Effect of Microstructure at Coating Layer on Fatigue Strength in Hot-Dip Galvanized Steel

Abstract

To understand the fatigue mechanism of hot-dip galvanized steel, the fatigue strength and fracture surface of hot-dip galvanized AISI 1045 steel(carbon steel) specimens were investigated. The galvanized coating layer was composed of δ₁-phase, ζ-phase and η-phase, and its thickness was about 100 µm. In the low cycle region (10⁴ cycles < N_f < 10⁵ cycles), the fatigue strengths of both the carbon steel and the galvanized steel corresponded with the static strength. The fatigue strength of the galvanized steel was lower than that of carbon steel. As the number of cycles increased, the difference between fatigue strength of the carbon steel and that of the galvanized steel increased. Also, the morphologies of the fatigue fracture were different in low cycle region and high cycle region. In the galvanized steel, the morphology of Stage II crack on the fracture surface at low cycle region exhibited crescent shape, and multiple crack initiation sites in low cycle region were observed. Whereas the morphology at high cycle region (N_f >10⁵ cycles) exhibited an ellipse shape, and the crack initiation site was single. At both regions, the crack initiation sites were in the coating layer. The mechanical properties of the microstructure in the coating layer had an effect on the fatigue strength. When η-phase was removed from the galvanized coating layer, the fatigue strength increased only in the high cycle region. Therefore, δ₁-phase and/or ζ-phase cause the fatigue strength to decrease in low cycle region, and η-phase causes it in high cycle region.

1. Introduction

Hot-dip galvanized steel is suitable for mass production because it can form a thick plating film on small to large carbon steel in a short time. Therefore, it is a base material for social infrastructure. Recently, due to the end of the useful life of social infrastructure, there are concerns about accidents caused by fatigue and corrosion. Hence, the strength evaluation of galvanized steel is required. Hot-dip galvanized steel is used in the fatigue in corrosive environment. However, to have a clear understanding of corrosion fatigue, it is necessary to understand, independently, the behavior of corrosion and fatigue, respectively. To date, a substantial amount of research has been carried out on corrosion of galvanized steel,^1,2,3) forming of alloy layer^4,5,6) and peeling of plating film,^7,8) to ensure high reliability. On the other hand, studies on the fatigue strength of hot-dip galvanized steels focus mainly on specific areas such as the effects of welds,⁹⁾ corroded parts,¹⁰⁾ substrate parts¹¹⁾ and the effects of plating film thickness.¹²⁾ As a result, little research has been done on the effect of hot-dip galvanized structure on fatigue strength. The microstructure of the hot-dip galvanized layer consists of a lamination structure in which there are a plurality of alloy layers and a pure zinc layer. Since the mechanical properties and crystal structures of their phases are different with each other,^13,14,15) it is considered that the fatigue strength is affected. Therefore, this study aimed to clarify the effect of the structure of galvanized film on fatigue strength. Fatigue tests were carried out on cold-worked steel, galvanized steel, steel with the same heat treatment as the heat history during plating, and galvanized steel in which pure zinc phase has been removed from hot-dip galvanized layer. By arranging the correspondence between the number of cycles to failure and the fracture surface morphology, the factors that caused differences in the fatigue strength of these steels were investigated.

2. Experimental Method

2.1. Test Specimen

The material used in this experiment was commercial cold-worked steel, equivalent to AISI 1045 steel. The specimen was cold worked as shown in Fig. 1. Four types of specimens were prepared as CS, HT3, GS and AT materials. CS (carbon steel) was as-recieved AISI 1045 steel, HT3 indicates that CS material was heat-treated for 3 minutes, GS means CS material was hot-dip galvanized, and AT indicates that η-phase was removed from the GS material by acid treatment. The chemical composition of the AISI 1045 steel material is shown in Table 1. The surface roughness, Ra of the cold-worked CS material with the shape as shown in Fig. 1 was 1.6 μm in arithmetic average roughness. After degreasing, acid cleaning and flux treatment, the CS materials were immersed in a hot-dip galvanizing bath at 450°C for 3 minutes. The composition of the hot-dip galvanizing bath was presented in Table 2. After dipping, the specimens were done air-cooled for 30 s and then water-cooled for 60 s, without post-treatment performed thereafter. HT3 specimen was heat-treated CS material in a salt bath at 450°C for 3 minutes and cooled in similar conditions as those of the hot-dip galvanized specimen. The η-phase was removed by immersing GS in a 3.5% hydrochloric acid solution with a corrosion inhibitor.

Fig. 1.

Configuration of specimens.

Table 1. Chemical composition of steel substrate.

(mass%)
S45C	C	Si	Mn	P	S	Cu	Ni	Cr	Fe
	0.44	0.20	0.66	0.021	0.02	0.01	0.01	0.12	bal.

Table 2. Chemical composition of hot-dip coating bath.

(mass%)
Al	Fe	Cd	Pb	Zn
0.0028	0.029	0.093	0.83	bal.

2.2. Tensile Test and Fatigue Test

Specimens with a gauge length of 35 mm as shown in Fig. 1, were used for tensile and fatigue tests, carried out at room temperature. The average values of the diameters and cross-sectional areas of the different specimens are CS material: 5.17 mm (cross-sectional area 20.99 mm²), GS material: 5.38 mm (cross-sectional area 22.73 mm²), AT material: 5.37 mm (cross-sectional area 22.64 mm²), HT3 material: 5.17 mm (cross-sectional area 20.99 mm²). The tensile tests were carried out at the initial strain rate of 4.76 × 10⁻⁴ s⁻¹, while the Fatigue tests were conducted at a load ratio of 0.01 at 10 Hz in an atmospheric environment. When the number of cycles to failure exceeded 1 × 10⁷ cycles, the test was terminated voluntarily and the fatigue limit was defined as the maximum stress at that time.

2.3. Microstructure and Fractography

An optical microscope was used to determine the grain size of the ferrite phase and the pearlite block. After polishing its surface to a mirror finish, the specimen was etched by Na₂S₂O₅ solution of 7%. Scanning electron microscopy was used to measure the lamellar interval of pearlite and to observe the fatigue fracture surface.

2.4. Micro-Vickers Hardness

For the Vickers hardness test, the center part of the specimen as shown in Fig. 1, was cut, embedded in a resin and polished to a mirror finish. The test was done on the galvanized layer and on the substrate. The measurement point of the plating layer was the central part of the thickness of each of the phases, while the substrate was measured at 100 μm pitch from the galvanized layer/substrate interface to the deep part. The 10 points of Vickers hardness of the specimens were measured with a load of 29.42 mN (3 gf) for 5 s.

3. Results

3.1. Microstructure and Static Strength

The optical micrographs of the cross-section and longitude sections of CS, HT3 and GS materials are presented in Fig. 2. The substrates of all materials were composed of perlite and ferrite phases, with no martensite phase observed. In the cross-section, perlite block grain size and ferrite grain size of each material were as follows; CS material: 7.4 μm and 7.3 μm, HT3 material: 6.8 μm and 7.4 μm, GS material: 7.3 μm and 6.6 μm (Figs. 2(a)–2(c)). The lamellar intervals were 0.31 μm (CS), 0.28 μm (HT3) and 0.32 μm (GS). The perlite block grain size, ferrite grain size and lamellar intervals of these materials had no difference, and there was no spheroidizing of the martensite phase besides. In the cross-section, the microstructure of the heat-treated specimen did not have much difference. On the other hand, in the longitudinal direction, CS material had been extended along the processing direction. However, ferrite grains of HT3 and GS materials became equiaxed grains (Figs. 2(e)–2(f)). This suggests that HT3 and GS material causes recrystallization. Additionally, after subjecting carbon steel to the same heat treatment as HT3 and GS materials by muffle furnace, the structure and hardness were evaluated and the results showed that the recrystallization completion time was 3 to 10 minutes. In the heat treatment by salt bath, the recrystallization was completed for 3 minutes, which was earlier than the recrystallization completion time of the heat treatment by muffle furnace. This difference is due to the difference in heat transfer of the heat medium. Figure 3(a) shows the optical micrograph of the galvanized layer in GS material. The galvanized layer is generally composed of three phases: The δ₁ phase has a high Fe concentration,^16,17) and adjacent to the substrate. The ξ phase is a columnar structure,¹⁸⁾ and there is the η phase in the outermost surface. The galvanized layer of GS made in this study also had three phases. The thickness of η, ζ, and δ₁ phases were 16 μm, 55 μm, and 19 μm, respectively. Figure 3(b) shows that the η phase was completely removed by acid treatment. The thickness of the galvanized layer in AT was approximately 77 μm. Figure 4 shows the stress-strain curves for CS, GS, AT and HT3 materials. The yield stress and ultimate tensile strength of HT3, which was heat-treated by salt bath were larger than those of CS material. Moreover, the elongation of HT3 was greater than that of CS. It is possible for strain aging to occur in this temperature range,¹⁹⁾ therefore the increase in yield stress and ultimate tensile strength had its origin from strain aging. In general, when strain aging occurs, elongation decreases.^20,21) However, the elongation of HT3 increased (Fig. 4). The ferrite phase of CS has expanded in parallel to the drawing direction. On the other hand, the ferrite phase became equiaxed in HT3. This suggests that recrystallization occurred due to heat treatment, resulting in increased elongation. A similar change in structure was observed in GS as well. The focus was on the tensile properties of the HT3, which had been heat-treated in similar conditions to the GS. Although the elongation of GS was equivalent to that of HT3, the tensile strength and yield stress of GS were lower than those of HT3 (Fig. 4 and Table 3). The Vickers hardnesses of η phase, ζ phase, and δ₁ phase in the galvanized layer were 83 HV, 277 HV, and 509 HV, respectively. The Vickers hardness of the ζ phase in our study accorded with what Han et al.²²⁾ did, however, they differed with those of Han et al. with regards to the δ₁ phase. In our study, the Zn content of the δ₁ phase was measured to be 87.9%. This value was different from those of Han et al. It is, therefore, believed that the difference in the content of Zn caused this discrepancy. When the Vickers hardness of GS was calculated considering the ratio of the area of each phase to the area of the substrate, its value was 303 HV, which was smaller than that of CS (333 HV). Furthermore, it correlated with the results of the tensile test. From the above results, the strength of GS was reasonable. By removing the soft η phase, the strength of AT increased slightly, and the uniform elongation decreased. When the η phase was removed from the galvanized layer, the outmost surface of the specimen became rough. It is considered that this roughness became a notch effect and the elongation decreased. From the above, we assessed there was a thermal effect on the substrate during the plating process. As a result of judging that the HT3 could be regarded as a comparison specimen, the effect of coating structure on fatigue strength was discussed including HT3.

Fig. 2.

Optical micrographs of substrates for (a)(d) CS, (b)(e) HT3 and (c)(f) GS materials.

Fig. 3.

Optical micrographs of galvanized layers in (a) GS and (b) AT materials.

Fig. 4.

Stress-strain curves for HT3, CS, GS, and AT materials.

Table 3. Mechanical properties of materials.

Material	σYS/MPa	σUTS/MPa	El./%
CS	742	883	7.3
HT3	824	944	9.4
GS	741	855	10.4
AT	780	859	8.7

3.2. Fatigue Strength and the Shape of Fracture Surface

The shape of the fracture surface varied depending on the number of cycles to failure. The low cycle region and the high cycle region were defined by the difference in the fracture surface. The number of cycles to failure is classified that not less than 10⁴ cycles to below 10⁵ are low cycle regions and not less than 10⁵ cycles are high cycle regions. Figure 5 shows the S–N curves for each material. In fractography, if the fracture surface had a radial pattern, the crack initiation site was judged as the radial pattern start point or the position opposite to the rapid fracture region, or the center of the circular arc, which formed on the outermost layer of the fracture surface in the Stage II region, was determined to be the crack initiation site. In low cycles, the fatigue strengths of GS and AT which were lower than that of CS. The fracture surfaces of CS were the same regardless of the cycles to failure. On the other hand, the fracture surface of GS and AT had different shapes in low cycles and high cycles. Figure 6 shows the fracture surface of CS in low cycles. There was a crack initiation site in the outmost surface (Fig. 6(b)), the shape of fracture surface at Stage II was an elliptical. The fracture surfaces of HT3 were also the same as those of CS. Figure 7 shows the fracture surface of GS in low cycles. There were multiple crack initiation sites (Fig. 7(a)), all of which were in the coating layer (Fig. 7(b)). The shapes of fracture surfaces at Stage II were like crescents, where the outermost surface was a circular arc. In terms of GS, when the maximum stress was no more than 561 MPa, the shapes of fracture surfaces had a completely different form. Namely, the crack initiation site was single in the η phase, and its shape of fracture surface at Stage II became elliptical (Fig. 8). Such transitions of fracture surface were also observed in AT material more than 10⁵ cycles. The fatigue limit for each of the specimens was 563 MPa (CS), 586 MPa (HT3), 410 MPa (GS), and 521 MPa (AT) (Fig. 5). The fatigue limit of GS was lower than that of CS, that is, the fatigue limit decrease if AISI 1045 steel was coated. The fatigue limit of AT was higher than that of GS; therefore, as for removing the η phase from GS, the fatigue limit increased. The difference in fatigue limit between HT3 and CS became smaller as the cycles to failure drew closer to 10⁷ cycles. Figure 9 shows the fracture surface of GS in high cycles. There was a crack initiation site in the galvanized layer (Fig. 9(b)), the fracture surface of AT was the same as well. In the high cycle, there was no significant difference in the fracture surface morphology in any of the specimens. However, the strength of Galvanized steel had a different depending on the presence or absence of the galvanized layer and the presence or absence of the η phase. In both low cycle and high cycle, the fatigue strengths of GS and AT were less than that of CS. Because the fatigue strength of GS material was lower than that of HT3 that was heat-treated like GS, it can be said that the decrease in fatigue strength of GS is due to the occurrence of the plating layer. The effect of the plating layer on fatigue strength will be examined in the next chapter.

Fig. 5.

S–N curves of materials and types of crack initiation.

Fig. 6.

Secondary electron (SE) images of fracture surface in low cycle fatigue of CS material (a) (σ_max = 703 MPa, N_f = 30620 cycles) and its enlarged one near the crack initiation site (b).

Fig. 7.

SE images of fracture surface in low cycle fatigue of GS material (a) (σ_max = 690 MPa, N_f = 17880 cycles) and its enlarged one near the crack initiation site (b).

Fig. 8.

SE images of fracture surface in low cycle fatigue of GS material (a) (σ_max = 561 MPa, N_f = 51570 cycles) and its enlarged one near the crack initiation site (b).

Fig. 9.

SE images of fracture surface in high cycle fatigue of GS material (a) (σ_max = 518 MPa, N_f = 100030 cycles) and its enlarged one near the crack initiation site (b).

4. Discussion

4.1. The Effect of Static Strength on Fatigue Strength

The fatigue strength of GS and AT with a galvanized layer were lower than that of HT3, which had heat-treated as shown in Fig. 5. One possible reason for the decrease in fatigue strength is a lowering in static strength due to the presence of a plating film. There appears to be a correlation between the ultimate tensile strength and fatigue strength, which is particularly evident among steel materials.^23,24) Thus, to understand the reduction of fatigue strength caused by static strength, we evaluated that the effect of static strength by normalizing by fatigue strength against the ultimate tensile strength (σ_max/σ_UTS) (Fig. 10). The σ_max/σ_UTS of all specimens in N_f = 10⁴ cycles have shown to be equal. When N_f reached 2 × 10⁴ cycles, the difference began to appear. Moreover, the difference between σ_max/σ_UTS became larger as the cycle became higher. That suggests that the static strength is dominant against the fatigue strength in the low cycle region and the fatigue strength is decided by microstructure in the high cycle region. Especially, the decrease of fatigue strength for GS can be observed in comparison with other specimens, the plating microstructure is caused by the decrease of the fatigue strength. The σ_max/σ_UTS of AT was the same as GS in low cycle region, while the σ_max/σ_UTS of AT became remarkably higher than that of GS in high cycle region and this value of AT slightly lower than that of CS and HT3 with no plating films. From these facts, we understood that the factor causing the decrease of fatigue strength is the galvanized layer in both low cycle and high cycle and the fatigue strength in high cycle especially decreases due to the η phase.

Fig. 10.

Relationship between σ_max/σ_UTS and number of cycles to failure.

4.2. The Effect of a Stable Crack Growth Speed on the Reduction of Fatigue Strength

The decrease in fatigue strength and fatigue limit of galvanized steel was due to the presence of the plating layer, not the effect of static strength. The decrease in fatigue strength caused by the hot-dip galvanized process is discussed in this section. The basic process of fatigue failure after dislocation substructure formation by the repeated deformation is classified into four processes: damage accumulation due to deformation concentration, generation and growth of subcracks (Stage I), stable crack growth (Stage II) and rapid fracture (Stage III).²⁵⁾ Stage I and Stage II are dominant in the fatigue life in those process. First of all, considering Stage II, to evaluate the crack growth rate da/dN, we observed the striation spacing S that were occurred the same distance from the crack initiation site by using fractured specimens with the same level of stress amplitude (Fig. 11). Striation spacing corresponds to the crack development amount, it is represented by formula (1).   

$$S\cong da/dN$$(1)

Where S is striation spacing, a is the crack length and N is the number of cycles to failure. The striation spacing of GS, AT, and CS were S_GS = 2.00 × 10⁻⁷ m/cycle, S_AT = 2.06 × 10⁻⁷ m/cycle, S_CS = 1.89 × 10⁻⁷ m/cycle, respectively. In this way, there was no significant difference in striation spacing. Next, we discuss the fatigue fracture toughness value on the transition from a stable crack growth (Stage II) to a rapid fracture (Stage III). Fatigue fracture toughness value K which upon forming the shape of elliptical on Stage II process can be approximated by the following equation.²⁶⁾   

$$K_{IC} \cong 0.65\sigma \sqrt{\pi \sqrt{area}}$$(2)

Where σ is load stress, $\sqrt{area}$ is a fracture area at Stage II. These K values when the crack shape in Stage II forms an ellipse were as follows; $K_{IC}^{GS}=67.5\mathrm{\hspace{0.17em} }\text{MPa}\sqrt{m}$ (σ = 606 MPa), $K_{IC}^{AT}=67.7\mathrm{\hspace{0.17em} }\text{MPa}\sqrt{m}$ (σ = 608 MPa), $K_{IC}^{CS}=63.3\mathrm{\hspace{0.17em} }\text{MPa}\sqrt{m}$ (σ = 593 MPa). The value of K_IC was not affected by the plating process. From the above results, it was shown that as for the fatigue fracture toughness value K_IC in the tip of crack and crack growth rate $da/dN$, there were no significant differences between the three materials. It can be said that the decrease of fatigue strength for galvanized steel is not due to crack propagation speed in stable crack propagation region (Stage II) but due to generation and growth of subcrack in plating film. The influence of subcrack generation and growth greatly has affected the microstructure. In low cycle, the fatigue strength of GS and AT became a decrease than those of CS and HT3; nevertheless the fatigue strength of GS and that of AT material were comparable. It was found from the result that the in low cycle region the η phase did not concern the decrease of fatigue strength for galvanized steel. It was the crack generation in the alloy phase that was responsible for the decrease of fatigue strength in the low cycle region. On the other hand, the crack initiation site of GS that reduced in high cycle region the fatigue strength was the η phase at the outermost surface. Moreover, when the η phase was removed from the GS, its fatigue strength increased. That is to say, that structure that caused a decrease in fatigue strength in the high cycle region is the η phase. As mentioned above, depending on the number of cycles to failure, the structure in the plating film, which causes the fatigue strength of the plated structure to decrease, was different.

Fig. 11.

SE images of striations for (a) GS, (b) AT and (c) CS materials.

4.3. Initial Crack Forming Process and Fatigue Strength Decrease

The fracture surface in Stage II process which broke in the low cycle region formed like a crescent shape. When the side of the fracture surface was investigated, those end parts of the materials, which were not done the plating treatment, were formed the linear shape in the direction perpendicular to the tensile axis (Fig. 12(a)), whereas those of the galvanized steel formed uneven shapes (Fig. 12(b)). Also, some ratchet marks, which indicate the existence of steps, were radially from the crack initiation sites toward the center of the specimen (Fig. 13). Those results of observations suggest that multiple cracks generate and coalesce at the same time. It was observed that the shape of fracture surface in Stage II process was constituted like a crescent-shaped after those cracks combined. In other words, multiple cracks coalesced to form a crescent-shape before the shape of fracture surface in Stage II process formed an elliptical-shape. As a result, it leads to fracture toughness value K_IC. Supposing that the above hypothesis is true, the area A_CS in Stage II for CS and this one A_GS in Stage II for GS should be mostly the same size. For those reasons, taking pieces of areas in Stage II into account, A_GS was calculated by approximating with the formula (3).   

$$A_{GS} \cong {\sum }_i{A}_i$$(3)

Where i = 1,2,3, … represent the number of areas in Stage II, the symbol A_i is correspond to the ith area in Stage II. In the case of the fracture surface, Stage II was like a crescent at the low cycle region. Figure 14 shows the ratio of areas in Stage II compared to the whole area. There was a minor difference in the area in Stage II with all of the specimens, which bears out our assumption. Figure 15 shows the model of the crack propagation process in the low cycle region based on the result of fractography. In galvanized steel, multiple and subcracks occur in the direction perpendicular to the tensile axis when loading repeated stress continues (Fig. 15(a)). As the number of repetitions increased, multiple subcracks coalesced and the growth of cracks increased (Fig. 15(b)). This is how those subcracks were the main cracks. The growth and coalescence of multiple subcrack where not on the same plane make the specimen to fracture (Fig. 15(c)). Therefore, the side of fracture is constituted on uneven shapes like a Fig. 12(b), and it led to a fracture after these ratchet marks are formed (Fig. 15(d)). In the high cycle region, it is the η phase to make the fatigue strength decrease. Comparing with these the Vickers hardness of respective phases, η phase was softer than any other phase in plating film. For those reasons, it can be said that it is easy to occur the slip deformation in the η phase. During the fatigue deformation, the slip deformation at the surface cause to form uneven shapes by irreversible operation of dislocation. As it repeatedly deforms, it becomes a persistent slip band that has concave-convex shaped with developed deep into the material. After that, the persistent slip band becomes the initiation of fatigue fracture.²⁷⁾ As a result, it is considered the early formation of the subcrack in the η phase to cause the decrease of the fatigue strength. In this study, we clarified that in the high cycle region, it was the plating film that acts as a nucleating site for fatigue crack in galvanized steel. However, in the low cycle region, it is unclear which phase in the alloy layer reduced the fatigue strength. Therefore, there is a need for further analysis in the future.

Fig. 12.

SE images of specimen surface near the crack initiation site of fracture surface for (a) CS and (b) GS materials.

Fig. 13.

Arrows show the ratchet marks which appeared in the galvanized layer and substrate for GS at low cycle fatigue. (σ_max = 647 MPa, N_f = 34950 cycles)

Fig. 14.

Relationship between area ratio of Stage II to fracture surface and maximum cyclic stress.

Fig. 15.

Schematic illustrations of fracture process for galvanized steel in low cycles; (a) subcrack formation, (b) subcrack growth, (c) main crack formation by the coalescence of subcrack, and (d) main crack growth and fracture.

5. Conclusions

The fatigue strength of as-cold-worked steel and galvanized steel were evaluated, after that doing the fractography of crack initiation site and the area in Stage II. Besides, we created thermal processing material with similar to hot-dip galvanizing and a model material from which the η phase has been removed as we carried out those evaluations and analysis. From the results, it was verified galvanizing decreases fatigue strength. The main results obtained were as follows.

(1) As galvanizing the carbon steel decreased slightly its fatigue strength in low cycle region and reduced enormously in high cycle region.

(2) There is no difference in crack propagation speed and fracture toughness value between as-cold-worked steel and galvanized steel. The decrease of Fatigue strength is not because of a stable crack propagation region (Stage II) but because of the formation of initial cracking.

(3) The removal of the η phase from the plating layer did not cause any variation in the fatigue strength of galvanized steel in the low cycle region. Therefore, it is the alloy layer that caused a decrease in fatigue strength in the low cycle region. The cracks, which are generated from multiple crack initiation sites, are coalesced and become one main crack which resulted in a crescent-shaped fracture.

(4) In the high cycle region, the fatigue strength increased when the η phase was removed. The difference between the fatigue strength for carbon steel and galvanized steel is nearly eliminated. For these reasons, it is believed that the η phase is the contributing factor to the decrease in the fatigue strength decrease in the high cycle region. This is considered because the η phase is soft and the persistent slip band is easily formed.

Acknowledgments

This work was supported by JSPS KAKENHI Grant Number 15K18232. The authors would like to thank Dr. Yoshinori Ono and Mr. Masayuki Komatsu from NIMS (National Institute for Materials Science) for technical assistance to fractography and for lending their expertise.

References

• 1)   Y.  Takano,  T.  Izeki,  T.  Nakada and  K.  Takada: J. Soc. Mater. Sci., Jpn., 42 (1993), 934 (in Japanese). https://doi.org/10.2472/jsms.42.934
• 2)   S.  Yanagi and  H.  Matsuoka: Zairyo-to-Kankyo (Corrosion eng.), 45 (1996), 728 (in Japanese). https://doi.org/10.3323/jcorr1991.45.728
• 3)   T.  Miwa,  T.  Kudo,  M.  Watanabe,  M.  Matsumoto and  T.  Handa: Bosei Kanri (Rust Prev. Control), 56 (2012), 173 (in Japanese).
• 4)   J.  Inagaki,  M.  Sakurai and  T.  Watanabe: ISIJ Int., 35 (1995), 1388. https://doi.org/10.2355/isijinternational.35.1388
• 5)   M.  Saito,  Y.  Uchida,  T.  Kittaka,  Y.  Hirose and  Y.  Hisamatsu: Tetsu-to-Hagané, 77 (1991), 947 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.77.7_947
• 6)   H.  Koga,  Y.  Uchiyama and  T.  Aki: J. Jpn. Inst. Met., 42 (1978), 136 (in Japanese).
• 7)   K.  Lee,  I.  Lee,  C.  Lee and  H.  Ahn: Surf. Coat. Technol., 201 (2007), 6261.
• 8)   Y.  Nunomura and  T.  Takasugi: ISIJ Int., 43 (2003), 454. https://doi.org/10.2355/isijinternational.43.454
• 9)   T.  Mori,  K.  Minami and  H.  Horikawa: Q. J. Jpn. Weld. Soc., 22 (2004), 272 (in Japanese). https://doi.org/10.2207/qjjws.22.272
• 10)   M.  Buršák and  I.  Mamuzić: Metalurgija, 46 (2007), 277.
• 11)   A.  Dimatteo,  G.  Lovicu,  M.  DeSanctis,  R.  Valentini,  F.  D’Aiuto and  M.  Salvati: Proc. XXI IGF Conf. Italian Group of Fracture (IGF21), eds. by G. Ferro et al., Gruppo Italiano Frattura, Cassino, (2011), 283.
• 12)   J. B.  Vogt,  O.  Boussac and  J.  Foct: Fatigue Fract. Eng. Mater. Struct., 24 (2001), 33. https://doi.org/10.1046/j.1460-2695.2001.00365.x
• 13)   S.  Lazik,  C.  Esling and  J.  Wegria: Textures Microstruct., 23 (1995), 131. http://doi.org/10.1155/TSM.23.131
• 14)   Y.  Adachi,  M.  Arai and  T.  Nakamori: Tetsu-to-Hagané, 80 (1994), 225 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.80.3_225
• 15)   A. R.  Marder: Prog. Mater. Sci., 45 (2000), 191. http://doi.org/10.1016/S0079-6425(98)00006-1
• 16)   R.  Kainuma and  K.  Ishida: Tetsu-to-Hagané, 91 (2005), 349 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.91.3_349
• 17)   N.  Takata and  M.  Takeyama: Tetsu-to-Hagané, 100 (2014), 1172 (in Japanese). https://doi.org/10.2355/tetsutohagane.100.1172
• 18)   Y.  Bergengren and  A.  Melander: Int. J. Fatigue, 14 (1992), 154. https://doi.org/10.1016/0142-1123(92)90368-M
• 19)   T.  Kuboki,  M.  Murata and  D. R. J.  Owen: Comput. Struct., 88 (2010), 1162. https://doi.org/10.1016/j.compstruc.2010.07.007
• 20)   A.  Bülbül and  R.  Kaçar: Mater. Res., 20 (2017), 210. http://doi.org/10.1590/1980-5373-MR-2016-0496
• 21)   F.  Vodopivec: Metalurgija, 43 (2004), 143.
• 22)   K.  Han,  I.  Lee,  I.  Ohnuma,  K.  Okuda and  R.  Kainuma: ISIJ Int., 58 (2018), 1578. https://doi.org/10.2355/isijinternational.ISIJINT-2018-111
• 23)   S.  Nishijima: J. Soc. Mater. Sci., Jpn., 29 (1980), 24 (in Japanese). https://doi.org/10.2472/jsms.29.24
• 24)   H.  Tamaki,  R.  Kitazawa,  M.  Yoshida and  S.  Horibe: J. Jpn. Inst. Met., 77 (2013), 514 (in Japanese).
• 25)   O.  Umezawa and  K.  Nagai: ISIJ Int., 37 (1997), 1170. https://doi.org/10.2355/isijinternational.37.1170
• 26)   Y.  Murakami and  M.  Isida: Trans. Jpn. Soc. Mech. Eng. A, 51 (1985), 1050 (in Japanese). https://doi.org/10.1299/kikaia.51.1050
• 27)   F.  Ellyin: Fatigue Damage, Crack Growth and Life Prediction, Springer, Chapman and Hall, London, (1997), 10. https://doi.org/10.1007/978-94-009-1509-1