13C isotopic labeling to decipher the iron corrosion mechanisms in a carbonated anoxic environment

A two‐step corrosion experiment was performed on a ferritic steel (Armco) in a synthetic solution representing the Callovo–Oxfordian at 120°C. After the development of a carbonated corrosion product layer (CPL) during the first 15 days of the experimental step, corrosion front progression was investigated using 13C marked carbonate species during the second 15 days experimental step. CPL was characterized at each step, in terms of morphology (scanning electron microscopy), composition (energy‐dispersive spectroscopy), and structure (µ‐Raman). 13C corrosion product locations were analyzed by time‐of‐flight secondary ion mass spectrometry. Results evidenced that after a step of generalized corrosion, iron corrosion continues locally at the metal/CPL interface. These results suggest that although a protective siderite layer formed on the iron surface after 15 days, a local dissolution of the carbonate layer at the M/CPL interface occurred. A galvanic effect is developed between the bared surface (anode) and the covered one (cathode). This activates iron oxidation. The precipitation of carbonate corrosion products to the metal/CPL interface is possible by the diffusion of 13CO32− ions from the bulk through the siderite layer.


| INTRODUCTION
Iron corrosion in carbonate environment concerns a lot of applications, from the field of oil and gas production, [1] to nuclear waste disposal, [2] or concrete steel reinforcement. [3]Understanding corrosion phenomena in such an environment is an essential tool for the design, dimensioning, and maintenance of these metallic materials.
In a carbonate environment, the development of carbonate corrosion products such as siderite FeCO 3 has been widely studied in the literature.It is now accepted that the presence of such a corrosion product layer (CPL) reduces the iron corrosion rate by acting as a diffusion barrier to electrochemically active species. [4]The nucleation and growth process of the FeCO 3 formation have been considered in light of numerous factors influencing its physical properties and, thereby its protective nature.Among them are K sp , ionic strength, pH, [5][6][7] temperature, [8,9] brine chemistry, [10][11][12][13] CO 2 partial pressure, and iron carbide. [14,15]n a CO 2 environment, it is assumed that localized corrosion starts with local damage of the siderite film due to many reasons usually related to fluid properties (temperature, pressure, brine chemistry) and flow conditions (velocity, flow regimes). [16,17]Although a protective siderite layer has been formed on a mild steel surface, any localized damage to this layer may initiate localized corrosion.
Many studies have focused on the first steps of the formation of the carbonate layer to understand its protectiveness.][20] Several studies have questioned the properties of these bilayers as well as the steps of their formation during the corrosion processes.First Palacios et al. [18] suggested that the two scales, inner and outer, presented different compactness and adherence.Following their results obtained at 71°C, they suggested that the outer layer of siderite precipitated on the top of the inner one already formed.Moreover, the inner layer is adherent to the metal whereas the outer one is easily removed from the inner one.More recently Gao et al. [9] studied the nucleation and growth of the carbonate layer at 75°C and 90°C at duration until 240 h.They observed the formation of a double layer observed by surface and cross-section observations (scanning electron microscopy-SEM).They completed the study with linear polarization resistance and EIS monitoring.Their conclusions provided information on the formation process of this bilayer.After the first hours, before what the authors qualified as a critical point, the steel corroded showing an increase in the measured linear polarization resistance.This seems to be a significantly abrupt decrease in the corrosion rate.The inner layer formed during this first step as it was proven by EIS spectroscopy performed at the critical point.Then after this critical point, the outer layer grew on the top of the inner layer.This was observed by EIS, which was then modeled by a two-layer equivalent circuit.Authors specified thanks to SEM observations on transverse section of their samples that the two observed layers formed apart from the original surface of the metallic coupon which is still visible thanks to a local higher porosity on the cross sections.Thanks to their electrochemical analyses, they showed the high resistance properties of the inner layer to corrosion.The formation of the outer layer denoted for these authors that the protective layer reached a uniform stage for the whole metallic surface.Complementary to this study Li et al. [19] studied the corrosion of steel in carbonate solutions at 80°C and Tanupabrungsun et al. [21] at 120°C observed the formation of a double layer after 20 h to 10 days depending on the pH imposed in the solution of their set-up.De Motte et al. [20] assumed that the limit between the inner and outer layers formed during 10-12 days at 80°C in a carbonated solution under controlled pH (6-6.6)corresponded to the original surface of the steel coupons.The so-called "pseudo-passivation" occurred for these authors while the measured OCP increased.The layer formation mechanisms provided by these authors suggested that in the first stage, the crystal growth of the siderite layer dominated leading to the formation of a porous layer.Then, with increasing time of experiment, the inner layer crystals filled up by further precipitation of siderite, before the precipitation of the outer layer occurred.For the authors on the outer part of the corrosion layer, the iron carbonate supersaturation was lower inducing a higher porosity of the outer layer.Over time the diffusion limitation of the reactive species could be measured by the authors thanks to EIS spectroscopy and the layers became more compact after the SEM observations.Last, a study was conducted by Lotz et al. [22,23] at 120°C in a Callovo-Oxfordian (COx) porewater containing carbonate in solution.After 1 month it was shown that the bilayer of siderite containing calcium was still present at such duration showing a compact morphology on both ferritic and ferrite-pearlitic metallic samples.The location of original surface of the metallic samples was located at the interface of both parts of these siderite layers for both samples.Moreover, nanometric nodules of magnetite were detected by TEM inside the inner part of the siderite layer. [22,23]After 3 months, the same layout was observed with a bilayer composed of carbonate phases.The only difference was due to the formation of a local corrosion area at the metal/corrosion layer interface containing iron silicate phases.This allowed us to conclude that at this stage the corrosion front was located in these local corrosion areas. [23]ummarizing the results of the experiments performed in carbonated solutions presented in the literature, whatever the temperature, the protection mechanisms presented are based on the control of the diffusion of the carbonate species through the corrosion layer acting as a diffusion barrier toward the metal.This leads to the mass transfer limitation and the decrease of the anodic reaction.Iron corrosion front progression relies heavily on the cycle of iron dissolution and deposition.However, the corrosion front progression accompanying local damage at the metal/CPL interface remains poorly studied after the siderite bilayer is precipitated.Complementary to these observations, many studies investigated the influence of the presence of cementite inside the siderite layer on the diffusion barrier effect.In a recent study conducted by Owen et al. in a carbonated deaerated medium, it has been demonstrated that cementite present in the siderite layer induced a local acidification that can dissolve locally the siderite.This was explained by the possible spatial separation of the half-chemical corrosion reaction inducing the delocation of the cathodic reaction along the conductive cementite that generates a decrease of pH by the release of H + or consumption of OH − . [24]Thus in a system containing Fe, FeCO 3 , and Fe 3 C, it has been suggested that cementite was mainly responsible for galvanic interaction, rather than FeCO 3 . [25]However, it has been shown recently that the presence of magnetite islets in the siderite corrosion layer of a pure iron coupon raises questions on the possibility of delocalizing the cathodic reaction inside the siderite layer. [22]he objective of this study was to trace the iron corrosion front progression by investigating diffusion and precipitation of carbonate species while a carbonate bilayer is already formed after 15 days of corrosion in the same conditions than previous studies. [22,23]A two-step corrosion experiment in the presence of 13 C labeled compounds was set up to characterize the location, morphology, and nature of the corrosion product formed during the front progression mechanism of iron corrosion.

| Samples
A pure α-Fe named Armco® was chosen because of its microstructure free from iron carbide to focus on the corrosion of ferrite only.The chemical composition is given in Table 1.Before the corrosion experiment, samples were polished with SiC papers (from 800 down to 4000 mesh) using ethanol as a lubricant.Between each polishing paper, samples were rinsed in an ethanol beaker and placed for 2 min in an ultrasonic bath to remove SiC grains.

| Corrosion experiments
To locate carbonate species precipitation zones inside the corrosion layer once a carbonate layer formed, a two-step corrosion experiment was set up in the same cell during the whole experience.First, a corrosion experiment was performed in an autoclave at 120°C with P(CO 2 ) = 0.5 bars under anoxic conditions.The corrosion media corresponds to a synthetic solution representative of the COx porewater in equilibrium with Bure argillite according to Gaucher et al. [26] (Table 2).
Once the samples were immersed in the solution, the autoclave was sealed and primary vacuum pumping was performed for air removal.Then, to achieve an equilibrium pressure of 0.5 bar, CO 2 was injected into the autoclave at a pressure of 3.3 bars.This inlet pressure depends on the ratio between the volumes of gas and solution.
Speciation of carbonate species was calculated using the BRGM Thermoden database. [27]The IAPWS Guide to Henry's Constant site (IAPWS G7-04) [28] was used to determine the inlet pressure.Finally, an overpressure of 1 bar He was injected to prevent the solution from boiling at 120°C.
After 15 days, the autoclave was opened to remove the solution and replaced with 100% 13 C marked medium where all the carbonated species were doped with 13 C: gas ( 13 CO 2 gas, Eurisotop) and dissolved (NaH 13 CO 3 , Eurisotop).The same protocol was followed to obtain the same equilibrium pressure of CO 2 as above.The experiment was extended for another 15 days under the same experimental conditions (Figure 1).
Following these two steps, three ferritic samples were corroded at different stages of the protocol.The first one, named in the following 12 C reference sample, corroded during the first corrosion step, in a 12 C environment, and the second sample during the second step experiment, in a 13 C environment and named 13 C reference sample.Both 12 C reference sample and 13 C reference sample were analyzed by ToF-SIMS to determine the signal of carbonate phases formed in 12 C and 13 C pure solutions, respectively.Finally, the third sample named the 12 C 13 C sample was corroded during the whole two-step experiment and was analyzed to determine the location of precipitation of the carbonate phases during an already formed carbonate layer.Upon stopping the experiment, the samples were dried under a primary vacuum for 1 week.After the experiments and before the analyses, samples were embedded in epoxy resin (EpoFix, Struers) and then cut with a precision diamond saw (Minitom, Struers), in order, and the cross sections were observed.The cross-sections were ground with SiC paper (up to grade 4000) under ethanol and then polished up to 0.25 µm using diamond paste on a velvet disk.

| SEM-energy-dispersive spectroscopy
The morphology and chemical composition of the corrosion layers were investigated using a Field Emission Gun-Scanning Electron Microscope (JEOL SEM 7001F) coupled with an energy-dispersive X-ray (EDX) Spectroscope (INCA from Oxford Instrument).Analyses were performed at 10 kV accelerating voltage and with a probe current of around 6 nA.The acquisition and data treatment were performed using Aztec software developed by Oxford Instruments.

| µ-Raman spectroscopy
µ-Raman analyses were performed with a Renishaw Invia Reflex spectrometer equipped with a doubled Nd: YAG laser emitting at 532 nm and a microscope to focus the beam on the surface.The spectrometer was calibrated in energy with a silicon wafer (reference peak at 520 cm −1 ).The analyses were carried out with a beam intensity of the order of a few hundred microwatts to avoid the transformation of the analyzed phases under laser heating.Spectra were acquired using Wire 3.4 software with a resolution of 2 cm −1 between 200 and 1200 cm −1 and a dwell time of 60 s.

| Time-of-flight secondary ion mass spectroscopy
Time-of-flight secondary ion mass spectrometry (ToF-SIMS) was used to study the diffusion of carbonate species and the precipitation of the CPLs through 13 C isotopic mapping.
ToF-SIMS measurements were performed with a TOF-SIMS 5 spectrometer (IONTOF GmbH Germany) equipped with a bismuth liquid metal ion gun (LMIG) for primary ions and Cs + for sputtering.Acquisitions were first performed in burst mode to locate precisely the M/ CPL interface (lateral resolution of 200 nm with mass resolution M/ΔM < 2000).Although the spatial quality is excellent, this mode does not allow precise quantification of the elements.Thus, a second acquisition in bunch mode was then performed to quantify the 13 C enrichment and be able to avoid overlapping with 12 CH − (lateral resolution of 1 µm with mass resolution M/ ΔM > 10 000).
Before the analysis, the surface of interest was etched with Cs + 2 kV, 180 nA, rastered over an area of 500 × 500 µm 2 for 5 min.The purpose of this last operation was to clean the surface of any adsorbed hydrogen species that could affect the measurements and isotopic ratio calculations.The surfaces analyzed were about 200 × 200 µm 2 .At least 500 scans were done for each acquisition to obtain a good counting statistic.Noninterlaced mode (Bi + 25 kV, Cs + 2 kV, 500 × 500 µm 2 ) was used to avoid surface contamination redeposition.
For 13 C enrichment calculations, the reference ratio was taken from the 12 C sample.The isotopic splitting was measured from the relative deviation of the isotopic ratio of the sample from a reference value: δ I I I I = ( / ) /( / ) i i a n i a 0 with I i the peak intensity of the isotopic marker and I a the peak intensity of the reference atom.I I ( / ) i a n corresponds to the experimental isotopic ratio and I I ( / ) i a 0 , the reference isotopic ratio ( 12 C sample).Results were expressed with a 2σ error.Note that 13  the metallic substrate appeared entirely covered by a gray bilayer with a homogeneous color intensity.Chemical compositions by energy-dispersive spectroscopy (EDS) on the corrosion layer revealed a homogeneous distribution of Fe (47 wt%), O (41 wt%), and Ca (10 wt%) (Na, S, Mg <1 wt%).Brighter islets were also visible at the M/CPL interface and composed mainly of Fe (72 wt%) and O (28 wt%).
In the gray matrix, structural information acquired by Raman spectroscopy (Figure 3a) evidenced bands at 270 and 1090 cm −1 characteristic of carbonate groups.The correlation of chemical and structural information in this region allowed identifying siderite rich in Ca in both the inner and outer parts of the layer.In addition, Raman spectra acquired in the brighter islets at the M/CPL interface revealed band characteristics of magnetite (Figure 3b).
Thus, the CPL was composed of a Ca-siderite matrix with magnetite islets at the M/CPL interface.The morphology and nature of the CPL are in agreement with previous results acquired in the same corrosion environment after 1 month. [22]I G U R E 2 Cross-section observation of 12  Isotopic 13 C analysis performed by ToF-SIMS in the siderite matrix showed that a percentage of 1.06 ± 0.047% is measured while summing the signal obtained on the whole corrosion layer.This is close to the natural abundance percentage of 13 C, in agreement with literature data (see, e.g., [29][30][31] ).

| 13 C reference sample
The 13 C reference sample was corroded for 15 days only during the second step of the experiment.On the SEM micrograph appeared a gray bilayer matrix and light gray micrometric islets at the M/CPL interface (Figure 4).
Chemical compositions acquired by EDS revealed the homogeneous presence of Fe (47 wt%), O (41 wt%), and Ca (10 wt%) and less than 1 wt% of Na, S, and Mg.Brighter islets are also visible at the M/CPL interface and are composed mainly of Fe (72 wt%) and O (28 wt%).
Structural analysis acquired by Raman spectroscopy confirmed the presence of siderite and magnetite at the M/CPL interface as observed in the 12 C samples (Figure 5).
Isotopic mapping of 13 C performed by ToF-SIMS by considering the spectra obtained on a zone corresponding to the siderite matrix, allowed extracting the maximum 13 C enrichment in the siderite matrix, compared to the 12  | 791 doped abundance did not reach 100%, it is assumed that some carbonate species might have dissolved, from the remaining 12 C/ 13 C sample during the two steps or from adsorbed carbonates on the autoclave walls.In this study, this percentage is considered the maximum enrichment possible detectable on the siderite matrix.
3.3 | Location of 13 C enrichment in the 12 C CPL after 1 month experiment-12 C/ 13 C sample Once the 12 C natural abundance and the maximum 13 C enrichment are established for the siderite layer, it is possible to study the sample corroded first in the 12 C medium for 15 days and another 15 days in the 13 C one.ToF-SIMS mappings were performed on the cross-section of the ferritic sample corroded through the two-step experiment (Figure 6).The ionic map of O − in burst mode (Figure 5a) allows for differentiating the CPL from the metal and the resin.Figure 5b corresponds to the map of the 13 CHO 2 − / 12 CHO 2 − ratio in burst mode, which allows locating 13 C enrichment with a good spatial resolution.One can notice a slight 13 C enrichment (lighter pixels at some rare locations between the middle of the CPL and the M/CPL interface).This location corresponds very likely to the inner CPL characterized previously in a similar system by Lotz et al. [22] Although the enrichment is detected in burst mode, it is not possible to quantify it due to the very low number of counts of 13 CHO 2 − using this mode.To quantify the amount of 13 C in this region, the same ToF-SIMS mappings were acquired in bunch mode (Figure 7).
Results are presented in Figure 6 for the ionic fragment O − (a) and the 13 CHO 2 / 12 CHO 2 ratio (b).
To determine the δ 13 C enrichment in the CPL, four ROIs were selected (Figure 7a).The first ROI corresponds to the area where the 13 C enrichment was observed on the ionic image and the three others were selected at other places in the CPL (numbers in Figure 7a).The results are presented in Table 3.
For the ROI 1, the deviation of the isotopic ratio from the reference value (14.83 ± 1.15) was more than a factor of 10 times higher than those observed in the other three ROIs acquired at other places of the corrosion layer (from 0.24 to 1.12).This local enrichment in 13 C of the M/CPL interface indicated that CPL growth took place locally at the M/CPL interface.The 13 C doped abundance associated with this deviation was 16.84% in ROI 1.This abundance did not reach the maximum enrichment estimated on the 13 C reference sample.This result indicated the presence of both 12 C and 13 C atoms in the doped carbonate area at the M/CPL interface.This observation suggested that the local growth of the siderite layer is also coupled with a local destabilization of the already-formed siderite matrix after 15 days of corrosion. [32]This result was supported by iron corrosion studies in a CO 2 environment that linked localized corrosion to a partial breakdown of the protective FeCO 3 corrosion product film.
First, experimental studies performed in a CO 2 environment investigated the global stability of the carbonate layer.Li et al. [19] showed that pseudopassivation due to siderite formation occurred only above pH 6 at 80°C.At lower pH, the siderite layer is not stable and protective.In our study, a local decrease in the porewater pH might change the siderite stability and explain its local destabilization.In a complementary study of iron samples, it has been shown that magnetite nodules were present inside the inner part of the siderite layer.It can be questioned if these magnetite nodules could be electrically connected.Indeed this was observed on archeological artifacts corroded in anoxic carbonated medium. [33,34]These electrically connected magnetite nodules could constitute a conductive path inside the siderite matrix and act as the cementite to induce delocation of the cathodic reaction, thus a local decrease of pH inside the siderite. [25]Other studies were performed to study the carbonate layer removal.Using a rotating cylinder glass cell, Ruzic et al. [35,36] studied the mechanical and chemical removal of the iron carbonate layer.Their results evidenced that mechanical removal in undisturbed single-phase flow does occur locally while chemical removal is governed by mass transfer and depends mainly on pH value (local acidity) and fluid velocity.Using a similar improved experimental set-up, Han et al. [16,37] confirmed that iron carbonate can be partially removed by mechanical stresses (flow rate), chemical dissolution (undersaturated solution), or by the synergy of both removal modes.In our study, the absence of any fluid flow suggests that the preferential breakdown mechanism was chemical.Moreover, complementary to experimental data, modeling studies were performed to investigate the evolution of a metallic surface affected by generalized corrosion.Using the cellular automata model, Pérez-Brokate et al. [38] evidenced that the generalized corrosion process switched to localized corrosion due to a local increase of the electrolyte acidity at the cathodic site while half corrosion reactions are spatially separated.The competition between the generalized and localized corrosion induced cycles in the corrosion processes and impacted the global corrosion kinetics.Based on the literature, Figure 8 proposes hypotheses on the corrosion front progression steps when a carbonate layer of about 10 µm already precipitated.When the anodic and cathodic reactions occur homogeneously around the metallic surface, the corrosion process is uniform and a homogeneous siderite bilayer grows through crystal nucleation and growth processes [4,38] (Step 1).Partial breakdown of the carbonate inner layer is probably activated by a chemical stress such as under/over-saturation or local acidity increase as suggested in both experimental and modeling approaches discussed previously.The destabilization of the carbonate layer induces a local dissolution and formation of ionic species: FeCO 3 ↔ Fe 2+ + 12 CO 3 2− (Step 2). [32]39][40] T A B L E 3 13 C/ 12 C ratio and enrichment were calculated using the CHO 2 fragment on the selected ROIs on the siderite layer (located in Figure 6a) in the 12   ). [32]| CONCLUSION A two-step corrosion experiment was set up to investigate carbonate diffusion and precipitation in an already developed and homogeneous siderite matrix, thanks to 13 C labeled carbonate species.The results highlighted the following: -The CPL growth continues locally at the M/CPL interface.This confirms that carbonate species diffuse through the outer layer and that the formation of the siderite layer continues at the inner layer of siderite.However, the results indicate that after 15 days at 120°C the bilayer is already compacted and ensures a role of a barrier of diffusion as only one localized corrosion was identified on the whole corrosion layer observed on a cross-section.After 1 month, the corrosion rate seems drastically reduced.-The CPL growth is the result of a local destabilization of the siderite film at this interface, followed by iron oxidation activation through a galvanic cell and reprecipitation with additional carbonate ions that diffused from the environment to the M/CPL interface.Although no cementite is present, the role of the magnetite micro to nanometric islets observed after 1 month of corrosion in a similar set-up. [22]Their possible connection could favor the delocation of the cathodic reaction inside the inner part of the bilayer of siderite.

CHO 2 − 3 |
and12 CHO 2 − fragments were used instead of13 C/ 12 C for the enrichment observation due to 12 C and 13 C ionization issues.RESULTS AND DISCUSSION 3.1 | 12 C reference sample 12 C reference sample was corroded during the first 15 days of the experiment in a 12 C medium.The morphology and chemical distribution of the CPLs of this sample examined by SEM (Figure 2) showed that F I G U R E 1 Schematic diagram of the experimental protocol alterning from 12 C to 13 C based solution after 15 days of corrosion.[Color figure can be viewed at wileyonlinelibrary.com]

3
C reference sample, morphology of the CPL observed by BSE-SEM and below associated chemical distribution by EDS.Points a and b locate µ-Raman analysis of Figure 2. [Color figure can be viewed at wileyonlinelibrary.com]Raman spectra acquired on point a and b located on Figure 2.

F G U R E 5
C reference sample.The result, δ = 71.71± 3.58 corresponded to a doped abundance percentage of 79.5 ± 1.48%.As the F I G U R E 4 Cross-section observation of 13 C reference sample, top shows the morphology of the CPL observed by BSE-SEM, and below are associated chemical distribution, EDS.Points a and b locate µ-Raman analysis.[Color figure can be viewed at wileyonlinelibrary.com]Raman spectra acquired on point a and b. located on Figure 4.LOTZ ET AL.

F 2 −/ 12 CHO 2 −
I G U R E 6 Ionic maps of the 12 C/ 13 C sample-a corner of the coupon is observed: (a) the O − fragment and (b) the 13 CHO ratio acquired on the ferritic sample re-corroded for 15 days in a 13 C-doped environment after being corroded the first 15 days in a 12 C-doped environment, ToF-SIMS, burst mode.[Color figure can be viewed at wileyonlinelibrary.com]

7 12 CO 3 (
Ionic maps on cross-section of the12 C/13 C sample (corner of the sample): (a) the O − fragment with the region of interests (ROIs) location mentioned in Table3and (b) the13 CHO 2 − / 12 CHO 2 − ratio acquired on the ferritic sample re-corroded for 15 days in a 13 C-doped environment, ToF-SIMS, bunch mode.[Color figure can be viewed at wileyonlinelibrary.com]LOTZ ET AL. | 793 Meanwhile, 13 CO 3 2− diffuses from the solution to the M/CPL interface.We suggested that this carbonated supply added to the ferrous supply made it possible to reach again a saturation ratio of Kps in favor of the precipitation of siderite.Thus, the ferrous and carbonate ions precipitated together, forming a new corrosion layer.Fe 2+ + 12 CO 3 x + y = 1) (Step 4 1313C sample.