Construction and Building Materials 362 (2023) 129746

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Long-term corrosion and stress corrosion cracking of X65 steel in

H2O-saturated supercritical CO2 with SO2 and O2 impurities
Kaiyang Li , Yimin Zeng *
Natural Resources Canada, CanmetMATERIALS, Hamilton, Ontario L8P 0A5, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: In this study, four testing cycles (total 480 h) were conducted to investigate corrosion and stress corrosion
Supercritical carbon dioxide cracking (SCC) of X65 steel in H2O-saturated supercritical CO2 (s-CO2) containing SO2 and O2 impurities. The
SO2 results showed that X65 steels experienced severe corrosion and formed a porous corrosion layer composed of
Corrosion
iron sulfates, carbonates and oxy-hydroxides. Increasing time from 120 to 480 h resulted in the shift of dominant
Stress corrosion cracking
corrosion products from iron carbonates to iron oxy-hydroxides due to acid regeneration cycle. The mass loss of
X65 steel almost linearly increased with exposure time, indicating a nearly constant corrosion rate in the s-CO2
environment. In comparison with previous studies, timely replenishing s-CO2 fluid in a static autoclave is crucial
to accurately determine the long-term corrosion kinetic of pipeline steel. Based on literature survey and current
study, several factors (SO2 content, H2O content, addition of NO2) affect the long-term corrosion rate in s-CO2
environments, and a power-function-based empirical model is purposed to depict the long-term corrosion rate of
X65 steel in H2O-saturated s-CO2 + O2 + 100/2000 ppm SO2. The long-term SCC assessments showed that the
SCC susceptibility of X65 would be still quite low, which might be related to the O2-depressed hydrogen cracking
and/or the lack of localized stress concentration.

1. Introduction Due to the limitation of capture technologies applied and economic

consideration, impurities (H2O, NO2, O2, H2S, SO2, etc.) are inevitably
The increasing power/heat/steam demands for human activities found in the transported s-CO2, thus posing a great threat to pipeline
have resulted in huge amounts of fossil fuel consumption and conse­ integrity [2,5,6]. Among these impurities, SO2 is likely to be the most
quently significant CO2 emission, which is recognized as the main source notorious one to cause severe corrosion damage [6,7]. In dry environ­
of global climate change. Carbon Capture, Utilization and Storage ments, SO2 could hardly introduce any damage to pipe steels [8]. With
(CCUS) technology has thus been developed as an effective pathway to the simultaneous presence of H2O, however, the corrosion of steels often
remarkably reduce CO2 emission while keeping the core values of fossil increases greatly [9]. Such high corrosivity is credited to the dissolution
fuel combustion systems. The successful deployment of CCUS technol­ of SO2 in water and the subsequent formation of sulfurous and sulfuric
ogy relies on the safe, reliable, and efficient pipeline transportation of acids (H2SO3/H2SO4) [10]. Previous studies show that the addition of
CO2 captured from combustion plants to desired permanent storage only 1000 ppm SO2 in H2O-saturated s-CO2 could result in a high
regions [1]. During transportation, CO2 needs to be compressed into the corrosion rate of pipe steels up to 1.9 mm/y [11]. Besides, the syner­
supercritical state (T > 304.2 K and P > 7.39 MPa) to achieve high gistic effect of SO2 with other impurities (like NO2, O2 and H2S) will
operating efficiency [2]. Commercial carbon steels are seen as the best likely intensify the corrosion damage as the corrosion rate of pipe steels
candidates for pipeline construction because of their good mechanical in the mixture is usually higher than that in the s-CO2 with individual
behavior and low cost. Among them, API 5L X65 is a good candidate impurity [12]. As summarized in Table 1, SO2 poses a great corrosion
since it has been widely applied in oil and gas fields to deliver dense CO2 threat to pipe steels in s-CO2 streams containing impurities and the
streams mixed with H2O, acid, and other components [3,4]. corrosion rates are generally very high.

Abbreviations: CCUS, Carbon Capture, Utilization and Storage; SCC, Stress Corrosion Cracking; s-CO2, supercritical CO2; HSLA, High-Strength Low-Alloy; SEM,
Scanning Electron Microscopy; EDS, Energy Dispersive Spectroscopy; XRD, X-Ray Diffraction; DIC, Digital Image Correlation; FIB, Focused Ion Beam; FPB, Four Point
Bending; FEA, Finite Element Analysis.
* Corresponding author.
E-mail addresses: [email protected], [email protected] (Y. Zeng).

https://doi.org/10.1016/j.conbuildmat.2022.129746
Received 15 February 2022; Received in revised form 18 September 2022; Accepted 9 November 2022
Available online 18 November 2022
0950-0618/Crown Copyright © 2022 Published by Elsevier Ltd. All rights reserved.
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Despite these findings, however, significant knowledge gaps remain induced SCC in s-CO2 are quite limited up to now. Our preliminary work
on the role of SO2 impurity on corrosion in s-CO2 streams, which hinders indicated that X65 steel exhibited low SCC susceptibility during the
the development of international CO2 transportation pipeline standards. short-term exposure (137 h) in H2O-saturated s-CO2 with 100 ppm SO2
Among them, SO2-induced long-term corrosion and Stress Corrosion [27]. However, with extended time, the embrittlement effect by SO2
Cracking (SCC) are the two major issues that are urgently required to be may become more pronounced due to the accumulation of evolved
addressed. First, the long-term effect of SO2 on pipeline steel corrosion hydrogen and possibly accelerated corrosion. For instance, in a simu­
in s-CO2 needs to be clarified. As shown in Table 1, most previous studies lated S-containing marine environment with humidified 3.5 % NaCl +
were conducted within 120 h exposure. It is unclear whether the 0.01 mol/L NaHSO3 solution, U-bended E690 steel experienced SCC
corrosion kinetics will be enhanced or remain the same in s-CO2 with initiation right after 5 days and then enhanced cracking after 90 days
SO2 impurity during long-term exposure. Previous studies on SO2- [28]. In general, it is necessary to identify the long-term SCC suscepti­
induced corrosion at atmospheric pressure showed the corrosion rate bility of pipeline steels in s-CO2 streams.
could be considerably high at an early stage due to the fast and direct According to the above description, the long-term corrosion kinetic
chemical attacks on steel surface by SO2 and other impurities, and prediction and SCC susceptibility of s-CO2 pipeline steels still remain
subsequently be reduced with extended time because of the accumula­ unknown. Therefore, this study intends to investigate the long-term
tion of corrosion products, which might serve as a barrier to retard the corrosion and SCC behavior of X65 pipeline steel in H2O-saturated s-
diffusion of aggrieve agents from the environment to the rust/steel CO2 with SO2 and O2 for simulating the transported s-CO2 captured from
interface [17]. For example, when exposed to SO2 + air mixture with a advanced oxy-fuel combustion power plants. To achieve a relatively
relative humidity of 99% [8], the corrosion rate of low carbon steel was consistent chemical environment, cyclic autoclave testing was per­
comparatively high (9.7 mpy) for the initial 3 days, and then gradually formed and the s-CO2 fluid with impurities was completely replenished
reduced down to a constant value (about 0.6–0.7 mpy) with extended after each cycle. After a cyclic exposure, the average corrosion rates of
time of 200–500 days [8]. However, there is also a possibility that the X65 steel were evaluated and the formed corrosion products were
corrosion products formed in s-CO2 with SO2 are too porous/permeable thoroughly characterized using digital cameras, Scanning Electron Mi­
to protect the steel substrate, thus making the long-term corrosion ki­ croscopy (SEM) with Energy Dispersive Spectroscopy (EDS) and X-ray
netics almost linear [18]. For instance, at 25 ◦ C, the corrosion rate of Diffraction (XRD) techniques. In the testing autoclave, the SCC suscep­
unalloyed carbon steel in atmospheric air with 90% humidity and 10 tibility of X65 steel was also examined using four-point bending and
ppm SO2 was found to be almost constant during the three-week expo­ tensile loading methods, Digital Image Correlation (DIC), and Finite
sure [19]. Clearly, long-term corrosion kinetics of pipeline steels in s- Element Analysis (FEA). After exposure, SEM and Focused Ion Beam
CO2 streams shall be coherently related to the timely concentration of (FIB) techniques were employed to determine whether the cracks were
SO2, the presence of other impurities (such as H2O and O2), and the formed. It is anticipated that this study can clarify the long-term
properties of corrosion layers formed. It is necessary to formulate a corrosion and SCC behavior of s-CO2 pipeline steels and help industry
reliable corrosion rate prediction model to achieve the purpose of long- engineers to construct and operate s-CO2 pipelines in a safe and reliable
term safe operation of s-CO2 pipeline networks. manner.
Besides general corrosion, SO2–induced SCC is another serious threat
to the integrity of s-CO2 transportation pipelines as SO2 is a well-known 2. Materials and methods
agent to trigger SCC of pipeline steels in petroleum, chemical and other
related industries. For example, Ma et al. [20] found that SO2 could 2.1. Specimens preparation
promote the SCC of E690 steel by facilitating the initiation of micro­
cracks underneath the corrosion layer and enhancing hydrogen gener­ Commercial X65 steel was used in this study, and its chemical
ation. Wu et al. [21] simulated a SO2-polluted marine atmosphere by composition includes: 1.36 Mn, 0.21 Si, 0.012 Ni, 0.023 Cr, 0.075 C,
humidifying a solution of 3.5 % NaCl + 0.01 NaHSO3, and reported that 0.024 Cu, 0.062 Nb, 0.013 P, 0.03 Al, 0.0064 W and balanced Fe (all in
a High-Strength Low-Alloy (HSLA) steel suffered from SCC through the wt.%). The test sample was machined from an X65 pipe into different
trapped hydrogen, whose formation was enhanced by SO2− 4 -induced dimensions for corrosion and SCC studies. For corrosion study, the
cyclic acid regeneration process. Gong et al. [22] found the SCC initia­ rectangular corrosion coupon was made with a size of 38 mm long × 20
tion of X100 steel was promoted when exposed to humid SO2-containing mm wide × 3 mm thick and mounted on a PTFE sample holder (shown in
environments due to the localized stress concentration at pits. The Fig. 1(a)). For SCC tests, two types of specimens were machined based
dissolution of SO2 in liquid phase will generate HSO−3 , which promotes on ASTM standards (E8 or D6272) [29,30]: tensile specimen with a
hydrogen evolution and the following embrittlement [22]. In s-CO2 gauge size of 32 mm long × 6 mm wide × 2 mm thick (see Fig. 1(b)) and
pipeline systems, the contained SO2 could be up to 1500 ppm, which bent beam with a size of 115 mm long × 18 mm wide × 3 mm thick (see
consequently introduces a great SCC risk [5]. In addition, the pipeline Fig. 1(c)). The longitudinal direction of SCC samples was parallel to the
steels shall undergo loading from not only the internal high-pressure s- hoop direction of the pipe. After machining, the surfaces of each sample
CO2 streams (about 8–17 MPa) [23], but also external stress like sig­ were polished with a series of silicon carbide papers up to 600 grit,
nificant ground movement and vibration [24]. Therefore, several studies washed with distilled water, degreased with ethanol in an ultrasonic
predict that s-CO2 transportation pipeline steels would be vulnerable to cleaner, and finally dried with pressurized air. The weight of a prepared
SCC failure [25,26]. Unfortunately, the related studies regarding SO2- sample was measured using a digital balance with an accuracy of 0.1 mg

Table 1
Corrosion rates of pipeline steels in SO2-containing H2O-saturated s-CO2 environments.
T (◦ C) P (MPa) t (h) O2 (ppm) SO2 (ppm) NO2 (ppm) H2S (ppm) Corrosion rate (mm/y) Ref.

50 10 120 – 1000 – – 0.469 [13]

50 10 120 – 1000 – 1000 1.224 [13]
50 8 24 40,000 10,000 – – 7 [10]
50 10 120 1000 20,000 – – 0.9 [14]
35 8 48 20 100 – – 0.72 [15]
50 8 24 – 10,000 – – 5.6 [16]
50 10 120 – 1000 1000 – 0.46 [12]
50 10 120 1000 1000 1000 1000 1.16 [12]

2
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Fig. 1. (a) Corrosion coupon and the PTFE sample holder; (b) tensile specimen and the tensile loading device with an extensometer for strain measurement; (c) bent
beam and the FPB fixture for strain loading.

and the dimensions were examined using a caliper with an accuracy of inner alumina rods) when the center screw was tightened. The
0.01 mm. All the prepared samples were stored in a vacuum container displacement of the beam sample during loading was measured using a
before a corrosion test. depth micrometer at the bottom of the fixture. Based on NACE TM1077,
the relationship between displacement and applied surface stress on the
mid-portion of the specimen can be described in the following equation
2.2. SCC testing methodologies
[31]:
SCC tests were carried out by two constant-strain methods, including 12Ecd
σ= (1)
constant tensile loading and Four Point Bending (FPB). For the tensile 3H2 − 4U2
loading method, a fixture made of Hastelloy C-276 loaded tensile stress
on the specimen (shown in Fig. 1(b)). The tensile specimen has two where σ is the surface stress on the mid-portion of the specimen; d is the
drilled holes on two grip sections, which could be held firmly by two displacement, mm; E is the elastic modulus of the testing steel; c is the
pins made of Alloy C-276 during the loading process. By rotating the nut, thickness of the sample, mm; H is the distance between two outer
the tensile specimen experiences extension and the elongation rate is alumina rods, mm; U is the distance between inner and outer alumina
measured by an extensometer. To avoid galvanic effects with the fixture rods, mm.
or pin during corrosion tests, two grip sections of tensile specimens and FEA was conducted using commercial software ABAQUS 2020 to
the pins were wrapped with thin PTFE tape as the electro-insulator. study the stress/strain distribution on FPB-loaded bent beam as a
For the FPB method (shown in Fig. 1(c)), a fixture made of Hastelloy function of displacement. The SCC specimen was treated as elastic-
C-276 was used to load strain on the rectangular beam sample. Four plastic solid and modeled based on Von-Mises yield criterion and
electrically insulated alumina rods provide the contact points between Johnson-Cook plasticity model [32]:
the fixture and the specimen. Stress/strain could be homogeneously σ = (Y + Z εn ) (2)
applied to the center region of the beam sample (the region between two

3
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

where σ is the flow stress of the material, ε is the strain level, Y is the inlet valve. The amount of water was controlled to ensure a water-
yield stress, Z is the pre-exponential factor, n is the work-hardening saturated s-CO2 environment based on theoretical calculations while
exponent. The parameters were determined from quasi-static tensile avoiding the reaction of SO2 with extra water to form H2SO4 [27,35]. At
tests of X65 steel, during which the X65 was machined into the standard 25 ◦ C, O2 was charged to 0.179 MPa. Once heated to 45 ◦ C, SO2 was
specimen and tensiled at a strain rate of 1×10− 4 s− 1 based on ASTM E8 added through a mixture of CO2 + 774 ppm SO2 to make the total
[29]. Then, the relationship between strain/stress distribution on the pressure reaching 3.25 MPa, and CO2 was added by the supercritical CO2
bent beam and the loading displacement was acquired by ABAQUS. Due pump to make the final total pressure of 10 MPa. At 10 MPa and 45 ◦ C,
to the symmetry, only half of the SCC bent beam in the length direction the corrosion test lasted for 120 h as one cycle. After a cycle, the auto­
was modeled, with the symmetry plane perpendicular to the longitudi­ clave was cooled down to room temperature and the vented gas from the
nal direction of the bent beam. The geometry used 8-noded reduced autoclave passed through a gas-washing bottle before being emitted into
integration brick elements (C3D8R). The mesh was built based on linear the air. Four corrosion coupons and selected strained samples were
brick elements (0.4 mm in size). Based on the stress-strain curve of the removed from the autoclave for corrosion product characterization,
sample, the strain level is acquired when the corresponding stress is corrosion rate assessment and SCC examination while the remaining
known. samples were kept for the next cycle. The whole test consists of four
To validate the FEA simulation results, DIC method was used to cycles (480 h in total). The strained samples removed after each cycle
measure the strain levels with an optical strain mapping system. FPB are shown in Table 2.
sample surface was spray-painted with black speckles over a white
background, and in-situ displacements of these black speckles during 2.4. Corrosion product characterization
deformation were tracked by high-resolution cameras [33]. By
comparing the high-resolution images captured before and after defor­ Both mass gain and mass loss measurements were performed to
mation, Aramis V6.3 software acquires the movement of these speckles evaluate the corrosion degree. The mass gain value is the direct mass
and determines the surface strains of the bent beam. After DIC mea­ change (usually positive) during the corrosion exposure. The mass loss
surements, the surface paint patterns were thoroughly removed with value was acquired as the weight change between pristine state and
acetone and the cleaned samples were kept in a vacuum container before descaled state (after completely removing the formed corrosion prod­
corrosion tests. ucts). Based on ASTM G1-03 [36], the corrosion products were removed
by immersing corrosion coupons in a pickling solution of 500 mL hy­
drochloric acid + 5 g hexamethylene tetramine + 500 mL deionized
2.3. Corrosion experiment setup and procedure water. The immersion was terminated when the steel surface was free
from dark color caused by corrosion and reshowed the original metallic
Fig. 2 illustrates the setup of s-CO2 corrosion testing. Based on the color. Then, the coupons were washed with deionized water, dehydrated
flue gas chemical composition analysis from a pilot pressurized oxy-fuel with alcohol and dried with pressurized air before weighing. The
combustion power plant (3.2–6.0 % O2, 36–800 ppm SO2, H2O up to corrosion rate (CR, mm/y) was then calculated from three parallel
solubility) and the recommended gas compositions by several guidelines corrosion coupons in each cycle:
[34], the concentrations of impurities in the current study were set as 3
% O2, 100 ppm SO2 and saturated H2O to represent a typical CO2-rich
flue gas without further purification. The testing system includes gas
resources, a 5.5 L autoclave and a heating system. Corrosion coupons Table 2
The strain level of removed SCC samples after different exposing periods.
accommodated on a PTFE sample holder, bent beams with different
strain levels (0.15 %, 0.4 %, 0.5 %, 0.6 %), and tensile specimens with 120 h 240 h 360 h 480 h
0.4 % strain level were placed in the autoclave. After being sealed, the Bent beam 0.4 % 0.4 % 0.4 % 0.15 %, 0.4 %, 0.5 %, 0.6 %
autoclave was purged with CO2 for several cycles to remove the air, and Tensile specimen – – – 0.4 %
5.3 g CO2-saturated water was added into the autoclave through the gas

Fig. 2. Experiment setup for s-CO2 corrosion test.

4
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

8.76 × 104 × ΔW
CR = (3)
Sρt

where Δ W is the mass loss value, g; S is the exposed surface area of each
coupon, cm2; ρ is the density of X65 steel, 7.87 g/cm3; t is the corrosion
testing time, h.
Macroscopic images of corrosion coupons and SCC specimens were
taken both before and after corrosion tests. XRD identified the phase
compositions of the corrosion products by Bruker D8 diffractometer
with a Co-kα X-ray source. SEM/EDS techniques were conducted by FEI/
Philips XL-30 to characterize the morphology and chemical composition
of corrosion products from both surface and cross-sectional views. The
porosity of the corrosion products was statistically measured by calcu­
lating the area ratio of pores/corrosion products in the cross-section
region using Image Pro Plus software [37,38]. The cracks on SCC sam­
ples were also characterized by FIB with liquid gallium source and
gallium ions up to 30 k eV at the sample [39]. Finely focused gallium
ions as primary beam examined the corroded sample surface like con­
ventional SEM. Relative heavy gallium ions bombarded atoms in the
selected area to reveal the cross-sectional morphology of cracks.

3. Results and discussion

3.1. Corrosion behavior of X65 steel

Fig. 3 shows the macroscopic images of X65 steel corrosion coupons

before corrosion and after different exposure periods in H2O-saturated s-
CO2 with 100 ppm SO2 + 3 % O2 impurities at 45 ◦ C and 10 MPa. The
freshly prepared steel showed metallic luster. After the first corrosion
exposure, all the coupons turned brown, and yellow spherical corrosion
products appeared on the sample surfaces. As the exposure time
increased from 120 to 480 h, the sample color turned darker and more
spherical products merged on the surface, indicating the intensified
corrosion of X65 steel with time.
SEM was employed to reveal more details of the formed corrosion
products at higher magnifications and the results are shown in Fig. 4.
With extending time from 120 to 480 h, both the density and size of
spherical corrosion particles increased. Similar spherical particles were
also found on pipeline steels (X70 and X80) exposed to H2O-saturated s-
CO2 + SO2 + O2 [40,41], whose formation was thought to be related to
the condensation of water droplets on steel surfaces and the subsequent
precipitation of corrosion products within the droplet [41]. Besides, the Fig. 3. Macroscopic morphologies of X65 steel exposed to H2O-saturated s-CO2
corrosion layer was porous and some cracks were found. with 100 ppm SO2 and 3 % O2 impurities at 10 MPa and 45 ◦ C for different
Fig. 5 shows the cross-sectional SEM images and EDS analyses of X65 testing periods.
steels after different periods, which depict the structures and chemical
compositions of the corrosion layer and its evolution with time. On each and 480 h show the peaks of FeSO4⋅4H2O (PDF# 16–0699), FeCO3
sample, the corrosion layer is basically composed of a continuous layer (PDF# 29–0696) and FeOOH (PDF# 29–0713) as formed corrosion
and spherical corrosion products over it. According to EDS line scan compounds, and those of Fe from the steel substrate. The corrosion
results, both the continuous layer and spherical corrosion products products of FeSO4⋅4H2O, FeCO3 and FeOOH were also found by Sun
(marked by red dash circles) were composed of Fe, O and S. C was not et al. [44] on X65 steels after exposure to water-saturated s-CO2 + 100
shown due to its light element nature during EDS analysis. The thickness ppm SO2 + 1000 ppm O2. Besides, both Sun et al. [44] and Hua et al.
of the continuous corrosion layer on X65 steels increased almost linearly [45] also observed the presence of minor FeSO3 within the corrosion
with time (from 4 μm at 120 h to 14 μm at 480 h) and the spherical products after exposing X65 steels to SO2-containing s-CO2 environ­
products also show a growing trend with larger diameters, as shown in ments. The absence of FeSO3 in the current study could be explained by
Fig. 5. Cracks and pores could be found within the continuous layer, the high O2 content (3%) and the resulting oxidation of FeSO3 into
which are believed to provide the essential channels for the inward FeSO4 and FeOOH. With increasing time from 120 to 480 h, the intensity
transportation of reactants and outward transportation of corrosion of substrate Fe becomes lower, indicating the thickening of corrosion
products during the corrosion process [42]. The porosity of the contin­ layer and the accumulation of corrosion products. In addition, the
uous layer after different testing periods was counted using the image dominant corrosion products shifted from FeCO3 to FeOOH.
software mentioned above, which generally fell in the range of 16 % – In s-CO2 environments, the corrosion products could be formed
18 % and did not vary too much with time [37]. Further examination through two different pathways: gaseous chemical reaction and aqueous
shows that the interface of continuous layer and substrate was flat and electrochemical reaction [42,46]. The former refers to direct chemical
uniform on all the steels. No localized corrosion like pitting was reactions between the steel and chemical species at the interface of
observed, which has also been reported by other researchers in s-CO2 solid/gas interface, while the latter includes partial anodic and cathodic
environments [43]. processes in the surface condensed phase, featuring the transfer of
XRD spectra (Fig. 6) of the oxides on X65 steels corroded after 120

5
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Fig. 4. SEM images of X65 steels after being exposed to H2O-saturated s-CO2 with 100 ppm SO2 and 3 % O2 impurities at 10 MPa and 45 ◦ C for different testing
periods: (a) 120 h, (b) 240 h, (c) 360 h, and (d) 480 h.

electrons among different anodic/cathodic sites on the steel surface reaction path [43]:
[47]. By comparing the corrosion rates of steels obtained in gaseous and O2 +4H+ +4e− →2H2 O (Reaction 6)
aqueous environments with SO2, our former study found that the Besides, with extended time, O2 will further oxidize SO2− 2−
3 into SO4 ,
dominant pathways in SO2-containing s-CO2 environments are very which will react with Fe2+ to generate FeSO4. In the presence of abun­
likely to be electrochemical reactions [27]. In this study, theoretical dant H2O and O2, FeSO4 will be further oxidized into FeOOH and H2SO4.
calculation by OLI, a commercial software that provides computational The formed H2SO4 will corrode substrate Fe again, and the whole pro­
physical/chemical properties of the s-CO2 systems [44], shows that the cess is depicted as acid regeneration cycle [14,52]:
solubility of H2O in a 10 MPa and 45 ◦ C s-CO2 is 4344 ppm, and the 2SO2− 2−
3 +O2 →2SO4 (Reaction 7).
addition of 3 % O2 + 100 ppm SO2 reduces the value to 3366 ppm. It 2+ 2−
Fe +SO4 →FeSO4 (Reaction 8).
implies that in the current environment, condensation of H2O on steel 4FeSO4 +6H2 O +O2 →4FeOOH +4H2 SO4 (Reaction 9).
surface is more likely to take place with the introduction of these im­ As schematically shown in Fig. 7, these reactions continuously
purities. In addition, the formation of spherical corrosion products and generate FeOOH in a circulating path.
their growing trend with extended time also serve as the evidence of With extended time, a considerable amount of FeOOH will form and
liquid phase formation during corrosion. accumulate as a porous corrosion layer. Besides, the generated H2SO4 in
The electrochemical theory depicts the corrosion process as a series Reaction 9 may dissolve some unstable corrosion products like FeCO3 in
of electrochemical reactions in liquid phase. In the beginning, a thin the corrosion layer. Thus, the acid regeneration cycle could explain the
liquid film or water droplet formed on the steel surface due to chemical reduced amounts of FeCO3 and the shift of dominant corrosion product
coagulation, adsorption or condensation, which provided the necessary from FeCO3 to FeOOH with time extended from 120 to 480 h (Fig. 6).
electrolyte for electrochemical reactions to occur. Then, the dissolution Due to H2SO4-induced Fe/FeCO3 dissolution, more channels are gener­
of CO2 and SO2 acidified the solution by generating H+ and other related ated for the diffusion of other corrosive components into the substrate
anions (HCO–3, CO2– 2−
3 , HSO3 , and SO3 ) as by-products [48]. The sub­
−
steel. The generated FeSO4 has high hygroscopicity, which enhances the
strate iron was corroded through the following anodic reaction [49]: water adsorption behavior of the corrosion products [53]. This may help
Fe→Fe2+ +2e− (Reaction 1) to collect more water as electrolyte on steel surface for the occurrence of
In such acidified environment, proton could be reduced via cathodic electrochemical reactions. Because of these factors, the steel continu­
reaction, and the hydrogen atom could combine as hydrogen gas: ously experienced severe corrosion despite the extended time and
H+ +e− →H (Reaction 2) accumulation of corrosion products.
H +H→H2 (Reaction 3) In order to obtain the corrosion rates of X65 steel exposed to H2O-
Iron ions will combine with carbonate/sulfite to form iron carbon­ saturated s-CO2 system containing 100 ppm SO2 + 3 % O2, mass loss
ates/sulfites, which precipitate on the steel surface as corrosion products measurements were performed and values were found to be increased
once their concentrations are beyond the corresponding solubilities in linearly with time, i.e., from 4.07 mg/cm2 at 120 h to 13.99 mg/cm2 at
the liquid phase [10,13,14,50,51]: 480 h. Based on mass loss values and Eqn 3, the corrosion rate was
Fe2+ +CO2− 3 →FeCO3 (Reaction 4) calculated as a function of time and shown in Fig. 8. It is found that the
Fe2+ +SO2−3 →FeSO3 (Reaction 5) corrosion rate generally falls in the range of 0.3–0.4 mm/y, and
Due to the high content of CO2 in the environment, FeCO3 is the exhibited a slight decreasing trend from 0.37 mm/y at 120 h to 0.32
dominant corrosion product in the initial period, as shown by the XRD mm/y at 480 h. This shows a good correspondence to the darker color of
spectrum at 120 h (Fig. 6). the corroded surface (Fig. 3) and linearly increased oxide layer thickness
O2 accelerates corrosion by providing an additional cathodic with time. Based on Wanger’s theory, the linear mass change with

6
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Fig. 5. Cross-sectional SEM images and corresponding EDS line scans of X65 steels exposed to H2O-saturated s-CO2 with 100 ppm SO2 + 3 % O2 impurities at 10 MPa
and 45 ◦ C for different testing periods: (a) 120 h, (b) 240 h, (c) 360 h, (d) 480 h.

increasing time implies that the corrosion process is controlled by phase- the replenishment of fluids during tests. For instance, without s-CO2
boundary reaction instead of diffusion through the corrosion scale [54]. fluid replenishment, the corrosion rate of X70 steel decreased dramati­
It indicates that in H2O-saturated s-CO2 with SO2 and O2, the formed cally from an initial value of 1.95 mm/y over the first 24 h to 0.70 mm/y
corrosion products would fail to provide sufficient protection for the after 192 h in H2O-saturated s-CO2 + NO2 + O2 + SO2 [42]. With
substrate metal and severe corrosion is expected. replenishment after 48 h, the corrosion rate of pipeline steels only
To develop a long-term corrosion rate prediction model in SO2- slightly decreased from 0.71 to 0.65 mm/y when exposed to 8 MPa H2O-
containing s-CO2 environments, all of the public and in-house data saturated s-CO2 + SO2 + O2 [50]. To quantitively measure the effect of
including current study were collected and summarized in Table 3. It is fluid replenishment on corrosion rate evolution with time, a concept
found that the variation of corrosion rate with time is greatly affected by named corrosion rate change (CRC) is proposed and defined as the

7
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

change of corrosion rate within unit time:

(CRi − CRf ) 1
CRC = × × 100% (4)
CRi ti − tf

where CRC has a unit of %/h; ti and tf (h) are the initial and final time at
which the corrosion rates are measured; CRi and CRf (mm/y) are the
corresponding corrosion rates at time ti and tf, respectively. All the CRC
values in Table 3 are positive, indicating that the corrosion rates in SO2-
containing s-CO2 environments tend to reduce with time. Further
analysis found that the replenishment of fluid on a basis of 48–120 h will
lead to a relatively low CRC value (<0.2%/h), while that without
replenishment has a much higher value (up to 1.74%/h). In non-
replenished s-CO2 tests, the gradually reduced corrosion rates and
high CRC values could be attributed to the quick consumption of im­
purities with extended time rather than the protective effect of formed
corrosion products [50]. Thus, for future long-term studies, periodical
replenishment of impurities or a once-through system should be
employed to avoid the obvious deviation of impurity concentrations
from designed values so as to accurately measure the corrosion rates.
Fig. 6. XRD spectra of the residual oxides on X65 coupons after corrosion for Since the variation of corrosion rate with time, namely CRC value,
120 and 480 h in H2O-saturated s-CO2 system containing 100 ppm SO2 + 3% O2 basically reflects the protectiveness of the formed corrosion products on
at 10 MPa and 45 ◦ C. sample surface, it also indicates that SO2, together with other impurities
in s-CO2 streams, will cause great corrosion damage on pipeline steels
and the formed corrosion products offer limited protection.
To set up a prediction model, the corrosion rates of pipeline steels
collected from different references [41,42,45,50,55] are plotted as a
function of time and fitted by power function. This model is often used to
predict the corrosion behavior of low-alloy steels exposed to an atmo­
spheric environment, which shares certain corrosion features with SO2-
containing s-CO2 since in both environments steels are corroded under
thin and acidified surface liquid film [56]. Based on ASTM G101-04, the
relationship between corrosion rate and time could be described as [57]:

CR = AtB (5)
Fig. 7. Schematic diagram of acid regeneration cycle.
where CR is the corrosion rate after t hours of exposure, A and B are
constants. In Fig. 9, fitting results show that power function model
precisely depicts the variation of corrosion rate with time in SO2-

Fig. 8. Corrosion rate of X65 steel as a function of testing time in H2O-saturated s-CO2 with 100 ppm SO2 + 3 % O2 at 10 MPa and 45 ◦ C.

8
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Table 3
Experimental conditions of corrosion studies regarding SO2-containing s-CO2 and the calculated CRC.
Ref. Impurity concentration (ppm) Replenish of fluid Corrosion rate (mm/y) at certain time CRC (%/h)

H2O O2 SO2 NO2 others Initial final

[45] s 1000 100 – – – 3.51 (6 h) 0.94 (48 h) 1.74

[42] s 1100 20,000 – – – 1.95 (24 h) 0.7 (192 h) 0.38
[41] s 500 500 500 – – 16.8 (24 h) 1.55 (480 h) 0.39
[50] s 20 100 – – Every 48 h 0.71 (48 h) 0.65 (96 h) 0.18
[55] 1000 8100 70 100 750-CO Every 72 h 0.026 (168 h) 0.0025 (4464 h) 0.02
Current s 30,000 100 – – Every 120 h 0.38 (120 h) 0.32 (480 h) 0.04

s: saturated.

Fig. 9. The fitted corrosion rates in different SO2-containing s-CO2 environments by power function.

containing s-CO2 environments since the Adj. R-Square values are close with a corrosion rate higher than 5 mpy. Although the power function
to 1. Further analyses reveal that the corrosion rate is generally affected prediction model indicates that corrosion rates could gradually decrease
by several factors, including SO2 content, the addition of NO2, and H2O with extended time (Fig. 9), expecting the corrosion rate being reduced
content. While increasing SO2 content from 100 to 20000 ppm could to the acceptable low value with time is unrealistic since it may take very
increase the corrosion rate [42,45], the addition of 500 ppm NO2 has a long. In addition, the porous corrosion products may easily be peeled off
more pronounced effect [41]. In addition, corrosion could be greatly due to the flowing s-CO2-induced interrupting stress, which may further
inhibited by controlling H2O content to a certain point below the satu­ impair the protectiveness of the corrosion products and increase corro­
rated point [55]. As for O2, while the previous study found that its sion damage. Thus, the more practical corrosion control strategy is
introduction into SO2-containing s-CO2 stream will greatly increase the limiting the impurities below a certain level. Based on the above ana­
corrosion rate [13], the current study shows that for the same amount of lyses, it is likely that reducing water content below the critical point will
SO2 (100 ppm), further increasing O2 content from 20 to 30000 ppm be an effective way since the corrosion mechanism might be shifted from
does not introduce a remarkable change in the corrosion rate [45,50]. electrochemical reaction to gas chemical reaction, after which the
Therefore, in H2O-saturated s-CO2 + O2 + different SO2 contents, the corrosion rates are much lower [55,59]. With timely fluid replenish­
empirical model depicting the variation of pipeline steel corrosion rates ment, further studies are still needed to confirm the maximum allowable
with time could be summarized as: impurity concentrations and control the degradation rate of pipeline
for H2O-saturated s-CO2 + O2 + 100 ppm SO2,.CR = steels to an acceptable value from a long-term point of view.
11.16116x− 0.64556
for H2O-saturated s-CO2 + O2 + 20000 ppm SO2,.CR =
3.2. SCC studies in SO2-containing s-CO2 environment
5.61035x− 0.31258
Currently, no generally acceptable corrosion rate of pipeline steels
3.2.1. The applied strain levels on SCC samples
for s-CO2 transportation has been purposed, and 5 mpy (0.127 mm/y) is
Due to the high pressure of the testing system, SCC tests were con­
a permissible value since it is a widely used criterion for pipeline
ducted by constant strain method. The pre-loaded strain levels on X65
corrosion in oil and gas industry [58]. The corrosion rates from both
samples by FPB method cover both elastic and plastic zone. Based on the
previous publications (Table 1) and the current study (Fig. 8) indicated
stress-strain curve of X65 steel (shown in Fig. 10(a)), the yield point of
that H2O-saturated s-CO2 with SO2 could cause severe corrosion damage
X65 steel (0.2 % off-set) corresponds to a strength of 450 MPa and a

9
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Fig. 10. (a) Stress–strain curve of X65 steels and the designed strain levels for SCC tests; (b) DIC-measured strain maps of bent beams and tensile specimen; (c) FEA
simulation of von Mises strain distribution of X65 bent beam with 0.6% strain level.

strain level of 0.4 %. From the safety and integrity points of view,
pipelines are suggested to be operated in elastic deformation zone, and
to be more specific, below 72 % of the yield strain [60]. However, during
practical service, pipeline steels may experience plastic deformation
(≥100 % yield strain) due to external mechanical force or localized
damage by corrosion, which increases the susceptibility to SCC [61].
Therefore, in current study, the applied strains were set as 0.15 %, 0.4 %,
0.5 % and 0.6 %, corresponding to 35 %, 100 %, 125 % and 150 % of the
yield strain, respectively (shown by four red dots in Fig. 10(a)).
To pursue a higher accuracy on measurement, it is crucial to confirm
that the loaded strain on SCC sample is equal to or at least close to the
designed value. For tensile specimen, the applied strain (0.4 %) could be
confirmed directly by the extensometer during loading. For FPB bent
beam, the applied strains (0.15 %, 0.4 %, 0.5 % and 0.6 %) on sample
surfaces were measured by DIC and simulated by FEA. Fig. 10(b) pre­
sents the DIC results, and the strain values obtained from the central
surface area of each sample are reflected by the color change on strain
maps. Obviously, both the FPB and tensile devices could accurately load
designed strains on the SCC samples. For FEA simulation results (Fig. 10
(c)), the von Mises strain distribution in the bent beam with a designed Fig. 11. Mass gain of corrosion coupons and bent beams exposed to H2O-
strain level of 0.6 % is presented as a typical example. It is found that the saturated s-CO2 with 100 ppm SO2 + 3 % O2 impurities at 10 MPa and 45 ◦ C for
strain value of bent beam gradually decreased from the topmost surface different testing periods.
to the inner part. On the sample surface, the strain level as ~0.56 % is
very close to the designed value. One possible explanation for the slight be related to the oxide spallation since the formed corrosion layer is
deviation is that the Johnson cook model, which is used in FEA simu­ fragile. Several previous studies showed that the applied stress could
lation, may lose certain accuracy with higher strain levels in the plastic intensify corrosion by increasing the surface activity of steels and pro­
region [62]. moting the interactions between corrodents and the steel surface [63].
However, the high corrosivity induced by the synergistic effects of SO2,
3.2.2. SCC susceptibility evaluation H2O and O2 in s-CO2 may result in the quick formation of thick corrosion
To clarify the effect of applied stress on corrosion behavior of X65 products that cover the active site on the steel surface [64]. Due to the
steels, the mass gain values of corrosion coupons and bent beams with high corrosivity and reaction kinetics, the tested steels experience severe
different strain levels are listed and compared in Fig. 11. Generally, the corrosion at a very high speed and the contribution from stress might be
mass gain values of two types of samples show no obvious difference negligible.
after the same exposure time, and the expected higher mass gain was not Besides corrosion degree, it is also found that the applied strain
found on bent beams. After 480 h of exposure, the obviously lower mass didn’t affect the phase constituents of formed corrosion products on X65
gain of FPB bent beam with 0.4 % strain level than other samples might steels. Based on XRD characterization (Fig. 12), the corrosion products

10
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

formed on both bent beams and tensile specimens after 480 h are dissolution of metal at the crack tips [66]. Metal exposed to corrosion
identified as FeCO3, FeOOH, FeSO4⋅4H2O, which is consistent with the environment tends to develop a passive layer at the surface. However,
corrosion coupons. localized breakdown of the passive layer may result in the formation of
Fig. 13 shows the photographic images of all the strained samples cracks, and the chemical and mechanical effects will keep the metal at
after the exposures to H2O-saturated s-CO2 with 100 ppm SO2 + 3 % O2. the crack tip dissolved at an accelerated speed, leading to crack propa­
As the testing time increased from 120 to 480 h, the surface of bent gation [67]. Hydrogen-induced cracking theory credits the propagation
beams turned darker and more yellow particles appeared, which shares of crack to the cathodically generated hydrogen. In wet environments,
the same trend with corrosion coupons (Fig. 3). For both bent beams and the dissolved acid gases (SO2/CO2) in water form corresponding acids
tensile specimens after 480 h’ corrosion, no obvious difference in surface and generate H+ through dissociation. The H+ may be reduced to
color was observed, despite of the variation on loading method and hydrogen through localized cathodic reactions, then diffusing into the
strain value. Besides, no visible crack was found on the surfaces of tested metal substrate, occupying the interstitial locations or being trapped at
specimens from all the tests, indicating the low SCC susceptibility of X65 local defects (like dislocations, grain boundaries). With extended time,
steels in the environment. the accumulation of hydrogen becomes more pronounced, thus
Fig. 14 shows the SEM surface morphologies of bent beams and degrading the mechanical property of the metals and triggering cracking
tensile specimens in H2O-saturated s-CO2 containing 100 ppm SO2 + 3 at a relatively lower stress level [68,69]. In addition, as the by-product
% O2. Same with the unstressed corrosion coupons (Fig. 4), the surfaces of SO2-induced corrosion, FeSO4 in the corrosion product film could
of bent beams and tensile specimens were covered with spherical facilitate the cyclic regeneration of sulfuric acid (Reactions 7–9) as well
corrosion products. In addition, cracks with several hundreds of mi­ as the hydrogen evolution, which further intensifies the anodic disso­
crometers in length could be found within the corrosion products. lution and hydrogen-induced cracking [21].
Further cross-sectional SEM inspection (Fig. 15) on the bent beams and In the current testing system, due to the acidified environment and
tensile samples shows that the cracks were found only within the formation of massive corrosion products, both enhanced anodic disso­
corrosion products and didn’t penetrate into the substrate. Therefore, lution and hydrogen-induced cracking might take place. Yet results of
the formation of these cracks may be related to the brittle nature of the current study show that no SCC cracking was found on the substrate of
corrosion scale, and X65 steels are free from SCC cracks in the current the strained X65 samples in all tests, despite the variation on loading
testing environment [50]. levels (0.15–0.6 %) and testing time (120–480 h). For bent beams, the
To better examine whether the cracks in the corrosion products FEA simulation of strain distribution (Fig. 10(c)) shows that the strain
penetrated into the substrates, in-situ FIB cutting was performed on the decreased gradually from surface to substrate. Since the SCC suscepti­
surface of strained samples (both bent beams and tensile specimens) bility is reduced with lower strain levels [65], it is questioned whether
after the exposure to H2O-saturated s-CO2 + 100 ppm SO2 + 3 % O2 for the absence of SCC crack in the steel substrate is caused by the reduced
120 or 480 h. Fig. 16(a, c and e) shows that the formed surface cracks are strain levels as corrosion proceeds inwardly. As shown in Fig. 15, the
generally several hundred micrometers in length and several microme­ corrosion layer has a thickness of <80 μm. Based on Fig. 10(c), X65 steel
ters in width. FIB cuttings were performed on the red rectangular areas in this zone is still in plastic deformation with a strain level of ~0.55 %.
with bombarded Ga+ ions. The cross-sectional images after cutting It indicates that the formed corrosion layer is not thick enough to
(Fig. 16(b, d and f)) show all the cracks were developed only within the signally change the strain level in the substrate. In addition, for the
corrosion product layers, which further proves the low susceptibility of tensile specimens whose strain distribution is believed to be homoge­
X65 steel in current testing environments despite of the variation on neous, cross-sectional examination also shows that the underneath
strain level, way of strain loading (FPB/tensile) and testing time. substrate is free from crack.
SCC is depicted as the sudden failure of susceptible materials under Since the way of loading does not account for the low SCC suscep­
the synergistic effect of corrosive environment and applied tensile stress. tibility of X65 steels, another two explanations from anodic and cathodic
In SO2/CO2-containing environments, three mechanisms, including aspects are purposed. The first is the inhibited hydrogen embrittlement
enhanced anodic dissolution, hydrogen-induced cracking and their from the cathodic side. In the acidified environment, the presence of O2
combination, have been proposed to elucidate SCC behavior of pipeline in the s-CO2 could shift the dominant cathodic reaction from hydrogen
steels [65]. Anodic dissolution attributes the development of crack to the evolution (Reaction 2) to oxygen reduction (Reaction 6) due to the
higher standard electrode potential of the latter [70], and the amount of
H that may participate in hydrogen-induced cracking will be signifi­
cantly reduced [71]. The beneficial effect of O2 on inhibiting SCC in
acidified environments was also confirmed by other researchers. Li et al.
[72] found that the SCC susceptibility of X70 steel is inversely propor­
tional to the concentration of dissolved O2 in NS4 solution since the
dissolved oxygen could react with H to reduce hydrogen embrittlement.
Lunde et al. [73] also revealed that increasing O2 content from 10 to 50
ppm could alter the cathodic reaction to oxygen reduction and reduce
the SCC susceptibility of pipeline steels (St.52.4 and St.35.4) in the
aqueous solution. Even hydrogen atoms are reduced and then diffuse
into the substrate, there is still a possibility that the rate of hydrogen
accumulation cannot match that of corrosion. Before the permeated
hydrogen in the underneath substrate reaches a critical concentration to
trigger SCC crack, the corrosion front exceeds and turns the substrate
into porous corrosion products, thus releasing the accumulated
hydrogen into the nearby environment.
Besides, the lack of pitting as stress concentrators might be another
reason from the anodic side. The SCC cracks tend to initiate from
localized structural defects, like pitting, due to the concentrated stress
Fig. 12. XRD spectra of the residual oxides on X65 corrosion coupons (C), FPB level in these places [20]. Pitting is depicted as the initiation of localized
bent beam (F) and tensile specimen (T) after corrosion for 480 h in H2O-satu­ corrosion underneath the passive film and the following vertical
rated s-CO2 containing 100 ppm SO2 + 3 % O2 at 10 MPa and 45 ◦ C. development in one direction, which was mainly driven by the

11
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Fig. 13. Photographic morphologies of corrosion scales on X65 steel bent beams and tensile specimens after exposing to the H2O-saturated s-CO2 with 100 ppm SO2
+ 3 % O2 at 10 MPa and 45 ◦ C.

Fig. 14. SEM images of strained X65 steels (bent beams and tensile specimen) exposed to H2O-saturated s-CO2 + 100 ppm SO2 + 3 % O2 impurities at 10 MPa and
45 ◦ C for different testing periods. FPB: Four point bending bent beams; Tensile: Tensile specimen.

hydrolysis of dissolved metal ions and the continuous self-catalytic re­ the initial pit site. In the current testing environment, SO2 dissolves in
actions within the pitting cavity [25]. In current tests, cross-sectional the condensed water and forms H2SO4 as strong acid, and the pH of the
SEM images of strained X65 steels (Fig. 15) revealed that the interface condensed phase in H2O-saturated s-CO2 + 100 ppm SO2 + 3 % O2
of corrosion product/substrate is flat and free from pitting, which might calculated by OLI Analyzer software is as low as 2.24. In this way, the pH
be related to the special corrosion environment of the SO2-containing s- value of the whole surface water film is decreased and the localized
CO2 stream. First, in the formed droplet or a thin film solution on steel acidification in the initial pit site is less prominent. In addition, the
surface as a confined environment, the low ionic migration rate might porous structure of the corrosion products facilitates the diffusion of
make the inward ingression of corrosive ions and outward diffusion of corrosive components into the steel substrate, and the micro-
corrosion products less prominent, thus inhibiting the pitting growth environment underneath corrosion layer is unlikely to support the so-
[41]. Second, the localized acidification by metal ion hydrolysis in called autocatalytic effect [74]. Instead, the substrate steel only expe­
initial pit cavities is inhibited due to the formed strong acids in sur­ rienced active dissolution uniformly, and no SCC cracks were found.
rounding environments. The development of pitting relies on the
inconsistency of pH between the external and localized environment of

12
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

Fig. 15. Cross-sectional SEM images of SCC X65 steels (bent beams and tensile specimen) exposed to H2O-saturated s-CO2 + 100 ppm SO2 + 3 % O2 at 10 MPa and
45 ◦ C for different testing periods.

of cracks and pores, which could provide the diffusion path of cor­
rosive components. Therefore, the corrosion rates of X65 steel in the
current testing environment are almost constant despite the pro­
longed testing time.
• A comparison between current study and data from literature shows
that a timely replenishment of s-CO2 fluid with impurities is crucial
for accurate corrosion rate measurement. In addition, increasing SO2
content, the addition of NO2 and controlling H2O content could
effectively affect the long-term corrosion rates. For H2O-saturated s-
CO2 + O2 + SO2, the variation of O2 didn’t affect the corrosion rate
too much, and power-function-based empirical models depicting the
variation of corrosion rates with time have been built.
• The SCC susceptibility of X65 steels in SO2-containing s-CO2 media is
relatively low, which is likely due to the inhibited hydrogen-induced
cracking and/or the lack of pitting as a localized stress concentrator.

CRediT authorship contribution statement

Kaiyang Li: Data curation, Formal analysis, Investigation, Method­

ology, Validation, Visualization, Writing – original draft. Yimin Zeng:
Conceptualization, Funding acquisition, Methodology, Project admin­
istration, Resources, Supervision, Writing – review & editing,
Validation.

Declaration of Competing Interest

Fig. 16. Surface and cross-sectional images of cracks on strained samples after
different testing periods: (a, b) bent beam with 0.4 % strain after 120 h; (c, d) The authors declare that they have no known competing financial
bent beam with 0.6 % strain after 480 h; (e, f) tensile specimen with 0.4 % interests or personal relationships that could have appeared to influence
strain after 480 h. the work reported in this paper.

4. Conclusions Data availability

In the present work, the long-term corrosion behavior and SCC sus­ Data will be made available on request.
ceptibility of X65 steel in H2O-saturated s-CO2 with 100 ppm SO2 + 3 %
O2 were investigated. Based on the above results, comprehensive liter­ Acknowledgements
ature review and analyses, and detailed discussions, the following con­
clusions are made: This study was funded by Natural Resource Canada OERD CO2
Capture, Utilization and Storage (CCUS) program. The authors grate­
• SO2 can induce severe corrosion of pipeline steels in H2O-saturated s- fully acknowledge the technical support by CMAT staff, including Shi
CO2 + O2 with the corrosion products of iron sulphates, carbonates Chao from corrosion lab, Jie Liang from mechanical testing lab, Pei Liu
and oxy-hydroxides. With extended time, the dominant corrosion and Jian Li from advanced microscopy lab.
products shifted from iron carbonates to iron oxy-hydroxides due to
acid regeneration cycle. The non-protective corrosion layer has a lot

13
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

References [32] A. Shrot, M. Bäker, Determination of Johnson-Cook parameters from machining

simulations, Computational Materials Science 52 (1) (2012) 298–304.
[33] B.N. Clark, R. Rentz, J.D. McColskey, J.W. Sowards. Using Digital Image
[1] Y. Hua, S. Mohammed, R. Barker, A. Neville, Comparisons of corrosion behaviour
Correlation to Improve Stress-Corrosion Cracking Evaluation by NACE TM0177-B,
for X65 and low Cr steels in high pressure CO2-saturated brine, Journal of Materials
CORROSION 2017, NACE. 2017.
Science & Technology 41 (2020) 21–32.
[34] R.T. Porter, M. Fairweather, M. Pourkashanian, R.M. Woolley, The range and level
[2] V.E. Onyebuchi, A. Kolios, D.P. Hanak, C. Biliyok, V. Manovic, A systematic review
of impurities in CO2 streams from different carbon capture sources, International
of key challenges of CO2 transport via pipelines, Renewable and Sustainable Energy
Journal of Greenhouse Gas Control 36 (2015) 161–174.
Reviews 81 (2018) 2563–2583.
[35] A. Bamberger, G. Sieder, G. Maurer, High-pressure (vapor+ liquid) equilibrium in
[3] X. Hu, A. Neville, CO2 erosion–corrosion of pipeline steel (API X65) in oil and gas
binary mixtures of (carbon dioxide+ water or acetic acid) at temperatures from
conditions—A systematic approach, Wear 267 (11) (2009) 2027–2032.
313 to 353 K, The Journal of Supercritical Fluids 17 (2) (2000) 97–110.
[4] G.A. Zhang, Y.F. Cheng, On the fundamentals of electrochemical corrosion of X65
[36] ASTM, G1-03 Standard Practice for preparing, cleaning, and evaluating corrosion
steel in CO2-containing formation water in the presence of acetic acid in petroleum
test specimens. 2003.
production, Corrosion Science 51 (1) (2009) 87–94.
[37] W. Niu, C. Bai, G. Qiu, Q. Wang, Processing and properties of porous titanium using
[5] R. Barker, Y. Hua, A. Neville, Internal corrosion of carbon steel pipelines for dense-
space holder technique, Materials Science and Engineering: A 506 (1–2) (2009)
phase CO2 transport in carbon capture and storage (CCS)–a review, International
148–151.
Materials Reviews 62 (1) (2017) 1–31.
[38] A. Terzić, L. Pavlović, Correlation among sintering process, porosity, and creep
[6] L. Wei, Y. Zhang, X. Pang, K. Gao, Corrosion behaviors of steels under supercritical
deformation of refractory concrete, Journal of materials science 44 (11) (2009)
CO2 conditions, Corrosion Reviews 33 (3–4) (2015) 151–174.
2844–2850.
[7] Y. Xiang, M. Xu, Y.-S. Choi, State-of-the-art overview of pipeline steel corrosion in
[39] J. Li, M. Elboujdaini, V. Gertsman, M. Gao, D. Katz, Characterization of a stress
impure dense CO2 for CCS transportation: mechanisms and models, Corrosion
corrosion crack with FIB and TEM, Canadian metallurgical quarterly 44 (3) (2005)
Engineering, Science and Technology 52 (7) (2017) 485–509.
331–338.
[8] W. McLEOD, R.R. Rogers, Sulfurous acid corrosion of low carbon steel at ordinary
[40] Y. Xiang, Z. Wang, Z. Li, W. Ni, Effect of temperature on corrosion behaviour of
temperatures-I, Its nature, Corrosion 22 (5) (1966) 143–146.
X70 steel in high pressure CO2/SO2/O2/H2O environments, Corrosion Engineering,
[9] M. Xu, Q. Zhang, X. Yang, Z. Wang, J. Liu, Z. Li, Impact of surface roughness and
Science and Technology 48 (2) (2013) 121–129.
humidity on X70 steel corrosion in supercritical CO2 mixture with SO2, H2O, and
[41] Y. Xiang, C. Li, W. Hesitao, Z. Long, W. Yan, Understanding the pitting corrosion
O2, The Journal of Supercritical Fluids 107 (2016) 286–297.
mechanism of pipeline steel in an impure supercritical CO2 environment, The
[10] Y.-S. Choi, S. Nesic, D. Young, Effect of impurities on the corrosion behavior of CO2
Journal of Supercritical Fluids 138 (2018) 132–142.
transmission pipeline steel in supercritical CO2− water environments,
[42] Y. Xiang, Z. Wang, Z. Li, W. Ni, Effect of exposure time on the corrosion rates of
Environmental science & technology 44 (23) (2010) 9233–9238.
X70 steel in supercritical CO2/SO2/O2/H2O environments, Corrosion 69 (3) (2012)
[11] C. Sun, Y. Wang, J. Sun, X. Lin, X. Li, H. Liu, X. Cheng, Effect of impurity on the
251–258.
corrosion behavior of X65 steel in water-saturated supercritical CO2 system, The
[43] M. Xu, Q. Zhang, Z. Wang, J. Liu, Z. Li, Cross impact of CO2 phase and impurities
Journal of Supercritical Fluids 116 (2016) 70–82.
on the corrosion behavior for stainless steel and carbon steel in water-containing
[12] C. Sun, J. Sun, Y. Wang, P. Sui, X. Lin, H. Liu, X. Cheng, M. Zhou, Effect of impurity
dense CO2 environments, International Journal of Greenhouse Gas Control 71
interaction on the corrosion film characteristics and corrosion morphology
(2018) 194–211.
evolution of X65 steel in water-saturated supercritical CO2 system, International
[44] J. Sun, C. Sun, Y. Wang, Effects of O2 and SO2 on water chemistry characteristics
Journal of Greenhouse Gas Control 65 (2017) 117–127.
and corrosion behavior of X70 pipeline steel in supercritical CO2 transport system,
[13] C. Sun, J. Sun, Y. Wang, X. Lin, X. Li, X. Cheng, H. Liu, Synergistic effect of O2, H2S
Industrial & Engineering Chemistry Research 57 (6) (2018) 2365–2375.
and SO2 impurities on the corrosion behavior of X65 steel in water-saturated
[45] Y. Hua, R. Jonnalagadda, L. Zhang, A. Neville, R. Barker, Assessment of general
supercritical CO2 system, Corrosion Science 107 (2016) 193–203.
and localized corrosion behavior of X65 and 13Cr steels in water-saturated
[14] Y. Xiang, Z. Wang, C. Xu, C. Zhou, Z. Li, W. Ni, Impact of SO2 concentration on the
supercritical CO2 environments with SO2/O2, International Journal of Greenhouse
corrosion rate of X70 steel and iron in water-saturated supercritical CO2 mixed
Gas Control 64 (2017) 126–136.
with SO2, The Journal of Supercritical Fluids 58 (2) (2011) 286–294.
[46] Y. Zeng, K. Li, J. Luo, M. Arafin. Impacts of Impurities on Corrosion of Supercritical
[15] Y. Hua, R. Barker, A. Neville, The influence of SO2 on the tolerable water content to
CO2 Transportation Pipeline Steels, CORROSION 2018, NACE. 2018.
avoid pipeline corrosion during the transportation of supercritical CO2,
[47] X. Guan, D. Macdonald, Determination of corrosion mechanisms and estimation of
International Journal of Greenhouse Gas Control 37 (2015) 412–423.
electrochemical kinetics of metal corrosion in high subcritical and supercritical
[16] Y.-S. Choi, S. Nešic, Effect of impurities on the corrosion behavior of carbon steel in
aqueous systems, Corrosion 65 (6) (2009) 376–387.
supercritical CO2-water environments, CORROSION 2010, NACE. 2010.
[48] Q. Gong, Y. Xiang, J. Zhang, R. Wang, D. Qin, Influence of elemental sulfur on the
[17] W. Chen, L. Hao, J. Dong, W. Ke, Effect of sulphur dioxide on the corrosion of a low
corrosion mechanism of X80 steel in supercritical CO2-saturated aqueous phase
alloy steel in simulated coastal industrial atmosphere, Corrosion Science 83 (2014)
environment, The Journal of Supercritical Fluids 176 (2021), 105320.
155–163.
[49] Y. Xiang, C. Song, C. Li, E. Yao, W. Yan, Characterization of 13Cr steel corrosion in
[18] M. Seiersten, P. Kofstad, The high temperature corrosion of nickel in SO2 at
simulated EOR-CCUS environment with flue gas impurities, Process Safety and
500–800 oC, Corrosion Science 22 (5) (1982) 487–506.
Environmental Protection 140 (2020) 124–136.
[19] S. Oesch, The effect of SO2, NO2, NO and O3 on the corrosion of unalloyed carbon
[50] Y. Hua, R. Barker, A. Neville, Understanding the influence of SO2 and O2 on the
steel and weathering steel—The results of laboratory exposures, Corrosion Science
corrosion of carbon steel in water-saturated supercritical CO2, Corrosion 71 (5)
38 (8) (1996) 1357–1368.
(2014) 667–683.
[20] H. Ma, C. Du, Z. Liu, X. Li, Effect of SO2 content on SCC behavior of E690 high-
[51] C. Li, Y. Xiang, W. Li, Initial corrosion mechanism for API 5L X80 steel in CO2/SO2-
strength steel in SO2-polluted marine atmosphere, Ocean Engineering 164 (2018)
saturated aqueous solution within a CCUS system: Inhibition effect of SO2
256–262.
impurity, Electrochimica Acta 321 (2019), 134663.
[21] W. Wu, Z. Liu, X. Li, C. Du, Z. Cui, Influence of different heat-affected zone
[52] I.M. Allam, J.S. Arlow, H. Saricimen, Initial stages of atmospheric corrosion of steel
microstructures on the stress corrosion behavior and mechanism of high-strength
in the Arabian Gulf, Corrosion Science 32 (4) (1991) 417–432.
low-alloy steel in a sulfurated marine atmosphere, Materials Science and
[53] V. Petkova, Y. Pelovski, Investigation on The Thermal Properties of Fe2O(SO4)2.
Engineering: A 759 (2019) 124–141.
Part II, Journal of Thermal Analysis and Calorimetry 64 (3) (2001) 1037–1044.
[22] K. Gong, M. Wu, G. Liu, Stress corrosion cracking behavior of rusted X100 steel
[54] N. Birks, G.H. Meier, F.S. Pettit, Introduction to the high temperature oxidation of
under the combined action of Cl− and HSO−3 in a wet–dry cycle environment,
metals, Cambridge University Press, 2006.
Corrosion Science 165 (2020), 108382.
[55] O. Yevtushenko. Water impact on corrosion resistance of pipeline steels in
[23] M. Mohitpour, P. Seevam, K.K. Botros, B. Rothwell, C. Ennis, Pipeline
circulating supercritical CO2 with SO2-and NO2-impurities, CORROSION 2014,
transportation of carbon dioxide containing impurities, ASME, 2012.
NACE. 2014.
[24] L. Xu, Y. Cheng, Corrosion of X100 pipeline steel under plastic strain in a neutral
[56] M. Morcillo, B. Chico, I. Díaz, H. Cano, D. De la Fuente, Atmospheric corrosion data
pH bicarbonate solution, Corrosion Science 64 (2012) 145–152.
of weathering steels, A review, Corrosion Science 77 (2013) 6–24.
[25] V. Raja, T. Shoji, Stress corrosion cracking: theory and practice, Elsevier, 2011.
[57] ASTM, G101-04 Standard Guide for Estimating the Atmospheric Corrosion
[26] H. Ma, Z. Liu, C. Du, X. Li, Z. Cui, Comparative study of the SCC behavior of E690
Resistance of Low-Alloy Steels. 2015.
steel and simulated HAZ microstructures in a SO2-polluted marine atmosphere,
[58] NACE, RP0775-2005 Preparation and Installation of Corrosion Coupons and
Materials Science and Engineering: A 650 (2016) 93–101.
Interpretation of Test Data in Oilfield Operations. 2005.
[27] Y. Zeng, K. Li, Influence of SO2 on the Corrosion and Stress Corrosion Cracking
[59] Y. Xiang, Z. Wang, X. Yang, Z. Li, W. Ni, The upper limit of moisture content for
Susceptibility of Supercritical CO2 Transportation Pipelines, Corrosion Science
supercritical CO2 pipeline transport, The Journal of Supercritical Fluids 67 (2012)
108404 (2019).
14–21.
[28] H. Ma, C. Du, Z. Liu, W. Hao, X. Li, C. Liu, Stress corrosion behaviors of E690 high-
[60] M. McLamb, P. Hopkins, M. Marley, M. Nessim, A justification for designing and
strength steel in SO2-polluted marine atmosphere, Acta Metallurgica Sinica 52 (3)
operating pipelines up to stresses of 80% SMYS, International Pipeline Conference
(2016) 331–340.
(2002) 745–757.
[29] ASTM, E8 Standard test methods for tension testing of metallic materials. 2009.
[61] Y. Li, X. Guo, G. Zhang, Synergistic effect of stress and crevice on the corrosion of
[30] ASTM, D6272–02 Standard Test Method for Flexural Properties of Unreinforced
N80 carbon steel in the CO2-saturated NaCl solution containing acetic acid,
and Reinforced Plastics and Electrical Insulating Materials by Four-Point Bending.
Corrosion Science 123 (2017) 228–242.
2002.
[62] Y. Lin, X.-M. Chen, G. Liu, A modified Johnson-Cook model for tensile behaviors of
[31] NACE. TM0177-200 Laboratory testing of metals for resistance to sulfide stress
typical high-strength alloy steel, Materials Science and Engineering: A 527 (26)
cracking and stress corrosion cracking in H2S environments, Houston, TX. 2005.
(2010) 6980–6986.

14
K. Li and Y. Zeng Construction and Building Materials 362 (2023) 129746

[63] E.M. Gutman, Mechanochemistry of materials, Cambridge Int Science Publishing, [69] X. Zhang, Z. Sun, Z. Tang, M. Liu, B. Li, Study on stress corrosion cracking of
1998. aluminum alloys in marine atmosphere, Environment-Induced Cracking of
[64] R. Ren, S. Zhang, X. Pang, K. Gao, A novel observation of the interaction between Materials, Elsevier, 2008, pp. 351–357.
the macroelastic stress and electrochemical corrosion of low carbon steel in 3.5 wt [70] D.R. Lide, CRC handbook of chemistry and physics, CRC Press, 2004.
% NaCl solution, Electrochimica Acta 85 (2012) 283–294. [71] W. Zheng, Stress corrosion cracking of oil and gas pipelines in near neutral pH
[65] S. Ramamurthy, A. Atrens, Stress corrosion cracking of high-strength steels, environment: review of recent research, Energy Materials 3 (4) (2008) 220–226.
Corrosion Reviews 31 (1) (2013) 1–31. [72] Z.Y. Liu, X.G. Li, C.W. Du, L.X. Wang, Y.Z. Huang, Effect of Dissolved Oxygen on
[66] R.G. Song, W. Dietzel, B.J. Zhang, W.J. Liu, M.K. Tseng, A. Atrens, Stress corrosion Stress Corrosion Cracking of X70 Pipeline Steel in Near-Neutral pH Solution,
cracking and hydrogen embrittlement of an Al–Zn–Mg–Cu alloy, Acta Materialia Corrosion 66 (1) (2010) 015006.
52 (16) (2004) 4727–4743. [73] L. Lunde, R. Nyborg, The effect of oxygen and water on stress corrosion cracking of
[67] J. Woodtli, R. Kieselbach, Damage due to hydrogen embrittlement and stress mild steel in liquid and vaporous ammonia, Plant/Operations Progress 6 (1) (1987)
corrosion cracking, Engineering failure analysis 7 (6) (2000) 427–450. 11–16.
[68] G. Ghosh, P. Rostron, R. Garg, A. Panday, Hydrogen induced cracking of pipeline [74] Y. Huang, Y. Zhu, Hydrogen ion reduction in the process of iron rusting, Corrosion
and pressure vessel steels: A review, Engineering Fracture Mechanics (2018). science 47 (6) (2005) 1545–1554.

15