Reduction of Delayed Fracture Susceptibility of Tempered Martensitic Steel through Increased Si Content and Surface Softening

Abstract

Improvement of the surface layer as well as the microstructure has been needed to develop high-strength steels, since delayed fracture cracks initiate in the surface layer. In the present study, two approaches were taken to reduce the delayed fracture susceptibility of tempered martensitic steel with tensile strength of 1450 MPa. One was by increasing the Si content, which was intended to improve the microstructure. The other was by a surface-softening treatment, which was for improving the surface layer. Delayed fracture susceptibility was evaluated by conducting constant strain rate tensile tests (tensile tests) and constant load tests in a NH₄SCN aqueous solution. It was found that increasing the Si content from 0.2 mass% to 1.88 mass% prevented intergranular fracture and reduced delayed fracture susceptibility. One reason for this improvement is that the Fe₃C particle size on prior-γ grain boundaries and in the matrix decreases with increasing Si content, which implies that Si stabilizes dislocation structures. When the surface strength of surface-softened steel specimens was lowered to 1150 MPa, delayed fracture susceptibility was reduced further. This is attributed to not only a reduction of the Vickers hardness of the surface layer but also a reduced hydrogen concentration at the surface layer. The rearrangement and annihilation of dislocations and also the spheroidizing and coarsening of Fe₃C particles at the surface layer subjected to a high tempering temperature lead to a reduction of the hydrogen concentration at the surface layer.

1. Introduction

If stress is applied to high-strength steel in an atmospheric corrosion environment, delayed fracture sometimes occurs. Delayed fracture is caused by hydrogen entry through corrosion. Corrosion, moreover, often contributes to pit formation. Enhanced stress concentration ahead of the pits gives rise to crack initiation,^1,2) resulting in final fracture. Consequently, there is a need to develop steel that does not fracture even if pits form in addition to hydrogen entry. Thus, improving the surface layer^3,4,5,6,7) where cracks initiate as well as the microstructures of steel has been proposed to reduce delayed fracture susceptibility.

Various ways of improving the microstructures have been proposed: strengthening grain boundaries by adding boron,^8,9) trapping hydrogen by adding vanadium^10,11,12) and stabilizing dislocation structures through high-temperature tempering.¹³⁾ While boron reduces tempering embrittlement caused by phosphorus or sulfur segregating near grain boundaries,¹⁴⁾ there are inconsistent reports regarding the effect of boron on hydrogen-induced intergranular fracture. Some researchers have reported that boron prevents intergranular cracking caused by hydrogen,^8,9) while others have reported that it does not.¹⁵⁾ As for adding vanadium, vanadium carbide, which precipitates under high-temperature tempering, renders hydrogen harmless by trapping it strongly.¹⁰⁾ Fine vanadium carbide, in addition to its trapping effect, stabilizes dislocation structures,¹³⁾ thereby reducing delayed fracture susceptibility. Since high-temperature tempering has to be done to precipitate vanadium carbide, it is difficult to separate the effect of trapping hydrogen from that of the tempering process. Another concern about adding vanadium is the hydrogen content increment due to large binding energy between hydrogen and vanadium carbide. Hence, it is worth gambling on the possibility that some techniques such as high-temperature tempering can reduce delayed fracture susceptibility without enhancing hydrogen absorption. For example, it is known that adding silicon raises the tempering temperature^{16,17,18,19,20)} and reduces delayed fracture susceptibility.²¹⁾ However, it is not well known why silicon reduces delayed fracture susceptibility because increasing the Si content brings corresponding changes in tensile strength and heat treatment conditions.

Regarding the improvement of the surface layer, decarburization reduces delayed fracture susceptibility by lowering the surface hardness.^3,4) However, it remains to be proven whether the dominant factor is the surface hardness or carbon content. Furthermore, the distribution of the hydrogen content, which can correspond to hardness, has yet to be measured experimentally.

In this study, we examined both improving the microstructure and surface layer in order to reduce delayed fracture susceptibility of 1450 MPa-class tempered martensitic steel. As for the improvement of the microstructure, we evaluated the effect on delayed facture susceptibility of high-temperature tempering by increasing the silicon content. The reason why increasing the silicon content reduces delayed fracture susceptibility is discussed from the view point of the stability of dislocation structures and the distribution of Fe₃C particles. With regard to the improvement of the surface layer, we manufactured prototype surfaces-oftened steel specimens having a reduced surface layer tensile strength of 1150 MPa, using induction heating, which hardly changes the chemical composition. The softened surface layer was expected to reduce delayed fracture susceptibility further.

2. Experimental Procedure

2.1. Specimen Properties

Steel bars for prestressed concrete as specified in JIS G 3137 were prepared as specimens. Table 1 shows the chemical compositions of a low Si (L-Si) and a high Si (H-Si) steel specimen containing different amounts of silicon and boron. Both the L-Si and H-Si specimens were induction-quenched and -tempered to have tensile strength, σ_B, of around 1450 MPa. For a surface-softened H-Si specimen, the holding time at the maximum tempering temperature for heat transfer to the center was shortened in order that the tempering temperature would be much higher only near the surface. The heat-treatment cycles are shown in Fig. 1. Table 2 shows the maximum temperatures at the surface during tempering and the tensile strength of the specimens. The tempering temperature of the H-Si specimen at 610°C was much higher than that of the L-Si specimen at 350°C, although their tensile strengths were almost the same. On the other hand, the surface-softened H-Si specimen was tempered at a higher temperature of 755°C at the surface and the tempering temperature was not uniform. The holding time at the maximum tempering temperature was shortened as shown in Fig. 1 in order to vary the tempering temperature or the hardness between the surface and center of the specimen. Figure 2 shows the Vickers hardness distributions of the H-Si specimen and the surface-softened H-Si specimen. While the H-Si specimen had a uniform hardness distribution from the surface to the center, the surface-softened H-Si specimen was the softest at the surface and had a softened surface layer of 1 mm in depth. Because the center of the surface-softened H-Si specimen was slightly harder than the H-Si specimen, the tensile strength of former specimen was around 1450 MPa, which was the same as that of the latter specimen.

Fig. 1.

Schematic diagram of heat-treatment cycles of L-Si, H-Si and surface-softened H-Si specimens.

Fig. 2.

Vickers hardness from the surface to the center of H-Si and surface-softened H-Si specimens.

Table 1. Chemical compositions of specimens (mass%).

	C	Si	Mn	P	S	Cu	Ti	B
L-Si	0.30	0.21	0.72	0.019	0.002	0.01	0.03	0.0016
H-Si	0.35	1.88	0.73	0.015	0.008	0.02	0.03	0.0024

Table 2. Quenching temperatures, tempering temperatures and tensile strengths of various specimens.

Specimen	Quenching temp. [°C]	Tempering temp. [°C]	Tensile strength [MPa]
L-Si	960	350	1445
H-Si	1010	610	1458
Surface-softened H-Si	1010	755	1440

To compare the microstructures of the three kinds of steel specimens, we immersed mirror-polished specimens in nital containing 3 mass% HNO₃ and observed them using a scanning electron microscope (SEM). Furthermore, we made thin-film samples by a twin-jet electropolishing method and observed them using a transmission electron microscope (TEM) for the purpose of making a detailed examination of carbide precipitation and dislocation structures.

2.2. Delayed Fracture Susceptibility

The delayed fracture susceptibility of the three kinds of specimens was evaluated by conducting tensile tests and constant load tests. Tempered steel bars of 7.2 mm in diameter were notched circumferentially. The geometry and the dimensions of the specimens are shown in Fig. 3. The notch was 0.4 mm in depth (stress concentration factor, K_t, was 2.1), which was intended to imitate a pit formed by atmospheric corrosion. The notch was shallower than the softened surface layer, the depth of which was around 1.0 mm, as shown in Fig. 2. In these tests, nominal stress, σ, was defined as σ = F / A_min, where F is the tensile load and A_min is the net cross-sectional area of the notch in the specimen.

Before the tensile test, a specimen was immersed in a NH₄SCN aqueous solution with a concentration of from 0.5 mass% to 2.0 mass% at a temperature of 50°C for 48 h for hydrogen pre-charging. The tensile test was then carried out in a NH₄SCN aqueous solution with the same concentration as that for pre-charging in order to keep the hydrogen content in the specimen constant during the test. The crosshead speed of the tensile tests was varied from 0.05 to 20 mm/min. In the constant load tests, the constant nominal stress was varied from 0.4 to 0.95 σ_B and was applied in a NH₄SCN solution with a concentration of 2.0 mass% at a temperature of 50°C without hydrogen pre-charging. The time to fracture was recorded. If a specimen did not fracture, the test was discontinued after 100 h.

Fig. 3.

Geometry and dimensions of notched specimens for tensile testing.

2.3. Distribution of Hydrogen Content in Surface-softened Steel

In order to obtain the distribution of the hydrogen content in the surface-softened H-Si steel, we manufactured four kinds of specimens the tensile strength of which was varied as follows by changing the tempering temperature: a 1150 MPa (steel-1150), a 1350 MPa (steel-1350), a 1450 MPa (steel-1450) and a 1500 MPa (steel-1500) specimen. Each specimen had tensile strength equivalent to the surface and the center or two points between the surface and the center of the surface-softened H-Si specimen. Table 3 shows the tensile strength and Vickers hardness, 0.5HV, of the specimens. Vickers hardness of the steel-1150 specimen was 370, which was equivalent to the surface hardness of the surface-softened H-Si specimen; that of the steel-1500 specimen was 510, which was equivalent to the center of the surface-softened H-Si specimen. Vickers hardness of the steel-1350 specimen was 450 and that of the steel-1450 specimen was 480. These values were equivalent to a point between the surface and the center. These specimens were immersed in a 20 mass% NH₄SCN aqueous solution at a temperature of 50°C for 48 h. The hydrogen concentration was then measured by thermal desorption analysis (TDA) using a gas chromatograph at a heating rate of 100°C/h in the temperature range from room temperature to 300°C. A standard gas mixture of Ar +50 vol. ppm of H₂ was used for calibration of the hydrogen concentration. Assuming that the hydrogen content is the same for steels having the same chemical composition and strength, the hydrogen contents thus obtained from the steel-1150, -1350, -1450, and -1500 specimens corresponded to those of the points from the surface to the center of the surface-softened H-Si specimen. For example, the hydrogen content of the steel-1150 specimen was regarded as the hydrogen content at the surface of the surface-softened H-Si specimen. In this way, we determined the distribution of the hydrogen concentration in the surface- softened H-Si specimen.

Table 3. Tensile strengths and Vickers hardnesses of four kinds of H-Si specimen changed in strength corresponding to Vickers hardness from the surface to the center of the surface-softened H-Si specimen.

	H-Si				Surface-softened H-Si
	1150	1350	1450	1500
Tensile strength [MPa]	1156	1363	1458	1502	1440
Vickers hardness (HV0.5)	370	450	480	510	*

*  Refer to Fig. 2.

2.4. Stability of Dislocation Structures

The stability of dislocation structures was evaluated by conducting stress relaxation tests according to JIS Z 2276 and JIS G 3137. First, smooth bar specimens were loaded and held at the initial load for 120 s. The initial applied load, F_I, was 0.7σ_Bs·A_nom, where σ_Bs is the standardized tensile strength of 1420 MPa, and A_nom is the nominal cross-sectional area. After 120 s, the distance between the chucks was kept constant for 200 h as the reduction of the load was recorded, and the relationship between the load and elapsed time was plotted on a double logarithmic chart. The load after 1000 h was estimated by extrapolation. The relaxation value was defined as the estimated reduction rate (%) of the load after 1000 h.

3. Results

3.1. Delayed Fracture Susceptibility Evaluated by Tensile Tests

Figure 4 shows the stress-displacement curves at a strain rate of 0.1 mm/min and initial hydrogen content of specimens immersed in a 2.0 mass% NH₄SCN solution at 50°C. Because there was no difference among the stress-displacement curves of the three kinds of uncharged specimens, only that of the surface-softened H-Si specimen is shown, representing the three uncharged specimens. Hydrogen charging decreases fracture stresses. The L-Si specimen showed the largest decline in fracture stress due to hydrogen charging. The H-Si specimen displayed the second largest decline and the fracture stress and the ductility of the surface-softened H-Si specimen declined only a little. The fracture surfaces near the notch root of these hydrogen-charged specimens are shown in Fig. 5. While the L-Si specimen exhibited typical intergranular fracture, the H-Si and surface-softened H-Si specimens showed typical quasi-cleavage fracture. The fracture surface apart from the notch root was microvoid coales-cence regardless of the specimen type.

Fig. 4.

Stress-displacement curves of specimens with/without hydrogen at a strain rate of 0.1 mm/min and hydrogen contents of specimens immersed in a 2.0 mass% NH₄SCN solution before tensile testing.

Fig. 5.

Microscopic fracture surfaces near the notch tip after tensile testing in a 2.0 mass% NH₄SCN solution at a strain rate of 0.1 mm/min; (a) L-Si specimen shows intergranular fracture, (b) H-Si specimen shows quasi-cleavage fracture and (c) surface-softened H-Si specimen also shows quasi-cleavage fracture.

Figure 6 shows the relationship between the fracture stress obtained in the tensile tests in the 2.0 mass% NH₄SCN aqueous solution and the crosshead speed in a range from 0.02 to 20 mm/min. The fracture stress of the uncharged specimens was independent of the crosshead speed and was constant. In contrast, the fracture stress of the hydrogencharged specimens decreased with decreasing crosshead speed. At a crosshead speed of 20 mm/min, the fracture stress of the hydrogen-charged specimens was almost the same as that of the uncharged specimens, and there was no difference among the three kinds of specimens. Meanwhile, at a crosshead speed of 0.02 mm/min, the effect of hydrogen was the most obvious; the difference in fracture stress among the three kinds of specimens was the most pronounced. The specimen order of delayed fracture susceptibility was L-Si > H-Si > surface-softened H-Si.

Fig. 6.

Relationship between fracture stress and crosshead speed of uncharged and hydrogen-charged specimens immersed in a 2.0 mass% NH₄SCN solution.

Figure 7 shows the macroscopic fracture surfaces after tensile tests at various crosshead speeds with hydrogen charging. The area affected by hydrogen, which was intergranular (IG) in the L-Si specimen and quasi-cleavage (QC) in the H-Si and the surface-softened H-Si specimens, is highlighted in Fig. 7. The specimen order of the size of the brittle-like fracture surface, i.e., QC or IG, was L-Si > H-Si > surface-softened H-Si, which corresponded to that of delayed fracture susceptibility in Fig. 6. The brittle-like fracture became larger with decreasing crosshead speed. At 20 mm/min, where fracture stress did not decline, the fracture surface was not brittle but ductile, i.e., microvoid coalescence. This increase in brittle-like fracture with decreasing crosshead speed corresponded to the reduction of fracture stress as the crosshead speed decreased, as shown in Fig. 6.

Fig. 7.

Macroscopic fracture surfaces of specimens after tensile testing in a 2.0 mass% NH₄SCN solution at various crosshead speeds. Intergranular or quasi-cleavage fracture area is highlighted.

Figure 8 shows the relationship between fracture stress and hydrogen content at a constant crosshead speed of 0.1 mm/min. Facture stress decreased with increasing hydrogen content. At any hydrogen content, the specimen order of delayed facture susceptibility was always L-Si > H-Si > surface-softened H-Si.

Fig. 8.

Relationship between fracture stress and hydrogen content obtained by tensile testing in a NH₄SCN solutions with various concentrations at a strain rate of 0.1 mm/min.

3.2. Delayed Fracture Susceptibility Evaluated by Constant Load Tests

Figure 9 shows the relationship between the applied stress ratio and the time to fracture as found in constant load tests conducted in a 2.0 mass% NH₄SCN aqueous solution. The surface-softened specimen did not fracture after 100 h even if the applied stress was as large as 0.95 σ_B. For the H-Si specimen and the L-Si specimen, meanwhile, the time to fracture declined as the applied stress increased. When critical stress is defined as the maximum stress without fracture after 100 h, the critical stress of the surface-softened HSi specimen was above 0.95σ_B, that of the H-Si specimen was around 0.5σ_B, and that of the L-Si specimen was around 0.35σ_B. Hence, the specimen order of critical stress as found in the constant load tests was surface-softened H-Si > H-Si > L-Si; the specimen order of delayed fracture susceptibility was L-Si > H-Si > surface-softened H-Si.

Fig. 9.

Relationship between applied stress ratio and time to fracture obtained by constant load testing in a 2.0 mass% NH₄SCN solution.

In summary, the specimen orders of delayed fracture susceptibility in the tensile tests and the constant load tests were consistent with each other under the same hydrogen charging condition. However, the fracture stress values in the tensile tests and the constant load tests were not the same when hydrogen charging was conducted by immersion in a NH₄SCN aqueous solution. The reason for that difference will be discussed in a future paper.

3.3. Observation of Fe₃C and Dislocation Structure

Figure 10 shows the microstructures of the specimens, focusing especially on the nature of carbide precipitation and dislocation structures, as observed by SEM and TEM. In the L-Si specimen, there are plate-like Fe₃C particles with a length of 400 to 1000 nm on prior-γ grain boundaries and plate-like Fe₃C particles with a length of 200 to 300 nm in the matrix as shown in the upper row (SEM) and the middle row (TEM). In the H-Si specimen, on the other hand, there are thinner film-like Fe₃C particles on some portions of prior- γ grain boundaries as shown in the middle row (TEM). Additionally, the Fe₃C particles in the matrix of the H-Si specimmen are finer than those in the L-Si specimen. Thus, the H-Si specimen is characterized by fine and thin Fe₃C particles on prior-γ grain boundaries and fine Fe₃C particles in the matrix. For the surface-softened H-Si specimen, the observed area was in the softened surface layer at a depth of 0.4 mm from the surface. Because of the higher tempering temperature of 755°C at the surface, Fe₃C particles both on prior-γ grain boundaries and in the matrix spheroidized, as seen in the scanning and transmission electron micrographs.

Fig. 10.

Microstructures of specimens obtained by scanning electron microscopy and transmission electron microscopy. The micrographs of the surface-softened H-Si specimen were obtained in a surface-softened area at a depth of 0.4 mm from the surface.

Transmission electron micrographs focusing on dislocation structures are shown in the lower row of Fig. 10. All the specimens have such high dislocation densities that it is difficult to find distinct differences among them. However, while the L-Si and H-Si specimens display random dislocation structures, the surface-softened H-Si specimen exhibits a cell structure of 50 to 200 nm in size and with a lower dislocation density.

4. Discussion

4.1. Effect of Si and B on Delayed Fracture Susceptibility

The delayed fracture susceptibility of the H-Si specimen was lower than that of the L-Si specimen, as shown in Figs. 4, 6, 8 and 9. Corresponding to that, the H-Si specimen showed quasi-cleavage fracture while the L-Si specimen showed intergranular fracture as shown in Figs. 5 and 7. The two specimens differed not only in their silicon content but also in their boron content. We therefore tried to separate the effect of boron from that of silicon.

To do that, four other kinds of specimens were prepared: a HSi+B specimen contained 1.80% Si and 0.0023% B; a LSi+B specimen contained 1.03% Si and 0.0022% B; a HSi specimen contained 1.67% Si and a LSi specimen contained 0.26% Si. The HSi+B specimen had boron and much silicon. The LSi+B specimen had boron and a little silicon. The HSi specimen had no boron but much silicon, and the LSi specimen had no boron but a little silicon. The chemical compositions except for silicon and boron were almost the same among these four specimens. These specimens were induction-quenched and -tempered to have tensile strength of around 1450 MPa. They were immersed in a 2.0 mass% NH₄SCN aqueous solution at 50°C for 48 h for hydrogen charging; then, they were subjected to tensile tests at a crosshead speed of 0.1 mm/min. The results revealed that the specimens containing much silicon (the HSi+B and HSi specimen) fractured at a higher stress level of 1420 MPa and 1665 MPa, respectively. The specimen containing little silicon (the LSi+B specimen) fractured at a lower stress level of 972 MPa, although it contained boron. Figure 11 shows the fracture surfaces near the notch root after the tensile tests. For the specimens containing much silicon of over 1.67% in the present study, their fracture surface was quasicleavage regardless of whether they contained boron or not. The specimens with a small Si content, on the other hand, exhibited intergranular fracture regardless of whether they contained boron or not. These results indicated that the reduction of delayed fracture susceptibility and prevention of intergranular fracture susceptibility was mainly due to the higher silicon content. This is consistent with a previous study¹⁸⁾ that indicated silicon can prevent intergranular fracture. One reason why boron had no effect on delayed fracture susceptibility in the present study may be that intergranular fracture was caused by something other than impurity segregation, such as phosphorus, because impurities are not likely to segregate during the short heat treatment time of induction heating.

Fig. 11.

Effects of Si and B on fracture surface after tensile testing in a 2.0 mass% NH₄SCN solution. The specimens with a high Si content show quasi-cleavage fracture regardless of whether B was added or not.

4.2. Mechanism of Reduction of Delayed Fracture Susceptibility by Increased Si

The reason why increasing the silicon content prevents intergranular fracture and reduces delayed fracture susceptibility is discussed here. Increasing the silicon content can raise the tempering temperature without lowering tensile strength, owing to two functions of silicon: solid solution strengthening and resistance to tempering softening. In fact, the tempering temperature of the H-Si specimen at 610°C was 260°C higher than that of the L-Si specimen at 350°C. The higher tempering temperature presumably changes the microstructure as follows;

(i) Stabilizing dislocation structures.

(ii) Changing the form of carbide precipitation.

With regard to (i) stabilizing dislocation structures, stress relaxation tests are one of the methods for evaluating the stability of dislocation structures. Three kinds of specimens with different Si content (0.21, 1.03 and 1.80 mass% Si) were prepared for use in stress relaxation tests. They were all tempered at various temperatures (350, 478 and 580°C) to have tensile strength of around 1450 MPa. Figure 12 shows the relationship between the stress relaxation value and Si content at a temperature of 20° ± 0.5°C. As the Si content increased, the stress relaxation value declined; however, above a silicon content of 1 mass%, the relaxation value converged to a constant value. Kawasaki et al. have also reported that increasing the Si content reduces the stress relaxation value obtained in relaxation tests at a high temperature of 180°C.²²⁾ Since the relaxation value is associated with moving distance and the number of moving dislocations, the reduction of the relaxation values corresponds to the stabilization of dislocation structures. Hence, these findings indicate that increasing the Si content, i.e., high-temperature tempering can stabilize dislocation structures under constant tensile strength.

Fig. 12.

Correlation between stress relaxation value at a temperature of 20°C and Si content in martensitic steels with a tensile strength of 1450 MPa.

Although the relationship between the stability of dislocation structures and delayed fracture susceptibility is not well understood, it is known that delayed fracture is related to the slip of dislocations. One of the authors has reported that moving dislocations on slip planes in iron and Ni-based alloys drag hydrogen during plastic deformation especially at slow strain rates.²³⁾ This interaction between hydrogen and dislocations results in the formation of vacancies and vacancy clusters during the cutting of screw dislocations in the presence of hydrogen.²⁴⁾ A positron annihilation lifetime measurement and hydrogen measurement as a probe have substantiated that hydrogen enhances the formation of vacancy clusters in tempered martensitic steels as time passes, also under a constant elastic stress.²⁵⁾ These studies provide an explanation of quasi-cleavage fracture, which follows local plastic deformation. Nagumo and Matsuda have reported that intergranular fracture is also involved with the hydrogen-enhanced formation of lattice defects.²⁶⁾ The enhanced vacancy clusters can grow to be micro-voids that act as embryos for crack initiation and can reduce resistance to crack propagation. McMahon and Kameda have also reported that dislocations piled up at the second-phase particles serve as nucleus for intergranular crack initiation.²⁷⁾ Therefore, stabilized dislocation structures may prevent intergranular fracture by inhibiting the piling up of dislocations.

As for (ii) changing the form of carbide precipitation, increasing the silicon content produces finer Fe₃C participates both on grain boundaries and in the matrix, as shown in Fig. 10. Some studies show that Si diffusion controls the transformation of ε-carbides to Fe₃C during tempering when the Si content is large.²⁰⁾ Formation of Fe₃C is hindered as a result. On the other hand, the relationship between delayed fracture susceptibility and the presence of carbides on grain boundaries has been studied.^28,29,30) Because second-phase particles on grain boundaries prevent dislocations from slipping, coarse carbide particles on grain boundaries increase the concentration of stress or strain.³¹⁾ Thus, finer Fe₃C particles on grain boundaries should reduce the concentration of stress and strain, resulting in a reduction of delayed fracture susceptibility.

Regarding the effect of finer Fe₃C particles in the matrix on delayed fracture susceptibility, finer particles in the matrix, for instance, can lead to a reduction of the relaxation value. If the total amount of carbide precipitated in the L-Si and H-Si specimens was the same, finer carbides would mean a higher number density of Fe₃C particles, in other words, a smaller distance between the particles. Since the shear stress required for bypass through the second-phase particles is inversely proportional to the distance between particles, finer Fe₃C particles prevent dislocations from slipping, i.e., stabilize dislocation structures. Thus, finer Fe₃C particles in the matrix can act as a factor that reduces delayed fracture susceptibility.

The mechanism of reducing delayed fracture susceptibility by increasing the Si content is summarized in Fig. 13.

Fig. 13.

Reduction in delayed fracture susceptibility of tempered martensitic steel by Si addition.

4.3. Mechanism of Reduction of Delayed Fracture Susceptibility by Surface Softening

When the surface of the H-Si specimen was softened, delayed fracture susceptibility was reduced further, as shown in Figs. 4, 6, 8 and 9. The reason for that will be discussed here. Since surface softening is done through rapid heating, changes in the chemical composition are negligibly small. Consequently, the reason for the reduction of delayed fracture susceptibility was not due to changes in the chemical composition but to the reduction of hardness. In general, steels with tensile strength above 1200 MPa are susceptible to delayed fracture. Hence, the main factor in the reduction of delayed fracture susceptibility was presumably the lowering of tensile strength to around 1150 MPa at the surface. Possible reasons why the softened surface layer reduced delayed fracture susceptibility include:

(a) Reduction of hydrogen concentration in the softened surface layer.

(b) Blunting of the notches.

Regarding possible reason (a), the fractographs in Fig. 7 show that cracks initiated near the notch root in the softened surface layer. Cracks may initiate at the surface of the notch root, where strain is the largest, or inside the notch root where hydrostatic stress is the highest. We cannot detect the exact point of crack initiation. However, it is certain that the crack initiation point was very near the notch root. Thus, the distribution of the hydrogen concentration in the softened surface layer becomes an important factor.

First, we used a least-squares technique to describe the Vickers hardness distribution of the surface-softened H-Si specimen as follows:   

$$\text{HV}\approx 172\text{d}+329\mathrm{\hspace{0.17em} }(0\le \text{d}<1.0),\mathrm{\hspace{0.17em} }\text{HV}\approx 510\mathrm{\hspace{0.17em} }(1.0\le \text{d}<3.6)$$(1)

where HV is Vickers hardness and d is the depth from the surface (mm).

Using Eq. (1), we determined d corresponding to the hardness of the H-Si specimens of various strength in Table 3: a steel-1150, -1350, -1450 and -1500 specimen. The equilibrium content of hydrogen in these specimens was then measured by TDA. The relationship between the hydrogen content and d was plotted to obtain the hydrogen content distribution. The hydrogen contents of these H-Si specimens of various strengths and the surface-softened H-Si specimen are shown in Table 4, and the hardness and hydrogen content distribution are shown in Fig. 14. The layer from the surface to a depth of 0.7 mm had a hydrogen distribution that was 1.0 mass ppm lower than that at the center. In other words, surface softening reduced the hydrogen content near the surface by 60%.

Fig. 14.

Hydrogen content and Vickers hardness as a function of distance from the surface of surface-softened H-Si specimens.

Table 4. Hydrogen contents of H-Si specimens with various tensile strengths and the surface-softened H-Si specimen immersed in a 20 mass% NH₄SCN solution.

	H-Si				Surface-softened H-Si
	1150	1350	1450	1500
Hydrogen content [mass ppm]	1.5	1.5	1.9	2.5	2.1

This reduction of the hydrogen concentration can be attributed to the higher tempering temperature of 755°C only near the surface. The higher tempering temperature caused the spheroidizing and coarsening of Fe₃C particles, thereby reducing the area of the carbide/matrix interfaces acting as hydrogen trapping sites.³²⁾ Other effects of a higher tempering temperature include the rearrangement and annihilation of dislocation structures,³³⁾ which can lead to a reduction of the hydrogen content since dislocations serve as hydrogen trapping sites. In fact, spheroidized Fe₃C particles and a reduced dislocation density were observed in the softened surface layer, as shown in Fig. 10.

Regarding reason (b) above, the blunting of the notches, lower yield stress of the surface layer can cause such blunting owing to preferential plastic deformation near the surface. If that happens, the stress concentration factor could decline, i.e., the maximum local stress would be lower. To examine this hypothesis, we observed the change in the shape of the notches. Notched specimens were loaded to the maximum load and then unloaded. The cross-sectional shapes of the notches were observed by SEM. The results showed there was no difference in the notch shape between the surface-softened H-Si specimen and the H-Si specimen. However, the surface of commercially used steel actually has pits and flaws of various shapes; some of them may be blunted by surface softening.

5. Conclusions

The results obtained regarding the effect of increasing the silicon content and surface softening on delayed fracture susceptibility of tempered martensitic steel with tensile strength of around 1450 MPa can be summarized as follows:

(1) The results of tensile tests for measuring fracture stress indicated that delayed fracture susceptibility was reduced when the Si content was increased from 0.2 mass% to 1.88 mass%. The delayed fracture susceptibility of steel containing 1.88 mass% Si was reduced further when the tensile strength of its surface was lowered to around 1150 MPa. The difference in delayed fracture susceptibility among the three types of steel tested became more obvious as the crosshead speed was reduced. The specimen order of delayed fracture susceptibility as evaluated in constant load tests was consistent with that found in tensile tests.

(2) When the Si content was increased from 0.2 mass% to 1.88 mass%, the fracture mode changed from intergranular to quasi-cleavage. This indicates that increasing the silicon content prevents intergranular fracture. Adding boron, on the other hand, did not affect the fracture mode in this study.

(3) Raising the tempering temperature by increasing the Si content reduces the stress relaxation value, i.e., stabilizes dislocation structures. This is one reason for the reduction of delayed fracture susceptibility by increasing the silicon content. Another reason is the presence of finer Fe₃C particles both on grain boundaries and in the matrix.

(4) The distributions of Vickers hardness and hydrogen concentration were measured for surface-softened steel specimens. Hardness was lower from the surface to a depth of around 1 mm. The hydrogen concentration was 60% lower in the surface layer, the depth of which was 0.7 mm, than at the center. There are two reasons for that reduction. The first is that Fe₃C particles spheroidized and coarsened. The second is that dislocations were rearranged and annihilated near the surface. Thus, the reduced delayed fracture susceptibility of surface-softened steel can be attributed to not only lower hardness but also a reduction in the hydrogen concentration.

References

• 1)   T.  Nakayama,  F.  Yuse,  Y.  Tsubokawa and  J.  Matsui: ISIJ Int., 43 (2003), 589.
• 2)   I.  Park and  A. W.  Thompson: Metall. Trans. A, 22A (1991), 1615.
• 3)   Y.  Iwai and  J.  Arima: J. Soc. Mater. Sci. Jpn., 30 (1981), 1074.
• 4)   Y.  Iwai and  J.  Arima: J. Soc. Mater. Sci. Jpn., 32 (1983), 1114.
• 5)   H.  Shimizu,  K.  Funami and  G.  Etoh: J. Soc. Mater. Sci. Jpn., 38 (1989), 117.
• 6)    Y.  Watanabe,  N.  Hasegawa and  M.  Inoue: J. Soc. Mater. Sci. Jpn., 41 (1992), 933.
• 7)   A.  Hirose,  Y.  Arita,  Y.  Nakanishi and  K. F.  Kobayashi: Mater. Sci. Eng., A219 (1996), 71.
• 8)   S.  Li,  Z.  Zhang,  E.  Akiyama,  K.  Tsuzaki and  B.  Zhang: Corros. Sci., 52 (2010), 1660.
• 9)   Y.  Ohno,  H.  Fujii and  S.  Sekino: Tetsu-to-Hagané, 60 (1974), 1373.
• 10)   S.  Yamasaki and  T.  Takahashi: Tetsu-to-Hagané, 83 (1997), 454.
• 11)   Y.  Hagihara,  T.  Shobu,  N.  Hisamori,  H.  Suzuki,  K.  Takai and  K.  Hirai: ISIJ Int., 52(2012), 298.
• 12)   T.  Tsuchida,  T.  Hara and  K.  Tsuzaki: Tetsu-to-Hagané, 88 (2002), 771.
• 13)   M.  Nagumo,  T.  Tamaoki and  T.  Sugawara: Hydrogen Effects on Material Behavior and Corrosion Deformation Interaction, ed. by  N. R.  Moody, TMS, Warrendale, PA, (2003), 999.
• 14)   T.  Inoue and  Y.  Namba: Tetsu-to-Hagané, 77 (1991), 2155.
• 15)   S.  Komazaki,  S.  Watanabe and  T.  Misawa: ISIJ Int., 43 (2003), 1851.
• 16)   C. J.  Altstetter,  M.  Choen and  B. L.  Averbach: Trans. ASM, 55 (1962), 287.
• 17)   W. S.  Owen: Trans. ASM, 46 (1954), 812.
• 18)   R. M.  Hobbs,  G .W.  Lorimer and  N.  Rideley: J. Iron Steel. Inst., 210 (1972), 757.
• 19)   W. J.  Nam and  H. C.  Choi: Mater. Sci. Technol., 15 (1999), 527.
• 20)   T.  Nishimura: J. Jpn. Heat Treat., 20 (1980), 342.
• 21)   K.  Takai,  J.  Seki,  E.  Sakita and  K.  Takayama: Tetsu-to-Hagané, 79 (1993), 685.
• 22)   K.  Kawasaki,  S.  Mizoguchi,  Y.  Murayama and  E.  Yamashita: Bull. Iron Steel Inst. Jpn., 5 (2000), 38.
• 23)   H.  Shoda,  H.  Suzuki,  K.  Takai and  Y.  Hagihara: ISIJ Int., 50 (2010), 115.
• 24)   K.  Takai,  H.  Shoda,  H.  Suzuki and  M.  Nagumo: Acta Mater., 56 (2008), 5158.
• 25)   T.  Doshida,  H.  Suzuki,  K.  Takai,  N.  Oshima and  T.  Hirade: ISIJ Int., 52 (2012), 198.
• 26)   M.  Nagumo and  H.  Matsuda: Philos. Mag. A, 82 (2002), 3415.
• 27)   J.  Kameda and  C. J.  McMahon, Jr.: Metall. Trans. A, 11A (1980), 91.
• 28)   C. J.  McMahon, Jr.: Eng. Fract. Mech., 68 (2001), 773.
• 29)   M.  Hayakawa,  S.  Matsuoka,  K.  Tsuzaki,  H.  Hanada and  M.  Sugisaki: Scr. Mater., 47 (2002), 655.
• 30)   M.  Hayakawa,  S.  Terasaki,  T.  Hara,  K.  Tsuzaki and  S.  Matsuoka: J. Jpn. Inst. Met., 66 (2002), 745.
• 31)   M.  Nagumo,  T.  Yagi and  H.  Saitoh: Acta Mater., 48 (2000), 943.
• 32)   M.  Hirao,  T.  Inaba and  K.  Kawasaki: CAMP-ISIJ, 19 (2006), 1063.
• 33)   K.  Takai,  G.  Yamauchi,  M.  Nakamura and  M.  Nagumo: J. Jpn. Inst. Met., 62 (1998), 267.