**3. Results**

#### *3.1. Characterization of the Corrosion Product Layers Formed in Artificial/Natural Seawater*

As already reported in previous studies, the corrosion product layer forming on carbon steel in seawater is, in most cases, a bilayer, composed of an inner black stratum which is in contact with the metal surface and an outer orange stratum which is in contact with the marine medium [8,9,15–18]. The layers obtained in this study verified this general trend too.

The results given by μ-RS are described first, and two typical μ-RS spectra are displayed in Figure 1. Table 2 lists all the components identified by μ-RS in the corrosion product layers of the analyzed coupons. Because a large number of zones were analyzed in each layer, three kinds of components could be identified: the main ones, frequently occurring ones and minor ones which were rarely observed.

For the coupon left for 6 months in artificial seawater, the main identified component was magnetite Fe3O4. A typical spectrum is shown in Figure 1a. The three characteristic peaks of magnetite are clearly seen, with the most intense one at 671 cm<sup>−</sup><sup>1</sup> and two smaller ones at 308 and 543 cm<sup>−</sup>1, as reported in literature data [27,28,32]. Magnetite was mainly identified in the inner black stratum of the corrosion product layer. Lepidocrocite γ-FeOOH and aragonite were also frequently identified, mainly in the outer orange stratum. Aragonite is not a corrosion product, as it is a form of calcium carbonate (CaCO3), but it is often associated with corrosion products in the cathodic zones of the metal surface [14,17,18]. The small increase of the interfacial pH in these zones is sufficient to induce the precipitation of aragonite from the dissolved Ca2+ and carbonate species present in seawater. The main peak of aragonite at 1082 cm<sup>−</sup><sup>1</sup> [33] is visible on the spectrum of Figure 1a. More rarely,

green rusts, i.e., ferrihydrite (FeOOH·H2O) and chukanovite (Fe2(OH)2CO3), were also identified, but only in the inner black stratum.

**Figure 1.** Raman spectra (examples) obtained during the analysis of the corrosion product layers formed after 6 months in: (**a**) artificial and (**b**) natural seawater. M = magnetite, A = aragonite, Mck = nanocrystalline mackinawite and GR = green rust, with the position (in cm<sup>−</sup>1) of the corresponding Raman peak.

**Table 2.** μ-RS analysis of the corrosion product layers formed after 6 months in artificial/natural seawater: synthesis of the results.


1 Only the black inner stratum was analyzed.

For the coupon left 6 months in natural seawater, only the inner black stratum of the corrosion product layer was analyzed. In the outer orange layer, the biofouling mixed with the corrosion products induced an important fluorescence phenomenon that made it difficult to acquire useful data. The FeOOH phases, mainly present in the outer orange layer, thus do not appear in Table 2 for this coupon (except for ferrihydrite, identified as a minor component). In this case, the main compounds identified in the black inner stratum were nanocrystalline mackinawite (FeS) and magnetite. Nanocrystalline mackinawite is the iron sulfide that forms from the dissolved Fe(II) species [34,35] produced by the corrosion of steel, and the dissolved sulfide species produced by SRB. Its Raman spectrum is characterized by two peaks, i.e., the main one at 283 cm<sup>−</sup><sup>1</sup> and the other at 207 cm<sup>−</sup><sup>1</sup> [35], as illustrated by Figure 1b. The spectrum of (well) crystallized mackinawite is slightly different, with the main peak occurring at 300 cm<sup>−</sup><sup>1</sup> [35]. GR compounds were also frequently identified. In Figure 1b, the spectral signature of nanocrystalline mackinawite is accompanied by that of a GR compound that may be GR(SO4<sup>2</sup>−) or GR(CO3<sup>2</sup>−). Both GRs have similar spectra, with two main peaks at 430–535 cm<sup>−</sup><sup>1</sup> and 510–515 cm<sup>−</sup><sup>1</sup> and two smaller ones at ~220 cm<sup>−</sup><sup>1</sup> and ~260 cm<sup>−</sup><sup>1</sup> [36,37]. The characteristic sulfate ion peak at 991 cm<sup>−</sup><sup>1</sup> does not demonstrate that this GR is GR(SO4<sup>2</sup>−), as it could correspond to sulfate ions adsorbed on the surface of GR(CO3<sup>2</sup>−) crystals.

It must be kept in mind that Figure 1 only shows one selected spectrum from each coupon. Magnetite was identified by μRS as one of the main components of the corrosion product layer in both cases, as reported in Table 2. However, in some cases, given the small size of the zone analyzed by μRS, magnetite was not observed. This means that some small regions of the corrosion product layer did not contain magnetite, as illustrated in Figure 1b.

The results given by XRD, consistent with those given by μ-RS, are presented in Figure 2. The first pattern (a) is that of the corrosion product layer formed in stagnant artificial seawater. The most intense diffraction peaks are unambiguously those of magnetite. Numerous diffraction peaks of aragonite and lepidocrocite are clearly seen. Finally, both GR(SO4<sup>2</sup>−) and GR(CO3<sup>2</sup>−) may be identified, but only owing to their main diffraction peak (GR001 or GRC003) that is very weak.

**Figure 2.** XRD pattern of the corrosion product layers formed after 6 months in: (**a**) artificial and (**b**) natural seawater. GR = GR(SO4<sup>2</sup>−), GRC = GR(CO3<sup>2</sup>−), A = aragonite, G = goethite, L = lepidocrocite, M = magnetite and Q = quartz, with the corresponding Miller index.

For the coupon immersed in natural seawater, corresponding to pattern (b), the main diffraction peaks are those of magnetite and goethite. If compared with pattern (a), the diffraction peaks of aragonite are slightly less intense, while those of lepidocrocite and GR(CO3<sup>2</sup>−) appear slightly more intense. Finally, the diffraction peaks of GR(SO4<sup>2</sup>−) are much more intense in pattern (b). As noted previously [8,17], the XRD analysis did not allow us to detect mackinawite FeS, because this phase remains in a nanocrystalline state.

In conclusion, the main observed difference between both kinds of coupons is the presence of FeS in the corrosion product layer formed in natural conditions a result of bacterial (SRB) activity. This FeS phase is nanocrystalline and was only identified via μ-RS analysis. However, the XRD analysis revealed other differences. In particular, it confirmed that the formation of GR(SO4<sup>2</sup>−) was indeed favored in the natural seawater of the harbor site.

#### *3.2. Characterization of the Precipitate Obtained in Abiotic Conditions*

In this section, and in Sections 3.3 and 3.4, the precipitates obtained by mixing NaOH with Fe(II) and Fe(III) salts are characterized.

Under the abiotic conditions considered here, and as previously studied [23], the initial precipitate was composed of a mixture of GR(Cl−) and GR(SO4<sup>2</sup>−). After 1 week of ageing, the proportion of GR(Cl−) drastically decreased. The XRD pattern presented in Figure 3 confirms this result: only the main peaks of GR(Cl−) are seen and their intensity is very small with respect to that of the peaks of GR(SO4<sup>2</sup>−). However, the ageing induced the formation of a small amount of magnetite Fe3O4. This evolution was attributed to the respective stability of the three phases [23], i.e., magnetite is more stable than GR(SO4<sup>2</sup>−), which, in turn, is more stable than GR(Cl−) in the conditions considered here.

**Figure 3.** XRD pattern of the precipitate obtained without addition of bacteria or organic compounds, after 1 week of ageing at RT. GR = GR(SO4<sup>2</sup>−), GRCl = GR(Cl−) and M = magnetite, with the corresponding Miller index. \* = unidentified peak.

This XRD pattern will be used below as a reference. The intensities of the diffraction peaks M311, GRCl003, GR001 and GR112 were determined by computer fitting as described in Section 2.5. The results of this analysis are presented in Table 3, with the intensity of the main peak of GR(SO4<sup>2</sup>−), i.e., GR001, being arbitrarily set as 100 in each case.

**Table 3.** Abiotic precipitate aged for 1 week: characteristics of the diffraction peaks GR001 and GR112 of GR(SO4<sup>2</sup>−), GRCl003 of GR(Cl−) and M311 of magnetite; 2*θ* = diffraction angle, in degree, *d* = interplanar distance (Å), FWHM = full width a half maximum, in degree, and *I* = peak intensity, with *I* = 100 for GR001.


The interplanar distance *d*001 obtained for GR(SO4<sup>2</sup>−) corresponded to the *c* parameter of the hexagonal cell. It was determined here at 11.00 Å, a value consistent with literature data [20]. The intensity of the GR112 peak was abnormally small, due to the preferential orientation of the GR particles. These particles comprised thin hexagonal platelets perpendicular to the *c* axis of the crystal structure [16]. For this reason, they were usually parallel to the sample holder. This preferential orientation increased the intensity of the 00l diffraction peaks. The interplanar distance *d*003 found for GR(Cl−) was also consistent with literature data [38].

#### *3.3. Influence of Bacteria*

The results of the study of bacterial growth and quantification are presented in Table 4. Each bacterial strain grew rapidly in the culture medium, reaching between 1.7 and 3.3 × 10<sup>9</sup> CFU mL−<sup>1</sup> after 24 h at 30 ◦C, and showing similar cell concentrations for the three strains. All the initial concentrated suspensions of bacteria contained more than 3 × 10<sup>11</sup> CFU mL−<sup>1</sup> of bacteria prior to mixing with the reagents used to prepare the GR precipitate. However, the results obtained after precipitation proved to be highly dependent on the considered bacterial strain. For *Pseudoalteromonas* IIIA004, no viable culturable bacteria could be enumerated. The same result was obtained for this strain 1 week later in both the precipitate and supernatant. In the first case, the Fe solid phases were precipitated and aged in a medium where the bacteria did not grow, likely because of cellular death due to the immersion of the bacteria in the NaOH solution 1.

**Table 4.** Numeration of bacteria (CFU mL−1).


In contrast, both *Micrococcus* IVA008 and *Bacillus* IVA016 remained viable and culturable throughout the experiments, even though the cell concentration after precipitation decreased from the initial concentrated suspension. In both cases, the results obtained in the precipitate after 1 week of ageing were similar to those obtained right after precipitation. A slight increase of the bacterial concentration was even observed, in particular for *Micrococcus* IVA008. In contrast, no viable culturable bacteria could be enumerated in the supernatant after ageing. This shows that the bacteria were mostly associated with the solid phases, more likely bound to the particles of Fe compounds. In this second case, the Fe compounds were precipitated and aged in a medium where bacteria survived and even developed.

The XRD analysis of the precipitates obtained after 1 week of ageing in the presence of bacteria provided results independent of the bacterial strain. The pattern obtained for the precipitate aged with *Micrococcus* IVA008 is displayed in Figure 4. It is mainly composed of the diffraction peaks of GR(SO4<sup>2</sup>−) that may all be clearly seen. Numerous additional, very small peaks are present, but they do not correspond to other expected Fe compounds, i.e., GR(Cl−) and magnetite, that are formed in abiotic conditions (Figure 3). These peaks were likely due to the various compounds, organic and inorganic, present in the concentrated bacterial suspension introduced in the system (see Section 2.3 for instance, where the composition of the culture medium is given). Only one small peak could be tentatively identified: located at 2*θ*hkl = 13.756◦, i.e., *d*hkl = 7.47 Å, it may correspond to the main diffraction peak of GR(CO3<sup>2</sup>−) [30,39], i.e., GRC003, as mentioned in Figure 4. NaHCO3 is present in the culture medium and bacteria produce carbonate species through their metabolic activity by oxidizing organic matter.

The XRD patterns obtained for the precipitates aged with bacterial strains *Pseudoalteromonas* IIIA004 and *Bacillus* IVA016 are both displayed in Figure 5. These patterns, like the previous one, did not show any trace of the diffraction lines of GR(Cl−) or magnetite. Numerous additional small peaks are also seen, located at similar positions regardless of the bacteria species. The main peak of GR(CO3<sup>2</sup>−) was not seen in the case of *Bacillus* IVA016. It was very small in the case of *Pseudoalteromonas* IIIA004, but could nonetheless be identified.

**Figure 4.** XRD pattern of the precipitate obtained with bacterial strain *Micrococcus* IVA008 after 1 week of ageing at RT. GR = GR(SO4<sup>2</sup>−) and GRC = GR(CO3<sup>2</sup>−), with the corresponding Miller index. \* = unidentified peaks.

**Figure 5.** XRD pattern of the precipitates obtained with bacterial strain *Bacillus* IVA016 (**a**) and *Pseudoalteromonas* IIIA004 (**b**) after 1 week of ageing at RT. GR = GR(SO4<sup>2</sup>−) and GRC = GR(CO3<sup>2</sup>−), with the corresponding Miller index. \* = unidentified peaks.

In conclusion, the presence of bacteria and associated organic matter prevented the formation of magnetite during ageing. The absence of GR(Cl−) showed that the bacteria and associated organic matter either accelerated the transformation of GR(Cl−) to GR(SO4<sup>2</sup>−) or prevented the formation of GR(Cl−) during the precipitation reaction.

The precipitates were also aged for 2 months at RT. Once again, the results were similar for all bacterial strains. The pattern obtained for the precipitate aged 2 months with bacterial strain *Micrococcus* IVA008 is displayed in Figure 6 as an example. It was very similar to that of the precipitate aged 1 week, i.e., the main diffraction peaks were those of GR(SO4<sup>2</sup>−), and the peaks of GR(Cl−) and magnetite were not seen. This demonstrates that the effects of the bacteria and the associated organic matter can persist for long periods. This may explain why GR(SO4<sup>2</sup>−) is favored in natural marine environments, as shown in Section 3.1.

**Figure 6.** XRD pattern of the precipitate obtained with bacterial strain *Micrococcus* IVA008 after 2 months of ageing at RT. GR = GR(SO4<sup>2</sup>−) and GRC = GR(CO3<sup>2</sup>−), with the corresponding Miller index. \* = unidentified peaks.

Finally, it can be noted that the numerous unidentified diffraction peaks, as well as the main peak of GR(CO3<sup>2</sup>−), were, compared to those of GR(SO42), more intense after 2 months of ageing. This can be observed visually by comparing Figures 4 and 6. To be more accurate, the intensities of the GR001 and GRC003 peaks were determined in each case by computer fitting, as described in Section 2.5. If the intensity of the GR001 peak was arbitrarily set at 100 in each case, then the intensity of the GRC003 peak slightly increased from 0.40(±0.01) to 0.45(±0.01) during ageing (from 1 week to 2 months). This may be attributed to weak bacterial activity.

#### *3.4. Influence of Acetate Ions*

The XRD pattern of the precipitate obtained with sodium acetate added as a reactant and after 1 week of ageing is displayed in Figure 7. The main diffraction peaks are once again those of GR(SO4<sup>2</sup>−), even though the acetate to sulfate concentration ratio, [CH3COO−]/[SO4<sup>2</sup>−], was equal to 2. Actually, it is well known that the double layered structure of GR compounds exhibits a stronger affinity for divalent anions [40,41], which explains why GR(SO4<sup>2</sup>−) forms instead of GR(Cl−) in seawater, even though the [Cl−]/[SO4<sup>2</sup>−] is high (about 19). The formation of GR(SO4<sup>2</sup>−) in this experiment was consistent with the findings in previous works.

As for the precipitate obtained in abiotic conditions without acetate, most of the diffraction peaks of magnetite were seen, together with the main peaks of GR(Cl−), i.e., GRCl003 and GRCl006. From Figures 3 and 7, it can be seen that the addition of acetate decreased the intensity of the diffraction peaks of GR(Cl−) and magnetite with respect to those of GR(SO42). The data obtained via computer fitting of diffraction peaks GR001, GR112, GRCl003 and M311 confirmed this (Table 5). The intensity of the main peak of magnetite decreased from 4.6 (Table 3) to 2.0, and that of GR(Cl−) from 7.0 to 1.8. The intensity of the GR112 lines, in contrast, increased from 3.2 to 5.9, which shows that the

preferential orientation is less pronounced. If the intensity of the magnetite and GR(Cl−) diffraction peaks were expressed with respect to the 112 diffraction peak of GR(SO42), the decrease due to the acetate ions would appear more significant.

**Figure 7.** XRD pattern of the precipitate obtained with acetate after 1 week of ageing at RT. GR = GR(SO4<sup>2</sup>−), GRCl = GR(Cl−), M = magnetite, with the corresponding Miller index. \* = unidentified peaks.

**Table 5.** Precipitate obtained with acetate and aged 1 week: characteristics of the diffraction peaks GR001 and GR112 of GR(SO4<sup>2</sup>−), GRCl003 of GR(Cl−) and M311 of magnetite; 2*θ* = diffraction angle, in degree, *d* = interplanar distance (Å), FWHM = full width a half maximum, in degree, and *I* = peak intensity, with *I* = 100 for GR001.


In conclusion, acetate ions induced the same effects as bacteria, but these effects were smaller and did not completely prevent the formation of magnetite and GR(Cl−).
