Anisotropy in Hydrogen Embrittlement Resistance of Drawn Pearlitic Steel Investigated by in-situ Microbending Test during Cathodic Hydrogen Charging

Abstract

To investigate causes of superior hydrogen embrittlement resistance of drawn pearlitic steel, notched microcantilevers with different notch orientations with respect to the lamellar interface were fabricated by focused ion beam, and microbending tests were conducted in air and during cathodic hydrogen charging by electrochemical nanoindentation. In air, indentation load monotonically increased with increase in indentation displacement, and no crack appeared for any notch orientations. During hydrogen charging, indentation load declined, and a crack appeared. The load reduction with respect to the displacement was larger, and the crack was deeper for the notch parallel to the lamellar interface than that normal to the lamellar interface. Furthermore, stationary cracks in the microcantilevers were observed by scanning electron microscopy and scanning transmission electron microscopy. For the notch parallel to the lamellar interface, a sharp long crack was identified along the lamellar interface. The crack stopped at the position where the cementite lamellae are disconnected. In lattice images, cementite was identified in one side of the crack, and ferrite in another side of the same crack. On the other hand, for the notch normal to the lamellar interface, a blunt short crack was identified. Thus, it was concluded that the ferrite-cementite interface is a preferential crack path, and hydrogen embrittlement resistance in the direction parallel to the lamellar interface is superior to that normal to the lamellar interface. The present results also indicate that directional lamellar alignment of the drawn pearlitic steel suppresses crack propagation in the radial direction of the drawn wire, improving the hydrogen embrittlement resistance in the drawing direction.

1. Introduction

It is well known that hydrogen embrittlement (HE) resistance of drawn pearlitic steel is superior to that of martensitic steel with the same strength.^1,2) When pearlitic steel is cold drawn, the lamellae align such that their ferrite(α)-cementite(θ) interfaces become parallel to the drawing direction. In the α lamellae, dislocation density increases, and a <110> fiber texture develops; the <110> direction of α aligns parallel to the drawing direction.^3,4) In the θ lamellae, fragmentation occurs, and nanoscale grains are formed.^5,6) It is proposed that the superior HE resistance of the drawn pearlitic steel arises from such a fiber structure which possibly prevents the HE crack from propagating in the radial direction of the drawn wire.^1,2) It is also proposed that the high density of dislocations in the α lamellae, which trap hydrogen, possibly suppresses hydrogen diffusion to crack initiation sites.

However, no direct experimental evidence exists that rationalizes these mechanisms. For example, the HE resistance of drawn pearlitic steel was mainly studied by macroscopic mechanical tests, such as a slow-strain rate test and a constant load test.^{1,2,7,8,9,10,11)} Machining of macroscopic specimens with the tensile axis normal to the drawing direction is difficult because of a limited diameter of the drawn wire. As a result, the HE resistance in the radial direction of the drawn wire is hardly understood although that in the drawing direction was extensively studied. Moreover, preferential propagation path of the HE crack was investigated mainly on the basis of fractography, and not fully understood up to the present. Microscopy observations, especially scanning/transmission electron microscopy (S/TEM) observations, of a stationary HE crack are essential to explore the preferential crack propagation path. Nucleation of the stationary crack by the macroscopic mechanical tests is, however, difficult for high-strength steel because of low strain controllability of the tests; the crack does not stop, and crack initiation and fracture occur in a moment.

The authors evaluated HE resistance of an individual grain boundary in Ni–Cr bialloy by in-situ microbending test of microcantilevers (MCLs) during cathodic hydrogen charging.¹²⁾ The MCLs were fabricated by focused ion beam (FIB), and bent by electrochemical nanoindentation (EC-NI).^13,14,15) The size of the MCL is micrometer scale and comparable to size of a single pearlite colony. Moreover, the indenter of the EC-NI can be finely displaced at a nanometer step. It is very likely that the locality and the excellent strain controllability of the microbending test can be exploited to investigate anisotropy in the HE resistance of the drawn pearlitic steel, and also to explore the preferential crack propagation path by microscopy.

In this context, in the present study, notched-MCLs with different notch orientations with respect to the α-θ interface were fabricated on the drawn pearlitic steel by FIB. Then, the microbending tests were conducted for these MCLs both in air and during cathodic hydrogen charging. Stationary cracks purposely nucleated by the microbending tests were also observed by both scanning electron microscopy (SEM) and S/TEM.

2. Experimental

2.1. Materials

A wire rod with a diameter of 6 mm and a chemical composition in Table 1 was used as a sample. First, a homogeneous pearlitic structure was obtained by the patenting treatment, where the rod was heated at 1223 K for 600 s, and quenched in a lead bath at 873 K. Then, the rod was cold drawn into a wire with a diameter of 4 mm, and drawn pearlitic steel with tensile strength of 1.6 GPa and equivalent strain of 0.81 was obtained. The microbending test was conducted for this as-cold-drawn wire without annealing.

Table 1. Chemical composition of the sample. The unit is mass%.

C	Si	Mn	P	S	Fe
0.94	0.23	0.70	0.010	0.004	bal.

Figure 1 shows a SEM image of the longitudinal (L) cross-section of the obtained drawn wire after picral etching. It was confirmed that the α-θ interfaces are almost parallel to the drawing direction as reported previously,^3,4) and an average lamellar spacing is about 70 nm.

Fig. 1.

SEM image of the L-cross section of the drawn wire used for the microbending test.

2.2. Methods

Disc specimens with a diameter of 2–4 mm and thickness of 2 mm were machined from the drawn wire, and the disc surface was polished with buffing compounds, and finished by Ar⁺ ion sputtering. On these mirror-polished surfaces, notched-MCLs with a triangular cross-section as illustrated in Fig. 2(a) were fabricated by FIB. The notch was milled near the fixed end of the MCL by linearly scanning Ga⁺ ion beam at 30 keV. In this milling condition, a V-shape notch with an open angle of about 20 degree, an apex curvature radius of less than 40 nm, and a depth of about 700 nm was obtained.

Fig. 2.

(a) Shape and (b) location of the MCL fabricated for the microbending test. Schematics of the notch orientation in the types A-C MCLs are shown in (c). (Online version in color.)

The MCLs were regioselectively fabricated such that the notch locates inside a pearlite colony, where the α-θ interfaces align in the same direction as illustrated in Fig. 2(b). Figure 2(c) shows a schematic of the relation between the notch and the α-θ interfaces. To investigate anisotropy in the HE resistance of the lamellar structure, three kinds of MCLs (types-A to -C MCLs) with different notch orientations were prepared; the notch is parallel to the α-θ interfaces for the type-A MCL, and normal to the α-θ interfaces for the types-B and -C MCLs. The type-A and -B MCLs were fabricated on the transversal (T) cross-sections of the drawn wire, and the type-C MCLs on the L cross-sections of the drawn wire. It is noteworthy that fabrication of the type-C MCLs on the T cross-section is difficult because the α-θ interfaces are almost perpendicular to the surface for this cross-section (Fig. 1).

In Fig. 3, a schematic of experimental setup of the microbending test is shown. Similarly to the previous study by the authors, a cono-spherical diamond indenter with a radius of less than 1 μm and an open angle 60 degree was pushed near the free end of the MCL in air and in electrolyte during cathodic hydrogen charging at room temperature (RT).¹²⁾ The indentation was performed under the displacement control, and the loading time and the maximum displacement were set to 180 s and 3000 nm, respectively. During the indentation, load-displacement relation of the indenter was monitored in real time.

Fig. 3.

Experimental setup of the microbending test. (Online version in color.)

Hydrogen was charged under the current control with a platinum counter electrode, and cathodic current density was set to 8.0 A/m². For the electrolyte, borate buffer solution (pH. 8.62) was used, and 3 g/L NH₄SCN was added as a poison. Precharge was conducted for more than 1200 s at 8.0 A/m² prior to the first microbending test to saturate the MCLs of hydrogen. It was confirmed by atomic force microscopy capability of the EC-NI that no corrosion occurs on the MCLs during the cathodic hydrogen charging.

The MCLs subjected to the microbending test were alternatively rinsed in ethanol and hot ultra-pure water at about 350 K, and dried in inert gas to remove the electrolyte. After that, the MCLs were observed by SEM and aberration-corrected S/TEM in vacuum to investigate the crack propagation path. Detailed preparation procedure of the SEM and S/TEM specimens is described afterward in Section 3.4.

To investigate hydrogen content in the MCL, thermal desorption analysis (TDA) was also conducted. For the TDA specimen, a drawn wire with a diameter of 4 mm and length of 14 mm was used, where hydrogen was charged for 48 hours in the same condition as for the microbending test. Temperature was elevated from RT to 873 K at a rate of 100 K/hour in vacuum, and hydrogen desorbed from the specimen was quantified with a quadro-pole mass spectrometer.

3. Results

3.1. TDA

In Fig. 4, a TDA curve of the hydrogen-charged specimen is shown. Two peaks were identified at 390 K and 580 K, which agreed with the previous study.^8,16) From the TDA curve, total hydrogen content and diffusive hydrogen one between RT and 493 K were estimated to be 9.26 mass ppm and 7.60 mass ppm, respectively; the temperature of 493 K corresponds to the valley temperature of the two peaks. Thus, it was confirmed that hydrogen is successfully charged to the MCLs during the microbending tests in the present cathodic hydrogen charging condition.

Fig. 4.

TDA curve of the drawn wire, where hydrogen was charged for 48 hours in the same condition as for the microbending test. (Online version in color.)

3.2. FIB-fabrication of MCLs

To confirm reproducibility of the microbending test results, eight, eight, and twelve MCLs were fabricated in total for the type-A, -B, and -C MCLs, respectively. For convenience, these type-A, -B, and -C MCLs are referred to as MCLs A1–A8, MCLs B1–B8, and MCLs C1–C12, respectively. Figure 5 shows SEM images of the MCLs acquired before the microbending tests. It was stochastically difficult to find a pearlite colony with the α-θ interfaces perfectly parallel to the surface even for the L cross-section. Consequently, the notch of the type-C MCL was milled inside a pearlite colony with a large lamellar spacing, where the α-θ interfaces are not perfectly but almost parallel to the surface.

Fig. 5.

SEM image of (a) the MCLs on the T cross-section, and those of the notch part of (b) the type-A, (c) type-B, and (d) type-C MCLs. (Online version in color.)

Dimension of the fabricated MCLs measured from the SEM images is summarized in Table 2, and definition of the parameters is schematically shown in Fig. 6. In Table 2, a loading point determined from imprints on the MCLs after the microbending tests, and moment of inertia of area I calculated from the following equation are also shown:¹⁷⁾   

$$I=\frac{a{b}^3}{36}$$(1)

The bending load of the cantilever is proportional to I in the linear elastic theory, and I is an important parameter for interpretation of the test results. For the types-A and -B MCLs, I was commonly about 15 μm⁴. On the other hand, for the type-C MCL, there were also slim MCLs with I of less than 10 μm⁴ (e.g., the MCLs C7 and C8), suggesting that the indentation load becomes smaller during the microbending tests.

Table 2. Dimension of fabricated MCLs measured from the SEM images.

Type	Mark	a (μm)	b (μm)	L (μm)	L1 (μm)	L2 (μm)	I (μm4)	environment
Type A	A1	6.7	4.4	14.9	13.0	1.6	15.8	Air
	A2	6.5	4.2	15.6	14.5	0.7	13.3
	A3	6.4	4.2	14.8	13.4	0.8	12.8
	A4	6.7	4.2	15.1	13.7	1.0	13.8	Hydrogen
	A5	6.7	4.4	14.8	13.4	0.5	16.3
	A6	6.8	4.6	14.3	13.4	0.5	18.8
	A7	6.6	4.4	14.9	13.7	0.6	15.6
	A8	7.0	4.5	15.3	12.6	0.6	18.2
Type B	B1	6.7	4.6	14.9	12.6	1.1	18.5	Air
	B2	6.8	4.4	14.9	13.7	1.2	16.1
	B3	6.9	4.4	15.1	13.1	1.0	16.7
	B4	7.1	4.3	15.7	13.4	0.9	15.8
	B5	6.8	4.2	14.9	13.5	0.8	14.5	Hydrogen
	B6	6.8	4.4	15.0	13.0	0.8	15.7
	B7	6.7	4.2	15.0	12.9	0.9	13.8
	B8	6.6	4.2	15.5	14.2	0.7	13.7
Type C	C1	6.6	3.6	15.8	13.4	1.0	8.3	Air
	C2	6.5	3.7	15.8	13.5	1.1	9.1
	C3	6.4	3.6	15.4	13.2	0.6	8.4
	C4	6.8	4.3	15.3	15.3	0.7	14.8
	C5	6.7	4.1	15.8	15.8	0.8	13.1
	C6	6.5	4.1	15.1	15.1	0.6	12.4
	C7	6.4	3.7	14.8	13.0	1.1	9.2	Hydrogen
	C8	6.2	3.4	14.8	13.2	0.8	6.9
	C9	6.8	4.4	15.1	15.1	0.6	15.7
	C10	6.6	4.3	14.2	14.2	0.6	14.3
	C11	6.7	4.2	15.8	15.8	0.8	13.4
	C12	6.4	4.1	15.2	15.2	0.5	12.5

Fig. 6.

Definition of parameters in Table 2.

3.3. Microbending Test

The MCLs A1–A3, the MCLs B1–B3, and the MCLs C1–C6 were bent in air, whereas the MCLs A4–A6, the MCLs B4–B8, and the MCLs C7–C12 were bent during the cathodic hydrogen charging. In Fig. 7, load-displacement relations recorded during the bending are shown. In Figs. 8, 9, 10, SEM images of the types-A to -C MCLs after the microbending tests are also shown, respectively. In air, the load monotonically increased with increase in the displacement, and no crack appeared irrespectively of the notch orientations. The variation in the maximum load among the different MCLs in Fig. 7 is principally attributed to the variation in the MCLs’ dimension given in Table 2. The tendency was identified that the maximum load becomes larger for the MCLs with larger I. During the hydrogen charging, on the other hand, significant difference was identified among the different notch orientations. For the type-A MCL, the load plunged in a plastic deformation region, and became zero before the maximum displacement of 3000 nm. In the SEM image, a crack penetrating the MCL in the thickness direction was observed at the notch root (Fig. 8). For the type-B MCL, similarly to the type-A MCL, the load declined, and a crack appeared at the notch root (Fig. 9). However, the load reduction with respect to the displacement was smaller, and the crack was shorter than for the type-A MCL. For the type-C MCL, the load declined, and a crack appeared only for the MCLs C7 and C8 with small I (Fig. 10). The load reduction was trivial, and the crack was much shorter than the types-A and -B MCLs. The crack did not spread also in the width direction of the MCL. For the type-C MCLs with large I (i.e., the MCLs C9–C12), the load monotonically increased, and no crack appeared similarly to in air.

Fig. 7.

Load-displacement relations recorded during the microbending test of (a) the type-A, (b) type-B, and (c) type-C MCLs. (Online version in color.)

Fig. 8.

SEM images of the type-A MCLs bent (a, b) in air and (c, d) during the cathodic hydrogen charging. The images in (b) and (d) are magnified images of the notch parts in (a) and (c), respectively. (Online version in color.)

Fig. 9.

SEM images of the type-B MCLs bent (a, b) in air and (c, d) during the cathodic hydrogen charging. The images in (b) and (d) are magnified images of the notch parts in (a) and (c), respectively. (Online version in color.)

Fig. 10.

SEM images of the type-C MCLs bent (a, b) in air and (c–f) during the cathodic hydrogen charging. The SEM images in (c, d) are of the MCL C7 with small I, and those in (e, f) are of the MCL C10 with large I. The images in (b), (d) and (f) are magnified images of the notch parts in (a), (c), and (e), respectively. (Online version in color.)

Reduction in a cross-section area of the MCL due to crack propagation is known to bring about the load reduction in the load-displacement relation.^12,18,19) The load reduction rate with respect to the displacement is regarded as a crack propagation rate with respect to given strain. It turns out from the present microbending test results that the HE crack propagates more easily in the direction parallel to the α-θ interface than normal to the α-θ interface.

Figure 11 shows SEM images of the crack surface of the type-A and -B MCLs. For the type-C MCL, the crack opening was not large enough that the crack surface could not be observed by SEM. For the type-A MCL, the crack surface was comparably flat, and there observed wavy steps. On the other hand, for the type-B MCL, a stripe pattern with separation of about 160 nm was identified on the crack surface, suggesting that delamination fracture occurs during the bending.

Fig. 11.

SEM images of the crack surfaces of (a) the type-A and -B MCLs bent during the cathodic hydrogen charging. (Online version in color.)

3.4. SEM and S/TEM Observations

To confirm a propagation path of the HE crack under the notch root, the types-A and -C MCLs were milled into a shape as illustrated in Fig. 12 by FIB, and cross-sections normal to the width direction of the MCL were observed by SEM. Prior to the milling, a platinum layer was formed on the notch root by FIB deposition to prevent the crack from being damaged by Ga⁺ ion beam. As described in Section 3.3, the crack penetrated the type-A MCL in the thickness direction when the indenter was pushed to the maximum displacement of 3000 nm (Fig. 8). On the other hand, stationary cracks are essential for the identification of the crack propagation path by SEM. Consequently, for the type-A MCL, the MCL A7 was milled by FIB, where the indentation was interrupted at the displacement of about 1000 nm soon after the occurrence of the load reduction as shown in Fig. 13. For the type-C MCL, the MCL C8 with small I was milled by FIB, where the stationary crack was formed after the microbending test with the maximum displacement of 3000 nm.

Fig. 12.

FIB milling of the cross-section for the SEM observations of the stationary crack. (Online version in color.)

Fig. 13.

Load-displacement relations of the type-A MCLs used for the SEM and S/TEM observations of the stationary crack. (Online version in color.)

In Fig. 14, the SEM images of the cross-sections are shown. It was confirmed that a stationary crack was successfully obtained for the MCL A7. A sharp crack was observed along near the α-θ interface for the type-A MCL, and a blunt crack for the type-C MCL. The crack length was almost the same between the type-A and -C MCLs although the maximum displacement of the type-A MCL was about one third of that of the type-C MCL.

Fig. 14.

SEM images of the cross-sections of (a) the type-A and (b) type-C MCLs bent during the cathodic hydrogen charging. (Online version in color.)

It is still difficult to distinguish which of the following is the propagation path for the type-A MCL from the SEM image in Fig. 14(a): the α-θ interface, the α lamella near the α-θ interface, and the θ lamella. To clarify this, S/TEM observations were also conducted for the type-A MCL (MCL A8), where a stationary crack was formed in the same manner as for the MCL A7 (Fig. 13). In Fig. 15 preparation procedures of the S/TEM specimen from the bent MCL are illustrated. The MCL was extracted from the disc sample surface by FIB lift-out technique,²⁰⁾ and then milled into a thin plate by FIB and Ar⁺ ion sputtering.

Fig. 15.

Milling procedure of the MCL into the S/TEM specimen. (Online version in color.)

In Figs. 16(a) and 16(b), low- and high-magnification bright-field STEM images of the crack is shown, respectively. The θ lamellar position indicated by the dashed lines in Fig. 16(b) were determined from electron diffraction patterns and dark-field TEM images of θ. It is noted that the image in Fig. 16(b) is rotated by 90 degree in the clockwise direction due to space limitation. At a glance, the crack propagated along the α-θ interface, and stopped at the position where the θ lamellae are disconnected. Figure 16(c) shows a lattice TEM image of the rectangular area in Fig. 16(b). An amorphous structure due to re-deposition during the milling was observed inside the crack, and ordered structures with thickness of 2–3 nm at the crack subsurface. The θ (α) phase was never observed in both sides of the crack; the θ phase was identified in one side of the crack, and the α phase in another side of the same crack.

Fig. 16.

(a) low- and (b) high-magnification bright-field STEM images of the stationary crack in the type-A MCL. The high-magnification image in (b) was rotated by 90 degree in the clockwise direction due to space limitation. The image in (c) is a lattice image obtained by high resolution TEM corresponding to the area indicated by the rectangle in (b). (Online version in color.)

It seems that a crystalline orientation in the θ lamella affects the HE crack propagation. For this reason, dark-field TEM images of θ were also acquired. In Fig. 17, the observed dark-field TEM image is shown, and the θ lamella position is indicated by the dashed lines. In the θ lamella, electron beam intensity was inhomogeneous, and a fragmentated structure was observed; the fragments with strong electron beam intensity are indicated by the arrows. These results suggest that nanoscale grains develop in the θ lamella during the cold drawing as previously reported.^5,6)

Fig. 17.

Dark-field TEM image of θ. The arrows indicate the fragments with strong electron beam intensity in the θ lamellae. (Online version in color.)

4. Discussions

4.1. Influence of Hydrogen on Young’s Modulus and Flow Stress

As reported in the previous study by the authors, Young’s modulus E and average flow stress $\overline{\sigma }$ over the thickness direction of the MCL can be estimated from the load-displacement relations in the microbending test. More specifically, for triangular MCLs without a notch, E is calculated from gradient g of the load-displacement relation in the elastic deformation region, and $\overline{\sigma }$ from load P_plastic in the plastic deformation region:   

$$g=\frac{3I}{L_1}E=\frac{a{b}^3}{12L_1{}^3}E$$(2)

  

$$\overline{\sigma }=\frac{81}{8}\frac{P_{plastic}{L}_1}{a{b}^2}$$(3)

On the assumption that the fabricated MCLs have no notch, E and $\overline{\sigma }$ were calculated from the observed load-displacement relations in Fig. 7 and the actual MCL’s dimension in Table 2, and the influence of hydrogen on the mechanical properties was investigated.

In Table 3, the calculated E and $\overline{\sigma }$ are summarized together with g and P_plastic. Equation (3) is derived from momentum equilibrium at the fixed end of the MCL, and the fixed end seems to be fully plastically deformed at the maximum displacement.²¹⁾ Thus, load at 3000 nm was substituted into P_plastic. The calculation results were shown only for the type-C MCLs without the load reduction, where no crack appeared, because P_plastic can be precisely determined only for such MCLs.

Table 3. E and $\overline{\sigma }$ calculated from the observed load-displacement relations in Fig. 7 and the dimension of the MCLs in Table 1.

Mark	g (μN/nm)	Pplastic (μN)	E (GPa)	$\overline{\sigma }$ (MPa)	Avg. E (GPa)	Avg. $\overline{\sigma }$ (MPa)	Environment
C1	1.35	1165	129.6	1883	121.5±16.7	1832±260	Air
C2	1.36	1267	122.1	1952
C3	1.51	1192	139.0	1916
C4	1.57	1266	89.9	1408
C5	1.51	1639	126.5	2162
C6	1.72	1307	121.7	1673
C9	2.10	1865	102.5	1658	108.3±18.3	1731±115	Hydrogen
C10	1.79	1796	88.2	1888
C11	1.71	1778	132.1	1745
C12	1.60	1592	110.4	1632

In air, E and $\overline{\sigma }$ were 121.5±16.7 GPa and 1832±260 MPa, respectively. On the other hand, during the hydrogen charging, E and $\overline{\sigma }$ were 108.3±18.3 GPa and 1731±115 MPa, respectively. The actual MCLs has the notch, and the absolute values of E nor $\overline{\sigma }$ are not reliable. However, it is obvious that neither E nor $\overline{\sigma }$ are affected by hydrogen. In general, $\overline{\sigma }$ reflects dislocation motion (e.g., dislocation mobility). It seems that hydrogen does not significantly change the dislocation motion for the drawn pearlitic steel.

4.2. Preferential Path of HE Crack

In the present study, the MCLs were placed in alkaline borate buffer solution (pH. 8.6) without cathodic hydrogen charging for a few minutes when the MCLs were removed from the EC-NI apparatus for the SEM and S/TEM observations. On the other hand, A.V. Syugaev et al. measured polarization curves of θ synthesized by mechano-chemical reaction.^22,23) They reported that, similarly to α, a two-layer passivation film composed of an inner Fe₃O₄ layer and an outer γ-Fe₂O₃ one is formed on θ at anodic potentials, and the passivation potential decreases with increase in pH. Thus, the ordered structures with thickness of 2–3 nm observed at the crack subsurface in the lattice TEM image (Fig. 16(c)) are very likely to be self-passivation films formed on the α and θ lamellae, respectively, in the alkaline electrolyte after the crack propagated. The crack propagation along the α-θ interface at nanoscale described above (Fig. 16(c)), and the crack termination at the position, where the θ lamellae are disconnected (Fig. 15(b)), both indicate that the preferential crack propagation path is the α-θ interface but neither in the α lamella nor in the θ lamella.

Supposedly, the α-θ interface in drawn pearlitic steel is incoherent, and its cohesive energy is small. On the other hand, such incoherent interfaces trap much hydrogen. The separation of the α-θ interface seems to be caused by hydrogen-enhanced decohesion (HEDE) mechanism, where hydrogen on the α-θ interfaces lowers cohesive force between the α and θ lamellae.^24,25)

It was confirmed from the dark-field image of θ that the θ lamella is composed of nano-scale grains, and there exist numerous grain boundaries in the θ lamella (Fig. 17). In general, grain boundaries behave as obstacles against the crack propagation. K. Kawakami et al., on the other hand, investigated hydrogen trapping sites in θ by the first-principle calculations, and reported that diffusion barrier of hydrogen into θ is high.²⁶⁾ The absence of the HE crack in the θ lamellae is attributed to the high-density of grain boundaries and the low concentration of hydrogen therein.

4.3. Fracture Toughness

To quantify how easy for the crack to propagate along the α-θ interface, a fracture toughness K_c was analyzed for the type-A MCL (the MCL A7), where a stationary crack was nucleated.

The crack propagates such that a stress intensity factor K determined from the crack length x and the bending load P becomes equal to K_c during the bending. The observed HE crack along the α-θ interface is sharp, and approximated by an edge crack. For triangular cantilevers with such an edge crack, K in mode I is given by the following equation:   

$$K=\frac{12P{L}_3}{a{b}^2}\times 1.12\sqrt{\pi x}$$(4)

where L₃ is distance between the loading point and the crack.¹⁸⁾ This equation is valid when x ≪ b. For the MCL A7, P and x are found to be 750 μN and 0.83 μm, respectively, from the load-displacement relation in Fig. 13 and the SEM image of the cross-section in Fig. 14(a). On the other hand, a, b, and L₃(=L₁−L₂) of the MCL A7 are found to be 6.6 μm, 4.4 μm, and 13.1 μm, respectively, as summarized in Table 2. Thus, from Eq. (4), K_c of 1.7 MPa m^1/2 is readily obtained. The obtained value is rough estimation because x is possibly different along the width direction of the MCL. The order of the obtained value is, however, likely to be correct.

One might think K_c can be obtained by substituting the crack formation load, at which the load drop occurs in the load-displacement relation, and the notch depth into P and x, respectively, in Eq. (4). This approach is, however, not appropriate because the notch root is work-hardened by Ga⁺ ion during the FIB fabrication of the MCLs, which suppresses the crack formation. The finite curvature radius of the notch root also causes an over estimation of K_c.

It is also noted that obtaining K_c is non-straightforward for the type-C MCLs. Equation (4) can be used only when small scale yielding (SSY) conditions are satisfied, where plastic deformation is localized around the crack apex. For the type-A MCLs, the SSY conditions are likely to be satisfied because the sharp crack was formed at small indenter displacement just after the elastic deformation (Fig. 14(a)). In contrast, for the type-C MCLs, the SSY conditions are unlikely to be satisfied because the blunt crack was formed at large indenter displacement (Fig. 14(b)).

In fracture toughness tests using macroscopic specimens, K_c of hydrogen-charged tempered martensitic steel is reported to be about 20 MPa m^1/2.²⁷⁾ On the other hand, K_c of hydrogen-charged drawn pearlitic steel for the crack propagation in the radial direction of the drawn wire is reported to be about 40 MPa m^1/2.¹⁰⁾ Thus, K_c of 1.7 MPa m^1/2 obtained by the present microbending tests is found to be very small. This large difference in K_c between macroscopic and microscopic tests is not surprising because, in general, collective propagation of a number of microscopic cracks is necessary for propagation of a macroscopic crack.^28,29) Influence of microstructure boundaries (e.g. colony boundaries) is also reflected in K_c for the macroscopic tests, but not for the microscopic tests.

4.4. Cause of Superior HE Resistance of Drawn Pearlitic Steel

The microbending test results of the type-A MCL imply that, when the lamellar structure experiences tensile stress normal to the α-θ interface in hydrogen environment, a long crack appears along the α-θ interfaces (Fig. 14(a)). On the other hand, the microbending test results of the types-B and -C MCLs imply that, when the lamellar structure experiences tensile stress parallel to the α-θ interface in hydrogen environment, no crack appears; otherwise a short and blunt crack appears (Fig. 14(b)). In the drawn pearlitic steel, the α-θ interfaces are parallel to the drawing direction as shown in Fig. 1.^3,4) Thus, few crack appears when the drawn pearlitic steel is strained in the drawing direction in the hydrogen environment. It is concluded that the directional lamellar alignment plays a crucial role in the superior HE resistance of the drawn pearlitic steel in the drawing direction.

5. Summary

The MCLs with different notch orientations with respect to the α-θ interfaces were fabricated on the drawn pearlitic steel by FIB, and in-situ microbending tests were conducted in air and during the cathodic hydrogen charging by EC-NI. As a result, the following findings were obtained.

(1) The load reduction with respect to the indenter displacement was larger, and the crack was longer in the order of the type-A, -B, and -C MCLs. It was revealed that the HE crack propagates more easily in the direction parallel to the α-θ interface than normal to the α-θ interface.

(2) From the SEM and S/TEM observations of the stationary cracks in the MCL, the preferential propagation path of the HE crack was identified to be the α-θ interface.

(3) From the load-displacement relations of the microbending test and the dimension of the MCLs, E and $\overline{\sigma }$ were calculated. The cathodic hydrogen charging did not affect these mechanical properties.

(4) For the crack propagation along the α-θ interface in the hydrogen environment, K_c was considerably small and estimated to be 1.7 MPa m^1/2.

(5) The directional lamellar alignment of the drawn pearlitic steel suppresses HE crack propagation in the radial direction of the drawn wire. Consequently, the drawn pearlitic steel exhibits excellent HE resistance in the drawing direction.

Acknowledgements

The authors thank Yuji Sakiyama and Taizo Makino at Nippon Steel and Sumitomo Metal Corporation, and Kazuto Kawakami and Yasuteru Nawafune at Nippon Steel and Sumikin Technology for fruitful discussion and sample preparation.

References

• 1)   B.  Marandet: NACE, 5 (1977), 774.
• 2)   T.  Fujita and  T.  Yamada: NACE, 5 (1977), 736.
• 3)   J. D.  Embury and  R. M.  Fisher: Acta Metall., 14 (1966), 147.
• 4)   V. K.  Chandhok,  A.  Kasak and  J. P.  Hirth: Trans. Am. Soc. Met., 59 (1966), 288.
• 5)   D. A.  Porter,  K. E.  Easterling and  G. D. W.  Smith: Acta Metall., 26 (1978), 1405.
• 6)   M. H.  Hong,  W. T.  Reynolds, Jr.,  T.  Tarui and  K.  Hono: Metall. Mater. Trans. A, 30A (1999), 717.
• 7)   N.  Sarafianos: J. Mater. Sci. Lett., 8 (1989), 1486.
• 8)   K.  Takai,  J.  Seki and  Y.  Honma: Tetsu-to-Hagané, 81 (1995), 1025.
• 9)   J.  Toribio and  A.  Lancha: J. Mater. Sci., 31 (1996), 6015.
• 10)   D. G.  Enos and  J. R.  Scully: Metall. Mater. Trans. A, 33A (2002), 1151.
• 11)   E.  Proverbio and  P.  Longo: Corros. Sci., 49 (2007), 2421.
• 12)   K.  Tomatsu,  H.  Kawata,  T.  Amino,  T.  Omura,  N.  Maruyama and  Y.  Nishiyama: ISIJ Int., 57 (2017), 564.
• 13)   M.  Seo and  M.  Chiba: Electrochim. Acta, 47 (2001), 319.
• 14)   A.  Barnoush and  H.  Vehoff: Acta Mater., 58 (2010), 5274.
• 15)   K.  Tomatsu,  K.  Miyata and  T.  Omura: ISIJ Int., 56 (2016), 418.
• 16)   T.  Doshida and  K.  Takai: Acta Mater., 79 (2014), 93.
• 17)   D. E. J.  Armstrong,  A. J.  Wilkinson and  S. G.  Roberts: J. Mater. Res., 24 (2009), 3268.
• 18)   A. D.  Norton,  S.  Falco,  N.  Young,  J.  Severs and  R. I.  Todd: J. Eur. Ceram. Soc., 35 (2015), 4521.
• 19)   Y.  Takahashi,  H.  Kondo,  R.  Asano,  S.  Arai,  K.  Higuchi,  Y.  Yamamoto,  S.  Muto and  N.  Tanaka: Mater. Sci. Eng. A, 661 (2016), 211.
• 20)   L. A.  Giannuzzi,  J. L.  Drown,  S. R.  Brown and  R. B.  Irwin: Microsc. Res. Tech., 41 (1998), 285.
• 21)   C.  Motz,  T.  Schöberl and  R.  Pippan: Acta Mater., 53 (2005), 4269.
• 22)   A. V.  Syugaev,  S. F.  Lomaeva and  S. M.  Reshetnikov: Prot. Met., 44 (2008), 51.
• 23)   A. V.  Syugaev,  S. F.  Lomaeva,  A. N.  Maratkanova,  D. V.  Surnin and  S. M.  Reshetnikov: Prot. Met. Phys. Chem. Surf., 45 (2009), 81.
• 24)   R. P.  Frohmberg,  W. J.  Barnett and  A. R.  Troiano: Trans. Am. Soc. Met., 47 (1955), 892.
• 25)   R. A.  Oriani and  P. H.  Josephic: Acta Metall., 22 (1974), 1065.
• 26)   K.  Kawakami and  T.  Matsumiya: ISIJ Int., 53 (2013), 709.
• 27)   N. R.  Moody,  S. L.  Robinson and  W. M.  Garrison: Res Mechanica, 30 (1990), 143.
• 28)   J.  Kameda: Metall. Trans. A, 12A (1981), 2039.
• 29)   M.  Yamaguchi: Materia Jpn., 54 (2015), 111.