Search for articles:
Y.H. Fan
, J.Q. Wang
Corresponding authors:
Received: 2019-01-19
Revised: 2019-03-10
Accepted: 2019-03-25
Online: 2019-10-05
Copyright: 2019 Editorial board of Journal of Materials Science & Technology Copyright reserved, Editorial board of Journal of Materials Science & Technology
More
Abstract
The effect of grain size (in the range from 4 μm to 12 μm) on the hydrogen embrittlement (HE) of 304 austenitic stainless steel (ASS) was studied. HE susceptibility result shows that HE resistance increases with grain refinement. Electron backscattered diffraction kernel average misorientation (EBSD-KAM) mapping shows that the strain localization can be mitigated by grain refinement. Hence, strain localization sites which act as highways for hydrogen diffusion and preferred crack initiation sites can be reduced along with grain refinement, leading to a high HE resistance. Meanwhile, grain size shows no influence on the strain induced martensite (SIM) transformation during the hydrogen charging slow strain tensile test (SSRT). Hence, the SIM formed during hydrogen charging SSRT is not responsible for the different HE resistance of 304 ASSs with various grain sizes. Hydrogen diffusion is supposed to be controlled by a competition between short-circuit diffusion along random grain boundary (RGB) and hydrogen trapping at dislocations, leading to a maximum hydrogen diffusion coefficient in the 304 ASS with an average grain size of 8 μm.
Keywords：
Austenitic stainless steels (ASSs), especially stable ASSs in which martensite transformation can hardly occur during deformation due to their high austenite mechanical stability, are usually used for hydrogen components because of their high hydrogen embrittlement (HE) resistance [[1], [2], [3]]. However, their low strength and high nickel content make the components bulky and costly. Reducing the nickel content in ASS can lower the material cost. Nevertheless, the mechanical stability of austenite may decrease, which is likely accompanied by the decrease in nickel content, making the ASS metastable and HE sensitive [[3], [4], [5], [6]]. Meanwhile, the relative low strength of metastable ASS also restricts its application. Prestrain treatment to introduce strain induced martensite (SIM) is a feasible way to improve the strength of metastable ASSs [7] but its effect on HE resistance is still controversial. Zhang et al. [8] found that prestrain treatment improved the HE resistance of 304 ASS. They explained that the boundary of dynamic strain induced martensite (SIM, formed during deformation after prestrain) and austenite acted as a preferred crack initiation site after hydrogen charging. The prior SIM introduced during the prestrain treatment suppressed the formation of dynamic SIM, contributing to an increase in HE resistance. Similar results have been also reported by Perng et al. [9], who found that the threshold stress intensity of 301 ASS in hydrogen gas increased with the increase in the prior SIM. However, deleterious effect of prestrain treatment on HE resistance was also reported in literatures. For example, Wang et al. [10] found that prestrain treatment decreased the HE resistance of 304 L ASS. They explained that the prior SIM acts as a rapid diffusion path, accelerating the entering of hydrogen into alloy.
Grain refinement is another way to improve the strength of materials according to the well-known Hall‒Petch theory [11,12]. Meanwhile, it has been widely recognized to have a positive effect on the HE resistance of ASSs [2,[13], [14], [15], [16]]. Arnaud et al. found that grain refinement increased the HE resistance of metastable Fe-16Cr-10Ni ASS. They explained that grain refinement can relieve the hydrogen-enhanced strain localization, leading to a uniform strain distribution and high HE resistance [2]. Mine et al. [13] successfully produced ultrafine 304 ASSs with grain size in the range from 0.1 μm to 0.5 μm by using high-pressure-torsion (HPT) treatment. They found that specimen with grain size larger than 0.4 μm showed a high HE resistance and a good combination of strength and ductility. When the HE of ASSs is studied, it should be noted that SIM fraction can have a great influence on the HE resistance of ASSs since hydrogen is thought to accumulate at the boundaries between austenite and SIM, resulting in crack initiation along the phase boundaries [8,17,18]. It is well known that the stacking fault energy (SFE) plays an important role during the SIM transformation, since it controls the formation of shear bands, and the intersections of the shear bands are considered to be the nucleation sites for SIM [19]. Generally, with the decrease of SFE, deformation modes of ASS are successively as follows: dislocation slipping, twinning and martensite transformation [20]. As for the 304 ASS used in this study, it possesses a relative low SFE [21], and the SIM transformation may occur during tensile tests. In the case where SIM fraction varies with the grain size during tensile tests, the effect of grain size on HE resistance can be more complicated. Varma et al. [22] studied the effect of grain size on SIM formation in 304 ASS during room temperature tensile testing and found that SIM fraction decreased with the decrease of grain size. Whereas, Matsuoka et al. reported that the amount of SIM formed during the tensile test of Fe-16%Cr-10%Ni metastable ASS is independent of grain size [23]. However, the effect of grain size on the SIM fraction during hydrogen charging tensile test was seldom quantitatively studied in the previous literatures. Hence, the influences of grain size on the SIM fraction during the hydrogen charged slow strain rate test (SSRT) needs to be considered when we explore how grain size affects the HE resistance of metastable ASSs.
It is well known that grain boundaries have a crucial impact on hydrogen diffusion. Generally, random grain boundary (RGB) is considered to be a short-circuit diffusion path, which can accelerate hydrogen diffusion [[24], [25], [26]]. For example, Brass and Chanfreau found that hydrogen diffusion coefficient of polycrystalline nickel increases when the grain size decreases from 150 μm to 25 μm [25]. However, negative effect of RGBs on hydrogen diffusion has also been reported. For example, Yao and Cahoon [27] stated that RGBs as hydrogen trapping sites can trap hydrogen and retard hydrogen diffusion in nickel. Hydrogen diffusion can be virtually stopped at RGB when the hydrogen concentration is extremely low. Grain refinement can lead to an increase in RGBs, which should have an influence on hydrogen permeation. Consequently, to explore the effect of grain refinement on the HE resistance, it is also necessary to study the effect of grain size on the hydrogen diffusion behavior.
In this paper, cold rolling and subsequent annealing were used to produce metastable 304 ASSs with different grain size. The HE resistance was evaluated through SSRT in the air and hydrogen environment. The fracture surface features were observed by scanning electron microscopy (SEM). Meanwhile, electrochemical permeation (EP) tests were used to evaluate the hydrogen diffusion coefficient. Electron backscattered diffraction kernel average misorientation (EBSD-KAM) mapping was used to characterize strain distribution. X-ray diffraction (XRD) was used to assess the volume fraction of SIM. Finally, the effect of grain refinement on the HE resistance of metastable 304 ASS can be explored.
Commercial AISI 304 was used in the present study, and the chemical compositions are given in Table 1. After solution treatment at 1050 ℃ for 1 h, the 6 mm thick plates were cold-rolled to 3 mm thick plates. The materials were subsequently annealed at 800 ℃ for 10 min (800A), 900 ℃ for 10 min (900A) and 950 ℃ for 30 min (950A). For XRD phase analysis of the annealed materials, the specimen surfaces were finally polished in alumina suspension (0.02 μm) to remove mechanically damaged layers.
Table 1 Chemical compositions (wt%) of the investigated austenitic stainless steel.
| C | Cr | Ni | Si | Mn | P | S | Fe |
|---|---|---|---|---|---|---|---|
| 0.042 | 18.87 | 9.33 | 0.27 | 1.6 | 0.026 | 0.002 | Bal. |
Specimens for optical microscopy (OM) observation were electrolytically (1.2 V) etched in 65% nitric acid at 25 °C for 120 s. Then the grain size was determined based on the liner intercept method. XRD analysis was carried out using CuKα radiation at 50 kV. The volume fraction of SIM for specimens after tensile deformation was evaluated using the following equations [28],
where Iihkl is the integrated intensity for (hkl) plane of i-phase, Vi is the volume fraction of i-phase, K is the instrument factor, Rhkl is the material scattering factor, ν is the volume of unit cell, Fhkl is the structure factor of (hkl) plane, p is the multiplicity factor, e-2M is the temperature factor and K/(2μ) is constant in a given XRD scan.
HE susceptibility was evaluated using SSRT at a strain rate of 2 × 10-5 s-1. SSRT tests were performed in air and under electrochemical hydrogen charging condition according to ASTM G129-00. Electrochemical hydrogen charging was carried out in 1 M NaOH solution containing 1 g L-1 thiourea with a constant cathodic current density of 10 mA cm-2. Plate-type specimens with size of 1 mm × 2 mm × 18 mm were used in this study, and all specimens were abraded along the tensile direction by using 2000 grit abrasive papers before tensile test. The relative plastic loss, Iδ, is defined as the HE susceptibility [29],
Iδ=$\frac{δ_{sol}}{δ_{air}}$×100% (4)
where δsol and δair are the elongation in solution and in air, respectively.
To further explore the effect of grain refinement on the HE resistance, SSRT under hydrogen charging condition was terminated at 30% strain for 800A, 900A and 950A samples. Then the SIM fraction and strain distribution of the 30% strained samples were analysed by XRD and EBSD-KAM mapping, respectively. To exclude the influence of cold rolling on the strain distribution analysis of 30% strained samples, the as-annealed samples were also analysed by EBSD-KAM mapping. The EBSD scanning was performed with a step size of 0.2 μm at a voltage of 20 kV. EBSD-KAM mappings were calculated from the grain orientation mappings with a threshold misorientation of 5° by using the TSL OIM analysis 5 software.
EP tests were carried out using the Devanathan-Stachurski two-component permeation cell. Specimens used for EP tests were abraded by using 2000 grit abrasive papers to a final thickness of 500 ± 20 μm. Prior to hydrogen charging, specimens were electrochemically coated on the detection side with a thin nickel layer. Then 0.2 M NaOH solution was introduced in the dection component. A static potential of 200 mV vs saturated calomel electrode (SCE) was applied at the detection side to remove the original hydrogen existed in the specimen and the current density was recorded simultaneously. When the residual current density decreased to a few nA cm-2, 0.2 M NaOH solution was introduced in the charging component and a cathodic current density of 10 mA cm-2 was applied for hydrogen charging. Hydrogen passing through the membrance was recorded at the detection side during the whole charging procedure. And the effective diffusion coefficient of hydrogen Deff can be determined by the following equation [30]:
Deff=$\frac{L^{2}}{15.3t_{b}}$ (5)
where L is the sample thickness, tb is the “breakthrough time”, which is usually defined as the time when the transient current density equals 10% of the steady-state current density [24].
OM images of 800A, 900A and 950A samples are shown in Fig. 1. All the microstructures exhibit equiaxed grains and do not present any texture. The grain size of 800A, 900A and 950A samples are 4.0 ± 0.9 μm, 8.0 ± 1.3 μm and 12.0 ± 2.1 μm, respectively. Except 800A specimen which possesses a very low intensity of (110)α’ peak, all the materials were shown to be a fully austenitic microstructure according to the XRD results, as indicated in Fig. 2. The very low intensity of (110)α’ peak in 800A specimen is corresponding to less than 5% martensite fraction and can be ignored. The corresponding full width at half maximum (FWHM) of (111)γ peak for 800A, 900A and 950A is 0.26°, 0.24° and 0.22°, respectively. EBSD-KAM mappings of the as-annealed 800A, 900A and 950A samples are shown in Fig. 3. The black lines indicate the RGBs. It can be seen that the grains are equiaxed and have no texture, which is consistent with the OM observations. The strain is not uniformly distributed in the as-annealed samples, and the strain localization exists. Meanwhile, the degree of strain localization decreases with the increase of annealing temperature.
Fig. 1. Metallographic images of (a) 800A, (b) 900A and (c) 950A samples.
Fig. 2. XRD spectra of 800A, 900A and 950A samples using CuKα radiation.
Fig. 3. EBSD-KAM images of the as-annealed (a) 800A, (b) 900A and (c) 950A samples (black lines indicate the RGBs).
Typical EP test results are shown in Fig. 4. According to Eq. (5), the Deff of 800A, 900A and 950A sample are (2.9 ± 0.3)×10-13 m2/s, (7.7 ± 0.5)×10-13 m2/s and (1.2 ± 0.2)×10-13 m2/s, respectively. This indicates that there is no direct correlation between the grain size and Deff, which is quite different from the results reported in the previous studies [24,25,27]. Similar result was reported by Oudriss et al. [24,31], who demonstrated that the Deff increased first and then decreased as the grain size decresed from 100 μm to 10 μm in polycrystalline nickel.
Fig. 4. Typical hydrogen permeation curves of 800A, 900A and 950A samples (tb refers to the breakthrough time defined as the time when the transient current density equals 10% of the steady-state current density).
Fig. 5 shows the typical SSRT engineering stress and strain curves of 800A, 900A and 950A samples tested in air and under electrochemical hydrogen charging. In the uncharged samples, the yield strength increases along with grain refinement whereas the total elongation decreases as the grain size decreases. After hydrogen charging, there is no obvious change of the yield strength for all the samples. However, hydrogen charging causes an elongation loss and premature fracture for each sample, as indicated in Fig. 5. According to Eq. (4), the HE susceptibility of 800A, 900A, and 950A is 5.3%, 15.6% and 19.8%, respectively. This result indicates that HE resistance increases with grain refinement. The fracture surfaces of 800A, 900A, and 950A samples after SSRT are shown in Fig. 6. As seen from Fig. 6(a), (d) and (g), fracture surfaces of all the uncharged samples exhibit dimpled features. As for the fracture surfaces of hydrogen charged samples, quasi cleavage features are observed on the edge of the fracture surfaces while dimpled fracture are still the dominant fracture mode in the centers. Even though all the samples exhibit a combination of quasi cleavage and dimpled fracture feature under hydrogen charging condition, the region sizes of quasi cleavage fracture are quite different for 800A, 900A, and 950A samples. The maximum width of quasi cleavage fracture region (d) for 800A, 900A and 950A samples fractured under hydrogen charging are 50 μm, 192 μm and 285 μm, which shows the same tendency with the HE susceptibility.
Fig. 5. Typical engineering stress‒strain curves of 800A, 900A and 950A samples tested in air and hydrogen environment with a strain rate of 2 × 10-5 s-1.
Fig. 6. SEM images of fracture surfaces of the 800A samples (a, b, c), 900A (d, e, f) and 950A (g, h, i) after slow strain rate test. (a, d, g) samples tested in the air; (b, e, h) the fracture surfaces tested under hydrogen charging (D refers to dimple fracture and QC refers to quasi cleavage fracture); (c, f, i) magnifying SEM images of quasi cleavage fractures shown in (b, e, h).
XRD results of the 30% strained samples are shown in Fig. 7. According to Eqs. (2) and (3), the SIM fraction of 800A, 900A and 950A after 30% strain are 30%, 31% and 29%, respectively, by using (111)γ, (200)γ, (110)α’and (211)α’ peaks, showing no obvious grain size dependence. Similar result has been reported by Yoshikazu et al. [23], who studied the effect of grain refinement on the thermal and mechanical stability of austenite without hydrogen charigng. EBSD-KAM mappings of the 30% strained samples are presented in Fig. 8. It can be seen that the grains are elongated along the tensile direction, and the yellow region which represents a severe strain localization becomes more remarkable as the grain size increases.
Fig. 7. XRD results of 800A, 900A and 950A samples after 30% strain under hydrogen charging using CuKα radiation.
Fig. 8. EBSD-KAM images of (a) 800A, (b) 900A and (c) 950A samples after 30% strain under hydrogen charging (black lines indicate the RGBs).
304 ASSs with different grain sizes (800A: 4 μm, 900A: 8 μm, 950A: 12 μm) have been successfully produced through cold rolling and subsequent annealing treatments. HE susceptibility results indicate that grain refinement increases the HE resistance of 304 ASS, as shown in Fig. 5. According to the previous studies [2,22,23], grain size may have an influence on both the SIM transformation and the hydrogen enhanced strain localization, which subsquently affects the HE reistance of metastable ASSs. XRD results (Fig. 7) show that SIM fraction is nearly the same for the 800A, 900A and 950A materials after 30% strain under hydrogen charging condition. This result indicates that grain size (in the range from 4 μm to 12 μm) has no effect on the mechanical stability of austenite in 304 ASS. This is consistent with the result reported by Matsuoka et al [23]. The reasons are as follows. It is known that martensitic transformation can introduce anisotropic transformation strain into materials, and then the anisotropic transformation strain can suppress the subsequent martensitic transformation if it cannot be released. Generally, the martensite formed in metastable ASS has a Kurdjumov-Sachs (K-S) orientation relationship with the original austenite and there are 24 variants derived from cubic systems symmetry [32]. For the athermal martensitic transformation, the anisotropic transformation strain can be minimized through the multi-variant martensitic transformation, and the number of martensite variants decreases as grain size decreases due to the spatial restriction effect in the small grain. Hence, the martensitic transformation induced strain energy, which can suppress the subsequent martensitic transformation is much higher in the small grain than that in the large grain, leading to a high thermal stability of austenite with small grain. As for the SIM transformation, some specific martensite variants are selected during the SIM transformation so that the anisotropic transformation strain can be released by the tensile strain. Consequently, the multi-variant transformation is no longer necessary for the release of anisotropic transformation strain, leading to the grain size independence of mechanical stability of austenite. Hence, SIM formed during hydrogen charged SSRT in 800A, 900A and 950A marterials is not responsible for their different HE resistance. EBSD-KAM mappings of the as-annealed samples show that the strain localization decreases with the increase in grain size. However, after 30% strain, the strain localization becomes more severe with the increase in grain size, as shown in Fig. 8. Therefore, it can be conclued that the strain localization introduced during the hydrogen charged SSRT can be relieved by grain refinement.
When the HE of materials is studied, hydrogen diffusion should also be considered as it influences the hydrogen content entering into materials, and the hydrogen concentration can play an important role in the HE of materials [33,34]. EP tests manifest that grain size is not the only factor that affects the Deff, as shown in Fig. 4. Compared with 950A material, 900A and 800A materials with smaller grain sizes possess relatively larger Deff. This is mainly due to that RGB fraction increases along with grain refinement, and RGBs as amorphous structures can provide a large excess of free volume for hydrogen diffusion [24]. However, the maximum Deff is found in 900A instead of 800A. It is well known that dislocations acting as hydrogen trapping sites can slow down the hydrogen diffusion in metals [[35], [36], [37]], and that is why hydrogen diffusion coefficient was found to decrease with grain refinement in Ref. [27]. For the 304 ASSs used in this work, XRD results of the as-annealed samples show that the FWHM of 111γ increases with grain refinement. EBSD-KAM mappings of the as-annealed samples indicate that strain localiation increases slightly as grain size decreases. Both the broadening of FWHM [38] and the increase in microstrain [39] are considered to be related to an increase in dislocation density. Meanwhile, the geometrically necessary dislocation (GND) density caused by the misorientation between neighboring grains is also reported to increase with grain refinement [24]. Hence, according to the above analysis, it can be inferred that dislocation density increases with the decrease in grain size. Hence, a competition between short-circuit diffusion along RGBs and hydrogen trapping at dislocations may occur, leading to a largest Deff in 900A material with moderate grain size instead of 800A with the smallest grain size.
It is also noted that hydrogen causes a transition of fracture mode from dimple fracture to a mixture fracture of quasi-cleavage and dimple for all the materials, as indicated in Fig. 6(b), (e) and (h). Such features are commonly found in the hydrogen induced fracture of ferritic steels [[40], [41], [42]] and austentic steels [[43], [44], [45]]. For ASSs, it is generally accepted that hydrogen charging can cause a slip localization, which promotes microcrack or void nucleation. Furthermore, hydrogen segregation at the boundaries of SIM and austenite (due to the different solubility and diffusion efficient of hydrogen) can further enhance the slip localization, resulting in crack initiation at the SIM/austenite boundaries [8,17,18]. Consequently, the microcrack or void can coalesce along the slip planes accompanied by hydrogen-enhanced dislocation movement, causing a premature fracture with a quasi-cleavage feature [40], as shown in Fig. 6(c), (f) and (i). However, the d is quite different for the three materials. The d increase as grain size increases, showing the same tendency with the HE susceptibility.
The quasi-cleavage fracture is caused by the hydrogen charging, and the d can be understood as the width of hydrogen affected zone. For 800A, 900A and 950A samples, the enlongations under hydrogen charged SSRT are nearly the same. The hydrogen diffusion coefficient calculated from EP test is largest in the 900A, followed by the 800A and 950A. Meanwhile, the maximum diffusion distance of hydrogen at a given time (t) is $\sqrt{2D_{dff}t}$ [46]. Hence, hydrogen is supposed to diffuse farthest in the 900A sample during the hydrogen charged SSRT, leading to a maximum hydrogen affected zone in 900A sample. However, fracture surfaces observation shows that the d is largest in the 950A sample instead of the 900A sample with a largest Deff. The reasons are as follows: During the hydrogen charged SSRT, hydrogen can move with dislocations, and then the hydrogen diffusion can be accelerated by dislocations movement, leading to a diffusion rate appreciably higher than that for lattice diffusion [47]. As seen from the EBSD-KAM mapping of 30% strained samples (Fig. 8), strain localization becomes more remarkable as grain size increases, and these strain localization sites can act as highways for hydrogen diffusion due to high dislocation activities at these sites. Thus, hydrogen diffusion is significantly enhanced by the strain localization in the 950A sample, leading to a maximum d in 950A sample. Meanwhile, it is noted that a ciritical hydrogen concentration is needed for the hydrogen assisted cracking [48,49]. As for the hydrogen induced transgranular fracture in 800A, 900A and 950A samples, the RGBs are considered to hinder crack propagation [[50], [51], [52]]. Hence, the critical hydrogen concentration for hydrogen assisted cracking increases with grain refinement, leading to a smallest hydrogen affected zone in 800A sample.
According to the above analysis, the effect of grain size (in the range from 4 to 12 μm) on the HE resistance of 304 ASS can be disclosed as follows. Hydrogen enhanced stain localization sites, which act as highways for hydrogen diffusion can be reduced with the grain refinement, contributing to a relatively low hydrogen concentration and uniform deformation. Hence, HE susceptibility decreases along with grain refinement. Meanwhile, grain refinement shows no influence on the SIM transformation during the hydrogen charged SSRT. Hence, the SIM formed in 800A, 900A and 950A materials is considered to be not responsible for their different HE resistance.
(1) HE susceptibility decreases with grain refinement, and hydrogen causes a transition of fracture mode from dimple fracture to a mixture fracture of quasi-cleavage and dimple for 800A, 900A and 950A samples.
(2) Grain size (in the range from 4 to 12 μm) has no influence on the SIM transformation during the hydrogen charged SSRT. However, hydrogen enhanced strain localization can be mitigated by the grain refinement.
(3) The effective hydrogen diffusion coefficient of 304 ASS increases as grain size decreases from 12 μm to 8 μm, and then it decreases as the grain size is further reduced to 4 μm. This is mainly due to a competition between short-circuit diffusion along RGBs and hydrogen trapping at dislocations.
This work is financially supported by the National Natural Science Foundation of China (No. U1608257).
WeChat
/
| 〈 |
|
〉 |
