- Email: [email protected]
Electrochemical formation of green rusts in deaerated seawater-like solutions
Electrochimica Acta 56 (2011) 6481–6488
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrochemical formation of green rusts in deaerated seawater-like solutions Ph. Refait a,c,∗ , D.D. Nguyen a,d , M. Jeannin a,c , S. Sable b,c , M. Langumier a,b,c , R. Sabot a,c a
Laboratoire d’étude des matériaux en milieux agressifs (LEMMA), EA 3167, Université de La Rochelle, Bât. Marie Curie, Av. Michel Crépeau, F-17 042 La Rochelle Cedex 01, France Littoral, Environnement et Société (LiENSs), UMR 6250, CNRS-Univ. La Rochelle, Bât. Marie Curie, Av. Michel Crépeau, F-17 042 La Rochelle Cedex 01, France c Féd. de Recherche en Environnement et Développement Durable, FR CNRS 3097, France d Hue University’s College of Education, Hue, Viet Nam b
a r t i c l e
i n f o
Article history: Received 12 February 2011 Received in revised form 7 April 2011 Accepted 12 April 2011 Available online 7 May 2011 Keywords: Anoxic conditions Marine corrosion Steel Rust Sulphate-reducing bacteria
a b s t r a c t Carbon steel electrodes were polarised at a potential ∼150 mV higher than the open circuit potential, in a deaerated seawater-like electrolyte (0.5 mol dm−3 NaCl, 0.03 mol dm−3 Na2 SO4 , 0.003 mol dm−3 NaHCO3 ). X-ray diffraction and -Raman analysis demonstrated that a layer mainly composed of GR(SO4 2− ) had grown on the steel surface. GR(SO4 2− ) was accompanied by traces of GR(CO3 2− ). Similar experiments performed in a solution composed of 0.3 mol dm−3 of Na2 SO4 and 0.03 mol dm−3 of NaHCO3 led to the same result. The nature of the GR forming on steel is thus mainly linked to the sulphate to carbonate concentration ratio. Finally, carbon steel coupons immersed for 11 years in the harbour of La Rochelle (Atlantic coast) were removed from seawater for analysis. The inner part of the rust layer proved to be mainly composed of magnetite, GR(SO4 2− ) and iron sulphide FeS. This definitively confirms that GR(SO4 2− ), as Fe3 O4 and FeS, can form from steel in O2 -depleted environments. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Marine corrosion of carbon steel involves first the formation of the Fe(II–III) hydroxysulphate also known as sulphated green rust GR(SO4 2− ) [1]. Green rusts (GRs) are Fe(II–III) hydroxysalts characterised by a crystal structure that consists of the stacking of Fe(OH)2 -like layers carrying a positive charge due to the presence of Fe(III) and interlayers constituted of water molecules and anions that restore the electrical neutrality of the crystal [2,3]. Several GRs are known, and in particular those based on the main anions found in seawater, that is GR(Cl− ), GR(CO3 2− ) and GR(SO4 2− ). The chemical composition of GR(SO4 2− ) is for instance FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O, [3] sometimes developed as [FeII 4 FeIII 2 (OH)12 ]2+ [SO4 ·8H2 O]2− to remind that the crystal structure is built on positive hydroxide layers alternating with negative interlayers. In seawater, the [Cl− ]/[SO4 2− ] and [SO4 2− ]/[HCO3 − ] molar ratios are about 19 and 12, respectively (computed from [4]). Laboratory experiments were devoted to the oxidation by air of aqueous suspensions of Fe(OH)2 in the presence of sulphate and chloride ions [5,6]. It could be demonstrated that GR(SO4 2− ) formed instead of GR(Cl− ) even in solutions with large [Cl− ]/[SO4 2− ] molar
∗ Corresponding author at: Laboratoire d’étude des matériaux en milieux agressifs (LEMMA), EA 3167, Université de La Rochelle, Bât. Marie Curie, Av. Michel Crépeau, F-17 042 La Rochelle Cedex 01, France. Tel.: +33 5 46 45 82 27; fax: +33 5 46 45 72 11. E-mail addresses: [email protected], [email protected] (Ph. Refait). 0013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2011.04.123
ratios, as the layered structure of GRs presents a strong affinity for divalent anions. Green rusts are readily oxidised into FeOOH compounds by dissolved O2 . So, at the beginning of the process, the rust layer formed on steel in seawater should be composed of an outer layer of Fe(III) oxyhydroxides and an inner layer of GR(SO4 2− ). However, after 6 or 12 months of immersion, GR(SO4 2− ) was found systematically associated with iron sulphide FeS and sulphate reducing bacteria (SRB) [1]. This indicated that anoxic conditions were established at the steel/rust layer interface and inside the inner part of the rust layer. The presence of FeS on steel in seawater is a consequence of the metabolic activity of SRB, since in seawater sulphur is only present as sulphate in the absence of bacterial activity. The anoxic conditions are due to the fact that dissolved oxygen is consumed in the outer part of the rust layers by (i) aerobic micro-organisms and (ii) its reaction with GR(SO4 2− ) that mainly leads to Fe(III) oxyhydroxides. So, after some time, the kinetics of the corrosion is no more controlled by oxygen transport [7–10]. Since the process is then clearly linked to the activity of micro-organisms, it was for instance proposed that the availability and transport of the nutrients necessary for the micro-organisms could be the limiting step [10]. It is now demonstrated that SRB can reduce the sulphate ions coming from GR(SO4 2− ) [11,12]. This phenomenon induces the transformation of GR(SO4 2− ) into various compounds, and in particular iron sulphides such as mackinawite [12]. Therefore it can be expected that GR(SO4 2− ) is totally consumed and that after some time, it is totally absent of the rust layers. However, previous
6482
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
works performed in carbonated media indicated that the carbonated green rust, GR(CO3 2− ) could be obtained in anoxic or almost anoxic conditions [13–15]. So if GR(SO4 2− ) could still form on steel in anoxic conditions it would remain along with FeS one of the components of the rust layer. The aim of this work was then to study the formation of rust on steel in deaerated seawater. In order to focus on the role of anionic species, an electrolyte composed of 0.5 mol dm−3 NaCl, 0.03 mol dm−3 Na2 SO4 and 0.003 mol dm−3 NaHCO3 was considered as a “seawater-like” solution. The concentration ratios [Cl− ]/[SO4 2− ] (=16.6) and [SO4 2− ]/[HCO3 − ] (=10) were close to those characteristic of seawater. Electrochemical experiments (voltammetry, chronoamperometry) were performed in anoxic conditions and the components of the electro-generated rust layers were studied by -Raman spectroscopy, X-ray diffraction (XRD) and scanning electron microscopy (SEM) coupled with energy dispersive spectrometry (EDS). Finally, rust layers grown during 11 years on steel coupons in seawater were analysed by Raman spectroscopy. The behaviour of the real corrosion system could then be interpreted at the light of the information given by laboratory experiments.
2. Experimental 2.1. Electrochemistry Electrochemical experiments were carried out in a classical three-electrode glass cell. The working electrodes were carbon steel disks with 2 cm2 area. The steel approximate composition (in wt%) was: 98.2% Fe, 0.122% C, 0.206% Si, 0.641% Mn, 0.016% P, 0.131% S, 0.118% Cr, 0.02% Mo, 0.105% Ni and 0.451% Cu. The surfaces were polished with silicon carbide (particle size 25 m), rinsed thoroughly with Milli-Q water and rapidly dried in an air flow. The working electrodes were positioned horizontally with the active surface set upwards. They were set in the electrochemical cell just before the beginning of the experiment. Potentials were measured vs. the saturated calomel electrode (SCE). A large platinum grid was used as counter electrode. Voltammetric and chronoamperometric (anodic polarisation) experiments were performed with a PGP 201 potentiostat system (Radiometer analytical). For linear voltammetry, the scans were started at −1.20 V/SCE and ended at +0.5 V/SCE, with a scan rate dE/dt = 1 mV s−1 . Anodic polarisation experiments were performed at −0.60 V/SCE, i.e. approximately 0.15 V above the open circuit potential (OCP). OCP was measured during 30 min at the beginning of each experiment prior to voltammetry or chronoamperometry. It varied around −0.75 V/SCE. Since the study related mainly to the role of anionic species, a simplified electrolyte prepared with NaCl, Na2 SO4 , and NaHCO3 was used instead of artificial seawater. The sodium salts with 98% min. purity (Aldrich) were dissolved in Milli-Q water (resistivity 18.2 M cm). The concentrations, set as 0.5, 0.03 and 0.003 mol dm−3 for NaCl, Na2 SO4 , and NaHCO3 respectively, were close to those typical of seawater. The pH of this electrolyte was measured at the beginning of the experiments at an average value of 7.95. Note that the pH of ASTM D1141 artificial seawater is 8.2. Two procedures were considered for deaeration. First, experiments were achieved in the electrochemical cell under an argon flow. The solutions were deaerated by the argon flow during 45 min before the beginning of each experiment and the argon flow was maintained till the end. Secondly, experiments were performed in an electrochemical cell set inside a glove box (Jacomex P[box]-T4) filled with Argon. In this case, the atmosphere inside the box contained a residual oxygen content ≤1 ppm.
The electrolyte was not stirred during the experiments. However, the argon flow induced by itself a stirring of the electrolyte through its permanent bubbling. In contrast, during the experiments performed in the glove box, the electrolyte was really stagnant. Additional experiments were performed in a second electrolyte, composed of Na2 SO4 (0.3 mol dm−3 ) and NaHCO3 (0.03 mol dm−3 ). It kept the sulphate to carbonate concentration ratio of 10 of the seawater-like electrolyte but the concentrations were ten times larger. The experimental procedure was strictly identical to that used in the seawater-like electrolyte. 2.2. Steel coupons immersed in seawater 1999, a series of carbon steel coupons In (70 mm × 70 mm × 6 mm) were immersed in one the harbours of La Rochelle (Atlantic coast, France) in the frame of a PhD work. Most of the coupons were used at that time but it was decided to leave a few of them in place for future works. One of these coupons was removed for analysis in 2010. It was made of the carbon steel used for laboratory experiments (see Section 2.1) and was permanently immersed at a constant depth of 1 m during those 11 years. For -Raman spectroscopy analysis, the coupons were frozen at −80 ◦ C. This freezing procedure was necessary to preserve the reactive Fe(II) compounds from the oxidizing action of O2 . Moreover, the rust layers are easier to handle in this hardened frozen state than in their wet and slurry room temperature state. Slices of the corrosion products layer could then be easily extracted and their stratification could be investigated. 2.3. Characterisation of the rust layers Micro-Raman analyses were performed on a Jobin Yvon High Resolution Raman spectrometer (LabRAM HR) equipped with a microscope (Olympus BX 41) and a Peltier-based cooled charge coupled device (CCD) detector. The zones analysed through a 50× objective had a diameter of ∼6 m. Spectra were recorded with the acquisition LabSpec software at room temperature with a resolution of ∼0.1 cm−1 . Excitation was provided by a He–Ne laser (632.8 nm). Its power was 0.9 mW, that is 10% of the maximal power, in order to prevent an excessive heating that could have induced the transformation of the analysed sample into hematite ␣Fe2 O3 . The acquisition time was variable, around an average value of 1 min, and was chosen for each analysis so as to optimise the signal to noise ratio. XRD experiments were carried out with a classical powder diffractometer (Brucker AXS® D8-Advance), using Cu-K␣ wavelength ( = 0.15406 nm) in Bragg–Brentano geometry. The electrodes were removed from the electrochemical cell just before analysis. Since the expected Fe(II) compounds were sensitive to the oxidising action of air, the electrodes were immediately placed on the sample holder and the rust layer was coated with glycerol before the analysis. This procedure limits the oxidation of the Fe(II) compounds during the acquisition of the pattern. Finally, the morphology of the electrochemically generated rust layers was observed using scanning electron microscopy (SEM) with a PHILIPS FEI Quanta 200F apparatus. The composition of the main component of the rust layers was estimated by energy dispersive spectroscopy (EDS) coupled to SEM (acceleration voltage: 15 kV). The samples to be analysed were carried from the laboratory to the SEM apparatus inside a bag filled with argon. They were exposed to atmosphere when they were removed from the cell before to be set in the bag and when they were removed from the bag before to be set in the SEM apparatus.
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
Fig. 1. Electrochemical experiments performed with a carbon steel electrode in deaerated seawater-like electrolyte. (a) Polarisation curve obtained under argon flow. The dash lines are the Tafel straight lines used to determine jcor . (b) Chronoamperometric curves obtained during anodic polarisation at E = −0.6 V/SCE: curves (1) and (2) were obtained under argon flow and curve (3) was obtained with a stagnant electrolyte in a glove box under argon atmosphere. The black circles locate the various analyses presented in Figs. 2 and 3.
3. Results 3.1. Anodic polarisation of steel in deaerated seawater-like solutions Typical polarisation and chronoamperometric curves obtained with steel electrodes immersed in the deaerated seawater-like electrolyte are displayed in Fig. 1. The log|j| vs. potential curve of Fig. 1a clearly shows that both anodic and cathodic parts of the curve obey to Tafel law, with anodic Tafel slope ba ∼ 100 mV and cathodic Tafel slope bc ∼ 200 mV. So the corrosion current density jcor could be readily determined, at the intersection of the Tafel straight lines. It was found that jcor was equal to 2.5 × 10−5 A cm−2 , which corresponds to a corrosion rate Vcor of 0.29 mm per year. The corresponding corrosion potential was measured at Ecor = −0.78 V/SCE. Note that experiments performed in aerated conditions led to an initial corrosion rate of about 5.7 mm per year [16]. Anodic polarisation experiments were performed at an applied potential of −0.6 V/SCE, which would give rise to an anodic current density of the order of 1 mA cm−2 according to the polarisation curve of Fig. 1a. Three chronoamperometric curves are displayed in Fig. 1b. The initial current density value (or the maximum reached
6483
a few minutes after the beginning of the experiment) is in each case consistent with what was expected, and measured between 0.7 and 0.8 mA cm−2 . Curves 1 and 2 were obtained with the electrolyte deaerated by an argon flow and curve 3 was obtained with the electrochemical cell placed inside the glove box. The difference between curves 1 and 2 illustrates the variations observed with this methodology, mainly induced by the uncontrolled stirring due to the argon bubbling. However, the main trend is illustrated by curve 2. In this case, the current density decreased from 0.7 to ∼0.15 mA cm−2 in 4 h, remained constant during 10 h before to decrease again down to 0.02 mA cm−2 . The current density reached at the end of the experiment was in this case 35 times smaller than the initial current density. This indicates that the layer of corrosion products forming on the steel surface tended to block the surface. Finally curve 3, obtained in the glove box with a stagnant electrolyte, is rather similar to curve 2. The main difference is that the current density remained constant at ∼0.15 mA cm−2 after its initial decrease. However, the composition and morphology of the rust layers proved to be independent (see thereafter) of the methodology used for deaeration. Actually the differences observed between the chronoamperometric curves of Fig. 1b are not yet clearly interpreted. Typical Raman spectra and XRD pattern of the products generated by anodic polarisation in deaerated seawater-like electrolytes are displayed in Fig. 2. The Raman spectra of Fig. 2a and b were obtained after 15 min and 3 h of polarisation in the glove box. They are characteristic of the various spectra obtained via the analysis of different spots of the surface. Note that similar spectra (not shown) could be obtained on electrodes polarised 26 h, the longest polarisation time considered in this study. This means that the same compound is forming all along the polarisation process. Its Raman spectrum is characteristic of a green rust compound, with two main bands at 430–435 and 509 cm−1 [17–21]. Similar bands are present in the spectra of all green rusts. However, other bands are usually seen that depend on the nature of the green rust. For instance, GR(Cl− ) gives rise to three additional bands at 238, 325 and 363 cm−1 [21], and consequently the GR formed here, characterised by a sharp Raman peak at 219 cm−1 (Fig. 2b) is not GR(Cl− ). Actually, both GR(CO3 2− ) and GR(SO4 2− ) give rise to an additional vibration band at about 220 cm−1 [18–20]. So, Raman spectroscopy cannot discriminate between those two GR compounds. However, the presence of SO4 2− ions was evidenced in some cases (Fig. 2a) by a peak at 995 cm−1 . These anions are more likely adsorbed on the GR crystals. Actually, the anions intercalated in the GR structure are very difficult to observe [22], which is more likely due to a resonance phenomenon between the excitation laser and the vibration modes of the GR hydroxide layers. XRD analysis was then required and the pattern obtained with an electrode polarised for 16 h in the electrolyte under argon flow is presented in Fig. 2c as an example. Two very intense sharp lines are seen. The one located at 2 ∼ 45◦ is due to the main line of the predominant phase of the substrate, i.e. ␣-Fe. Other small lines c of the substrate can be seen. They correspond to Fe3 C. The other intense line, at 2 ∼ 8◦ , is the main line of GR(SO4 2− ) [1,20]. It is accompanied by the other characteristic diffraction lines of this green rust, denoted in Fig. 2c with the corresponding Miller index. Except for the lines of GR(SO4 2− ) and those of the substrate, only one small line due to another compound is seen at 2 ∼ 11.7◦ . It corresponds to the main diffraction line of the carbonated form of GRs, GR(CO3 2− ) [19], which indicates that a small amount of GR(CO3 2− ) was formed together with GR(SO4 2− ). XRD analysis is then consistent with Raman analysis since these GRs give rise to similar Raman spectra. Note that the Raman spectrum of Fig. 2a, obtained after only 15 min of polarisation, is slightly different from the spectra usually obtained for GR(SO4 2− ) [1,12,20], and for instance that of Fig. 2b.
6484
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
Fig. 2. XRD and -Raman spectroscopy analysis of the rust layers formed by anodic polarisation of a steel electrode in deaerated seawater-like solutions. -Raman spectra of layers obtained after (a) 15 min and (b) 3 h of polarisation (glove box). GR, green rust (c) XRD pattern of a layer obtained after 16 h of polarisation (argon flow). The diffraction lines of GR(SO4 2− ) are denoted by the corresponding Miller index. GRC = carbonate green rust and c = Fe3 C. E = −0.6 V/SCE in each case.
The third peak at 219 cm−1 seems to be split here in two peaks, at 195 and 227 cm−1 , another peak appears at 150 cm−1 , and a Raman band due to sulphate ions is clearly visible at 995 cm−1 . This discrepancy may be due to some peculiarity of the GR crystals forming first, e.g. poor crystallinity or non-stoichiometry. For instance, the Fe(II)/Fe(III) ratio in the GR could be larger at the first stage of the GR formation. Anyway, this spectrum demonstrates that 15 min of anodic polarisation were sufficient to form a detectable amount of GR. It was not possible to identify any solid phase precursor of the GR, for instance Fe(OH)2 that could be expected. In conclusion, the main component formed on steel under anodic polarisation in deaerated conditions is GR(SO4 2− ). This compound was also observed to form on steel coupons immersed in seawater [1] but in this case it was oxidised into lepidocrocite ␥-FeOOH. So the effect of the depletion in O2 was to suppress the oxidation of GR(SO4 2− ) into Fe(III) compounds, and the first stage of the corrosion process, that is the formation of GR(SO4 2− ), still took place. This result is consistent with previous works that reported the formation of GR(CO3 2− ) in deaerated carbonated solutions [13–15]. The morphology of the GR(SO4 2− ) layers obtained after 24 h of anodic polarisation was observed by SEM. Some photographs are displayed in Fig. 3. The first one (Fig. 3a) shows an assembly of platelets mainly perpendicular to the steel surface. Their width is
about 5–10 m and their thickness is less than 1 m. Fig. 3b is a zoom on the lower right corner of the previous photograph. One of the platelets is strongly tilted and its hexagonal shape is revealed. This morphology is typical of GR(SO4 2− ) crystals [23]. EDS analysis performed on such individual crystals gave a Fe/S atomic ratio close to 7, that is slightly larger than the value of 6 expected for GR(SO4 2− ) according to its chemical formula FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O [3]. However, it confirms that these crystals are those of GR(SO4 2− ). The discrepancy between measured and theoretical Fe/S ratios may be due to a slight oxidation of GR(SO4 2− ) during the transport of the sample from the electrochemical cell to the SEM apparatus (see Section 2). Such a slight oxidation would have formed a very thin FeOOH layer on the GR crystals. Fig. 3c shows more clearly how the platelets are bound one to each other, leading to a “house of cards” structure. The layered structure of GRs is made of positively charge hydroxide sheets alternating with negatively charged interlayers of anions and water molecules. The faces of the platelets are thus electrically charged. If the sides of the platelets are charged oppositely, then the platelets tend to assemble with such morphology. This implies that the platelets visible in Fig. 3, mainly orientated perpendicularly to the steel surface, are bound to particles that are parallel to the steel surface and so on, exactly as in a house of cards. So the GR(SO4 2− ) layer is not compact at all and the empty spaces left in its “house of
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
6485
Fig. 3. SEM photographs of the rust layer formed after 24 h of anodic polarisation of a steel electrode in anoxic seawater-like solution. The experiment was performed in an electrolyte deaerated by an argon flow. E = −0.6 V/SCE.
cards” structure could be favourable for the colonisation of the rust layers by SRB and the subsequent growth of the colony, as observed in marine corrosion processes [1]. It must be noted that chloride-containing iron compounds were not detected. In particular, the chloride variety of GRs did not form. This is consistent with analysis of rust layers formed on steel in natural seawater after 6–12 months of immersion [1] and with previous laboratory experiments performed in aerated solutions [5]. In contrast, archaeological iron objects left 2000 years in the bottom of the Mediterranean sea proved to be covered by a thick layer mainly composed of a Fe(II)–hydroxychloride, the  phase of Fe2 (OH)3 Cl [24] and green rust compounds were not detected. So the mechanisms of the formation of rust in seawater may be drastically different after some (long) time. 3.2. Role of the sulphate to carbonate concentration ratio The XRD pattern of Fig. 2b revealed that, even if GR(SO4 2− ) was the main component of the electrochemically formed layer, a small proportion of GR(CO3 2− ) was also obtained. This was already observed with steel coupons immersed in seawater [1]. Actually, in natural seawater as in the seawater like-electrolyte considered in this study, three anionic species are likely to give rise to a GR compound: Cl− , SO4 2− and HCO3 − /CO3 2− . It is generally admitted that the GR structure has a variable affinity for anions [5,25–27,6]. In particular, divalent anions are preferred to monovalent anions. This is the reason why, in seawater, GR(SO4 2− ) forms instead of GR(Cl− ) even if the chloride to sulphate concentration ratio is as high as 19. And, among divalent anions, carbonate species are preferred to any other anion, and in particular to sulphate. So, in seawater, GR(SO4 2− ) forms preferentially to GR(CO3 2− ) because (i) the carbonate concentration is low or (ii) the sulphate to carbonate concentration ratio is high. To elucidate this point, additional experiments were necessary. They were performed in an electrolyte composed of 0.3 mol dm−3 of Na2 SO4 and 0.03 mol dm−3 of NaHCO3 . Compared to the seawaterlike electrolyte considered in Section 3.1, this solution (i) does not contain chloride, (ii) keeps the same sulphate to carbonate concentration ratio of 10, but (iii) has sulphate and carbonate concentrations 10 times larger. The measured OCP values for steel in this electrolyte were similar to those found with the seawater-like electrolyte. The main results obtained are summarised in Fig. 4. In Fig. 4a, the chronoamperometric curve 1 obtained with an applied potential of −0.6 V/SCE is compared to curve 2 obtained in seawater-like electrolyte. Both experiments were performed in similar conditions, i.e. the electrolyte was deaerated by an argon flow. The general trend is the same. The current density decreases
Fig. 4. Anodic polarisation of a steel electrode in a deaerated electrolyte composed of 0.3 mol dm−3 Na2 SO4 and 0.03 mol dm−3 NaHCO3 . (a) Chronoamperometric curve (1) compared to that obtained in seawater-like electrolyte (2). (b) XRD analysis of the resulting rust layer. The experiments were performed in electrolytes deaerated by an argon flow. E = −0.6 V/SCE.
6486
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
Fig. 5. -Raman spectroscopy analysis of the rust layer formed on a steel coupon immersed 11 years in seawater. Schematic representation of the rust layer and -Raman spectra of the main components found in the three strata (a), (b) and (c) of this layer. G, goethite; GR, green rust; Mn , nanocrystalline mackinawite; and Mgt, magnetite.
first down to a value close to 0.1 mA cm−2 , stabilises at this value for 4 h and decreases again. The initial current density is however smaller and the decrease of the current faster than in the seawaterlike electrolyte. Fig. 4b displays the XRD pattern results from the analysis of the electrode surface after 22 h of polarisation. The three main diffraction lines of GR(SO4 2− ) are seen, together with the main line of the substrate. They are accompanied by the main diffraction line of GR(CO3 2− ), denoted as GRC. This line is sharper and more intense than in the XRD pattern obtained in seawater-like electrolyte (Fig. 2b). Actually, the intensity of the main line of GRC is 1/5 of the intensity of the main line (0 0 1) of GR(SO4 2− ) in this pattern and 1/13 in the previous one. So, it is clearly the high sulphate to carbonate concentration ratio that favours GR(SO4 2− ) at the detriment of GR(CO3 2− ), but the low carbonate concentration of seawater seems to increase this effect. 3.3. Analysis of a steel coupon immersed 11 years in seawater The result of the -Raman analysis is summarised in Fig. 5. Only 3 of the 50 acquired Raman spectra are presented so as to illustrate the main trends. Globally, the steel surface was covered by a thick layer composed of four main strata. The outer part of this layer mainly contained macro-organisms (shells for instance), micro-organisms, alga, sediments, etc. It was removed from the steel surface before Raman analysis. The three other strata were mainly composed of corrosion products. Note that such rust layers are also colonised by bacteria [1]. The outer stratum (a) exhibited an orange–brown colour. In agreement, Raman analysis revealed the presence of Fe(III) oxyhydroxides, such as goethite ␣-FeOOH (spectrum 3a), and magnetite (Fe3 O4 ). The thicker intermediate stratum (b) was in fact a complex mixture of three main phases, magnetite, mackinawite (FeS) and GR(SO4 2− ), as revealed by spectrum 3b. Mackinawite was in most cases present in its nanocrystalline state, characterised by a Raman spectrum made of
two sharp lines at 208 and 283 cm−1 [28]. The GR spectrum, characterised by three bands at 219, 435 and 508 cm−1 was of course attributed to GR(SO4 2− ). Magnetite was mainly characterised by a large band around 670 cm−1 [29]. Actually, this thick intermediate stratum of corrosion products appeared to be very heterogeneous. In some area, only FeS was detected and in other regions GR(SO4 2− ) was predominating. In some cases, and even close to the metal, lepidocrocite ␥-FeOOH could be detected. This indicates that dissolved O2 could penetrate locally inside the rust layer, more likely through cracks. Finally, the Raman spectra obtained in a thin inner stratum (thickness smaller than 100 m) revealed mainly the presence of GR(SO4 2− ) and magnetite. In particular, FeS could not be detected so close to the steel surface. This indicates that SRB were located farther from the steel surface. This may be due to the fact that the metabolic activity and the growth of the micro-organisms have to be sustained by nutrients that can only come from seawater. So when the rust layers become very thick, SRB could be mainly located in anoxic regions far from the steel surface and close to their source of nutrients. The precipitation of iron sulphides would then also occur far from the steel surface. Note that chloride-containing iron compounds were not detected. This confirms that the presence of -Fe2 (OH)3 Cl in the rust layers of archaeological iron objects [24] is typical of corrosion processes occurring after a very long immersion time. This implies the existence of a transition period (between 11 and 2000 years) when the kinetics of the dissolution of steel and the mechanism of the formation of rust change progressively.
4. Discussion 4.1. Thermodynamics The results presented here confirm that GR(SO4 2− ) can be produced when anoxic or almost anoxic conditions are met at the steel
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
surface. Like Fe3 O4 , GR(SO4 2− ) is a Fe(II–III) mixed valence compound, and can be obtained via the oxidation by O2 of a precursor Fe(II) compound such as Fe(OH)2 [5]. However, such a process cannot take place in anoxic conditions. Actually, it was not possible to detect such a Fe(II) precursor and only GR(SO4 2− ) was observed, even after only 15 min of anodic polarisation. When an anodic polarisation is applied, both Fe(II) and Fe(III) may be electro-generated, leading to the direct formation of the GR from the metal, according to the following anodic reaction: 6Fe + SO4 2− + 20H2 O → FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O + 12H+ + 14e−
(1)
At OCP, during corrosion of steel in anoxic conditions, this reaction is necessarily accompanied by the reduction of water (H2 O or H+ ): 2H+ + 2e− → H2
(2)
The equilibrium conditions between Fe and GR(SO4 2− ) are given by [30]: Eeq (GRSO4 /Fe) = −0.15 − 0.0507 pH − 0.0042 log(aSO4 )
(3)
2−
The activity of SO4 ions, aSO4 , can be estimated using the activity coefficient of sulphate ions in seawater [31], that is SO4 = 0.12.
Then: aSO4 = SO4 × [SO4 2− ] = 0.12 × 0.03 = 0.0036 since the sulphate concentration of the electrolyte is 0.03 mol dm−3 . The pH of the electrolyte being equal to 7.95, the equilibrium potential is computed at: Eeq (GRSO4 /Fe) = −0.54 V/SHE This value was computed for the seawater-like electrolyte considered in this study but it may not differ significantly from the value corresponding to real seawater. It must be compared to the equilibrium potential of H2 O (or H+ ) and H2 , given by: E eq (H+ /H2 ) = 0.00–0.0591 pH
(4)
At pH 7.95, this equation gives Eeq (H+ /H2 ) = −0.47 V/SHE. So Eeq (GRso4 /Fe) < Eeq (H+ /H2 ), which demonstrates that the formation of GR(SO4 2− ) from steel in seawater, in anoxic conditions with water being the oxidizer, is indeed thermodynamically possible. The OCP values were measured accordingly at about −0.75 V/SCE, that is −0.51 V/SHE, and so: Eeq (GRso4 /Fe) < OCP < Eeq (H+ /H2 ). Note that magnetite Fe3 O4 can also form from steel in anoxic conditions. The equilibrium conditions between Fe and Fe3 O4 are given by [30,32]: E eq (Fe3 O4 /Fe) = −0.086–0.0591 pH
(5)
This gives Eeq (Fe3 O4 /Fe) = −0.55 V/SHE at pH = 7.95. Since this value is smaller than that of Eeq (GRSO4 /Fe), GR(SO4 2− ) is metastable with respect to magnetite in seawater, but the difference between the two equilibrium potential values, about 10 mV, is very small. 4.2. Mechanisms When anoxic conditions are established at the steel surface in natural seawater, the formation of GR(SO4 2− ) competes not only with that of Fe3 O4 but also with that of FeS. The formation of FeS is clearly linked to the metabolic activity of SRB. SRB reduce sulphate to sulphide, i.e. transform the species necessary for the formation of GR(SO4 2− ) into species required for the formation of FeS. In contrast, the formation of magnetite may or may not be directly linked to micro-organisms. First, like FeS, magnetite could be generated by microbial activity. It is now demonstrated that SRB can use the sulphate ions coming from GR(SO4 2− ) and consequently
6487
induce the transformation of the GR [11,12]. More precisely, the bacterial growth consumes dissolved sulphate species, which modifies the equilibrium conditions and induces progressively the dissolution of the GR: FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O → 4Fe2+ + 2Fe3+ + SO4 2− + 8H2 O + 12OH−
(6)
The reduction of 1 SO4 2− by SRB produces 1 S2− that precipitates with 1 Fe2+ to form FeS. So the process would leave in solution an excess of 3 Fe2+ and 2 Fe3+ , from which magnetite, a mixed valence Fe(II–III) compound, could form. This assumption would explain the association between the three phases, GR(SO4 2− ), FeS, and Fe3 O4 , observed in the main part of the rust layer. Of course, magnetite could also result like GR(SO4 2− ) from the dissolution of steel. The main parameters that govern the formation of GR(SO4 2− ) vs. Fe3 O4 may be the interfacial Fe2+ aq and SO4 2− concentrations and pH. Of course, the lower is the sulphate concentration the higher is the tendency to form magnetite at the detriment of GR(SO4 2− ). Similarly, it seems that high pH values and low dissolved Fe(II) species concentrations favour magnetite [30]. So, local variations of pH, sulphate, sulphide and dissolved Fe(II) species concentrations can explain the variability of corrosion products observed inside the rust layers of the steel coupon immersed for 11 years in seawater. The main source of heterogeneity inside the composite “rust/biofilm” layer has more likely a biological origin, even though biological heterogeneities can induce in turn electrochemical heterogeneities, i.e. create anodic and cathodic zones on the steel surface. It must also be reminded that transformations of the components of rust could also be induced by micro-organisms such as iron reducing bacteria (IRB) that can reduce Fe(III) compounds to Fe(II) or Fe(II–III) compounds. Actually, it was demonstrated that the simultaneous presence of SRB and IRB favoured the formation of GR(SO4 2− ) at the detriment of that of FeS [33]. Similarly, some micro-organisms are able to achieve direct electron transfer with the steel surface [34,35] or, via enzyme hydrogenase, catalyse proton reduction [36–38]. These micro-organisms can influence the rate of cathodic and/or anodic reactions which could in turn modify the nature of the corrosion products. It is however noticeable that GR(SO4 2− ) was also generated on carbon steel in aerated artificial seawater by cathodic polarisation at E = −1.0 V/SCE [39]. Taken into account the results obtained here under anodic polarisation, it appears that GR(SO4 2− ) can form on steel in seawater and similar electrolytes in a rather wide range of potential around OCP, that is whatever the kinetics of the elementary cathodic and anodic electrochemical reactions. 5. Conclusions It is now demonstrated that layers of GR(SO4 2− ) can be obtained by anodic polarisation of steel in deaerated seawater-like electrolytes. The oxidation of GR(SO4 2− ) into Fe(III) compounds does not take place due to the depletion in O2 which favours the accumulation of the GR onto the steel surface. This shows that GR(SO4 2− ) can still be produced during marine corrosion of steel once anoxic or almost anoxic conditions are established at the steel surface. Raman analysis of the rust layer that covered a steel coupon left 11 years in seawater confirmed it definitively. However, since these anoxic conditions lead to the growth of SRB in natural seawater, the reduction of sulphate to sulphide by these micro-organisms induces the formation of FeS at the detriment of GR(SO4 2− ). So, the formation of GR(SO4 2− ) can only occur locally where the metabolic activity of SRB is null or very low. It was also observed that GR(SO4 2− ) was accompanied by a minor proportion of GR(CO3 2− ). Anodic polarisation in an
6488
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
electrolyte keeping the sulphate to carbonate concentration ratio of seawater, but with a carbonate concentration 10 times larger, also led predominantly to GR(SO4 2− ). The sulphate to carbonate concentration ratio is then the major parameter to govern the formation of GR(SO4 2− ) and GR(CO3 2− ). In seawater this concentration ratio is high (∼10) which explains why GR(SO4 2− ) predominates even though carbonate ions are known to ensure a greater stability to the layered crystal structure of GRs. Acknowledgements This work, part of the PhD thesis of M. Langumier, was supported financially by the Conseil Général of Charente Maritime (17), and by the Conseil Régional of Poitou-Charentes (Master of Science of D.D. Nguyen). References [1] S. Pineau, R. Sabot, L. Quillet, M. Jeannin, Ch. Caplat, I. Dupont-Morral, Ph. Refait, Corros. Sci. 50 (2008) 1099. [2] Ph. Refait, M. Abdelmoula, J.-M.R. Génin, Corros. Sci. 40 (1998) 1547. [3] L. Simon, M. Franc¸ois, P. Refait, G. Renaudin, M. Lelaurain, J.M.R. Génin, Solid State Sci. 5 (2003) 327. [4] J.P. Riley, G. Skirrow, Eds., Chemical Oceanography, vol. 2, second ed., Academic Press, 1975. [5] Ph. Refait, J.B. Memet, C. Bon, R. Sabot, J.-M.R. Génin, Corros. Sci. 45 (2003) 833. [6] Ph. Refait, M. Abdelmoula, J.-M. Génin, R. Sabot, C.R. Geosci. 338 (2006) 476. [7] R.E. Melchers, Corrosion (NACE) 59 (2003) 319. [8] R.E. Melchers, Corros. Sci. 45 (2003) 923. [9] R.E. Melchers, R. Jeffrey, Corros. Sci. 47 (2005) 1678. [10] R.E. Melchers, T. Wells, Corros. Sci. 48 (2006) 1791. [11] A. Zegeye, L. Huguet, M. Abdelmoula, C. Carteret, M. Mullet, F. Jorand, Geochim. Cosmochim. Acta 71 (2007) 5450. [12] M. Langumier, R. Sabot, R. Obame-Ndong, M. Jeannin, S. Sablé, Ph. Refait, Corros. Sci. 51 (2009) 2694. [13] L. Legrand, S. Savoye, A. Chaussé, R. Messina, Electrochim. Acta 46 (2000) 111.
[14] L. Legrand, M. Abdelmoula, A. Géhin, A. Chaussé, J.-M.R. Génin, Electrochim. Acta 46 (2001) 1815. [15] S. Savoye, L. Legrand, G. Sagon, S. Lecomte, A. Chaussé, R. Messina, P. Toulhoat, Corros. Sci. 43 (2001) 2049. [16] J.B. Memet, PhD Thesis, University of La Rochelle, France, 2000. [17] N. Boucherit, A. Hugot-Le Goff, S. Joiret, Corros. Sci. 32 (1991) 497. [18] L. Legrand, G. Sagon, S. Lecomte, A. Chaussé, R. Messina, Corros. Sci. 43 (2001) 1739. [19] M. Reffass, R. Sabot, C. Savall, M. Jeannin, J. Creus, Ph. Refait, Corros. Sci. 48 (2006) 709. [20] R. Sabot, M. Jeannin, M. Gadouleau, Q. Guo, E. Sicre, Ph. Refait, Corros. Sci. 49 (2007) 1610. [21] Ph. Refait, P. Girault, M. Jeannin, J. Rose, Colloids Surf. A 332 (2009) 26. [22] Ph. Refait, M. Reffass, J. Landoulsi, R. Sabot, M. Jeannin, Colloids Surf. A 299 (2007) 29. [23] A. Géhin, C. Ruby, M. Abdelmoula, O. Benali, J. Ghanbaja, Ph. Refait, J.-M.R. Génin, Solid State Sci. 4 (2002) 61. [24] C. Rémazeilles, D. Neff, F. Kergourlay, E. Foy, E. Conforto, E. Guilminot,.S. Reguer, Ph. Refait, Ph. Dillmann, Corros. Sci. 51 (2009) 2932. [25] S. Miyata, Clays Clay Mine. 31 (1983) 305. [26] A. Mendiboure, R. Schöllhorn, Rev. Chim. Miner. 23 (1986) 819. [27] Ph. Refait, S.H. Drissi, J. Pytkiewicz, J.-M.R. Génin, Corros. Sci. 39 (1997) 1699. [28] J.A. Bourdoiseau, M. Jeannin, R. Sabot, C. Rémazeilles, Ph. Refait, Corros. Sci. 50 (2008) 3247. [29] D.L.A. De Faria, S.V. Silva, M.T.D. Oliveira, J. Raman Spectrosc. 28 (1997) 873. [30] Ph. Refait, A. Géhin, M. Abdelmoula, J.-M.R. Génin, Corros. Sci. 45 (2003) 656. [31] R.M. Garrels, C.L. Christ, Equilibres Des Minéraux Et De Leurs Solutions Aqueuses, Gauthiers-Villars, Paris, 1967. [32] G.H. Kelsall, R.A. Williams, J. Electrochem. Soc. 138 (1991) 931. [33] J. Duan, S. Wu, X. Zhang, G. Huang, M. Du, B. Hou, Electrochim. Acta 54 (2008) 22. [34] M. Mehanna, R. Basseguy, M.-L. Delia, A. Bergel, Electrochem. Commun. 11 (2009) 568. [35] M. Mehanna, R. Basseguy, M.-L. Delia, A. Bergel, Corros. Sci. 51 (2009) 2596. [36] R.D. Bryant, W.J. Jansen, J. Boivin, E.J. Laishley, W. Costerton, Appl. Environ. Microbiol. 57 (1991) 2804. [37] S. Da Silva, R. Basseguy, A. Bergel, Bioelectrochemistry 56 (2002) 77. [38] M. Mehanna, R. Basseguy, M.-L. Delia, L. Girbal, M. Demuez, A. Bergel, Electrochim. Acta 54 (2008) 140. [39] Ch. Barchiche, C. Deslouis, O. Gil, S. Joiret, Ph. Refait, B. Tribollet, Electrochim. Acta 54 (2009) 3580.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Electrochemical formation of green rusts in deaerated seawater-like solutions Ph. Refait a,c,∗ , D.D. Nguyen a,d , M. Jeannin a,c , S. Sable b,c , M. Langumier a,b,c , R. Sabot a,c a
Laboratoire d’étude des matériaux en milieux agressifs (LEMMA), EA 3167, Université de La Rochelle, Bât. Marie Curie, Av. Michel Crépeau, F-17 042 La Rochelle Cedex 01, France Littoral, Environnement et Société (LiENSs), UMR 6250, CNRS-Univ. La Rochelle, Bât. Marie Curie, Av. Michel Crépeau, F-17 042 La Rochelle Cedex 01, France c Féd. de Recherche en Environnement et Développement Durable, FR CNRS 3097, France d Hue University’s College of Education, Hue, Viet Nam b
a r t i c l e
i n f o
Article history: Received 12 February 2011 Received in revised form 7 April 2011 Accepted 12 April 2011 Available online 7 May 2011 Keywords: Anoxic conditions Marine corrosion Steel Rust Sulphate-reducing bacteria
a b s t r a c t Carbon steel electrodes were polarised at a potential ∼150 mV higher than the open circuit potential, in a deaerated seawater-like electrolyte (0.5 mol dm−3 NaCl, 0.03 mol dm−3 Na2 SO4 , 0.003 mol dm−3 NaHCO3 ). X-ray diffraction and -Raman analysis demonstrated that a layer mainly composed of GR(SO4 2− ) had grown on the steel surface. GR(SO4 2− ) was accompanied by traces of GR(CO3 2− ). Similar experiments performed in a solution composed of 0.3 mol dm−3 of Na2 SO4 and 0.03 mol dm−3 of NaHCO3 led to the same result. The nature of the GR forming on steel is thus mainly linked to the sulphate to carbonate concentration ratio. Finally, carbon steel coupons immersed for 11 years in the harbour of La Rochelle (Atlantic coast) were removed from seawater for analysis. The inner part of the rust layer proved to be mainly composed of magnetite, GR(SO4 2− ) and iron sulphide FeS. This definitively confirms that GR(SO4 2− ), as Fe3 O4 and FeS, can form from steel in O2 -depleted environments. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Marine corrosion of carbon steel involves first the formation of the Fe(II–III) hydroxysulphate also known as sulphated green rust GR(SO4 2− ) [1]. Green rusts (GRs) are Fe(II–III) hydroxysalts characterised by a crystal structure that consists of the stacking of Fe(OH)2 -like layers carrying a positive charge due to the presence of Fe(III) and interlayers constituted of water molecules and anions that restore the electrical neutrality of the crystal [2,3]. Several GRs are known, and in particular those based on the main anions found in seawater, that is GR(Cl− ), GR(CO3 2− ) and GR(SO4 2− ). The chemical composition of GR(SO4 2− ) is for instance FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O, [3] sometimes developed as [FeII 4 FeIII 2 (OH)12 ]2+ [SO4 ·8H2 O]2− to remind that the crystal structure is built on positive hydroxide layers alternating with negative interlayers. In seawater, the [Cl− ]/[SO4 2− ] and [SO4 2− ]/[HCO3 − ] molar ratios are about 19 and 12, respectively (computed from [4]). Laboratory experiments were devoted to the oxidation by air of aqueous suspensions of Fe(OH)2 in the presence of sulphate and chloride ions [5,6]. It could be demonstrated that GR(SO4 2− ) formed instead of GR(Cl− ) even in solutions with large [Cl− ]/[SO4 2− ] molar
∗ Corresponding author at: Laboratoire d’étude des matériaux en milieux agressifs (LEMMA), EA 3167, Université de La Rochelle, Bât. Marie Curie, Av. Michel Crépeau, F-17 042 La Rochelle Cedex 01, France. Tel.: +33 5 46 45 82 27; fax: +33 5 46 45 72 11. E-mail addresses: [email protected], [email protected] (Ph. Refait). 0013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2011.04.123
ratios, as the layered structure of GRs presents a strong affinity for divalent anions. Green rusts are readily oxidised into FeOOH compounds by dissolved O2 . So, at the beginning of the process, the rust layer formed on steel in seawater should be composed of an outer layer of Fe(III) oxyhydroxides and an inner layer of GR(SO4 2− ). However, after 6 or 12 months of immersion, GR(SO4 2− ) was found systematically associated with iron sulphide FeS and sulphate reducing bacteria (SRB) [1]. This indicated that anoxic conditions were established at the steel/rust layer interface and inside the inner part of the rust layer. The presence of FeS on steel in seawater is a consequence of the metabolic activity of SRB, since in seawater sulphur is only present as sulphate in the absence of bacterial activity. The anoxic conditions are due to the fact that dissolved oxygen is consumed in the outer part of the rust layers by (i) aerobic micro-organisms and (ii) its reaction with GR(SO4 2− ) that mainly leads to Fe(III) oxyhydroxides. So, after some time, the kinetics of the corrosion is no more controlled by oxygen transport [7–10]. Since the process is then clearly linked to the activity of micro-organisms, it was for instance proposed that the availability and transport of the nutrients necessary for the micro-organisms could be the limiting step [10]. It is now demonstrated that SRB can reduce the sulphate ions coming from GR(SO4 2− ) [11,12]. This phenomenon induces the transformation of GR(SO4 2− ) into various compounds, and in particular iron sulphides such as mackinawite [12]. Therefore it can be expected that GR(SO4 2− ) is totally consumed and that after some time, it is totally absent of the rust layers. However, previous
6482
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
works performed in carbonated media indicated that the carbonated green rust, GR(CO3 2− ) could be obtained in anoxic or almost anoxic conditions [13–15]. So if GR(SO4 2− ) could still form on steel in anoxic conditions it would remain along with FeS one of the components of the rust layer. The aim of this work was then to study the formation of rust on steel in deaerated seawater. In order to focus on the role of anionic species, an electrolyte composed of 0.5 mol dm−3 NaCl, 0.03 mol dm−3 Na2 SO4 and 0.003 mol dm−3 NaHCO3 was considered as a “seawater-like” solution. The concentration ratios [Cl− ]/[SO4 2− ] (=16.6) and [SO4 2− ]/[HCO3 − ] (=10) were close to those characteristic of seawater. Electrochemical experiments (voltammetry, chronoamperometry) were performed in anoxic conditions and the components of the electro-generated rust layers were studied by -Raman spectroscopy, X-ray diffraction (XRD) and scanning electron microscopy (SEM) coupled with energy dispersive spectrometry (EDS). Finally, rust layers grown during 11 years on steel coupons in seawater were analysed by Raman spectroscopy. The behaviour of the real corrosion system could then be interpreted at the light of the information given by laboratory experiments.
2. Experimental 2.1. Electrochemistry Electrochemical experiments were carried out in a classical three-electrode glass cell. The working electrodes were carbon steel disks with 2 cm2 area. The steel approximate composition (in wt%) was: 98.2% Fe, 0.122% C, 0.206% Si, 0.641% Mn, 0.016% P, 0.131% S, 0.118% Cr, 0.02% Mo, 0.105% Ni and 0.451% Cu. The surfaces were polished with silicon carbide (particle size 25 m), rinsed thoroughly with Milli-Q water and rapidly dried in an air flow. The working electrodes were positioned horizontally with the active surface set upwards. They were set in the electrochemical cell just before the beginning of the experiment. Potentials were measured vs. the saturated calomel electrode (SCE). A large platinum grid was used as counter electrode. Voltammetric and chronoamperometric (anodic polarisation) experiments were performed with a PGP 201 potentiostat system (Radiometer analytical). For linear voltammetry, the scans were started at −1.20 V/SCE and ended at +0.5 V/SCE, with a scan rate dE/dt = 1 mV s−1 . Anodic polarisation experiments were performed at −0.60 V/SCE, i.e. approximately 0.15 V above the open circuit potential (OCP). OCP was measured during 30 min at the beginning of each experiment prior to voltammetry or chronoamperometry. It varied around −0.75 V/SCE. Since the study related mainly to the role of anionic species, a simplified electrolyte prepared with NaCl, Na2 SO4 , and NaHCO3 was used instead of artificial seawater. The sodium salts with 98% min. purity (Aldrich) were dissolved in Milli-Q water (resistivity 18.2 M cm). The concentrations, set as 0.5, 0.03 and 0.003 mol dm−3 for NaCl, Na2 SO4 , and NaHCO3 respectively, were close to those typical of seawater. The pH of this electrolyte was measured at the beginning of the experiments at an average value of 7.95. Note that the pH of ASTM D1141 artificial seawater is 8.2. Two procedures were considered for deaeration. First, experiments were achieved in the electrochemical cell under an argon flow. The solutions were deaerated by the argon flow during 45 min before the beginning of each experiment and the argon flow was maintained till the end. Secondly, experiments were performed in an electrochemical cell set inside a glove box (Jacomex P[box]-T4) filled with Argon. In this case, the atmosphere inside the box contained a residual oxygen content ≤1 ppm.
The electrolyte was not stirred during the experiments. However, the argon flow induced by itself a stirring of the electrolyte through its permanent bubbling. In contrast, during the experiments performed in the glove box, the electrolyte was really stagnant. Additional experiments were performed in a second electrolyte, composed of Na2 SO4 (0.3 mol dm−3 ) and NaHCO3 (0.03 mol dm−3 ). It kept the sulphate to carbonate concentration ratio of 10 of the seawater-like electrolyte but the concentrations were ten times larger. The experimental procedure was strictly identical to that used in the seawater-like electrolyte. 2.2. Steel coupons immersed in seawater 1999, a series of carbon steel coupons In (70 mm × 70 mm × 6 mm) were immersed in one the harbours of La Rochelle (Atlantic coast, France) in the frame of a PhD work. Most of the coupons were used at that time but it was decided to leave a few of them in place for future works. One of these coupons was removed for analysis in 2010. It was made of the carbon steel used for laboratory experiments (see Section 2.1) and was permanently immersed at a constant depth of 1 m during those 11 years. For -Raman spectroscopy analysis, the coupons were frozen at −80 ◦ C. This freezing procedure was necessary to preserve the reactive Fe(II) compounds from the oxidizing action of O2 . Moreover, the rust layers are easier to handle in this hardened frozen state than in their wet and slurry room temperature state. Slices of the corrosion products layer could then be easily extracted and their stratification could be investigated. 2.3. Characterisation of the rust layers Micro-Raman analyses were performed on a Jobin Yvon High Resolution Raman spectrometer (LabRAM HR) equipped with a microscope (Olympus BX 41) and a Peltier-based cooled charge coupled device (CCD) detector. The zones analysed through a 50× objective had a diameter of ∼6 m. Spectra were recorded with the acquisition LabSpec software at room temperature with a resolution of ∼0.1 cm−1 . Excitation was provided by a He–Ne laser (632.8 nm). Its power was 0.9 mW, that is 10% of the maximal power, in order to prevent an excessive heating that could have induced the transformation of the analysed sample into hematite ␣Fe2 O3 . The acquisition time was variable, around an average value of 1 min, and was chosen for each analysis so as to optimise the signal to noise ratio. XRD experiments were carried out with a classical powder diffractometer (Brucker AXS® D8-Advance), using Cu-K␣ wavelength ( = 0.15406 nm) in Bragg–Brentano geometry. The electrodes were removed from the electrochemical cell just before analysis. Since the expected Fe(II) compounds were sensitive to the oxidising action of air, the electrodes were immediately placed on the sample holder and the rust layer was coated with glycerol before the analysis. This procedure limits the oxidation of the Fe(II) compounds during the acquisition of the pattern. Finally, the morphology of the electrochemically generated rust layers was observed using scanning electron microscopy (SEM) with a PHILIPS FEI Quanta 200F apparatus. The composition of the main component of the rust layers was estimated by energy dispersive spectroscopy (EDS) coupled to SEM (acceleration voltage: 15 kV). The samples to be analysed were carried from the laboratory to the SEM apparatus inside a bag filled with argon. They were exposed to atmosphere when they were removed from the cell before to be set in the bag and when they were removed from the bag before to be set in the SEM apparatus.
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
Fig. 1. Electrochemical experiments performed with a carbon steel electrode in deaerated seawater-like electrolyte. (a) Polarisation curve obtained under argon flow. The dash lines are the Tafel straight lines used to determine jcor . (b) Chronoamperometric curves obtained during anodic polarisation at E = −0.6 V/SCE: curves (1) and (2) were obtained under argon flow and curve (3) was obtained with a stagnant electrolyte in a glove box under argon atmosphere. The black circles locate the various analyses presented in Figs. 2 and 3.
3. Results 3.1. Anodic polarisation of steel in deaerated seawater-like solutions Typical polarisation and chronoamperometric curves obtained with steel electrodes immersed in the deaerated seawater-like electrolyte are displayed in Fig. 1. The log|j| vs. potential curve of Fig. 1a clearly shows that both anodic and cathodic parts of the curve obey to Tafel law, with anodic Tafel slope ba ∼ 100 mV and cathodic Tafel slope bc ∼ 200 mV. So the corrosion current density jcor could be readily determined, at the intersection of the Tafel straight lines. It was found that jcor was equal to 2.5 × 10−5 A cm−2 , which corresponds to a corrosion rate Vcor of 0.29 mm per year. The corresponding corrosion potential was measured at Ecor = −0.78 V/SCE. Note that experiments performed in aerated conditions led to an initial corrosion rate of about 5.7 mm per year [16]. Anodic polarisation experiments were performed at an applied potential of −0.6 V/SCE, which would give rise to an anodic current density of the order of 1 mA cm−2 according to the polarisation curve of Fig. 1a. Three chronoamperometric curves are displayed in Fig. 1b. The initial current density value (or the maximum reached
6483
a few minutes after the beginning of the experiment) is in each case consistent with what was expected, and measured between 0.7 and 0.8 mA cm−2 . Curves 1 and 2 were obtained with the electrolyte deaerated by an argon flow and curve 3 was obtained with the electrochemical cell placed inside the glove box. The difference between curves 1 and 2 illustrates the variations observed with this methodology, mainly induced by the uncontrolled stirring due to the argon bubbling. However, the main trend is illustrated by curve 2. In this case, the current density decreased from 0.7 to ∼0.15 mA cm−2 in 4 h, remained constant during 10 h before to decrease again down to 0.02 mA cm−2 . The current density reached at the end of the experiment was in this case 35 times smaller than the initial current density. This indicates that the layer of corrosion products forming on the steel surface tended to block the surface. Finally curve 3, obtained in the glove box with a stagnant electrolyte, is rather similar to curve 2. The main difference is that the current density remained constant at ∼0.15 mA cm−2 after its initial decrease. However, the composition and morphology of the rust layers proved to be independent (see thereafter) of the methodology used for deaeration. Actually the differences observed between the chronoamperometric curves of Fig. 1b are not yet clearly interpreted. Typical Raman spectra and XRD pattern of the products generated by anodic polarisation in deaerated seawater-like electrolytes are displayed in Fig. 2. The Raman spectra of Fig. 2a and b were obtained after 15 min and 3 h of polarisation in the glove box. They are characteristic of the various spectra obtained via the analysis of different spots of the surface. Note that similar spectra (not shown) could be obtained on electrodes polarised 26 h, the longest polarisation time considered in this study. This means that the same compound is forming all along the polarisation process. Its Raman spectrum is characteristic of a green rust compound, with two main bands at 430–435 and 509 cm−1 [17–21]. Similar bands are present in the spectra of all green rusts. However, other bands are usually seen that depend on the nature of the green rust. For instance, GR(Cl− ) gives rise to three additional bands at 238, 325 and 363 cm−1 [21], and consequently the GR formed here, characterised by a sharp Raman peak at 219 cm−1 (Fig. 2b) is not GR(Cl− ). Actually, both GR(CO3 2− ) and GR(SO4 2− ) give rise to an additional vibration band at about 220 cm−1 [18–20]. So, Raman spectroscopy cannot discriminate between those two GR compounds. However, the presence of SO4 2− ions was evidenced in some cases (Fig. 2a) by a peak at 995 cm−1 . These anions are more likely adsorbed on the GR crystals. Actually, the anions intercalated in the GR structure are very difficult to observe [22], which is more likely due to a resonance phenomenon between the excitation laser and the vibration modes of the GR hydroxide layers. XRD analysis was then required and the pattern obtained with an electrode polarised for 16 h in the electrolyte under argon flow is presented in Fig. 2c as an example. Two very intense sharp lines are seen. The one located at 2 ∼ 45◦ is due to the main line of the predominant phase of the substrate, i.e. ␣-Fe. Other small lines c of the substrate can be seen. They correspond to Fe3 C. The other intense line, at 2 ∼ 8◦ , is the main line of GR(SO4 2− ) [1,20]. It is accompanied by the other characteristic diffraction lines of this green rust, denoted in Fig. 2c with the corresponding Miller index. Except for the lines of GR(SO4 2− ) and those of the substrate, only one small line due to another compound is seen at 2 ∼ 11.7◦ . It corresponds to the main diffraction line of the carbonated form of GRs, GR(CO3 2− ) [19], which indicates that a small amount of GR(CO3 2− ) was formed together with GR(SO4 2− ). XRD analysis is then consistent with Raman analysis since these GRs give rise to similar Raman spectra. Note that the Raman spectrum of Fig. 2a, obtained after only 15 min of polarisation, is slightly different from the spectra usually obtained for GR(SO4 2− ) [1,12,20], and for instance that of Fig. 2b.
6484
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
Fig. 2. XRD and -Raman spectroscopy analysis of the rust layers formed by anodic polarisation of a steel electrode in deaerated seawater-like solutions. -Raman spectra of layers obtained after (a) 15 min and (b) 3 h of polarisation (glove box). GR, green rust (c) XRD pattern of a layer obtained after 16 h of polarisation (argon flow). The diffraction lines of GR(SO4 2− ) are denoted by the corresponding Miller index. GRC = carbonate green rust and c = Fe3 C. E = −0.6 V/SCE in each case.
The third peak at 219 cm−1 seems to be split here in two peaks, at 195 and 227 cm−1 , another peak appears at 150 cm−1 , and a Raman band due to sulphate ions is clearly visible at 995 cm−1 . This discrepancy may be due to some peculiarity of the GR crystals forming first, e.g. poor crystallinity or non-stoichiometry. For instance, the Fe(II)/Fe(III) ratio in the GR could be larger at the first stage of the GR formation. Anyway, this spectrum demonstrates that 15 min of anodic polarisation were sufficient to form a detectable amount of GR. It was not possible to identify any solid phase precursor of the GR, for instance Fe(OH)2 that could be expected. In conclusion, the main component formed on steel under anodic polarisation in deaerated conditions is GR(SO4 2− ). This compound was also observed to form on steel coupons immersed in seawater [1] but in this case it was oxidised into lepidocrocite ␥-FeOOH. So the effect of the depletion in O2 was to suppress the oxidation of GR(SO4 2− ) into Fe(III) compounds, and the first stage of the corrosion process, that is the formation of GR(SO4 2− ), still took place. This result is consistent with previous works that reported the formation of GR(CO3 2− ) in deaerated carbonated solutions [13–15]. The morphology of the GR(SO4 2− ) layers obtained after 24 h of anodic polarisation was observed by SEM. Some photographs are displayed in Fig. 3. The first one (Fig. 3a) shows an assembly of platelets mainly perpendicular to the steel surface. Their width is
about 5–10 m and their thickness is less than 1 m. Fig. 3b is a zoom on the lower right corner of the previous photograph. One of the platelets is strongly tilted and its hexagonal shape is revealed. This morphology is typical of GR(SO4 2− ) crystals [23]. EDS analysis performed on such individual crystals gave a Fe/S atomic ratio close to 7, that is slightly larger than the value of 6 expected for GR(SO4 2− ) according to its chemical formula FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O [3]. However, it confirms that these crystals are those of GR(SO4 2− ). The discrepancy between measured and theoretical Fe/S ratios may be due to a slight oxidation of GR(SO4 2− ) during the transport of the sample from the electrochemical cell to the SEM apparatus (see Section 2). Such a slight oxidation would have formed a very thin FeOOH layer on the GR crystals. Fig. 3c shows more clearly how the platelets are bound one to each other, leading to a “house of cards” structure. The layered structure of GRs is made of positively charge hydroxide sheets alternating with negatively charged interlayers of anions and water molecules. The faces of the platelets are thus electrically charged. If the sides of the platelets are charged oppositely, then the platelets tend to assemble with such morphology. This implies that the platelets visible in Fig. 3, mainly orientated perpendicularly to the steel surface, are bound to particles that are parallel to the steel surface and so on, exactly as in a house of cards. So the GR(SO4 2− ) layer is not compact at all and the empty spaces left in its “house of
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
6485
Fig. 3. SEM photographs of the rust layer formed after 24 h of anodic polarisation of a steel electrode in anoxic seawater-like solution. The experiment was performed in an electrolyte deaerated by an argon flow. E = −0.6 V/SCE.
cards” structure could be favourable for the colonisation of the rust layers by SRB and the subsequent growth of the colony, as observed in marine corrosion processes [1]. It must be noted that chloride-containing iron compounds were not detected. In particular, the chloride variety of GRs did not form. This is consistent with analysis of rust layers formed on steel in natural seawater after 6–12 months of immersion [1] and with previous laboratory experiments performed in aerated solutions [5]. In contrast, archaeological iron objects left 2000 years in the bottom of the Mediterranean sea proved to be covered by a thick layer mainly composed of a Fe(II)–hydroxychloride, the  phase of Fe2 (OH)3 Cl [24] and green rust compounds were not detected. So the mechanisms of the formation of rust in seawater may be drastically different after some (long) time. 3.2. Role of the sulphate to carbonate concentration ratio The XRD pattern of Fig. 2b revealed that, even if GR(SO4 2− ) was the main component of the electrochemically formed layer, a small proportion of GR(CO3 2− ) was also obtained. This was already observed with steel coupons immersed in seawater [1]. Actually, in natural seawater as in the seawater like-electrolyte considered in this study, three anionic species are likely to give rise to a GR compound: Cl− , SO4 2− and HCO3 − /CO3 2− . It is generally admitted that the GR structure has a variable affinity for anions [5,25–27,6]. In particular, divalent anions are preferred to monovalent anions. This is the reason why, in seawater, GR(SO4 2− ) forms instead of GR(Cl− ) even if the chloride to sulphate concentration ratio is as high as 19. And, among divalent anions, carbonate species are preferred to any other anion, and in particular to sulphate. So, in seawater, GR(SO4 2− ) forms preferentially to GR(CO3 2− ) because (i) the carbonate concentration is low or (ii) the sulphate to carbonate concentration ratio is high. To elucidate this point, additional experiments were necessary. They were performed in an electrolyte composed of 0.3 mol dm−3 of Na2 SO4 and 0.03 mol dm−3 of NaHCO3 . Compared to the seawaterlike electrolyte considered in Section 3.1, this solution (i) does not contain chloride, (ii) keeps the same sulphate to carbonate concentration ratio of 10, but (iii) has sulphate and carbonate concentrations 10 times larger. The measured OCP values for steel in this electrolyte were similar to those found with the seawater-like electrolyte. The main results obtained are summarised in Fig. 4. In Fig. 4a, the chronoamperometric curve 1 obtained with an applied potential of −0.6 V/SCE is compared to curve 2 obtained in seawater-like electrolyte. Both experiments were performed in similar conditions, i.e. the electrolyte was deaerated by an argon flow. The general trend is the same. The current density decreases
Fig. 4. Anodic polarisation of a steel electrode in a deaerated electrolyte composed of 0.3 mol dm−3 Na2 SO4 and 0.03 mol dm−3 NaHCO3 . (a) Chronoamperometric curve (1) compared to that obtained in seawater-like electrolyte (2). (b) XRD analysis of the resulting rust layer. The experiments were performed in electrolytes deaerated by an argon flow. E = −0.6 V/SCE.
6486
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
Fig. 5. -Raman spectroscopy analysis of the rust layer formed on a steel coupon immersed 11 years in seawater. Schematic representation of the rust layer and -Raman spectra of the main components found in the three strata (a), (b) and (c) of this layer. G, goethite; GR, green rust; Mn , nanocrystalline mackinawite; and Mgt, magnetite.
first down to a value close to 0.1 mA cm−2 , stabilises at this value for 4 h and decreases again. The initial current density is however smaller and the decrease of the current faster than in the seawaterlike electrolyte. Fig. 4b displays the XRD pattern results from the analysis of the electrode surface after 22 h of polarisation. The three main diffraction lines of GR(SO4 2− ) are seen, together with the main line of the substrate. They are accompanied by the main diffraction line of GR(CO3 2− ), denoted as GRC. This line is sharper and more intense than in the XRD pattern obtained in seawater-like electrolyte (Fig. 2b). Actually, the intensity of the main line of GRC is 1/5 of the intensity of the main line (0 0 1) of GR(SO4 2− ) in this pattern and 1/13 in the previous one. So, it is clearly the high sulphate to carbonate concentration ratio that favours GR(SO4 2− ) at the detriment of GR(CO3 2− ), but the low carbonate concentration of seawater seems to increase this effect. 3.3. Analysis of a steel coupon immersed 11 years in seawater The result of the -Raman analysis is summarised in Fig. 5. Only 3 of the 50 acquired Raman spectra are presented so as to illustrate the main trends. Globally, the steel surface was covered by a thick layer composed of four main strata. The outer part of this layer mainly contained macro-organisms (shells for instance), micro-organisms, alga, sediments, etc. It was removed from the steel surface before Raman analysis. The three other strata were mainly composed of corrosion products. Note that such rust layers are also colonised by bacteria [1]. The outer stratum (a) exhibited an orange–brown colour. In agreement, Raman analysis revealed the presence of Fe(III) oxyhydroxides, such as goethite ␣-FeOOH (spectrum 3a), and magnetite (Fe3 O4 ). The thicker intermediate stratum (b) was in fact a complex mixture of three main phases, magnetite, mackinawite (FeS) and GR(SO4 2− ), as revealed by spectrum 3b. Mackinawite was in most cases present in its nanocrystalline state, characterised by a Raman spectrum made of
two sharp lines at 208 and 283 cm−1 [28]. The GR spectrum, characterised by three bands at 219, 435 and 508 cm−1 was of course attributed to GR(SO4 2− ). Magnetite was mainly characterised by a large band around 670 cm−1 [29]. Actually, this thick intermediate stratum of corrosion products appeared to be very heterogeneous. In some area, only FeS was detected and in other regions GR(SO4 2− ) was predominating. In some cases, and even close to the metal, lepidocrocite ␥-FeOOH could be detected. This indicates that dissolved O2 could penetrate locally inside the rust layer, more likely through cracks. Finally, the Raman spectra obtained in a thin inner stratum (thickness smaller than 100 m) revealed mainly the presence of GR(SO4 2− ) and magnetite. In particular, FeS could not be detected so close to the steel surface. This indicates that SRB were located farther from the steel surface. This may be due to the fact that the metabolic activity and the growth of the micro-organisms have to be sustained by nutrients that can only come from seawater. So when the rust layers become very thick, SRB could be mainly located in anoxic regions far from the steel surface and close to their source of nutrients. The precipitation of iron sulphides would then also occur far from the steel surface. Note that chloride-containing iron compounds were not detected. This confirms that the presence of -Fe2 (OH)3 Cl in the rust layers of archaeological iron objects [24] is typical of corrosion processes occurring after a very long immersion time. This implies the existence of a transition period (between 11 and 2000 years) when the kinetics of the dissolution of steel and the mechanism of the formation of rust change progressively.
4. Discussion 4.1. Thermodynamics The results presented here confirm that GR(SO4 2− ) can be produced when anoxic or almost anoxic conditions are met at the steel
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
surface. Like Fe3 O4 , GR(SO4 2− ) is a Fe(II–III) mixed valence compound, and can be obtained via the oxidation by O2 of a precursor Fe(II) compound such as Fe(OH)2 [5]. However, such a process cannot take place in anoxic conditions. Actually, it was not possible to detect such a Fe(II) precursor and only GR(SO4 2− ) was observed, even after only 15 min of anodic polarisation. When an anodic polarisation is applied, both Fe(II) and Fe(III) may be electro-generated, leading to the direct formation of the GR from the metal, according to the following anodic reaction: 6Fe + SO4 2− + 20H2 O → FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O + 12H+ + 14e−
(1)
At OCP, during corrosion of steel in anoxic conditions, this reaction is necessarily accompanied by the reduction of water (H2 O or H+ ): 2H+ + 2e− → H2
(2)
The equilibrium conditions between Fe and GR(SO4 2− ) are given by [30]: Eeq (GRSO4 /Fe) = −0.15 − 0.0507 pH − 0.0042 log(aSO4 )
(3)
2−
The activity of SO4 ions, aSO4 , can be estimated using the activity coefficient of sulphate ions in seawater [31], that is SO4 = 0.12.
Then: aSO4 = SO4 × [SO4 2− ] = 0.12 × 0.03 = 0.0036 since the sulphate concentration of the electrolyte is 0.03 mol dm−3 . The pH of the electrolyte being equal to 7.95, the equilibrium potential is computed at: Eeq (GRSO4 /Fe) = −0.54 V/SHE This value was computed for the seawater-like electrolyte considered in this study but it may not differ significantly from the value corresponding to real seawater. It must be compared to the equilibrium potential of H2 O (or H+ ) and H2 , given by: E eq (H+ /H2 ) = 0.00–0.0591 pH
(4)
At pH 7.95, this equation gives Eeq (H+ /H2 ) = −0.47 V/SHE. So Eeq (GRso4 /Fe) < Eeq (H+ /H2 ), which demonstrates that the formation of GR(SO4 2− ) from steel in seawater, in anoxic conditions with water being the oxidizer, is indeed thermodynamically possible. The OCP values were measured accordingly at about −0.75 V/SCE, that is −0.51 V/SHE, and so: Eeq (GRso4 /Fe) < OCP < Eeq (H+ /H2 ). Note that magnetite Fe3 O4 can also form from steel in anoxic conditions. The equilibrium conditions between Fe and Fe3 O4 are given by [30,32]: E eq (Fe3 O4 /Fe) = −0.086–0.0591 pH
(5)
This gives Eeq (Fe3 O4 /Fe) = −0.55 V/SHE at pH = 7.95. Since this value is smaller than that of Eeq (GRSO4 /Fe), GR(SO4 2− ) is metastable with respect to magnetite in seawater, but the difference between the two equilibrium potential values, about 10 mV, is very small. 4.2. Mechanisms When anoxic conditions are established at the steel surface in natural seawater, the formation of GR(SO4 2− ) competes not only with that of Fe3 O4 but also with that of FeS. The formation of FeS is clearly linked to the metabolic activity of SRB. SRB reduce sulphate to sulphide, i.e. transform the species necessary for the formation of GR(SO4 2− ) into species required for the formation of FeS. In contrast, the formation of magnetite may or may not be directly linked to micro-organisms. First, like FeS, magnetite could be generated by microbial activity. It is now demonstrated that SRB can use the sulphate ions coming from GR(SO4 2− ) and consequently
6487
induce the transformation of the GR [11,12]. More precisely, the bacterial growth consumes dissolved sulphate species, which modifies the equilibrium conditions and induces progressively the dissolution of the GR: FeII 4 FeIII 2 (OH)12 SO4 ·8H2 O → 4Fe2+ + 2Fe3+ + SO4 2− + 8H2 O + 12OH−
(6)
The reduction of 1 SO4 2− by SRB produces 1 S2− that precipitates with 1 Fe2+ to form FeS. So the process would leave in solution an excess of 3 Fe2+ and 2 Fe3+ , from which magnetite, a mixed valence Fe(II–III) compound, could form. This assumption would explain the association between the three phases, GR(SO4 2− ), FeS, and Fe3 O4 , observed in the main part of the rust layer. Of course, magnetite could also result like GR(SO4 2− ) from the dissolution of steel. The main parameters that govern the formation of GR(SO4 2− ) vs. Fe3 O4 may be the interfacial Fe2+ aq and SO4 2− concentrations and pH. Of course, the lower is the sulphate concentration the higher is the tendency to form magnetite at the detriment of GR(SO4 2− ). Similarly, it seems that high pH values and low dissolved Fe(II) species concentrations favour magnetite [30]. So, local variations of pH, sulphate, sulphide and dissolved Fe(II) species concentrations can explain the variability of corrosion products observed inside the rust layers of the steel coupon immersed for 11 years in seawater. The main source of heterogeneity inside the composite “rust/biofilm” layer has more likely a biological origin, even though biological heterogeneities can induce in turn electrochemical heterogeneities, i.e. create anodic and cathodic zones on the steel surface. It must also be reminded that transformations of the components of rust could also be induced by micro-organisms such as iron reducing bacteria (IRB) that can reduce Fe(III) compounds to Fe(II) or Fe(II–III) compounds. Actually, it was demonstrated that the simultaneous presence of SRB and IRB favoured the formation of GR(SO4 2− ) at the detriment of that of FeS [33]. Similarly, some micro-organisms are able to achieve direct electron transfer with the steel surface [34,35] or, via enzyme hydrogenase, catalyse proton reduction [36–38]. These micro-organisms can influence the rate of cathodic and/or anodic reactions which could in turn modify the nature of the corrosion products. It is however noticeable that GR(SO4 2− ) was also generated on carbon steel in aerated artificial seawater by cathodic polarisation at E = −1.0 V/SCE [39]. Taken into account the results obtained here under anodic polarisation, it appears that GR(SO4 2− ) can form on steel in seawater and similar electrolytes in a rather wide range of potential around OCP, that is whatever the kinetics of the elementary cathodic and anodic electrochemical reactions. 5. Conclusions It is now demonstrated that layers of GR(SO4 2− ) can be obtained by anodic polarisation of steel in deaerated seawater-like electrolytes. The oxidation of GR(SO4 2− ) into Fe(III) compounds does not take place due to the depletion in O2 which favours the accumulation of the GR onto the steel surface. This shows that GR(SO4 2− ) can still be produced during marine corrosion of steel once anoxic or almost anoxic conditions are established at the steel surface. Raman analysis of the rust layer that covered a steel coupon left 11 years in seawater confirmed it definitively. However, since these anoxic conditions lead to the growth of SRB in natural seawater, the reduction of sulphate to sulphide by these micro-organisms induces the formation of FeS at the detriment of GR(SO4 2− ). So, the formation of GR(SO4 2− ) can only occur locally where the metabolic activity of SRB is null or very low. It was also observed that GR(SO4 2− ) was accompanied by a minor proportion of GR(CO3 2− ). Anodic polarisation in an
6488
Ph. Refait et al. / Electrochimica Acta 56 (2011) 6481–6488
electrolyte keeping the sulphate to carbonate concentration ratio of seawater, but with a carbonate concentration 10 times larger, also led predominantly to GR(SO4 2− ). The sulphate to carbonate concentration ratio is then the major parameter to govern the formation of GR(SO4 2− ) and GR(CO3 2− ). In seawater this concentration ratio is high (∼10) which explains why GR(SO4 2− ) predominates even though carbonate ions are known to ensure a greater stability to the layered crystal structure of GRs. Acknowledgements This work, part of the PhD thesis of M. Langumier, was supported financially by the Conseil Général of Charente Maritime (17), and by the Conseil Régional of Poitou-Charentes (Master of Science of D.D. Nguyen). References [1] S. Pineau, R. Sabot, L. Quillet, M. Jeannin, Ch. Caplat, I. Dupont-Morral, Ph. Refait, Corros. Sci. 50 (2008) 1099. [2] Ph. Refait, M. Abdelmoula, J.-M.R. Génin, Corros. Sci. 40 (1998) 1547. [3] L. Simon, M. Franc¸ois, P. Refait, G. Renaudin, M. Lelaurain, J.M.R. Génin, Solid State Sci. 5 (2003) 327. [4] J.P. Riley, G. Skirrow, Eds., Chemical Oceanography, vol. 2, second ed., Academic Press, 1975. [5] Ph. Refait, J.B. Memet, C. Bon, R. Sabot, J.-M.R. Génin, Corros. Sci. 45 (2003) 833. [6] Ph. Refait, M. Abdelmoula, J.-M. Génin, R. Sabot, C.R. Geosci. 338 (2006) 476. [7] R.E. Melchers, Corrosion (NACE) 59 (2003) 319. [8] R.E. Melchers, Corros. Sci. 45 (2003) 923. [9] R.E. Melchers, R. Jeffrey, Corros. Sci. 47 (2005) 1678. [10] R.E. Melchers, T. Wells, Corros. Sci. 48 (2006) 1791. [11] A. Zegeye, L. Huguet, M. Abdelmoula, C. Carteret, M. Mullet, F. Jorand, Geochim. Cosmochim. Acta 71 (2007) 5450. [12] M. Langumier, R. Sabot, R. Obame-Ndong, M. Jeannin, S. Sablé, Ph. Refait, Corros. Sci. 51 (2009) 2694. [13] L. Legrand, S. Savoye, A. Chaussé, R. Messina, Electrochim. Acta 46 (2000) 111.
[14] L. Legrand, M. Abdelmoula, A. Géhin, A. Chaussé, J.-M.R. Génin, Electrochim. Acta 46 (2001) 1815. [15] S. Savoye, L. Legrand, G. Sagon, S. Lecomte, A. Chaussé, R. Messina, P. Toulhoat, Corros. Sci. 43 (2001) 2049. [16] J.B. Memet, PhD Thesis, University of La Rochelle, France, 2000. [17] N. Boucherit, A. Hugot-Le Goff, S. Joiret, Corros. Sci. 32 (1991) 497. [18] L. Legrand, G. Sagon, S. Lecomte, A. Chaussé, R. Messina, Corros. Sci. 43 (2001) 1739. [19] M. Reffass, R. Sabot, C. Savall, M. Jeannin, J. Creus, Ph. Refait, Corros. Sci. 48 (2006) 709. [20] R. Sabot, M. Jeannin, M. Gadouleau, Q. Guo, E. Sicre, Ph. Refait, Corros. Sci. 49 (2007) 1610. [21] Ph. Refait, P. Girault, M. Jeannin, J. Rose, Colloids Surf. A 332 (2009) 26. [22] Ph. Refait, M. Reffass, J. Landoulsi, R. Sabot, M. Jeannin, Colloids Surf. A 299 (2007) 29. [23] A. Géhin, C. Ruby, M. Abdelmoula, O. Benali, J. Ghanbaja, Ph. Refait, J.-M.R. Génin, Solid State Sci. 4 (2002) 61. [24] C. Rémazeilles, D. Neff, F. Kergourlay, E. Foy, E. Conforto, E. Guilminot,.S. Reguer, Ph. Refait, Ph. Dillmann, Corros. Sci. 51 (2009) 2932. [25] S. Miyata, Clays Clay Mine. 31 (1983) 305. [26] A. Mendiboure, R. Schöllhorn, Rev. Chim. Miner. 23 (1986) 819. [27] Ph. Refait, S.H. Drissi, J. Pytkiewicz, J.-M.R. Génin, Corros. Sci. 39 (1997) 1699. [28] J.A. Bourdoiseau, M. Jeannin, R. Sabot, C. Rémazeilles, Ph. Refait, Corros. Sci. 50 (2008) 3247. [29] D.L.A. De Faria, S.V. Silva, M.T.D. Oliveira, J. Raman Spectrosc. 28 (1997) 873. [30] Ph. Refait, A. Géhin, M. Abdelmoula, J.-M.R. Génin, Corros. Sci. 45 (2003) 656. [31] R.M. Garrels, C.L. Christ, Equilibres Des Minéraux Et De Leurs Solutions Aqueuses, Gauthiers-Villars, Paris, 1967. [32] G.H. Kelsall, R.A. Williams, J. Electrochem. Soc. 138 (1991) 931. [33] J. Duan, S. Wu, X. Zhang, G. Huang, M. Du, B. Hou, Electrochim. Acta 54 (2008) 22. [34] M. Mehanna, R. Basseguy, M.-L. Delia, A. Bergel, Electrochem. Commun. 11 (2009) 568. [35] M. Mehanna, R. Basseguy, M.-L. Delia, A. Bergel, Corros. Sci. 51 (2009) 2596. [36] R.D. Bryant, W.J. Jansen, J. Boivin, E.J. Laishley, W. Costerton, Appl. Environ. Microbiol. 57 (1991) 2804. [37] S. Da Silva, R. Basseguy, A. Bergel, Bioelectrochemistry 56 (2002) 77. [38] M. Mehanna, R. Basseguy, M.-L. Delia, L. Girbal, M. Demuez, A. Bergel, Electrochim. Acta 54 (2008) 140. [39] Ch. Barchiche, C. Deslouis, O. Gil, S. Joiret, Ph. Refait, B. Tribollet, Electrochim. Acta 54 (2009) 3580.