Influence of Centerline Segregation Region on the Hydrogen Embrittlement Susceptibility of API 5L X80 Pipeline Steels

Abstract

The influence of the centerline segregation region (CSR) on the hydrogen embrittlement (HE) of two different API 5L X80 pipeline steel plates was investigated. The novelty of this work was to establish relationships between the CSR, microstructure, and distribution of localized fragile particles on HE susceptibility and on fracture morphology. This work intended to establish a relationship between centerline segregation and HE susceptibility in high-strength low-alloy steels submitted to inhomogeneous transformations. Microscopy, hydrogen permeation, and slow strain rate (SSR) tests were used to investigate hydrogen-related degradation. The solution used on the charging cell of the permeation tests—and on the SSR test cell—was 0.5 mol L⁻¹ H₂SO₄ + 10 mg L⁻¹ As₂O₃, and in the oxidation cell, 0.1 M NaOH was used as a solution. The CSR led the thicker plate to present the highest HE index (0.612) in analyses carried out in the mid-thickness; however, the same plate showed the lowest HE index in near-surface tests. The presence of hydrogen changed the fracture morphology from ductile to a brittle and ductile feature; this occurred due to the interaction with localized fragile particles and the significant reduction of the shear stress necessary for the dislocation movement.

1. Introduction

The oil and gas (OG) industry demands high-performance pipeline steels, due to severe operating conditions (including high pressures), sour environments, and corrosive crude oil [1,2]. The development of thermo-mechanical controlled processing (TMCP), which involves controlled rolling and subsequent rapid cooling, has yielded refined microstructures, thereby markedly enhancing the strength and toughness of steel [3]. To meet the growing global energy demand, significant investments have been directed toward improving API 5L X80 steels, which are recognized for their high strength, weldability, toughness, and resistance to corrosion. These characteristics make API 5L X80 one of the most widely used grades for oil and gas transportation [4,5].

However, component and structural failures associated with material degradation in service are reported [6]. Steels used in sour environments absorb atomic hydrogen produced by electrochemical reactions [7]. Hydrogen permeation can lead to hydrogen embrittlement (HE), a deleterious phenomenon that reduces mechanical properties. High-strength, low-alloy (HSLA) steels, such as API 5L X80, used for oil and gas pipelines can be more susceptible to HE, owing to their higher hardness values [8].

Microstructures are known to influence hydrogen diffusion and crack nucleation and propagation in pipeline steels [9]. Microstructural heterogeneities can reduce hydrogen mobility by acting as trapping sites [10]. Therefore, the microstructure plays a critical role in determining the susceptibility of pipeline steels to hydrogen-induced degradation, and microstructure design has been used as an embrittlement mitigation strategy [6,11].

Some researchers have used the hydrogen microprint technique and proposed that, in ferritic/pearlitic steels, the main hydrogen pathways are the lattice and at the carbide-ferrite interfaces, whereas, in martensitic steels, the main diffusion pathways are along the lath interfaces [12,13]. Also, the increase in the grain size decreases the hydrogen trapping by grain boundaries [14].

According to Park et al. [15], trapping efficiency in API 5L X65 increases in the following order: degenerated pearlite, bainite, and acicular ferrite. Moreover, interfaces between retained austenite and martensitic layers can act as trapping sites [14].

Pipeline steels produced in the 1960s through hot rolling typically exhibited a coarse microstructure of ferrite and pearlite, with manganese sulfide inclusions, oriented along the hot rolling direction [16]. Under a continuous supply of hydrogen, steels with high toughness and finer grain size (higher fraction of grain boundaries) can be considered more resistant to hydrogen embrittlement. This results from the reduced grain size, which hinders crack tip propagation and limits hydrogen access to more susceptible regions. Hence, acicular ferrite and bainitic ferrite can improve the resistance to hydrogen-related damage more than ferritic pearlite or martensitic microstructures [17].

Also, the chemical composition of HSLA steels and the possible presence of segregation regions play important roles in material behavior in sour service [18,19,20]. The centerline segregation region (CSR) is observed especially in thick plates subjected to continuous casting and is a well-understood phenomenon. Guo et al. [20] studied the effect of segregation on the low-temperature toughness and elongation of pipeline steel. Shant et al. [21] examined the adverse effect of delamination caused by the presence of CSR. Beidokhti et al. [4] investigated the influence of alloying elements, microstructure, and CSR on the susceptibility to hydrogen-related damage in API 5L X70 welds. Other researchers [5,18,22] have also observed that cracks initiate in the CSR, potentially leading to hydrogen-induced fracture, even in the absence of applied external stress.

The CSR exhibits higher hardness values due to chemical heterogeneity, such as the segregation of carbides and alloying elements [23]. Beidokhti et al. [4] found Vickers microhardness values of up to 358 HV in the CSR—a value that is more than 88% higher when compared to the average of the base metal (190 HV). The presence of high amounts of sulfur and phosphorus can also play a significant role in the quality problem of centerline segregation. Thus, centerline segregation is harmful to the resistance to hydrogen embrittlement of HSLA steels [4,23,24]. Nonetheless, although interest in hydrogen embrittlement mechanisms is increasing, information on the impact of hydrogen interactions with the CSR on the mechanical properties of API 5L X80 is limited. A deeper understanding of the factors leading to embrittlement and subsequent component fracture is needed. In this work, the influence of the centerline segregation region on the hydrogen embrittlement of two different API 5L X80 pipeline steels was investigated. The research was conducted to verify in an unprecedented way the joint influence of the CSR (in thick plates), microstructure, and distribution of localized fragile particles on the HE susceptibility and on the fracture morphology.

2. Materials and Methods

2.1. Materials

API 5L X80 pipeline steel plates of 20.0 mm (plate 1) and 38.1 mm (plate 2) thickness, produced by the TMCP process, were used. Their characteristics and chemical composition are presented in

Table 1. Characteristics of the two studied API 5L steel plates.

Identification	Thickness (mm)	Microhardness (HV0.2)	Average Grain Size (µm)
Plate 1	20.0	255 ± 10.54	3.75 ± 0.41
Plate 2	38.1	230 ± 23.5	4.19 ± 0.52

and

Table 2. Chemical composition of the two studied API 5L steel plates (wt.%).

		Main Chemical Elements Percentages (wt.%)
Plate 1	C	Si	Mn	P	S	Nb	V	Ti	Cu	Cr	Mo	Ni	Fe
	0.061	0.288	1.790	0.018	0.003	0.037	0.022	0.012	0.012	0.154	<0.002	0.032	Bal.
Plate 2	C	Si	Mn	P	S	Nb	V	Ti	Cu	Cr	Mo	Ni	Fe
	0.053	0.252	1.670	0.016	0.003	0.006	0.019	0.007	0.292	0.036	0.113	0.386	Bal.

, respectively.

2.2. Microstructural Characterization and Microhardness Investigations

Microstructural characterization analyses were performed by optical microscopy (OM) using an Olympus BX51 microscope (Tokyo, Japan). The metallographic samples were ground and polished using conventional methods. Sample surfaces were sequentially ground up to 1200 grit paper and polished up to a 1 μm before etching in 3% Nital for 15 s. The samples were washed with distilled water, then with isopropyl alcohol, and dried in a hot air stream. Scanning electron microscopy (SEM) and energy-dispersive x-ray spectrum (EDS) were used to observe the CSR in the thick plate and test the fracture surface of samples for hydrogen embrittlement susceptibility. A microhardness test was also performed using the Vickers method, applying 200 gf and dwell time of 15 s. A profile was obtained from the averages of measurements from seven equally spaced regions along the transverse section, which included the top, mid-thickness, and bottom. Twenty measurements were taken in each area.

2.3. Hydrogen Permeation Tests

The galvanostatic method was used for the hydrogen permeation tests at room temperature, under the ASTM G148 standard [25] requirements. In this technique, a constant cathodic electric current is applied to the charging cell to generate hydrogen atoms on the membrane surface exposed to the sour environment and promote hydrogen permeation; on the other hand, in the oxidation cell, hydrogen is oxidized, and the oxidation current is then obtained as a function of time, indicating the flow of hydrogen.

The disk samples were obtained by electrochemical discharge machining (EDM) from the surface of the plates. Both surfaces of the disk samples were sequentially ground up to 1000 grit paper and then polished with a 1 μm alumina. At the end of the preparation process, the samples were 40 mm in diameter and 1 mm thick. The solution used on the charging cell was 0.5 mol L⁻¹ H₂SO₄ + 10 mg L⁻¹ As₂O₃, and in the oxidation cell, 0.1 M NaOH was used as a solution. The solutions of both cells were purged with nitrogen before applying the current. The orientation of the hydrogen flux was transversal to the thickness.

At the beginning of the procedure, a potential of +300 mV was applied to the oxidation cell to ensure anodic polarization. Subsequently, a 5 mA cm⁻² cathodic current density was applied to electrolyze the solution from the charging cell and promote the entry of hydrogen.

The hydrogen flow through the membrane was measured in terms of current density at steady-state and converted to the hydrogen permeation flow [Jss, molH (s·cm²)⁻¹] according to Equation (1):

$$\mathrm{Jss}=i_p^{\infty }/F\times A,$$

(1)

where $i_p^{\infty }$ is the steady-state permeation current (A); F is the Faraday constant (C mol⁻¹); A is the area (cm²). On the other hand, the permeability [φ, molH (s·cm)⁻¹] is calculated using Equation (2):

$$\phi =J_{ss}L=\frac{i_p^{\infty }\times L}{F\times A},$$

(2)

where L is the thickness of the specimen (cm).

The apparent diffusivity coefficient (Dapp, cm² s⁻¹) was determined using the time-lag method. Therefore, Equation (3) was considered:

$$D_{app}=\frac{L^2}{6\times {\tau }_{lag}},$$

(3)

where ${\tau }_{lag}$ is obtained from the time taken for the permeation rate to reach 0.63 times the steady-state value. In this regard, it is important to point out that the term “apparent diffusivity” is more suitable for the present study, since it was calculated from the first permeation transient. Therefore, it includes the traps’ effects—which reduce the apparent diffusivity and increase the apparent solubility.

The solubility—or apparent solubility—(S, molH/cm³) is given by Equation (4):

$$\mathrm{S}=\frac{\mathsf{\phi }}{{\mathrm{D}}_{\mathrm{a}\mathrm{p}\mathrm{p}}},$$

(4)

2.4. Hydrogen Embrittlement Susceptibility Tests

The hydrogen embrittlement susceptibility was evaluated by slow strain rate tests (SSRT) according to ASTM G129-21 [26]. Cylindrical small-size specimens with a diameter of 2.5 mm and gauge length of 10.0 mm were obtained transversely to the rolling direction, in accordance with the ASTM-E8 standard [27]. In this study, specimens were obtained from the mid-thickness (MT) and the near-surface (NS) regions (as shown in Figure 1); they were tested in both air and an aqueous solution of 0.5 mol L⁻¹ H₂SO₄ + 10 mg L⁻¹ As₂O₃. Before the analysis, nitrogen gas was used to purge possible dissolved oxygen gas. The specimens tested in a hydrogen environment were firstly pre-charged in the electrochemical solution for 96 h. They were then loaded in the same hydrogen-charged environment at a strain rate of 3.35 × 10⁻⁶ s⁻¹. The current density of the electrochemical hydrogen charging experiment was 5 mA cm⁻².

Table 3. Experimental procedure for slow strain rate tests.

Identification	Material	Test Environment	Axis of the Test Specimen	ø (mm)
P1.NS—air	Plate 1	Air	Near-surface	2.5
P2.NS—air	Plate 2	Air	Near-surface
P1.MT—air	Plate 1	Air	Mid-thickness
P2.MT—air	Plate 2	Air	Mid-thickness
P1.NS—solution	Plate 1	Solution *	Near-surface
P2.NS—solution	Plate 2	Solution *	Near-surface
P1.MT—solution	Plate 1	Solution *	Mid-thickness
P2.MT—solution	Plate 2	Solution *	Mid-thickness

* 0.5 mol/L H₂SO₄ + 10 mg/L As₂O₃.

shows a summary of the experimental procedure.

The loss of ductility between equivalent samples, tested in solution and the air, was evaluated by calculating the hydrogen embrittlement index, I (ε), given by Equation (5):

$$\mathrm{I}\left(\mathrm{\epsilon }\right)=\frac{{\mathrm{\epsilon }}_{uncharged}-{\mathrm{\epsilon }}_{charged}}{{\mathrm{\epsilon }}_{uncharged}},$$

(5)

where ε_uncharged is the elongation to failure of the uncharged specimen (%), and ε_charged is the elongation to failure of charged specimen (%).

3. Results and Discussion

3.1. Microstructural Characterization

The microstructure analysis by optical microscopy of the thinner plate after etching with 3% Nital reagent, in ¼ of the thickness (Figure 2a) and in ½ of the thickness (Figure 2b), showed a uniform microstructure, with refined grains and with no microstructural banding. At ¼ of the thickness of plate 1, the average grain size was determined to be 3.63 ± 0.31 μm. Conversely, at mid-thickness, the average grain size increased to 3.87 ± 0.53 μm. The microstructure presented, according to Krauss’ and Thompson’s [28] nomenclature, was polygonal ferrite (α_P) grains, bainite (B), ferrite-carbide aggregates (FCA) and martensite-austenite (MA) microconstituent.

A refined microstructure is crucial for meeting the mechanical property requirements of HSLA pipeline steels. By initiating accelerated cooling below the Ar₃ temperature, a soft ferritic matrix with dispersed hard bainite was obtained, resulting in a complex microstructural arrangement that has the potential to enhance both strength and toughness. Ferrite plays a key role in reducing brittleness, nucleating at the edges of prior austenitic grains to suppress crack propagation [8]. Although both plates underwent similar processing routes (TMCP), differences in thickness and chemical composition led to variations in the bainite and ferrite structures along the transverse sections.

In ¼ of the thickness, plate 2 (38.1 mm thick) had a greater tendency to form polygonal ferrite, with remarkable grain growth, and regions enriched with carbon (Figure 2c). Similar events were observed by Cizek et al. [29] for products obtained from lower cooling rates. In the thickness center of plate 2, a segregation region (Figure 2d) presenting regions enriched with carbon (dark areas) was observed. CSR represented about 1.05% of the cross-sectional area. At ¼ of the thickness of plate 2, the average grain size was 4.15 ± 0.45 μm. In contrast, at mid-thickness, which included both the segregated region and the surrounding area (that cooled more slowly compared to the region at ¼ thickness), the average grain size increased to 4.23 ± 0.60 μm. According to Mishra and Datta [30], the segregation of carbon and manganese in the centerline favors the formation of pearlite and bainite, being bainite sites of crack nucleation. Additionally, HSLA steels are prone to form bainitic microstructures due to the presence of alloy elements [6]. Figure 3 presents in greater detail the transition between the CSR and the neighboring regions. The formation of CSR in steel plates subjected to continuous casting occurs during the solidification process due to uneven heat dissipation, leading to non-uniform redistribution of chemical elements such as C, Mn, and P. This results in concentration variations across the plate, with certain elements accumulating predominantly in the central region [19,23].

EDS analysis allowed analyzing Mn, P and S distribution in different regions of plate 2. Fractions of P and S were not verified in tests performed on the near-surface area (Figure S1). As pointed out by Mishra and Datta [30], sulfur and phosphorus, as well as carbon and manganese, have a strong tendency to segregate in the centerline. Analysis carried out on the CSR revealed 0.5% S, and the percentage of P reached 0.2 (Figure S2). Regarding the percentage of manganese, analyses carried out on the near-surface region indicated 1.9% Mn, while in the CSR, the % Mn was 4.0 (i.e., a 2.1-fold increase).

The CSR has predominantly dark regions (ferrite-carbide aggregates and bainite) and more refined ferritic microstructures. The delayed austenite-to-ferrite transformation in the CSR, due to Mn migration, promotes carbon segregation to the mid-thickness, leading to a higher fraction of ferrite–carbide aggregates and regions with increased hardenability.

SEM images (Figure 4) indicate the presence of localized fragile particles (bright particles) in the CSR of plate 2, such as martensite–austenite microconstituents, which are mostly associated with ferritic grain boundaries. Martensite–austenite microconstituents are primarily located along ferrite grain boundaries, due to carbon partitioning into the remaining austenite during the formation of ferrite and bainite. Some carbon-enriched austenitic regions do not fully transform into bainite and instead form martensite and retained austenite upon cooling [5].

For plate 1, the microhardness profile presents higher values in regions close to the top and bottom surfaces, with a drop in hardness in the mid-thickness, as shown in Figure 5. Along with the higher cooling rate at the surfaces during manufacturing, the rolling cylinders contribute to work hardening in the final stages, preventing ferritic recrystallization.

For plate 2, in non-segregated areas, the average microhardness was lower compared to plate 1 (Figure 5). This can be explained by the lower cooling rate during manufacturing, as well as by the lower percentage of alloying elements such as C, Nb, V, and Ti. Higher percentages of C can increase the hardness by solid solution and by the formation of carbides and carbonitrides of greater hardness. Ti, V, and Nb can increase hardness through grain refinement or precipitation [31].

The CSR presented a high average microhardness value (287 HV₀.₂)—almost 25% higher when compared to the average microhardness of plate 2. Ultra-microhardness testing procedures with a Vickers indenter were carried out on the CSR of plate 2. The average ultra-microhardness and standard deviation were 356.17 ± 31.30 HV₀.₂ for ferrite-carbide aggregates and 247.23 ± 27.45 HV₀.₂ for the ferritic areas. Figure 6 presents examples of measurements performed on the CSR.

3.2. Electrochemical Hydrogen Permeation Test

Figure 7 shows the hydrogen permeation curves. We highlight the existence of a current drop characteristic of the test, but it is not displayed in Figure 7 because it is associated with steel oxidation under anodic polarization in the oxidation cell and not with hydrogen permeation. A cathodic current density was applied to electrolyze the charge cell solution when the current reached negligible values. The point corresponding to zero seconds in hydrogen permeation curves corresponds to that moment of starting hydrogen generation.

After hydrogen generation was started, the curves grew fast, reached a maximum current density value, and remained constant for a short time. Subsequently, they decreased continuously, even though polarization continued. This typical behavior, according to Haq et al. [11], occurs due to the progressive formation of a passive oxide layer on the exit surface, which does not allow a steady-state permeation current density to be maintained for long periods, presenting itself as a decrease in the electrochemical permeation curve, even with the polarization conditions maintained. The values of the apparent diffusivity coefficient (Dapp), permeability (φ) and apparent solubility (S) were calculated and are shown in

Table 4. Parameters calculated from hydrogen permeation curves of plates 1 and 2.

	${\mathbf{D}}_{\mathbf{a}\mathbf{p}\mathbf{p}}$ (cm2/s)	Permeability [molH/(s·cm)]	Solubility (mol/cm3)
Plate 1	6.17 × 10−6 ± 4.49 × 10−7	2.97 × 10−11 ± 5.01 × 10−12	4.82 × 10−6 ± 1.82 × 10−7
Plate 2	8.07 × 10−6 ± 3.48 × 10−6	5.71 × 10−11 ± 5.41 × 10−11	7.08 × 10−6 ± 1.64 × 10−6

.

Previous works show that analyses performed on different solutions and experimental conditions lead to different permeation curve behaviors and, consequently, lead to some differences in parameter calculations. Zhang et al. [1] studied hydrogen permeation in API 5L X80 steel with experimental conditions similar to the present work and obtained a Dapp value equal to 5.75 × 10⁻⁶ cm² s⁻¹, showing a similarity in the results obtained. Additionally, Xue and Cheng [32] tested without poisons an API 5L X80 steel with microstructural constituents mainly composed of α_P, bainitic ferrite (${\alpha }_B^0$), and martensite–austenite microconstituent distributed in the grain boundaries and found a diffusivity coefficient equal to 2.00 × 10⁻¹¹ m² s⁻¹ (2 × 10⁻⁷ cm² s⁻¹). This result presents a more evident distinction when compared to the present study. At first, these results can show the poisoner’s action as promoters of the hydrogen entrance, lowering the reaction time and increasing the apparent diffusion coefficient. Secondly, they indicate different microstructure effects on the diffusion process.

Comparing the two plates studied in this work, plate 1 presented lower average val-ues of Dapp, permeability, and solubility. Since reversible hydrogen traps allow hydrogen to be released as diffusion occurs, apparent diffusivity is related to diffusivity in the crystalline lattice and reversible traps. In this regard, the plates showed different percentages of microconstituents. In the case of primary ferritic phases, plate 2 (38.1 mm thick) showed larger αp grains, implying greater diffusivity. In addition, interfaces between ferrite and ferrite–carbide aggregates were more frequent in plate 1 (20.0 mm thick) and can act as hydrogen trapping sites. Probably, such microstructure played a significant role in the lower Dapp from plate 1 by reducing hydrogen mobility [11].

Liu et al. [33], using hydrogen permeation in pure iron and two ferritic–pearlitic steels, reported that hydrogen permeation is smoother through ferrite/cementite interfaces in pearlitic microstructures, in comparison to grain boundaries and dislocations. The authors also reported the traps are stronger in grain boundaries and in dislocations. Therefore, more refined grains—as the thinner plate, in this study—further lower diffusivity coefficients, since grain boundaries are high-energy trapping sites.

3.3. Slow Strain Rate Tests

Figure 8 and Figure 9 show the curves obtained for all conditions evaluated for plate 1 and plate 2, respectively, in air and solution tests. The pre-charging ensured the saturation of the traps with hydrogen before deformation. In the second step, the tests were carried out at a low strain rate to allow the new dislocations formed by the plastic deformation to be filled by hydrogen [34]. As a result, there was a loss of ductility in all the conditions analyzed, represented by the decrease in elongation to failure of the specimens tested in a sour environment, indicating that the materials studied here are susceptible to hydrogen embrittlement.

In this study, such loss of ductility shows a relevant reduction of the energy-absorption ability due to the presence of hydrogen in the metal alloy [35]. The decrease in elongation-to-failure values of the specimens tested in air and solution were: from 17.39 to 6.74% (P1.MT—Plate 1, Mid-Thickness); from 21.40 to 9.03 (P1.NS—Plate 1, Near-Surface); from 21.47 to 7.91 (P2.MT—Plate 2, mid-thickness); from 20.61 to 9.57 (P2.NS—Plate 2, near-surface).

Table 5. Information extracted from tests in deaerated 0.5 mol/L H₂SO₄ + 10 mg/L As₂O₃ solution at slow strain rate (3.35 × 10⁻⁶ s⁻¹).

Material	Condition	YS (MPa)	UTS (MPa)	ε (%)	I (ε)
Plate 1 (20.0 mm)	P1.MT—air	535.2	641.1	17.39	0.612
	P1.MT—solution	528.6	633.2	6.74
	P1.NS—air	534.1	637.8	21.40	0.578
	P1.NS—solution	529.2	640.8	9.03
Plate 2 (38.1 mm)	P2.MT—air	453.0	574.9	21.47	0.632
	P2.MT—solution	448.6	465.1	7.91
	P2.NS—air	497.8	557.0	20.61	0.536
	P2.NS—solution	479.6	563.4	9.57

summarizes the results obtained from the SSRT curves. YS is the yield strength, UTS is the ultimate tensile strength, ε is the elongation to failure, and I (ε) is the hydrogen embrittlement index. A lower embrittlement index (i.e., values closer to 0) indicates less susceptibility to hydrogen embrittlement.

The P2.MT condition presented the highest I (ε). On the other hand, the P2.NS specimens presented the lowest I (ε) and were less susceptible to hydrogen embrittlement. The embrittlement rates decrease in the following order: P2.MT, P1.MT, P1.NS, P2.NS.

Regarding the influence of materials resistance in hydrogen embrittlement results, the values of the Vickers hardness and yield strength of plate 1 are higher. Therefore, the greater susceptibility showed by plate 1 compared to the P2.NS condition corroborates the results found by Hardie et al. [36]. However, the presence of CSR in the mid-thickness of plate 2 played the most important role in the hydrogen embrittlement susceptibility. The most susceptible condition was P2.MT, attributed to segregation in the mid-thickness, which created localized brittle particles and high-hardness regions (up to 400 HV₀.₂), primarily due to carbon and manganese segregation [18,19,20,23]. Furthermore, the interaction between carbon and hydrogen in the CSR led to the formation of fissures containing high-pressure methane gas in the steel.

According to ANSI/NACE MR0175-2021/ISO 15156:2020 [37], a maximum hardness of 22 HRC (or 248 HV) is recommended for carbon and low-alloy steels used in sour service, as higher values indicate increased susceptibility to cracking. Although hardness tests alone cannot predict material behavior in sour environments [12], the literature indicates that increased susceptibility to hydrogen embrittlement correlates with higher microhardness values [4]. Consequently, the average microhardness value was greater than 248 HV only in the 20.0 mm thick plate, which may justify its slightly greater susceptibility compared to the P2.NS condition.

Due to sulfur segregation at the center of the thicker plate, a band of complex FeMnS sulfide inclusions can form both within and along ferrite grain boundaries. It has been reported that globular MnS inclusions may inhibit hot cracking by binding Mn with S. However, elongated FeMnS inclusions can have detrimental effects, as they serve as potential crack initiation sites in steels exposed to sour environments [38]. Consequently, a high concentration of dispersed FeMnS sulfide inclusions could increase the susceptibility to hydrogen embrittlement [19].

According to Stroe [34], due to the theory of localized plasticity, the steps for the occurrence of fracture involve the diffusion and concentration of hydrogen around the dislocations. Under dynamic plastic deformation, dislocations are formed and move with the aid of hydrogen. Heterogeneous plastic deformation zones are created by facilitating the movement of dislocations, with failure occurring in localized fragile areas.

Nevertheless, in the present study, analyses near the surface presented a close hydrogen embrittlement index between the thinner and the thicker plates. Such results closeness can be attributed to the decrease in the dislocation’s movement due to the pinning effect caused by the presence of carbonitrides in microalloyed steels, although the addition of elements such as Ti, V, and Nb is small in both plates [39].

3.4. Fracture Morphology

The ductility loss changed the fracture morphology of the tested specimens. The air-tested specimen presented a ductile zone in the center of the specimen because of the nucleation, growth, and void coalescence. Once the cracking propagation was initiated, with consequent triaxiality loss, the external part of the specimen fractured at 45° to the tensile axis, under plane stress (Figure 10a). On the other hand, the fracture morphology of specimens tested in a 0.5 mol L⁻¹ H₂SO₄ + 10 mg L⁻¹ As₂O₃ solution indicates a shear fracture.

Figure 10b shows the characteristic fracture morphology. It is showing that cracking initiated on the transverse surfaces reduced by the necking. A crack of small dimensions propagated from the surface by a brittle fracture mode, perpendicular to the applied stress until reaching a critical size. The cracking progressed from one surface to the other by shearing (at 45° to the tensile axis), characterized by a ductile dimple fracture surface, with regions of microvoid nucleation and coalescence, and small cleavage areas. The brittle area has quasi-cleavage planes and secondary cracks. All the specimens experienced complete separation.

Regarding plate 1, Figure 11a,b presents a typically ductile fracture morphology for samples tested in the air. Figure 11c, Figure 11d, Figure 11e, and Figure 11f present the fracture surface of plate 1 samples tested in solution, where: c and e are obtained from the brittle zone; d and f are obtained from the ductile zone, for the P1.MT and P1.NS conditions, respectively.

The nucleation of microvoids in air tests can occur by decohesion of the particle–matrix interface or second-phase particle fracture. So, different second-phase particle shapes and sizes lead to a range of nucleation processes. Also, stress rate, interfacial cohesion, particle volume fraction, and material matrix resistance influence the conditions required for nucleation. Additionally, the plastic deformation allows the microvoids nucleated first to expand their volume and, at some point, to coalesce with voids nucleated later by smaller particles [40].

However, Figure 11d shows that, for hydrogenated specimens, in the microvoid nucleation regions, the presence of hydrogen increased the amount of microvoids compared to the ductile region of the samples tested in the air, especially in the center of the specimen. This occurred due to hydrogen trapping, which reduces the tension needed for microvoid nucleation by lowering the interfacial cohesion between the matrix and the particles.

Figure 12 shows SEM images of the thicker plate samples (P2.MT and P2.NS). Panels (a) and (b) depict the ductile fracture morphology of samples tested in air. For samples tested in solution, panels (c) and (e) illustrate the brittle zone, while panels (d) and (f) illustrate the ductile zone for P2.MT and P2.NS, respectively.

Probably, the fracture process initiation occurred according to the hydrogen-induced decohesion (HID) model, in which the presence of hydrogen can significantly affect the reduction of cohesive resistance in grain boundaries, phase interfaces, or even along the specific crystallographic planes [41].

The intergranular fracture occurred in phase interfaces or grain boundaries, due to the presence of localized fragile particles, such as the martensite–austenite microconstituent [5,9], which are mostly associated with ferritic grain boundaries. Intergranular cracking is characterized by a low-energy fracture mode with a brittle appearance, resulting from propagation along grain boundaries due to the precipitation of brittle phases, such as the martensite–austenite microconstituent [42].

According to Troiano [39], when stress is applied to a specimen, there is a stress concentration, especially in regions with metallurgical defects. Although this stress state may not be sufficient to cause fracture, there is stress-induced diffusion of hydrogen to the stress field of the dislocation array. When the critical hydrogen concentration is reached in a localized region and the microstructure is susceptible, localized fracture occurs. The crack propagates until it is interrupted, and then the sequence of stress-induced diffusion and localized fracture must be repeated.

The pinning mechanism originates from the migration of hydrogen atoms in vacancy, when it is cut by the moving edge dislocation [39]. The cohesive strength of the lattice is reduced by the segregation of interstitials at the region near the tip of the dislocation array. Thus, the dislocation array becomes a more active fracture embryo [39].

Due to the low hydrogen solubility in X80 steels, the decohesion processes presupposes the occurrence of segregations, which can be provided by the interactions between hydrogen and dislocations. The slow strain rate test allowed hydrogen redistribution around dislocations during the test, reducing the elastic interaction between dislocations and significantly increasing their mobility, as predicted by the hydrogen-enhanced localized plasticity (HELP) model. As a result, microcracks nucleated (as shown in the fractographic analysis of specimens tested in hydrogen solution), acting as stress concentrators sites and reducing the critical shear stress. The reduction of critical shear stress is probably due to the presence of high-density dislocations, since they start to move closer to each other [41].

Fractographic features reveal hydrogen-assisted microfractures caused by a significant reduction in the shear stress required for dislocation movement during SSRT. This led to microvoid coalescence along preferential planes and failure due to brittle point fracture [9]. Therefore, combining different models proposed in the literature (HID and HELP) provides a comprehensive understanding of the events leading to material failure due to the embrittlement from critical amounts of hydrogen.

4. Conclusions

In order to better comprehend the degradation of mechanical properties in API 5L X80 pipeline steels during service, this study examined the influence of centerline segregation on hydrogen embrittlement in two plates with different thicknesses and chemical compositions. The results highlight the importance for the pipeline industry to account for the effects of centerline segregation and variations in plate characteristics on material performance, particularly about hydrogen embrittlement, to improve the reliability and safety of pipeline systems in service. The findings are summarized as follows.

• The thicker plate exhibited a centerline segregation region (CSR) with an average microhardness value approximately 25% higher than that of the overall plate (including both segregated and non-segregated areas). The CSR constituted approximately 1.05% of the cross-sectional area.
• Larger α_p grain sizes and fewer ferrite–FCA (ferrite–carbide aggregates) interfaces in the thicker plate facilitated greater diffusivity in the non-segregated area, as grain boundaries serve as high-energy trapping sites.
• The centerline segregation region presented an important role in the hydrogen embrittlement resistance of the thicker plate due to its high hardness microconstituents and higher percentages of carbon, manganese, sulfur, and phosphorus. Except for mid-thickness samples, there was a tendency for greater resistance to hydrogen embrittlement among the specimens obtained from the thicker plate, evidencing a determinant role of the hardness to hydrogen embrittlement susceptibility.
• The specimens tested in the air presented ductile morphology, with the classical dimples featured, while the samples tested in hydrogen solution presented fracture behavior with brittle and ductile surfaces. The brittle area had quasi-cleavage planes and secondary cracks.
• The fracture process initiation occurred primarily through mechanisms aligned with the hydrogen-induced decohesion (HID) model. The presence of hydrogen was found to significantly decrease the cohesive strength at grain boundaries, promoting crack initiation at sites where localized brittle particles are present. Additionally, hydrogen redistribution under applied stress further influenced the fracture process through the hydrogen-enhanced localized plasticity (HELP) model. By reducing the elastic interactions between dislocations, hydrogen lowered the critical shear stress required for dislocation movement. Once the critical hydrogen concentration was reached in specific susceptible areas, localized fractures occurred.