Fatigue design curves for laser-metal-deposited type 420 stainless steel and effect of an interval during deposition process

Fully reversed axial loading fatigue tests were conducted using type 420J1 martensitic stainless steel. The specimens were additively manufactured by a laser metal deposition (LMD) process. The results were compared with conventionally manufactured (CMed) type 420J2 stainless steel. According to the axial loading fatigue test results, the fatigue strengths of the laser-metal-deposited (LMDed) specimens were nearly comparable to those of the CMed specimens. Fractographic analyses revealed that process-induced defects were hardly seen at the fatigue crack initiation sites of the LMDed specimens. It indicates that defect-free deposition was possible by the LMD process. On the other hand, when the LMD specimens experienced intervals during deposition processes, local softening occurred due to the tempering of the building plate. Fatigue tests revealed that the interval during LMD process had detrimental effect on the fatigue strengths due to the local softening. The upper and lower bounds of S-N curves were proposed as fatigue design curves for the samples with and without the interval during LMD process.


Introduction
Additive manufacturing (AM) is an attractive method to fabricate near-net-shape components with complex shapes [1,2]. Powder bed fusion (PBF) and directed energy deposition (DED) are two major methods in the several AM procedures, which are used for the fabrication of load-bearing mechanical components [3]. Laser metal deposition (LMD) is one of the DED processes, in which high power laser is used to melt metal powder [4][5][6]. LMD has some merits; for example, large components can be fabricated because inert gas chamber is not necessary like PBF. Furthermore, since powder supplying route can be changed easily, multi-layered components can be made [7,8]. LMD is sometimes used for the repairmen of dies and turbine blades [9,10], because it is easy to mold powder on the bulk components. Recently, multitasking LMD machines are developed [11], in which a 5-axis machining center is integrated into an LMD processing area. Multitasking LMD machine can fabricate final products by alternating procedures of AM and machining in the same processing area. It is expected that multitasking machines will be widely used for the fabrication of mechanical components. Therefore, it is important to understand the fatigue properties of additively manufactured (AMed) components. There are several works about fatigue properties of AMed materials of high-strength stainless steels [12][13][14][15][16][17]. For example, Yadollahi et al. [17] conducted axial loading fatigue tests and fatigue crack propagation tests using AMed 17-4 precipitation hardening stainless steel. They revealed that fatigue cracks predominantly initiated at process-induced defects, such as lack-of-fusion defects, and consequently, fatigue lives were successfully predicted using linear elastic facture mechanics assuming crack growth from initial defects. However, all samples were fabricated by PBF-type selective laser melting (SLM) in the above references [12][13][14][15][16][17]. Bandyopadhyay et al. investigated fatigue properties and effect of building conditions in Ti-6Al-4V fabricated by laser engineered net shaping (LENS TM ) [18], one of DED processes, but research about fatigue of the laser-metal-deposited (LMDed) highstrength stainless steels is missing.
It is known that several building conditions, such as laser hatching [19], building direction [14,16,20], strut orientation [21], scanning velocity [22], and laser power [23], to name a few, can affect mechanical properties of built samples. The process interval during deposition could be also the influencing factor on the mechanical properties [24]. Stoll et al. revealed that the tensile properties decreased 13~18 % by the interruption during building in selective-laser-melted (SLMed) type 316 austenitic stainless steel [24]. They also reported that the amount of degradation in tensile properties was dependent on the building direction. Consequently, it can be said that the interval during deposition process could be one of the influencing factors on the mechanical properties of AMed materials. However, the effect of building interval on the fatigue properties of LMDed high-strength stainless steels is not investigated yet.
In the present study, a multitasking LMD machine, MU-6300V LASER EX (OKUMA Corporation), was used for the fabrication of the samples using powder of type 420 martensitic stainless steel as shown in Fig. 1. Fatigue tests were conducted to evaluate fatigue properties of the LMDed highstrength stainless steel. In addition, deposition process was once interrupted to cool the building plate down to the room temperature, and then the deposition was resumed to evaluate the effect of the intervals during deposition process on the fatigue properties. Finally, design fatigue curves (DFC) were proposed, in which the effect of intervals was taken into account.

Materials and specimens
The powder material is type 420J1 martensitic stainless steel, whose particle size is 44~88 μm. The microscopic appearance of the powder is shown in Fig. 2. The reference material is conventionally manufactured (CMed), namely vacuummelted type 420J2. type 420J2 was received in the annealed condition. Therefore, some samples were annealed at 1040°C for 15 min followed by oil quenching and, subsequently, tempered at 150°C for 2 h followed by air cooling. The chemical compositions of the materials are summarized in Table 1. Figure 3 shows the LMD process. Firstly, type 420J1 powder was deposited on the carbon steel base plate to form the plate with the length of 130, height of 30, and thickness of 4 mm as shown in Fig. 3a. It should be noted that argon shielding gas was used during deposition. Subsequently, the LMDed plate was machined into the plate with the length of 130, height of 26, and thickness of 2.5 mm (Fig. 3b) in the same processing area of a multitasking LMD machine, whose LMD processing nozzle Cutting tool spindle   machining conditions are summarized in Table 2. Finally, the plate was cut from the base plate as shown in Fig. 3c. The fatigue specimen, whose configuration is shown in Fig. 4, was cut from the plate of Fig. 3c by wire cut discharge machining. The building direction was perpendicular to the longitudinal direction of a fatigue specimen. The LMD conditions are summarized in Table 3. In some samples, the LMD process was interrupted at the dotted line in Fig. 4, namely at the center of the specimen, to investigate the effect of the interval during deposition process. When the temperature of the building plate became the same with the room temperature, the deposition was resumed. The period of interruption was roughly about 30 min.

Experimental procedures
The microstructures were etched by Picric acid-ethanol solution and observed by optical and scanning electron microscopes. Hardness distribution was measured by a micro Vickers hardness tester at the load of 4.9 N and dwell time of 15 s with the adjacent spacing of 1 mm. Axial loading fatigue tests were conducted using an electro-hydraulic fatigue testing machine with the capacity of 50 kN. The fatigue load waveform was sinusoidal with the frequency of 20 Hz and stress ratio R = σ min / σ max = −1 (fully reversed loading). All fatigue specimens were polished before the fatigue tests by the emery paper from the grade of #80 to #2000 and buff finished to obtain mirror surface. One sample was used for each stress level in S-N diagram, which is enough to draw design fatigue curves because the approximation curve for discrete date points gives 50% probability curve of fatigue life. Residual stress was measured by an X-ray diffraction method using Bulker MO3XHF22. The conditions of X-ray diffraction are summarized in Table 4.

Microstructures
The microstructures of the reference material, CMed type 420J2, are shown in Fig (Fig. 5a). On the other hand, martensitic microstructure appeared in Fig. 5b due to the quenching and tempering. Carbides are dispersed in both microstructures, while the number of carbides is smaller in Fig. 5b, because carbon is under supersaturated solid solution in martensite microstructure. Figure 6 indicates the microstructures of the LMDed type 420J1 observed on the cross section as schematically shown in the upper figure. The interval during deposition procedure was given near the center of the cross section (at the arrow in Fig. 6a) to lower the temperature of the building plate to the room temperature. The magnified views of the microstructures at the locations "A" and "B" in Fig. 6a are shown in Fig. 6 b and c, respectively, which are far away from the interval where the effect interval on the microstructure is negligible. They reveal that martensitic microstructures developed at both locations due to re-solidification of melted powder. Figure 7 a shows the microstructure in the rectangular area of Fig. 6a including the location of interval. It should be noted that the shade of the gray color changed slightly across the interval. The magnified views at the areas "A," "B," and "C" in Fig. 7a are shown in Fig. 7 b, c, and d, respectively. The martensitic microstructures are dominant at all locations, and the microscopic appearances of martensitic microstructures are nearly the same in Fig.  6 b and d and Fig. 7b~d. As will be mentioned in the next section, local softening occurred by the interval in spite of the similar martensitic microstructures with and without interval. It means that cementite distribution in

Static mechanical properties
The average Vickers hardness of the reference material, CMed type 420J2, was 171 HV and 532 HV for the annealed and quenched samples, respectively, and the average hardness of the LMDed type 420J1 was 573 HV. Subsequently, Vickers hardness profiles were measured on the cross section of the LMDed samples with and without the interval as shown in Fig. 8. It should be noted that local softening occurred due to the interval, and the softened area was wider in the area before the interval (left side of the interval). Typical stressstrain curves of the CMed and LMDed specimens are shown in Fig. 9, and the properties are summarized in Table 5. The mechanical properties of the CMed specimen increased by the quenching, and those of the LMDed specimen are similar to the quenched reference material, while the elongation and reduction of area, namely ductility, were lower ( Table 5). The residual stresses parallel to the longitudinal direction were measured using the plate shown in Fig. 3c before the wire cutting of a fatigue specimen. The results are shown in Fig.  10 as a function of the distance from the center of the plate. Residual stress distributions are different depending on the building conditions, where the compressive residual stress is dominant without building interval, and tensile residual stress appears by the interval. The residual stress is strongly affected by the heat history, while the precise mechanism in the change of residual stress by the interval is not clear in the present case. However, it should be noted that similar compressive residual stresses appear at the center of the plates, namely the center of the gauge section of fatigue specimens, in both specimens with and without building interval, which would be the controlling factor of the fatigue strength shown in the next section.  The line where interval was given. without interval are nearly comparable to those of the quenched reference material. On the other hand, LMDed specimens with building interval had lower fatigue strengths than those without interval. Typical fatigue fracture surfaces of the quenched reference material and LMDed specimen without interval are shown in Figs. 12 and 13, respectively. Fatigue crack initiation sites are flat in both materials. It should be emphasized that defect is not seen at the crack initiation site of the LMDed specimen (Fig. 13). It is known that process-induced defects like pores could be fatigue crack initiation sites in the AMed materials, because fatigue crack initiation is sensitive to the stress concentration at defects [12,25]. In the present study, although fatigue crack initiation sites of six fatigue failed samples in Fig. 11 were examined in detail, defects were not found. Figures 14, 15, and 16 are the fatigue fracture surfaces of the LMDed specimens with intervals. In all cases, fatigue cracks initiated at the left side of the interval. A defect was found at the crack initiation site only in Fig. 16, where the size was as small as about 40 μm. As mentioned above, fatigue cracks predominantly initiate at defects in AMed materials. For example, one of the authors, Uematsu and Kakiuchi, conducted  fatigue tests using five samples of SLMed Type 630 martensitic stainless steel [12] and eleven samples of electron-beammelted Ti-6Al-4V [25], and fatigue cracks were initiated at the defects in all sixteen samples. However, we found a defect only one case in nine samples as shown in Fig. 16. It indicates that nearly defect-free fabrication could be possible in the present LMD process compared with the other conventional AM processes.

Discussion
As shown in Fig. 11, interval during deposition process had detrimental effect on the fatigue performances. Fractographic analyses revealed that fatigue cracks predominantly initiated at the area before the interval (Figs. 14, 15, and 16). The fatigue crack initiation sites are shown by the cross symbols in the hardness profile of Fig. 8. The sites do not exactly correspond to the location of minimum hardness, while it should be noted that the softened area is wider in the area before the interval. Consequently, the degradation of fatigue strength by the interval can be attributed to the local softening induced by the heat history during the interval. The heat history of the building process with interval is as follows. When the building is interrupted, the temperature of the building plate goes down to the room temperature at a slow cooling rate, and subsequently building is resumed, namely melted powder is deposited on the cooled-down plate again. The microstructural features near the interval are almost same as shown in Fig. 7, but it is considered that tempering occurred during slow cool down process during interval, resulting in the local softening. The tempering during interval was also responsible for the wider softened area before the interval (left side of the interval in Fig. 8). The softening would be recovered by a post-heat treatment. However, we assume the application of LMDed components in the as-built condition; therefore, the effect of post-heat treatment is not discussed in the present study. That is because martensitic stainless steel, type Fig. 7 a Microstructure in the rectangular area of Fig. 6a and b, c, and d are the magnified views at "A," "B," and "C" in a, respectively. 420, has high hardness even after the softening by the interval. In addition, sometimes large components are fabricated by LMD, and the post-heat treatment for large components would highly increase the cost.
In the above discussion, the heat history is a main factor of the interval. Although oxidation of the surface during interval could be another factor, we assume that the effect of the oxidation is minimal. The reasons are as follows. First, argon shielding gas was used during LMD process. Second, we focused on the softening by the interval as shown in Fig. 8, but the oxidation would bring about the embrittlement of material. However, embrittlement was not recognized. Third, we could not find any oxides on the fatigue fracture surfaces.       As schematically shown in Fig. 17, interval during deposition process is equivalent to the size of a building object. When the size of the object becomes large, the temperature at a location of the building object could go down to the room temperature during the movement of a laser processing nozzle. Therefore, the design fatigue curves for the LMDed samples can be given as shown in Fig. 18. The upper and lower bounds are design curves for the small and large objects, respectively. If building objects are small and the effect of the interval is negligible, the upper bound can be used for the fatigue design. If building objects are large, and the decrease of the temperature is unknown, the lower bound in Fig. 18 could be used as a conservative design curve. However, if the decrease of the temperature and local softening depending on   the size of the building object are monitored, the optimized lower bound can be proposed as schematically shown by a green broken-lined curve in Fig. 18.

Conclusion
Type 420J1 martensitic stainless steel plates were fabricated using a multitasking LMD machine consisting of additive manufacturing and 5-axis machining center. Subsequently, fully reversed axial loading fatigue tests were conducted using laser-metal-deposited (LMDed) and conventionally manufactured (CMed) samples. In some LMDed samples, building intervals were given to investigate the effect of intervals during deposition process on the fatigue properties and to propose design fatigue curves. The conclusions are as follows.
(1) The hardness and tensile strength of the LMDed samples were slightly higher than those of the quenched and tempered reference material. However, the fatigue strengths of both materials were nearly comparable.
(2) By giving building interval, local softening occurred near the location of the interval. Tempering occurred in the building plate during interval due to the slow cooling rate, resulting in the local softening. The softened area was wider in the area before the interval because of the tempering of the building plate during interval.
(3) Fatigue strengths of the specimens with intervals were inferior to those without interval due to the local softening. Defects were rarely seen at the crack initiation sites of the LMDed specimens irrespective of the intervals.
(4) S-N curves for the samples with and without intervals were proposed as design fatigue curves for large and small building objects, respectively. That is because building interval in an LMD process could be related to the size of the building objects.
Author contribution Y. Uematsu, T. Kakiuchi, and R. Sano planed and conducted fatigue test analyses. R. Sasaki, S. Yamamoto, and A. Zensho optimized building conditions and built samples. All authors discussed the results and contributed to the final manuscript.

Declarations
Ethics approval Not applicable.
Consent to participate All authors have approved to participate.

Consent for publication
The manuscript has been approved by all authors for publication.

Competing interests
The authors declare no competing interests.