Electrochemical behaviour of martensitic stainless steel after immersion in a H2S-saturated solution

Corrosion Science xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Corrosion Science journal homepage: www.elsevier.com/locate/corsci

Electrochemical behaviour of martensitic stainless steel after immersion in a H2S-saturated solution ⁎

Xiaowei Leia,c, Hongyan Wangb, , Feixiong Maoc, Junping Zhanga, Mifeng Zhaod, Anqing Fue, Yaorong Fenge, Digby D. Macdonaldc a

The Key Laboratory of Space Applied Physics and Chemistry, Ministry of Education, School of Science, Northwestern Polytechnical University, Xi’an 710072, China Key Laboratory of Macromolecular Science of Shaanxi Province, School of Chemistry and Chemical Engineering, Shaanxi Normal University, Xi’an 710062, China c Department of Materials Science and Engineering, University of California at Berkeley, Berkeley, CA, 94720, USA d Oil and Gas Engineering Research Institute of PetroChina Tarim Oilfield Company, Korla, Xinjiang, 841000, China e State Key Laboratory of Performance and Structural Safety for Petroleum Tubular Goods and Equipment Materials, Tubular Goods Research Institute, CNPC, Xi’an 710077, China b

A R T I C L E I N F O

A B S T R A C T

Keywords: Martensitic stainless steel EIS Potentiodynamic polarisation H2S Passivity

Time dependent experiments were carried out to study the corrosion behaviour of Super 13Cr martensitic stainless steel after immersion in a H2S-saturated solution. The Point Defect Model was employed to obtain key parameters. With increasing immersion time, the corrosion product layer thickened, but displayed a defective structure. The thickness of the barrier layer decreased with extending the H2S-exposure time, which notably lowered the corrosion resistance. The presence of the corrosion scale resulted in a thinner barrier layer in contrast to that without the scale. With/without the corrosion scale, no semi-conductivity and no passive region was observed for Super 13Cr after 96 h-immersion. The hydrogen ingress during H2S-exposure lowered the passivity of the bare alloy substrate.

1. Introduction The corrosion of tubular goods in the oil and natural gas industry in wet H2S environments has resulted in extremely high costs and sometimes even to the loss of life and property. It is generally believed that H2S is aggressive to carbon and low-alloy steels by accelerating the anodic dissolution reactions, resulting in severe weight-loss [1–3]. In addition, the cathodic process in wet H2S environments (e.g., hydrogen evolution) may also be catalysed on iron sulphide surfaces and lead to hydrogen embrittlement (HE), hydrogen induced cracking (HIC), and sulphide stress cracking (SSC) [4–6], all of which result from the entry of hydrogen into the alloy. In order to lower the risk of corrosion failure, compared with the low-grade steels, stainless steels have been widely employed in H2S-containing environments, due to their better corrosion resistance, although the presence of H2S also changed the semiconductor properties of passive film and increased the susceptibility of stainless steels to pitting corrosion [7–12]. Several investigations have contributed to developing an understanding of the influence of various corrosive media containing H2S on the passive behaviour of stainless steels [7,9]. In comparison, because of the high formation rate of corrosion products on the surface of electrodes in H2S-containing

⁎

media [1,9], the often-neglected effect of the corrosion products on the electrochemical properties of steel should be explored. Moreover, sulphide ions act as a hydrogen atom recombination poison, and the recombination process of atomic hydrogen can be significantly retarded, leading to the increased concentration of hydrogen absorbed by the substrate material [1]. Therefore, it is necessary to analyse the influence of hydrogen absorption on the passivity of stainless steels. The pre-charging method of cathodically polarizing the working electrode to a potential, at which hydrogen is evolved on the surface, was commonly applied in an acidic solution. Yang and Luo [13] found that hydrogen enhances the pitting susceptibility of Type 304 austenite stainless steel, which was rationalized by the authors in terms of “the enhanced substitution of chloride ion generated from the increase of hydrogen-containing species in the passive film” [14]. However, it is more likely that, due to recombination of hydrogen atoms within the blisters that are precursors to passivity breakdown, the blister ruptures at a shorter time after the initiation of cation vacancy condensation at the metal/barrier layer (m/bl) interface than would be the case in the absence of blister pressurization via hydrogen. Ningshen and Mudali [15] reported that hydrogen increased the passive current density and concentrations of the point defects in the passive film of

Corresponding author. E-mail address: [email protected] (H. Wang).

https://doi.org/10.1016/j.corsci.2017.10.015 Received 27 February 2017; Received in revised form 20 September 2017; Accepted 21 October 2017 0010-938X/ © 2017 Elsevier Ltd. All rights reserved.

Please cite this article as: Lei, X., Corrosion Science (2017), https://doi.org/10.1016/j.corsci.2017.10.015

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

Fig. 1. The surface and cross-sectional morphologies of Super 13Cr specimens after immersion for (a, d) 5 h, (b, e) 20 h, and (c, f) 96 h in NACE Solution A. The surface and cross-sectional images were obtained by using secondary and back-scattered electrons, respectively.

The surface and cross-sectional morphologies of the specimens after immersion were observed by using a Vega3 Easyprobe SEM with an EDS attachment. It was found that corrosion scales were generated on the surfaces of the WEs after immersion. For a detailed comparison of the scales formed under different conditions, the electrodes were divided into two groups to analyse the electrochemical behaviour of the stainless steel with and without corrosion scale. The specimens without corrosion scale were obtained by gently polishing the immersed WEs with a 1500 grit SiC paper. The criteria for stopping the polishing process was when the metal substrate was fully exposed. Note that all the “specimens without scale” mentioned in this study refer to this group of WEs. An aerated 3.5% NaCl aqueous solution was used to investigate the electrochemical properties of the H2S-immersed specimens. The experimental temperature was controlled at 40 °C by using a recirculating water bath. Potentiodynamic polarisation curves were recorded by using a CS370 electrochemical workstation produced by CorrTest Instruments Corporation in China. EIS and Mott-Schottky experiments were carried out using an EG&G M273A potentiostat/galvanostat attached with an M5210 lock-in amplifier. Two graphite electrodes and a saturated calomel electrode (SCE) were used as the counter electrode (CE) and the reference electrode (RE), respectively. Before the measurements, the WEs were immersed in the electrolyte and stabilized for 30 min. The EIS data were then measured from 10 kHz to 0.05 Hz with potential amplitude of 10 mV. The Mott-Schottky experiment was carried out with an applied potential from −0.2 VSCE to 0.15 VSCE, and the potential was swept at a rate of 12.5 mV/s. This procedure ensures that the barrier layer thickness remains constant and the point defect structure remains “frozen” over the duration of the experiment, thereby more accurately conforming to the assumptions behind Mott-Schottky theory. The potentiodynamic polarisation test was started from −0.2 V (vs. OCP) with a sweep rate of 1 mV/s.

316L austenite stainless steel. Guo et al. [16] revealed that hydrogen promotes the initiation and growth of pits in 2507 duplex stainless steel, and the pits tended to nucleate at austenite phase or austenite/ ferrite interface. Thus, from these studies, we infer that the absorbed hydrogen, which is generated by cathodic reaction in H2S-containing solutions, should have a significant impact on the electrochemical behaviour of stainless steel. However, to the best of our knowledge, few studies related to the impact of hydrogen ingress on the passivity of martensitic stainless steel in H2S environments have been reported. This paper presents a study of the corrosion behaviour of Super 13Cr martensitic stainless steel in a 3.5% NaCl solution after immersion into a saturated H2S solution. Study of the immersed specimens, without/ with removing the corrosion scale, not only reveals the impact of corrosion scale on the corrosion behaviour of Super 13Cr stainless steel in H2S environments, but also unravels the impact of hydrogen ingress that is induced by H2S on the electrochemical properties of Super 13Cr substrate. 2. Experimental The chemical composition of the Super 13Cr specimen was (wt.%): 0.027C, 12.87 Cr, 5.32 Ni, 2.20 Mo, 0.18 Si, 0.47 Mn, 0.022 P, 0.004 S and balance Fe. The working electrode (WE) with dimensions of Ф15 × 5 mm was spot-welded to copper wires and embedded in twocomponent epoxy resin, leaving an exposed area of 1.766 cm2. Before immersion, the WE was ground to 1500 grit, rinsed with deionized water, degreased in ethanol, and finally dried with a stream of cold air. Subsequently, the WE was immersed into a NACE Solution A [17], which comprises 5% NaCl and 0.5% CH3COOH in aqueous solution at ambient temperature and pressure. The solution was deoxygenized with high purity N2 gas for 4 h, then purged with H2S gas until saturated (approximately 2 h), and kept purged with H2S gas throughout the experiment. During the immersion process, the buffering effect of acetic acid [12] was used to ensure that the pH remains constant (equal to 2.7). The experimental temperature was 25 °C, and the immersion times were 5, 20, and 96 h. After immersion, the WE was rinsed with deionized water, dipped into ethanol, and dried with a stream of cold air. 2

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

are listed in Table 1. The compositional difference between the 20 h and 96 h specimens indicates that the Ni, Fe, S contents increased and the Cr content decreased with increasing immersion time. By comparing Fig. 1d–f, it is seen that a scale layer was generated and thickened with extended immersion time. After 5 h of immersion, the scale thickness is less than 1 μm. By extending the immersion time to 20 h and 96 h, the scale thickness increased to about 2 μm and 12 μm, respectively. It should be noted that the cross-sectional morphologies were observed using SEM in the back-scattered electron (BSE) mode. The brightness of corrosion scale is lower than that of the substrate, indicating the scale contains elements with lower atom weight. According to Table 1, elements S and O should mainly contribute to the lower brightness of corrosion scale (the atom weights of Fe, Ni, and Cr are very close to each other). Based on the observation from Marcus and Protopopoff, the chromium sulphides are thermodynamically unstable [18]. Instead of the formation of CrxSy in electrochemical reactions, the following anionassisted reaction mechanism may take place [10]:

Table 1 EDS analysis of corrosion scale formed on the electrode surfaces after immersion (wt.%). Element

C

Si

Cr

Ni

Fe

S

O

Al

Cl

Ca

Fig. 1b Fig. 1c

5.11 6.96

1.10 0.34

27.80 5.96

12.62 20.16

10.66 16.28

9.42 27.31

29.77 22.99

0.39 –

2.37 –

0.76 –

3. Results 3.1. Morphology and composition of the corrosion scales Fig. 1 shows the surface and cross-sectional morphologies of Super 13Cr specimens after immersion in NACE Solution A for 5, 20 and 96 h. Clearly, increasing immersion time results in the growth of the corrosion scale, as judged by the progressive disappearance of the surface abrasion marks. For the 5 h-immersed specimen, the abrasion scratches on the substrate are still apparent, indicating that the scale is very thin. After 96 h immersion, the electrode was fully covered by the corrosion scale, although many micro-cracks are evident in the scale, which may be attributed to the interior stress of the thick scale or/and drying of the film prior to SEM observation. The chemical composition of the corrosion scales in Fig. 1b and c was characterized by EDS, and the results

Cr+H2 S+H2 O⇔ (CrSH−)ads + H3 O+

(1)

(CrSH−)ads ⇔ (CrSH)ads + e−

(2)

(CrSH)ads → CrSH+ + e−

(3)

Fig. 2. EIS plots for the specimens after immersion for various time and analysed with the corrosion scale intact: (a) Nyquist plots; (b) and (c) are Bode plots vs. modulus and phase angle, respectively.

3

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

CrSH+ + H+ ⇔ Cr 2+ + H2 S

(4)

The chromium ions generated by Reactions (1)–(4) tend to be transported to the solution rather than stay in the corrosion scale, although the presence of a defective chromic oxide barrier layer is expected, judging from the behaviour of other chromium-containing Fe and Ni alloys. Thus, it is a “general rule of thumb” in the study of passivity that the component of an alloy which has the greatest affinity for oxygen (in this case, Cr) forms the barrier layer, provided that it is present in sufficient concentration (a few percent for chromium). In contrast, Fe and Ni, which have lower affinities for oxygen, are depassivated by sulphide, and will react with HS− (S2−) to form stable, non-protective sulphide compounds, e.g. FeS, FeS2, NiS, Ni3S2 scales [19,20]. As a result, the fractions of Fe- and Ni- sulphides in the corrosion scale increased with the immersion time, giving rise to the higher Fe, Ni, and S content in the scale on the 96 h-immersed specimen. These sulphides most likely exist in the porous, precipitated outer layer of the passive film (scale).

Fig. 4. Equivalent electrical circuit for optimisation of the PDM based upon the impedance data. Zf ^derivation of the Faradic impedance, W ^Warburg impedance due to the transportation of the point defects in the barrier layer, Cg ^geometric capacitance, Ze,h ^diffusional impedance of the electrons and/or electronic holes, Cdl ^capacitance of the double layer, Rct ^resistance of charge transfer, Col ^capacitance of the outer layer, Rol ^resistance of the outer layer, Rs ^resistance of the solution between the outer layer and top of the Luggin capillary.

universal circumstances (n + p type barrier layer). In Fig. 3, Reaction 2 creates the minimal unit of the barrier layer at the m/bl interface, while Reaction 5 destroys the unit at the bl/ol interface. When the rates of Reaction 2 and Reaction 5 are equal, the steady-state is reached. Based on this, the PDM derived the equations for calculating Lss, which is a function of applied potential and pH [24], and the steady-state passive current density (iss). The values for those quantities can be calculated by inputting the kinetic parameters that are obtained from optimisation of the PDM on the experimental impedance data. Theories and expressions for calculating the kinetic parameters and Lss and iss can be found in the References [23,25,26]. For optimising the PDM on EIS data and hence for determining optimal values for all model parameters, a geneticallyinspired algorithm, the IGOR-Pro software attached with GenCurvefit and Ellis 2 modules [24], was employed. This was done by inserting the expression for the passive film impedance, as derived from the PDM [27] into the equivalent electrical circuit (EEC) shown in Fig. 4. Here, the corrosion scale is deemed to be the outer layer in the EEC. According to the EEC, the total impedance of the system (Ztotal) can be calculated by the following equation:

3.2. Electrochemical analysis for electrodes with corrosion scale In order to investigate the influence of immersion treatment on the electrochemical behaviour of Super 13Cr, EIS measurements were conducted in a 3.5% NaCl solution, as shown in Fig. 2. Note that the “blank” specimen in the figures represents the electrode that did not undergo immersion and was tested with a fresh surface. Evidently, the diameters of capacitive arcs (Fig. 2a) and the modulus of impedance (Fig. 2b) decrease dramatically with increasing immersion time, suggesting a decrease in the corrosion resistance. In Fig. 2c, the phase angles of the blank specimen exhibit a wide and flat range that is close to −80° (under ideal circumstance the phase angle should be −90° for a purely capacitive interface), while the phase angles of the immersed specimens show “lag” feature with a rise in the phase angle at lower frequencies in contrast to the blank specimen, indicating lower resistance and higher capacitance of the immersed specimens [21]. The Point Defect Model (PDM) has been used extensively in the past to describe passivity and passivity breakdown on passive metals and alloys. Thus, the PDM was applied to obtain key parameters from the impedance data. The reactions that are envisioned to take place at the metal/barrier layer (m/bl) and barrier layer/outer layer (bl/ol) interfaces are depicted in Fig. 3. The steady-state thickness of the barrier layer (Lss) is determined by the lattice non-conservative PDM Reactions [22], which are Reaction 2 and Reaction 5 in Fig. 3. Note that the numbering of the PDM Reaction in this study is different from the model that includes the complete 7 Reaction set [23], as the later is for

(5)

Ztotal = Zbl+Zol+Rs where, −1

1 1 ⎤ + Zbl = ⎡jwCg+ ⎢ + + Z Z Z Zrandles ⎥ f W e,h ⎣ ⎦

(6)

−1

1 ⎤ Zrandles = ⎡jwCdl + ⎢ R ct ⎥ ⎦ ⎣

(7)

Fig. 3. Schematic of the reactions that take place within an n-type passive film based upon the Point Defect Model. m ^metal atom, Miχ + ^cation interstitial, V..O ^oxygen vacancy, Vm ^metal vacancy, Mδ+(aq) ^cation in outer layer/solution interface, MM ^metal cation on the metal sublattice of the barrier layer, OO ^oxide ion in anion site on the oxygen sublattice, MOχ/2 ^stoichiometric barrier layer oxide.

4

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

Table 2 Parameters from optimisation of the PDM on the EIS data of the specimens with the corrosion scale intact. The voltages for the optimisation are the open circuit potentials (vs. SHE) under different exposure time. Immersion time (h)

0 (blank)

5

20

96

α n α1 α2 α5 k1 (mol cm−2 s−1) k2 (mol cm−2 s−1) k5 (mol cm−2 s−1) CPE-Cg (S sα cm−2) CPE-α1 Cdl (F cm−2) Rct (Ω cm2) Col (F cm−2) Rol (Ω cm2) Rs (Ω cm2) Lss (nm) iss (μA cm−2)

0.7 0.6 0.09 0.57 0.71 2.13 × 10−11 1.56 × 10−14 2.48 × 10−10 1.26 × 10−4 0.88 4.90 × 10−3 2.67 × 107 5.78 × 10−4 111 1 0.76 6

0.7 0.6 0.02 0.33 0.47 2.92 × 10−11 7.61 × 10−13 7.62 × 10−10 3.03 × 10−3 0.91 3.14 × 10−2 4870 2.66 × 10−2 242 1.1 0.20 54

0.7 0.6 0.03 0.55 0.95 4.15 × 10−9 2.20 × 10−13 3.50 × 10−9 3.27 × 10−2 0.91 3.68 × 10−2 1653 5.03 × 10−2 288 1.6 0.11 91

0.7 0.6 0.58 0.06 0.37 1.89 × 10−8 1.70 × 10−12 2.70 × 10−8 6.47 × 10−1 0.83 1.05 × 10−1 147 1.77 × 10−1 320 2 0 221

−1

1 ⎤ Zol = ⎡jwCol + ⎥ ⎢ R ol ⎦ ⎣

values are very large for the long-term immersed specimens; however, as they can be regarded as being “model parameters” that mimic capacitors (not necessarily pure dielectric capacitors), the capacitance values can be used to indicate trends under specific circumstances. Finally, we note that the rate constant for the generation of metal interstitials (k1) is much larger than is the rate constant for oxygen vacancy generation (k2), as shown in Table 2. This demonstrates that the + dominant defect in the barrier layer is the metal interstitial (e.g., Fe2/3 , i Ni 2i +, and/or Cr 3i +), rather than the oxygen vacancy (V..O), which is consistent with previous findings for chromium-containing alloys [23]. Mott-Schottky measurements were carried out to ascertain the semiconductive properties of the passive film. Fig. 5 shows the MottSchottky plots of the specimens without removing the corrosion scale. As a common feature for martensitic stainless steels, n-type electronic character was observed in the blank and 5 h-immersed specimens [28]. The curves of the 20 h- and 96 h-immersed specimens are flat in contrast to the blank and 5 h-immersed specimens, as shown by the magnified plot in Fig. 5b. According to the Mott-Schottky theory, for an n-type semi-conductive film, the capacitance of the space-charge layer (Csc) is a function of the applied potential (E), which is given as [29]:

(8)

Zbl is the impedance of the barrier layer, Zol is the impedance of the outer layer, and Zrandles is the impedance due to the redox reactions taking place at the barrier layer/solution interface. Expressions for Zf, Zw, and Ze,h are given elsewhere [27] and are used directly in Eq. (6). j is the imaginary unit and ω is the angular frequency. The optimised parameters for the specimens with the corrosion scale are presented in Table 2, where α is the polarizability of the bl/ol interface, and n is the kinetic order of barrier layer dissolution. The parameters α1, α2, α5 and k1, k2, k5 are transfer coefficients and rate constants for the elementary Reactions 1, 2, and 5 in Fig. 3, respectively. For the blank specimen, the steady state thickness (Lss) of the barrier layer is 0.76 nm. With increasing immersion time, Lss gradually decreased from 0.2 nm (5 h-immersed specimen) to 0 (96 h-immersed specimen), indicating a depassivation feature after long-term immersion in the H2S-containing environment. Over that same period, the OCP decreased from −0.35 VSCE to −0.63 VSCE. One possible reason for the small thickness value of the barrier layer is that the EIS measurements were performed at OCP, without polarisation to avoid further impact to the corrosion scale and the dissolved hydrogen. Also, the current density at steady state (iss) increased dramatically with the immersion time. This is consistent with the phenomenon of resistive depassivation [25], which postulates that, if the passive current flows through the deposited, porous outer layer (the scale), the resulting IR potential drop must be subtracted from the potential at the outer layer/ solution (ol/s) interface (Vol/s) in determining the potential at the bl/ol interface (Vbl/ol), which determines the thickness of the barrier layer. Because the thickness of the barrier layer depends linearly on Vbl/ol, if Vbl/ol is sufficiently negative, the barrier layer thickness will tend to zero. The rate constants k1, k2, and k5 increased with the immersion time, suggesting a higher generation rates of the point defects and greater dissolution rate of the substrate metal through the semi-permeable barrier layer (k1) and the growth (k2) and dissolution (k5) of the barrier layer. With increasing immersion time, the charge transfer resistance (Rct) decreased from 2.67 × 107 Ω cm2 for the blank to 147 Ω cm2 after 96 h of exposure. The resistance of outer layer (Rol) shows an inverse trend; Rol progressively increased with exposure time, which is consistent with the thickening of the scale, as depicted in Fig. 1. However, the small values of Rol also reveal the poor protectiveness of the outer layer (the scale). Furthermore, the capacitance of the barrier layer (Cg), double layer (Cdl), and outer layer (Col) increased by one/two orders of magnitude for each interval of exposure time. Indeed, the capacitance

−2

Csc =

2 kT (E − Efb − ) εε0 eND e

(9)

where the dielectric constant (ε) is taken to be 15.6 [29], ε0 represents the vacuum permittivity, e is the elementary electron charge, ND is the donor density for n-type film, Efb is the flat band potential, k is the Boltzmann constant, and T is the Kelvin temperature. Hence, the M-S linear fitting method is applied to obtain the donor density (ND) of the blank specimen, and the value of ND is compared with the theoretical density to check if the linear fitting method is applicable for the specimens after immersing in such a harsh corrosion environment. The ND of the blank specimen, calculated via linear fitting method, is 5.17 × 1021 cm−3. Because the film is n-type, the dominant point defects should be cation interstitials or anion vacancies [Miχ + or V..O /V..S , depending upon whether the film is oxide (e.g., Fe2O3) or sulphide (e.g., FeS)]. Theoretically, the density of anions on the anion sublattices of Fe2O3 and FeS are 5.91 × 1022 cm−3 and 3.31 × 1022 cm−3, respectively, as calculated from the molecular weight, density, and stoichiometry. In comparison, the donor density of the blank specimen is very high, indicating an excess metal concentration of about 10%, in the case of oxide. Such a high donor density, while being physically realistic for a barrier layer, although perhaps not for a bulk chemical, raises concerns that the ND values obtained by linear fitting of M-S plots are not reliable in this case, as the specimen surfaces are covered with 5

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

Fig. 5. Mott-Schottky plots for the specimens after immersion for various time and analysed with the corrosion scale: (a) original image; (b) magnified plots for the 20 h and 96 h specimens.

of magnitude, and the zero-current potential (Ezc) decreased from −0.17 VSCE to −0.35 VSCE. Passivity breakdown is observed for the blank specimen, whereas the 5 h-immersed specimen appears to exhibit transpassive dissolution behaviour. Further increase of exposure time shows incrementally smaller changes in the current density, and Ezc kept decreasing to less noble (more active) values. The decrease in Ezc may be due to the formation of sulphides on the surface of the specimens. However, the 20 h- and 96 h-immersed specimens show active dissolution behaviour, which suggests that the specimens exhibit no passivity. According to the data summarized in Table 2, the thickness of the barrier layer for the specimens with corrosion scale is very small, and thus depassivation may occur after long-term immersion. Two experimental factors are assumed to dominate the form of the anodic polarisation curves. One is the formation of corrosion scale (outer layer) that has a deteriorative effect on passivation, due to the IR potential drop across the outer layer. Secondly, although the outer layer thickened with immersion time, the inner barrier layer which predominantly enables passivity of the stainless steel was gradually damaged by the immersion treatment. The second factor is supported by the optimised parameters based upon the PDM (Table 2).

thick corrosion products that also contribute to the capacitance of the interphase. While bulk chromic oxide is generally found to be p-type in electronic character, indicating the cation vacancy as being the dominant point defect, the defective chromic oxide barrier layer on chromium-containing alloys is invariably found to be n-type, except at high potentials, where some p-type behaviour has been reported, particularly on pure Cr [30]. The n-type behaviour found in the barrier oxide layers on Cr-containing alloys is accounted for by the presence of Fe2/ 3+ , Ni2+, and Cr3+ interstitials, which originate from the underlying alloy. From the M-S plots presented in Fig. 5, it is seen that ND increases with time after 5 h immersion. No apparent slope is observed for the 20 h- and 96 h-specimens, which indicates the film are either too defective to be regarded as a semi-conductive layer (probably best described as being semi-metals), or because of the absence of barrier layer altogether, which is consistent with the parameters given in Table 2. The increase in the defect concentration with time is also reported by Kong et al. [30] for chromium and by Ahn et al. [31] for nickel and iron, indicating that the defect structure of the barrier layer evolves slowly, as is the found in the present work. Generally, a more highly defective film will lower the corrosion resistance of the barrier layer, because of the enhanced flux of cation interstitials within the film [22]. Clearly, the immersion of Super 13Cr specimens into the H2S-saturated solution increased the density of point defects in the passive film and lowers its resistance to corrosion. The potentiodynamic polarisation curves for the specimens without removing the corrosion scale is shown in Fig. 6. After 5 h of immersion, the anodic current density dramatically increased by about two orders

3.3. Electrochemical analysis for electrodes without corrosion scale It is understandable that two processes occurred during the H2Simmersion process. One is the corrosion of the specimens which generated corrosion scale; the other one is the ingress of hydrogen into the film and substrate. Since the corrosion scale had been removed before the EIS measurements reported in this section, the difference between the immersed specimens should be due to the amount of hydrogen that has diffused into the substrate during the H2S-immersion process. Therefore, the influence of hydrogen ingress on the electrochemical behaviours of Super 13Cr in the H2S environment has been studied by the electrochemical analyses in this section. Fig. 7 shows EIS plots for specimens after H2S immersion and subsequent removal of the corrosion scale. In Fig. 7a and b, the diameters of the capacitive arcs and the modulus of impedance declined with increasing immersion time, indicating a decrease in corrosion resistance. The shape of the frequency vs. phase angle plots for the 0 h-, 5 h-, and 20 h-immersed specimens are similar, while that for the 96 himmersed specimen exhibits lower phase angles and resolved relaxations. For a detailed comparison and mechanistic interpretation, optimisation of the PDM was performed using the EEC in Fig. 4, and the parameter values obtained are listed in Table 3. With the longer immersion time, the barrier layer thinned and the current density increased (see Lss and iss values), though the variation is more moderate than that observed with the corrosion scale (Table 2). Due to the

Fig. 6. Potentiodynamic polarisation curves for the specimens after immersion for various time and analysed with the corrosion scale.

6

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

Fig. 7. EIS plots for the specimens after immersion for various time with the corrosion scale removed: (a) Nyquist plots; (b) and (c) are Bode plots vs. modulus and phase angle, respectively.

Table 3 Parameters from optimisation of the PDM on the EIS data of the specimens without the corrosion scale. The voltages for the optimisation are the open circuit potentials (vs. SHE) under different exposure time. Immersion time (h)

0 (blank)

5

20

96

α n α1 α2 α5 k1 (mol cm−2 s−1) k2 (mol cm−2 s−1) k5 (mol cm−2 s−1) CPE-Cg (S sα cm−2) CPE-α1 Cdl (F cm−2) Rct (Ω cm2) Col (F cm−2) Rol (Ω cm2) Rs (Ω cm2) Lss (nm) iss (μA cm−2)

0.7 0.6 0.09 0.57 0.71 2.13 × 10−11 1.56 × 10−14 2.48 × 10−10 1.26 × 10−4 0.88 4.90 × 10−3 2.67 × 107 5.78 × 10−4 111 1 0.76 6

0.7 0.6 0.46 0.88 0.57 6.47 × 10−12 1.51 × 10−13 2.40 × 10−9 1.36 × 10−4 0.92 9.0 × 10−3 8.3 × 106 5.16 × 10−4 288 2.2 0.30 15

0.7 0.6 0.79 0.93 0.14 4.03 × 10−12 5.22 × 10−13 8.29 × 10−9 1.96 × 10−4 0.88 9.88 × 10−2 4.33 × 105 2.24 × 10−2 71 1.5 0.25 23

0.7 0.6 0.22 0.63 0.46 2.69 × 10−10 2.92 × 10−12 4.64 × 10−8 1.17 × 10−3 0.59 7.30 × 10−2 2.25 × 105 4.63 × 10−2 760 1.5 0.14 77

7

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

Fig. 8. Mott-Schottky plots for the specimens after immersion for various time with the corrosion scale removed: (a) original image; (b) magnified plots for the 20 h and 96 h specimens.

were reported by Yao et al. [32]. In addition, the anodic branch transformed from passivity breakdown mechanism (blank) to transpassive dissolution (5 h and 20 h). More importantly, the anodic branch of the 96 h-immersed specimen exhibited active dissolution features, which means the specimen did not show passivity under the experimental condition. According to reports [13,32], hydrogen could decrease the pitting/transpassive potential of stainless steel, however, to the best of our knowledge, hydrogen-induced loss of passivity for a stainless steel has not been previously reported. This difference may be due to the much stronger hydrogen ingress into the specimen during H2S exposure than the regular electrochemical charging method. This is because the NACE Solution A (H2S-staturated) is a very aggressive hydrogen-injection medium, not only due to the abundant hydrogen source, but also because of the presence of sulphide ion, which is an effective hydrogen atom recombination poison [1]. On the other hand, as Super 13Cr is low-cost type of stainless steel, its passivity is not as good as those of 304/316/2205 stainless steels that have been previously investigated in hydrogen-charging studies, especially in a 3.5% NaCl solution (as used in this study), rather than those in the borate buffer solution, which generally provides a better passivation environment for stainless steels [15]. Nevertheless, the influence of H2S on the passivity of Super 13Cr stainless steel (Figs. 7–9) provides an extreme example in revealing the deteriorative effects that hydrogen can have on the corrosion resistance of stainless steels.

decrease in Lss, the resistance to charge transfer (Rct) reduced from 2.67 × 107 Ω cm2 (blank) to 2.25 × 105 Ω cm2 (96 h-immersion). It can also be found that, in contrast to that with the corrosion scale, the barrier layer is a little thicker when the corrosion scale is removed before the measurement. The rate constants did not vary notably except for k5, the rate constant for the dissolution of the barrier layer, which increased by two orders of magnitude over 96 h-immersion. Mott-Schottky plots for specimens after immersion and removing the corrosion scale are shown in Fig. 8. According to Fig. 8a, it is clear that 0 h-, 5 h-, and 20 h-immersed specimens exhibited n-type character; the magnified plot of the 96 h-immersed specimen is flat (Fig. 8b), which indicates the passive film is either too defective to be regarded as a semiconductor, or, the barrier layer does not exist after 96 h-immersion, which is consistent with the EIS results presented in Fig. 7. According to the slope of plots, it is seen that the donor density increased with immersion time. Since the corrosion scale on these specimens had been removed, the variation of donor density is due to the barrier layer alone or due to the ingress of hydrogen during the H2Simmersion process. By using cathodically hydrogen-charging approach, a couple of studies [14,15,32] have shown that the greater amount of hydrogen diffused into the steel surface, the higher concentration of interstitials and H (acts as an electron donor) will be in the passive film. These findings are in accordance with the M-S results in the present study. Fig. 9 shows the potentiodynamic polarisation curves for the specimens without corrosion scale. The zero-current potential dropped and the anodic current density increased with immersion time, similar to the results where the specimens were covered with corrosion scale (Fig. 6). Similar variation of corrosion potential and current density

4. Discussions With regard to the loss of passivity with increasing exposure time, we postulate that three processes may be involved: (1) thinning and destruction of the barrier layer due to the drift in the corrosion potential in the negative direction with exposure time; (2) increasing IR potential drop across the outer layer, which detracts from the applied voltage (OCP), giving rise to a decreasing voltage drop across the barrier layer; (3) cation vacancy condensation with hydrogen pressurization, which causes rupture of the barrier layer from the metal surface. 4.1. The effect of H2S-immersion For the immersed specimens with the corrosion scale intact, the passive film can be identified as a bi-layer type [23]. The corrosion scale is the outer layer, but it is the inner layer (barrier layer) that is generally responsible for the corrosion resistance of the metal substrate. Indeed, the thickened outer layer did not significantly enhance corrosion protection (see the Rol values in Table 2), but instead contributed to a loss of passivity, as indicated by the values for Lss and Iss with increasing exposure time. According to the PDM, it is the interstitial generation reaction

Fig. 9. Potentiodynamic polarisation curves for the specimens after immersion for various time with the corrosion scale removed.

8

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

Fig. 10. Schematic of the passive behaviour of the H2S-immersed specimen without the corrosion scale: (a) no immersion; (b) short-term immersion; (c) long-term immersion.

across the outer layer, which subtracts from the control voltage outside of the outer layer to yield the voltage at the bl/s (ol) interface, which drives film growth [25]. If the specimen is immersed for a sufficiently long time, the barrier layer should hardly exist, as explained previously, because of the increasing potential drop across the outer layer results in disappearance of the barrier layer, and only leaves the low-protective outer layer on the surface, as proven by the 20 h- and 96 h-immersed specimens that did not show semi-conductivity and passivity.

(Reaction 1 in Fig. 3) that is responsible for the majority of the passive current density (passive corrosion rate), because of the dominance of metal interstitials as the point defect in the barrier layer, while the generation of oxygen vacancies at the m/bl interface (Reaction 2 in Fig. 3) is responsible for barrier layer growth. On Cr-containing alloys, such as Super 13Cr, the electrochemically most active element in the alloy (Cr) forms the barrier layer (Cr2+xO3-y), but the layer contains a high concentration of Fe due to the interstitials being transmitted through the phase. Thus, while the barrier layer is primarily a defective chromic oxide, the outer layer comprises primarily iron oxides, oxyhydroxides, and hydroxides, with small Cr component that arises from the dissolution of the barrier layer and the transmission of Cr3+ interstitials. The barrier layer reaches a steady-state when the growth rate of barrier layer at the m/bl interface equals the dissolution rate of the barrier layer at the bl/ol interface. However, it is important to note that the outer layer may grow ad-infinitum, provided that conditions in the external environment remain conducive to the hydrolysis of transmitted metal interstitials and chromium ions generated by dissolution of the barrier layer and precipitation of the appropriate oxides, hydroxides, and oxyhydroxides. Once the specimen is scaled, the former steady state would be changed due to the existence of the scale, which may reduce the dissolution rate at the bl/s interface due to the blocking effect of the corrosion scale on dissolution reaction (Reaction 5 in Fig. 3). Accordingly, the barrier layer will reach another steady state, or, if the porosity of the outer layer decreased continuously, the barrier layer would increase in thickness continuously. Inhibition of Reaction 5 will result in a thickening of the barrier layer, not in the observed thinning. We note that the open circuit potential shifts in the negative direction by about 0.3 V over the exposure period of 96 h. This will have the effect of causing the barrier layer to thin, and if the thinning is sufficiently large, it may cause the barrier layer to disappear altogether. The effect of OCP decrease, coupled with the IR potential drop across the outer layer (scale), are postulated to result in the observed depassivation in the presence of the corrosion scale. To explain the effect of IR potential drop on the thickness of barrier layer, the third generation of PDM (PDM-III) [23,25] is employed. The first Point Defect Model [23] was proposed in the early 1980s, i.e., PDM-I, but it did not include metal interstitials and did not invoke dissolution of the barrier layer. These features were included in PDM-II and the porous outer layer was included in PDM-III. The advantage of PDM-III is that it considers the influence of resistive outer layer on the impedance of the interface and hence on the corrosion resistance of the system. According to the PDM-III [25], for a given applied voltage, the thickness of the barrier layer will decrease as the resistance of the outer layer increases. This is a direct result of the potential drop that develops

4.2. The effect of hydrogen ingress during H2S-immersion process Regarding the corrosion behaviour of the specimens that have had the corrosion scale removed, as discussed in Section 3.3, the cause of the different performance of the specimens is determined by the hydrogen that has entered the steel matrix during immersion. When the electrode was immersed into the H2S-saturated solution, the following hydrogen-related reactions should take place on the electrode/solution interface [33,34]:

2H+ + 2e− → Hads +Hads

(10)

Hads +Hads → H2 ↑

(11)

Hads ⇔ Habs → ingress into the metal

(12)

Based on these factors, a mechanism for the influence of hydrogen on the passive behaviour of stainless steel matrix is assumed to be as shown in Fig. 10. With medium immersion time (Fig. 10b), as there is small amount of hydrogen diffused into the metal, there will be an accumulation of hydrogen in various traps in the steel matrix, due to the existence of the defects inside the metal, such as vacancies, dislocations, grain boundaries, inclusions, precipitates, etc. [35]. The passive film may generate before the formation of very small hydrogen blisters. That hydrogen can induce blisters at the metal/film interface has been observed by an in-situ TEM study [36] and is an integral part of the PDM for passivity breakdown. The hydrogen blisters inhibited growth of the barrier layer at localized m/bl interface, which narrows the passive region and thus lowers the overall corrosion resistance. The deteriorative effect of diffused hydrogen on the corrosion resistance increased with extending immersion time, but the effect was not apparent until the amount of hydrogen within the substrate was extremely high after long-term immersion, as is shown in Fig. 10c. Since more hydrogen existed at the metal surface over longer H2S-immersion time, once there generated a small area of passive film on the metal surface, it should be accompanied by the formation of hydrogen blisters between the metal and the film in response to cation vacancy condensation. According to the PDM [22], Reaction 2 in Fig. 3 is 9

Corrosion Science xxx (xxxx) xxx–xxx

X. Lei et al.

interrupted due to the existence of hydrogen blisters, and the barrier layer detaches from the metal. Simultaneously, the barrier layer dissolves by Reaction 5 at the bl/ol interface, such that the barrier layer thins. At some point, the barrier layer “cap” ruptures, and passivity is lost. This process is aided by hydrogen atom recombining within the blister to cause blister pressurization. Therefore, it can be regarded that a layer of hydrogen blister that adsorbed on the electrode surface and blocked the generation/growth of the passive film (Reaction 2 in Fig. 3), as is depicted in Fig. 10c. Consequently, the specimen will not show passivity due to the absence of passive film, and this is what is exhibited by the polarisation curve of the 96 h-immersed specimen in Fig. 9.

[6] P. Wang, Z. Lv, S. Zheng, Y. Qi, J. Wang, Y. Zheng, Tensile and impact properties of X70 pipeline steel exposed to wet H2S environments, Int. J. Hydrogen Energy 40 (2015) 11514–11521. [7] J. Ding, L. Zhang, M. Lu, J. Wang, Z. Wen, W. Hao, The electrochemical behaviour of 316L austenitic stainless steel in Cl− containing environment under different H2S partial pressures, Appl. Surf. Sci. 289 (2014) 33–41. [8] Z.Y. Liu, C.F. Dong, X.G. Li, Q. Zhi, Y.F. Cheng, Stress corrosion cracking of 2205 duplex stainless steel in H2S–CO2 environment, J. Mater. Sci. 44 (2009) 4228–4234. [9] A. Davoodi, M. Pakshir, M. Babaiee, G.R. Ebrahimi, A comparative H2S corrosion study of 304L and 316L stainless steels in acidic media, Corros. Sci. 53 (2011) 399–408. [10] X. Cheng, H. Ma, S. Chen, L. Niu, S. Lei, R. Yu, Z. Yao, Electrochemical behaviour of chromium in acid solutions with H2S, Corros. Sci. 41 (1999) 773–788. [11] J.-S. Lee, Y. Kitagawa, T. Nakanishi, Y. Hasegawa, K. Fushimi, Effect of hydrogen sulfide ions on the passive behavior of type 316L stainless steel, J. Electrochem. Soc. 162 (2015) C685–C692. [12] W. He, O. Ø. Knudsen, S. Diplas, Corrosion of stainless steel 316L in simulated formation water environment with CO2–H2S–Cl−, Corros. Sci. 51 (2009) 2811–2819. [13] Q. Yang, J.L. Luo, The effects of hydrogen on the breakdown of passive films formed on type 304 stainless steel, Thin Solid Films 371 (2000) 132–139. [14] Q. Yang, J.L. Luo, The hydrogen-enhanced effects of chloride ions on the passivity of type 304 stainless steel, Electrochim. Acta 45 (2000) 3927–3937. [15] S. Ningshen, U.K. Mudali, Hydrogen effects on pitting corrosion and semiconducting properties of nitrogen-containing type 316L stainless steel, Electrochim. Acta 54 (2009) 6374–6382. [16] L.Q. Guo, Y. Bai, B.Z. Xu, W. Pan, J.X. Li, L.J. Qiao, Effect of hydrogen on pitting susceptibility of 2507 duplex stainless steel, Corros. Sci. 70 (2013) 140–144. [17] NACE Standard TM 0177-2005, Laboratory Testing of Metals for Resistance to Sulfide Stress Cracking and Stress Corrosion Cracking in H2S Environments, NACE, Houston, TX, 2005. [18] P. Marcus, E. Protopopoff, Potential-pH diagrams for sulfur and oxygen adsorbed on chromium in water, J. Electrochem. Soc. 144 (1997) 1586–1590. [19] P. Marcus, E. Protopopoff, Potential-pH diagrams for adsorbed species application to sulfur adsorbed on iron in water at 25° and 300 °C, J. Electrochem. Soc. 137 (1990) 2709–2712. [20] P. Marcus, E. Protopopoff, Potential pH diagrams for sulfur and oxygen adsorbed on nickel in water at 25 and 300 °C, J. Electrochem. Soc. 140 (1993) 1571–1575. [21] J.Q. Zhang, C.N. Cao, Study and evaluation on organic coatings by electrochemical impedance spectroscopy, Corros. Prot. 19 (1998) 99–104 (in Chinese). [22] D.D. Macdonald, The point defect model for the passive state, J. Electrochem. Soc. 139 (1992) 3434–3449. [23] D.D. Macdonald, The history of the point defect model for the passive state: a brief review of film growth aspects, Electrochim. Acta 56 (2011) 1761–1772. [24] S. Sharifi-Asl, M.L. Taylor, Z. Lu, G.R. Engelhardt, B. Kursten, D.D. Macdonald, Modeling of the electrochemical impedance spectroscopic behavior of passive iron using a genetic algorithm approach, Electrochim. Acta 102 (2013) 161–173. [25] D.D. Macdonald, G. Englehardt, The point defect model for bi-layer passive films, ECS Trans. 28 (2010) 123–144. [26] P. Lu, B. Kursten, D.D. Macdonald, Deconvolution of the partial anodic and cathodic processes during the corrosion of carbon steel in concrete pore solution under simulated anoxic conditions, Electrochim. Acta 143 (2014) 312–323. [27] S. Sharifi-Asl, D.D. Macdonald, A. Almarzooqi, B. Kursten, G.R. Engelhardt, A comprehensive electrochemical impedance spectroscopic study of passive carbon steel in concrete pore water, J. Electrochem. Soc. 160 (2013) C316–C325. [28] Y. Zhang, M. Urquidi-Macdonald, G.R. Engelhardt, D.D. Macdonald, Development of localized corrosion damage on low pressure turbine disks and blades: I. Passivity, Electrochim. Acta 69 (2012) 1–11. [29] Y.F. Cheng, J.L. Luo, Electronic structure and pitting susceptibility of passive film on carbon steel, Electrochim. Acta 44 (1999) 2947–2957. [30] D. Kong, S. Chen, C. Wang, W. Yang, A study of the passive films on chromium by capacitance measurement, Corros. Sci. 45 (2003) 747–758. [31] S. Ahn, H. Kwon, D.D. Macdonald, Role of chloride ion in passivity breakdown on iron and nickel, J. Electrochem. Soc. 152 (2005) B482–B490. [32] J. Yao, C. Dong, C. Man, K. Xiao, X. Li, The electrochemical behavior and characteristics of passive film on 2205 duplex stainless steel under various hydrogen charging conditions, Corrosion 72 (2015) 42–50. [33] Y. Qi, H. Luo, S. Zheng, C. Chen, D. Wang, Effect of immersion time on the hydrogen content and tensile properties of A350LF2 steel exposed to hydrogen sulphide environments, Corros. Sci. 69 (2013) 164–174. [34] S. Zheng, Y. Qi, C. Chen, S. Li, Effect of hydrogen and inclusions on the tensile properties and fracture behaviour of A350LF2 steels after exposure to wet H2S environments, Corros. Sci. 60 (2012) 59–68. [35] W.A. Counts, C. Wolverton, R. Gibala, First-principles energetics of hydrogen traps in α-Fe: point defects, Acta Mater. 58 (2010) 4730–4741. [36] D.G. Xie, Z.J. Wang, J. Sun, J. Li, E. Ma, Z.W. Shan, In situ study of the initiation of hydrogen bubbles at the aluminium metal/oxide interface, Nat. Mater. 14 (2015) 899–903.

5. Conclusions The electrochemical behaviour of Super 13Cr stainless steel was analysed in a 3.5% NaCl solution after immersion in a H2S-saturated solution. The results reveal that with increasing immersion time, the corrosion scale thickened, accompanied by increasing contents of Feand Ni- sulphides in the scale. The passive film showed n-type semiconductivity, and the donor density increased with immersion time. After 96 h of immersion, the steel with and without corrosion scale showed no semiconductor character and the polarisation curve exhibited no passive region, indicating the loss of passivity for Super 13Cr. In the presence of corrosion scale, the thickness of the barrier layer on Super 13Cr declined with increasing immersion time, which results from the IR potential drop across the outer layer that reduced the effective film-formation potential across the barrier layer. The barrier layer is a little thicker when the corrosion scale is removed before the electrochemical measurement. Moreover, the steel substrate exhibited decreased corrosion resistance with increasing immersion time, possibly due to the hydrogen ingress induced by H2S-exposure. The dissolved hydrogen inhibited the generation and growth of passive film, possibly aided by pressurization of cation vacancy voids at the m/ bl interface, and resulted in the depassivation character of Super 13Cr over sufficient immersion in H2S solution. Acknowledgements This work is supported by the National Natural Science Foundation of China (21402113, 51641409), the China Postdoctoral Science Foundation (No. 2015M581349), and the Fundamental Research Funds for the Central Universities (GK201503033, 3102017OQD104). Furthermore, the authors would like to thank Liyun Pan, Pengfei Huang, and Qingwei Ma for the experiments that involved H2S gas. Xiaowei Lei would like to acknowledge Jizheng Yao for helpful discussions. References [1] R.D. Kane, M.S. Cayard, Roles of H2S in the behavior of engineering alloys: a review of literature and experience, Corrosion-NACE Proceedings (1998) (pp. Paper No. 274). [2] P. Bai, S. Zheng, C. Chen, Electrochemical characteristics of the early corrosion stages of API X52 steel exposed to H2S environments, Mater. Chem. Phys. 149–150 (2015) 295–301. [3] F. Mao, C. Dong, D.D. Macdonald, Effect of octadecylamine on the corrosion behavior of Type 316SS in acetate buffer, Corros. Sci. 98 (2015) 192–200. [4] L.W. Tsay, M.Y. Chi, Y.F. Wu, J.K. Wu, D.Y. Lin, Hydrogen embrittlement susceptibility and permeability of two ultra-high strength steels, Corros. Sci. 48 (2006) 1926–1938. [5] W.K. Kim, S.U. Koh, B.Y. Yang, K.Y. Kim, Effect of environmental and metallurgical factors on hydrogen induced cracking of HSLA steels, Corros. Sci. 50 (2008) 3336–3342.

10