A review of hydrogen embrittlement in gas transmission pipeline steels

2.1 Hydrogen entry

Two commons sources of hydrogen absorbed into steels are (i) gaseous hydrogen in steel pressure vessels and (ii) hydrogen produced at the steel surface by a chemical or electrochemical reaction.

Hydrogen gas is used as a fuel in the aerospace industries, as a feedstock in oil refining and in producing chemicals such as ammonia and methanol, and more recently in steelmaking (Lynch 2011a). Steel vessels, casings and pipelines are used widely due to the cost efficiency of steel. In oil and gas wells, steel can be exposed to environments containing hydrogen sulphide (H₂S) which produces high amounts of hydrogen and can cause HE (Shadravan and Amani 2019). Gaseous hydrogen is expected to become more widespread due to its use in hydrogen fuel cells, in domestic heating and cooking, and as an energy vector in the hydrogen economy (Edwards et al. 2007).

Exposure to chemically produced hydrogen is often a consequence of some other chemical process. During steelmaking, hydrogen can form by the reaction of moisture with molten steel (Carter and Cornish 2001; Sezgin et al. 2019). A similar reaction can cause hydrogen entry into steel during welding if moisture is present in the air, in welding consumables, on the welding electrodes, or on the steel surfaces (Carter and Cornish 2001; Lee 2016). Hydrogen is produced electrochemically when steel is exposed to acidic solutions, such as those used in pickling, etching, cleaning, phosphating, and paint-stripping. Other electrochemical sources of hydrogen include cathodic protection, electroplating, and aqueous corrosion (Carter and Cornish 2001; Jewett et al. 1973; Lynch 2011a; Venezuela et al. 2016a).

Hydrogen entry into steel from gaseous hydrogen requires dissociation of the hydrogen molecule (H₂) into hydrogen atoms. The process consists of three steps: physisorption, chemisorption, and absorption (Barnoush 2009; Oriani 1978).

In physisorption, the hydrogen molecule is drawn to the metal surface by van der Waals forces. The second step, chemisorption, involves the cleavage of the hydrogen molecule into two hydrogen atoms, and results in covalent bonds between adsorbed hydrogen atoms and surface atoms (Barnoush 2009; Christmann 1981). Absorption occurs when the hydrogen atom enters the steel lattice (Barnoush 2009). In iron at room temperature, the rate-limiting step of gaseous hydrogen entry is likely the final absorption step (Oriani 1978). Absorbed hydrogen can escape from steel by the reverse of these three steps.

Electrochemical hydrogen entry occurs by the production of atomic hydrogen at the steel surface by the cathodic partial reaction known as the hydrogen evolution reaction (HER), which can be written as:

The adsorbed hydrogen atoms on the metal surface can either combine with other hydrogen atoms and form hydrogen molecules (which is called desorption or recombination) or be absorbed into the metal.

Some chemical species slow the rate of the recombination of adsorbed hydrogen and result in more hydrogen being absorbed into the steel. These species are called poisons, as they act as poisons for hydrogen recombination and promote hydrogen entry into the steel (Venezuela et al. 2016a). These species include compounds of arsenic, phosphorus, sulphur, antimony, selenium, tellurium and tin, and carbon compounds such as carbon sulphide (CS₂), carbon monoxide (CO), cyanide (CN⁻), thiocyanate (CNS⁻), urea (CON₂H₄), thiourea (CSN₂H₄), and hydrogen sulphide (H₂S) (Barnoush 2009; Berkowitz et al. 1976; Bockris et al. 2001; Venezuela et al. 2016a). Relatively small concentrations are effective in promoting hydrogen entry.

The influence of internal hydrogen depends solely on the hydrogen content in the steel. This means that for IHE, once hydrogen is contained in the steel, the effects are the same regardless of whether the hydrogen was from gaseous hydrogen or from an electrochemical reaction (Atrens et al. 1980; Lee 2016; Liu et al. 2014). HE studies are therefore often conducted using electrochemical hydrogen charging, and the electrochemical hydrogen charging condition can be equated to an equivalent gaseous hydrogen pressure. This topic is discussed more in Section 4.2.

2.2 Hydrogen diffusion

Hydrogen atoms inside the steel randomly jump between interstitial sites in the lattice. This random movement leads to hydrogen diffusion through the crystal lattice according to the hydrogen concentration gradient. The hydrogen flux, J, can be evaluated using Fick’s first law of diffusion (Crank 1967; Marchi and Somerday 2008):

where D is the diffusivity (also known as the diffusion coefficient) and dc/dx is the hydrogen concentration gradient. The dimensions of J are amount of substance per unit area per unit time, so the dimensions of D are area per time. For a constant concentration gradient, a larger diffusivity results in greater permeation of hydrogen.

As an example, consider a thin sheet of thickness L and infinite width. The concentration C at any distance x through the sheet can be evaluated from (Crank 1967):

where C0 and C2 are the concentrations at the entry and exit surfaces respectively, C1 is the initial uniform concentration throughout the sheet, and t is the time. Equation (4) gives the concentration through the steel sheet specimens used in permeation experiments, as discussed in Section 4.2. In the case of the first polarisation in a permeation experiment, the initial concentration (C1) is 0. The exit surface concentration (C2) is 0 because hydrogen is quickly oxidised upon exiting. Therefore, Equation (4) reduces to:

If the subsurface concentration C0 is known, the concentration C at a distance x through the membrane can be evaluated using Equation (5). Equation (5) indicates that the hydrogen concentration is dependent on Dt/L2. This means that, as the thickness of the sheet increases, the time taken to reach a given concentration increases quadratically. The fraction Dt/L2 is sometimes called the dimensionless time parameter (Crank 1967).

Figure 1 shows the normalised hydrogen concentration evaluated using Equation (5) for different values of the dimensionless time parameter. Once the dimensionless time parameter reaches 0.4, the concentration is close to the steady-state linear distribution.

Figure 1:

Diffusive normalised hydrogen concentration across a sheet of infinite width, with hydrogen entering at one surface (x=L) and exiting at the other surface (x=0). L is the thickness. The normalised hydrogen concentration is represented by a ratio of the concentration at a particular position divided by the concentration just below the specimen surface (C0), calculated using Equation (5). Each line represents the normalised hydrogen concentration for a particular value of the dimensionless time parameter (Dt/L2). This figure was plotted using MATLAB R2018b.

A second example is a cylindrical tensile specimen with an initial uniform concentration C1 (which, for an uncharged specimen, is close to 0) being charged with hydrogen at a constant hydrogen fugacity so that the surface has a constant concentration C0. The concentration throughout the cylinder approaches a uniform concentration of C0 over time. The concentration C at time t and radial distance from the centre r can be calculated using (Crank 1967):

where a is the radius of the cylinder, D is the diffusivity, and J0 and J1 are the zeroth and first order Bessel functions, which can be calculated using the MATLAB function ‘besselj’. The values for αn are calculated from the roots of the zeroth order Bessel function using Equation (7), which can also be found using MATLAB. As was the case for Equation (4), the concentration distribution can be related to the dimensionless time parameter, which in this case is Dt/a2.

Figure 2 shows the normalised hydrogen concentration (C/C0) calculated using Equation (6) for different values of the dimensionless time parameter. As Dt/a2 increases, the concentration ratio approaches an asymptote of a uniform value of 1.0 throughout the specimen. Once Dt/a2=0.8, the concentration is close to uniform throughout, with C/C0 never lower than 0.98. If the diffusivity of a steel is known, this can be used to approximate the charging time it takes for a uniform concentration to be reached.

Figure 2:

Diffusive normalised hydrogen concentration across a cross-section of a cylindrical specimen, which initially contains no hydrogen. The specimen is charged with a constant hydrogen fugacity resulting in a constant surface hydrogen concentration (C0). The x-axis is the position (r) divided by the radius (a). The normalised hydrogen concentration was calculated using Equation (6). Each line represents the normalised hydrogen concentration for a particular value of the dimensionless time parameter (Dt/a2). The figure was plotted using MATLAB R2018b.

Hydrogen diffusivity varies with material, crystal structure, temperature, and stress (Lynch 2011a). Diffusivity increases with increasing temperature, as hydrogen atoms have greater kinetic energy and can more easily jump between interstitial sites. For a given material, the microstructure can also lead to variation in the diffusivity, especially in materials with complex microstructures. Ferritic steels, for example, have diffusivities that span 3 to 4 orders of magnitude at room temperature (Lynch 2011a). This is attributed to differences in hydrogen trap types and densities, which are discussed in greater detail in the following section on hydrogen trapping.

2.3 Hydrogen trapping

Hydrogen atoms are slightly larger than the interstitial sites between iron atoms that they occupy. This strains the lattice surrounding a hydrogen atom. Therefore, hydrogen tends to move to imperfections in the metal where the crystal lattice is slightly expanded (Birnbaum 2003; Louthan 2008). Such imperfections include (i) solute atoms, (ii) vacancies, (iii) boundaries such as grain boundaries and phase boundaries, (iv) dislocations, (v) interstitial sites in the first few atomic layers below free surfaces, (vi) interfaces between precipitates or inclusions and the matrix, (viii) crack tips, and (ix) voids. Hydrogen has a lower-potential energy at these sites, and as a result, these sites are known as traps (Carter and Cornish 2001; Lynch 2011a; Venezuela et al. 2016a). The binding energies of various traps in iron and steels have been tabulated by Ohaeri et al. (2021), Bhadeshia (2016), and Szost et al. (2013).

The average residence time of hydrogen at a trap increases with increasing binding energy. Increasing temperature increases the kinetic energy of the hydrogen atom, increasing the possibility of the hydrogen atom jumping out of the trap. Traps are referred to as irreversible if the binding energy exceeds a threshold binding energy. In irreversible traps, hydrogen atoms remain essentially completely trapped at room temperature.

The diffusivity of hydrogen at room temperature decreases with increasing number of traps and increasing trap strength (Lynch 2011a). Therefore, traps can reduce the degree of HEE by preventing hydrogen accumulation at potential crack initiation sites (Bhadeshia 2016). Some hydrogen traps, however, can also exacerbate HE by (i) accumulating hydrogen at regions of concentrated stress, (ii) by allowing hydrogen to recombine at large voids which causes pressure damage, or (iii) by acting as sources of internal hydrogen after diffusible hydrogen has egressed (Bhadeshia 2016; Lynch 2011a; Venezuela et al. 2016a). For these reasons, the interactions between hydrogen and trap sites often determine the mechanisms and manifestations of HE (Barnoush 2009; Barrera et al. 2018). The effective diffusivity (Deff) of hydrogen accounts for hydrogen traps and stress fields that can influence diffusion and is smaller than the lattice diffusivity.

2.4 Hydrogen solubility

The hydrogen concentration inside the steel in equilibrium depends on how much hydrogen is available at the surface. For steel exposed to pure gaseous hydrogen, the equilibrium hydrogen concentration in the steel depends on the external hydrogen partial pressure (Barthélémy 2006). If hydrogen trapping is ignored (for example, in pure annealed iron), the hydrogen concentration is related to hydrogen pressure by Sieverts’ law such that the hydrogen concentration, CH, is given by (Helmi 2016; Marchi and Somerday 2008; Oriani 1978):

where S is the solubility constant, which depends on the material and temperature, and fH is the fugacity of the hydrogen gas, which has the same units as pressure. For a real gas, the fugacity is the pressure that the gas would exert if the gas were an ideal gas. The hydrogen fugacity is essentially the same as the partial pressure up to a hydrogen pressures of 100 bar, above which the hydrogen fugacity is greater than the hydrogen partial pressure (Atkins et al. 2018; Oriani 1978). The fugacity can be estimated using tabulated fugacity coefficients such as those found in Holley et al. (1958) or from charts such as Figure 3.

Figure 3:

Fugacity/pressure plotted against hydrogen pressure. Reproduced from Venezuela et al. (2018) with permission.

For steels, hydrogen trapping needs to be incorporated into Sieverts’ law (Equation (8)) (Alraeesi and Gardner 2021; Lee 2016; Venezuela et al. 2018) by adding a trapped hydrogen concentration CT, so that the equilibrium hydrogen concentration in the steel is given by:

This modified version of Sieverts’ law can be derived using statistical mechanics, as shown by Venezuela et al. (2018). Figure 4 shows the equilibrium hydrogen concentration for gaseous hydrogen charging of a medium-strength steel (3.5NiCrMoV), two advanced high-strength steels (980DP and MS1500) compared with literature data for pure annealed Fe (designated as Fe). The vertical intercepts in Figure 4 represent CT, showing that at a pressure of 2 bar or above, there is a substantial hydrogen concentration in the hydrogen traps. The hydrogen solubility data for the steels indicates that the hydrogen solubility in steels is significantly greater than in pure iron, and much of this hydrogen is trapped hydrogen.

Figure 4:

The equilibrium hydrogen concentration for gaseous hydrogen charging of a medium-strength steel (3.5NiCrMoV) (Venezuela et al. 2018), two advanced high-strength steels (980DP and MS1500) (Liu et al. 2018; Venezuela et al. 2017) compared with literature data for pure annealed Fe (designated as Fe) (Kiuchi and McLellan 1983). The equilibrium hydrogen concentration was measured using thermal desorption spectroscopy (TDS) and includes hydrogen in traps. Reproduced from Atrens et al. (2022) with permission.

3.1 Manifestations of HE

Typical manifestations of HE include a decrease in ductility, a decrease in fracture toughness, subcritical crack growth and altered fracture surfaces. Commonly, hydrogen causes a change from ductile to brittle fracture.

Subcritical crack growth is the propagation of a crack at an applied stress intensity factor below the critical stress intensity factor for the steel (the fracture toughness). Cracks can initiate at hydrogen traps at defects in the steel lattice or surface defects such as scratches (Lee 2016; Louthan 2008; Serebrinsky et al. 2004). The combination of a stress concentrator and hydrogen can lead to localised fracture, extending the crack tip. Hydrogen can then migrate to the new crack tip, and the process can repeat until the crack grows to a critical length whereupon brittle fracture occurs suddenly (Atrens et al. 2020; Louthan 2008; Serebrinsky et al. 2004). Subcritical cracking in the presence of hydrogen can be characterised using the threshold stress σH (or threshold stress intensity factor KIH) which is the critical stress (or critical stress intensity factor) above which subcritical crack growth begins (Atrens et al. 2020). The threshold stress and threshold stress intensity factor may be significantly smaller than the yield stress and fracture toughness.

Subcritical crack growth can lead to hydrogen-induced delayed fracture, which occurs when an applied stress (typically below the yield strength) leads to sudden fracture (Bhadeshia 2016; Jewett et al. 1973; Louthan 2008). The degree of HE increases with decreasing strain rate. Delayed fracture is the extreme case at a strain rate of 0, with a constant applied stress (Louthan 2008; Serebrinsky et al. 2004).

In low strength steels, HE appears not to result in subcritical crack growth (Atrens et al. 2020). Hydrogen can, however, decrease mechanical properties, particularly ductility, fracture properties and strength. HE is often characterised using the decrease in reduction-in-area (RA) and elongation (Jewett et al. 1973; Lee 2016; Oriani 1978). In low strength steels the yield strength can be reduced by up to 30% for moderate hydrogen contents (Atrens et al. 2020; Hirth 1980). Hydrogen generally does not affect the modulus of elasticity (Jewett et al. 1973; Lee 2016).

Hydrogen introduces new cracking mechanisms and decreases the fracture toughness of steels. Hydrogen decreases the strength of notched tensile steel specimens (Jewett et al. 1973; Lee 2016). Hydrogen reduces the stress intensity factor required for crack growth (i.e. subcritical crack growth) and larger amounts of hydrogen may also increase the subcritical crack velocity (Barnoush 2009; Louthan 2008). Hydrogen also increases susceptibility to fatigue and creep failure (Jewett et al. 1973; Lee 2016; Louthan 2008). The influence of hydrogen on fatigue depends on the frequency, as a lower frequency results in a lower strain rate which increases susceptibility to HE (Louthan 2008).

HE often changes the appearance of fracture surfaces. However, there is no single fracture mode common to all hydrogen embrittled materials as the fracture varies according to the steel and the environment (Louthan 2008; Oriani 1978). Intergranular fracture – fracture along grain boundaries – is common in hydrogen embrittled high-strength steels (Louthan 2008; Lynch 2011a; Serebrinsky et al. 2004). Intergranular fracture surfaces are bright and faceted, and can be covered in microscopic dimples. For some martensitic steels embrittled by hydrogen, the fracture is transgranular, as in the absence of hydrogen, but the dimples on the fracture surface are smaller and less deep (Lynch 2011a). Iron and low-carbon steels typically undergo ductile fracture without hydrogen, and undergo brittle quasi-cleavage in the presence of hydrogen. Quasi-cleavage is a kind of transgranular fracture, similar in appearance to cleavage but not along any known cleavage plane (Lynch 2011a; Xu and Rana 2008).

Other fractographical features in hydrogen embrittled steel include: (i) many surface cracks on the external surface perpendicular to and near the fracture; (ii) fish-eyes, especially in lower-strength steels, which are small, bright, circular regions of cleavage-like fracture that form during tensile failure, centred around hydrogen traps such as inclusions; and (iii) flakes or shatter cracks, which appear as small shiny regions on the fracture surface and form during solidification (Jewett et al. 1973; Louthan 2008; Lynch 2011a).

3.2 Steel attributes

In general, stronger steels are more susceptible to HE. High-strength martensitic steels are highly susceptible to HE, while lower-strength ferritic steels are less so (Lee 2016; Lynch 2011a). However, many attributes other than strength can influence HE susceptibility, such as microstructure, residual stresses, impurities, and hydrogen traps.

HE in martensitic steels (especially IHE) has been well studied due to (i) the embrittling influence of even small hydrogen concentrations and (ii) the potential for catastrophic delayed failure. In contrast, lower-strength ferritic steels require higher hydrogen concentrations for significant HE. Martensitic steels can undergo significant HE at hydrogen concentrations of ∼0.5 wt ppm, whereas ferritic steels may require hydrogen concentrations of ∼10 wt ppm (Lynch 2011a).

Low strength, stable austenitic steels have low susceptibility to HE (Barthélémy 2006; Lee 2016; Lynch 2011a; Michler et al. 2012). In addition, the austenitic FCC crystal structure has a much lower hydrogen diffusivity and higher hydrogen solubility than ferritic BCC and martensitic BCT structures (Lynch 2011a; Venezuela et al. 2016a). Many of the alloying elements that stabilise austenitic steel, such as nickel, carbon and manganese, have been reported to reduce HE (Barthélémy 2006).

Impurities such as sulphur, phosphorus and oxygen can exacerbate HE (Barthélémy 2006). These impurities can segregate at grain boundaries, weakening them, and large inclusions such as sulphides and oxides can trap enough hydrogen to act as crack initiation or void nucleation sites (Barthélémy 2006; Lynch 2011a). There is also some evidence that the presence of some impurities is necessary for hydrogen to alter the fracture mode by forming localised microvoids (Oriani 1978). The presence of strong hydrogen traps that do not serve as crack initiation sites can reduce HE susceptibility. This can be achieved by the presence of solute atoms, such as chromium, manganese, molybdenum, vanadium, and rare earth metals (Barthélémy 2006; Bhadeshia 2016). Titanium solute atoms trap hydrogen by forming carbides. Aluminium in martensitic steels also reduces HE (Bhadeshia 2016), which may be due to surface aluminium oxides hindering hydrogen entry.

Resistance to HE in steels can also be improved by subsequent processes that affect the microstructure. Heat treatments such as tempering, and normalising and quenching have been found to improve hydrogen resistance (Barthélémy 2006; Bhadeshia 2016). Cold working can create dislocations and internal cracks that may impede hydrogen diffusion (Bhadeshia 2016; Hirth 1980). In addition, component design features have a large influence on susceptibility. For example, stress concentrators can act as crack initiation sites in HEE (Barthélémy 2006).

3.3 Environmental factors

Environmental factors influencing HE include (Barthélémy 2006; Lynch 2011a; Venezuela et al. 2016a):

1. Hydrogen content

2. Stress, both applied and residual

3. Strain rate

4. Temperature

5. Surface condition of the steel

6. Presence of chemicals that facilitate or inhibit hydrogen uptake at the external surface.

3.3.1 Hydrogen concentration

The degree of hydrogen embrittlement increases with increasing hydrogen concentration. Figure 5 shows the notched tensile strength of a typical hot stamping steel against hydrogen concentration. The decrease in the notched tensile strength was accompanied by the change of fracture mode from microvoid coalescence for no hydrogen or for a low hydrogen concentration) to quasi-cleavage (at 2.6 wt ppm) to quasi-cleavage plus intergranular fracture for higher hydrogen concentrations (Lovicu et al. 2012).

Figure 5:

Notched tensile strength for a typical hot stamping steel decreased with increasing hydrogen concentration. Reproduced from Lovicu et al. (2012) with permission.

Even though HE generally increases with increasing hydrogen concentration, at high concentrations there exists a saturation point beyond which the effects of HE are relatively constant (Barthélémy 2006; Lynch 2011a). There also exists a similar minimum HE concentration, below which there are few HE effects (Bhadeshia 2016; Louthan 2008; Venezuela et al. 2016a).

3.3.4 Temperature

The influence of temperature on HE is associated with two important effects. First, hydrogen mobility increases with temperature, so that there is little hydrogen mobility at sufficiently low temperatures. Second, hydrogen is more likely to accumulate in traps at lower temperatures (Louthan 2008). In any case, there exists an intermediate temperature range over which HE may occur, as shown in Figure 6 for tensile tests carried out by Xing et al. (2021). The maximum HE in steels usually occurs near room temperature (Barthélémy 2006; Lee 2016; Louthan 2008). Above 100 °C HE is often negligible (Barthélémy 2006).

Figure 6:

Influence of temperature on HE measured using tensile tests of X90 pipeline steel as measured by reductions in elongation (δ) and RA (Ψ). The hydrogen was charged electrochemically. Reproduced from Xing et al. (2021) with permission.

3.4 Mechanisms of HE

Numerous mechanisms have been proposed for the HE of steels. The most popular include hydrogen enhanced local plasticity (HELP), hydrogen enhanced decohesion (HEDE), adsorption-induced dislocation emission (AIDE), hydrogen-enhanced strain-induced vacancy (HESIV), and the defactant concept.

It has also been proposed that in most cases, a combination of mechanisms occurs (Djukic et al. 2019; Lynch 2011a), and that the relative contributions of the various mechanisms vary according to material and environmental factors. HE mechanisms have been the subject of several recent in-depth reviews (Barrera et al. 2018; Birnbaum 2003; Djukic et al. 2019; Lynch 2011a; Robertson et al. 2015).

4.1 Tensile testing

Tensile testing is used to assess (i) how hydrogen influences strength and ductility and (ii) the HE mechanism from an evaluation of the fractography. Tensile testing to study HE is usually carried out at a slow strain rate because HE is exacerbated at slow strain rates. Tensile testing allows the evaluation of the influence of hydrogen on strength, ductility, and fracture mode. Tensile testing can also be used to detect subcritical crack growth by applying a constant stress below the yield stress for an extended period.

The ductility can be measured using the elongation (ε) or reduction in area (RA) at fracture (Briottet et al. 2012; Lee 2016). HE can be quantified using the hydrogen embrittlement index (I), calculated from:

where the subscript indicates tests in air or in the hydrogen environment. I closer to 100% indicates high HE, while I closer to 0% indicates low HE. Other hydrogen susceptibility indices can be calculated for reductions in other mechanical properties such as the ultimate tensile strength (UTS), and fracture toughness (Briottet et al. 2012; Venezuela et al. 2016b, 2020, 2021). The hydrogen susceptibility indices IUTS and IFM are defined as:

The influence of strain rate on SCC and HE led to the development of the slow strain rate test (SSRT), standardised in ASTM G129 (ASTM International 2013). The SSRT is also designated the constant extension rate test (CERT). The extension of the specimen is typically increased linearly, so the (engineering) strain rate is kept constant in the initial linearly elastic region of the tensile test. The strain rate is usually between 10⁻⁵ s⁻¹ and 10⁻⁷ s⁻¹ (Henthorne 2016; Martínez-Pañeda et al. 2020). Throughout the test, the specimen is exposed to the embrittling environment. Various specimen designs have been used: (i) rectangular cross-section specimens; (ii) circular cross-section specimens; (iii) notched specimens to investigate the influence of stress-concentration; and (iv) hollow specimens to maintain a corrosive environment in a small volume inside the specimen.

An alternative to the SSRT is the linearly increasing stress test (LIST) in which the load is increased at a constant rate. The LIST and SSRT produce equivalent results in the initial elastic part of the tensile test up to the threshold stress (Atrens et al. 1993; Liu and Atrens 2013). The LIST has the advantage that it is concluded when the stress reaches the UTS, so it is typically much shorter than the SSRT. Like the SSRT, the LIST is used to expose specimens to a corrosive environment to study HE and SCC. Typically, LISTs are conducted along with the potential drop technique. The potential drop technique allows measurement of the threshold stress for the initiation of subcritical crack growth (or the initiation of stress corrosion cracking), or the yield stress if the HE begins at a stress greater than the yield stress (Ramamurthy and Atrens 1993; 2010; Ramamurthy and Atrens 2013; Ramamurthy et al. 2011).

4.2 Permeation testing

Permeation testing is used to determine the subsurface hydrogen concentration and the hydrogen diffusion coefficient, which determines the speed of hydrogen diffusion within the steel. Hydrogen permeation experiments typically use the Devanathan-Stachurski cell (Devanathan and Stachurski 1962, 1964). A schematic of the Devanathan-Stachurski cell is shown in Figure 7.

Figure 7:

Design for the Devanathan-Stachurski cell used to measure hydrogen permeation electrochemically. The cell contains two three-electrode cells, each connected to a potentiostat. CE is the counter electrode, RE is the reference electrode, and WE is the working electrode. Nitrogen is bubbled through the solution to purge oxygen to low concentrations. Other inert gases can be used for this purpose, such as argon.

The Devanathan-Stachurski cell is made up of two three-electrode electrochemical cells, joined together by the thin steel specimen which is the working electrode for both cells. At the specimen surface in the cathodic cell (typically the left-hand cell), a constant potential is often applied which electrochemically produces a constant hydrogen fugacity so that hydrogen diffuses into the specimen. At the specimen surface in the anodic cell (typically the right-hand cell), a constant potential is applied, which oxidises the hydrogen atoms exiting the specimen. The diffusive hydrogen flux (J) can be calculated from the measured current (I) at the anodic surface (McBreen et al. 1966):

where A is the exposed area of the specimen, z is the number of electrons transferred in the oxidation reaction, and F is the faraday constant. I/A gives the current density in units of A m−2. This current density is usually denoted by i. The units of flux are therefore mol m−2s−1. The hydrogen concentration at the entry side of the specimen (referred to as the subsurface concentration C0) can be calculated from the flux using Fick’s first law (Equation (3)), which can be related to the fugacity using Sieverts’ law (Equation (9)).

The analysis assumes that the subsurface concentration at the cathodic (entry) side is a constant (C0) while the concentration at the anodic (exit) side is 0 (Barrera et al. 2018; Devanathan and Stachurski 1962). With increasing time, the hydrogen flux forms a sigmoid curve that approaches the steady-state flux, and the concentration throughout the specimen increases until it approaches a linear distribution from C0 to 0 (see Figure 1).

The advantage of the Devanathan-Stachurski technique over other methods is that the accuracy depends on the accuracy of the measurement of the current, which can be more accurate than other measurement devices such as pressure transducers and mass flow meters used in gaseous permeation (Barrera et al. 2018; Devanathan and Stachurski 1962). This allows the measurement of small amounts of permeating hydrogen. The Devanathan-Stachurski cell can be inaccurate if hydrogen does not immediately exit the exit side of the specimen, causing the exit concentration to be greater than 0. Thus, the exit side is often electrochemically plated with a thin layer of palladium to facilitate hydrogen egress and oxidation (Oriani 1978). Other issues with the method can arise if the specimen surfaces are corroded due to the presence of a corrosive electrolyte.

Multiple methods have been used to evaluate the diffusivity from a permeation test (Haq et al. 2013). The breakthrough time (tb) is the time it takes for the first hydrogen to reach the exit side of the specimen. The diffusivity is evaluated using:

The time lag method evaluates the diffusivity from the time (tL) it takes for the permeation rate to reach 0.6399 of the steady state value (Crank 1967; Devanathan and Stachurski 1962; Haq et al. 2013):

A better approach is to use the whole permeation transient to evaluate the diffusivity. This method uses the Laplace transformation of Fick’s first law. The method evaluates the diffusivity which gives the best fit of the measured permeation transient to the theoretical permeation transients given by (Fallahmohammadi et al. 2013; Liu et al. 2014; Zakroczymski 2006):

where i∞ is the steady-state current density, i0 is the starting current density, and i is the current density at time t. In permeation experiments, the charging side of the specimen is often initially polarised at a constant potential for a period of time to allow the reduction of oxides present at the specimen surface, after which a reproducible series of rise and decay transients can be conducted (Venezuela et al. 2017).

Often in permeation experiments the area of the specimen available for hydrogen permeation is limited by the mounting of the specimen in the permeation cell. This permeation area is typically defined by O-rings that provide the seal that contains the electrolyte in each permeation cell. Such an arrangement means that the exposed area of the specimen is smaller than the total area of the specimen. Some hydrogen can diffuse into the outside regions of the specimen, a phenomenon known as an ‘edge effect’. Edge effects introduce some errors into the permeation evaluations. For a circular specimen, a thicker specimen (larger L) with a smaller exposed radius (smaller r) results in a greater error due to edge effects. Crank (1967) gives a formula for estimating the error incurred by edge effects. A specimen with a thickness to radius ratio (L/r) of 0.2 results in about 6% error, and Crank recommends this as the maximum safely ignorable edge effect error.

Permeation experiments can also be conducted using a specimen exposed to gaseous hydrogen. These experiments are less common than electrochemical permeation experiments due to the difficulty in accurately measuring the small amounts of hydrogen exiting the specimen to determine the hydrogen flux. Gaseous hydrogen egress has been measured using a mass flow meter (Pizzi et al. 2008), a mass spectrometer (Frauenfelder 1968), an ion pump (Tsong-Pyng and Altstetter 1986), or a pressure transducer (Robertson and Thompson 1980). An example of a gas phase permeability cell that measures hydrogen flux using a pressure transducer is given in Figure 8.

Figure 8:

Diagram of a cell for gas phase hydrogen permeation. The steel specimen is identified by the grey colour. Hydrogen gas enters the cell at the top and diffuses to below the specimen. The pressure increase in the volume below the specimen is used to measure the amount of hydrogen diffusing through. Reproduced from Atrens et al. (2022) with permission.

The hydrogen permeation equations derived for electrochemical permeation experiments also apply to gas-phase permeation experiments, with the current densities replaced by fluxes in Equations (17) and (18), as indicated by:

Gas-phase permeation experiments give information that complements that from electrochemical permeation experiments. Gas-phase experiments allow the relationship between hydrogen partial pressure and subsurface hydrogen concentration to be evaluated. This allows the results of other HE studies that use electrochemical charging, such as tensile and fracture toughness studies, to be related to hydrogen pressure. Gas-phase experiments allow testing at elevated temperatures, whereas the temperature range is limited to below 100 °C in electrochemical experiments which use aqueous solutions. Gas-phase experiments also allow experimentation using gas mixtures containing hydrogen, such as mixtures of gases likely to be encountered in service, or the use of adsorption inhibitors such as oxygen.

If the same hydrogen concentration is introduced into a steel by both gas-phase and electrochemical charging, and the temperature is the same in both experiments, then the hydrogen fugacity during electrochemical charging and gas phase charging are the same (Liu et al. 2014; Venezuela et al. 2018). This allows evaluation of the hydrogen fugacity during electrochemical charging as indicated in Figure 9 (Liu et al. 2018; Venezuela et al. 2017).

Figure 9:

Relationship between hydrogen fugacity and overpotential during electrochemical hydrogen charging. The right-hand scale provides the equivalent hydrogen pressure. The fugacity of a gas corresponds to the pressure that the gas would have if the gas behaved as an ideal gas. Reproduced from Liu et al. (2018) with permission.

4.3 Fracture and fatigue

The impact of hydrogen on fatigue and fracture can be evaluated by comparing the results of experiments in air to those in hydrogen. Fatigue properties include the fatigue limit, the fatigue life, and the fatigue crack growth rate. Fracture mechanics experiments can measure crack velocity and the threshold stress intensity factor (KIH) (Liu and Atrens 2013). KIH is the stress intensity factor above which there occurs either subcritical crack growth or rapid fracture if there is no subcritical crack growth, in which case KIH is equivalent to the fracture toughness K.

4.4 Measuring hydrogen content

The hydrogen content can be measured using permeation experiments, hot extraction methods, and using Thermal Desorption Spectroscopy (TDS). Hot extraction methods typically involve measuring the amount of hydrogen released from a specimen as the specimen is heated (sometimes quickly melted) using methods such as thermal conductivity or mass spectrometry. Thermal desorption spectroscopy (TDS) is used to measure the amount of hydrogen in a specimen while recording the temperatures at which hydrogen is released from the specimen, which typically produces a series of peaks at different temperatures. Each peak typically corresponds to a particular hydrogen trap, and TDS therefore allows the characterisation of the trap binding energies (Atrens et al. 2018; Tapia-Bastidas 2016). Other methods used to measure the amount of diffusible hydrogen include measurement of the volume of hydrogen released when the specimen is submerged in a liquid such as glycerol (Wang 2009), liquid paraffin (Dong et al. 2010) or silicone oil (Wang et al. 2015a).

4.5 Microanalysis

The location of hydrogen within steel is of interest in the study of HE. Direct imaging of hydrogen within iron and steel is difficult due to the low solubility of hydrogen and the weak interactions between hydrogen and electrons or X-rays. Imaging using neutron scattering, transmission electron microscopy (TEM), and helium bombardment can identify high concentrations of hydrogen, but not individual hydrogen atoms (Barrera et al. 2018). Individual deuterium atoms and their locations have been imaged using atom probes.

The hydrogen microprinting technique (HMT) is used to indirectly identify the location of hydrogen at a specimen surface. The exit surface of a (permeation) specimen is coated with a salt such as silver bromide (AgBr). Hydrogen exiting the specimen reduces the silver ions to silver crystals. The unreacted salt can then be washed away and the silver crystals can be visualised. HMT provides insight into the distribution of hydrogen and hydrogen trapping, however, it is limited to surface features and cannot be used to image hydrogen within the steel (Barrera et al. 2018; Ovejero-García 1985).

4.6 Stress corrosion cracking

Stress corrosion cracking (SCC) manifests as subcritical cracking under an applied stress in a corrosive solution. Well known SCC conditions include (i) steels in hydroxide solutions (caustic cracking), (ii) brass in ammonium solutions (season cracking), (iii) high pH ground water on the outside of gas-transmission pipelines, and (iv) high-strength steels in water, particularly sea water. SCC likely encompasses many mechanisms including the metal dissolving at the crack tip (frequently called anodic dissolution), or adsorbed ions at the crack tip weakening metallic bonds (Lee 2016; Lynch 2011b; Ramamurthy and Atrens 2013). HE is one of the proposed mechanisms of SCC and is considered the likely SCC mechanism for high-strength steels in water. SCC can be studied using some of the same experimental techniques as HE, such as slow tensile tests.

5.1 Pipeline steel grades

The most common steel grades used in (Australian) natural gas transmission pipelines are those standardised in the American Petroleum Institute (API) 5L standard, ‘Specification for Line Pipe’ (American Petroleum Institute 2018; Ohaeri et al. 2021). These include the low-strength grades A25, A, and B, as well as the X series which ranges from X42 to X120. The numbers in the X series steels designate the minimum yield strength in the non-SI unit of kilo-pounds per square inch (ksi).

API 5L grades are low-carbon steels (in part to ensure good weldability). The maximum carbon content varies across the different grades from 0.21 to 0.28 wt %. The API 5L standard also specifies maximum percentages for alloying elements, as well as a maximum ‘carbon equivalent’ that estimates the contributions of alloying elements to the hardness of the steel and to the weldability of the steel. The heat treatments described in the API 5L standard are tempering, normalising (or annealing), stress relieving, quenching and tempering, and age hardening (American Petroleum Institute 2018). The microstructure of pipeline steels is mostly ferrite and pearlite, with some steels containing bainite and martensite (Ohaeri et al. 2021; Xu 2012). Higher strength pipeline steels (X70 and above) are more recent grades, developed using processing techniques including continuous casting, controlled thermo-mechanical rolling, micro-alloying, and quenching (Ohaeri et al. 2021), although these processes are also used for the production of the lower strength steel grades.

Pipelines may be extruded as pipe or fabricated from rolled metal plate joined longitudinally using a seam weld. Seam welds may be produced with a filler metal, as is the case for submerged-arc welding (SAW) and gas metal-arc welding (GMAW), or without a filler metal, as is the case for continuous welding, electric welding and laser welding. Pipes with seam welds may have one longitudinal weld, two longitudinal welds, or a helical seam weld. Welds which join two pieces of pipe circumferentially are known as girth or jointer welds (American Petroleum Institute 2018).

During welding, the microstructure of the steel on either side of the weld may be altered. This area is known as the heat-affected zone (HAZ). The weld may have higher hardness than the base metal, with hardness through the HAZ decreasing with distance from the weld (Olabi and Hashmi 1996). The weld metal frequently contains bainite and different morphologies of ferrite, including acicular ferrite, which resembles needle-shapes and has high toughness (Boumerzoug et al. 2010; Chaveriat et al. 1987; Olabi and Hashmi 1996). The welding method influences the microstructure of the weld and HAZ.

The longest natural gas pipeline in Australia is the 1539 km Dampier Bunbury Natural Gas Pipeline (DBNGP) in WA. Much of the main pipe length is X65 steel (Dampier Bunbury Pipeline (DBP) 2011). Other major Australian pipelines include (i) the Amadeus Gas Pipeline in NT, constructed mostly from X60 steel (APT Pipelines 2015), (ii) the Goldsfields Gas Pipeline in WA, constructed mostly from X70 steel (Goldfields Gas Transmission Joint Venture 1994), and (iii) the South East Australia Gas (SEAGas) pipeline in SA, constructed mostly from X70 steel (Langley et al. 2010). Over time, the construction of gas pipelines in Australia has tended from lower strength to higher strength pipeline steels (Langley et al. 2010), although this trend may change as hydrogen transmission becomes a consideration.

In order to establish the safe limits of hydrogen blending into existing natural gas pipelines, the resistance of these pipeline steels to HE must be evaluated, including the relationship between susceptibility and steel grade, microstructure, and the susceptibilities of the weld and HAZ.

5.2 Summary of HE research

A literature review using the Web of Science found 256 articles relevant to HE in pipeline steels. The number of articles is shown in Figure 10, with articles grouped by topic. The largest share of research articles has focused on the effects of hydrogen on mechanical properties of the pipeline steels. These mechanical properties included tensile properties, fracture and fatigue. ‘Hydrogen charging’ studies typically involved hydrogen induced cracking (HIC) or permeation experiments. Articles listed as ‘SCC’ focused on stress-corrosion cracking but involved hydrogen charging, typically using simulated soil solutions. Articles listed as ‘Other’ studied topics that were tangential to HE such as sulphate-reducing microorganisms and methods of taking field measurements of hydrogen content in pipeline walls. Table 1 lists the specific types of experiments that have been conducted on pipeline steels charged with hydrogen.

Figure 10:

Results of a Web of Science search for hydrogen in pipeline steels, showing the number of journal articles published each year from 1995 to 2022. Articles are group according to their focus. SCC articles expose steel to a corrosive environment where hydrogen is an electrochemical by-product and not the primary focus.

Figure 11 shows the steel grade investigated in journal articles identified in the literature review. The most frequently researched steels were X70 and X80. Both of these steels have been used for natural gas pipelines (Langley et al. 2010), and are more susceptible to HE than lower strength pipeline steels. Fewer articles have been published on lower strength pipeline steels such as X42 and X46. This is presumably because these steels are resistant to HE and therefore research into their hydrogen susceptibility would produce negative results. Some studies investigated steel grades from standards other than API 5L, such as F22 from the ASTM A182 standard (Fassina et al. 2011), A333Gr6 from the ASTM A333 standard (Zheng et al. 2012), and P110 from the API 5CT standard (Majchrowicz et al. 2019). Some researchers investigated steels that did not conform to a standard, listed in Figure 11 under ‘Novel steels’. These included steels with high copper contents (Shi et al. 2016), and steel microalloyed with niobium and titanium (Torres-Islas et al. 2013).

Figure 11:

Number of articles relating to different grades of pipeline steel. Novel steels are new steels created for the research that are not graded. Unknown steels were referred to by the authors as pipeline steels but the grade was not defined.

There have been previous reviews of HE in pipeline steels. Eliassen (2004) and Cabrini et al. (2015) focused on hydrogen generated by cathodic protection in submarine pipelines. Findley et al. (2015) reviewed the mechanisms behind the initiation of HIC, concluding that reversibly-trapped hydrogen was chiefly responsible for crack initiation. Ghosh et al. (2018) also reviewed HIC susceptibility and the influence of microstructure and chemical composition on HIC. Ohaeri et al. (2018) covered the many kinds of hydrogen degradation in pipeline steels and their relationships with microstructure. Xu (2012) reviewed the HE susceptibility of low-carbon steels generally, including low-strength pipeline steels. These steels generally had low HE susceptibility, although some mechanical properties were reduced – particularly fracture toughness and elongation to failure. For lower-strength carbon steels, studies reported similar HE susceptibility for the base metal, HAZ and weld. In higher-strength pipeline steels, there was some evidence that the weld and HAZ were most susceptible to HE. No subcritical crack growth was recorded in the base metal of X42 and X70 steels at hydrogen pressures up to 310 bar (Cialone and Holbrook 1985), although there was some evidence of subcritical crack growth in the weld and HAZ. Chemical composition influenced HE: carbon content increased HE susceptibility, as did manganese, chromium, molybdenum and vanadium. There was some evidence that microalloying niobium and titanium decreased HE. In reviewing the HE mechanisms, Xu concluded that HELP was the dominant mechanism in low-carbon steels, due in part to the high diffusivities of these steels which allow hydrogen to follow moving dislocations.

Ohaeri et al. (2021) reviewed hydrogen permeation and the influence of hydrogen on strength and fatigue properties of pipeline steels. Nanninga et al. (2010) reviewed the effect of hydrogen on fatigue crack growth in pipeline steels. They concluded that the crack growth rates in these steels were significantly affected by hydrogen, even in low-strength steels such as X42, and there was no strong relationship between pipeline steel strength and HE fatigue susceptibility. Although fatigue properties are an important consideration in the design of gas transmission pipelines carrying hydrogen, the literature on fatigue has been covered in detail by the previously mentioned reviews. Therefore, fatigue properties are not included in this review. This review will instead give an overview of previous research into tensile properties, hydrogen permeation, and fracture properties.

5.3 Pipeline degradation

A consideration in the repurposing of natural gas pipelines for hydrogen transmission is the possibility of degradation of pipelines during time in service. The mechanical properties of pipeline steels after field operation have been compared to samples of the same steel that have been kept in reserve. This has allowed the effects of service exposure to be isolated. Service exposure has been reported to reduce the impact toughness, fracture toughness, and ductility (Gabetta et al. 2008; Nykyforchyn et al. 2018, 2022b). The results for yield and tensile strength vary, as operation has been found to decrease or increase both yield and tensile strength (Gabetta et al. 2008; Nykyforchyn 2021; Nykyforchyn et al. 2022b; Zvirko et al. 2022). Nykyforchyn et al. (2022b) found service exposure caused the electric potential of the steel fracture surface to become more negative, from which they suggested that electrochemical testing of pipelines in the field could be used to estimate the degree of degradation.

In addition, the HE susceptibility of pipeline steels might increase over the steel lifetime. Figure 12 shows the results of Nykyforchyn et al. (2022a) who found that the effect of hydrogen on strength (YS and UTS) and ductility (elongation and reduction of area) of a VST3ps pipeline steel after 52 years of operation was greater than a reserve sample of the same pipeline steel. Zvirko et al. (2022) conducted tensile tests of 17H1S pipeline steel and found that the steel after 36 years of operation had a hydrogen embrittlement index of 38% compared with 6% for reserve steel.

Figure 12:

Results for strength (a) and ductility (b) in VST3ps steel as-received and after 52 years of service exposure (designated operated). White bars are without hydrogen, and black bars are after hydrogen charging for 100 h in an aqueous H₂SO₄ solution (pH 3.5) at a current density of 1 mA/cm². Reproduced from Nykyforchyn et al. (2022a) with permission.

Multiple mechanisms were proposed for the observed pipeline degradation. These include (i) natural aging, (ii) stress corrosion cracking, (iii) HIC during service in the field, and (iv) the production of internal defects, such as voids, in the steel over time due to stress (Nykyforchyn 2021; Nykyforchyn et al. 2010, 2018). Nykyforchyn et al. proposed the void model due to the observed increase in HE susceptibility, which they suggested was due to the creation of hydrogen traps. Gabetta et al. (2008) found field operation resulted in reduced hydrogen diffusivity, which would support this theory as traps reduce hydrogen diffusion.

5.4 Tensile properties

The previous sections of this review indicated that low-carbon ferritic steels, including low-strength pipeline steels, were resistant to HE. The literature review also indicated that higher-strength steels tend to be more susceptible to HE. It was therefore expected that studies into the tensile properties of pipeline steels in hydrogen would find that pipeline steels have low values of the hydrogen embrittlement index (I), which would increase with increasing pipeline steel strength.

Table 2 summarises the results of studies that conducted tensile tests of pipeline steels in hydrogen, listed in order of publication date. These include studies that used gaseous hydrogen and studies that generated hydrogen electrolytically. The I values (the hydrogen embrittlement index as calculated from either elongation or RA) and changes in strength (YS and UTS) that resulted from the charging conditions are included. In the electrolytic studies, the most frequently used electrolyte was sulfuric acid (H₂SO₄), used in almost all studies. Most experimenters also used a recombination poison, such as arsenic trioxide (As₂O₃), potassium arsenate (KH₂AsO₄), thiourea (SC(NH₂)₂), and ammonium thiocyanate (NH₄SCN). Charging current densities ranged from 0.1 mA/cm² to 66 mA/cm². In gaseous hydrogen studies, pressures ranged from 1 to 690 bar. Following Table 2 is an elaboration of these studies, and thereafter there is a synthesis of trends revealed by these studies.

Tensile experiments have also been used to investigate SCC of pipeline steels. SCC may involve the electrochemical generation of hydrogen alongside other corrosive processes (Ramamurthy and Atrens 2013). Numerous studies have conducted tensile tests with cathodic hydrogen charging in a corrosive solution (Li et al. 2006; Liang et al. 2009; Liu et al. 2012, 2016; Sant’Anna et al. 2016; Song et al. 2020a, b; Torres-Islas and Gonzalez-Rodriguez 2009; Wan et al. 2017; Zhu et al. 2016). Other studies have investigated the effects of hydrogen sulphide (H₂S) gas on pipeline steel tensile properties which also involves simultaneous corrosion and HE (Anijdan et al. 2021; Bai et al. 2020b; Silva et al. 2021; Titov et al. 2019; Wang et al. 2015a, b). Though these findings are important for pipeline safety, they are not pertinent to hydrogen pipelines and hydrogen blending into natural gas pipelines and therefore are not included in Table 2.

Gu et al. (1999) investigated IHE as part of a study of the SCC in X52 and X80 pipeline steels. The IHE study involved conducting SSRTs for specimens in air and precharged with hydrogen. The parameters of this precharging are given in Table 2; the resulting hydrogen concentration was not measured. For X80 steel, the hydrogen precharging did not affect the YS or UTS but reduced the strain at fracture by 9%. Both uncharged and hydrogen precharged specimens had similar fracture surfaces that were indicative of ductile fracture. The results for SSRTs of X52 steel were not given, presumably because there was no evidence of HE in this lower-strength steel.

Torres-Islas et al. (2005) investigated the effect of heat treatments on IHE in X70 using SSRT. The specimens were precharged electrolytically and tested in air. The heat treatments investigated were spray quenching, water quenching, and quench and tempering. The as-received steel had a mostly ferritic microstructure, while the water sprayed steel had a mostly bainitic microstructure with some acicular ferrite section. These specimens underwent ductile fracture and had low susceptibility to HE with I values of 10% and 5%, respectively. The water quenched steel and quench and tempered steel had martensitic microstructures, and underwent brittle fracture after hydrogen charging with respective I values of 85% and 90%. In another study, Torres-Islas et al. (2013) investigated HE in a novel microalloyed steel similar to the API 5L class pipeline steels. This study compared tensile specimens from the longitudinal and transverse directions with respect to the pipe axis. Transverse specimens are likely more representative of real pipeline failure as this relates to the probable orientation of fracture in a pipe wall due to hoop stresses. The transverse specimens were somewhat less susceptible to HE than the longitudinal specimens, although this trend was inconsistent.

Hardie et al. (2006) studied the effect of IHE on X60, X80 and X100 pipeline steel using tensile tests precharged with current densities varying from 11 to 70 mA/cm², with a charging time of 15 min. The hydrogen concentrations resulting from these current densities were not measured. The results were highly variable, possibly due to the low charging time. The changes in RA after hydrogen charging were similar for the three steel grades, dropping from an average of around 80% in the uncharged specimen to around 30% in the specimen charged at 10 mA/cm². Reductions in elongation after charging in the X80 and X100 were greater than in X60. As the current density increased further, RA and elongation remained mostly constant – this is likely due to the hydrogen concentration reaching the saturation level (Barthélémy 2006; Lynch 2011a). At high current densities (>60 mA/cm²) ductility may have decreased further in the X80 and X100 steels, although this finding may be unreliable due to the variation in results, and the authors had no explanation for this observation. Their fractographic analysis indicated both intergranular and transgranular cracking in the charged specimens, although the fracture was primarily transgranular. Hydrogen induced cracking had occurred on the specimen surface, and the cracks were oriented in the rolling direction of the pipe.

The authors argued that the absorbed hydrogen inside the specimen was the primary cause for losses in ductility. Their evidence was that a specimen charged at 66 mA/cm² resulted in severe surface cracking, but after the same specimen was allowed to outgas hydrogen for a week, the RA and elongation were similar to the uncharged specimen.

In the same paper, Hardie et al. (2006) reported no HE for X60 and X80 specimens precharged in gaseous hydrogen at 325 bar, and then tensile tested in gaseous hydrogen at 1.7 bar. Given the severe embrittlement that occurred after electrochemical charging, this suggests that the hydrogen concentrations achieved in the electrochemical charging conditions were much larger than in these gaseous conditions and were not relevant to gas pipeline conditions where the maximum gas pressure is 200 bar.

Several studies have focused on the relationship between electrolytic precharging time and IHE. Dong et al. (2009b) carried out IHE tensile tests of X100 after charging specimens for 1 h–8 h. The reduction in area decreased from approximately 33% in the uncharged specimen to 17%, 13%, and 8% after 1, 5, and 8 h of precharging, respectively. The decrease in ductility is attributed to increasing hydrogen concentrations resulting from increasing charging time. A similar decrease in ductility was reported in a later study by Dong et al. (2010) using X80 steel.

Rahman et al. (2019a) found that electrochemical hydrogen precharging X60, X60 stainless (X60SS), and X70 steels between 1 and 24 h increased both the UTS and YS by approximately 10%. Ductility reduced up to a maximum at 15 h, when the I was 58%, 71%, and 45% in the three steels. HE declined at longer charging times, which was attributed to a lower hydrogen concentration in the specimens due to the particular electrolytic hydrogen charging conditions.

Han et al. (2019) found that electrochemical hydrogen charging X100 pipeline steel reduced both YS and UTS, in contrast to the hydrogen-strengthening results from Rahman et al. (2019a) in which strength was increased by hydrogen in the lower-strength pipeline steels. I increased with charging time, up to 46% at 12 h, the maximum charging time tested. Increasing pre-strain hardened the steels, leading to increased YS and UTS and decreased elongation and RA. Pre-strain also increased the effects of HE, leading to reduced ductility.

Moro et al. (2010) found that the YS and UTS of X80 was unchanged in gaseous hydrogen up to 300 bar, but the ductility was decreased, with I of around 68% between 100 and 300 bar. At 50 bar I was 41%. At 1 bar there was only a minor hydrogen effect on the elongation, and no effect on reduction in area. The fracture surfaces of the embrittled specimens displayed less necking and quasi-cleavage fracture. As in the work of Hardie et al. (2006), the brittle fracture surface contained cracks oriented in the rolling direction. The authors attributed this to separation at the interfaces between pearlite and ferrite in the X80 steel. The specimens also showed external cracks close to the fracture surface running circumferentially around the surface.

Lee et al. (2011) found a negligible effect on the YS and UTS for X65 exposed to 200 bar hydrogen gas inside hollow tensile specimens. Hydrogen reduced the strain to fracture from 12% to 9%. Reducing the strain rate reduced the strain to fracture slightly further. Fractography showed a mixture of quasi-cleavage brittle fracture propagating from the inside of the specimen exposed to hydrogen and transitioning to a ductile fracture near the outer specimen surface. The amount of brittle fracture increased with lower strain rates.

Nanninga et al. (2012) found that X52, X65 and X100 in gaseous hydrogen at pressures ranging from 2 to 690 bar had negligible changes in YS and UTS. HE ductility losses increased with increasing steel strength. For X52, X65, and X100 at 138 bar hydrogen, I (as calculated from elongation) was 22%, 28%, and 50%, respectively. These results showed that even low-strength pipeline steels such as X52 can undergo HE at a hydrogen partial pressure of 138 bar. Transverse specimens were only slightly more susceptible to HE, even though they were generally stronger and more brittle than longitudinal specimens. X100 in hydrogen possibly underwent a small increase in YS and UTS, although the increases were within experimental error. 2 bar hydrogen reduced the reduction in area of X100 from 76% to 68% but had no effect on the elongation. Higher hydrogen pressures reduced reduction in area further: 55 bar hydrogen reduced RA to 30%; 276 bar hydrogen reduced RA to 23%. At 690 bar hydrogen saturation likely occurred. The authors found that there was no strong relationship between decreasing strain rates and HEE.

During fracture beyond the UTS, there was surface cracking in the necking region of the longitudinal specimens, extending in rings perpendicular to the strain direction. In the transverse specimens, the surface cracks were instead at an angle to the specimen axis, such that the cracks travelled in a helix along the specimen surface. This anisotropic surface cracking indicates that cracking occurred along interfaces in the steel microstructure that depended on the rolling direction – possibly (as noted by the authors) ferrite-pearlite interfaces. SEM imaging of the embrittled fracture surfaces showed a central ductile fracture region and a brittle region around the edges of the cylindrical specimen (the region expected to have the highest hydrogen concentration). The authors reported that the hydrogen fracture mode was quasi-cleavage (Nanninga et al. 2012).

Xiong et al. (2015) investigated X80 steel in a gaseous mixture of H₂ maintained at 1 bar and CO₂ from 2 bar to 10 bar. At 10 bar CO₂ there was no HE. At 5 bar CO₂ there was reduced YS (from approximately 660 MPa–600 MPa), UTS (approximately 710 MPa–650 MPa), and strain to fracture (11%–10%). At 2 bar CO₂ partial pressure, strain to fracture decreased further (9%), although the effect on the strength was negligible. These results indicate that HE can be significantly reduced by limiting the access of hydrogen to the steel surface by mixing with another gas with a higher partial pressure. Microanalysis of the fracture surface of the baseline nitrogen specimen showed necking and some deep MVC dimples characteristic of ductile pipeline steels. The embrittled specimens had less necking and a ductile central region and a brittle outer region. The central ductile region also had shallower dimples than those in nitrogen and some quasi-cleavage surfaces.

Hejazi et al. (2016) found for hollow X70 tensile specimens charged with 100 bar gaseous hydrogen at room temperature that I (calculated from elongation) was 25%. YS was unaffected and the UTS was reduced by ∼4%. Increasing the temperature to 50 °C reduced I to 18%, and at 100 °C I was 14%, which was attributed to steel softening at the increased temperatures.

Meng et al. (2017) found for X80 in gas mixtures of nitrogen and hydrogen, with the total pressure maintained at 120 bar and hydrogen partial pressures of 6, 12, 24 and 60 bar, that (i) YS and UTS were unaffected by hydrogen, (ii) I increased with increasing hydrogen partial pressure, up to 16% at 24 bar hydrogen, and (iii) beyond 24 bar, the degree of HE remained essentially constant, with an I value of 17% at 60 bar hydrogen. This indicates that saturation of HE occurred at around 24 bar hydrogen. Furthermore, this X80 steel had some resistance to HE because the losses of ductility were small even when saturated with hydrogen.

The results of Zhou et al. (2019) are shown in Table 3. They found that X80 precharged at 5 mA/cm² (Test 2) resulted in a 35% I, while hydrogen charging during the test using the same conditions (Test 3) resulted in a 79% I, and hydrogen precharging plus charging during the test (Test 4) increased I further to 88%. Increasing the strain rate from 2 × 10⁻⁵/s to 2 × 10⁻⁴/s reduced I (i) from 35% (Test 2) to 10% (Test 5), and (ii) from 99% (Test 4) to 80% (Test 6). Hydrogen had almost no effect on the YS across all tests. Annealing the X80 increased the YS and reduced the susceptibility to HE (Tests 7–9). Microanalysis of the fracture surfaces showed a reduced MVC dimple size in the IHE specimen (Test 2). The HEE (Test 4) fracture surface had few dimples and was quasi-cleavage fracture, and the fracture crack had extended inwards from the specimen surface.

Xing et al. (2021) found reduced ductility for X90 charged with hydrogen in NS4, a solution used to simulate hydrogen charging due to corrosion in soil, at temperatures from 0–60 °C. The degree of HE was a maximum at 40 °C, in agreement with the common observation of a maximum HE at around room temperature in steels (Lee 2016; Louthan 2008). The maximum decrease in elongation was small, from around 14% in air to 11% under HE. YS slightly increased, although there was no relationship between temperature and YS. At 30 °C, the fracture surface contained quasi-cleavage features. At 40 °C, the temperature with the greatest embrittlement, the fracture surface was evenly brittle with no dimples. From 50 °C onwards, dimples reappeared and necking resumed (Xing et al. 2021).

Figure 13 presents the dependence on hydrogen content of the hydrogen embrittlement index (I), evaluated from the reduction in elongation at failure from studies where this data is available. The higher-strength X80 (Zhang et al. 2017b; Zhou et al. 2019) had higher I values than the lower-strength steels such as X42 and X65 (Bai et al. 2020a; Neeraj et al. 2012; Singh et al. 2019), despite the X65 having higher hydrogen concentrations. This indicates that X65 is significantly less susceptible to HE than X80 for hydrogen concentrations up to 5.5 wt ppm. Lunarska et al. (1997) found that 13CrMo44 pipeline steel had no decrease in ductility with up to 0.2 wt ppm hydrogen, and I was 11% at 0.5 wt ppm, and 17% at 1.5 wt ppm. These results support the existence of a minimum hydrogen concentration below which HE does not occur. Zhang et al. (2017b) found an internal hydrogen concentration of 1 ppm was enough to cause significant embrittlement in X80 steel. Increasing the current density beyond that which produced 1 wt ppm resulted in significant damage to the specimen, including blistering even with no applied stress.

Figure 13:

Hydrogen embrittlement index (I) against subsurface hydrogen concentration.

The lower strength steels (13CrMo44, X42, and X65) had similar I values across a wide range of hydrogen concentrations (from 0.27 to 5.4 wt ppm). This suggests that hydrogen saturation occurred under increasing internal hydrogen concentration.

Figure 14 shows that the pipeline steels (X70, X80, X100) exposed to 1–2 bar hydrogen typically indicated a small degree of HE (Moro et al. 2010; Nanninga et al. 2012; Nguyen et al. 2020a). Not included in Figure 14 is the data of Xiong et al. (2015) who found that X80 charged at 1 bar hydrogen had an I value of 22%, but this was in the presence of 2 bar CO₂ which may have had a confounding effect. At 100 bar hydrogen and above, there was significant embrittlement (I> 50%) in X70 (Nguyen et al. 2020a), X80 (Moro et al. 2010), and X100 (Nanninga et al. 2012), but less embrittlement for X52, X60 and X65 exposed to 100–150 bar hydrogen which had lower I values between 20 and 30% (Boot et al. 2021; Lee et al. 2011; Nanninga et al. 2012). This indicated a general relationship between increasing steel strength and increasing HE susceptibility. However, an exception to this trend was the I of 25% for X70 in 100 bar hydrogen (Hejazi et al. 2016). This variation indicates that HE resistance can vary significantly among steels of the same grade with similar strengths and chemical compositions.

Figure 14:

HE index (I) against hydrogen partial pressure in tensile studies that used gaseous hydrogen.

Hydrogen saturation is indicated by the hydrogen pressures beyond which HE did not continue to increase (or continued to increase slowly). Figure 15 shows estimates of hydrogen saturation conditions. The maximum pressure shown was the pressure which exhibited the largest embrittling effect. Greater pressures caused a similar degree of embrittlement. The minimum pressure shown is the next lowest pressure tested below the saturation pressure. This gives a range over which hydrogen saturation may have occurred. All gaseous tensile studies that tested multiple pressures of pure hydrogen found evidence of hydrogen saturation, indicating that this is a general phenomenon. Included in Figure 15 is the work of Xu and Rana (2008), who conducted tensile tests in low-carbon steels similar to pipeline steels, ASME SA-106 and ASME SA-372, and found that saturation was reached at or below 69 bar.

Figure 15:

Hydrogen pressure ranges at which hydrogen saturation may have occurred in tensile studies. Higher hydrogen pressures tested beyond the maximum pressures shown caused no additional embrittling effect.

Most of the studies indicated that hydrogen saturation occurred at below 55 bar. The results of Nanninga et al. (2012), however, indicated that saturation occurred at a pressure between 55 bar and 138 bar. These results may indicate that X100 was sufficiently different from lower-strength pipeline steels to cause a noticeable difference in its HE susceptibility.

Not shown in Figure 15 are the results of electrochemical tensile studies. Bai et al. (2020a) found saturation occurred at 2.5 mA/cm² for an X42 steel. Hardie et al. (2006) found HE reached a maximum at a current density of 10 mA/cm² for X60, X80 and X100 steels, although ductility may have decreased further at much higher current densities (>60 mA/cm²). Stalheim et al. (2010) found similar results for X60, X70 and X80. These results indicate that hydrogen saturation may be largely independent of strength and microstructure, and the variation in results across different studies may be due to differences in charging or testing conditions. Similarly to Hardie et al. (2006), Nanninga et al. (2012) found evidence of further embrittlement occurring at very high concentrations beyond saturation, in their case at high pressures above 500 bar.

Figure 16 shows the HE index calculated from the UTS, charted against hydrogen partial pressure. The percentage changes were small across all pressures and steel grades. Some studies found hydrogen increased strength slightly, and some found hydrogen reduced strength slightly, but there was no relationship between increased or reduced strength and other variables such as steel strength or hydrogen partial pressure. Previous reviews of HE have noted that hydrogen generally does not affect the strength of high-strength martensitic steels but can reduce the strength of low-strength ferritic steels (Atrens et al. 2020; Hirth 1980), but there was no clear relationship between steel strength and strength change in hydrogen in the literature reviewed.

Figure 16:

Percentage reduction of the UTS after charging in hydrogen compared to the UTS in air, charted against hydrogen partial pressure. A negative HE index value indicates that UTS was increased after hydrogen charging.

Figure 17 shows that UTS was generally reduced by a larger amount in the electrochemical studies than in the gaseous studies shown in Figure 16. Unlike the gaseous charging studies, there is evidence that hydrogen affects the UTS in the electrochemical charging studies for substantial hydrogen concentrations above ∼0.5 wt ppm. However, there is no clear relationship between increasing internal hydrogen concentration and UTS.

Figure 17:

Percentage reduction in UTS in electrochemical studies that evaluated the internal hydrogen concentration. A negative HE index value indicates that UTS was increased after hydrogen charging.

It has been proposed that relatively low hydrogen concentrations generated by gaseous charging reduce steel strength, while the higher concentrations typically generated by electrochemical charging increase strength (Zhao et al. 2015). The data shown in Figures 16 and 17 instead support the opposite, that low hydrogen concentrations can increase strength while high hydrogen concentrations reduce strength, as a greater number of studies of gaseous hydrogen charging found hydrogen strengthened the steel (as evidenced by a negative HE index). Meng et al. (2017), studying X80, and Nanninga et al. (2012), studying X100, both found that UTS was increased by exposure to low hydrogen partial pressures below 10 bar. Zhang et al. (2017b) found that X80 electrochemically charged with 1 wt ppm hydrogen had UTS reduced by 18%.

Xu (2012) concluded from a literature review that HELP was the dominant HE mechanism in low-carbon steels. Moro et al. (2010) instead considered HEDE the primary mechanism of embrittlement for X80. They observed interfaces between layers of ferrite and pearlite in the fracture surface of the rolled steel, and proposed that hydrogen transported by dislocations had accumulated at these interfaces and caused HE by decohesion. Han et al. (2019) indicated that HE was caused by HELP at low hydrogen concentrations in X100, which was evidenced by a reduction in the YS of the steel, while they proposed a mixture of HELP and HEDE caused HE at higher concentrations, where HEDE was evidenced by secondary cracks on the fracture surfaces.

5.4.1 Tensile properties of welds

There has been limited research into the tensile properties of pipeline welds in hydrogen gas. Nguyen et al. (2020c) found that the strength of X70 weld metal was unaffected by hydrogen, but ductility was reduced, with elongation decreasing from approximately 26% in air to 21%, and reduction in area decreasing from approximately 70% in air to 50%. The specimens were exposed to a gas mixture of 1 bar hydrogen and 99 bar methane. The ductility of the base metal was unaffected by hydrogen. These results indicated that these pipeline welds were more susceptible to HE than the base metal. The authors used SEM to analyse the fracture surfaces. The weld specimen fractured in air underwent ductile failure and the surface contained MVC dimples. After hydrogen charging, there was less necking. The fracture surface was split between a brittle region at the surface (the region expected to have higher hydrogen content) showing evidence of quasi-cleavage fracture and a central ductile region. There were surface cracks in the necking region, which in some specimens were parallel to the fracture surface, while in others were at a 45° angle to the fracture surface. The possible causes for the different crack orientations were not fully explored.

The welds were girth pipe welds created by orbital welding. The microstructures of the base and weld metal are shown in Figure 18. The X70 base metal microstructure was ferritic, consisting mostly of acicular and polygonal ferrite. The weld metal had a less uniform microstructure, with interspersed sections of dendritic ferrite and bainite as well as various inclusions (Nguyen et al. 2020c).

Figure 18:

Microstructures of X70 steel base metal and weld metal. Reproduced from Nguyen et al. (2020c) with permission.

To explain the reduced hydrogen tolerance in the weld, Nguyen et al. (2020c) proposed several contributing microstructural differences: (i) the more homogeneous microstructure of the base metal allowed hydrogen to uniformly diffuse and not become concentrated at crack initiation sites; (ii) acicular ferrite (in the base metal) contains more hydrogen traps than bainite, as reported by Park et al. (2008), and these traps reduced the amount of diffusible hydrogen which was responsible for HE; (iii) the alternating dendritic ferrite and bainite in the weld metal led to differences in hardness which can localise strain, leading to hydrogen concentration and crack initiation; and (iv) inclusions in the weld metal served as hydrogen concentration sites and therefore crack initiation sites.

Nguyen et al. (2021) produced similar results in a subsequent study of notched tensile specimens: the X70 base metal was unaffected by hydrogen but the weld was embrittled. The stress concentration in the notched tensile specimens correlated with increased HE, and base metal and weld metal were similarly affected by the stress concentration.

Boot et al. (2021) found that notched, hollow X60 base metal specimens in 100 bar hydrogen had an I value (as measured from reduction in area) of 28%, while the weld specimens had an I of 11%. These results show that, despite the weld being stronger and more brittle, the base metal was more susceptible to hydrogen than the weld. The YS and UTS of the base metal or weld were unaffected by hydrogen. Reducing the tested hydrogen pressure from 100 bar to 30 bar produced no change in the ductility, suggesting saturation occurred at a hydrogen pressure below 30 bar.

Figure 19 summarises the reduction in ductility in the pipeline weld tensile studies. Nguyen et al. (2020c) found that the weld material was more susceptible to HE than the base metal, while Boot et al. (2021) found the base metal was more susceptible to HE than the weld. More studies are required to reconcile this inconsistency. Figure 20 shows that hydrogen had a negligible effect on UTS in both the weld metal and the base metal.

Figure 19:

HE index resulting from hydrogen partial pressure in tensile studies of pipeline steel welds. HE index is calculated from elongation at fracture.

Figure 20:

Percentage reduction of the UTS in base metal and weld metal. A negative HE index value indicates that UTS was increased by hydrogen charging.

5.5 Permeation experiments

Table 4 summarises the experimental conditions used in electrochemical permeation experiments in pipeline steels. All studies used the Devanathan-Stachurski cell design discussed in Section 4.2. The most frequently researched pipeline steels were X70 and X80, with some studies on X65, X100 and F22 steels. As with the tensile experiments discussed in Section 5.4, the most frequently used electrolyte for hydrogen generation was sulfuric acid (H₂SO₄). Recombination poisons were used frequently; commonly used poisons included arsenic trioxide (As₂O₃), thiourea (SC(NH₂)₂), and sodium sulphide (Na₂S·9H₂O).

Table 5 summarises the effective diffusivities from the permeation studies. Diffusivities were most frequently calculated from the time lag method. Table 5 indicates that there may be a general relationship between decreasing diffusivity and increasing strength. This trend is highlighted by Figure 21. Figure 21 shows that the hydrogen diffusivities varied greatly within steel grades. For example, the hydrogen diffusivities found by studies of X80 varied from 5.3 × 10⁻⁹ to 9.8 × 10⁻⁵ cm²/s.

Figure 21:

Values of hydrogen diffusivity in permeation studies grouped by steel grade. The geometric mean diffusivity for each steel grade is shown by a diamond.

Numerous researchers investigated the influence of heat treatment on hydrogen permeation. Torres-Islas et al. (2005) found that three heat treatments of X70 all increased the hydrogen permeation flux significantly, with the largest permeation flux after quenching, then quenching and tempering, then water-spraying. The researchers did not calculate the hydrogen diffusivities or hydrogen concentrations. Fallahmohammadi et al. (2014) found that the hydrogen diffusivities of X65 were highest when annealed at 930 °C (2.0 × 10⁻⁵ cm²/s), followed by quenched and tempered (4.7 × 10⁻⁶ cm²/s), and quenched from 930 °C (1.4 × 10⁻⁶ cm²/s). They found that the hydrogen concentration increased as the diffusivity decreased (attributed to increased trapping). Mohtadi-Bonab et al. (2014) evaluated the effective diffusivities (which includes the effect of trapping), with the highest diffusion coefficient in the as-received steel (3.9 × 10⁻⁶ cm²/s), followed by water-sprayed from 850 °C (2.9 × 10⁻⁶ cm²/s), tempered at 350 °C (2.6 × 10⁻⁶ cm²/s), and quenched from 850 °C (2.2 × 10⁻⁶ cm²/s). They also found that the hydrogen concentration had an inverse relationship with diffusivity. Zhou et al. (2019) found that heat treating X80 at 200 °C increased diffusivity by ∼65%. Li et al. (2020b) varied the tempering temperature of an X80 steel from 450 °C to 700 °C, finding that increasing tempering temperature reduced the diffusivity and increased the hydrogen concentration. The available heat treatment research indicated that annealed pipeline steels had significantly higher diffusivities, and quenched steels had the lowest diffusivities.

Other researchers have investigated the influence of deformation on hydrogen permeation. Li et al. (2020a) found that the diffusivity increased slightly with increased compression ratio of finish rolling from 20% to 50% of X80, but there was little influence on hydrogen solubility. Thomas and Szpunar (2021) found for X70 increasing the degree of cold rolling from 0 to 80% lowered the effective diffusivity from 2.5 × 10⁻⁶ to 1.1 × 10⁻⁷ cm²/s, and the lattice diffusivity from 3.3 × 10⁻⁶ to 4.2 × 10⁻⁷ cm²/s, attributed to the significantly increased number of traps.

Figure 22 shows that most studies have found that hydrogen diffusivity increased with increasing temperature. The one exception was Liu et al. (2019) who found a decrease of diffusivity with increasing temperature for an X65 steel, ascribed by the authors to a high hydrogen concentration producing voids, which impeded hydrogen diffusion. However, most studies (as indicated in Figure 22) have found increasing diffusivity with increasing temperature attributed to the increased temperature providing more energy for the elemental jumps of the hydrogen atom which is the basis of hydrogen diffusion.

Figure 22:

Hydrogen diffusivity typically increases with increasing temperature in permeation studies of pipeline steels. The data indicate no general tendency with steel strength.

Two papers have investigated hydrogen permeation in the weld and HAZ of pipeline steels. Skjellerudsveen et al. (2010) found the diffusivity in the weld averaged ∼150% of the X70 base metal. The HAZ had the smallest diffusivity, averaging ∼16% of the base metal. The hydrogen concentrations in the base metal, weld and HAZ averaged 0.27, 0.08 and 1.05 ppm, respectively. In contrast, Xue and Cheng (2013) found in X80 that the base metal had the highest diffusivity, followed by the weld metal, then the HAZ. The weld had a similar hydrogen concentration as the HAZ. The authors also evaluated the number of traps, finding that the HAZ had the highest number of traps, followed by the weld, then the base metal.

Table 6 gives results for subsurface hydrogen concentrations in studies that gave details on the applied current or potential. Figure 23 shows that there was no clear relationship between current density and concentration across the permeation studies. This variation may be attributed to differences in charging conditions and differences in steel processing greatly influencing the number of traps. There was a significant degree of variation in the hydrogen concentrations, with the majority ranging from 0.029 to 197 wt ppm. Li et al. (2020a, 2020b) found that the hydrogen concentration varied from 0.62 to 6.3 wt ppm in X80 depending on heat treatment and amount of rolling reduction. Dong et al. found a hydrogen concentration of 197 wt ppm in X70 Dong et al. (2010) and 34.3 wt ppm in X100 (Dong et al. 2009b) steels using the same electrolyte and current density as in Li et al. (2020a). This may suggest that more hydrogen is generated at the surface of X70 steels than X100 steels, or it may be due to variations in the processing of these steels. An anomalously low concentration of 1.06 × 10⁻⁷ wt ppm was reported by Liu et al. (2019), which is assumed to be an error, which is supported by the reported anomalous diffusivities in this paper.

Figure 23:

Subsurface hydrogen concentration resulting from applied current density in permeation studies of pipeline steels. The legend denotes the steel grade, electrolyte used, and reference number. Details on the concentrations are given in Table 6.

Studies of gaseous hydrogen permeation are rare, and no studies of gaseous permeation in pipeline steels were found in this literature review. For this reason, the relationship between hydrogen gas pressure and subsurface concentration for pipeline steels could not be evaluated. The equivalent fugacity resulting from cathodic overpotential (discussed in Section 4.2) for pipeline steels has also not been evaluated.

5.6 Fracture toughness

Table 7 tabulates the effect of hydrogen on the fracture toughness of pipeline steels. Only studies that included results for fracture toughness in air are included so that results can be compared by the hydrogen embrittlement index, I_FM. I_FM can be calculated from numerous fracture parameters including the stress intensity factor (K), the J-integral, and the crack-tip opening displacement (CTOD) at fracture. The I_FM values listed in Table 7 are based on the reported fracture parameters. The specimen orientation uses the two-letter codes defined in ASTM E399 (ASTM International 2020). The first letter defines the direction perpendicular to the crack plane (which is also typically the longest dimension of the specimen). The second letter defines the direction of crack propagation. The directions are L for longitudinal, T for transverse, and S for short transverse (pointing into the pipe wall). For example, LS designates a specimen cut from the longitudinal plane with the notch directed into the pipe wall.

Li et al. (2018) found that hydrogen significantly reduced the crack tip opening displacement (CTOD) for all specimens tested, which were single edge notch tensile (SENT) specimens. I_FM increased significantly as the thickness-to-width ratio was increased, from 66% at a thickness-to-width ratio of 0.5 to 91% at a ratio of 2. This indicates that the influence of hydrogen on fracture toughness in this case depends upon the specimen dimensions.

Hardie et al. (2006), using tensile specimens, found that hydrogen embrittlement saturation appeared at 10 mA/cm² as HE did not increase beyond this point. In contrast, Wang (2009) and Kyriakopoulou et al. (2020a) found that fracture toughness was further reduced at higher currents up to 100 mA/cm², which was the maximum current tested (Wang 2009). Kyriakopoulou et al. (2020a) found fracture toughness was reduced up to the maximum current tested, 30 mA/cm², and the number of blisters increased with current density above 10 mA/cm², indicating that these applied current densities were producing hydrogen fugacities greater than the strength of the steel and were directly damaging the steel.

Nguyen et al. (2020a) found that the fracture toughness measured with compact tension (CT) specimens was reduced by 30% in a gaseous environment of 1 bar hydrogen and 99 bar methane. This indicates that small amounts of hydrogen in natural gas pipelines (whether sourced from methane enrichment or electrochemical processes such as cathodic protection) can significantly reduce fracture toughness. This contrasts with the tensile results in the same environment given in the same article, where this hydrogen partial pressure only negligibly reduced the ductility.

Generally, the hydrogen embrittlement index (I) values evaluation using the fracture toughness studies were greater than those evaluated using ductility from the tensile studies in Table 2, indicating that hydrogen affects fracture properties more than ductility in pipeline steels. For example, Cabrini et al. (2019) using the same electrolytic hydrogen charging solution found that (i) at the open circuit potential the fracture toughness of X70, X80, and X100 was reduced by 13–42%, whereas there was no reduction in ductility in the tensile tests, and (ii) at an applied potential of −2 V the fracture toughness was reduced significantly (68–89%) compared with a lower reduction in the reduction in area (37–48%) in the tensile tests. Nguyen et al. (2020a) found for X70 in gaseous hydrogen that (i) at 1 bar hydrogen, the RA in the tensile tests was reduced by 3% while the fracture toughness was reduced by 30%, and (ii) at 100 bar hydrogen, RA was reduced by 33% and fracture toughness was reduced by 49%.

Table 7 indicates that most studies found hydrogen reduced fracture toughness significantly, indicating that pipeline steels are susceptible to hydrogen induced fracture. The single anomalous article was that by Chatzidouros et al. (2018) which found that (i) the stress intensity factor (K_Q) was unaffected by the hydrogen concentration for pre-charged specimens, but (ii) the CTOD was significantly reduced by 33% for 0.5 ppm hydrogen, 78% ppm for 1 ppm hydrogen, and 82% for 2 ppm hydrogen. The inconsistencies in these results could be because the specimens were pre-charged as it was found by Wang (2009) that pre-charging could produce inconsistent results. However, similar results were found in the other fracture toughness studies by Chatzidouros et al. (2011), (2014) which featured in-situ charging: the fracture parameter K_Q was also unaffected by hydrogen, while other fracture toughness parameters, such as the J-integral, were reduced by hydrogen.

This may indicate that K_Q is not affected by hydrogen in three-point bending (TPB) tests. In contrast, Kyriakopoulou et al. (2020a), (2020b) found that K_Q was significantly reduced by hydrogen in similar TPB tests on X65 charged in-situ at 10 mA/cm². Kyriakopoulou et al. (2020b) found the J-integral was reduced by 28%, and K_Q was reduced by 17%. Chatzidouros et al. (2014) found the J-integral was reduced by 85% in the hydrogen charged-specimens, whereas K_Q actually increased by 9%. In two of their studies, Chatzidouros et al. (2018) and Kyriakopoulou et al. (2020a) used identical electrolytes yet still found conflicting results for the effect of hydrogen on K_Q. This discrepancy may indicate that the two groups used different methods for calculating K_Q.

Figure 24 shows the influence of the hydrogen concentrations on the hydrogen embrittlement index based on the fracture toughness. The irregular results from Chatzidouros et al. (2018) are included. The results for X70 from Wang (2009) and Hejazi et al. (2012) indicate that I_FM varies from 15% to 42% at ∼2 wt ppm. This shows the variability of the measured effect of hydrogen on fracture toughness for similar pipeline steels. The results of Wang (full circles) indicate the hydrogen embrittlement index increases rapidly for small amounts of hydrogen, and thereafter increases more slowly. The results of Wang (2009) and Li et al. (2018) indicate high values of the hydrogen embrittlement index for high hydrogen concentrations of 2–2.5 wt ppm.

Figure 24:

Influence of hydrogen concentration on the hydrogen embrittlement index evaluated from the fracture toughness for pipeline steels.

Hejazi et al. (2012) found that the simulated HAZ of X70 had a substantially higher hydrogen embrittlement index based on the fracture toughness than the base metal. In contrast, Chatzidouros et al. (2011) found that the hydrogen embrittlement index based on fracture toughness of the HAZ was less than that of the base metal. Kyriakopoulou et al. (2020a) found that for X65, hydrogen had a similar influence on the fracture toughness of the weld and base metal.

Chatzidouros et al. (2011) found that (i) the base of metal of X52 consisted of bands of alternating ferrite and pearlite, and (ii) the HAZ also consisted of alternating ferrite and pearlite but in a non-banded structure with varying grain sizes. Numerous authors have found that the microstructure of X65 steel typically consists of ferrite and bainite phases, with some islands of degenerated pearlite (Chatzidouros et al. 2014, 2018, Kyriakopoulou et al. 2020a, b). Hejazi et al. (2012) reported that an X70 consisted of ferrite and pearlite, while Nguyen et al. (2020a) (2020b), found an X70 consisted of bands of ferrite and bainite. Chatzidouros et al. (2011) found the HAZ of X70 had transformed some of the pearlite in the base metal into ferrite. Cabrini et al. (2019) found X80 was mostly bainite with some islands of ferrite, and X100 consisted of martensite and bainite.

Comparison of the I_FM values of different steel grades or microstructures is made possible by studies that conducted experiments on multiple steel grades and used consistent testing procedures and fracture parameters. Chatzidouros et al. (2011) found that the X70 base metal (banded ferrite and bainite/pearlite) was slightly more susceptible to HE than the X52 base metal (banded ferrite and pearlite), whereas the I_FM values of the two HAZs were similar. Cabrini et al. (2019) found no correlation between steel strength and I_FM for X70, X80, and X100, with the X80 steel having the lowest I_FM and the X70SS steel having the highest I_FM. These two studies found no strong relationship between steel strength and I_FM, suggesting that the influence of hydrogen on fracture toughness is not strongly related to pipeline steel grade (or microstructure).

Numerous studies found that fracture toughness specimens fractured in air showed ductile features such as dimples and microvoids. After hydrogen charging, specimens showed features indicative of quasi-cleavage, with a mixture of dimpled and cleavage regions (Hejazi et al. 2012; Kyriakopoulou et al. 2020a, b; Li et al. 2018; Nguyen et al. 2020a; Wang 2009). Dimples in hydrogen-charged specimens were smaller, shallower, and elongated (Hejazi et al. 2012; Kyriakopoulou et al. 2020a). At higher hydrogen concentrations, more of the fracture surface showed cleavage facets, with larger steps between facets (Hejazi et al. 2012; Wang 2009).

Wang (2009) reasoned that the HELP mechanism was responsible for HE in specimens with reduced dimple size, as HELP reduces the stress required for ductile fracture at the crack tip, leading to smaller dimples. Wang proposed that the HEDE mechanism was responsible for HE in specimens with brittle cleavage facets. Hejazi et al. (2012) investigated X70 with varying manganese content and came to the same conclusion. They stated that manganese content in steels impedes dislocation motion, restricting the HELP mechanism, and these steels, therefore, predominantly undergo HEDE and show cleavage surfaces.