OH)3 should give rise to an additional O-H stretching signal, and, looking closely to the corresponding FT-IR spectra, a shoulder is visible around 3000 cm−1, on the low wavenumber side of the broad band mainly due to H2O molecules. This shoulder is more pronounced for Q = 0.88. The hydrogen bond between H atoms of hydroxide ions of both Fe(OH)2 and Cr(OH)3 and O atoms of water molecules should also give a signal in this spectral region.

Fig. 6
figure 6

FT-IR spectra of Fe(OH)2 precipitates obtained without Cr(III) species or with 8 at.% Cr(III), for Q ratios equal to 1, 0.88, and 1.136, after 1 week of ageing at RT in anoxic conditions and drying under a N2 flow.

Full size image

The other detected peaks, at 3627 and 487 cm−1, are those typical of the Fe(II) hydroxide44. The peak at 3627 cm−1 is sharp and corresponds to the O-H stretching mode while the peak at 487 cm−1 corresponds to a lattice (T’) mode44. Actually, this second peak is exactly at 487 cm−1 for Fe(OH)2 obtained without Cr but shifted to a slightly higher wavenumber, about 502 cm−1, for Fe(OH)2 obtained with 8% Cr. A similar shift of this vibration band to 505–508 cm−1 was observed previously for partly oxidized Fe(OH)244, i.e., it is associated with the presence of Fe(III) in the crystal lattice of Fe(OH)2. In the present case, this systematic shift observed with 8% Cr can be attributed to the presence of Cr(III) ions in the lattice of Fe(OH)2.

The precipitation of a hydrated Cr(III) hydroxide together with Fe(OH)2 can explain partially the stronger H2O signal of the FT-IR spectra, but not that of the sulfate ions. This very intense H2O signal, associated with a similarly intense sulfate signal, shows that Cr(OH)3 actually helped forming a gel-like structure that not only retains the water molecules bound to the Cr(OH)3 matrix, but sulfate ions as well (and more likely Na+ and Cl- ions that cannot be evidenced with FT-IR spectroscopy).

Consequently, FT-IR analysis of the precipitates after 1 week of ageing in anoxic conditions demonstrates that the only Fe-bearing phase was Fe(OH)2 whereas Cr(III) appears to be present both as a hydrated Cr(OH)3 phase and as Cr3+ ions present in the crystal lattice of Fe(OH)2. Note that Cr(OH)3 is actually a phase repeatedly identified in the corrosion product layers covering Cr-containing low alloy steels8,16,18,20,21.

XRD monitoring of compact thin layers of Fe(OH)2

A schematic representation of the thin layers analyzed with this methodologic approach is displayed in Fig. 7. The precipitate, aged 1 week in anoxic conditions, was not filtered and thoroughly dried, but filtered to a wet paste and mixed as such with glycerol. The resulting oily paste was spread on the sample holder to form a dense thin (280 µm) layer. Consequently, the particles of Fe(OH)2, represented here as hexagons according to the crystal structure44, are partly surrounded by water and partly surrounded by glycerol. This procedure limited the oxidation of Fe(OH)2 without preventing it completely. Preliminary tests demonstrated that the wetter the filtered paste was before mixing with glycerol, the higher was the oxidation rate of Fe(OH)2 during the acquisition of the XRD pattern. The acquisition time was restricted to 30 min as it was visually observed that the analyzed sample was almost dry, i.e., only solid phases and glycerol remained present, after such duration of exposure to the X-ray beam.

Fig. 7
figure 7

Schematic representation of the thin compact layers of Fe(OH)2 used for the in situ XRD monitoring of the oxidation process.

Full size image

The results obtained with Fe(OH)2 precipitates prepared without Cr at Q = 1 and Q = 0.88 are summarized in Fig. 8. Depending on the quality of the XRD pattern, the first pattern was recorded after 30 s (Q = 0.88) or 3 min (Q = 1). In both cases, this first pattern only revealed the presence of the characteristic diffraction peaks of Fe(OH)2, in agreement with FT-IR analysis. These peaks are only accompanied by a broad band, centered on 2θ = 26°, which is due to glycerol45. This signal is clearly visible on the pattern recorded after 3 min for Q = 1. Then, oxidation products of Fe(OH)2 were progressively revealed. For Q = 1, the main oxidation product proved to be the chloride green rust, GRCl, and its four main peaks could then be distinguished on the pattern recorded after 30 min. At the end of the experiment, two additional very small peaks could be seen, and are attributed to feroxyhyte, i.e., δ-FeOOH. For Q = 0.88, feroxyhyte was the only oxidation product to form.

Fig. 8
figure 8

XRD patterns (Co-Kα wavelength) of thin layers of Fe(OH)2 obtained in the absence of Cr(III) species for Q ratios equal to 1 and 0.88, acquired after 30 s (Q = 0.88) or 3 min (Q = 1), and 30 min for both Q ratios. FH are the diffraction peaks of Fe(OH)2, GRCl those of the chloride-GR, and δ-FeOOH those of feroxyhyte, denoted with the corresponding Miller index.

Full size image

Feroxyhyte forms when dissolution-reprecipitation processes cannot take place. It is commonly obtained via a violent oxidation of Fe(OH)246,47, e.g., using hydrogen peroxide. It is structurally similar to its precursor phase Fe(OH)2 in the sense that it keeps the hexagonal close packing of anions46,47,48. It can also be obtained by other methods implying that a solid state transformation takes places, and for instance can be obtained by exposure of dried Fe(OH)2 to oxygen47. Therefore, in our experiments, feroxyhyte could result from the oxidation of the Fe(OH)2 particles remaining unoxidized after the evaporation of water. This would explain its formation at the end of the experiment for Q = 1. However, a phase structurally similar to δ-FeOOH, which was called α’-FeOOH, was identified as the precursor phase to goethite α-FeOOH in the oxidation process of Fe(OH)2 in alkaline conditions35,49. Based on the characteristic XRD pattern of feroxyhyte, now commonly admitted to be composed of four diffraction peaks, i.e., (100), (101), (102), and (110) as illustrated in Fig. 8 for Q = 0.88, α’-FeOOH and δ-FeOOH are actually the same compound. Besides, the more recent study about the crystal structure of δ-FeOOH demonstrated that feroxyhyte shared key structural features with goethite50. Older works also emphasized that the transformation of δ-FeOOH to α-FeOOH only required a re-arrangement, or a re-ordering, of Fe3+ cations in the octahedral sites of the hexagonal close packing of O atoms46. Thus, feroxyhyte must be considered as a precursor of the formation of goethite, which explained why it is obtained as the only oxidation product in the early stages of the process for Q = 0.88, i.e., a process that led, for the corresponding aqueous suspension, to goethite (Fig. 2).

Figure 9 displays the XRD patterns obtained for a Fe(OH)2 precipitate prepared without Cr at Q = 1.136. Three patterns are displayed. The first one, recorded after 30 s, shows that GRCl is present, as a minor phase, together with Fe(OH)2. GRCl was not be detected by FT-IR spectroscopy, more likely because its signal was too weak. It is also possible that GRCl formed during the filtration procedure but it was systematically and only observed in these conditions (i.e., Cr = 0% and Q = 1.136). The other assumption is that the small initial amount of Fe(III), present because the solutions were not deaerated, induced the formation of GRCl during precipitation, as the formation of GR compounds is favored for Q > 133,35. Then, after 3 min of experiment, the main peaks of the sulfate green rust GRSO4 were in turn clearly visible. As the oxidation process went on, the intensity of the diffraction peaks of GRSO4 increased, relatively to those of Fe(OH)2 and GRCl. This process is similar to that observed with the aqueous suspensions, i.e., Fe(OH)2 is first oxidized to GRCl which is in turn oxidized to GRSO4. Finally, as for the other Q ratios, feroxyhyte was also obtained after 30 min.

Fig. 9
figure 9

XRD patterns (Co-Kα wavelength) of thin layers of Fe(OH)2 obtained in the absence of Cr(III) species for Q = 1.136, acquired after 30 s, 3 min, and 30 min. FH are the diffraction peaks of Fe(OH)2, GRCl those of the chloride-GR, GRS those of the sulfate-GR, and δ-FeOOH those of feroxyhyte, denoted with the corresponding Miller index.

Full size image

The results obtained for precipitates prepared with 8% Cr are summarized in Fig. 10. Only the patterns obtained after 30 min are displayed because those recorded after 30 s or 3 min were similar, whatever the acquisition time and the Q ratio, and only showed the diffraction peaks of Fe(OH)2. As revealed by the XRD patterns finally obtained, this is due to the fact that the formation of green rust compounds did not take place. Consequently, feroxyhyte was the only observed oxidation product, may be accompanied by another poorly crystallized Fe(III) compound. Two additional weak broad peaks can be seen for Q = 1.136, and maybe for Q = 0.88. These broad diffraction peaks may correspond to so-called “2-line” ferrihydrite51, the most poorly ordered and poorly crystallized form of Fe(III)-(oxy)hydroxides. This results clearly demonstrates that Cr(III) hinders the oxidation of Fe(OH)2 to green rust compounds.

Fig. 10
figure 10

XRD patterns (Co-Kα wavelength) of thin layers of Fe(OH)2 obtained with 8 at.% of Cr(III) for Q ratios equal to 1, 0.88, and 1.136, for an acquisition time of 30 min. FH are the diffraction peaks of Fe(OH)2 and δ-FeOOH those of feroxyhyte, denoted with the corresponding Miller index. Fd-1 and Fd-2 correspond to ferrihydrite.

Full size image

The XRD data were also used to study the morphology of the Fe(OH)2 particles. Using the XRD patterns obtained after 30 min of acquisition, the 6 main diffraction peaks of Fe(OH)2 were computer fitted with Gauss functions to determine the angular position, the intensity, and the full width at half maximum of each peak. The width of the peak was used to determine the particle size using the Sherrer equation. Table 1 gives the data obtained for Q = 1 at both Cr contents.

Table 1 XRD data for the 6 main peaks of Fe(OH)2 at Q = 1 (30 min. acquisition), 0% Cr and 8% Cr: dhkl is the interplanar distance, Ihkl is the intensity of the corresponding diffraction peak (with 100 for the main one) and Dhkl is the particle size in the direction perpendicular to the hkl planes.
Full size table

The obtained relative intensities of the diffraction peaks can first be compared to those expected for randomly orientated particles (reference Ihkl values in Table 1). No significant differences are observed, which indicates that the Fe(OH)2 particles are indeed randomly orientated inside the 280 µm thick analyzed layer. Looking at the dimension of the particles, it can first be seen that these particles are significantly smaller with 8% Cr. The interplanar distances are similar, except d001 that is significantly smaller with Cr. Due to the smaller size of Cr3+ ions with respect to Fe2+ ions52, the presence of Cr3+ in the crystal structure of Fe(OH)2 could indeed lead to the decrease in the lattice cell size. It could also explain the observed decrease in particle size. Actually, FTIR analysis already indicated that part of the Cr3+ ions were present inside the crystal structure of Fe(OH)2.

As for the shape of the Fe(OH)2 particles, it can be noted that D001, i.e., the dimension along the basal plane, is much higher (approximately three times as much) than D100, i.e., the dimension along the c axis of the hexagonal layered structure of Fe(OH)2. The Fe(OH)2 particles may be hexagonal platelets, as represented in Fig. 7, but they are thick ones.

Discussion

If the corrosion process is uniform, both anodic and cathodic reaction rates are identical at each point of the steel surface. For a Cr-containing low alloy steel, the anodic reaction can be written, with x measuring the amount of Cr atoms oxidized per oxidized Fe atom, as:

$${{\rm{FeCr}}}_{{\rm{x}}}\to {{\rm{Fe}}}^{2+}+x{{\rm{Cr}}}^{3+}+(2+3x){{\rm{e}}}^{-}$$
(1)

Assuming the cathodic reaction is O2 reduction, if the same number of electrons is involved, then the reaction is:

$$\left(\frac{1}{2}+\frac{3x}{4}\right){{\rm{O}}}_{2}+\left(1+\frac{3x}{2}\right){{\rm{H}}}_{2}{\rm{O}}+(2+3x){{\rm{e}}}^{-}\to (2+3x){{\rm{OH}}}^{-}$$
(2)

In this case, when 1 Fe2+ and x Cr3+ cations are produced by the anodic reaction, (2 + 3x) OH- ions are simultaneously produced by the cathodic reaction. Then, the ratio Q (see Eq. 3 in the “Methods” section) is equal to 1. Consequently, in anodic zones of the surface where Fe2+ and Cr3+ cations are in excess with respect to OH- ions, Q is larger than 1, while in cathodic zones where OH- ions are in excess, Q is smaller than 1. As Q has a strong influence on the obtained corrosion products, the composition of the corrosion product layer may differ in anodic and cathodic zones. This phenomenon was clearly observed for carbon steel immersed in seawater, and it was shown that magnetite was favored in cathodic zones while GRSO4 and FeOOH were favored in anodic zones31,32. The influence of Cr may then be different in anodic zones (Q > 1), cathodic zones (Q < 1) or if the process is uniform and does not involve persistent anodic and cathodic areas (Q = 1).

The combination of two methodologic approaches, one based on the oxidation of Fe(OH)2 aqueous suspensions, the other based on the XRD monitoring of the oxidation of thin layers of Fe(OH)2, revealed various effects of Cr(III) on the formation of Fe-corrosion products in marine environments and the associated mechanisms.

One of the effects of Cr(III) on the oxidation processes of Fe(OH)2 is to hinder the formation of GR compounds. Actually, in a comparative study of marine corrosion processes of 2.2% Cr low alloy steel and mild steel, it was observed that the predominant compounds in the inner stratum of the corrosion product layers obtained after 320 days of immersion in natural seawater differed significantly53: For mild steel, the main components of the inner stratum were GR compounds and amorphous FeS. For 2.2% Cr low alloy steel, the main components were various iron sulfides, and green rust compounds were not identified. The authors did not comment on that particular point but their results confirm that chromium has a negative impact on GR formation.

The FT-IR analysis of the initial precipitates revealed that Cr3+ ions were mainly present as a hydrated Cr(OH)3 compound forming a gel-like structure hence binding water molecules and ions from the solution to the solid phases. FT-IR also revealed that part of the Cr3+ ions were present in the crystal lattice of Fe(OH)2. The in situ XRD monitoring of the oxidation of thin layers of Fe(OH)2 clearly revealed that the oxidation process leading to GR compounds was inhibited by Cr(III), which led to the formation of feroxyhyte δ-FeOOH and a compound similar to ferrihydrite, both phases corresponding to poorly ordered forms of Fe(III)-(oxy)hydroxides. Feroxyhyte is the oxidation product of Fe(OH)2 when dissolution-reprecipitation processes cannot occur, because δ-FeOOH is obtained from Fe(OH)2 via a solid state transformation. This result shows that the formation of GR compounds is hindered by Cr(III) because dissolution-reprecipitation processes are hindered. This phenomenon may be attributed to the gel-like Cr(OH)3 matrix because the water molecules it retains, which also surround the Fe(OH)2 crystallites, are bound to the solid phase. The formation of Cr(OH)3, commonly reported for Cr-containing low alloy steel8,16,18,20,21, seems then to be a key point in the mechanisms involved in the beneficial effect of Cr on the resistance to corrosion of these alloys.

Both ferrihydrite and feroxyhyte are metastable Fe(III)-compounds and precursors of goethite. The inhibition of dissolution-precipitation processes due to the Cr(OH)3 gel-like matrix necessarily hinders the transformation of both phases and the growth of the finally obtained goethite particles. Consequently, Cr(III) would globally favor the formation of smaller FeOOH particles, which was indeed often reported for Cr-containing low alloy steels, in particular for atmospheric corrosion11,12,19,20,21,22. This result was not obtained during the study of Fe(OH)2 in aqueous suspension. However, the effect of the Cr(OH)3 gel-like matrix may be much less important for Fe(OH)2 particles scattered in a large volume of solution. Reported effects of Cr(III) such as densification of the rust layer cannot indeed be observed with the methodology based on oxidation of aqueous precipitates but may be revealed with the new approach proposed here, i.e., the in situ XRD monitoring of the oxidation of dense thin layers of Fe(OH)2 precipitates.

Note also that the presence of some Cr3+ ions in the crystal lattice of Fe(OH)2, as indicated by both FT-IR analysis and XRD analysis, implies that goethite finally obtained is Cr-substituted goethite, considering in particular that goethite is obtained via a solid state transformation.

Finally, the last point is the formation of feroxyhyte δ-FeOOH, obtained in any case during the in situ XRD monitoring of the oxidation of thin layers of Fe(OH)2. In the absence of Cr(III), feroxyhyte more likely formed after the complete evaporation of the aqueous liquid phase. During an atmospheric corrosion process, it could form similarly at the end of the drying periods, when Fe(OH)2 particles, whether formed by the corrosion of steel or resulting from the reduction of FeOOH phases, are found in a dry state, no more surrounded by any electrolyte. Poorly crystallized or (sometimes called amorphous) FeOOH phases were reported as an important component of the rust layer resulting from atmospheric corrosion22,24,25,26,41,54. For Cr-containing low alloy steel, the formation of phases such as ferrihydrite and feroxyhyte, both observed during the in situ XRD monitoring of the oxidation of thin layers of Fe(OH)2, should be favored even during the wet period, because of the positive effect of Cr(III) on the solid state transformation pathway. As already noted, it was repeatedly reported that the rust layers of Cr-containing low allow steels subjected to atmospheric corrosion contained a higher proportion (with respect to carbon steel) of poorly crystallized/small sized particles of FeOOH phases11,12,19,20,21,22.

The study of the oxidation of aqueous suspensions gave fundamental information. First, the results obtained with Fe(OH)2 precipitates without Cr illustrate the three main pathways for the oxidation of Fe(OH)2, that led to the most common corrosion products of iron and steel, i.e., lepidocrocite, goethite, and magnetite, depending on the ratio Q. For Q = 0.88, i.e., in alkaline conditions, Fe(OH)2 was oxidized to goethite. According to previous work35, goethite is obtained via a precursor similar to δ-FeOOH for ratios around Q = 0.83. As δ-FeOOH is obtained from Fe(OH)2 via a solid state transformation, this suggests that, for Q = 0.88, this transformation mechanism was favored due to the lower solubility of Fe(OH)2 in alkaline conditions. According to thermodynamic data55, the solubility of Fe(OH)2 is the lowest around pH = 12, which corresponds to the conditions met for Q = 0.88. At such pH, the activity of Fe(II) dissolved species in equilibrium with the solid phase is very low, smaller than 10−7.

In contrast, for Q = 1.136, in the presence of excess dissolved Fe(II) species, dissolution-reprecipitation processes are favored and lepidocrocite is obtained via the formation of GR compounds as intermediate transient compounds. For Q = 1, a specific mechanism leading to magnetite, discussed in previous studies35, takes place.

Secondly, with 8% of Cr(III), it is observed that, for Q = 1.136, lepidocrocite is not the only oxidation product, as magnetite and goethite are obtained too. This shows that even in aqueous suspensions, Cr(III) partially hinders the transformation pathways involving GR compounds and then lepidocrocite. Consequently, the two other transformation pathways can take place, leading to goethite and magnetite. For Q = 1, magnetite is not the only oxidation product and goethite is mainly obtained. This shows that Cr(III) favors the formation of goethite, because it favors the solid state transformation pathway as discussed above. Thus, goethite remained the only oxidation product for Q = 0.88. This finding is consistent with previous work, which correlated the proportions of Fe3O4 and FeOOH compounds in the rust layer with the Cr content of low alloy steels56. It was demonstrated that for 4–8 wt% Cr, the resistance to atmospheric corrosion was improved because Cr promoted the formation of FeOOH and decreased the proportion of magnetite.

From these results, some conclusions can be drawn about the influence of Cr on marine corrosion of low alloy steel.

In the case of steel permanently immersed in seawater, which is the main topic addressed here, the influence of Cr may be beneficial in the case of a uniform corrosion process, where the Q ratio is expected to be around 1. Cr should favor the formation of non-conductive FeOOH compounds at the detriment of the conductive Fe3O4 and thus decrease the risk of formation of corrosion cells, as described in previous work56. A higher FeOOH to Fe3O4 ratio was also observed for the corrosion product layer formed on 1 wt% Cr + 0.5 wt% Al low alloy steel after permanent immersion in stagnant artificial seawater6. In laboratory conditions or in artificial seawater, or during the aerobic phase as defined by the phenomenological model of R. Melchers for natural exposure conditions1,2,3,4, Cr should then improve the resistance to corrosion of low alloy steel.

The main corrosion products of actual seawater corrosion of steel are GRSO4, magnetite, goethite, lepidocrocite, and, in natural environments, FeS that result from SRB metabolic activity32. The oxidation of Fe(OH)2 can lead to each of these compounds, except FeS. The present study gives direct information about the role of Cr on the formation of most of the actual corrosion products, but not on the formation of FeS. However, it demonstrates that Cr hinders the formation of GR compounds. This should favor the formation of other Fe(II)-based compounds. In natural exposure conditions, the formation of FeS should then be favored, as indeed observed in previous work53. Because FeS is an electronic conductor, the effect of Cr could then be detrimental once the corrosion process is mainly controlled by SRB activity, i.e., at long-term during the anaerobic phase of the process.

In the case of localized corrosion, both anodic and cathodic zones have to be considered separately. It has been demonstrated that for carbon steel permanently immersed in seawater, the localized corrosion processes could persist because the composition of the corrosion product layer differed in anodic and cathodic zones31,32,57. In cathodic zones, the corrosion product layer is enriched in magnetite, an electronic conductor, and depleted in GRSO4 and FeOOH phases that are insulators. Consequently, dissolved O2 can be readily reduced at the corrosion product layer/seawater interface, and the underlying metal surface can act as cathode. Conversely, the anodic zones are depleted in Fe3O4 and enriched in GRSO4 and FeOOH compounds. Moreover, the corrosion being more active in anodic zones, the corrosion product layer is thicker. Consequently, the corrosion product layer acts as a barrier against O2 diffusion, increasing the anodic character of the underlying metal surface. In this situation, the effect of Cr may depend on the alloy composition.

For low Cr contents, the effects of Cr should be limited to those highlighted by the present study. In the cathodic zones, where Q should be lower than 1, Cr may favor the formation of poorly crystallized FeOOH phases at the detriment of Fe3O4, consequently favoring the formation of a corrosion product layer acting as a barrier against O2 diffusion. In the anodic zones, where Q should be higher than 1, Cr may hinder the formation of GR compounds hence favor the formation of magnetite. Globally, Cr would induce effects that oppose to the persistence of cathodic and anodic zones. The process describes above involves mainly dissolved O2 and relates to the aerobic phase of the phenomenological model1,2,3,4.

For larger Cr amounts, the enrichment in Cr at the steel seawater interface may induce the well-known effect of Cr, involved in stainless steel resistance to corrosion, that is a trend to passivation. This trend should be favored in the cathodic zones of the steel surface, where the interfacial pH is increased, so that localized corrosion should be favored, as confirmed by big data analysis in the case of atmospheric corrosion5. This phenomenon should also be compared to the formation of Cr oxides or Cr-rich Fe-oxides on the surface of low alloy steels reported in various studies dealing whether with marine corrosion6,8 or corrosion in other saline solutions7,9,10. Such oxides were not observed in the present study which shows that their formation does not result from interactions between Fe(OH)2 and Cr(III) species but involves surface reactions that can lead to the formation of an oxide phase directly on the metal surface, e.g., through a process similar to the formation of a passive oxide film. Such a process does not involve the precipitation of solid compounds in the aqueous phase.

In conclusion, our results are consistent with the conclusions drawn by R. Melchers in its thorough study devoted to the role of alloying elements on the corrosion of low alloy steel in natural seawater1. For low Cr amounts, typically Cr < 3 wt%, Cr may be beneficial at short-term (aerobic phase), if associated with Mo or Al, and detrimental at long term (anaerobic phase). The remaining question then relates to the seemingly important role of Al and Mo upon the influence of Cr1.

Methods

Preparation and oxidation of Fe(OH)2 aqueous suspensions

Fe(OH)2 was obtained by mixing a 100 mL solution of FeCl2 ∙ 4H2O, NaCl, and Na2SO4 ∙ 10H2O with a 100 mL solution of NaOH. To study the influence of Cr(III) species, part of FeCl2 ∙ 4H2O was replaced by CrCl3 ∙ 6H2O so that the proportion in Cr was about 8 at.% (exactly 8.33%). Preliminary experiments were performed with 2 at.% Cr but only some results, obtained with an initial ratio of reactants (see below) Q = 1, are presented.

The considered concentrations of reactants are listed in Table 2. They are based on those previously used for the study of the influence of Cr(III) and Al(III) on the precipitation and oxidation of GRSO428. The NaCl and Na2SO4 ∙ 10H2O concentrations were chosen so that the overall Cl- (0.55 mol L−1) and SO42- (0.03 mol L−1) concentrations corresponded to those of seawater58. The overall Fe(II)+Cr(III) concentration was set constant at 0.12 mol L−1 and the NaOH concentration was then varied to modify the ratio between the reactants. The considered ratio, denoted Q, is defined by the following equation:

$$Q=\frac{2[{{{\rm{Fe}}}}^{2+}]+3[{{{\rm{Cr}}}}^{3+}]}{[{{{\rm{OH}}}}^{-}]}$$
(3)
Table 2 Concentrations (mol L−1) of reactants used for the synthesis of Fe(II)/Cr(III) hydroxide(s).
Full size table

With this definition, the value Q = 1 corresponds to the stoichiometric conditions of the precipitation of Fe(OH)2 and Cr(OH)3 according to the reaction:

$${{\rm{Fe}}}^{2+}+{x{\rm{Cr}}}^{3+}+(2+3x){{\rm{OH}}}^{-}\to {{\rm{Fe}}({\rm{OH}})}_{2}+{x{\rm{Cr}}({\rm{OH}})}_{3}$$
(4)

In this writing, the main considered experimental conditions, i.e., Cr = 0% and Cr = 8%, correspond to x = 0 and x = 0.087, respectively. If Q > 1, then Fe2+ and Cr3+ cations are in excess with respect to OH- ions whereas if Q < 1, OH- ions are in excess. The three cases were considered, with Q values of 1, 1.136, and 0.88. Note that the stoichiometric conditions are the same whether Cr(III) precipitates as Cr(OH)3, CrOOH or Cr2O3, e.g.,:

$${{\rm{Fe}}}^{2+}+{x{\rm{Cr}}}^{3+}+(2+3{\rm{x}}){{\rm{OH}}}^{-}\to {{\rm{Fe}}({\rm{OH}})}_{2}+x{\rm{CrOOH}}+{x{\rm{H}}}_{2}{\rm{O}}$$
(5)

The solutions were not deaerated and were mixed in air, stirred ~30 s without specific precaution, so that a small proportion of Fe(II) may have been oxidized during the precipitation process. However, the amount of dissolved oxygen initially present is negligible with respect to the overall amount of Fe(II), as large concentrations are used (0.11 and 0.12 mol L−1). Then, once the precipitation was achieved:

  1. (1)

    The obtained aqueous suspension was oxidized completely. In this case, the oxidation of the precipitate was achieved by a magnetic stirring (470 rpm, rod 4-cm long and 6-mm diameter) of the suspension in open air, which ensured a progressive homogeneous process. The 200 mL suspension was aerated at the liquid-air interface in a beaker of 9-cm diameter set in a thermostatic bath at a controlled temperature of 25 ± 0.5 °C. The oxidation reaction was monitored by recording the potential E of a platinum electrode immersed in solution, using an Ag-AgCl-3M electrode as a reference (E = +0.210 V vs SHE at 25 °C). The E vs time curves were acquired using a VSP Bio-Logic potentiostat.

  2. (2)

    The obtained aqueous suspension was poured in a sealed flask, filled to the rim so that no air remains and aged 1 week before FT-IR spectroscopy analysis under N2 atmosphere. The procedure used to avoid the oxidation of the precipitate was already used previously45 and is described thereafter.

  3. (3)

    The obtained suspension was aged 1 week in anoxic conditions as described above, to increase the crystallinity of the precipitated solid phases and allow an in situ XRD monitoring of the oxidation of thin compact layers of Fe(OH)2.

Characterization of the Fe(OH)2 precipitates

The Fe(OH)2 precipitates were characterized by FT-IR spectroscopy using a Thermo-Nicolet iS50 spectrometer equipped with a KBr beamsplitter, a DTGS detector, and an attenuated total reflectance (ATR) iTX accessory with diamond crystal.

A small amount of the aqueous suspension to be studied was rapidly filtered (a few seconds). The obtained slightly wet paste was then rinsed with ethanol, sampled with a spatula and spread on the diamond crystal of the ATR accessory. The sample holder was covered with a hemispheric plastic cell insulating it and allowing a nitrogen flow to pass through. After a complete evaporation of the ethanol and, if possible (see section “Results”), of the water remaining in the sample, which was controlled via the previsualization function of the data collecting and processing associated software (OMNIC), the acquisition of the spectrum was carried out with 128 scans at a resolution of 8 cm−1. This resolution was sufficient to distinguish all the characteristic vibration bands of Fe(OH)2 and was chosen to minimize the acquisition time.

A background spectrum was acquired each hour and used for all samples studied within this hour. Like those of the samples, the background spectra were acquired under nitrogen flow.

Characterization of the oxidation products of Fe(OH)2 aqueous suspensions

The end products of the oxidation of aqueous suspensions of Fe(OH)2 were filtered, air-dried, and finely crushed in an agate mortar for µRS and XRD analysis. A Horiba High Resolution Raman spectrometer (LabRAM HR) equipped with a microscope (Olympus BX 41) and a Peltier-based cooled charge coupled device (CCD) detector was used for µRS analysis. The zones analyzed through a ×50 objective had a diameter of ~5 µm. Spectra were recorded at room temperature with a resolution of ~0.1 cm−1. Excitation was provided by a green laser diode (533 nm) whose power was reduced to 2.5% of the maximum (~0.5 mW) in order to prevent an excessive heating able to induce the transformation of the analyzed Fe-compound into α-Fe2O3 (hematite)39. The acquisition time was 1 min in most cases, but was increased up to 2 min when necessary to optimise the signal-to-noise ratio.

The XRD analysis was performed with a Thermo Scientific ARL-INEL EQUINOX 6000 diffractometer using Co-Kα radiation (λ = 0.17903 nm). The system is equipped with a curved detector (CPS-590) designed for the simultaneous detection of the diffracted photons on a 2θ range of 90°. The acquisition of each XRD pattern was made with a constant angle of incidence (7°) during 45 min.

XRD monitoring of the oxidation of thin layers of Fe(OH)2

The XRD monitoring was carried out with the Thermo Scientific ARL-INEL EQUINOX 6000 diffractometer also used for the characterization of the final oxidation products, since it allows to record the XRD pattern obtained on a 90° 2θ-range at any time.

The suspension to be studied was filtered while sheltered from air with a plastic membrane during the process. The obtained wet paste was mixed with a few droplets of glycerol, a procedure initially designed to prevent the oxidation by O2 of GR compounds59. The resulting oily wet paste was then set in the sample holder of the XRD system, thus forming a 280 µm thick compact layer.

The first preliminary tests demonstrated that, depending on the remaining water content of the paste obtained after filtration and the additional amount of glycerol, the oxidation process could be only slowed down and not totally hindered. Consequently, this process could be monitored in situ using XRD. For that purpose, an XRD pattern was recorded after different increasing acquisition times, namely 30 s, 3, 13, and 30 min.

The various phases were identified via the ICDD (International Center for Diffraction Data) PDF-4 database. The files used for the various compounds were 01-073-6991 and 00-013-0089 for Fe(II)-hydroxide, 00-029-0713 (goethite), 01-044-1415 (lepidocrocite), 01-082-1533 (magnetite), 01-077-0247 (feroxyhyte), 00-040-0127 (chloride green rust), 01-074-3400 (sulfate green rust), 00-005-0628 (halite, i.e., NaCl) and 00-016-0817 (Cr(OH)3 ∙ 6H2O). The diffraction lines of sulfate green rust, chloride green rust, and feroxyhyte were, however, indexed according to the reported studies of their crystal structures34,38,46,48,50.