Characterization of Hydrogen Trapping Systems and HIC Susceptibility of X60 Steel by Traditional and Innovative Methodologies

Abstract

Hydrogen interaction with steels and metallic alloys, in general, is an old issue, but the interest in the phenomenon is incredibly increased in recent years. From the ’90s of the last century up to today, the papers about Hydrogen Embrittlement (Scopus, Elsevier, Amsterdam, [1]) are almost tripled! This establishes the growing attention on this topic, due also to the development of hydrogen-based economy. In Oil & Gas industry, hydrogen–steel interaction and related phenomena are often related to corrosion reactions, a consequence of the severe sour environments typical of this industrial sector. This is the reason why corrosion is the second most frequent cause of pipeline failures (Yang et al., Reliab Eng Syst Saf 159, 214–222, [2]), with the peculiar result of dangerous substances released in the environment, and it has translated into a continuous development of new technologies to monitor and control the ‘corrosion state’. The present work aims at characterizing the hydrogen susceptibility of X60 steel, a High-Strength Low-Alloy (HSLA) steel, widely used in Oil & Gas industry. The study is carried out by means of a rigorous approach based on traditional scientific techniques; moreover, an innovative solution was developed, validated and proposed to approach the possibility of on-field monitoring of operating pipelines.

12.1 Introduction

Hydrogen embrittlement is a phenomenon of great interest for Oil&Gas industry, due to the use of higher grades steels, as well as low-purity service fluids, containing H$_2$S, for example. In this scenario, the possibility of approaching the plants and pipelines on-field monitoring looks very attractive and could be the real turning point for reducing failures and manage safe planned maintenance.

The mechanisms of hydrogen embrittlement and hydrogen interaction with metal lattice are still under investigation, but some theories are overall accepted to demonstrate the dangerous and deteriorating outcome of hydrogen absorption on steels’ mechanical properties and behaviour. Nevertheless, the effect of hydrogen is mainly and deeply investigated for low-carbon steels, stainless steels, TRIP steels, DP steels, pipeline steels and so on [3], it is well known that the embrittlement effect is particularly severe for high-strength and ultra-high-strength steels. Typically, a UTS higher than 1000 MPa is assumed as threshold value to significantly appreciate it.

The most diffused models that describe the hydrogen embrittlement in metals are the HELP (Hydrogen-Enhanced Local Plasticity) and HEDE (Hydrogen-Enhanced DEcohesion). These models are quite different, HELP is related to hydrogen enhancement of dislocations’ generation and motion, giving rise to local plasticized zones and material softening. Other literature studies state the opposite phenomenon, that is the pinning of dislocations due to the presence of hydrogen atoms. Katzarov et al. [4] and Taketomi et al. [5] observed and demonstrated the coexistence of both actions. Hydrogen affects dislocation mobility as a function of several factors, including hydrogen concentration, tension stress state and temperature. HEDE model is instead based on the reduction of metal lattice chemical bounds due to the presence of hydrogen, which promotes crack formation and propagation. It is strictly correlated to locally reach the critical hydrogen concentration.

Currently, the most recent researches propose a new approach which consists of the synergistic contribution of HELP and HEDE theories, the so-called HELP-mediated HEDE model [6]. A first HELP stage is followed by a combination of HELP and HEDE mechanisms, where hydrogen enhances the mobility of dislocations (HELP), giving rise to high-density dislocations pile up and plasticization at the crack tip. At the same time, hydrogen transportation by dislocation increases the local concentration in the metal leading to decohesion of chemical bonds (HEDE).

Hydrogen-induced cracking (HIC) is another phenomenon typical of Oil & Gas industry. It usually takes place near to precipitates or non-metallic inclusions, vacancies or other microstructural defects, and consists in crack nucleation and propagation even without applied load to the part or component. When hydrogen accumulates in these areas, it can recombine into molecular hydrogen and reach a local overpressure level able to crack the metal according to hydrogen pressure theory [7].

Peng et al. [7] stated the HIC nucleation depends on the type and shape of the inclusions: MnS, as well as Al, Ti and Ca oxides, are very active initiation sites, while inclusions rich in Si present a low risk. Recent studies [7] showed a better resistance of pipeline steels with controlled precipitation of carbides and nitrides (as V$_4$C$_3$, TiC, NbC and nanosized (Ti,Mo,C) aimed at reducing particles size and increasing their number [8].

Sour environment, typical of petrochemical plants, is characterized by the presence of a strongly reactive and dangerous compound, the hydrogen sulphide. H2S inhibits the recombination of atomic hydrogen into molecular hydrogen according to the following reactions:

$$\begin{aligned} \mathrm{H}_2\mathrm{S} + e^- \rightarrow \mathrm{H} + \mathrm{HS} \end{aligned}$$

(12.1)

$$\begin{aligned} \mathrm{HS}^- + e^- \rightarrow \mathrm{H} + \mathrm{S}^{2-} \end{aligned}$$

(12.2)

$$\begin{aligned} \mathrm{H}^+ + e^- \rightarrow \mathrm{H}. \end{aligned}$$

(12.3)

Equation (12.3) allows atomic hydrogen to be adsorbed on the metal surface, considerably increasing the probability of absorption in the bulk material and the risk of hydrogen damage [9].

12.2 Materials and Methods

The present work focuses on API X60 and Iron-Armco§. API X60 is HSLA pipeline steel that combines good resistance, ductility, weldability and corrosion resistance. The strengthening mechanisms consist of the partial precipitation of inclusions, coupled with grain size reduction and it is subjected to hot rolling. Fe Armco is instead cold rolled and annealed. Mechanical properties and chemical compositions are listed in Tables 12.1 and 12.2 respectively.

Table 12.1 Mechanical properties of API X60

Table 12.2 Composition of Iron Armco §  and API X60

For permeation tests, rectangular sheet specimens of different thickness (0.5, 1 and 2 mm) were used, for HIC tests and TPD analyses, cylindrical specimens were manufactured, of diameter, respectively, 13 and 6 mm.

12.3 Experimental

A series of experimental tests were carried out in order to characterize the steel behaviour in a hydrogen environment and its susceptibility to hydrogen embrittlement.

12.3.1 Permeation Tests

Permeation tests were performed to evaluate hydrogen diffusion in steels, determining the hydrogen diffusion coefficient.

The traditional test method, based on Devanathan–Stachurski double cell [10], was used. The Devanathan–Stachurski technique involves the application of a Pd coating to the sample surface. A solution of PdCl$_2$ (5 g/L) and NH3 (28% in volume) was used and a current density of 4 mA/cm$^2$ [11] was applied, after a prior surface polish. The sample is placed between two semi-cells and surface passivation of one side is achieved by using 0.1 N NaOH solution and imposing 200 mV with respect to Ag/AgCl reference electrode.

Permeation tests were also carried out by means of HELIOS 2 (Hydrogen Embrittlement Line Instruments and Operative Sensors), an innovative equipment patented by Letomec S.r.l [12] (patent EP2912452B1), and results were compared.

HELIOS 2 is based on an electrochemical cell, where the metallic sample, the cathode, is in contact with the electrolytic solution through a hole in the cell with calibrated diameter. Applying a certain current between the anode and the cathode, hydrogen is produced on the exposed surface of the sample. A probe, containing a solid-state gas sensor, registers the hydrogen flux coming out from the other side (see Fig. 12.1). HELIOS method since there is no electrolytic solution on the detection side, doesn’t require the passivation of the metallic sample.

Fig. 12.1

Helios 2 equipment scheme

After the preliminary comparison among Devanathan and HELIOS 2 methods on the reference material (Iron Armco), permeation tests were repeated on API X60 HSLA steel. Tests were executed by varying operative temperature, to estimate the low-energy traps binding energy and, in a second phase, tests on samples with different thicknesses were executed. Finally, for what concerns permeation tests, a real case of cracking was simulated. A sheet of 2 mm thickness has been put in contact with H$_2$SO$_4$ 1 N and As$_2$O$_3$ 10 mg/L and a current density of 50 mA/cm$^2$ was applied.

12.3.2 TPD Tests

Hydrogen trapping systems were studied by the temperature-programmed desorption technique. Pre-charged samples were heated from room temperature up to 600 $^\circ$C. Various heating rates were applied, and binding energy of traps was extrapolated. According to Choo and Lee model [13], the heating rate can be expressed as follows:

$$\begin{aligned} \dot{z}=A(1-z)e^{E_a/RT} \end{aligned}$$

(12.4)

with $z=(C_0-C(t))/C_0$ and where $C_0$ and C(t) represent, respectively, the initial and the instantaneous hydrogen concentration in steel, A is a proportionality constant, R the gas constant, T the absolute temperature and $E_a$ the activation energy of a trap. Deriving the Equation (12.4), the following expression is obtained:

$$\begin{aligned} \frac{d\dot{z}}{dt}=-A\dot{z}e^{-\frac{E_a}{RT}}+A\left( 1-z\right) e^{-\frac{E_a}{RT}}\frac{E_a}{RT^2}\ \frac{dT}{dt} \end{aligned}$$

(12.5)

imposing $\frac{d\dot{z}}{dt}=0$, the maximum of the function ($\frac{dT}{dt}=\phi$) can be determined.

$$\begin{aligned} -A\dot{z}e^{-\frac{E_a}{RT_p}}+\dot{z}\frac{E_a}{RT^2_p}\phi =0\ {{{\mathop {\Longrightarrow }\limits ^{\left( T_p=\mathrm{Peak\ Temperature}\right) }}}}-\frac{E_a}{RT_p}=\ln \left[ \left( \frac{E_a}{ART^2_p}\right) \phi \right] \ \end{aligned}$$

(12.6)

In a diagram where $\left( {\frac{1}{T_p};\ \ln \left[ \left( \frac{\phi }{T^2_p}\right) \right] \ }\right)$ were plotted, experimental points settle on a straight line and the binding energy can be estimated from the slope of that line.

12.3.3 HIC Tests

HIC is an extremely dangerous and unpredictable phenomenon, related to corrosion phenomena and consequent hydrogen absorption in the metal lattice. There are a lot of factors influencing HIC phenomenon like environment, microstructure, shape of the grain boundaries, number and type of inclusions [14].

That is why, a series of HIC nucleation tests were carried out, in order to determine the critical hydrogen concentration able to give rise to crack initiation. Samples were electrochemically charged by varying electrolytic solution and current density to introduce a different amount of hydrogen in the metal lattice, then they were cut into two parts. A half was observed under light microscope to check the presence of cracks, the other half was used for hydrogen measurement by means of thermal desorption analysis. Figure 12.2 shows an example of cracked material in a hydrogen-induced cracking test.

Fig. 12.2

API X60 sample, hydrogen crack was highlighted with the white arrow

12.3.4 Numerical Analysis

A series of numerical simulations were also performed to determine hydrogen trapping parameters in the bulk steel. Typical diffusion behaviour, explained by Fick’s law [15, 16], is not verified in the presence of hydrogen trapping sites in the metal lattice. McNabb and Forster developed a theory to consider this effect and this was implemented in a simulation model to determine all kinetic parameters [17].

$$\begin{aligned} \left\{ \begin{array}{l} \frac{\partial C}{\partial t}=D\frac{{\partial }^2C}{\partial x^2}-N_r\frac{\partial \nu }{\partial t}-N_i\frac{\partial w}{\partial t} \\ \frac{\partial \nu }{\partial t}=k_r C(1-\nu )-p\nu \\ \frac{\partial w}{\partial t}=k_iCt(1-w) \\ \end{array} \right. \end{aligned}$$

(12.7)

where C and D are the hydrogen concentration and the diffusion coefficient, respectively; $N_r$ and $N_i$ are the reversible and irreversible traps concentration, $\nu$ is the occupied reversible trap fraction, w the irreversible one; $k_r$ is the reversible trapping rate, $k_i$ the irreversible one and p is the release amount for reversible traps.

Solve Eq. 12.7, combined with the corresponding initial and boundary conditions, allows to determine the outlet hydrogen flux defined as $J(t)=-D{\partial }_xC$. The initial conditions correspond to reversible and irreversible traps initially empty and material free of diffusible hydrogen. Boundary conditions consist of a constant concentration $C_0$ on the face of the specimen in contact with the electrolyte solution and zero concentration on the opposite face. The mathematical formulation is reported in Eq. (12.8) and Eq. (12.9), respectively, where a is the sample thickness.

$$\begin{aligned} C(t=0, x) = 0; \ \ v(t=0, x) = 0;\ \ w(t=0, x) = 0 \end{aligned}$$

(12.8)

$$\begin{aligned} C(t, x=0) = C_0; \ \ C(t, x=a) = 0 \end{aligned}$$

(12.9)

The hydrogen outlet flux is a function of time and seven other parameters $C_0$, D, $k_r$, $N_r$, $N_i$, $k_i$, p. The iron lattice diffusion coefficient is constant at room temperature and equivalent to 7$\cdot 10^{-9}$ m$^2$/s [18]. $C_0$ can be easily obtained from the hydrogen flow during steady state. The $k_r/p$ ratio can be determined from the knowledge of the lattice number of sites per unit of volume $N_L$ and from the activation energy of traps $E_a$ known by the TPD technique using Eq. (12.10) [19].

$$\begin{aligned} \frac{ k_r}{p}=\frac{1}{N_L}\exp \Big (\frac{E_ a}{RT}\Big ) \end{aligned}$$

(12.10)

The remaining parameters were determined by minimizing the cost function f defined in Eq. 12.8.

$$\begin{aligned} F = \min [J_\mathrm{simulated} - J_\mathrm{experimental}]^2 \end{aligned}$$

(12.11)

12.4 Results

12.4.1 Permeation Tests

Permeation experiments were first executed on Iron Armco. Samples were tested with Devanathan method, with and without Pd coating, besides HELIOS 2 technique on bare specimens. Results were presented in Table 12.3.

Table 12.3 Permeations test results obtained with different detecting techniques on Fe-Armco

Note that Palladium coating on the reading side is widely shown and demonstrated in the international literature to be a fundamental feature to get significant results [11], and this effect is clearly confirmed in Fig. 12.3.

Fig. 12.3

Comparison of exiting hydrogen flux in three different test conditions and for various sample thicknesses

12.4.1.1 API X60 Steel

Helios test results as a function of temperature were shown in Fig. 12.4. From these tests, the binding energy of the traps [20] was extrapolated and resulted equal to 18 kJ/mol.

Fig. 12.4

Permeation tests at different temperatures

Figure 12.5 indicated the stationary hydrogen flux determined during experimental results as a function of sample thickness.

Fig. 12.5

Hydrogen flux varying sample’s thickness

Another test performed has been the permeation in crack condition (cracking conditions evaluated from HIC studies): so the conditions in terms of solution and current density able to generate microvoids due to the hydrogen interaction with metal lattice.

The permeation curve through the metal wall of the cracking case was reported in Fig. 12.6.

Fig. 12.6

Permeation in crack condition

After reaching the maximum (point 1 in Fig. 12.6), the hydrogen signal decreases as an effect of crack nucleation: hydrogen accumulates in the cracks, while the amount exiting from the sheet decreases (point 2); when the new traps are filled the permeation flux starts to increase again and the signal goes up (point 3). Once determined the hydrogen concentration in steel, correspondent to the maximum peak of Fig. 12.6 (5.1 ppmw), and with reference to Fig. 12.7, the hydrogen accumulation next to the crack can be calculated. Considering the theoretical profile of hydrogen concentration along with the specimen thickness during permeation, [21] (Fig. 12.7(a)), the relationship between thickness and hydrogen concentration can be assumed linear:

Fig. 12.7

a Hydrogen concentration profile and b crack and its distance from sample surface

$$\begin{aligned} C_{\mathrm{H}}=C_{\mathrm{H,0}}\left( 1-\frac{x}{L}\right) \end{aligned}$$

(12.12)

where $C_\mathrm{H,0}$ is the hydrogen concentration on the input surface, 5.1 ppmw (found from the HIC tests, where the sample was saturated, so the surface concentration is equal to the core one); x is the distance shown in Fig. 12.7(b) and L the sample thickness (2 mm). Given these considerations, $C_\mathrm{H}$ was estimated equal to 4.6 ppmw, and the result was in accordance with HIC results, shown in the following paragraphs.

Figure 12.8 shows the permeation curve of 2 mm thickness sample returned by HELIOS 2. The outcoming hydrogen flux, measured by the device, is represented as a function of time. Signal steady state is equal to 9.4$\cdot 10^{-6}$ mol/s m$^2$ and the curve was also used to identify the diffusion parameters through the previously exposed numerical model.

Fig. 12.8

Permeation curve obtained using HELIOS 2 on a 2mm thick sample

12.4.2 TPD Tests

Different heating rates have been set for TPD analysis: 2, 5, 10 and 20 $^\circ$C/min. The results were plotted in Fig. 12.9, according to the model of Choo et Lee [13].

Fig. 12.9

TPD test results

According to paragraph 12.3.1, starting from the slope of regression line, the binding energy can be easily estimated and it was equal to 24 kJ/mol.

12.4.3 HIC Tests

HIC tests were carried out on cylindrical samples, after electrochemical hydrogen charging. Coupons were both observed by metallographic analysis to check the presence of cracks and in parallel the hydrogen content was determined by TDA.

The critical hydrogen concentration was estimated to be equal to 4.5 ppmw: this is the hydrogen amount able to crack the material without external load or applied stresses (Fig. 12.10).

Fig. 12.10

Results of HIC tests on API X60 steel

12.4.4 Numerical Analysis

Starting from experimental data on API X60 steel with thickness 2 mm, a series of studies were carried out by numerical analysis. Figure 12.11 shows the comparison between the permeation experimental curve and the simulated one. The model allowed to calculate the occupancy of reversible traps. Figure 12.12 shows instead the simulated percentage occupancy of reversible traps as a function of the spatial coordinate for different permeation times. As can be seen, the occupancy of reversible traps remains small over time. The fractional occupation of the trap much smaller than one during the permeation test. Table 12.4 shows the diffusion and trapping parameters obtained by the simulation.

Fig. 12.11

Permeation curve. Comparison with the simulated one

Fig. 12.12

Simulated percentage occupancy of reversible traps at different test times

Table 12.4 Diffusion and trapping parameters identified by the permeation test

Equation (12.13) is a kinematic expression of Oriani’s effective diffusion coefficient derived from the local equilibrium between atomic populations in the normal lattice and traps when the occupancy of trapping sites is very low [22].

$$\begin{aligned} D_\mathrm{eff}=D{\left( 1+N_r\frac{k_r}{p}\right) }^{-1} \end{aligned}$$

(12.13)

The reversible trap concentration calculated using the Oriani model leads to a value equal to 210 mol/m$^3$. This value is of the same order of magnitude as that identified by the numerical model.

12.5 Discussion

Pd coating is typically applied for Devanathan–Stachurcki permeation tests, it avoids the recombination of hydrogen atoms into molecular hydrogen before their oxidation in H$^+$ (electron release), with consequent reduction of detected current in the reading semi-cell.

In this case, the hydrogen flux exiting from the sample surface can be calculated as follows:

$$\begin{aligned} J_\mathrm{De}=\frac{i}{F} \end{aligned}$$

(12.14)

where i is the current density and F the Faraday’s constant. Considering the results of Table 12.3, it can be noticed that without the Pd coating the underestimation of hydrogen flux and concentration is not negligible.

HELIOS 2 working principle is different and there is no need for Pd surface deposition since hydrogen escaping out of the metallic sample after recombination into molecular H$_2$ reacts with a solid-state gas sensor. This case shows the variation of sensor resistivity which returns an electrical signal in volts. Translating the signal into a hydrogen concentration and flux, the comparison with Devanathan method was possible and is reported in Fig. 12.3i

The flux through HELIOS 2 probe was calculated after a calibration of the instrument, evaluating molar amount per second in stationary conditions:

$$\begin{aligned} J_\mathrm{He}=\frac{n}{A} \end{aligned}$$

(12.15)

with A is the sample area in contact with the probe. HIC tests highlighted a good resistance to hydrogen damage of API X60 steel. Even though the quality of steel manufacturing can play a crucial role, a critical value near 4.5 ppmw was found. The result was also validated by the permeation test in severe environment, where the hydrogen flux was registered, and crack nucleation was identified through the sensor signal progress.

12.6 Conclusion

Devanathan-Stachurski method can be used for the calculation of hydrogen diffusion coefficient, but the Palladium coating deposition is needed for a reliable estimation of hydrogen flux coming out of the sample, especially for high values of hydrogen flux. This pointed out the elevated precision of the double cell for laboratory investigation but also the impossibility to apply the method to a real pipeline.

HELIOS technique, based on a solid-state gas sensor, is demonstrated to be very practical. Thanks to the high sensitivity device it can be applied directly to the pipe wall. The laboratory investigation carried out in the present study, was aimed at approaching and validating the first step toward online monitoring.

The measurement carried out with HELIOS II of the hydrogen outlet flux was interpreted through a hydrogen diffusion and trapping numerical model providing results in line with the classical theory.

However, the realization of permeation tests with Devanathan-Stachurski on Pd coated specimens, can be successfully used for calibration and validation alternative methods.

Through the TPD and HIC tests, it has been possible a complete characterization of the material in terms of hydrogen trapping and binding energy [23] and the critical hydrogen concentration giving rise to cracks in the material.

Furthermore, the validation of new sensors and hydrogen evaluation suitable for the industrial application have been carried out and achieved.