Hydrogen Embrittlement of Steel Armor Wires Containing Globular and Lamellar Cementite

Abstract

This work addresses the hydrogen-microstructure interaction and hydrogen embrittlement (HE) susceptibility in commercial hypoeutectoid ferrite/cementite tensile armor wires used in flexible risers. Electrochemical hydrogen permeation was performed using H₂SO₄-based electrolyte under cathodic potential of -1.5V/SCE. Microstructures containing globular cementite exhibited behavior closely aligned with the theoretical Fickian diffusion. The pearlitic wires required additional time shifts for curve fitting, corresponding to trapped hydrogen concentrations of 1.92 and 0.65 mol m⁻³. In alloys containing globular cementite, the microstructure with higher mechanical strength, in tensile tests, and 89% greater amount of desorbed hydrogen, as measured by thermal desorption spectroscopy, exhibited significantly lower HE susceptibility. An opposing tendency was observed in pearlitic microstructures. The high-strength fine pearlite has shown critical losses of ductility and brittle behavior in hydrogen-charged samples submitted to tensile tests. When heat-treating and deforming a pearlitic microstructure, it was seen that deformation played a more significant role in hydrogen diffusion and embrittlement, than interlamellar spacing. The effect of prior corrosion in NaCl 3.5% saturated with CO₂ was also assessed, showing that Fe₃C-rich remaining structures increased hydrogen generation in cathodic polarization tests and permeated hydrogen flux, enhancing the observed loss of ductility in tensile tests due to hydrogen charging.

Keywords:
Hydrogen embrittlement; flexible risers; CO₂ corrosion; hydrogen trapping

1. Introduction

Flexible risers are conductor tubes used for fluid transportation in the oil and gas industry, connecting flowlines at wellheads to floaters on the surface. They are made of consecutive steel armor wires and polymeric layers, each of them responsible for performing a specific role. Polymeric layers provide sealing, protecting the metallic layers from direct contact with the environment, and reducing frictional forces between adjacent layers. Metallic reinforcements uphold different loads and pressures¹-³.If damage occurs in the outer shield sheath, seawater may enter the annulus space, while permeation of molecules originating from the well and sea also tends to occur. The result is a wet confined space between consecutive layers, which is filled with molecules such as H₂S, CO, CO₂, CH₄, and H₂O²-⁶.

Cathodic protection, usually applied in flexible risers to prevent corrosion, and corrosion reactions can lead to hydrogen evolution on the armor wire’s surface. Hence, the steel reinforcements are prone to hydrogen embrittlement (HE), especially in the extreme conditions created in wet confined spaces. The steel structures can, as a result, collapse prematurely due to hydrogen-induced cracking (HIC), sulfide stress corrosion cracking (SSCC), and hydrogen-assisted stress corrosion cracking (HASCC)⁵-⁹. Leakage and rupture of these structures, in turn, may lead to severe environmental damage, safety risks, and extensive financial losses¹⁰-¹².

Tensile armors are the metallic layers responsible for resisting the axial loading generated by various external forces. They are made of flat wires of high-strength carbon steels, generally with a proeutectoid composition and microstructures consisting of ferrite and cementite. The morphology and distribution of cementite, as well as the straining condition and the associated deformation mode, strongly influence a steel’s HE susceptibility. Fine and spheroidized cementite has been reported to reduce hydrogen-related degradation during fatigue cycling in a low-carbon steel¹³ and to be related to higher resistance to HE in tempered steels¹⁴-¹⁶. On the other hand, rod-like cementite has been observed to be a more effective hydrogen trap than globular cementite¹⁷.

In contrast, lamellar cementite has exhibited higher resistance to delamination under low hydrogen fugacity and delayed fracture under high hydrogen fugacity when compared to the spheroidite microstructure¹⁸. Previous studies also show that hydrogen tends to accumulate mainly in regions of high microhardness¹⁹ and to soften ferrite and cementite due to a reduction in the energy for dislocation motion in pearlite²⁰, whereas pre-straining ferrite/cementite interfaces can result in improved or undermined HE susceptibility, depending on the hydrogen’s level²¹. In a martensitic, tempered steel, the presence of plate-shaped cementite increased toe resistance to HE by hindering hydrogen diffusion towards critical sites. However, this was lost when samples were fully saturated with H before testing²². In this manner, different microstructural aspects are interconnected and do not behave monotonically. Variations in the chemical composition and manufacturing route of the steel wires lead to simultaneous alterations in these features. This brings complexity to studies on hydrogen-related fracture mechanisms in steel components¹⁸,²³-²⁵.

On the other hand, approaches to emulate different conditions that can evolve within the annulus must be considered. Slow strain rate tests (SSRTs) in NaCl-based systems to emulate seawater are usually employed to explore HE in these components²³,²⁶,²⁷. Reductions of areas ranging from near 0.2 to near 0.6 have been observed in armor wires of different microstructures at cathodic potentials of -1.1 and -1.4 V vs Ag/AgCl²³ in comparison to SSRTs performed in air. H₂S-saturated solution or H₂S bubbling is applied, in turn, in SSRT, HIC, SSCC and permeation tests to characterize the alloy’s behavior in sour media⁹,¹⁹,²⁸-³¹. H₂S leads to increased hydrogen uptake due to the prevention of hydrogen recombination on the metallic surface, resulting in exacerbated hydrogen embrittlement measured by increased losses of ductility and intense SSC and HIC²⁸,²⁹,³²-³⁴.

The presence of CO₂ within the annulus space can also be relevant⁹,²⁸,²⁹,³⁴-⁴⁰. Even though gaseous CO₂ does not traditionally cause a catastrophic failure mode as that associated with H₂S_(g), CO_2(g) can provoke the acidification of the aqueous medium, potentially increasing hydrogen fugacity and HE, leading to increases in the generated and permeated hydrogen²⁹,³⁵,³⁶,³⁹, and to high corrosion rates⁴,⁴¹-⁴³. CO₂ corrosion products at the alloy’s surface, in turn, can generate distinct impacts on hydrogen uptake and embrittlement, depending on their nature, and the literature still shows divergent and unresolved trends³³,³⁵,⁴⁴-⁴⁶. For instance, the formation of a Fe₃C-rich layer due to the preferential dissolution of ferrite during the corrosive process has been reported to affect the absorption and desorption of hydrogen³³. Opposite tendencies of increasing the permeated current³⁵,³⁹ and decreasing hydrogen uptake⁴⁶ were previously reported to be related to this Fe₃C-rich layer.

Although cementite morphology has been associated with hydrogen transport and fracture in carbon steels, its contribution is commonly coupled with chemical composition, mechanical strength, interfacial density, and deformation history. Consequently, the variables controlling the response of commercial flexible-riser armor wires remain difficult to distinguish. Therefore, this work combines experimental approaches to investigate the hydrogen–microstructure interaction in commercial hypoeutectoid ferrite/cementite armor wires while preserving their original dimensions and manufacturing-induced characteristics.

First, microstructures containing globular and lamellar cementite were compared under a high-hydrogen-fugacity electrolyte, which was employed as an experimental tool to maximize hydrogen charging and enable a direct comparison of the intrinsic response of the different commercial microstructures. Subsequently, one of the pearlitic wires was heat-treated and plastically deformed to investigate the individual influence of lamellar spacing and deformation while maintaining a constant chemical composition. Finally, the effect of prior CO₂ corrosion was evaluated in 3.5 wt.% NaCl solution saturated with CO₂, providing conditions more representative of the wet annulus environment in flexible risers. By combining full-scale analyses, controlled microstructural modifications, and service-representative environments, we aim to provide a comprehensive framework to investigate the interplay between cementite morphology, hydrogen trapping, deformation, and corrosion history in the HE in these commercial components.

2. Materials and Experimental Procedure

2.1. Materials

Four flexible riser tensile armor wires with flat cross-sections of varying dimensions from two different manufacturers were studied, as listed in Table 1. From manufacturer A, a 6-mm-thick wire was investigated. From manufacturer B, wires with thicknesses of 4, 5, and 7 mm were studied. Samples were designated after the main cementite morphology as CG1, CG2, P1, and P2, as depicted in Table 1. The wires were received in coiled condition and straightened for sample selection. Their chemical compositions were determined by optical emission spectroscopy in a SPECTRO MAX LMF04 spectrometer, as presented in Table 2. Volume fractions of cementite, f_Fe3C, were obtained from the measured chemical composition by thermodynamic calculations using Thermo-Calc®.

Table 1
Studied armor wires’ identification.

ID	Manufacturer	Width (mm)	Thickness (mm)	Microstructure
GC1	A	14	6	Ferrite + globular and fragmented cementite
P1	B	8	4	Fine pearlite
GC2	B	12	5	Ferrite + globular and fragmented cementite
P2	B	12	7	Coarse pearlite + fragmented cementite

Table 2
Chemical composition of the studied armor wires obtained by optical emission spectroscopy.

Chemical Composition (wt%)
	GC1	GC2	P1	P2
C	0.70	0.67	0.63	0.59
Si	0.19	0.17	0.29	0.29
Mn	0.83	0.49	0.66	0.69
P	0.01	0.01	0.01	0.01
Cr	0.05	0.03	0.04	0.03
Ni	0.05	0.03	0.04	0.04
Mo	0.01	0	0	0
Al	0.02	0	0.03	0.05
Cu	0.01	0.01	0.02	0
Co	0	0.01	0	0
Nb	0.01	0	0	0
W	0	0.04	0.04	0.05
Sn	0	0.01	0.01	0.01
As	0	0	0	0.01
Bi	0	0	0	0.01
Fe	98.1	98.5	98.2	98.2
fFe3C	10.6	10.1	9.5	8.9
(vol %)

2.2. Armor wires’ characterization

After mounting, grinding, polishing, and etching with Nital 2%, the microstructure in the transverse area of each wire was evaluated using scanning electron microscopy (SEM) at a Tescan Vega 3 microscope in the secondary electron mode. For mechanical characterization, tensile tests were performed in a mechanical MTS EXCEED E45 testing machine. Samples of 300 mm length were selected, maintaining the wire’s original cross-section. A useful length of 180 mm was used with a strain rate of 5.5 × 10^-5 s^-1. Elongations were measured by the crosshead displacement, and engineering stress vs strain curves were obtained from the sample’s initial dimensions. The elastic region was replaced during data treatment, considering an elastic modulus of 210 GPa.

2.3. Hydrogen/microstructure interaction

Each microstructure was evaluated in respect to its interaction with hydrogen using electrochemical tests. A 0.1 M H₂SO₄ + 2 mg/L As₂O₃ solution was used as an electrolyte. Besides being a strong acid with a pH of nearly 1, As₂O₃ acts as a poisoner. It diminishes hydrogen recombination, forming gaseous H₂, thereby increasing electrolysis efficiency and hydrogen fugacity. Thus, it plays an important role in increasing the amount of available atomic hydrogen and hydrogen fugacity.

Electrochemical permeation tests were performed through each wire’s original thickness. Samples were ground and mechanically polished before the tests. Tests were conducted in two replicates for validation. The test setup is based on the Devanathan-Stachurski method⁴⁷. It consists of two compartments connected by the permeated sample, as illustrated in Figure S1(a). Each compartment is equipped with a reference electrode (saturated calomel electrode - SCE) and a counter electrode (Pt wire). The sample acts as the common working electrode for both compartments and has an exposed elliptical area of 0.457 cm².

The first compartment is filled with 0.1M H₂SO₄ + 2 mg/L As₂O₃. A constant cathodic polarization of -1.5 V/SCE is established, promoting hydrogen generation. This side, referred to as the hydrogen generation side, is where atomic hydrogen is adsorbed and absorbed into the sample. The second compartment is filled with 0.1M NaOH. An anodic potential equal to the open circuit potential (OCP) of the system, measured in a pre-step of the test, is applied. This condition promotes the oxidation of all hydrogen atoms diffusing from the entry surface through the sample, maintaining a concentration equal to zero at the exit surface. This is referred to as the detection side.

The test records the anodic current, i, required to promote the oxidation of all hydrogen atoms that diffuse through the sample thickness. Measured values are converted into a hydrogen flux, J_H, by Equation 1⁴⁸, where F is the Faraday constant, and A is the exposed area. The apparent hydrogen diffusivity, D_app,H, was calculated from the breakthrough time, t_b. This parameter corresponds to the time required for hydrogen to reach the detection side. It is extracted from the permeation curves as the intersection of the tangent at the inflection point of the curve with the x-axis. For that, Equation 2 was used, where L is the sample’s thickness. Hydrogen apparent solubility, S_app,H, was calculated using Equation 3, where J_∞ is the steady-state hydrogen flux obtained in the permeation curves.

$J_H=\frac{i}{FA}$ (1)

$t_b=0.5\frac{L²}{\pi ²D_{app,H}}$ (2)

$J_{\infty }=\frac{D{S}_{app,H}}{L}$ (3)

The evolution of hydrogen permeation flux with time through the crystal structure of metallic alloys, denoted as J(t), can be described by Equation 4. This equation is derived by applying the boundary conditions for the non-steady-state potentiostatic time-lag method to Fick’s Second Law⁴⁸. It represents an ideal behavior and does not account for the contribution of trapping sites along the hydrogen diffusional path. Therefore, deviations between experimental curves and those predicted by Equation 4 were used to evaluate the influence of trapping sites on hydrogen diffusion in the investigated armors.

$J\left(t\right)=J_{\infty }\left[1+2{\displaystyle \sum }_1^{\infty }{\left(-1\right)}^n{e}^{\left(\frac{-n²\pi ²Dt}{L²}\right)}\right]$ (4)

Thermal desorption spectroscopy (TDS) tests were performed, in replicates, to characterize hydrogen trapping at the different microstructures. Samples with a cylinder shape were extracted from the central region of the armor wires by electron discharge machining (EDM), with diameters of 5.3 mm, 3.9 mm, 5.2 mm, and 4.77 mm, respectively for CG1, P1, CG2, and P2. The diameter of the cylindrical samples was limited by the diameter of the reactor used in the TDS apparatus, of 6 mm.

The specimens were hydrogen charged using 0.1M H₂SO₄ + 2 mg/L As₂O₃ at the cathodic potential of -1.5 V/SCE for 24 h. Immediately after charging, they were cleaned and placed in the reactor which was then inserted inside of a furnace. The temperature was increased from room temperature to 800 °C, at a heating rate of 10 °C/min. Gaseous helium was used to drag the desorbed hydrogen, which is directed to a mass spectrometer, where gases are detected according to their molar masses. The hydrogen concentration in each microstructure was obtained in relation to a reference sample, PdH₂.

Hydrogen embrittlement susceptibility of each armor wire was evaluated based on the loss of ductility of hydrogen-charged samples subjected to mechanical tensile tests in comparison to its respective as-received (AR) condition. For that, hydrogen charging was carried out by immersing ground and polished tensile test samples in 0.1M H₂SO₄ + 2 mg/L As₂O₃ under the cathodic potential of -1,5 V/SCE for 18 h. In this step, only the useful length is in contact with the electrolyte, as the regions that would be in the grip are protected by seal tape. The hydrogen embrittlement susceptibility was calculated by the hydrogen embrittlement index (HEI), described in Equation 5. $e_{f,AR}$ is the engineering deformation at fracture in the AR condition and $e_{f,H}$ is the engineering deformation at fracture in the hydrogen-charged condition. The fracture surfaces of these wires were analyzed using SEM.

$HEI=\frac{e_{f,AR}-e_{f,H}}{ef,AR}\times 100$ (5)

2.4. Effect of heat treatments and deformation on pearlitic microstructure

Heat-treatments were performed in P1, aiming to obtain fine and coarse pearlitic microstructure at a fixed chemical composition. The heat treatments were conducted in two steps using two furnaces. Austenitization was performed in a Carbolite muffle furnace at 900 °C for 30 min. Immediately after, samples were transferred to a second muffle furnace (Quimis Q-318M21) for austenite decomposition into ferrite and cementite at either 450 °C or 700 °C. Final cooling was carried out in oil. Samples were named according to the second heat-treatment temperature as HT450 and HT700, respectively. The effect of deformation was then evaluated in HT450 by submitting it to a 10% thickness reduction by cold-rolling. This strained condition will be referred to as HT450D.

HT700, HT450, and HT450D samples were evaluated by electrochemical hydrogen permeation and tensile tests in non-charged and hydrogen-charged states, which were performed as described in section 2.3.

2.5. Impact of CO₂ corrosion on the hydrogen embrittlement susceptibility in NaCl 3.5%

CG1 and P1 were used for evaluating the effect of CO₂ corrosion prior to the cathodic charging in a microstructure containing globular cementite and another containing lamellar cementite. Besides the high hydrogen fugacity electrolyte based on H₂SO₄, a simplified seawater, made of a NaCl 3.5% solution, containing CO_2(g) is used in this section. Saturation of CO_2(g) was promoted before any performed test and maintained throughout the experimental procedure by continuous gas bubbling. 0.5 ml/L of HCl was added to guarantee a pH of 4.

The as-received (AR) wires were ground and mechanically polished before every electrochemical test that will be further described. Samples at this stage of preparation characterize the as-polished condition. To obtain the corroded condition, polished samples were submitted to corrosion in NaCl 3.5% electrolyte saturated with gaseous CO₂. This step was conducted in a capped beaker with a setup consisting of the sample as the working electrode, a counter electrode (coiled platinum wire), and a reference electrode (saturated calomel electrode – SCE), as schematized in Figure S1(b). Aiming to obtain a Fe3C-enriched surface, an anodic potential of -0.3 V/SCE was applied for 3 hours at room temperature³⁵. CO_2(g) was continuously bubbled to keep the electrolyte’s saturation at atmospheric pressure. The resultant samples’ surface morphology was assessed using SEM. Atomic force microscopy (AFM) was used to analyze the surface rugosity. X-ray diffraction (XRD) measurements were performed on both corroded and as-polished samples to identify changes in the detected phases. These tests were performed using Cu – Kα (λ= 0.15406 nm ) with 2θ ranging from 20 to 90°, with 5s/step.

Anodic and cathodic potentiodynamic polarization tests were performed from open circuit potential (OCP) to 2 and -2 V/SCE, respectively, at a scanning speed of 600 mV/h. As-polished samples were tested in NaCl 3.5% saturated with CO₂ and in H₂SO₄ 0.1M + As₂O₃ 2mg/L. To assess the effects of prior corrosion on the measured currents, corroded samples were tested in NaCl 3.5% saturated with CO₂.

Hydrogen permeation tests were promoted by electrochemical means. In this case, permeation was promoted through half the wire’s thickness. These samples were obtained by a longitudinal separation through electro-discharging machining (EDM), dividing them through the half-thickness of their original cross-section. They were ground and polished on both sides, and the external surface was faced to the hydrogen generation side

To evaluate hydrogen permeation in the corroded surface condition, permeation samples were ground, polished, and cold-mounted with resin in order to protect the internal, as-cut, side of the sample, Figure S1(c). The external side, the surface of the wire, was submitted to the corrosion step in NaCl 3.5% saturated with CO_2(g). The resin was then mechanically removed so the sample could be placed in the permeation setup, where the corroded surface was in contact with the hydrogen entrance side, and the detection surface was in the as-polished condition.

Tensile tests were conducted on as-polished and corroded wires, as described in section 2.3. In this case, hydrogen charging prior to mechanical testing was conducted in NaCl 3.5% + CO₂ medium for the same time, 18 h. To ensure the maintenance of the solution’s pH throughout this step, a 30 L reservoir was used to provide fresh solution to the charging cell using a peristaltic pump, as demonstrated in Figure S1(d).

3. Results and Discussion

3.1. Characterization of globular and fragmented cementite-containing microstructures

GC1 contains 0.7 wt% and GC2, 0.67 wt% of carbon, as obtained by optical emission spectroscopy. Cementite fractions obtained by CALPHAD calculations were, respectively, 10.6 vol% and 10.1 vol%. Secondary electron images (SEI) obtained by SEM are shown in Figures 1a and 1b. Ferrite can be seen as the gray matrix phase, which is consumed during etching, in contrast to the non-consumed white cementite. The carbide phase exhibits a globular and fragmented nature and is embedded in a ferritic matrix. The average particle sizes of 0.14 μm and 0.11 μm were obtained, respectively, for CG1 and CG2 using ImageJ. Previous austenitic grain boundaries are marked by cementite precipitation, which indicates finer grain sizes in GC1.

Figure 1
SEI of (a) CG1 and (b) CG2 armor wires containing globular and fragmented cementite; and (c) P1 and (d) P2 armor wires containing lamellar cementite.

Hydrogen permeation curves obtained for GC1 and GC2 are presented in Figures 2a and 2b. The apparent hydrogen diffusivity, D_app,H, and solubility, S_app,H, calculated from the curves of each microstructure can be seen, respectively, in Figures S2-a and 2b. Experimental data exhibited behaviors closely aligned with those predicted by theoretical curves, presented as dashed lines, indicating little or no effect of trapping sites on the permeation curves. In CG2, the hydrogen diffusivity calculated from experimental curves, D_app,H, matched the value used to fit the theoretical curve, D_fit,H. In contrast, for CG1, the hydrogen diffusivity calculated from the experimental curve was higher than that used for fitting the theoretical curve, implying some influence of microstructural defects.

Figure 2
(a-d) Electrochemical permeation curves, (e-h) TDS spectra, and (i-l) engineering tensile stress vs strain curves in as-received (AR) and hydrogen charged (H) (a,e,i) CG1, (b,f,j) CG2 armor wires, (c,g,k) P1 and (d,h,l) P2 armor wires.

Measurements of desorbed hydrogen flux normalized by the mass of the sample, obtained from TDS analyses, are plotted against temperature in Figures 2e and 2f. Experimental data are presented as points shading curves fitted through mathematical treatment, also highlighting the deconvolution of peaks. The desorption of hydrogen on the tensile armor wires is characterized by the presence of two peaks. Desorption peak temperatures and the amount of desorbed hydrogen obtained from TDS curves are depicted in Figure S2(c, d). Total desorbed hydrogen contents of 9.0 mol m^-3 and 4.9 mol m^-3 were obtained for CG1 and CG2, respectively.

The presence of multiple hydrogen desorption peaks is attributed to the wide distribution of hydrogen binding energies associated with different microstructural trapping sites, including grain boundaries, dislocations, phase interfaces, and vacancies⁴⁹. For both samples, the first desorption peak, observed at 134 °C for CG1 and 127 °C for CG2, is likely associated with hydrogen trapped at ferrite/cementite interfaces, as previously reported⁵⁰-⁵². The second peak, observed at 258 °C for CG1 and 234 °C for CG2, has been attributed to hydrogen trapped at defects introduced by plastic deformation⁵⁰ as well as deformation-induced defects in cementite⁵¹. The higher amount of desorbed hydrogen measured for CG1 compared with CG2 also indicates a higher density of hydrogen trapping sites, particularly ferrite/cementite interfaces and defects, introduced by plastic deformation.

Tensile stress-strain curves of hydrogen-charged samples (H) are presented in Figures 2i and 2j, in comparison with the respective as-received (AR) condition. The mechanical response of AR wires shows higher mechanical strength in CG1, marked by higher yield strength (CG1 - 1020 MPa, CG2 - 865 MPa) and ultimate tensile stress (CG1 - 1160 MPa, CG2 - 1044 MPa), with no harm in ductility in comparison to CG2. The higher strength in CG1 implies that this microstructure contains a higher number of crystallographic defects, such as a finer grain size or higher dislocation density. Its higher carbon content, in turn, also leads to a higher fraction of cementite and ferrite/cementite interfaces. All these features could be explanations of the observed stronger effect of trapping sites in CG1, affecting hydrogen diffusion and leading to larger amounts of retained hydrogen during hydrogen charging.

In the hydrogen-charged state, CG1-H still presented relevant ductility and an HEI of 40%, while CG2-H presented virtually no plastic deformation, resulting in a superior response regarding hydrogen embrittlement in the first microstructure. The fracture morphologies of CG1-H and CG2-H, presented in Figure 3, differ significantly from each other. When compared to their respective AR states (Figures 3a and 3c), CG1 is marked by increased microdelaminations (Figure 3b) and CG2, by the presence of quasi-cleavage features (Figures 3e). In different regions of the tested sample, both CG1-H and CG2-H presented shallow and fine dimples (Figures 3c and 3f), with CG1 also presenting a large fraction of delamination features.

Figure 3
Fractographies of (a,b,c) CG1 and (d, e, f) CG2 samples submitted to tensile tests in the (a,d) as-received and (b, c, e, f) hydrogen-charged (H) states.

The morphology of the fracture surface of CG1 – H points out to hydrogen acting locally, leading to particle decohesion, pointed out by a white arrow, and multiple hydrogen-enhanced delamination sites, exemplified by yellow arrows in Figure 3b. The mechanism behind this effect can be hydrogen atoms locally enhancing the mobility of dislocations and promoting void nucleation and growth by a largely reported mechanism named hydrogen-enhanced localized plasticity (HELP)⁵³-⁵⁵. This phenomenon arises from the interaction of hydrogen atoms with stress fields surrounding dislocations, which increases their mobility and reduces mutual interactions. As a result, dislocation slip along specific crystallographic planes is facilitated, enhancing local ductility in regions with higher hydrogen concentrations⁵⁶.

In turn, CG2 is characterized by transgranular quasi-cleavage facets, highlighted in Figure 3e. This type of fracture has been previously related to hydrogen embrittlement in ferritic microstructures. Its occurrence has been linked to HELP, wherein hydrogen accumulation along slip planes enhances dislocation motion, resulting in severe strain localization. This localized deformation promotes the nucleation, growth, and coalescence of microvoids, culminating in fracture⁵⁵,⁵⁷.

One effective way to minimize the deleterious effects of hydrogen on steels is through the introduction of trapping sites, which help limit the diffusible hydrogen content⁵⁸. In this context, the distribution, quantity, and binding energy of these traps are crucial. Homogenously distributed defects with intermediate binding energies can effectively delay migration and critical accumulation of hydrogen at stress concentrators, thereby minimizing embrittlement⁵⁹. The fracture aspects and high density of trapped hydrogen in CG1-H indicate that its fracture mechanism is influenced by trapping sites controlling diffusible hydrogen content, resulting in the lower susceptibility to hydrogen embrittlement. In CG2, it appears that trapping sites are less effective in concentrating hydrogen locally, leading to cleavage and significant hydrogen embrittlement.

Thus, in the microstructures containing globular and fragmented cementite, the more significant role of hydrogen trapping sites in CG1 led to significantly lower hydrogen embrittlement susceptibility. In this case, the mechanical strength and carbon content were inversely related to the alloy’s hydrogen embrittlement. In previous studies, fine and spheroidized cementite has been related to a homogeneous distribution of the hydrogen clouds around cementite/ferrite interfaces and to reduced hydrogen-related degradation¹³-¹⁵,⁶⁰. In this work, it has been observed that this cementite morphology does not guarantee that the alloy will present low HEI, as seen in CG2. For this microstructure to be less susceptible to hydrogen embrittlement, the effect of trapping sites must be effective, as seen in CG1.

3.2. Characterization of pearlitic wires

Carbon contents of 0.63 wt% and 0.59 wt% were obtained for P1 and P2, respectively, resulting in calculated cementite fractions of 9.5 and 8.9 vol%, both values lower than those previously obtained for CG1 and CG2. P1 is mainly constituted by fine and deformed pearlite, also containing proeutectoid ferrite, as can be seen in Figure 1c. An average interlamellar spacing of 0.15 ± 0.07 μm was estimated by image analysis. The pearlitic and proeutectoid ferrite microconstituents are present with a coarser aspect in P2, Figure 1d. The average estimated interlamellar spacing was 0.22 ± 0.05. In this armor, regions containing degenerated cementite embedded in a ferritic matrix can also be observed, as circled in Figure 1 (d’).

Hydrogen permeation curves for P1 and P2 are shown in Figures 2c and 2d. P2 presented apparent hydrogen diffusivities similar to those previously observed in CG1 and CG2, in the order of magnitude of 10^-10 m² s^-1, while P1 presented a significantly lower diffusivity, in the order of 10^-12 m² s^-1. In the theoretical curve fitting, in contrast to the microstructures containing globular cementite, a time interval Δt had to be added to the theoretical curves, shifting them to the right, to better align with the experimental data. This delay at the onset of the transient regime in experimental hydrogen permeation curves can be attributed to the filling of trapping sites, postponing the arrival of the first hydrogen atoms at the detection side. The delay Δt was used to calculate the amount of hydrogen trapped in the microstructure, Q_H, before its detection according to Equation 6⁶¹.

$Q_H=\frac{J_{\infty }\Delta t}{L}$ (6)

The horizontal shift between the experimental and theoretical curves measured by Δt was much more pronounced in P1, corresponding to a concentration of 1.92 mol m^-3 of trapped hydrogen. In this armor, the hydrogen diffusivity value used in Equation 5 to fit the theoretical curve, D_fit,H, is still significantly lower than in the other armor wires, in the order of 10^-11 m² s^-1_. The deviation of this calculated value from what would be expected for diffusion in ferrite is an indication that trapping sites are still affecting the material’s diffusivity even in the transient regime. For P2, the Δt derived 0.65 mol m^-3 of trapped hydrogen. In this case, D_fit,H, was 2.6 times higher than the value obtained from the experimental permeation curve, D_app,H. This suggests that once the trapping sites are filled, they no longer influence hydrogen permeation through the microstructure, leading to accelerated diffusion.

Hydrogen desorption of previously hydrogen-charged samples of the pearlitic wires also exhibits two peaks, Figures 2g and 2h. The peak temperatures and desorbed hydrogen are depicted in Figure S2(c,d). P1 presented 7.5 mol m^-3 of desorbed hydrogen, while P2 showed 5.0 mol m^-3. Similar to CG1 and CG2, the first desorption peak observed for P1 and P2 falls within the temperature range associated with hydrogen desorption from ferrite/cementite interfaces. The second peak appears over a broad range, varying from 192°C for P1 to 295°C for P2. This peak has been attributed to hydrogen trapping in defects produced by plastic deformation⁵¹ and deformation in cementite⁵⁰. The difference in its temperature implies stronger trapping in P2 when compared to P1, despite the lower Q_H.

The first armor, P1, presented significantly higher mechanical strength and lower ductility, as observed from its tensile stress vs strain curves in Figures 2k and 2l. These mechanical properties indicate a strain-hardened state. Associated with the high fraction of cementite/ferrite interfaces in its refined microstructure, this could explain the more prominent action of trapping sites in hydrogen permeation curves and the higher desorbed hydrogen content.

In the tensile tests of hydrogen-charged samples, P1 fractures in the merge of the plastic regime, whereas P2 exhibited an HEI of 69%. The fracture surface of P1 changes from mainly dimples (Figures 4a and 4d) in P1 to presenting microdelamination (pointed out in Figure 4b), and quasi-cleavage (as limited by a rectangle in Figure 4d and magnified in 4(f)) in P1-H. Microdelaminations may be a result of HELP mechanisms occurring on the strain-hardened pro-eutectoid ferrite or hydrogen-enhanced decohesion (HEDE) between adjacent pearlite colonies⁶². HEDE may also result in debonding between cementite and ferrite lamellae. Enos and Scully⁶³ showed that shear cracking due to the reduction of cohesive strength at the cementite/ferrite interface results in regions containing many small areas with voids, with the width corresponding to the interlamellar spacing. These features have also been seen in P1-H, as exemplified by the circled area in Figure 4b, magnified in Figure 4e.

Figure 4
Fractographies of (a-f) P1 and (g-i) P2 samples submitted to tensile tests in the (a,d,g) as-received and (b, c, e, f, h, i) hydrogen-charged (H) states.

P2-H exhibited the dominant presence of quasi-cleavage (Figures 4h and 4i) in contrast to shallow dimples in the non-charged condition (Figure 4g). Similar to CG2, it also implies an easy hydrogen diffusion through the ferritic phase, which could be attributed to discontinuities caused by the coarse nature of pearlite and proeutectoid ferrite, as well as the presence of fragmented cementite. The presence of intergranular cracks, pointed out in Figure 4i, can be due to microvoid formation and coalescence in the proeutectoid ferrite due to HELP. The migration of hydrogen toward grain boundaries may lead to significant dislocation pile-up at these interfaces, promoting high local stress concentrations that can trigger failure⁶⁴.

In this manner, comparing the two pearlite-containing microstructures, P1 and P2, the refined and strain-hardened P1 was strongly prone to hydrogen’s detrimental effects. The reduced density of lamellar ferrite/cementite interfaces in the coarse microstructure and the lower mechanical strength of P2 were associated with improved resistance to hydrogen embrittlement. Different from the microstructures containing globular cementite (CG1 and CG2), the results for P1 and P2 imply that in a microstructure containing lamellar ferrite/cementite interfaces, the high density of interfaces and microstructural defects created by plastic deformation are related to the development of a critical concentration for the hydrogen embrittlement phenomena.

3.2.1. Effect of lamellar spacing and deformation on pearlitic P1 wire

SEIs of P1 heat-treated under different conditions are presented in Figures 5a and 5b. Comparatively, P1-HT700 is constituted by coarser and P1-HT450, by finer pearlite. Mean spacings of 0.322 ± 0.02 and 0.195 ± 0.02 were obtained, respectively, by image analysis. These estimated values imply a 39.3% refinement in the pearlite obtained at 450 °C. 10% deformation of P1-HT450D’s thickness did not lead to relevant visible microstructural changes, Figure 5c.

Figure 5
(a-c) SEI; (d-f) hydrogen electrochemical permeation curves, and (g, h, i) engineering stress x strain curves of P1 (a,d,g) HT700, (b, e, h) HT450 and (c, f, i) HT450D.

Experimental and fitted hydrogen permeation curves are presented in Figures 5d-f. Lamellae refinement led to delays in the permeation curve, which translated into lower D_app,H. Moreover, while no Δt was necessary for fitting HT700, one of 1500 s was necessary in the refined HT450. This experimental curve, HT450, also presented greater deviation in the transient regime from the Fickian curve. Both of these effects on D_app,H and the Δt were much more pronounced in the deformed HT450D, as follows. In the deformed condition, the D_app,H was of 3.8 x 10^-11 m²s^-1, one magnitude lower than its undeformed counterpart, 2.5 x 10^-11 m²s^-1. In turn, Δt in HT450D was 105000 s, much higher than that in HT450C.

Tensile stress-strain curves of previously hydrogen-charged samples are presented in comparison to their respective non-charged conditions in Figures 5g-i. Little plastic deformation was measured in HT700-H and HT450-H, which were characterized by similar ductilities. HT450D-H, which exhibited much greater mechanical strength and smaller ductility than the others, fractured within the elastic regime. These results also point out to a more important effect of deformation in comparison with lamellar spacing.

3.3. Effect of cementite morphology on the studied armor wires

The relationship between microstructure and hydrogen embrittlement susceptibility is not straightforward when analyzing different armor wires. This is because several microstructural features must be considered simultaneously, including carbon content, mechanical strength, and cementite morphology. In the present work, neither mechanical strength nor carbon content alone determined the hydrogen–microstructure interaction or the resulting hydrogen embrittlement susceptibility.

Instead, cementite morphology was intimately related to hydrogen permeation behavior and appeared to dictate distinct relationships between hydrogen trapping and hydrogen embrittlement. It is worth mentioning that other parameters not assessed in this study, such as the strengthening mechanism, dislocation density, and the nature of ferrite/cementite interfaces, may also contribute to the response of each microstructure to hydrogen charging.

The microstructures containing globular cementite, CG1 and CG2, presented the lowest apparent hydrogen solubility, S_app,H, and permeation curves that closely aligned with theoretical predictions. Comparing CG1 and CG2, CG1 exhibited a greater difference between D_fit,H and D_app,H, indicating a stronger influence of hydrogen trapping sites on hydrogen diffusion. This interpretation is supported by the significantly higher amount of trapped hydrogen measured by TDS.

CG1 is the alloy with the highest carbon content, implying a high fraction of cementite. Hence, a high density of cementite/ferrite interfaces is expected, which could explain the high Q_H obtained in peak 1, as highlighted in Figure S2(d). The Q_H measured from the second peak is also higher, indicating a higher fraction of trapping sites induced by plastic deformation than in CG2. The higher strength of CG1 may result from differences in grain size, dislocation density, cementite fraction, or other features inherited from thermomechanical processing. Because these variables were not quantified, their individual contributions cannot be distinguished.

This armor wire, CG1, exhibited low susceptibility to HE. In this manner, the more significant role of hydrogen trapping sites in the microstructure containing globular cementite led to lower hydrogen embrittlement susceptibility. In this case, the mechanical strength and carbon content were inversely related to the HEI.

On the other hand, the microstructures where pearlite was present, P1 and P2, showed significant deviations from theoretical curves during hydrogen permeation. While it implies that lamellar cementite alters the permeation of hydrogen through ferrite more significantly, this effect loses strength once hydrogen trapping sites are filled in the coarse pearlite of P2. The result is the highest calculated D_fit,H.

A significant amount of hydrogen trapping could be measured in P1 by the delay in hydrogen permeation and the intense desorption peaks in TDS tests. In the second desorption peak, an important difference between the two pearlitic microstructures, P1 and P2, has been observed. Even though P2 presented lower measured Q_H, it also presented a much higher peak temperature than P1, suggesting stronger hydrogen trapping. In turn, in P2, the contributions of peak 1 and peak 2 are similar, indicating a more homogeneous distribution of hydrogen through the different trapping sites.

In this case, hydrogen embrittlement was less pronounced in the alloy with lower hydrogen trapping capability, but greater homogeneity of hydrogen trapped at different sites, P2. In contrast, the density and morphology of the lamellae, concomitant with the strain hardening resultant from thermomechanical processing, lead to significant stress concentrations at the ferrite/cementite interfaces and within the ferritic phase in P1. This scenario was critical for the hydrogen embrittlement phenomenon. The behavior of the pearlitic wires resulted in mechanical strength and carbon content directly related to the HEI.

By submitting P1 to heat treatments and deformations, it was possible to isolate the effects of lamellar spacing and strain hardening. A 10% deformation presented a greater effect on delaying hydrogen permeation and lowering hydrogen embrittlement resistance than a 39% refinement of pearlitic lamellae.

Taken together, these results demonstrate that the hydrogen embrittlement susceptibility of commercial ferrite/cementite armor wires is controlled by the interplay between cementite morphology and the microstructural condition. Although globular cementite is generally regarded as beneficial for improving resistance to hydrogen embrittlement, the present results show that cementite morphology alone is insufficient to determine the material response. Instead, cementite morphology dictates distinct relationships between hydrogen trapping, mechanical strength, and hydrogen embrittlement susceptibility. While higher mechanical strength and greater hydrogen trapping were associated with increased susceptibility in the pearlitic microstructures, the opposite trend was observed in the ferritic microstructures containing globular cementite. These findings demonstrate that neither mechanical strength nor hydrogen trapping can be interpreted as universal indicators of hydrogen embrittlement susceptibility without considering the underlying cementite morphology.

3.4. Impact of prior corrosion in CG1 and P1

XRD patterns of the as-polished wires, presented in Figure 6a, are characterized by the presence of ferrite (Fe-α) peaks. The corroded samples’ diffraction patterns, Figure 6d, are characterized by a decreased peak intensity of the ferrite peaks and the presence of Fe₃C peaks. This is an indication that the formation of a carbide-rich film has occurred. These peaks are more intense in armor CG1, where 10.6 vol% of cementite was calculated in the thermodynamic analysis in comparison to the 9.5 vol% in armor P1.

Figure 6
(a,d) XRD patterns, (b,c) microstructure, and (e,f) corroded surfaces of (a-c) wet-ground and (d-f) corroded surfaces of CG1 and P1 armor wires. In the inset of (e,f), AFM images of the corroded surfaces.

SEI of the corroded surfaces are presented in Figures 6e and 6f in contrast to the respective as-polished and etched microstructures in Figures 6b and 6c. In CG1, the presence of fragmented veins of a light gray tone can be pointed out, as indicated by arrows in Figure 6e. The rugosity of this surface, seen in the SEM analysis, does not appear to be related to the distinct behaviors of ferrite and cementite during the corrosive process. In contrast, in P1, the structure of the unconsumed layer after corrosion resembles a network of straight and fragmented segments, pointed in Figure 6f, contrasted at a high relief from the background ferrite.

Because the initial steps of the corrosive process occur through the selective dissolution of ferrite, the morphology of the oxidized surface can be directly related to the distribution of carbides in the steel’s microstructure²⁵,⁴¹,⁶⁵. In fact, the morphologies of the corroded surfaces obtained in CG1, where cementite appears spheroidized and fragmented, and in P1, characterized by the presence of lamellar cementite, are significantly distinct and seem to be associated with their respective microstructures. The preferential dissolution of ferrite in eutectoid and hypoeutectoid microstructures in NaCl solution was widely explored by Katiyar, Misra, and Mondal⁶⁶.

They observed that the relief of the corroded surface is formed by the lamellae of cementite. Accordingly, the structure in the corroded surface of P1 is likely constituted by a porous cementite network, where the voids are regions at which the previous ferrite was consumed²⁵. In turn, in armor CG1, where cementite is finely distributed with a non-continuous nature, the network aspect is not evident, and the Fe₃C fragments appear to have been partially revealed. Measurements using AFM show that the corroded surface of CG1 is characterized by a much higher local roughness than P1, as observed in the index images of Figures 6e and 6f. This can be explained by the fine and fragmented aspect of the cementite dispersed in the ferrite matrix in the first microstructure.

Potentiodynamic polarization, hydrogen permeation, and tensile stress-strain curves obtained for CG1 and P1 in the as-polished and corroded conditions are presented in Figure 7. In the polarization curves, Figures 7a and 7b, the current density under high cathodic potential, where a current density plateau takes place with increasing applied potential, can be directly related to the intensity of hydrogen evolution reactions. An extreme condition is given by the reference high-fugacity electrolyte, H₂SO₄ + As₂O₃. Previous corrosion led to increases in the measured current density in NaCl + CO₂ solution in CG1. However, this effect was not relevant in P1. The surface morphology after corrosion of CG1 presents lower relief separation from ferrite and cementite phases and higher local rugosity in comparison to P1, as seen in SEM and AFM images. From this point of view, a higher overall external surface area occurs in CG1, which could contribute to the increase in hydrogen evolution reactions in comparison with as-polished samples.

Figure 7
(a,b) Cathodic polarization curves, (c-f) hydrogen permeation curves and (g,h) engineering tensile stress-strain curves of (a, c, d, g) CG1 and (b, e, f, h) P1 armor wires.

Alterations in the shape of the cathodic polarization curve can also be noted in the potential range close to the OCP of the cathodic curves. A step on the current density that is not present in the H₂SO₄-based electrolyte is observed in this transition range of the curve obtained in NaCl + CO₂. This characteristic has been previously associated with a mass transport limitation in regions close to OCP due to the occurrence of two dissociation reactions of CO₂. The denoted buffering effect induced by the dissolved gas can result in a reduction of hydrogen ions on the steel’s surface due to limitations on the diffusion³⁵,⁶⁷.

The shape of the curve close to the OCP of corroded samples shows an intermediate figure between that of as-polished samples in NaCl 3.5% + CO₂ and in H₂SO₄-based electrolyte for both microstructures. This behavior indicates that the corroded surfaces appear to diminish the buffering effect in comparison with as-polished samples in the same medium, which could be due to the higher hydrogen generation and/or the higher available surface area compensating for the hydrogen diffusion limitation on the surface.

In the hydrogen permeation curves of previously corroded samples, Figures 7d and 7f, a slight increase in the permeated hydrogen flux can be observed in relation to the as-polished ones for both microstructures, Figures 7c and 7d. In previous studies, two antagonistic effects of a Fe₃C-rich layer have been discussed. One is the increase in the generated hydrogen due to enhancing hydrogen evolution reactions by a larger relative cathodic area³⁵,³⁹. The other is a barrier effect to hydrogen adsorption due to a greater distance between the hydrogen generated in the cementite and the superficial ferrite, where hydrogen is effectively absorbed into the microstructure³³,⁴⁶. The prevalence of one or the other may depend on aspects such as the thickness of the corroded layer and the shape and distribution of the residual cementite. In the present study, the permeated flux indicates that the previously proposed barrier effect was not sufficient to lead to decreases in the permeated hydrogen flux, not even in the pearlitic P1, where a Fe₃C network is present. Instead, increases in the steady-state hydrogen permeation flux could be seen, especially in the armor containing globular and fragmented cementite.

Moreover, hydrogen permeation in corroded CG1 exhibited Fickian behavior with no need to adjust D_app,H, in comparison to a small increase in D_fit,H in contrast to D_app,H in the as-polished condition. In both surface conditions of P1, D_fit,H higher than D_app,H was used. In turn, a significant decrease in Δt, required for fitting in this microstructure, from 7000 s in the as-polished to 5500 s in the corroded surfaces, was observed. This means that the barrier effect previously proposed is not relevant in delaying hydrogen entrance in P1. Both these effects indicate greater efficiency for hydrogen entrance due to the reminiscent cementite-rich surface and faster filling of hydrogen trapping sites.

In the tensile stress-strain curves, Figures 7g and 7h, as-polished samples that were pre-hydrogen-charged for the same time in NaCl + CO₂ medium have shown no significant HE effects measurable by the loss of ductility. In the corroded samples, losses of ductility of 12 and 26% were measured, respectively, for CG1 and P1, in comparison with their AR condition. These differences indicate that the occurrence of corrosion prior to hydrogen charging reveals the HE susceptibility of the steels. The greater HE susceptibility of P1 is in agreement with what was measured after hydrogen charging in H₂SO₄ + As₂O₃. In this manner, the presence of the residual cementite layer contributed to a higher predisposition to hydrogen embrittlement mechanisms in both microstructures in NaCl + CO₂ medium. This effect was accompanied by higher cathodic current densities and hydrogen permeation fluxes. These changes imply enhanced hydrogen entry at the corroded, Fe₃C reminiscent layers.

4. Conclusions

• Hydrogen permeation behavior was strongly governed by cementite morphology. Wires containing globular cementite exhibited lower apparent hydrogen solubility and permeation curves close to theoretical predictions, whereas pearlitic wires showed delayed transient responses, indicating distinct hydrogen trapping mechanisms.

• The interplay between mechanical strength, hydrogen trapping and hydrogen embrittlement depended on the underlying microstructure Although increasing mechanical strength was associated with higher HE susceptibility in the pearlitic wires, the opposite trend was observed in wires containing globular/fragmented cementite.

• In the wires assessed in this study, the globular and fragmented cementite morphology was only effective in ensuring HE susceptibility when hydrogen trapping was great and effective. In contrast, in pearlitic steels, higher hydrogen trapping correlated with increased HE susceptibility.

• Heat treatment and 10% cold rolling of pearlitic P1 indicated a stronger contribution of prior deformation than of the changes in interlamellar spacing associated with the transformation temperature.

• Prior CO₂ corrosion promoting the formation of a Fe₃C-rich surface led to increased cathodic current densities, steady-state hydrogen permeation flux, and, hence, hydrogen entry into both microstructures.

• Although no significant loss of ductility was observed after hydrogen pre-charging polished wires in 3.5 wt.% NaCl + CO₂, in the armors where a Fe₃C-rich layer was formed by prior corrosion, HE was observed in this low-hydrogen-fugacity medium.

From a practical perspective, the present results show that both globular/fragmented cementite in a ferritic matrix and pearlitic microstructures in hypoeutectoid steels may exhibit either high or low HE susceptibility, depending on their associated properties and hydrogen trapping behavior. These findings highlight that the optimization of flexible riser armor wires should rely on tailoring the interplay between cementite morphology, hydrogen trapping characteristics, and deformation history, rather than on mechanical strength or cementite morphology alone. On the other hand, tests performed in NaCl+CO₂ medium have shown that external factors, such as the existence of a corroded layer, can reveal the susceptibility of steel armor wires to HE and lead to unexpected failure

5. Acknowledgments

The authors would like to thank Petrobras, Conselho Nacional de Desenvolvimento Científico (CNPq), and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for the financial support.

6. Data Availability

The dataset supporting the findings of this study will be available upon request to the corresponding author.

7. Supplementary Material

The following online material is available for this article:

Figure S1

Schematic representation of (a) permeation setup; (b) polarization setup; (c) corrosion setup of a permeation sample; and (d) images of the apparatus for hydrogen charging in NaCl 3.5% saturated with CO_2.

Figure S2

(a) Apparent hydrogen diffusivity and (b) solubility extracted from hydrogen permeation tests; (c) desorption peak temperatures and (d) amount of desorbed hydrogen obtained from TDS tests.

8. References

• ¹ Bai Y, Bai Q. Subsea engineering handbook. Cambridge: Elsevier; 2018. Subsea production risers; p. 885-918. https://doi.org/10.1016/B978-0-12-812622-6.00030-0
  » https://doi.org/10.1016/B978-0-12-812622-6.00030-0
• ² Narzul P, Dumay JM, Rigaud J. Contribution of flexible pipes and riser systems to the development of subsea hydrocarbon resources. In: Offshore Technology Conference; 1997; Houston, TX. Proceedings. Richardson: OnePetro; 1997. p. 5-8. https://doi.org/10.4043/8561-MS
  » https://doi.org/10.4043/8561-MS
• ³ Li X, Jiang X, Hopman H. A review on predicting critical collapse pressure of flexible risers for ultra-deep oil and gas production. Appl Ocean Res. 2018;80:1-10. https://doi.org/10.1016/j.apor.2018.08.013
  » https://doi.org/10.1016/j.apor.2018.08.013
• ⁴ Dugstad A, Palencsár S. Bernstsen T, Statoil LB. Corrosion of steel armour wires in flexible pipes: history effects. In: SPE International Oilfield Corrosion Conference and Exhibition; 2018; Aberdeen, Scotland, UK. Proceedings. Aberdeen: Society of Petroleum Engineers; 2018. p. 1-16.
• ⁵ Ohaeri E, Eduok U, Szpunar J. Hydrogen related degradation in pipeline steel: a review. Int J Hydrogen Energy. 2018;43(31):14584-617. https://doi.org/10.1016/j.ijhydene.2018.06.064
  » https://doi.org/10.1016/j.ijhydene.2018.06.064
• ⁶ API: American Petroleum Institute. Recommended practice for flexible pipe. Washington, D.C.; 1998.
• ⁷ Wasim M, Ngo TD. Failure analysis of structural steel subjected to long term exposure of hydrogen. Eng Fail Anal. 2020;114:104606. https://doi.org/10.1016/j.engfailanal.2020.104606
  » https://doi.org/10.1016/j.engfailanal.2020.104606
• ⁸ Hirth JP. Effects of hydrogen on the properties of iron and steel. Metall Trans, A, Phys Metall Mater Sci. 1980;11:861-90.
• ⁹ Ponciano Gomes JAC, Silva SC, Campos T. Stress corrosion cracking susceptibility of armour layers in CO₂ annulus environments – SSRT experimental simulation. Eng Fail Anal. 2022;139:106451. https://doi.org/10.1016/j.engfailanal.2022.106451
  » https://doi.org/10.1016/j.engfailanal.2022.106451
• ¹⁰ Christou M, Konstantinidou M. Safety of offshore oil and gas operations: lessons from past accident analysis. Luxembourg: European Commission; 2012.
• ¹¹ Aljaroudi A, Khan F, Akinturk A, Haddara M, Thodi P. Risk assessment of offshore crude oil pipeline failure. J Loss Prev Process Ind. 2015;37:101-9. https://doi.org/10.1016/j.jlp.2015.07.004
  » https://doi.org/10.1016/j.jlp.2015.07.004
• ¹² Al-Rasheed M, Brell A, Al-Qaffas S, Al-Marri M, Aramco S. Pipeline rupture consequences mitigation comprehensive study. In: PSIG Annual Meeting; 2010; Bonita Springs, FL. Proceedings. Richardson: OnePetro; 2010.
• ¹³ Watanabe T, Yamashita T, Tsuchida Y, Tokuno K. Effect of hydrogen charging on low cycle fatigue life and its dependence on cementite morphology. ISIJ Int. 2012;52(2):222-7. https://doi.org/10.2355/isijinternational.52.222
  » https://doi.org/10.2355/isijinternational.52.222
• ¹⁴ Nagao A, Hayashi K, Oi K, Mitao S. Effect of uniform distribution of fine cementite on hydrogen embrittlement of low carbon martensitic steel plates. ISIJ Int. 2012;52(2):213-21. https://doi.org/10.2355/isijinternational.52.213
  » https://doi.org/10.2355/isijinternational.52.213
• ¹⁵ Saha SK, Moon B, Lee C, Lee S, Chung J, Kang N. Enhancing the hydrogen embrittlement resistance of medium‑carbon high-strength steel by optimizing the tempering temperature. Mater Charact. 2024;207:113530. https://doi.org/10.1016/j.matchar.2023.113530
  » https://doi.org/10.1016/j.matchar.2023.113530
• ¹⁶ Kim SG, Hwang B. Effect of Si on cementite morphology and hydrogen embrittlement resistance of tempered martensitic steels. Met Mater Int. 2025;31(5):1313-22. https://doi.org/10.1007/s12540-024-01826-2
  » https://doi.org/10.1007/s12540-024-01826-2
• ¹⁷ Latypova R, Javaheri V, Claeys L, Verbeken K, Depover T, Kömi J, et al. Effect of rapid tempering and cementite morphology on hydrogen diffusion and trapping in a medium-carbon advanced high-strength steel. Eng Fract Mech. 2025;326:111376. https://doi.org/10.1016/j.engfracmech.2025.111376
  » https://doi.org/10.1016/j.engfracmech.2025.111376
• ¹⁸ Karimi S, Taji I, Hajilou T, Barnoush A, Johnsen R. Evaluation of the cementite morphology influence on the hydrogen induced crack nucleation and propagation path in carbon steels. Int J Hydrogen Energy. 2022;47(30):14121-9. https://doi.org/10.1016/j.ijhydene.2022.01.222
  » https://doi.org/10.1016/j.ijhydene.2022.01.222
• ¹⁹ Seixas Campos E, Silva de Sá J, Campos TS, Souza EA, Ponciano Gomes JAC. Influence of metallurgical factors on the hydrogen induced cracking of carbon steel wires in H₂S-containing environments. Eng Fail Anal. 2024;155:107739. https://doi.org/10.1016/j.engfailanal.2023.107739
  » https://doi.org/10.1016/j.engfailanal.2023.107739
• ²⁰ Niu R, Li H, Liu PY, Burr P, Feng Y, Yen HW, et al. Hydrogen-enhanced deformation in pearlite. Acta Mater. 2024;281:120327. https://doi.org/10.1016/j.actamat.2024.120327
  » https://doi.org/10.1016/j.actamat.2024.120327
• ²¹ Zhang B, Cao Z, Zhou J, Li W, Chen Y, Jia Y, et al. Improving the hydrogen embrittlement resistance by straining the ferrite/cementite interfaces. Acta Mater. 2024;270:119850. https://doi.org/10.1016/j.actamat.2024.119850
  » https://doi.org/10.1016/j.actamat.2024.119850
• ²² Pinson M, Springer H, Depover T, Verbeken K. The role of cementite on the hydrogen embrittlement mechanism in martensitic medium-carbon steels. Mater Sci Eng A. 2022;859:144204. https://doi.org/10.1016/j.msea.2022.144204
  » https://doi.org/10.1016/j.msea.2022.144204
• ²³ Skilbred ES, Lootz SA, Johnsen R. Hydrogen embrittlement susceptibility of steel armor wires for flexible pipes. In: International Corrosion Conference; 2020; Houston, TX. Proceedings. Houston: NACE International; 2020. https://doi.org/10.5006/C2020-14489
  » https://doi.org/10.5006/C2020-14489
• ²⁴ Skilbred ES, Kappes M, Iannuzzi M, Johnsen R. Hydrogen uptake and diffusivity in steel armor wires with different chemical composition, carbide distribution, grain size, and degree of deformation. Mater Corros. 2022;73(3):326-45. https://doi.org/10.1002/maco.202112615
  » https://doi.org/10.1002/maco.202112615
• ²⁵ Karimi S, Taji I, Hajilou T, Palencsár S, Dugstad A, Barnoush A, et al. Role of cementite morphology on corrosion layer formation of high-strength carbon steels in sweet and sour environments. Corros Sci. 2023;214:111031. https://doi.org/10.1016/j.corsci.2023.111031
  » https://doi.org/10.1016/j.corsci.2023.111031
• ²⁶ Truschner M, Deutsch J, Pengg J, Mori G, Gruber P, Köberl H. Influence of cold deformation on hydrogen embrittlement by determining the critical hydrogen concentration in high strength steel for wire applications. In: AMPP Annual Conference & Expo; 2022; San Antonio, TX. Proceedings. Houston: AMPP; 2022. https://doi.org/10.5006/C2022-17639
  » https://doi.org/10.5006/C2022-17639
• ²⁷ Zhang T, Chen Z, Li K, Zhang N. Investigation on the hydrogen induced cracking behaviour of heat-treated pipeline steel. Eng Fail Anal. 2024;157:107909. https://doi.org/10.1016/j.engfailanal.2023.107909
  » https://doi.org/10.1016/j.engfailanal.2023.107909
• ²⁸ Carvalho MM, Lopes NF, dos Santos CA, Schroeder RM, da Costa EM. Hydrogen embrittlement susceptibility of AISI 4140 alloy-steel under cathodic protection and in H₂S and CO₂ environment. Eng Fail Anal. 2024;161:108327. https://doi.org/10.1016/j.engfailanal.2024.108327
  » https://doi.org/10.1016/j.engfailanal.2024.108327
• ²⁹ Silva SC, Silva AB, Ponciano Gomes JAC. Hydrogen embrittlement of API 5L X65 pipeline steel in CO₂ containing low H₂S concentration environment. Eng Fail Anal. 2021;120:105081. https://doi.org/10.1016/j.engfailanal.2020.105081
  » https://doi.org/10.1016/j.engfailanal.2020.105081
• ³⁰ Huang F, Cheng P, Zhao XY, Liu J, Hu Q, Cheng YF. Effect of sulfide films formed on X65 steel surface on hydrogen permeation in H₂S environments. Int J Hydrogen Energy. 2017;42(7):4561-70. https://doi.org/10.1016/j.ijhydene.2016.10.130
  » https://doi.org/10.1016/j.ijhydene.2016.10.130
• ³¹ Henthorne M. The slow strain rate stress corrosion cracking test-a 50 year retrospective. Corrosion. 2016;72(12):1488-518. https://doi.org/10.5006/2137
  » https://doi.org/10.5006/2137
• ³² Wen X, Bai P, Luo B, Zheng S, Chen C. Review of recent progress in the study of corrosion products of steels in a hydrogen sulphide environment. Corros Sci. 2018;139:124-40. https://doi.org/10.1016/j.corsci.2018.05.002
  » https://doi.org/10.1016/j.corsci.2018.05.002
• ³³ Karimi S, Taji I, Palencsár S, Dugstad A, Hajilou T, Barnoush A, et al. Evaluation of microstructural and environmental effects on the hydrogen uptake and desorption in high-strength carbon steels: a thermal desorption spectroscopy study. Corros Sci. 2023;219:111210. https://doi.org/10.1016/j.corsci.2023.111210
  » https://doi.org/10.1016/j.corsci.2023.111210
• ³⁴ Jeon J, Ahmed R, Elgaddafi R, Teodoriu C. Hydrogen embrittlement of high-strength API carbon steels in H₂S and CO₂ containing environments. J Nat Gas Sci Eng. 2022;104:104676. https://doi.org/10.1016/j.jngse.2022.104676
  » https://doi.org/10.1016/j.jngse.2022.104676
• ³⁵ Silva SC, Souza EA, Pessu F, Hua Y, Barker R, Neville A, et al. Cracking mechanism in API 5L X65 steel in a CO₂-saturated environment. Eng Fail Anal. 2019;99:273-91. https://doi.org/10.1016/j.engfailanal.2019.02.031
  » https://doi.org/10.1016/j.engfailanal.2019.02.031
• ³⁶ Plennevaux C, Kittel J, Frégonèse M, Normand B, Ropital F, Grosjean F, et al. Contribution of CO₂ on hydrogen evolution and hydrogen permeation in low alloy steels exposed to H₂S environment. Electrochem Commun. 2013;26:17-20. https://doi.org/10.1016/j.elecom.2012.10.010
  » https://doi.org/10.1016/j.elecom.2012.10.010
• ³⁷ Majchrowicz K, Brynk T, Wieczorek M, Miedzińska D, Pakieła Z. Exploring the susceptibility of P110 pipeline steel to stress corrosion cracking in CO₂-rich environments. Eng Fail Anal. 2019;104:471-9. https://doi.org/10.1016/j.engfailanal.2019.06.016
  » https://doi.org/10.1016/j.engfailanal.2019.06.016
• ³⁸ Sun C, Yang W, Sun J, Yan X, Lin X, Xu X. Hydrogen permeation and SCC susceptibility of X70 pipeline steel in CO₂-saturated water environment containing acidic impurity. Corros Sci. 2024;236:112259. https://doi.org/10.1016/j.corsci.2024.112259
  » https://doi.org/10.1016/j.corsci.2024.112259
• ³⁹ Silva SC, Silva AB, Folena MC, Barker R, Neville A, Ponciano Gomes JAC. Cracking mechanism in API 5L X65 steel in a CO₂-saturated environment – part II: a study under cathodic polarisation. Eng Fail Anal. 2020;113:104550. https://doi.org/10.1016/j.engfailanal.2020.104550
  » https://doi.org/10.1016/j.engfailanal.2020.104550
• ⁴⁰ Sadeghi MA, Javidi M. Investigation on stress corrosion cracking of API 5L X65 steel in CO₂ corrosion medium. Mater Sci Eng A. 2021;824:141856. https://doi.org/10.1016/j.msea.2021.141856
  » https://doi.org/10.1016/j.msea.2021.141856
• ⁴¹ Kermani MB, Morshed A. Carbon dioxide corrosion in oil and gas production: a compendium. Corrosion. 2003;59(8):659-83. https://doi.org/10.5006/1.3277596
  » https://doi.org/10.5006/1.3277596
• ⁴² Campos T, Gomes JA, Seiersten M, Palencsár S, Dugstad A. Corrosion of armor wire steel in the annulus of flexible pipes at near neutral pH. In: Corrosion Conference & Expo; 2019; Nashville, TN. Proceedings. Houston: NACE International; 2019. https://doi.org/10.5006/C2019-13302
  » https://doi.org/10.5006/C2019-13302
• ⁴³ López DA, Schreiner WH, De Sánchez SR, Simison SN. The influence of carbon steel microstructure on corrosion layers: an XPS and SEM characterization. Appl Surf Sci. 2003;207(1-4):69-85. https://doi.org/10.1016/S0169-4332(02)01218-7
  » https://doi.org/10.1016/S0169-4332(02)01218-7
• ⁴⁴ Zhang T, Zhao W, Zhao Y, Ouyang K, Deng Q, Wang Y, et al. Effects of surface oxide films on hydrogen permeation and susceptibility to embrittlement of X80 steel under hydrogen atmosphere. Int J Hydrogen Energy. 2018;43(6):3353-65. https://doi.org/10.1016/j.ijhydene.2017.12.170
  » https://doi.org/10.1016/j.ijhydene.2017.12.170
• ⁴⁵ Paolinelli LD, Pérez T, Simison SN. The effect of pre-corrosion and steel microstructure on inhibitor performance in CO₂ corrosion. Corros Sci. 2008;50(9):2456-64. https://doi.org/10.1016/j.corsci.2008.06.031
  » https://doi.org/10.1016/j.corsci.2008.06.031
• ⁴⁶ Skilbred ES, Palencsár S, Dugstad A, Johnsen R. Hydrogen uptake during active CO₂-H2S corrosion of carbon steel wires in simulated annulus fluid. Corros Sci. 2022;199:110172. https://doi.org/10.1016/j.corsci.2022.110172
  » https://doi.org/10.1016/j.corsci.2022.110172
• ⁴⁷ Devanathan MAV, Stachurski Z. The adsorption and diffusion of electrolytic hydrogen in palladium. Proc R Soc Lond. 1962;270:90-102.
• ⁴⁸ Boes N, Züchner H. Electrochemical methods for studying diffusion, permeation and solubility of hydrogen in metals. J Less Common Met. 1976;49:223-40. https://doi.org/10.1016/0022-5088(76)90037-0
  » https://doi.org/10.1016/0022-5088(76)90037-0
• ⁴⁹ Silverstein R, Eliezer D, Tal-Gutelmacher E. Hydrogen trapping in alloys studied by thermal desorption spectrometry. J Alloys Compd. 2018;747:511-22. https://doi.org/10.1016/j.jallcom.2018.03.066
  » https://doi.org/10.1016/j.jallcom.2018.03.066
• ⁵⁰ Kawakami K, Matsumiya T. Ab-initio investigation of hydrogen trap state by cementite in bcc-Fe. ISIJ Int. 2013;53(4):709-13. https://doi.org/10.2355/isijinternational.53.709
  » https://doi.org/10.2355/isijinternational.53.709
• ⁵¹ Takai K, Nozue A. Hydrogen permeation and trapping characteristics of cold-drawn pure iron and eutectoid steel evaluated by thermal desorption spectrometry. In: CORROSION 2000; 2000; Orlando, FL. Proceedings. Houston: NACE International; 2000. https://doi.org/10.5006/C2000-00468
  » https://doi.org/10.5006/C2000-00468
• ⁵² Truschner M, Pengg J, Loder B, Köberl H, Gruber P, Moshtaghi M, et al. Hydrogen resistance and trapping behaviour of a cold-drawn ferritic-pearlitic steel wire. Int J Mater Res. 2023;114(6):439-52. https://doi.org/10.1515/ijmr-2022-0376
  » https://doi.org/10.1515/ijmr-2022-0376
• ⁵³ Toribio J. HELP versus HEDE in progressively cold-drawn pearlitic steels: between Donatello and Michelangelo. Eng Fail Anal. 2018;94:157-64. https://doi.org/10.1016/j.engfailanal.2018.07.026
  » https://doi.org/10.1016/j.engfailanal.2018.07.026
• ⁵⁴ Abraham DP, Altstetter CJ. Hydrogen-enhanced localization of plasticity in an austenitic stainless steel. Metall Mater Trans, A Phys Metall Mater Sci. 1995;26(11):2859. https://doi.org/10.1007/BF02669644
  » https://doi.org/10.1007/BF02669644
• ⁵⁵ Martin ML, Fenske JA, Liu GS, Sofronis P, Robertson IM. On the formation and nature of quasi-cleavage fracture surfaces in hydrogen embrittled steels. Acta Mater. 2011;59(4):1601-6. https://doi.org/10.1016/j.actamat.2010.11.024
  » https://doi.org/10.1016/j.actamat.2010.11.024
• ⁵⁶ Birnbaum HK, Sofronis P. Hydrogen-enhanced localized plasticity-a mechanism for hydrogen-related fracture. Mater Sci Eng A. 1994;176(1-2):191-202. https://doi.org/10.1016/0921-5093(94)90975-X
  » https://doi.org/10.1016/0921-5093(94)90975-X
• ⁵⁷ Michler T, Naumann J. Microstructural aspects upon hydrogen environment embrittlement of various bcc steels. Int J Hydrogen Energy. 2010;35(2):821-32. https://doi.org/10.1016/j.ijhydene.2009.10.092
  » https://doi.org/10.1016/j.ijhydene.2009.10.092
• ⁵⁸ Takahashi J, Kawakami K, Kobayashi Y. Origin of hydrogen trapping site in vanadium carbide precipitation strengthening steel. Acta Mater. 2018;153:193-204. https://doi.org/10.1016/j.actamat.2018.05.003
  » https://doi.org/10.1016/j.actamat.2018.05.003
• ⁵⁹ Wei FG, Tsuzaki K. Hydrogen trapping phenomena in martensitic steels. In: Gangloff RP, Somerday BP, editors. Gaseous hydrogen embrittlement of materials in energy technologies. Cambridge: Woodhead Publishing; 2012. p. 493-525. (vol. 2). https://doi.org/10.1533/9780857093899.3.493
  » https://doi.org/10.1533/9780857093899.3.493
• ⁶⁰ Hagi H. Effect of interface between cementite and ferrite on diffusion of hydrogen in carbon steels. Mater Trans. 1994;35:168-73.
• ⁶¹ Cardenas AL, Silva RO, Eckstein CB, dos Santos DS. Hydrogen effect on 2.25Cr–1Mo–0.25V bainitic steel under aging heat treatment. Int J Hydrogen Energy. 2018;43(33):16400-10. https://doi.org/10.1016/j.ijhydene.2018.07.047
  » https://doi.org/10.1016/j.ijhydene.2018.07.047
• ⁶² Toribio J, Ovejero E. Hydrogen assisted cracking in progressively drawn pearlitic steel. Corros Sci. 2007;49(9):3539-56. https://doi.org/10.1016/j.corsci.2007.03.016
  » https://doi.org/10.1016/j.corsci.2007.03.016
• ⁶³ Enos DG, Scully JR. A critical-strain criterion for hydrogen embrittlement of cold-drawn, ultrafine pearlitic steel. Metall Mater Trans, A Phys Metall Mater Sci. 2002;33(4):1151-66. https://doi.org/10.1007/s11661-002-0217-z
  » https://doi.org/10.1007/s11661-002-0217-z
• ⁶⁴ Li X, Zhang J, Cui Y, Djukic MB, Feng H, Wang Y. Review of the hydrogen embrittlement and interactions between hydrogen and microstructural interfaces in metallic alloys: grain boundary, twin boundary, and nano-precipitate. Int J Hydrogen Energy. 2024;72:74-109. https://doi.org/10.1016/j.ijhydene.2024.05.257
  » https://doi.org/10.1016/j.ijhydene.2024.05.257
• ⁶⁵ Farelas F, Galicia M, Brown B, Nesic S, Castaneda H. Evolution of dissolution processes at the interface of carbon steel corroding in a CO₂ environment studied by EIS. Corros Sci. 2010;52(2):509-17. https://doi.org/10.1016/j.corsci.2009.10.007
  » https://doi.org/10.1016/j.corsci.2009.10.007
• ⁶⁶ Katiyar PK, Misra S, Mondal K. Corrosion behavior of annealed steels with different carbon contents (0.002, 0.17, 0.43 and 0.7% C) in freely aerated 3.5% NaCl solution. J Mater Eng Perform. 2019;28(7):4041-52. https://doi.org/10.1007/s11665-019-04137-5
  » https://doi.org/10.1007/s11665-019-04137-5
• ⁶⁷ Remita E, Tribollet B, Sutter E, Vivier V, Ropital F, Kittel J. Hydrogen evolution in aqueous solutions containing dissolved CO₂: quantitative contribution of the buffering effect. Corros Sci. 2008;50(5):1433-40. https://doi.org/10.1016/j.corsci.2007.12.007
  » https://doi.org/10.1016/j.corsci.2007.12.007

*

e-mail: [e-mail]

Associate Editor:

Hugo Sandim.

Editor-in-Chief:

Luiz Antonio Pessan.

Publication Dates

• Publication in this collection
  18 Sept 2026
• Date of issue
  2026

History

• Received
  13 May 2026
• Reviewed
  18 Aug 2026
• Accepted
  20 Aug 2026