Journal of Materials Science 【-逻*辑*与-】amp; Technology, 2020, 49(0): 166-178 doi: 10.1016/j.jmst.2020.01.016

Research Article

## Near-neutral pH corrosion of mill-scaled X-65 pipeline steel with paint primer

Shidong Wang, Lyndon Lambornb, Karina Chevilc, Erwin Gamboac, Weixing Chen,a,*

a Department of Chemical and Materials Engineering, University of Alberta, Edmonton, T6G 2G6, Canada

b Enbridge Pipelines Inc., Edmonton, T5J 3N7, Canada

c TC Energy Corporation, Calgary, T2P 5H1, Canada

Corresponding authors: * E-mail address:weixing@ualberta.ca(W. Chen).

Received: 2019-06-9   Revised: 2019-09-17   Accepted: 2019-10-4   Online: 2020-07-15

Abstract

The corrosion behaviour of mill-scaled X65 pipeline steel with and without a primer layer was studied in a simulated near-neutral pH soil solution. Results revealed a three-stage corrosion process of the mill-scaled pipeline steel surface. The first stage included an initial preferential dissolution of goethite (α-FeOOH) and lepidocrocite (γ-FeOOH) in mill scale. The second stage was marked by enhanced localized corrosion and pit-formation because of either galvanic corrosion or acidic dissolution in areas enclosed by mill scale. The final stage was general corrosion after the mill scale flaked off the steel surface. When the primer layer was applied, localized corrosion was significantly enhanced on the steel surface and persisted for an extended period as compared to the mill-scaled condition. The precipitation of siderite (FeCO3) was observed at flawed locations of mill scale, although the bulk chemistry is not favorable for its formation on the steel surface free of mill scale. The local precipitation of siderite formed a capped mill scale enclosure where localized corrosion can be further enhanced.

Keywords： Pipeline steel ; Primer paint ; Mill scale ; Localized corrosion ; Siderite precipitation

Export EndNote| Ris| Bibtex

Shidong Wang, Lyndon Lamborn, Karina Chevil, Erwin Gamboa, Weixing Chen. Near-neutral pH corrosion of mill-scaled X-65 pipeline steel with paint primer. Journal of Materials Science & Technology[J], 2020, 49(0): 166-178 doi:10.1016/j.jmst.2020.01.016

## 1. Introduction

The occurrence of near-neutral pH stress corrosion cracking (SCC) of buried pipeline steels starts when the protective coatings on pipeline steels are damaged or become disbonded, and the steel surface comes in direct contact with groundwater [1]. A primer layer, which allows for optimal adhesion between the protective coatings and the steel surface, was commonly applied to the as-fabricated steel surface with mill scale before the application of protective coatings [2,3]. The primer layer commonly applied to pipeline steel surfaces is a dilute solution of an adhesive in an organic solvent, generally consisting of a blend of solvents, petroleum resin, butyl rubber compound and additive agent [4,5]. The liquid adhesive (primer) is sprayed onto the pipe surface, producing a dried layer with a thickness of 1.5-50 μm [6]. It was found that the primer layer would likely remain on the mill-scaled surface, even though the protective coating was damaged or disbonded [5,7]. This scenario is quite common for operating buried pipelines, especially for tape-coated pipelines [4]. Near-neutral pH SCC occurrences in Canadian pipeline systems were mainly found in tape-coated pipeline steels [1]. Although many laboratory investigations have been made [1,[7], [8], [9], [10], [11], [12]], the mechanism on near-neutral pH SCC crack initiation and early stage crack growth remains to be determined [1]

Mill scale is typically composed of different iron oxide layers formed on the surface of steel during the high temperature rolling of steel plates [5,13]. It generally consists of hematite (Fe2O3), magnetite (Fe3O4) and wüstite (FeO), depending on the steel type, rolling temperature and cooling rate [5,13]. It is widely accepted that corrosion of steels can be accelerated due to the galvanic coupling established between mill scale (cathode) and the steel (anode), providing that the mill scale was porous or cracked in nature [[7], [8], [9],[14], [15], [16]]. By using an X52 pipeline steel, Shirband et al. found that the penetration of solution through cracked mill scale to the steel substrate occurred readily and caused localized corrosion under the mill scale in simulated near-neutral pH soil solution [7]. Qin et al. reported that for an X65 pipeline steel, porous mill scale caused local separation of anodic and cathodic sites, which resulted in continuous acceleration of localized corrosion [8]. Some researchers believed that mill scale or oxide layers could provide protection to the steel surface, if they were compact, adherent and well-distributed [17,18]. Zhang et al. reported that in a simulated near-neutral pH solution containing Cl- and SO42-, the corrosion rate reduced with time due to the increasing thickening and densification of FeOOH layer on the steel surface [17]. Furthermore, the degradation of mill scale or oxide layers was inevitable when they were exposed to aggressive solutions [7,8,19]. In anaerobic soil solutions, reductive dissolution of pre-formed Fe(III) corrosion products could become coupled to the dissolution of the steel surface exposed at the localized corrosion areas, resulting in a gradual removal of Fe(III) corrosion products [19]. Qin et al. implied that porous mill scale started to dissolve within 2 d in simulated groundwater [8]. Despite of long-term exposure under aggressive mechanical loading, Shirband et al. found that crack initiation under the mill-scaled surface was very difficult since the mill scale on the sample surface was very prone to flake off during corrosion immersion as a result of galvanic corrosion [7].

Apart from mill scale, other factors that cause the corrosion of buried pipeline steels in near-neutral pH environment were extensively studied recently [1,[7], [8], [9], [10],15,[20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31]]. It revealed that inclusions and pre-existing defects on steel surface, persistent slip bands induced by mechanical loading, high tensile residual stress, coating damage or disbondment, fluctuations of cathodic potential, alternating current and microorganisms were the potential factors that contribute to the localized corrosion and SCC of buried pipeline steels [1,[7], [8], [9], [10],15,[20], [21], [22], [23], [24], [25]]. Corrosion of an X80 pipeline steel in near-neutral pH soil solution under disbonded coating was investigated by Yan et al., it was found that the active dissolution rate of steel under disbonded coating decreased toward the bottom of a disbonded holiday [25]. With the same steel and experimental setup, Wu et al. reported that the occurrence of localized corrosion was quite sensitive to the sulfate-reducing bacteria (SRB) activities [21]. In addition, alternating current interference could decrease the cathodic protection (CP) effectiveness and increase the corrosion rate of pipeline steel [20,32]. The past corrosion or SCC investigations have been performed on specimens with either a polished surface or mill-scaled surface without a primer layer applied [7,10]. To the authors’ knowledge, no attention has been paid to the effect of a primer-coated mill scale on localized corrosion when analyzing crack initiation of pipeline steels.

The primer layer, essentially permeable to water, is unable to protect the substrate from corrosion as well as protective coatings [5,33]. Although it is applied to achieve better coating adhesion, it is suspected that a primer layer once applied could preserve the mill scale on the steel surface for an extended period. The attachment of mill scale to the steel surface promotes localized corrosion and pit-formation, as such, it is hypothesized that the application of a primer layer plays a key role in the initiation of near-neutral pH SCC. This hypothesis is reinforced by the fact that the modern fusion bonded epoxy (FBE) coating is immune to SCC in the field; FBE is applied to the steel surface after sandblasting (mill scale being removed) [4]. The removal of mill scale from pipe surface could be a key factor that prevents the occurrence of SCC on pipeline steels with FBE, besides the benefit of cathodic current penetration of FBE coating as generally believed. The goal of this investigation is to validate the hypothesis of primer-assisted localized corrosion and pit-formation. SCC crack initiation from localized corrosion on mill-scaled surface with and without the influence of primer coating will be communicated in the near future.

## 2. Experimental procedures

### 2.1. Material preparation

This study was conducted on mill-scaled X65 pipeline steel with a wall thickness of 14.2 mm and an outer diameter of 914 mm. The chemical composition (wt%) of the steel was as follows: C: 0.06, Mn: 1.43, P: 0.011, S: 0.005, Si: 0.31, Ni: 0.063, Cr: 0.066, Mo: 0.02, Cu: 0.21, V: 0.034, Nb: 0.07, Ti: 0.014, Al: 0.025, Fe: balance. Samples with a dimension of 20 mm × 15 mm × 10 mm were cut from the pipeline steel without damaging the mill scale on the surface. For comparison, samples with a 1 μm surface finish were prepared by polishing to remove any of the mill scale. All mill-scaled and polished samples were degreased and rinsed with ethanol before any testing or characterization.

An commercial oil-based ALLPRO® paint primer (vinyl toluene alkyd copolymer) was applied by a spray method on the surface of mill-scaled samples. The chemical composition of the primer paint is listed in Table 1. The primer was applied by nozzle-spraying for ~ 5 s at a speed of ~ 2 ml/min and a distance of ~ 30 cm from the sample surface to reach a cured thickness of about ~ 10 μm; the primer was allowed to air cure at room temperature for 2 d.

Table 1   Chemical composition of the primer paint used in the current investigation (wt%).

IngredientConcentrationCAS number
Acetone24.2367-64-1
Propane13.8674-98-6
N-butane8.14106-97-8
VM&P Naphtha5.5864742-89-8
TiO25.1313463-67-7
Toluene5.07108-88-3
Talc (Mg3Si4O10(OH)2)4.4914807-96-6
Xylene (mix)4.081330-20-7
Ethyl alcohol3.9864-17-5
N-butyl acetate2.78123-86-4
Mineral spirits2.5564742-47-8
Isobutyl acetate1.62110-19-0
Isopropyl alcohol1.0267-63-0

The oxide scales on the sample surface were examined by using Rigaku Geigerflex X-ray Diffractometer (XRD) equipped with a cobalt tube (wavelength, λKα1 = 1.78900 Å and λKα2 = 1.79283 Å), graphite monochromator, and scintillation detector. The XRD data was recorded at 38 kV and 38 mA in the 2θ range of 10°-90° at a step of 0.02° and a scanning speed of 2°/min. The surface and polished cross-sectional morphologies of samples were observed by using scanning electron microscopy (SEM; Sigma 300 V P, Zeiss) equipped with a Bruker energy dispersive X-ray spectroscopy (EDS) system. All sample surfaces were observed without any conductive coatings to avoid any impact on brightness or contrast during backscattered electron (BSE) imaging. Raman spectroscopy (Thermo Fisher Scientific DXR2, USA) was selected to analyze the initial composition of the mill scale. The spectra were excited with 532 nm radiation from a solid-state laser. The spectra were collected using a Thermo Fisher confocal microscope with a maximum magnification of ×1000 to select the laser beam focal location. The laser power was 1 mW, and each data collection time was 100 s.

### 2.2. Testing environment

The test solution in this study was a simulated soil solution, namely C2 solution, with the chemistry shown in Table 2. A stable pH of ~ 6.3 was established by purging with 5 % CO2 balanced with nitrogen for two days prior to the test and continuously during the test [34,35]. The solution was not refreshed during the corrosion tests, which is consistent with the fact that the corrosion was caused by solution trapped within the disbonded coatings in the field. Previous reports showed that a dilute bicarbonate solution with a neutral pH in the range of 5.5-7.5 is commonly found during pipeline excavations where near-neutral pH SCC has been detected [1].

Table 2   Chemical composition of the C2 solution used in the current investigation (g/l).

KClNaHCO3CaCl2·2H2OMgSO4·7H2OCaCO3
0.00350.01950.02550.02740.0606

### 2.3. Mass loss and pH measurement

Mass loss was monitored during the entire course of corrosion exposure with regular measurements. Samples with a dimension of 20 mm × 15 mm × 10 mm were coated with epoxy, leaving only one surface exposed to the environment (Fig. 1). The immersion test was conducted in the C2 solution at 20 °C for up to 90 d. The ratio of sample surface area (cm2) to the volume of C2 solution (ml) was fixed at 1/100. The samples were withdrawn from the immersion solution at regular time intervals for mass measurements using an analytical scale with a precision of 0.01 mg after rinsing with ethanol and blow drying with cold air. At least three parallel tests were carried out for each condition in the same testing solution. The mass loss was measured without removal of mill scale, primer layer and corrosion products. The change of sample mass would reflect the mass gain of corrosion products and the mass loss due to iron dissolution and the exfoliation of surface layers. Separation of each corrosion processes is not possible because of the unknown mass of each component prior to corrosion exposure. The pH of the testing solution was measured prior to and after 30, 60 and 90 d of immersion, respectively, using XL60 Dual Channel pH Meter that was calibrated with buffer solutions of pH 4 and 7 prior to each measurement. At least three measurements were taken for each condition. Samples after corrosion were analyzed using XRD and SEM. The corrosion products of immersed samples were characterized by using Raman spectroscopy. The distribution of pit depth was statistically analyzed by measuring pit-depths on the cross-section of a sample that was sequentially ground with each grinding step of ~ 0.2 mm.

### Fig. 1.

Fig. 1.   Schematic illustration of the samples used in the current investigation for (a) mill-scaled and (b) primer pre-coated samples.

### 2.4. Electrochemical testing

Electrochemical measurements were performed in the C2 solution at 20 °C using a Gamry Reference 600 electrochemistry workstation. A classical three-electrode cell was used with Pt as a counter electrode, saturated calomel electrode (SCE) as a reference electrode, and the samples mounted using epoxy resin with an exposed area of 1 cm2 as a working electrode. The ratio of sample surface area (cm2) to the volume of C2 solution (ml) was 1/1000. The solution was not refreshed, but the solution was continuously purged with 5 % CO2-95 % N2 during the 90-day test. The corrosion potential (Ecorr) and electrochemical impedance spectroscopy (EIS) of the samples were monitored regularly. The corrosion potential was recorded after a 2 h stabilization period to allow the solution to achieve a relatively stable anaerobic condition. EIS measurements were performed at the open circuit potential (OCP), and the EIS scanning frequency ranged from 100 kHz to 10 mHz with an AC amplitude of 10 mV (peak-to-zero). The EIS spectra were fitted using the ZSimpWin 3.10 software.

## 3. Results

### 3.1. Microstructural characterization

Fig. 2 shows microstructural characterization of samples before test. Surface morphology of mill-scaled and primer pre-coated samples are shown in Fig. 2(a) and (b). The top layer of mill scale appears to comprise spherical particle-like oxides (Fig. 2(a)). For primer pre-coated sample, a layer of primer paint evenly covers the surface, and no spherical particle-like oxides are visible (Fig. 2(b)). Cross-sectional surface morphologies are shown in Fig. 2(c) and (d). The mill-scaled sample consists of two oxide layers. The outer layer is porous, while the inner layer is relatively compact and composed of a mixture of cracked oxides with different contrasts, as marked in Fig. 2(c). The non-uniform contrast reflects the change of oxide chemistries. Raman spectra from different regions on the cross-sectional surface of mill-scaled samples are shown in Fig. 2(e). According to the literature [17], the peaks at about 220, 250, 301, 380, 526 and 651 cm-1 indicate the presence of lepidocrocite (γ-FeOOH) in the outer layer. For dark regions of the inner layer, the peaks at about 250, 301, 388, 482 and 1003 cm-1 can be assigned to goethite (α-FeOOH), besides the identical peaks associated with γ-FeOOH [36,37]. A single peak at about 282 cm-1 could not be assigned to α-FeOOH [36]. Previous studies indicated that this peak was probably due to interference from impurities affecting the crystallinity of α-FeOOH [38]. As for the bright inner layer, the peaks at about 308, 542 and 669 cm-1 indicate the presence of magnetite (Fe3O4) [37]. As for primer pre-coated sample, a layer of primer paint with a thickness of ~ 10 μm can be observed on the top of the two oxide layers.

### Fig. 2.

Fig. 2.   Microstructural characterization of samples before test: (a, b) secondary electron imaging of the surfaces of mill-scaled and primer pre-coated samples, respectively; (c, d) backscattered electron imaging of the cross-sectional surfaces of mill-scaled and primer pre-coated samples, respectively; (e) Raman spectra of the outer and inner layers of mill scale.

### 3.2. Mass loss and pH variation during immersion test

Fig. 3 shows mass loss of samples after immersion for up to 90 d. The corrosion process of samples can be divided into two stages based on their distinct slopes. The mass loss data for the two different stages are fitted by the following equation which is widely used for quantifying corrosion kinetics of steels [[39], [40], [41]]:

logΔW = logA + mlogt

where ΔW is the mass loss (mg cm-2), A and m are constants, t is the corrosion time. The slope m reflects the characteristic of the corrosion kinetics [42], with m = 1 for constant mass loss rate, m > 1 for accelerated mass loss rate, m < 1 for decelerated mass loss rate, m < 0 for mass gain. For mill-scaled samples, the value of m is ~ 1 during the first 5-days, indicating a constant mass loss rate. Afterwards, the mass loss rate decreased with time (m = 0.40). For primer pre-coated samples, the mass loss rate decreases with increasing time within the first 45-days of immersion (m = 0.64). Interestingly, no further mass loss, instead, a slight mass gain, is observed after exposure for 45 days (m = -0.09). The pH variation of the testing solution reveals that its pH remains almost unchanged (~ 6.29) during the entire course of immersion (Fig. 4). It indicates that the testing solution is maintained as a buffer solution due to continuous purging of the solution with 5 % CO2/N2.

### Fig. 3.

Fig. 3.   Mass loss of the mill-scaled and primer pre-coated samples immersed in C2 solution for up to 90 d. Standard deviation of the measured data was provided. The symbols and lines denote the experimental and fitted data, respectively.

### Fig. 4.

Fig. 4.   pH variation of the testing solution for up to 90 d.

### 3.3. XRD analysis of samples after immersion test

Fig. 5 shows XRD patterns collected from the surface of samples after exposure to the C2 solution for various lengths of time. Prior to the corrosion test, the main composition of the mill-scaled sample is γ-FeOOH, α-FeOOH and Fe3O4. For primer pre-coated samples, rutile (TiO2) and talc (Mg3Si4O10(OH)2) which are the composition of primer (Table 1), are detected besides all the phases observed on mill-scaled samples. For both samples during the immersion process, the strong peaks for γ-FeOOH, α-FeOOH and Fe3O4 weaken or even disappear with time. For mill-scaled samples, the peaks for γ-FeOOH and α-FeOOH are not detected after respectively 5 and 30 d of immersion, while only γ-FeOOH peaks for primer pre-coated samples disappear until after 90 d of immersion. Additionally, peaks for siderite (FeCO3) on both sample surfaces could be detected after 30 d of immersion. For polished samples, no other peaks except for that of Fe are detected after immersion test.

### Fig. 5.

Fig. 5.   XRD patterns of samples after exposure to C2 solution for up to 90 d: (a) mill-scaled, (b) primer pre-coated and (c) polished samples. JCPDS standard cards are provided, and the strongest lines are marked correspondingly (Fe, No. 65-4899; γ-FeOOH, No. 74-1877; α-FeOOH, No. 08-0097; Fe3O4, No. 79-0419; FeCO3, No. 29-0696; TiO2, No. 87-0710; Mg3Si4O10(OH)2, No. 83-1768).

### 3.4. Surface morphology variation during immersion test

Fig. 6 shows surface morphologies of mill-scaled samples exposed to the C2 solution for various lengths of time. Prior to the corrosion test, the surface is relatively rough, with a layer of spherical particle-like oxides (Fig. 6(a)). In some regions, the rough oxide layer has exfoliated, revealing the underlying oxide layer (Fig. 7(a)). After 5 d of immersion, the outer oxide layer completely disappears, while cracked inner-layer oxides with different contrasts can be observed (Fig. 6(b)). The majority of the inner-layer oxides appear bright, while some ribbon-like regions are dark. Meanwhile, the brightest regions correspond to the exposed metal surface which is found in regions where mill scale has exfoliated. The main composition of the bright, dark and brightest regions on sample surface can be verified by their corresponding EDS results (Fig. 7(b-d)). When the immersion time is 30 d, few dark inner-layer oxides are visible. Instead, the area of exposed metal surfaces increases appreciably, while some regions of localized corrosion areas appear on the exposed metal surfaces. These localized corrosion areas appear to be predominately shallow round pits. After 90 d of immersion, only some isolated bright inner-layer oxides are left and surrounded by exposed metal surfaces. Interestingly, the exposed metal surface exhibits a uniform appearance of general corrosion even within the previously discussed pits, and no additional distinct localized corrosion areas are visible (Fig. 6(d) and (e)). High-magnification imaging shows some cubic crystals that have deposited at the mouth of a flaw in the remaining mill scale (Fig. 6(f)). Based on Raman results, the composition of the cubic crystals is FeCO3, and peaks for Fe3O4 from surrounding mill scale are also detected (Fig. 6(f)) [37,43].

### Fig. 6.

Fig. 6.   Backscattered electron imaging to surfaces of mill-scaled samples exposed to C2 solution for (a) 0 d, (b) 5 d, (c) 30 d and (d) 90 d; (e) secondary electron observation of the image (d); (f) high-magnification imaging of the marked area “f” in image (e). The inset in image (f) is a Raman spectrum from the corresponding marked area in image (f).

### Fig. 7.

Fig. 7.   EDS analysis results obtained within the marked areas in Fig. 6. Images (a), (b), (c) and (d) are EDS results of the corresponding marked areas “a”, “b”, “c” and “d” in Fig. 6 (a, b), respectively. Note that C element was removed from EDS results, as it might come from the contamination.

As for the primer pre-coated samples, a layer of primer paint is uniformly distributed on the surface prior to immersion, as shown in Fig. 8(a). After 5 d of immersion, the primer layer is exfoliated at localized areas where mill scale underneath can be observed (Fig. 8(b)). Also, exposed metal surface appears at locations where mill scale has exfoliated. After 30 d of immersion, non-primer areas expand with more cracked mill scale being exposed. At the same time, localized corrosion starts to occur at non-primer coated areas (Fig. 8(c)). When the immersion time is 90 d, corrosion of underlying metal surface can be observed at every location where the primer coating has exfoliated (Fig. 8(d) and (e)). Despite an enlargement of the non-primer coated areas, the majority of the exposed mill scale remains on the sample surface, probably due to the confinement by the primer layer. High-magnification imaging shows that some cubic crystals locate at the mouth of mill scale cracks adjacent the opening of a pit (Fig. 8(f)).

### Fig. 8.

Fig. 8.   Backscattered electron imaging to surfaces of primer pre-coated samples exposed to C2 solution for (a) 0 d, (b) 5 d, (c) 30 d and (d) 90 d; (e) secondary electron observation of the image (d); (f) high-magnification imaging of marked area “f” in image (e). The inset in image (f) is the backscattered electron observation of the marked area “g” in image (f).

### 3.5. Cross-sectional observation of immersed samples

Fig. 9 shows representative cross-sectional images of samples after immersion test. After exposure for 30 d, the outer-layer oxides on mill-scaled samples become absent and a majority of the inner-layer oxides flake off (Fig. 9(a)). Localized corrosion occurs at the interface between the mill scale and steel substrate and locations where mill scale is absent, forming shallow pit-like features. After exposure for 90 d, mill scale is rarely observed (Fig. 9(b)). The metal surface appears to be relatively flat with shallow localized corrosion features.

### Fig. 9.

Fig. 9.   Representative backscattered electron images of cross-sectional surfaces of samples after exposure to C2 solution for various time: (a, c) mill-scaled and primer pre-coated samples after 30 d of immersion, respectively; (b, d) mill-scaled and primer pre-coated samples after 90 d of immersion, respectively.

For primer pre-coated samples, most of the mill scale is intact after 30 days of immersion (Fig. 9(c)). However, deep pits occur on the metal surface where the surrounding mill scale and primer layer are defective. After 90 d of immersion, severe localized corrosion in the form of large and deep pits can be observed at locations where mill scale is absent (Fig. 9(d)).

Fig. 10 shows the pit-depth distribution after 90 days of immersion. It is clearly seen in Fig. 10 that both the pit-depth and pit-density are much higher on the samples with primer coating. For the later sample, the maximum depth was measured to be 60 μm, which is twice deeper than the maximum depth found on the mill-scaled sample without primer coating.

### Fig. 10.

Fig. 10.   Pit-depth distribution measured on 1 cm long cross-section of each sample after corrosion exposure for 90 d. Error bars show the standard deviation of the measured data.

### 3.6. Electrochemical evaluated corrosion response

Fig. 11 shows the variation of OCP with the time of corrosion exposure. For mill-scaled samples, the OCP values are ~ -0.584 V/SCE initially, but decrease gradually to a stable value of ~ -0.701 V/SCE after immersion for ~ 5 d. Similarly, the OCP values of primer pre-coated samples are ~ -0.567 V/SCE initially, and then slowly decrease to a stable value of ~ -0.697 V/SCE at immersion time of ~ 30 d of exposure. It is clear that the OCPs for both samples after 30 d of corrosion exposure appear comparable, which is likely related to the OCP of the underlying steel. For polished samples, the OCP value is initially at -0.735 V/SCE, but rises slowly to a stable value of about -0.702 V/SCE after immersion for ~10 d; the latter value is quite close to that of mill-scaled samples, which suggests that similar surface conditions and corrosion behavior have reached between the mill-scaled sample and the polished sample.

### Fig. 11.

Fig. 11.   Open circuit potential (OCP) variation of samples in C2 solution for up to 90 d.

Fig. 12 shows the EIS curves of mill-scaled samples with different immersion time. For samples after 5 d of immersion, the Nyquist plot consists of two capacitive loops respectively in the medium and low-frequency ranges, which is in good agreement with two time constants in the respective Bode plots of phase angle (θ) vs. logf (Fig. 12(d)). For primer pre-coated samples (Fig. 13), the Nyquist plots consist of three capacitive loops in the high, medium and low-frequency ranges, respectively, which is consistent with the three time constants in the according Bode plots of phase angle (θ) vs. log f (Fig. 13(d)).

### Fig. 12.

Fig. 12.   Electrochemical impedance spectra of mill-scaled samples with different immersion time: (a) Nyquist plots, (b) enlarged graph of (a), (c) Bode plots of log |Z| vs. log f and (d) Bode plots of phase angle (θ) vs. log f. The symbols and lines in images (a-d) denote the experimental and fitted data, respectively.

### Fig. 13.

Fig. 13.   Electrochemical impedance spectra of primer pre-coated samples with different immersion time: (a) Nyquist plots, (b) enlarged graph of (a), (c) Bode plots of log |Z| vs. log f and (d) Bode plots of phase angle (θ) vs. log f. The symbols and lines in images (a-d) denote the experimental and fitted data, respectively.

In order to fit the EIS data, different electrochemical equivalent circuit models were utilized, as shown in Fig. 14. It is difficult to separate the outer and inner-layer oxides in the EIS curves because the physical boundary is not well-defined, and the outer-layer oxides can be dissolved or flaked off quickly. Thus, the two-layered oxides are regarded as one entity when analyzing the EIS curves. For mill-scaled samples, the EIS curves can be described by the equivalent circuit in Fig. 14(a). Rs is the solution resistance; CPEo and Ro are respectively the constant phase element (CPE) and resistance of the reduction process of oxides (medium-frequency capacitive loop); CPEi and Ri are respectively the CPE and resistance of the dissolution of iron (low-frequency capacitive loop). CPE is usually used as a substitute for a capacitor to compensate for the nonideal capacitive response. The CPE impedance is defined as [44]:

${{Z}_{\text{CPE}}}={{\left[ Q{{\left( j\omega \right)}^{n}} \right]}^{-1}}$

where Q is the constant of CPE, ω is the angular frequency, and n is the dispersion coefficient of CPE (0 < n ≤ 1). Note that for n = 1, CPE is identical to a capacitor [45]. For the primer pre-coated samples, the EIS curves can be described by the equivalent circuit in Fig. 14(b). CPEc and Rc are, respectively, the CPE and resistance of the primer layer (high-frequency capacitive loop).

### Fig. 14.

Fig. 14.   Equivalent circuit models used for fitting the impedance spectra of samples: (a) mill-scaled samples and (b) primer pre-coated samples. (Rs: solution resistance; CPEo: CPE of the reduction process of oxides; Ro: resistance of the reduction process of oxides; CPEi: CPE of the dissolution of iron; Ri: resistance of the dissolution of iron; CPEc: CPE of the primer layer; Rc: resistance of the primer layer).

Based on the proposed equivalent circuit models, all the EIS curves were fitted with the values of fitting parameters listed in Table 3, Table 4. As for the mill-scaled samples immersed for a period from 5 d to 30 d, the values of Ro are increased from 223 Ω cm2 to 690 Ω cm2 (Table 3). The increase might correspond to the reduction of Fe3O4 after a total removal of γ-FeOOH and α-FeOOH (Fig. 5). Afterwards, the Ro values decrease with time, whereas the Qo values increase with time. These changes can be attributed to the acceleration of cathodic reduction caused by an increased area of metal surface (Fig. 6). In accordance with the decrease of Ro, Rialso decreases from 3078 Ω cm2to 2268 Ω cm2and Qiincreases from 173 μΩ-1 cm-2 sn to 373 μΩ-1 cm-2 sn, demonstrating that the resistance of iron dissolution decreases with immersion time.

Table 3   Fitting results of the EIS for mill-scaled samples with various immersion time.

TimeRs
(Ω cm2)
Qo
(μΩ-1 cm-2 sn)
noRo
(Ω cm2)
Qi
(μΩ-1 cm-2 sn)
niRi
(Ω cm2)
χ2×104
5 d5824120.762236580.7645351.01
30 d5614840.796901730.8030780.54
60 d5396240.804492930.7922780.43
90 d5296710.824073730.8022680.39

Table 4   Fitting results of the EIS for primer pre-coated samples with various immersion time.

TimeRs
(Ω cm2)
Qc
(μΩ-1 cm-2 sn)
ncRc
(Ω cm2)
Qo
(μΩ-1 cm-2 sn)
noRo
(Ω cm2)
Qi
(μΩ-1 cm-2 sn)
niRi
(Ω cm2)
χ2×104
5 d5880.50.679231700.5913924260.6578751.49
30 d5690.80.659111830.6311313540.7788331.43
60 d5411.20.616311890.778083570.8167401.71
90 d5121.40.605551910.774053800.8259101.70

For primer pre-coated samples, the Rc values decrease whereas the Qc values increase with time (Table 4), which is ascribed to the local exfoliation of the primer layer (Fig. 8) [46]. The value of Ro, initially at 1392 Ω cm2, decreases with time. Accordingly, the Qo values increase from 170 μΩ-1 cm-2 sn to 191 μΩ-1 cm-2 sn. These changes should be attributed to the increased area of oxide cathode caused by extensive exfoliation of primer layer (Fig. 8). Meanwhile, Ri value increases slightly to 8833 Ω cm2 at 30 days but decreases gradually to 5910 Ω cm2 after immersion for 90 d, indicating that the dissolution of iron is accelerated with time.

When comparing Ro values between the two different samples, it is clear that the Ro value of primer pre-coated samples is higher than that of mill-scaled samples initially, but both decrease with time and become close after 90 d of immersion. It indicates the degradation of primer layer is beneficial for the reduction of oxides on the sample surface. Similarly, the Ri value of mill-scaled samples is always lower than that of primer pre-coated samples, indicating that the primer layer can reduce the rate of overall metal dissolution.

## 4. Discussion

### 4.1. The formation of FeCO3 in dilute near-neutral pH solution

Pourbaix diagrams of Fe-H2O-CO2 system at 25 °C were calculated using HSC software [47] with a consideration of similar ones developed by Dong et al. [48]. Values of standard Gibbs free energy ΔG° (298.15 K) of the compounds or species being involved are listed in Table 5 [[48], [49], [50], [51], [52]]. According to the Pourbaix diagram in Fig. 15, FeCO3 is not stable when the concentrations of the involved species in C2 solution are at low levels. When the concentration of anions is at 10-2 mol/l, the minimum concentration of Fe2+ is determined to be ~ 10-4 mol/l in order to stabilize FeCO3 (Fig. 15(a)).

Table 5   Thermodynamic data of used species in E-pH calculations at 298.15 K.

SpeciesΔG0
(kJ mol-1)
Ref.SpeciesΔG0
(kJ mol-1)
Ref.
H20[52]CO32--527.9[52]
H+0[52]Fe2+-78.9[48]
Fe0[52]H2CO3-623.2[52]
FeCO3-666.67[48]HCO3--586.85[52]
Fe3O4-1015.4[48]H2O-237.141[49]
α-FeOOH-485.3[50]Fe2(OH)2CO3-1169.3[48]
γ-FeOOH-480.1[51]Fe6(OH)12CO3-3650[48]

### Fig. 15.

Fig. 15.   E-pH diagrams for the Fe-H2O-CO2 system at 25 °C, pressure =1 bar: (a) [Fe2+] = 10-6, 10-5, 10-4 and 10-3 mol/l and the concentrations of the anions are assumed to be 10-2 mol/l, (b) [Fe2+] = 10-6 mol/l, [anions] = 10-3, 10-2 and 10-1 mol/l. The superposed red and blue triangles respectively represent OCP values of mill-scaled and primer pre-coated samples at various immersion time.

In this investigation, the total concentration of CO2 (aqueous), H2CO3, HCO3- and CO32- was determined to be 2 × 10-3 mol/l (equilibrium concentration was calculated according to [53]), and CO32- concentration was 3.2 × 10-10 mol/l at the equilibrium condition. If the Fe2+ concentration was at 10-6 mol/l, the bulk concentration of the involved species in C2 solution is not favorable for FeCO3 formation according to the Pourbaix diagram (Fig. 15). However, FeCO3 was detected after immersion for 30 d (Fig. 5), which is rationalized below.

In the current investigation, carbonic acid (H2CO3) was generated through the carbon dioxide hydration reaction [53]:

CO2 + H2O ↔ H2CO3

The H2CO3 could then be dissociated into bicarbonate (HCO3-) and carbonate (CO32-) through a two-step process [53]:

H2CO3 ↔ H+ + HCO3-
HCO3- ↔ H+ + CO32-

The Fe2+ can be generated through iron dissolution (Reaction (8)) or oxides reduction (Reactions (9)-(11)), as shown in Table 6.

Table 6   Possible electrode reactions and corresponding expressions for their equilibrium potential used in the currently investigation (pH = 6.29, [Fe2+] = 10-6 mol/l).

No.Electrode reactionEquilibrium potential equationE (V/SCE)
(8)Fe→Fe2++2e-EFe2+/Fe=-0.6527+0.0296log[Fe2+]-0.830
(9)γFeOOH+3H++e-→Fe2++2H2OEγFeOOH/ Fe2+=0.5136-0.0592log[Fe2+]-0.178pH-0.251
(10)αFeOOH+3H++e-→Fe2++2H2OEαFeOOH/ Fe2+=0.4597-0.0592log[Fe2+]-0.178pH-0.305
(11)Fe3O4+8H++2e-→3Fe2++4H2OEFe3O4/Fe2+=0.6364-0.0888log[Fe2+]-0.237pH-0.322
(12)2H++2e-→H2EH+/H2=-0.2438-0.0592pH-0.616

When the concentration of [Fe2+] and [CO32-] is high enough, FeCO3 can precipitate according to Reaction (6):

Fe2+ + CO32- → FeCO3

A parameter frequently used in literature to determine precipitation of FeCO3 is the saturation ratio of iron carbonate ($S{{R}_{\text{FeC}{{\text{O}}_{3}}}}$), defined as [54,55]:

$S{{R}_{\text{FeC}{{\text{O}}_{3}}}}=\left[ \text{F}{{\text{e}}^{2+}} \right]\left[ \text{CO}_{3}^{2-} \right]/{{K}_{\text{sp}}}$

where [Fe2+] is the concentration of ferrous ions, [CO32-] is the equilibrium concentration of the carbonate ions, Ksp is the solubility product. The Ksp is reported to be in a range from 1.17 × 10-11 mol2/l2 to 3.72 × 10-11 mol2/l2 at room temperature [54]. Based on the mass loss data in Fig. 3, and assuming that all the mass loss was caused by iron dissolution and all other factors were insignificant, the maximum [Fe2+] and [CO32-] were calculated to be ~ 2 × 10-3 mol/l and ~ 3.2 × 10-10 mol/l, respectively, in the bulk solution after 90 days of immersion [53]. Assuming Ksp = 1.17 × 10-11 mol2/l2 and substituting the above data into Eq. (7) yields a value of $S{{R}_{\text{FeC}{{\text{O}}_{3}}}}$ of ~ 5 × 10-2, which is about 20 times smaller than 1, the minimum ($S{{R}_{\text{FeC}{{\text{O}}_{3}}}}$) in theory for the precipitation of FeCO3 [56].

Despite the above inconsistency, FeCO3 was detected on the surface of mill-scaled and primer pre-coated samples (Fig. 5). It is believed that either a higher concentration of Fe2+ or CO32- was achieved on the surface of both samples with mill scale, for example, within the narrow region on the metal surface underneath the mill scale where the electrolyte is nearly stagnant [57]. The transport of Fe2+, generated from metal dissolution and reduction of mill scale, is severely limited [57]. With increasing immersion time, high concentration of Fe2+ can be achieved. Precipitation of FeCO3 is possible at porous or cracked location of mill scale where $S{{R}_{\text{FeC}{{\text{O}}_{3}}}}$ becomes sufficiently high. The fact that FeCO3 was not detected on polished sample surface confirms the critical role played by the mill scale in the precipitation of FeCO3 (Fig. 5).

A slight mass gain on primer pre-coated samples was observed after 45 days of immersion (Fig. 3). Compared with mass loss of mill-scaled samples, the slight mass gain of primer pre-coated samples is a combined effect of the following factors: (I) mass loss due to exfoliation of mill scale, and active corrosion at the bottom of cracks/pores in mill scale (Fig. 8, Fig. 9), and (II) mass gain caused by FeCO3-formation (Fig. 5). The latter might be the sole reason for the slight mass increase observed.

### 4.2. Primer-assisted localized corrosion of samples with mill scale

The pH = 6.29 and [Fe2+] = 10-6 mol/l were used for the following analysis. According to the OCP evolution of mill-scaled samples (Fig. 11, Fig. 15), the OCP measured after 2 h of immersion was ~ -0.584 V/SCE, which was in the corrosion region of Fe2+ in the Pourbaix diagram, indicating that the penetration of solution through mill scale to the metal surface was achieved [8]. Table 6 lists some of the possible electrode reactions that can occur spontaneously when the OCP remains in the corrosion region of Fe2+ [[58], [59], [60]]. Since the OCP value is between the equilibrium potential of the anodic and cathodic half reactions [61], a galvanic couple should be formed between the metal (oxidation) (Reaction (8)) and oxide (reductive dissolution) (Reactions (9)-(11)). The reductive oxide dissolution would cause the removal of the outer oxide layer and formation of pores or flaws in the inner oxide layer (Fig. 6).

After 5 d immersion, the OCP value became lower than the equilibrium potential of hydrogen evolution reaction (HER) and remained in the corrosion region of Fe2+ (Fig. 15). Under these circumstances, HER might be involved in the above galvanic process. Since the outer-layer oxides and γ-FeOOH peaks were not observed, it is clear that dissolution of α-FeOOH and Fe3O4, and hydrogen evolution were the possible cathodic reactions after 5 d immersion (Fig. 5, Fig. 6(b)).

After 30 d of immersion, α-FeOOH peaks disappeared (Fig. 5), indicating Fe3O4 reduction (Reaction (11)) and HER were the two cathodic reactions. By this stage, the flaws in mill scale are enlarged due to the dissolution of Fe3O4. The HER will become predominant when majority of mill scale is dissolved or flaked off [8].

The above analysis was made based on the possible electrode reactions listed in Table 6. There are other reactions that could coexist. For instance, reducing from γ-FeOOH to the stable phase of Fe3O4 in the presence of Fe2+ is much easier than from α-FeOOH [17,62,63], which could explain the fast disappearance of γ-FeOOH (Fig. 5). Note that all the subsequent discussions are made on the basis of the electrode reactions proposed in Table 6.

Based on the above analysis, a three-stage process of corrosion of mill-scaled samples can be envisioned and also schematically illustrated in Fig. 16.

### Fig. 16.

Fig. 16.   Schematic illustration for the corrosion process on cross-sections of (a, c, e, g) mill-scaled and (b, d, f, h) primer pre-coated samples in C2 solution.

Stage I: Autoreductive dissolution of selective oxide scales and early stage of steel corrosion. Upon the start of immersion, the outer layer, i.e. γ-FeOOH, can be quickly removed by autoreductive dissolution (Fig. 16(a) and (c)), which is coupled with metal dissolution. The term “autoreductive dissolution” is used exclusively for the situations where oxide layer and substrate dissolution processes are coupled [63]. The removal of the outer layer is followed by direct dissolution of the inner-layer oxides due to the autoreductive dissolution. As analyzed by SEM and XRD, γ-FeOOH and α-FeOOH are not stable and will be either dissolved or exfoliated prior to Fe3O4 (Fig. 5, Fig. 6). The above galvanic process leads to the formation of flawed mill scale, localized corrosion of the underlying metal surfaces, and the start of pit-formation (Fig. 16(c)). In addition, a higher concentration of Fe2+ is accumulated gradually at the corroded metal surface and in the flawed mill scale. At these locations, local precipitation of FeCO3 is possible, as discussed in Section 4.1.

Stage II: Enhanced localized corrosion and pit-formation. With increasing the immersion time, more inner-layer oxides will be released to the solution and a large area of the exposed metal surface will be created (Fig. 16(e)). Consequently, the HER on the exposed metal surface underneath the flawed mill scale or on the surface of conductive mill scale, starts to play a role [8]. At this stage of immersion, the reductive dissolution of oxides continues as demonstrated by the roughing surface of the mill scale with increasing exposure time (Fig. 6).

In Stage II, precipitation of FeCO3 continues at flawed locations of mill scale (Fig. 5). This precipitation creates a capped enclosure that blocks the path of mass transport to and from the bulk solution. Such a configuration in principle may favour localized corrosion and pit-formation [14,[64], [65], [66], [67], [68]]. A large area of conductive Fe3O4 adjacent to mill scale flaws would allow the flow of electrons from the metal surface to the cathodic sites [69]. Thus, the anodic dissolution of iron can continue, which causes an accumulation of Fe2+ ions and the subsequent hydrolysis of Fe2+ ions. The latter reaction would increase H+ ions in the solution. Concurrently, Cl- ions migrate to the local areas to maintain the charge balance. As a result of both processes, a high acidity condition is achieved, and dissolution of Fe is accelerated, resulting in the fast formation of corrosion pits. The above corrosion process would come to a stop when the mill scale has flaked-off.

Stage III: General corrosion after removal of mill scale. As the dissolution of oxides continues, total removal of the mill scale could be observed (Fig. 16(g)), and HER as a cathodic reaction would shift from mill-scale-covered surfaces to the exposed metal surfaces [70]. Under the circumstances, the anodic and cathodic reactions are occurring at the same location, and corrosion becomes uniform in nature, and therefore a shift occurs from localized corrosion to general corrosion (Fig. 6, Fig. 9).

The mill-scaled surface with primer coating applied also experiences the above three-stage process with some uniqueness:

(1) The primer layer is not an effective barrier to water solutions and the same three-stage corrosion process as seen for the non-primer coated mill scale occurs except that primer layer slows down the degradation of mill scale and therefore would preserve the localized corrosion for an extended period.

(2)Similar to mill-scaled samples, FeCO3 deposits at flawed locations of mill scale creating occluded areas, which are favorable for localized corrosion and pit-formation. Moreover, the defected primer layer can also act as capped enclosures and further aggravates the local acidity in occluded areas (Figs. 8(f) and 16 (f)).

(3)Even though a total removal of the outer-layer oxides (γ-FeOOH) is observed after long-term immersion, the majority of mill scale (Fe3O4 and α-FeOOH) remains on the metal surface due to the strong confinement of the primer layer (Figs. 16(h) and 5). The prolonged longevity of the mill scale leads to a significantly extended period of enhanced anodic dissolution of metal at the bottom of cracks or pores, and, subsequently, the formation of deep corrosion pits.

## 5. Conclusions

The corrosion behavior of a mill-scaled X65 pipeline steel with and without primer layer was investigated in parallel in simulated near-neutral pH soil solution; the following conclusions can be drawn:

(1)The mill scale on sample surface appeared in two layers, a porous outer layer and a relatively dense inner layer with cracks. The outer layer mainly consisted of lepidocrocite (γ-FeOOH). The inner layer consisted of predominately magnetite (Fe3O4), a small portion of goethite (α-FeOOH), and lepidocrocite (γ-FeOOH).

(2)The porous outer layer (γ-FeOOH) on the mill-scaled sample surface could be quickly dissolved by autoreductive dissolution, and preferential dissolution of α-FeOOH and γ-FeOOH in the inner layer could occur subsequently, which caused localized corrosion and pit-formation. Finally, the dense Fe3O4 in the inner layer would be gradually dissolved by reductive dissolution.

(3)For the mill-scaled sample condition, localized corrosion and pitting occurred during the initial period of immersion, but general corrosion prevailed as most of the mill scale flaked off during the subsequent long period of exposure.

(4)The application of the primer layer significantly reduced the degradation of the mill scale. However, severe localized corrosion and pitting caused by anodic dissolution dominated at locations where the primer coating was damaged. Localized corrosion proceeded due to the persistent acceleration from the cathodic reduction of oxides and water on the exposed mill scale confined by the nearby primer layer.

(5)Local precipitation of siderite (FeCO3) was found at flawed locations of mill scale, although the bulk chemistry is not favorable for its formation on the steel surface without mill scale.

## Acknowledgments

The authors wish to thank TransCanada Pipelines Limited, Enbridge Pipelines Inc., Natural Science and Engineering Research Council of Canada, Pipeline Research Council International for the financial support.

## Reference By original order By published year By cited within times By Impact factor

W. Chen, Corrosion 72 (2016) 962-977.

H.L.H. Cuong, T.V. Dieu, N.D. Thanh, D.T.Y. Oanh, Vietnam J. Chem. 54 (2016) 419-423.

V. Valencia-Goujon, T. Hawkins, E. Maya-Visuet, H. Castaneda,

Corrosion 2015, NACE International

Dallas, Texas, 2015, March 15-19.

I. Thompson, J.R. Saithala, Corros. Eng. Sci. Technol. 51 (2016) 118-135.

D.A. Bayliss, D.H. Deacon, Steelwork Corrosion Control, Spon Press, London, 2014.

S. Ebnesajjad, A.H. Landrock, New York, 2014.

Z. Shirband, University of Alberta, 2016.

Z. Qin, B. Demko, J. Noël, D. Shoesmith, F. King, R. Worthingham, K. Keith, Corrosion 60 (2004) 906-914.

P. Ghods, O.B. Isgor G.A. McRae, J. Li, G.P. Gu, Corros. Sci. 53 (2011) 946-954.

Y.F. Cheng, New Jersey, 2013.

B. Fang, E. Han, J. Wang, W. Ke, Corrosion 63 (2007) 419-432.

Y.Z. Jia, J.Q. Wang, E.H. Han, W. Ke, J. Mater. Sci. Technol. 27 (2011) 1039-1046.

M.C. Bagatini, V. Zymla, E. Osório A.C.F. Vilela, ISIJ Int. 51 (2011) 1072-1079.

G. Zhang, Y. Cheng, Electrochim. Acta 56 (2011) 1676-1685.

H. Liu, Y.F. Cheng, Electrochim. Acta 266 (2018) 312-325.

L. Shi, X. Yang, Y. Song, D. Liu, K. Dong, D. Shan, E.H. Han, J. Mater. Sci. Technol. 35 (2019) 1886-1893.

X. Zhang, K. Xiao, C. Dong, J. Wu, X. Li, Y. Huang, Eng. Fail. Anal. 18 (2011) 1981-1989.

Corrosion products of carbon steel in a simulated solution containing Cl- and SO42- were investigated by in situ Raman spectroscopy with a self-designed device. The results of Raman test and the SEM/EDS analysis of samples in different immersion times indicate that the corrosion products on the surface of the carbon steel consist of double rust layers, i.e. a porous outer layer (lepidocrocite gamma-FeOOH) and a compact inner layer (magnetite Fe3O4). Aggressive ions such as Cl- and SO42- promote corrosion leading to the formation of lepidocrocite gamma-FeOOH. Weight loss vs. time curve was also measured to study the effects of corrosion products. Corrosion products on a parallel sample were tested by Raman spectroscopy in air environment, which shown a distinct result from the one of in situ test. (C) 2011 Published by Elsevier Ltd.

X. Nie, X. Li, C. Du, Y. Huang, H. Du, J. Raman Spectrosc. 40 (2009) 76-79.

B.W.A. Sherar, P.G. Keech, D.W. Shoesmith, Corros. Sci. 53 (2011) 3643-3650.

J. Xu, Y. Bai, T. Wu, M. Yan, C. Yu, C. Sun, Eng. Fail. Anal. 100 (2019) 192-205.

T. Wu, M. Yan, L. Yu, H. Zhao, C. Sun, F. Yin, W. Ke, Corros. Sci. 157 (2019) 518-530.

L. Yu, M. Yan, B. Wang, Y. Shu, J. Xu, C. Sun, J. Chem. Soc. 38 (2018) 10-17.

K. Yin, H. Liu, Y.F. Cheng, Corros. Sci. 145 (2018) 271-282.

T. Wu, C. Sun, J. Xu, M. Yan, F. Yin, W. Ke, Corros. Eng. Sci. Technol. 53 (2018) 265-275.

M. Yan, J. Xu, L. Yu, T. Wu, C. Sun, W. Ke, Corros. Sci. 110 (2016) 23-34.

K. Yin, Y. Yang Y. Frank Cheng, Constr. Build. Mater. 192 (2018) 20-27.

U. Eduok, E. Ohaeri, J. Szpunar, Mater. Sci. Eng. C 105 (2019), 110095.

R. Jia, T. Unsal, D. Xu, Y. Lekbach, T. Gu, Int. Biodeter. Biodegr. 137 (2019) 42-58.

E. Ohaeri, U. Eduok, J. Szpunar, Int. J. Hydrogen Energy 43 (2018) 14584-14617.

Q. Hu, S. Lu, J. Liu, F. Huang, Anti-Corros. Methods Mater. 66 (2019) 651-660.

K. Yin, University of Calgary, 2018.

L. Xu, X. Su, Y. Cheng, Corros. Sci. 66 (2013) 263-268.

R.M. Souto, Y. González-Garc$\acute{i}$a, S. González, G.T. Burstein, Corros. Sci. 46 (2004) 2621-2628.

K. Chevil, W. Chen, G. Van Boven, R. Kania, J. Been,

10th International Pipeline Conference, American Society of Mechanical Engineers Digital Collection, Calgary

Alberta, 2014, September 29-October 3.

A. Eslami, R. Kania, B. Worthingham, G.V. Boven, R. Eadie, W. Chen, Corros. Sci. 53 (2011) 2318-2327.

D. de la Fuente, J. Alcántara, B. Chico, I. Díaz, J.A. Jiménez, M. Morcillo, Corros. Sci. 110 (2016) 253-264.

D. Neff, S. Reguer, L. Bellot-Gurlet, P. Dillmann, R. Bertholon, J. Raman Spectrosc. 35 (2004) 739-745.

M. Legodi D. De Waal, Dye. Pigment. 74 (2007) 161-168.

J. Dong, E. Han, W. Ke, Sci. Technol. Adv. Mater. 8 (2007) 559-565.

L. Hao, S. Zhang, J. Dong, W. Ke, Corros. Sci. 59 (2012) 270-276.

C. Qiao, L. Shen, L. Hao, X. Mu, J. Dong, W. Ke, J. Liu, B. Liu, J. Mater. Sci. Technol. 35 (2019) 2345-2356.

L. Hao, S. Zhang, J. Dong, W. Ke, Corros. Sci. 53 (2011) 4187-4192.

C. Lee, M. Odziemkowski, D. Shoesmith, J. Electrochem. Soc. 153 (2006) B33-B41.

F. King, Virginia, 2010.

M. Nobial, O. Devos, O.R. Mattos, B. Tribollet, J. Electroanal. Chem. 600 (2007) 87-94.

A. Lasia,

Electrochemical Impedance Spectroscopy and its Applications, in: Modern Aspects of Electrochemistry

Springer, Boston, Massachusetts, 2002, pp. 143-248.

A. Roine, Finland, 2002.

J. Dong, T. Nishimura, T. Kodama, Mater. Res. Soc. Symp. Proc. 713 (2002) 105-112.

J. Malcolm, W. Chase,

NIST-JANAF Thermochemical Tables, in: Part I, Al-Co, 4th ed.

Gaithersburg, Maryland, Monograph, 1998.

B. Beverskog, I. Puigdomenech, Corros. Sci. 38 (1996) 2121-2135.

J. Majzlan, K.-D. Grevel, A Navrotsky, Am. Mineral. 88 (2003) 855-859.

D.D. Wagman, W.H. Evans, V.B. Parker, R.H. Schumm, I. Halow, S.M. Bailey, K.L. Churney, R.L. Nuttall, J. Phys. Chem. Ref. Data 18 (1989) 1807-1812.

M. Nordsveen, S. Neˇsic, R. Nyborg, A. Stangeland, Corrosion 59 (2003) 443-456.

R. Barker, D. Burkle, T. Charpentier, H. Thompson, A. Neville, Corros. Sci. 142 (2018) 312-341.

R.A. De Motte, R. Barker, D. Burkle, S.M. Vargas, A. Neville, Mater. Chem. Phys. 216 (2018) 102-111.

W. Sun, K. Chokshi, S. Nesic,

Corrosion 2005, NACE International, Houston

Texas, 2005,April 3-7.

F. Farelas, B. Brown, S. Nesic,

Corrosion 2013, NACE International, Orlando

Florida, 2013, March 17-21.

Y. Cheng, Electrochim. Acta 52 (2007) 2661-2667.

M. Li, Y. Cheng, Electrochim. Acta 52 (2007) 8111-8117.

C. Dong, A. Fu, X. Li, Y. Cheng, Electrochim. Acta 54 (2008) 628-633.

V.S. Bagotsky, New Jersey, 2005.

H. Antony, L. Legrand, L. Maréchal, S. Perrin, P. Dillmann, A. Chaussé, Electrochim. Acta 51 (2005) 745-753.

Y.H. Huang, T.C. Zhang, Water Res. 40 (2006) 3075-3082.

R. Avci, B. Davis, M. Wolfenden, I. Beech, K. Lucas, D. Paul, Corros. Sci. 76 (2013) 267-274.

H. Liu, X. Zhong, H. Liu Y. Frank Cheng, Electrochem. Commun. 90 (2018) 1-5.

Z. Liu, X. Gao, L. Du, J. Li, P. Li, C. Yu, R.D.K. Misra, Y. Wang, Electrochim. Acta 232 (2017) 528-541.

B. Zhang, X.L. Ma, J. Mater. Sci. Technol. 35 (2019) 1455-1465.

F. Xue, X. Wei, J. Dong, C. Wang, W. Ke, J. Mater. Sci. Technol. 35 (2019) 596-603.

G.S. Parkinson, Surf. Sci. Rep. 71 (2016) 272-365.

F. King,

Overview of a Carbon Steel Container Corrosion Model for a Deep Geological Repository in Sedimentary Rock, in: Nuclear Waste Management Organization Report, Nuclear Waste Management Organization

Toronto, Ontario, 2007.

ISSN: 1005-0302
CN: 21-1315/TG
Editorial Office: Journal of Materials Science & Technology , 72 Wenhua Rd.,
Shenyang 110016, China
Tel: +86-24-83978208
E-mail:JMST@imr.ac.cn