Available online at www.sciencedirect.com

Corrosion Science 50 (2008) 431–435 www.elsevier.com/locate/corsci

Effect of the final annealing of cold rolled stainless steels sheets on the electronic properties and pit nucleation resistance of passive films J. Amri, T. Souier, B. Malki *, B. Baroux Institut National Polytechnique de Grenoble, CNRS/ SIMAP 1130, rue de la piscine BP 75, 38402 Saint Martin d’He`res Cedex, France Received 8 February 2007; accepted 17 August 2007 Available online 7 September 2007

Abstract Semiconducting properties of passive films formed on AISI 304 stainless steel grade were investigated by capacitances measurements in chloride containing aqueous solutions for different surface finishes: BA (bright annealing in hydrogen containing atmospheres) and 2B (standard annealing in oxidising atmospheres followed by pickling in acid, then water rinsing). Mott–Schottky analysis shows that for high enough electrode potential, and whatever the surface finish, the films behave like n-type semiconductors. 2B passive film appears to be more donor-doped than BA one and the density of donor states increases with chloride concentration. The electron donor levels are assumed to be generated by negatively charged cations vacancies produced by the chloride ions reaction with the outer passive film. This reaction looks easier for 2B than BA condition, which explains why BA resists better than 2B to pit nucleation. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: A. Stainless steels; A. Industrial surface finishes; B. Mott–Schottky; C. Pitting corrosion; C. Semiconducting properties

1. Introduction The localized corrosion resistance (pitting and crevice) of a given stainless steel grade dramatically depends on the final annealing performed on the sheet and the subsequent pickling process, which are the source of different surface conditions and properties of the passive films. It was hardly to find up to now a metallurgical parameter that acts separately on the sensitivity to pit nucleation and not on the other pitting stages. Investigating semiconducting properties of two stainless steels sheets made of the same material but having undergone different final treatments seems a better way to overcome this difficulty [1]. In this purpose, we considered two specimens drawn for two AISI 304 cold rolled sheets that were annealed in two different ways:

*

Corresponding author. Tel.: +33 4 76 82 67 29; fax: +33 4 76 82 67 67. E-mail address: [email protected] (B. Malki).

0010-938X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.corsci.2007.08.013

(i) Bright annealing (BA), performed in a hydrogen containing atmosphere. No pickling is required. However, due to the presence of a small amount of residual water vapour in the furnace, a thin superficial oxide (some nm) is formed, resembling to a passive film [2]. (ii) Annealing in an oxidising atmosphere followed by an acidic pickling (2B condition). The passive film is formed by rinsing in water, reinforced after a flash treatment in nitric acid and further natural ageing in (water containing) air. In chloride containing aqueous solutions the BA sheets exhibit a higher pitting potential and a smaller number of metastable pitting transients at open circuit [1] than the 2B ones. In this case, the pitting potential likely directly reflects the sensitivity to passive film breakdown in the pit nucleation stage. How far the semiconducting properties of the passive films generated by the industrial finishing treatment affect this stage is the main objective of the study. As for how to handle the question, Mott–Schottky method supported by XPS chemical analysis is the appropriate

432

J. Amri et al. / Corrosion Science 50 (2008) 431–435

technique to use. Both pH and chloride concentration effects are investigated. On the basis of the obtained results and previous modelling approaches of passive films, an attempt to explain the correlation between semiconducting properties of the surface oxide film and pit nucleation sensitivity is proposed. 2. Experimental AISI 304 0.8 mm thick and 15 mm diameter specimen are cut in industrial stainless steel sheets (see composition in Table 1) with two different surface finishes (2B and BA) as mentioned above. In addition, a mechanically polished AISI 304 sample was used as a reference point (polishing under water with SiC paper up to 1200 grit size). Prior to each experiment, samples were degreased in an ultrasonic acetone/ethanol mixed bath, rinsed with distilled water, dried, and finally aged for 24 h in ambient air. XPS analyses have been carried out on the two surface finishes 2B and BA at 30° and 90° takeoff angles (see Table 2a). The average chemical composition and thickness of the passive films were obtained from the integrated peak intensities of oxygen, iron and chromium. Only oxidised forms of iron and chromium were considered. Carbon, nitrogen, silicon and phosphorus were also detected; however, no nickel was detected in both passive films. It appears from these results (see Table 2b) that the iron/chromium ratio increases significantly from the inner part (90°) to the outer Table 1 Chemical composition of stainless steel type AISI 304 (wt.%, except for O and S) Cr

Ni

Mn Si

Cu

Mo

C

N

Nb

O

one (30°), particularly for BA condition, and that the 2B film seems richer in iron than the BA one. Whatever the surface condition, hydroxides were found predominantly at external part, whereas oxides are rather concentrated at the internal part. The 2B film is richer in hydroxides in its outer part (ratio = 0.58) than that BA one (ratio = 0.76), a non-surprising result regarding the hydrated atmosphere in which the 2B films were formed. The average thickness of the two films was found in the range of 3–4 nm. In brief, these findings confirm the duplex structure of the passive film formed on stainless steels: an outer hydroxides rich layer and an inner iron and chromium oxides rich layer [3]. They will be used later to discuss the Mott–Schottky results. Mott–Schottky experiments were conducted at fixed frequency in NaCl solutions using standard three electrodes set-up monitored by VoltaLab PGZ 301 potentiostat. The pH is controlled by addition of aqueous HCl and NaOH solutions. In order to avoid (or limit) the oxygen cathodic reduction, the solutions were deaerated with pure nitrogen for 2 h. After immersion, the working electrode was maintained under open circuit during 15 min in order to reach a stationary condition. Then, the electrode was polarised cathodically at 1000 mV for 4 min before starting the capacitance measurements. The potential was then scanned in the anodic direction between 1000 mV and 200 mV, a range chosen in order to avoid pitting potentials. The specimen were polarised with consecutive steps of 20 mV min1 during which sinewave signals of 5 mV were applied. All experiments were performed at room temperature (23 ± 2 °C). 3. Results and discussion

S

18.19 8.67 1.42 0.407 0.39 0.209 0.04 0.036 0.006 37 ppm 10 ppm

3.1. Mott–Schottky analysis

Table 2a Angle resolved XPS analysis; elementary chemical compositions of passive films

The total capacitance CT measured by Mott–Schottky method is generally regarded as the contribution in series of both space charge depletion layer (CSC) at the film/electrolyte interface and the classical Helmholtz layer (CH):

Passive film

Angle

C

O

Fe

Cr

2B

30° 90°

11.48 9.56

72.80 71.80

7.79 9.50

7.90 9.13

BA

30° 90°

16.34 12.78

79.58 79.67

2.07 3.50

1.99 4.03

Note that no nickel amount was detected.

Table 2b Angle resolved XPS analysis of oxides and hydroxides of passive films formed on AISI 304 Passive film

Angle

Ratio Fe–ox/Cr–ox

Ratio ox/hydrox

2B

30° 90°

1.66 1.36

0.58 1.16

BA

30° 90°

1.17 0.89

0.76 1.13

Ox = oxide form, hydrox = hydroxide form.

1 1 1 ¼ þ C T C SC C H

ð1Þ

Given a correct value of CSC capacitance the electronic properties are usually deduced from Mott–Schottky relation [4,5]:   1 2 k:T ¼  U  U FB þ ð2Þ q C 2SC e  e0  q  N q where e is the dielectric constant of the passive film, e0 the vacuum permitivity, q the elementary charge (e for an electron and +e for electron hole), Nq the density of charge carriers (NA for acceptors and ND for donors), U the applied potential, UFB the flat-band potential, k the Boltzmann constant and T the temperature in Kelvin. We neglected the term k  T/q as it is only about 25 mV at room temperature and we took the dielectric constant of the pas-

J. Amri et al. / Corrosion Science 50 (2008) 431–435

8 1000 Hz 100 Hz

6

10 Hz

4

2

-2

4

9

1/Csc (F cm x 10 )

sive film equal to 12 as generally used in the literature [3,6–9]. From another hand, CT is generally found frequency dependent [10] and different interpretations have been reported in the literature [11–17]. They are mainly based on electronic surface states contribution (adsorption of chloride for example) or on the presence of traps levels (due to structure order and chemical composition of the film), which may not be ionised at the same frequency. A possible dielectric relaxation in both oxide film and Helmholtz layer is also advanced. Two major models emerge from these analyses. McCann and Badwal [18] introduced the so-called Constant Phase Element (CPE) and deduced a logarithmic law of CT(f), while Young [19] described the oxide films as composed of infinite number of RC units in series with an exponential variation of its resistivity where he deduced an inverse logarithmic law of CT(f). To make sure that one can have confidence in the calculation of CT from, we choose to perform EIS experiments as a function of the frequency. Fig. 1 reports 1/CT vs. ln(f) obtained for a passive film formed on mechanically polished surface in 0.02 M NaCl solution at pH 6.6 and at 200 mV. The experimental result fits quite closely the prediction of Young’s model. The use of superposition principle of infinite number of RC units is still correct. It remains now to estimate the real value of CSC. For this end, we assume that at the flat-band potential the Helmholtz capacitance CH is equal to CT since no depletion layer exists at this potential [20]: C H ¼ C TðU ¼U FB Þ . The capacitance CSC can then be easily deduced from Eq. (1). Fig. 2 shows the Mott–Schottky plots obtained at different frequencies for a passive film formed on mechanically polished surface in 0.02 M NaCl solution at pH 6.6. As expected the Mott–Schottky plots are frequency dependent. One can note however that at 100 Hz or below the experimental data are quite stable and this allowed us to set a threshold frequency in the experimental protocol before conducting definitively any experiments on the two surface finishes.

2 0 -1

-0.4

-0.2

0

0.2

Fig. 3 shows the Mott–Schottky plots at 100 Hz for the three surface finishes in 0.02 M NaCl solution at pH 6.6. Three zones were highlighted, depending on the surface condition. For zones 1 and 3, all films behave like semiconductors in the sense of Mott–Schottky. In the cathodic zone (1), the negative slopes are characteristics of p-type semiconductors, whereas, in the anodic zone (3) the positive slopes hold for n-type semiconductors, which is frequently attributed to the p-type behaviour of the chromium oxides and n-type behaviour of the iron oxides [19,21]. The two zones are separated by a narrow potential zone (2) where the electronic band structures of the films are in the so-called ‘‘flat-band” condition. The corresponding high capacitances values (Helmholtz capacitances) were estimated to about 22 and 15 lF cm2, respectively for 2B and BA films. Furthermore, the calculated acceptor densities NA as well as flat-band potentials are found to be the same whatever the surface finish (see Table 3) but the donor density ND is larger for the 2B condition than for the BA one. Now if we plot the evolution of the thickness of space charge layers W as a function of the applied potential using

9

8

4

6

1

2

3 BA 2B Polished

4

2

-2

0.3

-0.6

Fig. 2. Mott–Schottky plots at different frequencies of the passive films formed on mechanically polished surface in NaCl 0.02 M, pH 6.6.

1/Csc (F cm x 10 )

1/Csc (µF-1cm2)

Experimental data Fitted curve

-0.8

Potential (V/SCE)

10

0.4

433

0.2 1/Csc = 0.021 ln(f) + 0.116

0.1

2 0 -1

-0.8

-0.6

-0.4

-0.2

0

0.2

Potential (V/SCE) 0

0

2

4 ln [f(Hz)]

6

8

Fig. 1. (a) Linear variation of 1/CTvs. ln(f) according to Young’s Model, where f is the frequency. Measurements are obtained by EIS on AISI 304 mechanically polished surface at 200 mV in NaCl 0.02 M, pH 6.6.

Fig. 3. Mott–Schottky plots of the passive films formed on different surface finishes in NaCl 0.02 M, pH 6.6, and at f 100 Hz. The films behave similarly as p-type semiconductor in the cathodic range and as n-type semiconductor in the anodic range. The p-type behaviour of polished surface is not strongly marked due to the insufficient development of the chromium oxides layer.

434

J. Amri et al. / Corrosion Science 50 (2008) 431–435

Table 3 Flat-band potentials and doping densities of passive films formed on AISI 304

Table 4a Effect of pH on semiconducting properties of passive films formed on AISI 304

Passive film

UFB (mV vs. SCE)

NA (1020 cm3)

ND (1020 cm3)

Passive film

pH

UFB (mV vs. SCE)

NA (1020 cm3)

ND (1020 cm3)

2B BA

250 240

5.30 5.58

8.74 5.11

2B

6.6 4

250 60

5.30 3

8.74 10.47

BA

6.6 4

240 55

5.58 3.56

5.11 10.52

The 2B Film is more donor-doped than the BA one. UFB = flat-band potential, NA = acceptor density, ND = donor density.

UFB = flat-band potential, NA = acceptor density, ND = donor density. NaCl 0.02 M, f = 100 Hz.

12

concentration at 0.02 M while varying the pH from 6.6 to 4 (for each value of the pH, measurements were performed on new samples) then, in a second time keeping the pH at 6.6 while varying the chloride concentration towards higher values (0.2 M then 2 M).

W (m x 10-10)

10 8 6 2B

4

3.2. pH and chloride effects

BA

2 0 -0.2

-0.1

0

0.1

0.2

Potential (V/SCE) Fig. 4. Evolution of the thickness of space charge layer in the anodic potential range. The film formed on BA finish develops a thickest space charge layer. NaCl 0.02 M, pH 6.6, f = 100 Hz.

the equation W ¼

h

2ee0 qN D

ðU  U FB Þ

i12

one can see (Fig. 4)

that BA film develops the thickest space charge layer. A result which is in agreement with the XPS analyses suggesting a relatively thicker BA film [22]. Finally, the main difference between the two surface finishes is rather related to the n-type behaviour: 2B film is more donor-doped than BA one. Metallurgical heat treatment conditions may probably be basics for this result [8,23]. Considering the crucial role of H+ protons and chloride concentrations on passive films stability we choose to conduct a parametric study of their effects on the semiconducting properties of our films. We initially fixed the chloride

Table 4a shows the effect of pH on semiconducting properties of both BA and 2B films. One can note that flat-band potentials increase towards acidic pH whatever the surface finishes. The change of UFB per unit pH is approximately the same but far from the Nernstian behaviour of standard semiconductors at room temperature [5]. Moreover, by decreasing the pH, NA and ND densities vary in opposite direction for both films. Decreasing the acidity leads undoubtedly to depletion in acceptor species and enrichment in donor ones. The shift with pH of the flat-band potential for 2B and BA films might be explained by hydrogen adsorption/desorption mechanism. In this view, H+ protons (acceptor species [24]) would adsorb at the film/electrolyte interface and react with the oxygen generating increasingly anionic oxygen vacancies [25,26]. As for chloride effects, the results show when considering n-type behaviour the SC capacitance increases while increasing the chloride concentration, see Fig. 5. However, no noticeable effect was found on p-type behaviour. The corresponding values of UFB, NA and ND are reported in 10

10

AIS I 304 2B 8

8 9 4

2M

4

0.02 M 6

0.2 M 2M

-2

0.2 M

2

2

1/Csc (F cm x 10 )

0.02 M

6

-2

4

9

1/Csc (F cm x 10 )

AISI 304 BA

2

4

2

0

0

-1

-0.8

-0.6 -0.4 -0.2 Potential (V/SCE)

0

0.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

Potential (V/SCE)

Fig. 5. Mott–Schottky plots of the passive films at different chloride concentrations. Note that only n-type behaviour is affected. NaCl, pH 6.6, f = 100 Hz.

J. Amri et al. / Corrosion Science 50 (2008) 431–435 Table 4b Effect of chloride concentration on semiconducting properties of passive films formed on AISI 304 Passive film Cl(M) UFB(mV vs. SCE) NA(1020 cm3) ND(1020 cm3) 2B

BA

0.02 0.2 2

250 200 120

5.30 – –

8.74 11.25 11.60

0.02 0.2 2

240 230 240

5.58 – –

5.11 11.30 12.97

Table 4b. The donor density ND increased with the chloride concentration for both 2B and BA films, whereas the flatband increases for 2B film and seems to be unchanged for BA film (UFB  240 mV). According to the original Point Defect Model [27] the chloride are transferred from the electrolyte to the oxygen vacancies in the outer passive film, decreasing then their concentration and increasing in turn the number of cation vacancies. Alternatively, the chloride ions can adsorb from the electrolytic solution onto the metallic sites (M) of the oxide, producing then cations vacancies (VM) according to the following reactions (written for a divalent metal): þ

þ2



MCl ðadsÞ ! M ðsolÞ þ Cl ðsolÞ

ð6aÞ ð6bÞ

When the Fermi level decreases, the negatively charged cation vacancies loss their electrons and act as donor species [3,28]: 0  V2 M ! VM þ 2e

between the highest acceptor level and the Fermi level and its observed increase with the chloride concentration for 2B film could be then the signature of the presence of such chloride-induced acceptor levels. 4. Conclusions

UFB = flat-band potential, NA = acceptor density, ND = donor density.

MðfilmÞ þ Cl ðsolÞ ! MClþ ðadsÞ þ V2 M ðfilmÞ

435

ð6cÞ

Therefore, the donor density ND characterises the affinity of chloride ions for the passive film (reaction (6a)). From another hand, one may assume that the sensitivity to pit nucleation increases with this affinity, or occurs when the cation vacancies concentration exceeds a critical value, so that ND also features the pit nucleation ability. The fact that 2B film is more donor-doped than BA reveals then a larger sensitivity to pit nucleation, related either to an easier adsorption chloride ions on the metallic sites of the 2B film or, following the original PDM model, an easier incorporation of chloride ions in the oxygen vacancies present in the outer passive film. From another viewpoint, Sato postulated that adsorbed chlorides may generate localized electron acceptor levels below the Fermi level [21]. When, under anodic polarization, the Fermi level reaches a chloride-induced acceptor level, a tunnelling transfer of electrons between the passive film and the metal would occur, leading to a pinning of the Fermi level, then concentrating all further electrode potential increase at the film electrolyte interface, which increases in turn the dissolution rate. Following this view, the flatband potential may be considered as the energetic distance

A comparison of semiconducting properties of two industrial surface finishes (2B and BA) passive films formed on AISI 304 stainless steel was done using Mott–Schottky method supported by XPS analysis. Mott–Schottky results reveal that acceptor densities NA as well as flat-band potentials UFB do not significantly depend on the surface condition. The 2B passive film however exhibits a higher donor density ND that the BA one, which may reflect a larger concentration in cations vacancies, generated by chloride ions at the film solution interface and annihilated at the metal film interface. Assuming that pit nucleation results from an accumulation of such point defects, it should explain why 2B surface finishes are less resistant than BA ones. References [1] G. Berthome´, B. Malki, B. Baroux, Corr. Sci. 48 (2006) 2432. [2] D. Gorse, J.C. Joud, B. Baroux, Corr. Sci. 33 (1992) 1455. [3] N.E. Hakiki, M. Da Cunha Belo, A.M.P. Simoes, M.G.S. Ferreira, J. Electrochem. Soc. 145 (1998) 3821. [4] W. Schottky, Z. Physik 118 (1942) 539. [5] S.R. Morrison, Electrochemistry at Semiconductor and Oxidized Metal Interfaces, Plenium Press, 1980, Ch. 8. [6] M. Da Cunha Belo, N.E. Hakiki, M.G.S. Ferreira, Electrochim. Acta 44 (1999) 2473. [7] L. Hamadou, A. Kadri, N. Benbrahim, Appl. Surf. Sci. 252 (2005) 1510. [8] U. Stimming, J.W. Schultz, J. Phys. Chem. 80 (1976) 129. [9] A. Di Paola, Electrochim. Acta 34 (1989) 203. [10] J.P. Petit, L. Antoni, B. Baroux, in: Modifications of Passive Films, European Federation of Corrosion, The Institute of Materials, London, 1994, pp. 91–95, Book 577. [11] M.W. Peterson, B.A. Parkinson, J. Electrochem. Soc. 133 (1986) 2538. [12] M.H. Dean, U. Stimming, Corr. Sci. 29 (1989) 199. [13] M.H. Dean, U. Stimming, J. Phys. Chem. 93 (1989) 8053. [14] R. Babic, M. Metikos-Hukovic, J. Electroanal. Chem. 385 (1993) 143. [15] W.P. Gomes, D. Vanmackelbergh, Electrochim. Acta 41 (1996) 967. [16] M. Gevers, F.K. Du Pre´, Trans. Faraday Soc. 42 (1946) 47. [17] M. Gevers, Phillips Res. Rept. 1 (1946) 447. [18] J.F. McCann, S.P.S. Badwal, J. Electrochem. Soc. 129 (1981) 551. [19] L. Young, Trans. Faraday Soc. 51 (1955) 1250. [20] N.E. Hakiki, M. Da Cunha Belo, J. Electrochem. Soc. 143 (1996) 3088. [21] N. Sato, J. Electrochem. Soc. 129 (1982) 255. [22] V.A. Alves, C.M.A. Brett, Electrochim. Acta 47 (2002) 2081. [23] K. Azumi, T. Ohtsuka, N. Sato, J. Electrochem. Soc. 134 (1987) 1352. [24] S. Virtanen, P. Schmuki, H. Bo¨hni, P. Vuoristo, T. Ma¨ntyla¨, J. Electrochem. Soc. 142 (1995) 3067. [25] E. Sikora, D.D. MacDonald, J. Electrochem. Soc. 147 (2000) 4087. [26] D.D. MacDonald, Pure Appl. Chem. 71 (1999) 951. [27] D.D. MacDonald, J. Electrochem. Soc. 139 (1992) 3434. [28] M. Froment (Ed.), U. Stimming in Passivity of metals and semiconductors, Elsevier, New York, 1983, p. 477.