Corrosion Science 47 (2005) 1097–1117 www.elsevier.com/locate/corsci

Mechanism of copper action on pitting phenomena observed on stainless steels in chloride media T. Sourisseau

a,b,*

, E. Chauveau b, B. Baroux

a,c

a

b

Laboratoire de Thermodynamique et de Physico-Chimie M
etallurgiques de Grenoble, 1130, rue de la piscine, 38402 Saint Martin d’Heres, France Centre de Recherche d’Ugine Savoie-Imphy (groupe ARCELOR), Avenue Paul Girod, 73400 Ugine, France c Stainless Steel Durability Research Manager, ARCELOR, France Received 27 June 2003; accepted 10 May 2004 Available online 21 December 2004

Abstract The influence of the addition of copper on the resistance to pitting corrosion of stainless steels has been investigated using different experimental techniques––current transient analysis, polarization curves in acidic media, pitting and repassivation potential measurements, XPS and SEM observations––so that pit initiation, propagation and repassivation could be analysed separately. Copper addition is shown to act in three different ways on pitting corrosion. On the one hand, copper reduces steel dissolution rates in acidic chloride media and also pit propagation rates. On the other hand, copper addition in steel is shown to lower repassivation potentials in neutral chloride environments and also to delay pit repassivation. Lastly, when copper is injected into solution as CuCl2 or when the steel is polarized at anodic potentials so that copper can dissolve from the steel into solution, pit initiation close to sulfide inclusions is prevented. A model is proposed for these three different actions of copper, showing that the role of this element is complex and that no relevant information can be drawn from only considering its effect on the pitting potential.
2004 Published by Elsevier Ltd. Keywords: Pitting corrosion; Copper; Stainless steel; Repassivation

*

Corresponding author. Address: Laboratoire de Thermodynamique et de Physico-Chimie Me´tallurgiques de Grenoble, 1130, rue de la piscine, 38402 Saint Martin d’He`res, France. 0010-938X/$ - see front matter
2004 Published by Elsevier Ltd. doi:10.1016/j.corsci.2004.05.024

1098

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1. Introduction Copper is considered either as an undesirable or as a beneficial element in stainless steels, being correlated to the problem of hot shortness causing damage during hot rolling of the steel because of its relatively low melting temperature [1]. Copper is also used as a substitute element for nickel (like carbon, nitrogen or manganese) in austenitic stainless steels even if its austenizing power is relatively weak compared to carbon and nitrogen. In other circumstances, its additions are made to provide substantial precipitation hardening effects in martensite stainless steels [2]. Moreover, copper is sometimes added to ferritic, austenitic or duplex steels in order to improve their resistance to uniform corrosion in sulfuric media [3–6]. Indeed, copper is said by some authors to be enriched at the surface during the anodic dissolution of the alloy and to reduce the corrosion rate of the latter [7,8]. Yet, if the beneficial influence of this element on the corrosion behaviour of stainless steels in sulfuric acid is well understood, its effects in chloride media appear far more complex and remain unclear. Some authors [9,10] claim that copper additions can inhibit the harmfulness of soluble sulfide inclusions by producing insoluble CuX S and hence increases the resistance of the steel to pitting initiation and propagation. Others [7,11] found no substantial effect on the pitting potentials of 18/8 austenitic steels in neutral dilute chloride media nor on the critical pitting temperature of 18/20 austenitic steels in FeCl3 when 2% Cu is added to the materials. Furthermore, the addition of 2% Cu in a 18/10 austenitic alloy in acidic chloride environments was assumed [12] to delay the enrichment of chromium at the surface in the first stage of passivation and also lower the pitting resistance of the steel. The main effects of copper on every step of pit evolution proposed by different authors in the literature are listed in Table 1. Even if a review on the subject shows that the influence of copper on pitting corrosion depends upon both experimental conditions––pH, chloride concentration and temperature––and composition of the steel––especially molybdenum and sulfur

Table 1 Effects of copper on pitting corrosion of steels Pit evolution step

Effect of lowering pitting resistance

Effect increasing pitting resistance

Initiation

Presence of copper enriched particles [13] Lowering of Cr content of passive film [14]

Inhibition of sulfide inclusion harmfulness by copper sulfide formation [9,10]

Promotion of chromium carbide precipitation [13] (leading to local dechromization) Propagation

Increasing anodic dissolution of the steel by acting as cathode [15]

Repassivation

Delay of surface enrichment in Cr [14]

Inhibition of steel dissolution by a metallic copper film at the steel surface [16,17]

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1099

contents––, no comprehensive model has been yet proposed to account for the different results found in the literature. Hence, the present study aims at putting into evidence the major effects of copper on pitting phenomena, focusing on austenitic stainless steels, and explaining them from a mechanistic point of view. In this context, the influence of this element on each step of pitting (i.e. pit initiation, propagation and repassivation) has been studied separately, by using surface analysis techniques (SEM, TEM, XPS) as well as different electrochemical testings. Among these latter, analysis of current transients (signatures of the initiation, propagation and then repassivation of a metastable pit [18–22]) with amplitudes of a few nanoamps or more during potentiostatics gave pieces of information about each step of the pitting process. Those experimental data were correlated to pitting potential that was considered as a global criterion correlated to the probability of appearance of a stable pit [23] (i.e. a global result of the steel resistance to pit initiation and pit propagation and the steel tendency to repassivation). Current transient analysis has been proved to be a powerful technique to lead comprehensive studies on localized corrosion phenomena and the influence of alloying elements [24–26]. As mentioned previously, here copper is added to austenitic steels as an austenizer in order to replace nickel without lowering the stability of austenite. Hence the pitting behaviour of AISI 304 and low-nickel grades is analysed. By isolating the major effects of copper on each step of pitting and by comparing the influence of this element on AISI 304 and low-nickel grades, a mechanism of the action of copper on pitting phenomena is finally proposed, taking into account synergistic effects of copper with other alloying elements like nickel and sulfur.

2. Experimental 2.1. Samples Four grades of experimental austenitic stainless steels produced in the laboratory were studied: (i) Two grades A and A-Cu containing about 8% nickel and whose copper contents are, respectively, 0.2% and 3%; (ii) Two austenitic grades with higher nitrogen and manganese contents which counterbalance a lower nickel one. Both grades B and B-Cu differ from each other by their copper content, respectively, 0.2% and 3%. Their chemical analysis is listed in Table 2. They were produced as 25 kg laboratory heats which were then forged at 1250 C, hot-rolled at 1240 C and finally cold-rolled in order to get 1 mm thick plates. These plates were then heat-treated at 1100 C before being quenched. Samples were cut as 15 mm diameter disks, polished at 3 lm with diamond paste, washed with acetone/alcohol, rinsed with water and

1100

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Table 2 Steel compositions Investigated steels

C

Si

Mn

Ni

Cr

Mo

Cu

S (ppm)

N

A A-Cu B B-Cu

0.03 0.04 0.07 0.06

0.4 0.4 0.8 0.7

1.4 1.4 7.3 7.5

8.6 8.7 1.7 1.6

18.2 18.3 16.4 16.4

0.2 0.2 0.1 0.1

0.2 3.0 0.2 3.0

36 45 9 6

0.1 0.1 0.2 0.2

dried. They were finally left for 24 h in ambient air so that stable passive films could form before any immersion test. SEM observations of sample surfaces show that the four grades contain oxides (either Cr and Mn oxides or alumina) and manganese sulfide inclusions. Addition of copper does not seem to modify inclusion densities. Residual ferrite contents of the four grades were measured by magnetism to ensure that they were all austenitic. 2.2. XPS analysis XPS analysis of passive films formed in different conditions allowed the content of the different metallic elements (Fe, Cr, Mn, Cu) present in oxide and hydroxide layers to be estimated. Passive films were analysed after immersion and rinsing of the samples with deionized water. XPS spectra were provided by a Vacuum Generators XR3E2 spectrometer using a 1253.6 eV Mg X-ray source [27]. Au4f 7=2 pike at 84 eV and Cu2p 3=2 pike at 932.7 eV were considered as reference energies. Charge effect was corrected by taking the C1s energy at 285.0 eV. Energy values were determined with a ±0.l eV accuracy. The main parameters (EL bounding energy, LMH pike width at half-height) used to decompose XPS spectra are summed up in Table 3. 2.3. Electrochemical measurements Pitting potentials were measured in deaerated 0.02 M NaCl pH ¼ 6.6 electrolyte at 23 C by leaving the sample at free potential for 15 min and then running a potentiodynamic scan at a constant sweeping rate (equal to 100 mV/min) until the current reached a value of 50 lA where Vpit pitting potential was taken. When potentiodynamic went on, current still increased until it reached a given Iback value (which might be equal to 100 lA or 1 mA). At this time, potentiodynamic was reverse at the same sweeping rate and the current fell back to its initial value at a potential defined as Vrep repassivation potential. Tests were reproduced three times and dispersions were reported for both pitting and repassivation potentials. Pitting potentials were also measured after 24 h prepolarization treatments at constant Vprep potential. Polarization curves were measured in deaerated acidic NaCl solutions in order to simulate the nature of the confined electrolyte inside pits. According to Isaacs [35], electrolytes whose chloride concentration varies from 0.5 to 4 M and whose pH lies

Elements

Fe0

Fe II

Fe III ox

Fe III hyd

Cr0

Cr III ox

Cr III hyd

Mn0

Mn II

Mn III

Cu0 or Cu I

Cu II

EL (eV) LMH (eV) References

707.0 1.6 [27–29]

709.5 3.0

711.0 3.0

711.8 2.8

574.1 1.8 [28,30]

576.6 2.6

577.2 2.7

639.0 1.7 [31–33]

640.9 3.8

641.8 3.3

932.8 1.8 [33,34]

933.8 3.7

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Table 3 Values of parameters used for the treatment of XPS spectra

1101

1102

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Table 4 Experimental techniques and parameters used in the present study Pit evolution step

Experimental technique

Measured parameters

Initiation

Current transient analysis XPS analysis

Number of transients for 24 h Passive film composition

Propagation

Potentiodynamics Current transient analysis

icrit and ð1=Rp Þ measured in acidic chloride media Transient average current density

Potentiodynamics Current transient analysis ‘‘Potential jump’’ technique

Vrep repassivation potential Transient average charge Time s

Repassivation

between 1.5 and 3 were selected. These potentiodynamic scans were carried out at 10 mV/min. From these tests were deduced critical passivation anodic current density icrit and polarization resistance at corrosion potential Rp . Both criteria icrit and ð1=Rp Þ are correlated to alloy dissolution rate in acidic chloride solutions, hence to pit propagation rate. Moreover, ‘‘potential jump’’ experiments were carried out in 2 M NaCl pH ¼ 1.5 electrolyte at 23 C in order for passivation kinetics in pit-like solutions to be analysed. These tests consist in three successive polarizations: (i) a first one at )0.6 V/SCE (cathodic domain of the four steels) for 5 min so that unstable hydroxides of the passive film were eliminated; (ii) their average current density calculated at maximum current assuming that pits are hemispherical (see Appendix A), which is an indication of the propagation rate of metastable pits; (iii) their average charge, correlated to both pit propagation and repassivation. In conclusion, different experimental techniques allowed several parameters to be determined, each parameter being connected to one of the three steps of pitting (i.e. pit initiation, propagation and repassivation) as summed up in Table 4. Because of the value of current chosen to define the pitting potential (50 lA, i.e. that at least one stable pit has been formed), this latter was considered to characterize the overall resistance of the steel to pitting corrosion, including the different steps (i.e. pit initiation, propagation and repassivation). 3. Results 3.1. Pitting potentials and current transients In order to assess the effect of 3% Cu addition on the overall resistance of the steel to pitting corrosion in neutral dilute chloride media, the values of pitting potentials

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1103

of the four austenitic grades A, A-Cu, B and B-Cu were experimentally determined in deaerated 0.02 M NaCl pH ¼ 6.6 at 23 C (scanning rate: 100 mV/min). In every case, copper addition is observed to lower pitting potentials. Hence, pitting potential of grade A is found to be between 520 and 550 mV/SCE whereas that of grade A-Cu has a value from 440 to 480 mV/SCE, which corresponds to a drop of 80 mV. This drop caused by copper addition is larger (equal to 180 mV) for grades B and B-Cu. This is rather surprising since the effect of copper on the corrosion resistance of stainless steels is often reported to be beneficial. However, it has to be mentioned that the experimental conditions which were chosen for measuring pitting potentials did not allow the formation of a very protective passive film before the test, so that the rest potential is likely to be much lower than the rest potential prevailing in real service conditions. This parameter was to be considered and the pitting potentials were then measured again, but after a 24 h polarization at +100 mV/SCE before the potentiodynamic scan. Such a procedure allows the formation of a more protective passive film before the test and overall the zero current potential is larger in these conditions. The result is that the pitting potentials of all grades A, B and B-Cu are found to lie between 500 and 550 mV/SCE with no detrimental effect of copper addition. From these preliminary results, it appears clearly that the effect of copper is rather complex and that the pitting potential measurement, which includes in a single criterion the different stages of the pitting process, is not a relevant parameter for assessing the behaviour of Cu containing stainless steels. To go further and to find out the mechanisms that lead to these observations, the different assumptions of copper action on each step of pit evolution (cf. Table 1) have been analysed. As far as pit initiation is concerned, TEM analyses of steel thin layers at a ·800,000 magnification show no sign of copper enriched particles in the materials, even in A-Cu and B-Cu grades. Besides, these analyses reveal far higher densities of M23 C6 type carbide precipitates in low copper content grades than in those with 3% Cu, due to the higher carbon content of the former in comparison to the latter. Both observations rule out the assumption according to which copper could enhance pitting initiation by promoting chromium carbide precipitation or by precipitating as metallic copper particles as an explanation for the lowering of pitting potentials by copper addition. Moreover, XPS analyses have been performed on passive films formed on both B and B-Cu grades after being polarized for 24 h in 0.02 M NaCl pH ¼ 6.6 at 23 C and at different potentials. Fig. 2 shows that the chromium contents of passive film oxide layers is slightly lowered (less than 3 at.%) when 3% Cu is added in the matrix (regardless of the value of polarization potential) but this drop is likely too small to engender a fall of 180 mV for the pitting potential as previously observed. Lastly, current versus time signals were analysed during 24 h long potentiostatics in passive conditions in 0.02 M NaCl pH ¼ 6.6 electrolyte at 23 C. A typical current versus time signal is described in Fig. 1. It shows a succession of current fluctuations or transients. Each current transient was considered as the signature of the apparition, propagation and repassivation of a metastable pit [20,23,26] (i.e. a pit that has

1104

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Fig. 1. Typical current versus time signals and current transient.

Fig. 2. Chromium contents of the oxide layer of passive films formed on grades B and B-Cu polarized for 24 h in 0.02 M NaCl pH ¼ 6.6 at 23 C and at different potentials.

repassivated after a few seconds of propagation). A sampling frequency of 18.75 Hz was chosen to get a sufficient resolution for these events to be studied. Two different morphologies of transients have been reported in the literature [23,26,36,37]: type 1 (characterized by a slow increase followed by a sharp fall) and type 2 (quick increase followed by a slow decrease). Type 1 transients are signatures of pits associated with dissolution of Manganese sulfide inclusions whereas type 2 are correlated to pits initiated on passive film defaults or others [37]. In the signals recorded on the grades studied here, type 2 transients were rare and current transients were almost exclusively type 1. We focused in particular on three parameters connected with these transients:

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1105

(i) the number of transients for 24 h whose amplitude is higher than 5 nA, which is directly correlated to the steel resistance to pit initiation; (ii) a second step (called ‘‘propagation step’’) at a potential in the active range (for all the grades studied here) equal to )500 mV/SCE and for a given dt time, in order to simulate pit propagation; (iii) a last polarization at V2 potential, located in the passive domain of the steel for 100 s, during which current values were recorded at a sampling frequency of 100 Hz in order for passive film formation kinetics to be studied. Current typically decreased with time since film formation tended to block steel dissolution and s was defined as the time required for the current to be divided by 10. The smaller the time s, the higher the repassivation kinetics and the better the aptitude of the steel to repassivation. These first results suggest that copper has no significant influence on pit nucleation. That is confirmed by the analysis of current transients during potentiostatics on the four grades: adding 3% Cu either in steel A or in steel B does not modify noticeably the number of metastable pits for 24 h of polarization at different potentials in 0.02 M NaCl pH ¼ 6.6 at 23 C. 3.2. Dissolution rates in acidic chloride electrolytes If the 200 mV drop of pitting potentials observed on grade B when 3% Cu is added is not due to a lower resistance to pit initiation, copper may increase pit propagation rate. In this context, critical current densities in acidic chloride media at different pH values were determined on the four grades and compared to each other as shown in Fig. 3. In this graph does the addition of 3% Cu appear to reduce pit dissolution rates, even if this influence can be negligible when pH > 2. In no case does copper increase pit dissolution rates.

Fig. 3. Critical current densities versus pH for grades A, A-Cu, B and B-Cu in 2 M NaCl at 23 C.

1106

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

3.3. Repassivation If copper does not enhance either pit initiation or pit propagation, the drop of pitting potentials due to 3% Cu addition should come from an effect of this element on pit repassivation. In this context, repassivation potentials and average charges of current transients in a neutral dilute chloride electrolyte were analysed and the results are shown in Figs. 4 and 5. Copper addition in grade B engenders a drop of repassivation potentials whatever Iback may be (100 lA or 1 mA). Moreover, the

Fig. 4. Repassivation potentials of grades B and B-Cu in 0.02 M NaCl pH ¼ 6.6 at 23 C (scanning rate: 100 mV/min––Iback ¼ 1 mA and 100 lA).

Fig. 5. Average charges of metastable pits for grades A, A-Cu, B and B-Cu polarized for 24 h at different potentials in 0.02 M NaCl pH ¼ 6.6 at 23 C.

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1107

presence of 3% Cu in the steels tends to increase the average values of current transient charges, meaning that copper delays repassivation––since it does not enhance propagation rates as seen above. All these results confirm that the drop of the global pitting resistance of the steel when copper is added is connected to a delay of repassivation caused by copper. In order to analyse more closely the way following which copper acts on repassivation kinetics inside a pit, Pourbaix diagrams of Cu–Cl–H2 O system [38] are referred to. In pit-like electrolytes (i.e. acidic chloride media), copper may be stable either as metallic copper, as soluble CuCl
2 complex, as insoluble CuCl salt film or as Cu2þ ions depending on potential value (cf. Fig. 6). Polarization curves of the four steels in pit-like solutions (with [Cl
] between 0.5 and 4 M, and with pH between 3 and 1.5) show that the steel electrode potential at pit bottom is too cathodic for copper to be stable as CuCl nor Cu2þ during pit propagation, and that only metallic copper redeposition on pit walls (as observed in sulfuric acid [5–7]) or dissolution of copper as soluble chloride complexes can occur. In order to determine which of these two last species can act on repassivation kinetics, a first series of ‘‘potential jump’’ experiments on both grades B and B-Cu in pit-like electrolytes were carried out for dt ¼ 0 (i.e. no ‘‘propagation step’’ since dt represents the time of the polarization step in the active range of the steels) and for different values of V2 (V2 being the polarization potential of the third and last step in the passive domain of the steels). On the time s versus V2 plots (cf. Fig. 7), it can be

Fig. 6. Pourbaix diagram of Cu–Cl–H2 O ([Cl
] ¼ 0.1 M) [38].

1108

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Fig. 7. Time s versus potential V2 plots deduced from ‘‘potential jump’’ experiments on grades B and B-Cu in 2 M NaCl pH ¼ 1.5 at 23 C (dt ¼ 0).

first observed that the values of s measured on grade B-Cu are higher than those measured on grade B, meaning that copper addition in steel reduces passivation kinetics, which is coherent with all the previous experimental results. From Fig. 7, it can also be noticed that the more V2 cathodic, the higher this reduction of passivation kinetics due to copper, showing that copper has a pronounced effect on repassivation when metallic and a negligible one when dissolved as chloride complexes during pit propagation. A second polarization aimed at simulating the pit propagation phase was added to these ‘‘potential jump’’ experiments: it consisted in polarizing the steel electrode at )500 mV/SCE (in the active domain of the steel) for different dt times, each one corresponding to one Qdis charge of dissolved metal that was calculated by integrating the current versus time curve between 0 and dt. Fig. 8 represents the evolution of s determined during the third step at )300 mV/SCE (potential located in the passive domain of both steels B and B-Cu and in the domain of metallic copper stability) versus the Qdis charge of dissolved metal during the second step at )500 mV/SCE for grades B and B-Cu in 2 M NaCl pH ¼ 1.5 at 23 C. Whereas s remains rather low and constant when Qdis varies in the case of grade B, s increases with Qdis in the case of grade B-Cu before reaching a plateau. This last observation shows that metallic copper enrichment at the steel surface, which takes place during dissolution and which was revealed by XPS analysis, enhances the reduction of passivation kinetics. This suggests that metallic copper delays passive film formation inside one pit by redepositing on the pit wall and preventing metallic atoms (iron and chromium atoms) underneath from being oxidized. This assumption about the action of copper on repassivation agrees with the higher drop of pitting potentials due to 3% Cu addition in grade B than in grade A as seen previously. Indeed, if potential at pit bottom during pit propagation is assumed to lay in the active domain of the steel immersed in pit-like electrolyte, Fig. 9 shows that copper would be stable as metallic copper in most of the pits formed on grade

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1109

Fig. 8. Time s versus dissolved charge Qdis plots for grades B and B-Cu polarized at three different successive potentials in 2 M NaCl pH ¼ 1.5 at 23 C: (i) at )600 mV/SCE (cathodic domain of both steels), (ii) at )500 mV/SCE (active domain of both steels) for different dt times, (iii) at )300 mV/SCE (passive domain of both steels).

Fig. 9. Polarization curves of grades B-Cu and A-Cu in 2 M NaCl pH ¼ 1.5 at 23 C (scanning rate: 10 mV/min).

B-Cu whereas it would dissolve as CuCl
2 in most of the pits initiated on grade A-Cu, due to the difference of nickel content between both steels. However, in the latter case, metallic copper present in the first layers of pit walls (underneath the surface) in grade A-Cu would perturbate (even if this effect would be very weak) the reconstruction of the passive film in the pit, causing the pitting potential of grade A-Cu to be lightly lower than that of grade A. This means that the action of copper in delaying repassivation––and also lowering global resistance to pitting corrosion––would be more pronounced for grades B and

1110

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

B-Cu than for steels A and A-Cu, which in agreement with the results previously found on pitting potentials. In conclusion, a first effect of copper on steel repassivation has been evidenced: when copper is redeposited as metal on the pit wall during pit propagation, oxidation of iron and chromium atoms is likely to be prevented and repassivation is delayed, by this way decreasing the overall resistance of the steel to pitting corrosion. Since dissolution potential of copper in the pit solution depends on chloride content and temperature, and potential at the pit bottom varies with steel composition (and especially nickel content), it has to be pointed out that this effect of copper is dependent upon both steel composition and experimental conditions. This mechanism could hence account for different results obtained by various authors in the literature [7,11,12,39], for austenitic as well as ferritic steels. 3.4. Interaction of copper with sulfide inclusions A second effect of copper has been evidenced in the present study by injecting copper not via the steel electrode as before but via the aggressive electrolyte as dissolved CuCl2 . The numbers of current transients counted for a 24 h polarization of grade B at +100 mV/SCE (potential at which copper is stable mostly as dissolved Cuþ ions) in two different electrolytes, i.e. 0.02 M NaCl and 0.01 M CuCl2 but at the same pH ¼ 5 and at 23 C, have been compared to each other (cf. Fig. 10). What differs from one solution to the other is simply the addition of cuprous ions into solution in the case of CuCl2 , since pH, chloride concentration and temperature remain identical. Fig. 10 shows that this addition causes a noticeable drop of the number of metastable pits.

Fig. 10. Number of current transients counted for a 24 h polarization of grade B at +100 mV/SCE in two different electrolytes at same pH ¼ 5 and at 23 C.

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1111

Fig. 11. Pourbaix diagram of the Cu–S–Cl–H2 O system calculated with Thermocalc and DCAQ database (for [Cl
] ¼ 0.1 M).

According to the Pourbaix diagram of the Cu–S–Cl–H2 O system calculated with Thermocalc and DCAQ database 1 (cf. Fig. 11), dissolved cuprous ions can react with aqueous sulfur species (produced by dissolution of Manganese sulfide inclusions) to form insoluble Cu2 S which would prevent sulfur species adsorption on steel surface and then film breakdown according to Marcus and Teissier [40]. Moreover, this effect of copper was enhanced by prepolarization treatments undergone by steel electrodes in neutral dilute chloride media. Indeed, pitting potentials of the four steels after 24 h polarization at a given Vprep potential in 0.02 M NaCl pH ¼ 6.6 and at 23 C were determined. By repeating the experiment for different values of Vprep , pitting potential versus Vprep curves were drawn for the four grades (cf. Fig. 12). The influence of Vprep on pitting potential is found to be rather negligible in the cases of grades A and B unlike for grade B-Cu. For this latter, Vpit remains low as long as Vprep is more cathodic than )150 mV/SCE (i.e. close to the equilibrium potential for the transition Cu0 /CuCl
2 in 0.02 M NaCl pH ¼ 6.6 electrolyte [38]) because of the first effect of metallic copper on repassivation. However, when Vprep becomes more anodic than )150 mV/SCE, copper from pits initiated during prepolarization can dissolve into solution and react with sulfur species to form insoluble copper sulfide and prevent sulfur from adsorbing on passive film and promoting pit initiation. By inhibiting pit initiation, the first effect of metallic copper on repassivation is also avoided, which makes Vpit increase up to the values of grades A and B as observed in Fig. 12. Lastly, copper sulfides formed after 24 h polarization

1

DCAQ database from Thermocalc for aqueous species.

1112

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Fig. 12. Vpit versus Vprep curves for grades A, B and B-Cu after 24 h polarization at Vprep in 0.02 M NaCl pH ¼ 6.6 at 23 C (sweeping rate for Vpit determination: 100 mV/min).

at potentials more anodic than )150 mV/SCE on grade B-Cu were observed by microprobe as shown in Fig. 13. Concentration profiles suggest that Cu2 S rather

Fig. 13. Dissolved MnS after 24 h prepolarization of grade B-Cu at +100 mV/SCE in 0.02 M NaCl pH ¼ 6.6 at 23 C analysed by microprobe with corresponding concentration profiles.

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1113

Fig. 14. Mechanism of action of copper on stainless steel resistance to pitting corrosion.

than CuS are formed adsorbing on steel surface and hence sulfide inclusions from promoting pit initiation. This observation is in agreement with Wranglen’s ideas [10]. It has to be pointed out that since copper acts in three different ways on pitting resistance and these actions are dependent upon both steel composition and environmental conditions, this makes it difficult to predict the influence of copper addition on pitting behaviour of stainless steels in real service conditions. Fig. 14 sums up the mechanism of action of copper on pitting corrosion observed on stainless steels––since austenite structure is not involved in any of these effects, this mechanism should also apply to ferritic steels. This mechanism, based on the three different effects mentioned previously, should explain the apparently controversal results found by several authors who studied the influence of copper on the behaviour of stainless steels in localized corrosion.

4. Discussion Three opposite effects of copper on pitting phenomena on austenitic stainless steels have been evidenced in the present study.

1114

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Firstly, the addition of 3% Cu in steels lowers pitting potentials in neutral and dilute chloride media. Since this addition does not affect the number of metastable pits neither increases dissolution rates in acidic chloride environments, it is deduced that copper does delay repassivation. This result is confirmed by repassivation potential measurements (which are lowered by copper addition in the steels) and by metastable pit charges deduced from current transients (which increase when the steel contains 3% Cu). Hence, copper lowers stainless steel resistance to pitting not by destabilising the passive film as Ujiro et al. [17] suggest but by hindering repassivation and promoting pit stabilisation. Moreover, ‘‘potential jump’’ experiments show that this effect of copper is more pronounced when this element is stable under its metallic state. It is also suggested that repassivation is delayed when copper is stable as metal on pit walls––i.e. potential at the pit bottom is more cathodic than that of copper dissolution in the pit electrolyte. Under these conditions, enrichment of metallic copper is likely to take place on the pit walls during pit propagation and this metallic barrier is assumed to prevent iron and chromium atoms from being oxidized and forming a stable passive film. This lowers the overall steel pitting resistance characterized by pitting potential. When nickel content is lowered, potential at the pit walls is also more cathodic (according to potentiodynamics run on stainless steels with different nickel contents immersed in pit-like solutions) and metallic copper is more likely to be stable and redeposit inside the pit. This observation is in agreement with the results found on the four investigated stainless steels, showing that this detrimental effect of copper is more pronounced on grades with lower nickel contents. The use of microelectrodes and local chemical analysis measurements would be very useful to put redeposition of metallic copper on pit wall into evidence. On the other hand, a beneficial effect of copper has been observed on austenitic stainless steels: copper addition tends to lower dissolution rates in acidic chloride solutions and also pit propagation rates. This observation supports the assumption (made by several authors) that redeposition of metallic copper at the active steel surface may form a barrier which would inhibit metal dissolution [8,16]. This assumption has been proposed for the effect of copper when steel is immersed in non-chloride acidic environments, but can also account in acidic chloride media [17]. This observation is in agreement with the fact that this effect shown here to be more pronounced on grades with a lower nickel content, for which copper is supposed to be stable as metallic copper on the pit wall as previously mentioned.

5. Conclusions Copper added as alloying element in austenitic stainless steels may have three different effects on pitting corrosion resistance, the first two being beneficial and the last one detrimental.

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1115

(i) It tends to lower dissolution rates in acidic chloride solutions and then decreases pit propagation rates. (ii) When dissolved into solution, it is found to inhibit the detrimental effect of dissolved sulfur species on pit initiation via the formation of insoluble copper sulfides. (iii) When potential at pit bottom is more cathodic than that of copper dissolution in the pit electrolyte, copper is believed to hinder pit repassivation, then promoting pit stabilisation at lower pitting potentials. • Then, nickel content has an indirect beneficial effect on the pitting resistance by increasing the potential at the pit bottom, favouring the dissolution of copper ions rather than its precipitation in the metallic state. Acknowledgements The authors wish to acknowledge the ARCELOR group, Ugine Savoie-Imphy company (now part of ARCELOR group) and the French government (A.N.R.T.) for their financial support. They are also grateful to the staff of Ugine Research Center and L.T.P.C.M. for their contribution to this work. Appendix A. Calculation of transient current density A current density has been calculated for each current transient at the maximum current. For this purpose, the charge Qp correlated to the current transient at maximum current has been determined as in Fig. 15. If charge Qp is assumed to be entirely correlated to dissolution of metallic elements into cations (proportionally to the contents of different elements in the steel) and if the pit is supposed to be hemispherical, then the volume V of the metastable pit at the maximum current of the transient is equal to 4 ðA:1Þ V ¼ pr3 3

Fig. 15. Charge Qp of a type 1 transient.

1116

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

Fig. 16. Metastable pits initiated near MnS inclusions.

and V ¼

Meq Qp qZeq F

ðA:2Þ

with • M Peq : equivalent molar mass associated with the dissolved elements Meq ¼ i %i  Mi with %i steel content of element i and Mi molar mass of element i and i ¼ Fe, Cr, Mn, Ni. • Zeq : equivalent oxidation state of cations assuming that Cr3þ , Fe3þ , Mn3þ and Ni2þ are produced Zeq ¼ 2  %Ni þ 3  ð%Fe þ %Cr þ %MnÞ • q: steel density equal to 7900 kg/m3 [1] and F : Faradic constant. Pit radius is deduced by combining Eqs. (A.1) and (A.2): sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3Meq 3 r¼ Qp 2pqFZeq Current density i at the maximum current of the transient is then equal to Imax i¼ 2 pr SEM analyses showed that numerous repassivated pits appeared to be hemispherical, as shown in Fig. 16.

References [1] D.A. Melford, Philosophical Transactions of the Royal Society A 295 (1980) 89. [2] D.T. Llewellyn, Ironmaking and Steelmaking 22 (1) (1995) 25. [3] G. Hultquist, M. Seo, T. Leitner, C. Leygraf, N. Sato, Corrosion Science 27 (9) (1987) 937.

T. Sourisseau et al. / Corrosion Science 47 (2005) 1097–1117

1117

[4] B. Postrach, I. Garz, H.H. Stehblow, Werkstoffe und Korrosion 45 (1994) 508. [5] J.C. Bavay, P. Damie, M. Traisnel, in: Froment (Ed.), Passivity of Metals and Semiconductors, Elsevier, Amsterdam, 1963, p. 613. [6] L. Fedrizzi, A. Molinari, F. Deflorian, A. Tiziani, British Corrosion Journal 26 (1) (1991) 46. [7] H. Lin, W. Tsai, J. Lee, C. Huang, Corrosion Science 33 (5) (1992) 691. [8] M. Seo, G. Hultquist, C. Leygraf, N. Sato, Corrosion Science 26 (11) (1986) 949. [9] D.M. Buck, Transactions of the American Electrochemical Society 39 (1921) 109. [10] G. Wranglen, Corrosion Science 9 (1969) 585. [11] R.J. Brigham, E.W. Tozer, Corrosion 30 (5) (1974) 161. [12] Y. Jiangnan, W. Lichang, S. Wenhao, Corrosion Science 33 (6) (1992) 851. [13] J. Banas, A. Mazurkiewicz, Materials Science and Engineering A 277 (2000) 183. [14] A. Yamamoto, T. Ashiura, Corrosion Engineering 37 (1988) 345. [15] M. Streicher, Materials Performance 37 (May) (1983). [16] U. Heubner, M.B. Rockel, E. Wallis, ATB Metallurgie 25 (3) (1985) 235. [17] T. Ujiro, S. Satoh, R.W. Staehle, W.H. Smyrl, Corrosion Science 43 (2001) 2185. [18] P.C. Pistorius, G.T. Burstein, Corrosion Science 36 (3) (1994) 525. [19] S.P. Mattin, G.T. Burstein, Philosophical Magazine Letters 76 (5) (1997) 341. [20] G.T. Burstein, S.P. Mattin, in: Proceedings of the Symposium on Critical Factors in Localized Corrosion II, vol. 95-15, The Electrochemical Society, 1995. [21] T. Suter, H. B€ ohni, Electrochimica Acta 43 (19/20) (1998) 2843. [22] Y. Kobayashi, S. Virtanen, H. B€ ohni, Journal of the Electrochemical Society 147 (1) (2000) 155. [23] C. Gabrielli, F. Huet, M. Keddam, R. Oltra, Corrosion 46 (4) (1990) 266. [24] C. Boulleret, J.L. Pastol, J. Bigot, B. Baroux, D. Gorse, Journal de Physique IV 5 (nov) (1995). [25] C. Boulleret, D. Gorse, B. Baroux, Corrosion Science 38 (6) (1996) 999. [26] G.T. Burstein, P.C. Pistorius, S.P. Mattin, Corrosion Science 35 (1993) 57. [27] C. Vittoz, thesis, INP Grenoble, France, 1996. [28] F. Gaben, thesis, Universit
e de Bourgogne, France, 1992. [29] D. Brion, Applications of Surface Science 5 (1980) 133. [30] E. De Vito, P. Marcus, Surface and Interface Analysis 19 (1992) 403. [31] J.C. Carver, G.K. Schweitzer, T.A. Carlson, Journal of Chemical Physics 57 (2) (1972) 973. [32] G.C. Allen, S.J. Harris, J.A. Jutson, J.M. Dyke, Applied Surface Science 37 (1989) 111. [33] D. Briggs, M.P. Seah, Practical Analysis by Auger and X-ray Photoelectron Spectroscopy, Wiley and Sons Publishers, NY, 1983. [34] M. Lenarda, R. Ganzerla, L. Storaro, R. Frattini, S. Enzo, R. Zanoni, Journal of Materials Research 11 (1996) 325. [35] H. Isaacs, Corrosion Science 29 (1989) 313. [36] C. Gabrielli, F. Huet, M. Keddam, R. Oltra, in: Proceedings of the 2nd International Conference on Localized Corrosion, NACE, Orlando, 1987, pp. 93–108. [37] C. Boulleret, thesis, Paris XI, France, 1997. [38] D. Tromans, R. Sun, Journal of the Electrochemical Society 138 (11) (1991) 3235. [39] J.P. Audouard, Technical Report RCA 92601, IRSID Unieux, 1992. [40] P. Marcus, A. Teissier, J. Oudar, Corrosion Science 24 (4) (1984) 259.