Hydrogen Effects on Fracture Resistances of Bulk Cementite Evaluated by in-situ Microbending Test during Cathodic Hydrogen Charging

Abstract

Hydrogen absorption characteristics and mechanical properties in hydrogen environment of cementite were evaluated by low-temperature thermal desorption analysis and in-situ microbending tests during cathodic hydrogen charging using bulk cementite plates obtained through a vacuum carburizing process. In the low-temperature thermal desorption analysis, no hydrogen desorption was identified up to 1073 K. In the microbending test, notched microcantilevers experienced cleavage fracture in an elastic deformation range in air. The cathodic hydrogen charging increased fracture load (i.e., fracture toughness) with appearance of plasticity while it did not change the fracture surface morphology and Young’s modulus. In the present hydrogen charging conditions, the hydrogen atoms are present only near the specimen surface because of high hydrogen migration energy in the cementite. It seems that no hydrogen desorption is detected because the hydrogen atoms are absent in most regions of the specimens. The invariance of the Young’s modulus and the fracture surface morphology can be explained by the same reason. On the other hand, it is considered that the fracture toughness is improved because the hydrogen atoms charged near notch bottom of the microcantilever enhance dislocation nucleation and glide, and cause blunting of the notch during the bending.

1. Background

Development of high-strength steel requires overcoming hydrogen embrittlement (HE), which becomes evident with increase in the steel strength.^1,2,3,4) On the other hand, most of high-strength steel contain cementite (Fe₃C), which is a carbide of iron, and it is important to elucidate hydrogen absorption characteristics and mechanical properties in hydrogen environments of cementite itself. However, these remain yet to be clarified because the cementite exists in the form of fine particles or thin plates in steel and direct measurements are difficult to be conducted.

Regarding the hydrogen absorption characteristics, Terasaki et al. investigated reaction of cementite powder in high-pressure hydrogen gas by X-ray diffraction (XRD). They found that the hydrogen gas neither change the cementite lattice volume nor form a new phase up to 14 GPa and 1973 K, and suggested that hydrogen atoms are not absorbed by the cementite.⁵⁾ Kawakami et al. investigated trapping energies and migration ones of hydrogen atoms in the cementite by first-principles calculations.⁶⁾ They found that an octahedral (O) site surrounded by six iron atoms is the most stable one for the hydrogen atom. The migration energy between the neighboring O sites is high and more than 55 kJ/mol (0.57 eV), and it is predicted that few hydrogen atoms are trapped in the cementite at room temperature. On the other hand, Adachi et al. conducted electrochemical hydrogen permeation tests for a bulk cementite plate obtained through a sintering process.^7,8) They found that no hydrogen permeation is detected during the cathodic hydrogen charging at least for 3.5 days for the plate specimen with a 0.5 mm thick. This experimental result also implies that the hydrogen diffusion coefficient of the cementite is considerably small.

Regarding the mechanical properties in hydrogen environments, the authors conducted in-situ microbending tests of drawn pearlitic steel during cathodic hydrogen charging and investigated HE crack propagation paths in the bent specimens by scanning electron microscopy (SEM) and transmission electron microscopy (TEM).⁹⁾ It was found that the HE crack preferentially propagates along the ferrite-cementite interface but cannot across the cementite. On the other hand, Pinson examined roles of cementite in HE for martensitic medium carbon steel using sample materials with different cementite densities.¹⁰⁾ They found that protrusions corresponding to the cementite particles are seen on hydrogen-induced intergranular fracture surfaces and suggested that ferrite-cementite interfaces are preferentially separated by the hydrogen atoms. Both these two experiments indicate that the mechanical properties of cementite are not degraded by the hydrogen atoms differently from the ferrite matrix in steel. However, strength or toughness of cementite itself in the hydrogen environments has not been quantitatively and directly evaluated, so far.

Morita et al. demonstrated that when low-alloy steel supplemented with chromium is carburized, its surface is covered with cementite without any graphite nucleation.¹¹⁾ This bulk cementite has fewer voids and larger grains than that obtained through the sintering process⁸⁾ and seems to be a suitable material for investigating the characteristics of cementite itself. In this context, in the present study, we investigated the hydrogen absorption characteristics and the mechanical properties in the hydrogen environments using the bulk cementite obtained through the vacuum carburizing process.

Low-temperature thermal desorption analysis (LT-TDA) was used to evaluate the hydrogen absorption characteristics. On the other hand, as described later, the bulk cementite contained a ferrite phase although its fraction is small. Such a different phase hinders accurate evaluation of the mechanical properties of the cementite because the phase boundaries can provide crack initiation sites.^9,10) According to the calculations by Kawakami et al.,⁶⁾ it is also very likely that the hydrogen atoms are charged only near the specimen surface for practical hydrogen charging times because of their high migration energy. Therefore, in-situ microbending tests during cathodic hydrogen charging^9,12,13) were used that allows us to selectively evaluate the mechanical properties of the small volume near the specimen surface.

2. Experimental

2.1. Materials

The bulk cementite was obtained by carburizing low-carbon steel with a chemical composition given in Table 1. According to the study by Morita et al., chromium suppresses graphite nucleation and stabilizes cementite during the carburizing process as described above.¹¹⁾ Thus, 4.9 wt.% of chromium was added in the present study, that is an optimal value determined from phase diagrams calculated with a Thermos-calc software. The sample steel was cold-rolled into plates with a 0.3 mm thick, and then vacuum carburized with the heating pattern shown in Fig. 1. That is, the steel plate was homogenized at 1373 K for 1800 s, exposed to acetylene gas at the same temperature for 12 h, and finally gas cooled to room temperature using nitrogen gas.

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

C	Si	Mn	Cr	Fe
0.19	0.003	0.003	4.9	Bal.

Fig. 1. Heating pattern of vacuum carburizing. (Online version in color.)

2.2. Experimental Procedures

To confirm whether the bulk cementite is successfully obtained through the vacuum carburizing process, phase distributions of the sample materials were evaluated by electron backscattering diffraction (EBSD). Then, the hydrogen absorption characteristics were evaluated by LT-TDA and the mechanical properties (i.e., toughness and Young’s modulus) in the hydrogen environments by the in-situ microbending tests during the cathodic hydrogen charging. When exposed to high temperature hydrogen gas, cementite releases ammonium gas (NH₄) and is reduced to steel.¹⁴⁾ Such reduction or chemical damages due to the electrolyte can influence the test results. Thus, atomic force microscopy (AFM) and energy-dispersive X-ray spectroscopy of SEM (SEM-EDS) were also conducted to confirm surface conditions of the specimens. The detailed test conditions of each test in the present study are described below.

2.2.1. EBSD

The plate surface (i.e., the surface exposed to the acetylene gas) was finished with electrochemical polishing. Then, EBSD were conducted and the phase type was analyzed with a commercial software (OIM Analysis^TM) assuming three typical crystalline structures in steel: a body-centered cubic structure of ferrite, a face-centered cubic one of retained austenite, and an orthorhombic one of cementite.

2.2.2. LT-TDA

Specimens measuring 10 mm×7 mm×0.3 mm size were cut from the sample material, and all the sides of the specimen was polished with emery papers. Then, cathodic hydrogen charging was performed at room temperature, and the absorbed hydrogen content was evaluated by the LT-TDA. Until just before the analysis, the specimens were kept in a freezer at 223 K to prevent hydrogen desorption. The hydrogen was charged for 48 h in 3% NaCl solution with 3 g/L NH₄SCN at a current density of 10 A/m². In the LT-TDA, the specimen was heated from 173 K up to 1073 K at a heating rate of 100 K/h, and the desorbed hydrogen molecules were detected by gas chromatography. Hydrogen desorption from a sample stage or a furnace wall of the LT-TDA apparatus is not negligible at high temperatures more than about 800 K. Therefore, after completely removing the hydrogen atoms from the specimen at 1073 K, it was again heated from 173 K up to 1073 K and these background hydrogen atoms were also evaluated.

2.2.3. In-situ Microbending Tests

The plate surface of the sample material was electrochemically polished and microcantilevers (MCLs), which are the specimens of the microbending tests, were fabricated by focused ion beam (FIB) thereon. Dimension and shape of the MCL are schematically shown in Fig. 2(a) and fabrication positions in Fig. 2(b). Similarly to the MCLs in the previous authors’ study,⁹⁾ the MCLs with a triangular cross-section were used, and a notch was added to the vicinity of their fixed ends by linearly scanning Ga⁺ ion beam at an energy of 30 keV. As described later, the bulk cementite contained trace amounts of ferrite phases. Thus, the MCLs were selectively fabricated in the cementite phase avoiding the ferrite one. Furthermore, to eliminate possible crystalline orientation effects, all the MCLs were placed such that their notches are contained in the same cementite grain and have the same orientation. The grain boundaries were identified by obtaining secondary electron (SE) images of SEM and cross-checked by obtaining inverse pole figure (IPF) images of EBSD after the microbending tests.

Fig. 2. (a) Shape and dimension of the MCLs and (b) their fabrication positions. (Online version in color.)

In Fig. 3, an experimental setup of the in-situ microbending test is schematically shown. The tests were conducted by pressing an indenter onto the MCL with a nanoindentation apparatus (Hysitron Triboindenter-950), with an electrochemical cell for the cathodic hydrogen charging (electrochemical nanoindentation: EC-NI).⁹⁾ For the indenter, a cono-spherical diamond indenter with a 60° open angle and a radius of curvature less than 1 μm was used. The indenter was pressed in the displacement control mode, and the maximum displacement and the indentation time were set to 3 μm and 180 s, respectively. The microbending tests were conducted multiple times both in air and during cathodic hydrogen charging at room temperature, and influences of hydrogen on Young’s modulus and toughness were evaluated from load-displacement curves of the indenter during the bending. The cathodic hydrogen charging was performed in borate buffer solution (pH 8.6) with 3 g/L NH₄SCN at a current density of 5 A/m². Because the hydrogen diffusion coefficient of the cementite is likely to be small at room temperature as described above, hydrogen charging was also performed for 4 h in the same conditions prior to the microbending tests. After the microbending tests, the plate specimen with the MCLs was alternatively rinsed in ethanol and ultrapure water at about 353 K to remove the electrolytes, and the MCLs were observed by SEM in vacuum.

Fig. 3. Schematic of in-situ microbending tests during the cathodic hydrogen charging. (Online version in color.)

2.2.4. AFM

Topography images of electro-chemically polished sample surfaces were obtained both in air and during cathodic hydrogen charging by using the AFM capability of the EC-NI apparatus. The cathodic hydrogen charging conditions were kept the same as in the in-situ microbending tests described above. On the other hand, a cube-corner diamond indenter was used instead of the cono-spherical one, allowing us to detect fine irregularities on the surface.

2.2.5. SEM-EDS

Chemical compositions of electrochemically polished surfaces of the sample material before and after the cathodic hydrogen charging were analyzed by SEM-EDS. The cathodic hydrogen charging conditions were kept the same as in the in-situ microbending described above. The specimen after the hydrogen charging was alternatively rinsed in ethanol and ultrapure water at about 353 K to remove the electrolytes prior to the SEM-EDS.

3. Results

3.1. Microstructure of Sample Material

Phase and IPF images of the plate surface of the sample material obtained through the carburizing process are shown in Fig. 4. As expected, the primary phase was cementite, and its grain size was approximately 100 μm. Massive ferrite phases were also seen both in the grain interior and on the grain boundary while retained austenite phase was absent. The area fraction of the ferrite phases was about 5%. It was found that the ferrite phase is small enough to selectively fabricate the MCLs in the cementite phase by FIB. It was also confirmed that the phase distributions are homogeneous in the thickness direction of the plate (not shown).

Fig. 4. (a) Phase and (b) IPF images of the sample material. (Online version in color.)

3.2. Hydrogen Absorption Characteristics

Hydrogen desorption spectra of the bulk cementite obtained in the LT-TDA are shown in Fig. 5. The migration energy of 0.56 eV between the O sites reported by Kawakami et al. corresponds to desorption temperatures around 600 K.¹⁵⁾ However, no hydrogen desorption was identified around this temperature while it was identified at higher temperatures above 800 K. On the other hand, the hydrogen desorption spectra of hydrogen-charged and hydrogen-removed specimens were very similar. Therefore, the hydrogen molecules desorbed above 800 K were assigned to those from the LT-TDA apparatus and background. It was found that the cementite does not absorb hydrogen atom, or the content of the charged hydrogen atoms is too small to be detected with the gas chromatography detector.

Fig. 5. Hydrogen desorption spectra of LT-TDA of the hydrogen-charged and hydrogen-removed specimens. (Online version in color.)

3.3. Dimensions and Positions of MCLs

A SE image of the MCLs before the microbending tests is shown in Fig. 6(a), and an IPF image after the microbending tests in Fig. 6(b). Totally eight MCLs were successfully fabricated in the cementite phase avoiding the ferrite one. It was also confirmed that all the MCLs have their notches in the same cementite grain with the same orientation. The crystalline plane of the cementite grain, where the notches were fabricated, was (-17 11 -9) and the longitudinal direction of the MCL was approximately parallel to the <1 -4 -7> direction. Because of limitations on the grain size, two types of MCL with the opposite orientations to each other were fabricated. It is notable that, in principle, the fracture load obtained by the microbending tests is the same between these MCLs because the stress direction with respect to the crystalline planes under the notch is approximately the same.

Fig. 6. (a) SE and (b) IPF images of the MCLs. The SE and IPF images were obtained before and after the microbending tests, respectively. The red dashed lines in (b) represent the notch positions. (Online version in color.)

Dimensions of the eight MCLs measured from their SE images are given in Table 2, and definitions of the parameters in Table 2 are schematically shown in Fig. 7. The variations in the width a and the height b of the MCLs were 6.7±0.1 μm and 4.7±0.1 μm, respectively. In a linear elasticity theory, the indentation load is proportional to second moment of area I given by the equation,¹⁶⁾

  

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

The variation in I of the MCLs was 19.4±1.3 μm⁴. Thus, the dimensional variation of the MCLs was also successfully reduced to a low level.

Table 2. Dimensions of the fabricated MCLs. Definitions of the parameters are given in Fig. 8.

Mark	a (μm)	b (μm)	L (μm)	L1 (μm)	L2 (μm)	L (μm4)	$\overrightarrow{v}$	Test environment
MCL A1	6.6	4.5	16.4	14.5	1.1	16.7	[1 -4 -7]	air
MCL A2	6.6	4.7	16.4	14.4	1.0	19.0	[-1 4 7]
MCL A3	6.7	4.7	16.9	14.9	1.1	19.3	[1 -4 -7]
MCL H1	6.5	4.7	17.1	15.0	1.0	18.7	[-1 4 7]	hydrogen
MCL H2	6.8	4.8	17.2	15.2	1.8	20.9	[1 -4 -7]
MCL H3	6.6	4.8	16.7	14.7	1.1	20.3	[-1 4 7]
MCL H4	6.7	4.7	16.8	14.8	1.3	19.3	[-1 4 7]
MCL H5	6.7	4.8	17.0	15.0	1.5	20.6	[1 -4 -7]

Fig. 7. Definitions of the parameters in Table 2.

Fracture load in the microbending tests is sensitive to a notch shape. To measure the notch shape, a dummy notch was fabricated on a free space inside the cementite grain in the same way as for the MCL’s notch. Then, a protective platinum layer was deposited on the dummy notch by FIB deposition, and its cross-section fabricated by FIB was observed by SEM. A SE image of the cross-section obtained by tilting the plate specimen by 45° is shown in Fig. 8(a), and a schematic of the dummy notch shape in Fig. 8(b). The depth and the radius of curvature of the notch were measured to 200 nm and 90 nm, respectively.

Fig. 8. (a) SE image of the cross-section of the dummy notch and (b) schematic of the dummy notch shape.

3.4. Microbending Test Results

Out of the eight MCLs in Table 2, three (MCLs A1 to A3) were bent in air and five (MCLs H1 to H5) during the cathodic hydrogen charging. In Fig. 9, obtained load-displacement curves of the indenter during the bending are shown. All the MCLs fractured irrespectively of the test environments. The fracture loads are summarized in Table 3. As described later, the MCL H5 was fractured between the free end and the notch. Therefore, the value of the MCL H5 (2263 μN) was excluded from the average in Table 3. The average of the fracture loads was 1549±70 μN in air and 2429±473 μN during the hydrogen charging. Interestingly, cathodic hydrogen charging increased the fracture load by about 55%.

Fig. 9. Load-displacement curves in the microbending tests (a) in air and (b) during the hydrogen charging. The dashed lines represent the results of linear fitting for MCLs A1 and H4 between 0 nm and 1000 nm. (Online version in color.)

Table 3. Fracture loads of the MCLs.

Mark	Fracture load (μN)	Avg. fracture load (μN)
MCL A1	1660	1549±70
MCL A2	1530
MCL A3	1549
MCL H1	1935	2429±473
MCL H2	2586
MCL H3	2180
MCL H4	3013
MCL H5	2263	–

It is known that, in bending tests, the load and deflection have a linear relationship in an elastic deformation range, and the load approaches a constant value as plastic deformation progresses.¹²⁾ In Fig. 9, dashed lines are drawn to indicate an approximate range of elastic deformation, that are obtained by fitting the experimental data between 0 nm and 1000 nm with lines for the MCLs A1 and H4. In air, the load-displacement curves were linear up to the fracture load for all the MCLs. On the other hand, during the hydrogen charging, they deviated from the dashed line near the fracture load especially for the MCLs with the higher fracture load (e.g., the MCLs H2 and H4). It was also found that the MCLs were fractured in the elastic deformation range in air and the elastoplastic deformation one during the cathodic hydrogen charging although the plasticity is small.

3.5. Fracture Surfaces of MCLs

SE images of the MCLs A1, H2 and H5 after the microbending tests and that of the MCL H5 before the microbending tests are shown in Fig. 10 as representatives. These three MCLs have the same orientation (i.e., $\overrightarrow{v}$ value) as shown in Table 2. For the MCLs A1 and H5, their fractured free-ends were lost during the rinse process and not seen in the images. The fractures occurred at the notch for all the MCLs excepting for the MCL H5. The fracture locations in the MCLs A1 and H2 did not perfectly match the notch bottom and approximately matched a surface trace of the (110) cleavage planes of the cementite indicated by the dashed lines.¹⁷⁾ No significant difference was identified in the fracture location between in air and during the cathodic hydrogen charging. For the MCL H5, the fracture occurred between the free end and the notch. When the grain boundary is extrapolated in the IPF image in Fig. 6(b), it approximately matches the fracture location in the MCL H5. An elongated small defect was also present at the fracture location as indicated by the arrow in the SE image obtained before the microbending tests, which is possibly a grain boundary ferrite. It seems that, for the MCL H5, the grain boundary ferrite served as an initial point, and the grain boundary was fractured.

Fig. 10. SE images of the MCLs A1, H2, and H5 after the microbending tests and that of the MCL H5 before the microbending tests. The dashed line represents a surface trace of the (110) cleavage planes of cementite. The arrow represents an elongated small defect in the MCL H5.

SE images of the fracture surfaces of the MCLs A3 and H3 with the same $\overrightarrow{v}$ value are shown in Fig. 11, that were obtained by tilting the plate specimen by 60°. A river pattern associated with cleavage fractures was seen on these fracture surfaces. The flow direction of the river and the notch bottom are indicated by red arrows and dashed circles in the image, respectively. From the river pattern, it was found that the crack initiates at the notch bottom and then rapidly propagates in the thickness direction of the MCL. No significant difference was identified in the fracture surface morphology between in air and during the cathodic hydrogen charging.

Fig. 11. SE images of fracture surfaces of the MCLs A3 and H3. The red arrows and the dashed circles represent the flow direction of the river and the notch bottoms, respectively. (Online version in color.)

3.6. Surface Damage during Hydrogen Charging

AFM topography images and the corresponding gradient ones of the bulk cementite surfaces obtained in air and during the cathodic hydrogen charging obtained with the EC-NI apparatus are shown in Fig. 12. The surface was very flat, and the cathodic hydrogen charging did not influence its topography.

Fig. 12. AFM topography images and the corresponding gradient ones of the bulk cementite surface obtained in air and during the cathodic hydrogen charging with the EC-NI apparatus. (Online version in color.)

Qualitative and quantitative analysis results of SEM-EDS are shown in Fig. 13 and Table 4, respectively. The values in Table 4 are averages of six data obtained at different positions on the specimen surface. Chemical elements other than carbon, chromium, and iron were not detected both before and after the cathodic hydrogen charging. Fractions of these three elements were also almost the same. Thus, it was confirmed that neither the reduction of the specimen due to the hydrogen nor the chemical damages due to the electrolyte occurs during the cathodic hydrogen charging.

Fig. 13. Qualitative analysis results of SEM-EDS conducted before and after the cathodic hydrogen charging. (Online version in color.)

Table 4. Quantitative analysis result of SEM-EDS of the bulk cementite before and after the hydrogen charging.

	Fraction (wt. %)
	Fe	Cr	C
Before H charge	85.1±0.2	4.2±0.1	10.8±0.2
After H charge	85.1±0.1	4.2±0.1	10.6±0.1

4. Discussion

4.1. Influence of Cathodic Hydrogen Charging on Young’s Modulus

It is known that, in microbending tests of MCLs without a notch, Young’s modulus and a E gradient g of the load-displacement curve in the elastic deformation range has the following relationship,

  

$$g=\frac{3I}{L_1}E,$$(2)

where L₁ is distance between the fixed end and the indentation point as schematically shown in Fig. 7.¹²⁾ The E values were calculated from Eq. (2) using the g values obtained from the load-displacement curves in Fig. 9 and the I and L₁ values in Table 2. In the present study, the MCLs with the notch were used, and Eq. (2) underestimates the Young’s modulus. However, it is possible to qualitatively discuss the change in the Young’s modulus due to the cathodic hydrogen charging. The g and E values of each MCL are summarized in Table 5. The average of E was 75.2±6.0 GPa and 76.8±8.4 GPa in air and during the cathodic hydrogen charging, respectively, and it was found that the cathodic hydrogen charging does not influence the Young’s modulus.

Table 5. g and E values of the MCLs.

Mark	g (μN/nm)	E (GPa)	Average E (GPa)
MCL A1	1.27	77.3	75.2±6.0
MCL A2	1.31	68.5
MCL A3	1.40	79.9
MCL H1	1.25	75.0	76.8±8.4
MCL H2	1.32	74.0
MCL H3	1.47	76.8
MCL H4	1.62	90.6
MCL H5	1.24	67.8

4.2. Influence of Cathodic Hydrogen Charging on Fracture Toughness

As shown in Table 3, the cathodic hydrogen charging increased the fracture load by about 55% compared to in air. In order to quantitatively evaluate how much the toughness was improved, a critical stress intensity factor K_c in mode I for the fracture was calculated following the steps below.

For a MCL with an edge notch and a triangular cross-section, a stress intensity factor K_edge at the edge notch bottom is given by

  

$$K_{edge}=\frac{12P\left(L_1-L_2\right)}{a{b}^2}\times 1.12\sqrt{\pi d},$$(3)

where P and d are the indentation load, and the notch depth, respectively.¹⁸⁾ First, the K_edge value at the fracture (K_edge,C) was calculated from Eq. (3) by substituting the fracture load in Table 3 into P, the dimensions of the MCL in Table 2 into a and b, and the notch depth (200 nm) into d. The K_edge,C value becomes larger than the actual K_C value because the fabricated notch is not edge shape and the radius of curvature of the notch bottom is finite. On the other hand, Norton et al. conducted microbending tests of MCLs of alumina (Al₂O₃) with a notch with different radii of curvature and obtained the relationship between the K_edge,C and K_C values.¹⁸⁾ According to this relationship, for the radius of curvature of 80 nm, K_C is given by,

  

$$K_C=0.55K_{edge,C}.$$(4)

Thus, finally, the K_C values were obtained from Eq. (4) using the calculated K_edge,C values.

In Table 6, the K_edge,C and K_C values of all the MCLs excepting for the MCL H5 are summarized, that was fractured between the free end and the notch. The average of K_C in air was 0.88±0.08 MPa·m^1/2 and close to the value of graphite (~1 MPa·m^1/2).¹⁹⁾ On the other hand, the average of K_C during the cathodic hydrogen charging was 1.29±0.23 MPa·m^1/2. Thus, it was found that the cathodic hydrogen charging increases K_C by about 45% while the absolute value is still small compared to values of steel (100–200 MPa·m^1/2).²⁰⁾

Table 6. K_edge,C and K_C values of the MCLs.

Mark	Kedge,C (MPa·m1/2)	KC (MPa·m1/2)	Average KC (MPa·m1/2)
MCL A1	1.77	0.98	0.88±0.08
MCL A2	1.50	0.82
MCL A3	1.54	0.85
MCL H1	2.01	1.11	1.29±0.23
MCL H2	2.36	1.30
MCL H3	2.08	1.14
MCL H4	2.93	1.61

4.3. Hydrogen Distributions in Specimens

The LT-TDA evaluates average hydrogen content of the entire plate specimen. On the other hand, in the microbending tests, the Young’s modulus is determined by the deflection of the entire MCL and K_C by local deformation near the notch bottom. Thus, it is necessary to consider hydrogen distributions in the specimens to accurately understand the present experimental results.

Diffusion distance x of hydrogen atoms in materials at time t is given by²¹⁾

  

$$x=\sqrt{Dt}.$$(5)

D is a hydrogen diffusion coefficient and described in an Arrhenius equation form,

  

$$D=D_0\text{exp}\left(-\frac{E_a}{k_{B}T}\right),$$(6)

where D₀, E_a, T, and k_B are a diffusion prefactor, a migration energy of the hydrogen atom, temperature, and a Boltzmann’s constant, respectively.²¹⁾ As far as we know, there is no available data for D₀ of cementite. Thus, assuming that D₀ is the same as of iron (0.6×10⁻⁷ m²/s),²²⁾ the hydrogen distributions in the specimen were roughly estimated from Eqs. (5) and (6) using the E_a value (0.57 eV) calculated by Kawakami et al.⁶⁾ and the actual hydrogen charging times: 48 h in the LT-TDA and 4 h in the microbending tests. The x values were calculated to 1.1 μm and 0.3 μm in the LT-TDA and the microbending tests, respectively.

The x values of 1.1 μm is much smaller than the thickness of the plate specimen (0.3 mm). It is very likely that the hydrogen desorption was not detected in the LT-TDA because the hydrogen content in the region near the specimen surface was very low and below the detection lower limit of the LT-TDA. The detection lower limit is about 0.01 wt.ppm for the present plate specimen.

The x values of 0.3 μm is also much smaller than the thickness of the MCL (4.7 μm). The Young’s modulus is determined by the deflection of the MCL as described above. It is likely that the E value was not changed by the cathodic hydrogen charging because the hydrogen atoms are absent in most regions in the MCL. This localized hydrogen distributions also explains the similarity in the fracture surface between in air and the during the hydrogen charging in Fig. 11.

On the other hand, the x values of 0.3 μm is found to be approximately the same as the notch depth (200 nm) of the MCL. This means that the K_C determined by the local deformation near the notch bottom can be changed by the charged hydrogen atoms. In the Section 4.4., the causes of the observed improvement in K_C by the cathodic hydrogen charging will be discussed in detail.

4.4. Cause of Improvement of Fracture Toughness by Cathodic Hydrogen Charging

Both the AFM and SEM-EDS results indicated that neither the reduction of the specimen due to the hydrogen atoms nor the chemical damages due to the electrolyte occurs during the cathodic hydrogen charging (Figs. 12 and 13). On the other hand, it was estimated that hydrogen is charged in the region near the notch, that strongly influences on the fracture behavior, as described in the Section 4.3. Thus, it seems that the observed improvement in K_C by the cathodic hydrogen charging is attributed to the charged hydrogen atoms.

In general, three mechanisms are conceivable for the improvement in K_C by the charged hydrogen atoms: formation of hydride with higher toughness, increase in cohesive energy, and enhancement of dislocation motion. If the dislocation nucleation and glide during the bending are enhanced, the notch apex becomes blunt, increasing the fracture load.

It has been shown by XRD that the cementite does not form hydride in high pressure hydrogen gas up to 14 GPa and 1973 K.⁵⁾ Chromium added to stabilize the cementite phase is also an element with low hydrogen affinity unlike titanium or vanadium.²³⁾ Thus, the hydride formation can be excluded from the cause of the observed improvement in K_C.

When the cohesive energy is large and bonding of the lattice atoms is strong, the dislocation motion is suppressed. That is, plastic deformation is hindered, and the fracture occurs in the elastic deformation range. This contradicts the present experimental results shown in Fig. 9, where plasticity appeared prior to the fracture during the cathodic hydrogen charging. In addition, as far as we know, there are very few studies reporting that the hydrogen atoms increase the cohesive energy, while there are numerous studies reporting that the hydrogen atoms decrease the cohesive energy for many metals.^24,25) It is very unlikely that cohesive energy is exceptionally increased by the hydrogen atoms for the cementite.

On the other hand, the enhancement of dislocation motion explains the appearance of the plasticity during the cathodic hydrogen charging. It is considered that the observed improvement of K_C is attributed to the notch blunting due to the hydrogen-enhanced dislocation motion. Atomic simulations on dislocation motion or TEM observations of dislocation structures under the notch would be helpful to further clarify this phenomenon, which are future works.

5. Summary

The hydrogen absorption characteristics and the mechanical properties in hydrogen environments of cementite were evaluated by the LT-TDA and the in-situ microbending tests during the cathodic hydrogen charging using the bulk cementite obtained through the vacuum carburizing process. Consequently, the following insights were obtained.

(a) In the LT-TDA, the hydrogen desorption spectra were very similar between the hydrogen-charged and hydrogen-removed specimens. Hydrogen desorption from the cementite was not identified at temperatures between 173 K and 1073 K.

(b) In the in-situ microbending tests, the cathodic hydrogen charging increased the fracture load by 55%. The fractures occurred in the elastic deformation range in air and in the elastoplastic range during the cathodic hydrogen charging.

(c) The fracture location corresponded to the surface trace of the (110) cleavage planes of the cementite, and the river pattern associated with the cleavage fracture was observed on the fracture surfaces. No significant difference was identified in the fracture surfaces between in air and during the cathodic hydrogen charging.

(d) From the obtained load-displacement curves during the bending, E were calculated to 75.2±6.0 GPa and 76.8±8.4 GPa in air and during the cathodic hydrogen charging, respectively, On the other hand, K_C was calculated to 0.88±0.08 MPa·m^1/2 and 1.29±0.23 MPa·m^1/2 in air and during the cathodic hydrogen charging, respectively. Thus, the cathodic hydrogen charging did not change the Young’s modulus while it improved K_C by 45%.

(e) It was estimated that the hydrogen atoms are charged only near the specimen surface in the present cathodic hydrogen charging conditions. It seems that no hydrogen desorption from the cementite is identified because the hydrogen atoms are absent in the most regions of the specimens. The invariance of the Young’s modulus and the fracture surface morphology can be explained by the same reason. On the other hand, it seems that K_C was improved by the cathodic hydrogen charging because the charged hydrogen atoms near the notch bottom enhance the dislocation nucleation and glide and cause the notch blunting.

References

• 1)   W. H.  Johnson: Proc. Royal Soc. London, 23 (1875), 168. https://doi.org/10.1098/rspl.1874.0024
• 2)   R. A.  Oriani: Ann. Rev. Mater. Sci., 8 (1978), 327. https://doi.org/10.1146/annurev.ms.08.080178.001551
• 3)   H. K.  Birnbaum and  P.  Sofronis: Mater. Sci. Eng. A, 176 (1993), 191. https://doi.org/10.1016/0921-5093(94)90975-X
• 4)   M.  Nagumo: ISIJ Int., 41 (2001), 590. https://doi.org/10.2355/isijinternational.41.590
• 5)   H.  Terasaki,  Y.  Shibazaki,  K.  Nishida,  R.  Tateyama,  S.  Takahashi,  M.  Ishii,  Y.  Shimoyama,  E.  Ohtani,  K.  Funakoshi and  Y.  Higo: ISIJ Int., 54 (2014), 2637. https://doi.org/10.2355/isijinternational.54.2637
• 6)   K.  Kawakami and  T.  Matsumiya: ISIJ Int., 53 (2013), 709. https://doi.org/10.2355/isijinternational.53.709
• 7)   N.  Adachi,  H.  Ueno,  K.  Onoe,  S.  Morooka and  Y.  Todaka: ISIJ Int., 61 (2021), 2320. https://doi.org/10.2355/isijinternational.ISIJINT-2021-134
• 8)   M.  Umemoto,  Z. G.  Liu,  H.  Takaoka,  M.  Sawakami,  K.  Tsuchiya and  K.  Masuyama: Metall. Mater. Trans. A, 32 (2001), 2127. https://doi.org/10.1007/s11661-001-0024-y
• 9)   K.  Tomatsu,  T.  Amino,  T.  Chida,  S.  Uji,  M.  Okonogi,  H.  Kawata,  T.  Omura,  N.  Maruyama and  Y.  Nishiyama: ISIJ Int., 58 (2018), 340. https://doi.org/10.2355/isijinternational.ISIJINT-2017-461
• 10)   M.  Pinson,  H.  Springer,  T.  Depover and  K.  Verbeken: Mater. Sci. Eng. A, 859 (2022), 144204. https://doi.org/10.1016/j.msea.2022.144204
• 11)   T  Morita and  T  Hanyuda: Tetsu-to-Hagané, 92 (2006), 268 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.92.4_268
• 12)   K.  Tomatsu,  H.  Kawata,  T.  Amino,  T.  Omura,  N.  Maruyama and  Y.  Nishiyama: ISIJ Int., 57 (2017), 564. https://doi.org/10.2355/isijinternational.ISIJINT-2016-540
• 13)   T.  Hajilou,  Y.  Deng,  B. R.  Rogne,  N.  Kheradmand and  A.  Barnoush: Scr. Mater., 132 (2017), 17. https://doi.org/10.1016/j.scriptamat.2017.01.019
• 14)   M.  Egashira,  T.  Tsuchiyama and  S.  Takai: Mater. Sci. Forum., 426–432 (2003), 4239. https://doi.org/10.4028/www.scientific.net/MSF.426-432.4239
• 15)   M.  Nagumo,  K.  Takai and  N.  Okuda: J. Alloys and Compound, 293–295 (1999), 310. https://doi.org/10.1016/S0925-8388(99)00322-9
• 16)   D. E. J.  Armstrong,  A. J.  Wilkinson and  S. G.  Roberts: J. Mater. Res., 24 (2009), 3268. https://doi.org/10.1557/jmr.2009.0396
• 17)   A.  Inoue,  T.  Ogura and  T.  Masumoto: Trans. Jpn. Inst. Metals, 17 (1976), 663. https://doi.org/10.2320/matertrans1960.17.663
• 18)   A. D.  Norton,  S.  Falco,  N.  Young,  J.  Severs and  R. I.  Todd: J. Eur. Ceram. Soc., 35 (2015), 4521. https://doi.org/10.1016/j.jeurceramsoc.2015.08.023
• 19)   M.  Sakai,  Y.  Goto and  M.  Inagaki: Yogyo-Kyokai-Shi, 94 (1986), 163. https://doi.org/10.2109/jcersj1950.94.163
• 20)   J.  Yamabe,  H.  Itoga,  T.  Awane,  T.  Matsuo,  H.  Matsunaga and  S.  Matsuoka: J. Pressure Vessel Technol., 138 (2016), 011401. https://doi.org/10.1115/1.4030086
• 21)   M.  Nagumo: Fundamentals of Hydrogen Embrittlement, Springer, Singapore, (2016), 70. https://doi.org/10.1007/978-981-10-0161-1
• 22)   W.  Beck, J. O’. Bockris, J. McBreen and L. Nanis: Proc. Roy. Soc., A 290 (1966), 220. https://doi.org/10.1098/rspa.1966.0046
• 23)   H.  Mizoguchi,  S.-W.  Park and  H.  Hosono: J. Am. Chem. Soc., 143 (2021), 11345. https://doi.org/10.1021/jacs.1c06307
• 24)   A. R.  Troiano: Metall. Micro. Anal., 5 (2016), 557. https://doi.org/10.1007/s13632-016-0319-4
• 25)   R. A.  Oriani: Corrosion, 43 (1987), 390. https://doi.org/10.5006/1.3583875