Introduction

Sound concrete corresponds to an environment favouring the passivation of steel reinforcement and its subsequent protection against corrosion and degradation. However, the pollution of concrete by aggressive species such as chloride and carbon dioxide leads to a decrease of the pH and a breakdown of the passive film. It results in the corrosion of the steel reinforcing bars (rebars) and, in the long term, in the deterioration of the concrete. Carbonated and chloride polluted concrete corresponds mainly to basic aqueous media with large concentrations in both chloride and carbonate anions. Iron (II) and iron (III) salts as well as green rusts (GRs), that is iron (II)–iron (III) hydroxy-salts, constitute some examples of important corrosion products possibly involved in the corrosion of steel rebars in concrete. According to previous works devoted to the chloride and carbonate forms of GRs [1–4], concrete would be one of the most favourable media for their appearance. A first study [5] was then devoted to the early stages of the corrosion of steel in electrolytes simulating concrete and to the ability of phosphate to inhibit this degradation. It confirmed that GRs are the main products of the initial stages of the corrosion process and demonstrated the efficiency of PO 3−4 as an inhibitor.

Since calcium nitrite Ca(NO2)2 is an inhibitor commonly used in concrete today, the accurate knowledge of the possible interactions between GRs and NO −2 ions is a clue for the understanding and optimisation of this inhibition. This work was then focused on the mechanisms of the inhibition of the corrosion of iron in concrete pore waters based on saturated Ca(OH)2 solution added with chloride, carbonate species and nitrite. A previous work has shown that electrochemical results in saturated calcium hydroxide were similar to those obtained in concrete [6]. The present study was performed by means of electrochemical methods, such as galvanostatic, polarisation curves, and impedance spectroscopy experiments, coupled with analyses of the corrosion products by transmission Mössbauer spectroscopy (TMS). The role of nitrite will be compared with that of phosphate.

Experimental

Electrochemical tests

Three types of solutions were used for simulating the conditions met in various types of concrete. Their molar composition and pH are given in Table 1. Solution S1 is intended to simulate a fresh concrete containing some chloride ions; solutions S2 and S3 simulate carbonated concrete containing Cl− ions. NaNO2 was used to provide nitrite ions. The amount added was related to the Cl− concentration and expressed with the ratio R = [NO −2 ]/[Cl−]. Three values were considered: R = 0, i.e. the reference solution without nitrite, R = 0.4 and R = 1.

Table 1 Composition and pH of the solutions considered in order to simulate concrete media
Full size table

α-Iron foils of 50 μm thick, provided by Goodfellow® were used to prepare 34 mm diameter (i.e. a surface of 9 cm2) working electrodes. These electrodes were cleaned with acetone and distilled water and dried in warm air. Their sides were coated with an electric insulator while a copper wire was welded at an extremity to provide the connection to the electrical circuit needed for electrochemical testing. It was finally set in a three-electrode cell containing 250 ml of solution, with a saturated calomel electrode for reference and a platinum wire as counter-electrode. The solution was slightly agitated and maintained at a temperature of 25 ± 1 °C during tests. A Radiometer-Tacussel® PGP201 potentiostat–galvanostat monitored by the Voltamaster-1® software was used for galvanostatic and polarisation curves measurements. In each case, the corrosion potential E cor was measured and plotted against time during 1 h before experiments. For galvanostatic test, a current density of 0.5 mA cm−2 was then maintained for 5 h to rapidly induce the formation of a sufficient amount of corrosion products. Polarisation curves were drawn from the corrosion potential up to an anodic current of 200 μA cm−2 at a scanning rate of 25 mV min−1, followed by a similar scan in the reverse direction down to −1255 mVSHE. Impedance spectroscopy was performed within the same cell. The equipment consisted of a potentiostat (Solartron SI 1286®), an analyser of transfer function (Solartron FRA SI 1250®), a filter (Kemo VBF8®) and a PC computer with FRACOM software for plotting impedance by controlling the analyser of transfer function and a SIMPLEX software for modelling experimental data by an “Equivalent Electric Circuit”. Plotting on the same graph the experimental and calculated Nyquist diagrams checked the validity of the calculated results. Steel was polarised at ±10 mV around its zero-current potential. Frequency was ranging between 65 kHz and 10 mHz, with 5 frequency values per decade.

It can be noticed that for reproducibility, each electrochemical test was repeated three times. The experimental error did not exceed 10%.

Analyses of the solid phases

The products obtained after galvanostatic tests were scraped from α-iron foils and analysed by TMS at room temperature. In order to avoid any subsequent oxidation, these products were immediately set in the sample holder and kept under inert helium atmosphere during spectrum accumulation. The Mössbauer apparatus consists of a conventional constant acceleration spectrometer calibrated with a 25 μm α-iron foil at room temperature.

Results

Electrochemical tests

Figure 1 shows the behaviour of the corrosion potential E cor with time for α-iron dipped in S1, S2 and S3 solutions with ratio R = [NO −2 ]/[Cl−] = 0, 0.4 and 1. It appears that the stabilisation (±30 mV) occurred rapidly, after approximately half an hour. In the absence of nitrite (R = 0), the potential decreases with time, which denotes an active behaviour due to chloride ions. Moreover, the measured values decrease when the pH increases. In the presence of nitrite, the potential stabilises at more noble values. The increase in E cor, e.g. about 300 mV for R = 1, suggests a predominant anodic inhibition.

Fig. 1
figure 1

Influence of nitrite on the evolution of the corrosion potential E cor versus time in (a) solution S1, (b) solution S2 and (c) solution S3. R = [Cl−]/[NO2 −]

Full size image

Figure 2 gives the electrochemical impedance diagrams at zero current potential in the Nyquist’s plot for α-iron dipped in S1, S2 and S3 solutions with ratio R = [NO −2 ]/[Cl−] = 0, 0.4 and 1. In all cases, they are made of a capacitive loop, which is slightly centred off the real axis. When adding nitrite ions, the outlook of the impedance diagrams is not modified. A simple Randles equivalent circuit R e(R t C dl) (electric resistance, R e, charge transfer resistance, R t and double layer capacitance, C dl) was used for fitting the EIS Nyquist plots. It must be noted that the best fit was obtained by introducing a symmetrical Cole–Cole distribution on the time constant. In all cases, a discrepancy, which was less than 5%, was observed between experimental and calculated values of the impedance. Figure 2d shows the comparison between experimental and calculated typical EIS Nyquist plot. Thus, the electric model describes the interface and consequently the experimental diagrams can be used to determine the values of the various elements in the electric circuit. Table 2 gathers the values obtained in each case for these parameters. It appears that the electrolyte resistance R e, which is in series with the other elements, decreases slightly with nitrite addition. The solution becomes more conducting but its nature does not change. The resistance R t that corresponds to the diameter of the capacitive arc increases strongly with nitrite content The interface capacity C dl is close to or higher than the double layer capacitance. According to some works [7], the increase of the C dl capacity can be explained either by an increased iron surface caused by the corrosion process or by the higher concentration of charges in the diffuse double layer. Generally speaking, the resistance R t increases and the capacitance C dl decreases with inhibitor addition. Thus, the inhibitor diminishes drastically the metal dissolution rate and consequently the corrosion intensity. The interface response can be related to a charge transfer process.

Fig. 2
figure 2

Nyquist diagrams of α-iron in (a) solution S1, (b) solution S2 and (c) solution S3, with or without nitrite added as an inhibitor. (d) Experimental and calculated EIS diagrams: example of α-iron in solution S1 at R = 0. R = [Cl−]/[NO2 −]

Full size image
Table 2 Characteristic parameters of the α-iron-solution-inhibitor interface
Full size table

As pitting corrosion is important in chloride environment, the pitting potential E p and the repassivation potential E rep are of interest. It is to be noticed that E p is the potential where a first pit appears, when no initial pit exists before the potential scanning starts. E rep is the potential where the reverse scan intersects with the forward scan. It corresponds to the potential value that must be applied in order to repassivate the well-grown pits where the electrolyte has changed towards more severe conditions. In such pits, the pH decreased and the chloride concentration increased. In the absence of pitting, forward and reverse scans coincide. Figure 3 displays a typical example of the polarisation curves for α-iron dipped in S1 solution with ratio R = [NO −2 ]/[Cl−] = 0, 0.4, 1 and Table 3 gathers these specific potentials for all the studied systems. It appears that E p and E rep shift towards more noble values whereas the passivation plateau increases with the inhibitor concentration. Therefore, nitrite ions increase the resistance of iron to pitting corrosion in solutions simulating polluted concrete.

Fig. 3
figure 3

Polarisation curves of α-iron dipped is solution S1 at various R = [Cl−]/[NO −2 ] values

Full size image
Table 3 Pitting potential E p and repassivation potential E rep at R = [NO −2 ]/[Cl−] = 0, 0.4 and 1 in S1, S2 and S3 solutions
Full size table

Figure 4 displays the galvanostatic curves. In each case, the evolution is typical of an active system. The potential drops rapidly then it stabilises at potential values approximately 200 mV larger than the corrosion potential. In each case, the metal surface is covered with corrosion products at the end of the experiment.

Fig. 4
figure 4

Potential versus time curves during galvanostatically induced corrosion of α-iron dipped in various solutions at various R = [Cl−]/[NO −2 ] values. (a) Solution S1, (b) solution S2 and (c) solution S3

Full size image

Analyses of the corrosion products

TMS analyses of the products obtained without NO −2 (R = 0) with or without phosphate [5]

TMS analysis at room temperature of the corrosion product formed after galvanostatic experiments on the iron foil in the solution S2 free of nitrite (R = 0) is presented in Fig. 5. This product did not adhere to the metal and was easily removed from the surface. Its spectrum is mainly composed of two ferrous quadrupole doublets, D 1 and FC, characterised by large isomer shifts δ and quadrupole splittings Δ (Table 4), and one ferric doublet D 3 that corresponds to smaller δ and Δ values. The ferrous doublet D 1 and the ferric doublet D 3 are attributed to a green rust compound. The Δ value of the ferrous component, i.e. 2.35 mm s−1, was already observed for such compounds at room temperature [4]. Moreover, the D 1/D 3 intensity ratio, that is the Fe(II)/Fe(III) ratio, is about 3, a value often met in GRs [1, 4, 8–10]. Since CO 2−3 ions are preferred over any other anion, and especially monovalent ones [11, 12], the GR observed in solution S2 is necessarily GR(CO 2−3 ). For instance, in NaHCO3 + NaCl solutions with [Cl−]/[HCO −3 ] ratio of 40, the corrosion process was demonstrated to produce GR(CO 2−3 ) [3]. Doublet FC is characterised by a smaller Δ value of 1.81 mm s−1. Such a value at room temperature is characteristic of the iron (II) carbonate FeCO3 [13].

Fig. 5
figure 5

Mössbauer spectrum measured at room temperature of the corrosion product obtained after galvanostatic experiment in solution S2 without any inhibitor (R = 0)

Full size image
Table 4 Mössbauer hyperfine parameters at room temperature of the corrosion products obtained after galvanostatic experiments
Full size table

The analysis related to solution S3 (not represented) confirmed the presence of both GR(CO 2−3 ) and FeCO3, related to solution S1, revealed the presence of GR(Cl−) and Fe(OH)2.

Similar experiments were performed in the presence of phosphate with ratios [Cl−]/[PO 3−4 ] equal to 0.4 or 1 [5]. The TMS analysis at room temperature of a typical corrosion product formed after galvanostatic experiments is presented in Fig. 6 and the corresponding data in Table 4. This product, obtained in the solution S2 for [Cl−]/[PO 3−4 ] equal to 1, was found in small amounts localised around a few pits on the metal and adhered to the surface. It was then scraped from the iron foil, which explains the presence of the sextet due to α-Fe. The main difference between this spectrum and the previous one (Fig. 5) is the presence of an important Fe(III) doublet F. Additional analyses revealed that it corresponded to ferrihydrite, the most poorly ordered form of Fe(III) oxyhydroxides. GR(CO 2−3 ), identified by its Fe(II) doublet D 1, is still present, indicating that the corrosion process is the same. The average oxidation number of iron increased indicating that the extent of the oxidation reaction is larger. This is mainly due to the fact that the corrosion was localised on a few pits, implying a larger current density, which is a more rapid oxidation process. The presence of ferrihydrite as the final product of this oxidation may result from a specific action of phosphate. It was actually demonstrated that orthophosphate could induce the transformation of GR(CO 2−3 ) into ferrihydrite [14].

Fig. 6
figure 6

Mössbauer spectrum measured at room temperature of the corrosion product obtained after a galvanostatic experiment in solution S3 with PO 3−4 as an inhibitor

Full size image

TMS analyses of the products obtained for R > 0

The corrosion products obtained galvanostatically in the presence of NO −2 ions adhered to the iron surface. They were scraped from the metal, revealing that a lot of pits have formed under the rust layer. Two typical Mössbauer spectra of these products, namely that related to solution S1 with R = 1 and that related to solution S3 with R = 0.4 are presented in Fig. 7. Other spectra (not represented) are similar whatever the solutions considered and whatever the value of R. They all consist of the spectral component of α-Fe accompanied by Fe(III) quadrupole doublets F 1 and F 2, and in some cases, by a minor Fe(II) doublet D 1. The hyperfine parameters of D 1 (Table 4) are typical of Fe(II) in GRs. F 1 is a typical doublet of FeOOH phases. It could be attributed whether to γ-FeOOH lepidocrocite [15], to poorly crystallised goethite, which gives rise to a so-called “superparamagnetic” spectrum at room temperature [16], or ferrihydrite [17]. Each of these compounds is a by-product of the oxidation of GRs [1–4, 14]. F 2, with a large Δ value, is characteristic of poorly crystallised and poorly ordered FeOOH phases.

Fig. 7
figure 7

Mössbauer spectra measured at room temperature of the corrosion products obtained after galvanostatic experiments in solutions containing NO −2 as an inhibitor. (a) Solution S1 with R = [Cl−]/[NO −2 ] = 1, (b) solution S3 with R = [Cl−]/[NO −2 ] = 0.4, (c) enlargement of the previous spectrum in the region showing one peak of doublet D1

Full size image

The TMS spectra of the corrosion products resulting of the action of nitrite can be compared to the spectrum of Fig. 6 that presents the analysis of the corrosion products obtained in the same conditions with phosphate instead. The trend is similar, the Fe(III) compounds are in both cases favoured due to the localisation of the process. But the effects are quantitatively different. With NO −2 , only traces of GR(CO 2−3 ) are present. With PO 3−4 , GR(CO 2−3 ) is still present in non-negligible proportion. Moreover, when only a few pits were observed on the steel surface immersed in PO 3−4 containing electrolytes, a lot of pits and a greater amount of corrosion products were seen on surfaces immersed in NO −2 containing electrolytes. The corrosion is more localised when phosphates are present but, in contrast, the corrosion products have a larger average oxidation number when nitrites are present.

Finally, it must be noted that some α-iron foils were also immersed at open circuit potential for 1 week in the concrete-simulating solutions with NO −2 at R = 1. They did not show any corrosion, confirming the inhibiting effect of nitrite ions.

Discussion

Mechanisms of the corrosion of α-iron in concrete simulating solutions

Five hours of galvanostatically induced corrosion led to the formation of iron (II) and iron (II)–iron (III) compounds. In any case, the corrosion of iron in concrete-simulating solutions involved GR compounds. In the presence of oxygen, GRs are unstable and oxidised into the usual components of rust, that is mainly γ-FeOOH, α-FeOOH, Fe3O4 [1–4], or even β-FeOOH [18], depending on various parameters such as temperature, dissolved oxygen concentration, pH, etc. Besides GRs, various iron (II) compounds were observed, namely Fe(OH)2 and FeCO3. Fe(OH)2 would appear only at lower carbonate concentrations and/or higher pH values, as it is the case for solution S1. This compound is known to be a precursor of green rust iron (II)–iron (III) hydroxy-compounds, such as GR(Cl−) [1] or GR(CO 2−3 ) [2]. According to the Pourbaix diagram previously drawn for iron in chloride-containing solution with [Cl−] close to 1 mol l−1 [4], GR(Cl−) could form in a range of pH from 7.8 to 12.8. The identification of GR(Cl−) as a corrosion product of iron in solution S1 with a pH value of 12.4 is then in full agreement with the behaviour predicted by the diagram.

Finally, even though GR(CO 2−3 ) is metastable with respect to FeCO3 at most pH values and carbonate concentrations, the formation of GR(CO 2−3 ) from iron was reported to occur easily [3], and, actually, this compound was first observed as the corrosion product of cast iron water pipes [19]. In general, the formation of this GR should be favoured instead of FeCO3 when the pH increases or when the concentration of carbonate species decreases.

Influence of nitrite ions on the corrosion processes

The absence of any detectable corrosion process during the immersion of the α-iron foils in the various nitrite containing solutions at R = 1 testifies to the effectiveness of the inhibition. This inhibiting effect allows the passive film of iron oxide, more likely γ-Fe2O3 [20–22], to remain.

The results obtained by impedance spectroscopy show that at OCP, the Nyquist plots are formed of one capacitive loop. Its diameter increases with the inhibitor content and the corresponding capacitances decrease. This confirms that the nitrite ions do not form an inhibitor film at the metallic surface.

The galvanostatic experiments allowed us to induce pitting corrosion in the presence of NO −2 . Since traces of a GR compound were seen on some of the samples, it can be assumed that the corrosion process is similar to what happens without NO −2 , that is, schematically:

Fe \( \xrightarrow{{(1)}} \) Fe 2+aq \( \xrightarrow{{(2)}} \) FeCO3 and GR(CO 2−3 ) \( \xrightarrow{{(3)}} \) FeOOH, in carbonated media, and:

Fe \( \xrightarrow{{(1)}} \) Fe 2+aq \( \xrightarrow{{(2)}} \) Fe(OH)2 and GR(Cl−) \( \xrightarrow{{(3)}} \) FeOOH, in chlorinated media.

where (1) is dissolution of iron; (2) is precipitation of Fe(II) bearing phases and (3) oxidation of these phases into FeOOH.

However, the presence of NO −2 increases the kinetics of oxidation of the GRs and other Fe(II) compounds, leading to the rapid formation of FeOOH phases. This can be related to the reactivity of GRs towards NO −2 ions [23]. In particular, the oxidising effect of NO −2 is evidenced by comparing the results obtained here to those obtained previously with PO 3−4 as an inhibitor [5]. In Fig. 8 the E cor and E p values are reported versus the ratio R = [inhibitor]/[Cl−]. It can be seen that phosphate produces a considerable increase of the pitting potential, which testifies to the favourable influence of this inhibitor upon the stability of the passive film. In contrast, it has almost no influence on the corrosion potential E cor, as PO 3−4 is not an oxidising inhibitor. The slight increase of E cor may reflect an increase of the passive film thickness.

Fig. 8
figure 8

Evolutions of the corrosion potential E cor, the pitting potential E p and the potential measured during galvanostatic experiments E galv with the inhibitor concentration. Comparison between the values obtained with PO 3−4 as an inhibitor [5] and the values obtained here using NO −2 as an inhibitor

Full size image

The effects of NO −2 on those potential values are quite different. The increase of E p is significant but in no way comparable to that produced by phosphate. E cor is also increased and the difference E cor−E p is almost constant. This indicates clearly that nitrite acts as an anodic inhibitor, favouring the oxidation of Fe(II)-containing solids and thus preserving the Fe(III)-based passive film. Note finally that the potential corresponding to the galvanostatic experiments E galv is close to the E p value, except in one case (R = 0.4 with PO 3−4 ) where it is larger.

The oxidising action of nitrite ions is also revealed by the Mössbauer analyses. Without inhibitors, uniform corrosion occurred during galvanostatic polarisation and Fe(OH)2, FeCO3 and GRs were formed. The effect of phosphate was to localised the corrosion in a few pits, where a mixture of GRs and FeOOH phases (mainly ferrihydrite) were formed. The effect of nitrite was to accelerate the oxidation, leading to FeOOH phases with, in some cases, traces of GRs. Therefore, the effects of NO −2 can be summarised as follows:

  • When pitting corrosion occurs, NO −2 facilitate the transformation of GRs and other Fe(II) solids into FeOOH phases.

  • In the absence of stable pitting corrosion processes, NO −2 may promote repassivation by transforming Fe(II) containing species into the spinel oxides Fe3O4 and γ-Fe2O3 that constitute the passive film. This is similar to a previous model proposed by Berke et al. [23] where the repassivation is attributed to the action of NO −2 upon Fe2+ aq:

$$ 2{\text{ Fe }}^{2 + } + 2{\text{ OH}}^ - + 2{\text{ NO}}_2 ^ - {\text{ }} \to {\text{ }}2{\text{ NO}} + {\text{ Fe}}_2 {\text{O}}_3 + {\text{H}}_2 {\text{O }} $$

  • The action of NO −2 may be generalised to most of the reactive Fe(II) containing corrosion products.

Conclusions

Electrochemical tests and Mössbauer analyses of the corrosion products were coupled successfully to study the behaviour of α-iron in solutions simulating concrete, with and without NO −2 as an inhibitor. It was deduced that:

  • Nitrite ions are anodic inhibitors that increase the resistance to pitting attack and decrease the corrosion rate when chlorides are present. It can be guaranteed for a large nitrite concentration, matching that of chloride ions, whereas some pitting may occur, should this concentration be too small.

  • The early stages of the corrosion processes involve GR compounds and other Fe(II) containing phases, namely Fe(OH)2 and FeCO3. The oxidising action of nitrite favours and accelerates the oxidation of Fe2+ aq and Fe(II) containing phases, which facilitates repassivation and improves the corrosion resistance.