Evaluation of Hydrogen-induced Cracking Behavior in Duplex Stainless Steel by Numerical Simulation of Stress and Diffusible Hydrogen Distribution at the Microstructural Scale

Abstract

Duplex stainless steels possess ferrite and austenite microstructures, which exhibit different mechanical properties. The strength level and hydrogen diffusion constant of the phases are different; therefore, it is expected that the microscopic stress and hydrogen concentration distribution are inhomogeneous. Assuming that hydrogen-induced cracking occurs at locally stress-concentrated and hydrogen-accumulated locations, it is important to consider the influence of the microstructure in the evaluation of hydrogen-induced cracking. In order to observe crack locations at the microstructural scale, a slow strain rate test of the hydrogen-charged specimen was performed and the cross-section of the specimen was observed following the test. Hydrogen-induced cracks were mainly observed in the ferrite phase. A numerical simulation was performed to determine the contribution of the stress and hydrogen concentration distribution to the initiation of hydrogen-induced cracks. A microstructure-based finite element model consisting of ferrite and austenite phases was designed based on the micrograph of the duplex stainless steel used. The stress–strain curves of the ferrite and austenite phases were used and macroscopic tension was applied to calculate the microscopic stress distribution. The microscopic distribution of hydrogen concentration was calculated by incorporating the stress distribution into the hydrogen diffusion simulation as one of the driving forces. From the simulation results, the stress concentration and hydrogen accumulation occurred at the ferrite phase or at the ferrite/austenite boundary. This tendency corresponds closely to the experimentally observed results; therefore, the above approach can be applied to the evaluation of hydrogen-induced cracking at the microstructural scale.

1. Introduction

Duplex stainless steel has been increasingly used in recent years, particularly in severely corrosive environments, due to its superior strength and corrosion resistance. Although duplex steel is often used under cathodic protection, numerous problems of hydrogen cracking have been reported.^{1,2,3,4,5,6,7,8)} Hydrogen solubility, hydrogen diffusion coefficient, stress/strain distribution,⁹⁾ and the corresponding hydrogen diffusion behavior are considered to be strongly influenced^10,11) by the heterogeneous microstructure consisting of two phases with different mechanical properties. The factors influencing the behavior of diffusible hydrogen in duplex stainless steels are: the phase ratio of the ferrite to the austenite phase,^12,13,14) the grain size,¹⁵⁾ the difference in steel microstructure with respect to the hydrogen penetration direction,^16,17,18,19) the difference in the hydrogen diffusion path due to the difference in microstructure,^20,21,22,23) and the effect of prestrain.^24,25) The distribution of stress and diffusible hydrogen concentration at the microstructure scale should be assessed based on their relatedness to the occurrence of hydrogen cracking. However, due to the challenges associated with the direct measurement of the stress/strain distribution and hydrogen concentration at the microstructure scale, the relationship between the aforementioned parameters has not been adequately elucidated. Therefore, in this study, the influence of the microstructure on the hydrogen-induced cracking characteristics at the microstructure scale was primarily investigated using a numerical simulation method.

Herein, specimens of the base and weld metals were prepared, and a slow strain rate tensile (SSRT) test was performed using specimens that were electrochemically charged with hydrogen. The relationship between the initiation site of hydrogen-induced cracking in the duplex stainless steel and the microstructure was examined by observing the fracture surface and cross-section of the specimen. Additionally, a microstructural finite element model was generated to obtain the distributions of stress and diffusible hydrogen concentration at the microstructure scale. Elasto-plastic and hydrogen diffusion analyses, which consider the stress gradient as the main driving force, were performed. Then, the relationship between the location of hydrogen-induced cracking and the microstructure was examined based on the resultant stress distribution and diffusible hydrogen concentration distribution. This paper focuses on the effect of the microstructure on the hydrogen cracking behavior in the base metal. A similar study was conducted for the weld metal, and the results are described in another paper.²⁶⁾

2. Experimental and Numerical Simulation Methods

2.1. Slow Strain Rate Tensile Test after Hydrogen Charging

Hydrogen-charged tensile test specimens were fractured using slow strain rate tensile testing to investigate the effect of microstructure on the hydrogen-induced cracking characteristics of duplex stainless steel. The fracture surface and cross-section were observed after the test.

2.1.1. Test Materials and Hydrogen Charging Conditions

The test material was a rolled plate of 22% Cr duplex stainless steel (UNS No. S31803) with a thickness of 12 mm (hereinafter referred to as the base metal). Table 1 shows the chemical composition of the base metal. Figure 1 shows the microstructure of a cross-section parallel to the rolling direction. A cylindrical tensile test specimen (diameter 3 mm, length 25 mm) was extracted from the base metal.

Table 1. Chemical composition of duplex stainless steel used (mass%).

C	Si	Mn	P	S	Ni	Cr	Mo	Co	N
0.015	0.33	0.93	0.024	0.001	5.5	22.6	3.2	0.11	0.17

Fig. 1.

Microstructure of duplex stainless steel used.

The specimen was electrochemically charged with hydrogen. The hydrogen charging conditions were: sulfuric acid (pH 2.5) + 1% NH₄SCN aqueous solution (25°C), current density of 1.0 mA/cm², and a charging period of one and three weeks. As a preliminary experiment, a round bar sample (diameter 3 mm and length 25 mm), which was identical to the parallel portion of the tensile test specimen, was hydrogen charged under the same conditions, and the amount of the charged hydrogen was measured using the temperature-programmed desorption gas analysis method. The measurement was performed from room temperature to 600°C at a heating rate of 100°C/h. The concentrations of the charged diffusible hydrogen within one and three weeks were 86 and 136 ppm, respectively. Based on this result, the round bar tensile test specimen used in SSRT testing was considered to be charged with comparable diffusible hydrogen. In addition to these one- and three-week hydrogen-charged test specimens, SSRT tests were also performed on hydrogen-uncharged test specimen for comparison.

2.1.2. Slow Strain Rate Tensile Test Method

SSRT tests were performed on both the hydrogen-charged and uncharged specimens at a crosshead displacement rate of 0.005 mm/min. The test commenced immediately after hydrogen charging was complete; plating on the specimen surface to prevent the release of diffusible hydrogen was not performed. However, since there was concern that diffusible hydrogen was released from the specimen during SSRT testing, a section of the specimen was cut and the diffusible hydrogen concentration was measured after the test. The measurement was conducted on the test specimen with a charging period of three weeks and the section was selected from the region of the specimen that did not interfere with the fracture surface and cross-sectional observation. As a result, 116 ppm of diffusible hydrogen was retained, even after the specimen underwent SSRT testing with a charging period of three weeks. The SSRT test was performed in the presence of sufficient diffusible hydrogen compared to the preliminary charging test, which had a diffusible hydrogen concentration of 136 ppm. The apparent strain rate in the SSRT test was 3.3 × 10^–6 s^–1, assuming that the parallel portion (25 mm in length) was subjected to the same displacement rate (0.005 mm/min) in the SSRT test. In the numerical simulation described in Section 2.2, this apparent strain rate was applied as the strain rate for the numerical simulation model.

After SSRT testing, the fracture surface and cross-section were observed. Scanning electron microscopy (SEM) was used to observe the fracture surface. The cross-section of the fractured test specimen was observed using optical microscopy on a plane parallel to the longitudinal direction of the specimen. The plane comprised of the rolling and plate thickness directions of the test material. The fractured specimen was embedded in the resin, wet-polished to #2000 emery paper, buffed with diamond paste (1, 3, and 6 μm), and finish-polished with alumina suspension. The microstructure was exposed by electrolytic corrosion with 33% aqueous potassium hydroxide solution.

2.2. Numerical Simulation

2.2.1. Overview of Numerical Simulation Method

The hydrogen cracking behavior of the duplex stainless steel is impacted by the microstructure containing ferrite and austenite phases, which have different strength properties and hydrogen diffusion characteristics. At the microstructure scale, the distribution of stress and diffusible hydrogen concentration is non-uniform, which is expected to be a probable reason for the numerous cracks that were mainly observed in the ferrite phase. Therefore, in this study, numerical analysis was utilized to evaluate the hydrogen cracking behavior based on the distributions of stress and diffusible hydrogen concentration at the microstructure scale. The objective of the numerical simulation is to understand the characteristics of the inhomogeneous distribution of stress and diffusible hydrogen concentration expected to occur at the microstructure scale. The finite element model was generated using the morphology of microstructure, and elasto-plastic and hydrogen diffusion analyses were performed.

2.2.2. Microstructural Finite Element Modeling

The phase map obtained by the electron backscatter diffraction (EBSD) method was used to create the microstructural finite element model. The EBSD measurement was performed with a step size of 0.5 μm after polishing with colloidal silica. Figures 2(a) and 2(b) depict the phase map and the generated finite element model, respectively. Figure 2(c) depicts a magnified section of the model that shows the element and phase distributions. The finite element analysis software OOF2 (version 2.1.12),²⁷⁾ provided by the National Institute of Standards and Technology (NIST), was used to develop the finite element model. This software has the ability to generate a finite element model based on microstructure images and perform various numerical simulations, but only the function for generating a model was used. After generating the model, a general-purpose finite element analysis software Abaqus/Standard (version 2018) was used for elasto-plastic and hydrogen diffusion analyses.

Fig. 2.

Phase map of duplex stainless steel (a) and generated finite element model (b and c). (Online version in color.)

The dimensions of the inset (microstructure photograph) shown in Fig. 2 is 200 μm × 200 μm; this region was divided by a 1 μm × 1 μm two-dimensional four-node quadrilateral element. The two-dimensional four-node plane strain element was used for elasto-plastic analysis, whereas the two-dimensional four-node mass diffusion element was used for hydrogen diffusion analysis. Each element was classified into either ferrite or austenite phases based on the microstructure photograph, and the material properties were set. Phases other than the ferrite and austenite phases as well as the existence of carbide or nitride and impurities were not considered in this study. At the interface between the ferrite and austenite phases, both phases were considered to simply exist adjacent to each other; no special modeling for the boundary was done in this simulation model.

2.2.3. Elasto-plastic and Hydrogen Diffusion Analyses

In the numerical simulation, the elasto-plastic and hydrogen diffusion analyses were performed sequentially. First, the elasto-plastic analysis involving a macroscopic tensile load applied to the created microstructural finite element model was conducted, and the stress distribution at the microstructure scale was simulated. Subsequently, hydrogen diffusion analysis was performed by incorporating the obtained stress gradient as the driving force for diffusion.

Figure 3 shows the stress–strain curves of the ferrite and austenite phases used in the elasto-plastic analysis. These were estimated based on the stress–strain curve obtained during static tensile testing of the duplex stainless steel, assuming that a linear mixture rule holds. The stress–strain curve of duplex stainless steel was determined by Swift’s approximation until uniform elongation and was extrapolated beyond uniform elongation. The area percentage (S_F:S_A) and hardness (HV_F/HV_A) ratios of the ferrite/austenite phases obtained from microstructural observation were 57:43 and 1.1, which were used for the estimation by a linear mixture rule. The stress–strain curves shown in Fig. 3 were set to the elements corresponding to each phase of the microstructural finite element model shown in Figs. 2(b) and 2(c), and a tensile load was applied. Boundary conditions were set to ensure that the upper, lower, left, and right ends of the finite element model would fit straight lines, and forced displacement was applied in the loading direction. From the results of the SSRT test, the hydrogen-charged specimens showed a significant decrease in elongation and that hydrogen cracking at the microstructure scale occurs, even if the macroscopically applied strain does not reach the fracture strain. Therefore, in the numerical simulation, a strain was applied to achieve uniform elongation. The applied strain was 9.8%, which was determined by dividing the crosshead displacement (2.45 mm) at the maximum load by the length (25 mm) of the parallel section of the tensile test specimen in the SSRT test. The strain rate was calculated to be 3.3 × 10^–6 s^–1 using the crosshead displacement rate and the parallel part length. In the elasto-plastic analysis, the strain rate dependency was not taken into consideration and the time in the simulation could be arbitrarily set. However, in this study, since the hydrogen diffusion analysis with time dependence was subsequently performed, the displacement was given in consideration of the strain rate.

Fig. 3.

Stress–strain curve of ferrite and austenite phases used in the simulation. (Online version in color.)

Hydrogen diffusion analysis was performed considering the non-uniform stress distribution at the microstructure scale obtained by the abovementioned elasto-plastic analysis. The hydrogen diffusion analysis was performed based on the following formulation.²⁸⁾ First, the mass conservation law of diffusible hydrogen is expressed by the following equation:   

$${\int }_V\frac{dc}{dt}dV+{\int }_S\mathit{n}\cdot \mathit{J}dS=0$$(1)

where, V is the arbitrary volume region, c is the concentration, t is the time, S is the surface of arbitrary volume region, n is the normal vector of surface S, and J is the concentration flux. Assuming that the driving force for diffusion is the gradient of chemical potential (μ), the concentration flux (J) is expressed by the following equation:   

$$\mathit{J}=-\frac{\mathit{D}c}{RT}\cdot \frac{\partial \mu }{\partial x}$$(2)

where, J is the concentration flux, D is the diffusion coefficient, c is the concentration, R is the gas constant (8.31 J/K·mol), T is the absolute temperature, μ is the chemical potential, and x is the position vector. The chemical potential (μ) is defined by the following equation:   

$$\mu ={\mu }^0+RTln\phi +p\overline{V_H}$$(3)

where, μ⁰ is the standard chemical potential, φ is the normalized concentration (c/s, s: solubility), p is the hydrostatic pressure, $\overline{V_H}$ is the partial molar volume of diffusible hydrogen. Comparable to a similar numerical analysis, a $\overline{V_H}$ of 2.00 × 10³ mm³/mol was used.^20,21,28) This value is for the ferritic iron, but it is not clear whether this value can be applied to the austenite phase. However, in the numerical analysis used in this paper, this value was also used for the austenite phase. From Eqs. (2) and (3),   

$$\mathit{J}=-s\mathit{D}\left(\frac{\partial \varphi }{\partial x}+\phi ln\mathrm{\hspace{0.17em} }\phi \frac{\partial lnT}{\partial x}+\frac{\phi \overline{V_H}}{RT}\frac{\partial p}{\partial x}\right)$$(4)

is obtained. The terms in parentheses on the right side of Eq. (4) indicate that hydrogen diffusion proceeded with a gradient of normalized concentration (φ = c/s), temperature (T), and hydrostatic pressure (p) as a driving force. Since the temperature gradient was not considered in this study, the diffusion equation is expressed by the following equation, and the driving force for hydrogen diffusion is the normalized concentration gradient and the hydrostatic pressure gradient.   

$$\mathit{J}=-s\mathit{D}\left(\frac{\partial \phi }{\partial x}+\frac{\phi \overline{V_H}}{RT}\frac{\partial p}{\partial x}\right).$$(5)

The literature values for duplex stainless steels (Table 2) were used in the hydrogen diffusion analysis for the hydrogen diffusion coefficient and hydrogen solubility of the ferrite and austenite phases.²¹⁾ The initial conditions in the hydrogen diffusion analysis were set so that the local equilibrium state was reached in the numerical simulation model and an equivalent normalized concentration (φ) was set within the model. The normalized concentration (φ) is defined as c/s and the hydrogen solubility (s) differs in the ferrite and austenite phases. Hence, the diffusible hydrogen concentration is different in the ferrite and austenite phases. In the numerical simulation, the initial normalized diffusible hydrogen concentration (φ) of the microstructure model was set to 6.13 × 10^–3 to reproduce the average diffusible hydrogen concentration of the one-week hydrogen-charged specimen (86 ppm). As a result, the initial diffusible hydrogen concentrations in the ferrite and austenite phases were set to 0.2 ppm and 199.2 ppm, respectively. In addition, since the objective is to study the change in the local diffusible hydrogen concentration distribution, the amount of diffusible hydrogen in the entire numerical analysis model was preserved. In other words, the inflow and outflow of diffusible hydrogen were not considered, but the change in distribution within the numerical simulation model will be discussed.

Table 2. Material properties used in hydrogen diffusion simulation.

	Diffusion coefficient, D (m2/s)	Solubility, s (ppm·mm/N1/2)
Ferrite	6.0 × 10−11	0.033
Austenite	1.4 × 10−16	32.51

3. Results and Discussion

3.1. Slow Strain Rate Tensile Test after Hydrogen Charging

3.1.1. Slow Strain Rate Tensile Test

Figure 4 shows the nominal stress–crosshead displacement curve of the base metal obtained by the SSRT test. The elongation decreases as the amount of charged diffusible hydrogen increases. Based on the nominal stress–crosshead displacement curve, the effect of diffusible hydrogen can be confirmed, even in a one-week charged specimen. Therefore, subsequent observations of fracture surface and cross-section were performed on the uncharged and one-week charged (hereinafter referred to as “charged”) specimens.

Fig. 4.

Stress–crosshead displacement curve of uncharged and charged specimens of base metal. (Online version in color.)

3.1.2. Fracture Surface Observation after Slow Strain Rate Tensile Test

The fracture surface of the specimen was observed by SEM after the SSRT test. The uncharged and charged specimens are shown in Figs. 5(a)–5(c) and Figs. 5(d)–5(f), respectively. Figure 5 depicts images of the entire fracture surface and characteristic fracture surface. The duplex stainless steel base metal used exhibits anisotropy due to its rolled microstructure (Fig. 1) and the outline of the fracture surface after the SSRT test is elliptic. The uncharged specimen is more elliptical than the charged specimen, because the former specimen continues deformation for a longer duration until fracture.

Fig. 5.

Fracture surface after SSRT test of uncharged (a, b and c) and one-week charged base metal specimens (d, e and f).

In the uncharged specimen, dimples can be observed over the entire fracture surface (Figs. 5(b) and 5(c)) and the typical ductile fracture occurs. On the other hand, in the charged specimen, there is a region where dimples are confirmed (Fig. 5(e)), but a quasi-cleavage fracture surface also occurs (Fig. 5(f)). Figure 5(d) shows the regions of the dimple-based and quasi-cleavage fracture surfaces. The quasi-cleavage fracture surface was mainly confirmed in the vicinity of the outer edge of the fracture surface and the influence was more prominent near the surface due to the charging of diffusible hydrogen from the surface.

3.1.3. Cross-sectional Observation after Slow Strain Rate Tensile Testing

The ductility of the duplex stainless steel base metal was degraded by the influence of diffusible hydrogen and the fracture morphology was changed. Therefore, the cross-section was observed by focusing on the relationship between the crack initiation site and the microstructure of the fractured specimen.

Figure 6 shows that the microstructure of the uncharged specimen was elongated in the tensile direction and ductile fracture occurred. In addition, no cracks were observed in the area away from the fracture surface. In contrast, Fig. 7 shows that cracks were observed on the surface for the entire length of the parallel section of the charged specimen. Majority of the cracks occurred perpendicular to the longitudinal direction of the specimen, i.e., perpendicular to the loading direction. The maximum principal stress is considered the controlling factor of the crack initiation.

Fig. 6.

Cross-section of uncharged SSRT test specimen.

Fig. 7.

Cross-section of one-week charged SSRT test specimen.

The cross-sectional observation was performed on the fractured specimen; thus, numerous cracks with large openings were found. It is challenging to analyze the effect of microstructure on cracking by considering the cracks with large openings. Therefore, in this study, focus is placed on relatively small cracks, which are clearly influenced by microstructure. In particular, cracks with lengths over several ferrite or austenite phases are examined in detail. These relatively small cracks occur in both the ferrite and austenite phases, but the cracks attributed to the ferrite phase are dominant. There is no cracking in the austenite phase adjacent to the cracked ferrite phase, but cracking is observed in the ferrite phase beyond the adjacent austenite phase, which is possibly due to the difference in the crack initiation properties of the ferrite and austenite phases.

3.2. Numerical Simulation

The simulation results of the stress distribution at the microstructure scale are presented. Figure 8 shows the hydrostatic pressure distribution, which is the driving force of hydrogen diffusion. Figure 9 shows the maximum principal stress distribution, which is considered to be one of the controlling factors for hydrogen cracking. The resultant microstructures at the applied strain of 1.5% are shown in Figs. 8(a) and 9(a), and those at the applied strain of 9.8%, at which the maximum load is applied, are shown in Figs. 8(b) and 9(b). In Figs. 8 and 9, the boundary between the ferrite and austenite phases is indicated by a white line. Since the hydrostatic pressure exhibits a positive value for compression, the tensile stress acts in the region with a large negative value (Fig. 8). Moreover, the distribution tendency for the hydrostatic pressure and maximum principal stress is almost similar (Figs. 8 and 9). The ferrite phase demonstrates large values of hydrostatic pressure and a high maximum principal stress, which is evident from the stress–strain curve used in the numerical simulation. High stress tends to occur in the region of the ferrite phase where the thickness in the plate thickness direction is smaller than that in the surroundings, which indicates that the microstructure morphology also influences the stress distribution.

Fig. 8.

Hydrostatic pressure distribution in base metal microstructure at applied strain (a) 1.5% and (b) 9.8%. (Online version in color.)

Fig. 9.

Maximum principal stress distribution in base metal microstructure at applied strain (a) 1.5% and (b) 9.8%. (Online version in color.)

Figure 10 depicts the hydrogen concentration distribution as a result of the hydrogen diffusion analysis, based on the assumption that the hydrostatic pressure obtained by the above elasto-plastic analysis is the driving force for diffusion. In addition to the initial diffusible hydrogen concentration distribution (Fig. 10(a)), the distributions at the applied strains of 1.5 and 9.8% are shown in Figs. 10(b) and 10(c), respectively. As mentioned above, there is a large difference in the diffusible hydrogen concentration between the two phases compared to the initial conditions due to the difference in solubility between the ferrite and austenite phases. For example, the initial diffusible hydrogen concentration distribution is shown in Fig. 10(a); the austenite phase has a high diffusible hydrogen concentration and is displayed in light gray. For this reason, it was difficult to represent the distributions of both phases so that they can be distinguished in one contour map. First, the contour range was set so that the distribution in the ferrite phase becomes apparent because this phase is considered to be the initiation site of hydrogen-induced cracking. The range was set to display the initial value of the diffusible hydrogen concentration in the center (green) of the legend and the region where the diffusible hydrogen accumulated and increased from the initial state is indicated by warm colors, whereas the decreased region is indicated by cold colors. Figures 10(b) and 10(c) show that the region of high diffusible hydrogen concentration occurs in the ferrite phase; this region corresponds to regions of high tensile stress. This is because the hydrostatic pressure gradient is the driving force of hydrogen diffusion and the diffusible hydrogen ultimately accumulates in the region of high tensile stress.

Fig. 10.

Hydrogen concentration distribution in ferrite phase of base metal microstructure at (a) initial condition, (b) applied strain 1.5%, and (c) 9.8%. (Online version in color.)

Figure 11 shows the diffusible hydrogen concentration distribution in the austenite phase at applied strains of 1.5 and 9.8%. The region indicated by dark gray is the ferrite phase with low diffusible hydrogen concentrations. In the austenite phase, the concentration of diffusible hydrogen typically increases due to diffusion and accumulation, but there is also a region where it decreases. On the other hand, in the ferrite phase, the diffusible hydrogen concentration increased in majority of the regions, as shown in Fig. 10. The diffusible hydrogen not only redistributes within a phase, but also diffuses between phases. Hence, the diffusible hydrogen accumulated at the ferrite/austenite phase boundary, as shown in Fig. 11(b). In the SSRT test conducted in this study, the longitudinal and loading directions of the ferrite and austenite phases were identical, and the load perpendicular to the ferrite/austenite phase boundary was considered relatively small. Consequently, cracks that undoubtedly occurred at the boundary were not observed, but the phase boundary was proposed as the possible crack initiation site.

Fig. 11.

Hydrogen concentration distribution in austenite phase of base metal microstructure at applied strain (a) 1.5% and (b) 9.8%. (Online version in color.)

Based on the numerical analysis results of the stress and diffusible hydrogen concentration distributions at the microstructure scale, the hydrogen cracking characteristics of the duplex stainless steel base metal are considered. First, diffusible hydrogen, which is initially distributed in the equilibrium state at the microstructure scale after hydrogen charging, diffuses and accumulates using the stress distribution at the microstructure scale generated by tensile load as a driving force, thus yielding locally high hydrogen concentration. The diffusible hydrogen accumulates at the ferrite phase where the hydrogen diffusion coefficient is large and the hydrostatic pressure is high. Since the microscopic stress distribution is greatly influenced by the microstructure, the value of the maximum principal stress is also high in the region where the hydrostatic pressure is high and the diffusible hydrogen is accumulated. Hydrogen cracking occurred due to the high maximum principal stress acting on the ferrite phase, which resulted in accumulated diffusible hydrogen. The relatively large number of cracks observed in the ferrite phase corresponds qualitatively to the numerical analysis result. Importantly, it is possible to consider the distributions of stress and diffusible hydrogen concentration for different microstructures using the numerical analysis. In the future, this method, in tandem with more detailed observations, can be used to estimate the critical conditions that induce hydrogen cracking at the microstructure scale.

4. Conclusion

In this paper, the hydrogen-induced cracking behavior of duplex stainless steel containing ferrite and austenite phases with different strength and hydrogen diffusion characteristics were investigated based on the distributions of stress and hydrogen concentration at the microstructure scale. The main results are presented below:

(1) An SSRT test was conducted using hydrogen-charged and uncharged specimens of a duplex stainless steel base metal to study the effect of diffusible hydrogen on cracking. As a result of the cross-sectional observation after SSRT testing, the cracks perceived to be due to the influence of diffusible hydrogen were mainly confirmed in the ferrite phase.

(2) A finite element model of the microstructure was generated and elasto-plastic and hydrogen diffusion analyses performed to investigate the crack initiation site at the microstructure scale based on the distributions of stress and hydrogen concentration. As a result, the diffusible hydrogen accumulates in the ferrite phase and at the ferrite/austenite phase boundary, and high maximum principal stress occurs in the ferrite phase.

(3) Based on the results of (1) and (2), diffusible hydrogen accumulates in the ferrite phase and a high maximum principal stress occurs in the hydrogen-charged duplex stainless steel base metal; thus, a relatively large number of cracks were observed in the ferrite phase.

Acknowledgement

This work was partly supported by JSPS KAKENHI, Grant Numbers 16K05977 and 19K04075.

References

• 1)   P.  Woollin and  W.  Murphy: Proc. Corrosion 2001, NACE, Houston, (2001), Paper No. 01018.
• 2)   S.  Huizinga,  B.  McLoughlin,  I. M.  Hannah,  S. J.  Paterson and  B. N. W.  Snedden: Proc. Corrosion 2006, NACE, Houston, (2006), Paper No. 06145.
• 3)   R. N.  Gunn: Duplex Stainless Steels, Abington Publishing, Cambridge, UK, (1997), 187.
• 4)   R.  Francis,  G.  Byrne and  G. R.  Warburton: Corrosion, 53 (1997), No. 3, 234. https://doi.org/10.5006/1.3280465
• 5)   A. J.  Griffiths,  A.  Turnbull: Proc. Corrosion 96, NACE, Houston, (1996), Paper No. 96262.
• 6)   M.  Aursand,  L. A.  Marken,  G.  Rørvik and  I. M.  Kulbotten: Proc. Corrosion 2013, NACE, Houston, (2013), Paper No. 2490.
• 7)   A. J.  Leonard,  R. N.  Gunn and  T. G.  Gooch: Proc. Stainless Steel World Duplex America 2000 Conf., KCI Publishing, Zutphen, (2000), 345.
• 8)   P.  Woollin and  A.  Gregori: Proc. 23rd Int. Conf. on Offshore Mechanics and Arctic Engineering (OMAE2004), ASME, New York, (2004), OMAE2004-51203, 777. https://doi.org/10.1115/OMAE2004-51203
• 9)   A.  Yousefi and  G.  Itoh: ISIJ Int., 58 (2018), 561. https://doi.org/10.2355/isijinternational.ISIJINT-2017-631
• 10)   M.  Nagumo: ISIJ Int., 52 (2012), 168. https://doi.org/10.2355/isijinternational.52.168
• 11)   H. K. D. H.  Bhadeshia: ISIJ Int., 56 (2016), 24. https://doi.org/10.2355/isijinternational.ISIJINT-2015-430
• 12)   R. F. A.  Jargelius and  F. C.  Gan: Proc. Int. Conf. on Stainless Steels, ISIJ, Tokyo, (1991), 692.
• 13)   A.  Turnbull and  R. B.  Hutchings: Mater. Sci. Eng. A, 177 (1994), 161. https://doi.org/10.1016/0921-5093(94)90488-X
• 14)   R. B.  Hutchings,  A.  Turnbull and  A. T.  May: Scr. Metall. Mater., 25 (1991), 2657. https://doi.org/10.1016/0956-716X(91)90134-M
• 15)   S. L.  Chou and  W. T.  Tsai: Mater. Sci. Eng. A, 270 (1999), 219. https://doi.org/10.1016/S0921-5093(99)00174-4
• 16)   A.  Mikkelsen,  S.  Wastberg,  R.  Johnsen,  B.  Nyhus and  T.  Rogne: Proc. Corrosion 2006, NACE, Houston, (2006), Paper No. 06499.
• 17)   E.  Dabah,  Th.  Kannengiesser,  T.  Mente,  K.  Beyer and  S.  Brauser: Weld. World, 57 (2013), No. 4, 561. https://doi.org/10.1007/s40194-013-0051-5
• 18)   S.  Mukai,  H.  Okamoto,  T.  Kudo and  A.  Ikeda: J. Mater. Energy Syst., 5 (1983), No. 1, 59. https://doi.org/10.1007/BF02833508
• 19)   W. C.  Luu,  P. W.  Liu and  J. K.  Wu: Corros. Sci., 44 (2002), 1783. https://doi.org/10.1016/S0010-938X(01)00143-3
• 20)   V.  Olden,  C.  Thaulow and  R.  Johnsen: Mater. Des., 29 (2008), 1934. https://doi.org/10.1016/j.matdes.2008.04.026
• 21)   V.  Olden,  A.  Saai,  L.  Jemblie and  R.  Johnsen: Int. J. Hydrog. Energy, 39 (2014), 1156. https://doi.org/10.1016/j.ijhydene.2013.10.101
• 22)   T.  Mente and  T.  Bollinghaus: Weld. World, 56 (2012), No. 11–12, 66. https://doi.org/10.1007/BF03321397
• 23)   Z.  Ghorani,  A.  Yousefi and  G.  Itoh: ISIJ Int., 59 (2019), 1901. https://doi.org/10.2355/isijinternational.ISIJINT-2019-121
• 24)   S. S.  Chen,  T. I.  Wu and  J. K.  Wu: J. Mater. Sci., 39 (2004), 67. https://doi.org/10.1023/B:JMSC.0000007729.14528.a8
• 25)   A. M.  Elhoud,  N. C.  Renton and  W. F.  Deans: Int. J. Hydrog. Energy, 35 (2010), 6455. https://doi.org/10.1016/j.ijhydene.2010.03.056
• 26)   G.  Ogita,  K.  Matsumoto,  M.  Mochizuki,  Y.  Mikami and  K.  Ito: submitted to ISIJ Int 61 (2021), 1236. https://doi.org/10.2355/isijinternational.ISIJINT-2020-400
• 27)  National Institute of Standards and Technology (NIST): Finite Element Analysis of Microstructures OOF2 (Version 2.1.12), https://www.ctcms.nist.gov/oof/oof2/, (accessed 2019-09-24).
• 28)   P.  Sofronis and  R. M.  McMeeking: J. Mech. Phys. Solids, 37 (1989), No. 3, 317. https://doi.org/10.1016/0022-5096(89)90002-1