Dwell-fatigue crack propagation in additive manufactured Hastelloy X

Author’s Accepted Manuscript Dwell-fatigue crack propagation in additive manufactured Hastelloy X Jonas Saarimäki, Mattias Lundberg, Håkan Brodin, Johan J. Moverare www.elsevier.com/locate/msea

PII: DOI: Reference:

S0921-5093(18)30309-5 https://doi.org/10.1016/j.msea.2018.02.091 MSA36178

To appear in: Materials Science & Engineering A Received date: 15 January 2018 Accepted date: 26 February 2018 Cite this article as: Jonas Saarimäki, Mattias Lundberg, Håkan Brodin and Johan J. Moverare, Dwell-fatigue crack propagation in additive manufactured Hastelloy X , Materials Science & Engineering A, https://doi.org/10.1016/j.msea.2018.02.091 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Dwell-fatigue crack propagation in additive manufactured Hastelloy X Jonas Saarimäki

a

∗ a a,b a , Mattias Lundberg , Håkan Brodin , Johan J. Moverare

a,

Division of Engineering Materials, Department of Management and Engineering, Linköping University, SE-58183 Linköping, Sweden b Siemens Industrial Turbomachinery AB, Finspång, SE-612 31, Sweden

Abstract

Additively manufactured Hastelloy X by laser-powderbed fusion is a superalloy used in for example burners and non-rotating parts in gas turbines. Turbines are often subjected to dwell-fatigue as a result of an operating prole including load cycles with long constant power output. The eect of building direction and heat treatments on dwell-fatigue crack propagation in additively manufactured Hastelloy X has not yet been thoroughly investigated. Crack propagation behaviour was characterized using compact tension samples cut from as-built and heat treated material blocks. Samples were machine with the notch parallel and perpendicular to the building direction enabling the investigation of building direction on crack behaviour and crack prop-

◦ agation rates. The samples were subjected to dwell-fatigue tests at 700 C with 90 s or 2160 s dwell-times at maximum load. Microstructural characterization was conducted using light optical microscopy and scanning electron microscopy techniques such as electron channelling contrast imaging and electron backscatter diraction. The additively manufactured alloy exhibits anisotropic behaviour caused by the directionally solidied microstructure. Cracks propagated intergranularly and preferably

∗

[email protected] +46 73 778 41 84

Preprint submitted to Material Science & Engineering A

March 1, 2018

through streaks of topologically close-packed phases. Keywords:

AM, Nickel-based superalloys, Fatigue, Mechanical characterization,

Electron microscopy

1. Introduction

Additive manufacturing (AM) can be used to produce advanced geometries that are impossible to manufacture by conventional manufacturing processes such as machining and casting.

AM was primarily developed for simpler materials such as

thermoset plastics and plaster. The AM equipment originally used could only melt materials with low melting points for example brass and was not powerful enough to completely melt steel.

This did not meet the requirements for parts made of

superalloys [1]. With time, process control was improved and more powerful lasers were developed. The new lasers made it possible to create a microstructure with a low amount of porosity and less internal defects such as solidication cracks or poor bonding [2]. There is a wide variety of alloys designed for selective laser melting including aluminum, titanium, tool steel, stainless steel, and heat resistant materials that are cobalt- and nickel-based [35]. When it comes to the process of melting powder, laser melting is the dominating manufacturing process, often denoted selective laser melting (SLM), direct metal deposition (DMD) or direct metal laser fabrication (DMLF), depending on the company's manufacturing the equipment for laser melting. The building process can be explained by the three following steps. Firstly the lasers work envelope can be preheated <400

◦

is distributed evenly over the work surface.

When the rst layer has been spread

C and the rst thin layer of powder

the laser will melt the contour and hatch of the rst layer from the sliced model. The third manufacturing step is lowering the lasers work surface, and the process is

2

repeated by coating a new thin layer of powder that will be melted to the previous layer. When the part build is complete the part is removed from the work surface, usually by electrical-discharge machining (EDM). If necessary, the built part is post processed using for example machining, blasting and heat treatments such as stress relieve. AM is appealing to industries targeting low volume production of highly customized parts for specic applications such as medical implants [6].

Surgical in-

struments and patient/injury-specic implants can be generated via AM reducing patient wait times and accelerating their post-implant healing. More individualized medical equipment can be generated and delivered quicker [7, 8]. AM also provides the ability of remote manufacturing and repair in space and at sea on demand as well as manufacturing of functionally-graded parts. The full potential of AM to provide new means for manufacturing load bearing structural parts is however not yet fully realized [9]. The main industrial challenge for implementing AM is the uncertainty in mechanical properties of AM parts [911]. This uncertainty is due to heterogeneities and randomly dispersed defects [9] and unknown mechanical properties as well as variations in as-received powder characteristics, recycling of powder, building procedure and AM systems. This becomes even more dicult taking in to account the many involved process parameters such as laser power, laser speed, layer thickness which all will aect the thermal history during fabrication [10, 11]. Thermal history from melt pool temperature, thermal gradient, cooling rate and cyclic reheating during the AM process aects the microstructural properties such as grain size, morphology, texture, defect type, defect size, defect distribution, residual stresses, etc., and consequently, the mechanical properties [9]. Superalloys show a positive combination of mechanical properties and corrosion

3

resistance, which is why they are used in some of the worlds most aggressive working conditions namely the hot-section of gas turbines and aircraft engines. These harsh conditions are harmful to the alloy in the form of fatigue, creep and oxidation. All of which deteriorates components. AM EOS NickelAlloy HX superalloy is a solid solution strengthened alloy also containing strengthening carbides. EOS NickelAlloy HX is recurrently used for high temperature components such as gas turbine burners. Gas turbine burners are surrounded by temperatures above the melting temperature of the material but are protected by thermal barrier coatings eectively reducing the working temperature of the metal down to

∼ 700 ◦ C. At these high temperatures the mechanical properties

start to degrade considerably [12]. Components subjected to combinations of high temperatures and sustained periods of high tensile loads can experience accelerated crack propagation rates due to dwell-times. Not only dwell-times eect crack propagation but so does environmental aspects. The inuence of an oxidizing environment has long been recognized to be the main reason for accelerated crack propagation rates [1315]. Accelerated crack propagation rates are often accompanied by a change in crack propagation mechanism from transgranular to intergranular. This is particularly true for ne grained and high strength alloys [1316]. For lower strength alloys, the eects of creep deformation should be considered carefully. Stress relaxation ahead of the crack tip could lower the mechanical driving forces for crack propagation. If the degree of crack tip creep deformation is signicant, the local creep deformation blunts the crack tip and may reduce the crack propagation rates [1719]. A similar retardation eect has also been observed when an overload is applied prior to the dwell-time [20, 21]. A phenomenon which has been observed but is rarely discussed in the open literature is out of plane crack propagation [2225] which seems to be more pronounced

4

during sustained-loads or dwell-fatigue testing of ne grained superalloys. Moverare and Gustafsson found that crack deection occurred during in-phase thermomechanical fatigue testing at 550

◦

C when long dwell-times (6 h) were introduced but not

for short (5 min) or no dwell-times [24]. No conclusive explanation to this behavior is available at the moment and further studies are needed. The present study is motivated by the fact that no work exist that systematically investigates the eect of anisotropy as a result of build-direction on the crack propagation behaviour during dwell-fatigue in AM SLM nickel-based alloys. A test program was carried out to characterise the dwell-fatigue crack propagation in different directions of SLM EOS NickelAlloy HX at 700

◦

C using compact tension (CT)

samples. Detailed metallographic investigations of the tested samples were performed in order to investigate the crack propagation mechanisms. The aim of the present study is to examine dwell-fatigue properties of SLM EOS NickelAlloy HX.

2. Experimental procedure

The material used in the current study is manufactured from EOS NickelAlloy HX powder and tested in the as-built condition as well as heat treated at both 900

◦

C and 1177

◦

C for one hour respectively. Previous work by Brodin et al. [4] on

the Hastellloy X alloy has shown that material bulk properties meet or exceed the properties of both hot-rolled and cast Hastelloy X in heat treated condition. powder material is gas atomized and sieved to a fraction (10-45 the SLM process.

µm)

The

suitable for

Blocks of EOS NickelAlloy HX were manufactured in a Eosint

M270DM machine. The nominal composition of EOS NickelAlloy HX is presented in Table 1. CT samples were cut from as-built material and heat treated blocks to investigate the inuence building direction and dwell-fatigue at 700

5

◦

C. The specimen and

Table 1: Chemical composition, EOS NickelAlloy HX, values given in wt.%. Ni

Cr

Fe

Mo

Co

Nb

Mn

Ti

Al

C

Si

W

B

56.9

19.6

10.3

8.7

2.24

0.04

0.1

0.01

0.01

0.05

0.06

1.5

0.05

potential drop (PD) instrumentation is illustrated in Fig. 1 (a). Sample measurements are shown in Fig. 1 (b) and (c) and given in Table 2. The orientations W and S indicate how the samples were notched in relation to the building direction. W indicates that the sample was notched parallel and S perpendicular to the building direction. Precracking was performed at ambient temperature using the compliance method for crack length measurements. PD was used during testing at 700

◦

C for

crack length measurements. Both precracking and the subsequent elevated temperature testing were performed according to ASTM E647-08. were used.

EDM starter notches

All samples were precracked 2 mm at room temperature.

All loading

conditions are listed in Table 2. One specimen was used for each test condition.

[Figure 1 about here.]

High temperature crack propagation tests were conducted at 700

◦

C

using 90 s

and 2160 s dwell-times, with a constant ramp up/ramp down rate yielding a 10 s single ramp time for loading and unloading. The maximum loads were successively increased step wise until noticeable crack propagation could be detected.

Testing

was done using a 100 kN Zwick servo electric Kappa 50DS tensile testing machine, equipped with a three zone furnace and a 20 A pulsed direct current potential drop (DCPD) system. Data was evaluated using evaluation code for CT-samples [26]. regarding the test setup are covered in reference [27]. deviate by more that 10

◦

More details

A crack is not allowed to

from its point of initiation at the machined notch for a test

6

Table 2: Summary of elevated temperature crack propagation tests, with the constants: thickness

B = 10 mm, width W = 25 mm, temperature 700 ◦ C, 2 mm fatigue precrack and load ratio R = 0.05. Material condition

Direction

Dwell-time [s]

∆P [N]

an

1177 ◦ C

W

2160

4000

8.5

1177 ◦ C

S

2160

4500

8.5

1177 C

W

90

2500

8.5

1177 ◦ C

S

90

4500

8.5

900 C

W

2160

2200

8.5

900 ◦ C

S

2160

3500

8.5

900 C

W

90

1500

8.5

900 ◦ C

S

90

2700

8.5

as-built

W

2160

1000

10.5

as-built

S

2160

2100

10.5

as-built

W

90

1000

10.5

as-built

S

90

1500

10.5

Cast

-

2160

4500

10.5

◦

◦

◦

7

to be considered valid according to ASTM E647-08. If the crack deviates by more

◦ than 10 , the valid crack length is underestimated and the analytical solution of the stress intensity factor,

K

[MPa

√ m],

for CT-samples obtained from ASTM E647-08

will be inaccurate. All but one sample presented in Table 2 were run to fracture.

After which all

samples were sectioned as-is, parallel to the centerline and perpendicular to the surface of the crack, so that the crack path could be studied in a cross-section. These samples were prepared by grinding and careful mechanical polishing but no etching. A Hitachi SU70 FEG analytical scanning electron microscope (SEM) was used together with various techniques such as electron channelling contrast imaging (ECCI) [28] of the crack appearance and microstructure and electron back scatter diraction (EBSD) to analyze grain orientation with orientation imaging maps (OIM) with a step size of

1 µm,

20 mm working distance and 15 kV acceleration voltage. Optical

microscopy was used to study crack paths. In order to better understand the inuence of heat treatment on the microstructure of the alloy, the development of thermodynamically stable phases at dierent temperatures were predicted using the Thermo-Calc software. The simulations were made with TC version 2017b and the TCNi8 database.

3. Results

Crack propagation rate d a/dN versus

Kmax

∆K = Kmax − Kmin ,

are minimum and maximum stress intensity factors

where

√ [MPa m]

Kmin

and

respectively

are presented in Fig. 2. A test is considered invalid if the main crack propagates out of plane by more than 10 08.

◦

according to both ASTM E399-97 and ASTM 647-

The dashed lines in Fig. 2 illustrates which samples cracked out of plane by

8

more than 10

◦

perpendicular to the loading direction.

Rendering the evaluation

method according to ASTM E399-97 and ASTM 647-08 invalid due to the change in stress mode resulting in an underestimation of the crack length

a.

All W samples

propagated perpendicular to the loading direction compared to the S samples which propagated almost immediately out of plane. In the W direction an increase in dwelltime drastically increase the crack propagation rate for the as-built and the 900 heat treated samples for the same

∆K .

The 1177

◦

◦

C

C heat treated W samples exhibit

similar propagation rates but the 2160 s dwell-time sample needed a higher

∆K

in

order to propagate the crack. Since all samples in the S direction propagated out of plane similarly makes it plausible to compare their propagation rates even though the data evaluation is not fully valid. The longer dwell-time resulted in faster crack propagation rates for all S samples. The cast sample was added to illustrate that AM materials could be considered as a new material group.

[Figure 2 about here.]

All samples exhibited intergranular cracking regardless of the dwell-time being 90 s or 2160 s long. illustrated in Fig.

The W samples all exhibited in plane crack propagation as

3 (a).

Fig.

3 (b) and (c) shows the general out of plane crack

propagation exhibited by the S samples.

Where the crack either propagated fol-

lowing a single intergranular crack as seen in Fig. 3 (b) or splitting into two large intergranular cracks depicted in (c).

[Figure 3 about here.]

General microstructure images of samples tested at 700

◦

C are displayed in Fig.

4 where column one shows images of as-built material, column two, material heat treated at 900

◦

C, and column three, material heat treated at 1177

9

◦

C, with the

building direction indicated by the arrows. (a)(c) depicts the general microstructure of all material conditions.

(d)(h) shows intergranular fracture as a result of

creep damage in the form of cavitation and cracking following streaks of detrimental topologically close-packed (TCP) phases. (e), (f ), (h) and (i) shows the general distribution of precipitates as well as streaks of TCP-phases located in grain boundaries. Testing at 700

◦

C leads to the formation of TCP-phases and carbides. Creep

damage can also be seen ahead of a crack tip as exposed in Fig. 4 (g).

[Figure 4 about here.]

The large scale OIMs of samples tested at 700

◦

C, utilizing cubic coloring and

resulting pole density plots, are depicted in Fig. 5 and show that all material conditions consists of multiple grains with preferred orientations that are elongated to a large degree. This gives evidence of the directional solidication as a result of the layer by layer building technique. Fig. 5 also shows that the main crack propagated intergranularly for all material conditions. No big microstructural dierence can be seen between the as-built material in (a) and the heat treated at 900 in (b). The main dierence is revealed in the 1177

◦

◦

C material

C heat treated material by the

dashed circle in (c) where an indication of recrystallization can be seen as well as the existence of annealing twins. All three material conditions exhibit the same type of texture as indicated by the green bands in the pole density plots.

[Figure 5 about here.]

4. Discussion

The evaluated crack propagation rates d a/dN versus

∆K

in Fig. 2 shows that

with the introduction of the 2160 s dwell-time crack propagation rates increased

10

compared to the 90 s dwell-time samples. This is most likely due to the increased amount of creep damage generated during the 2160 s dwell-time. This conforms with the ndings regarding accelerated dwell-fatigue crack propagation rates reported for other polycrystalline superalloys [25, 27, 29, 30]. The dashed lines in Fig. 2 illustrates which samples cracked out of plane by more than 10

◦

perpendicular to the loading direction rendering the evaluation method

according to ASTM E399-97 and ASTM 647-08 invalid due to the change in stress mode resulting in an underestimation of the crack length

a.

In the W direction an

increase in dwell-time drastically increases the crack propagation rate for the asbuilt and the 900

◦

C heat treated samples for the same

∆K .

The 1177

∆K

C heat

◦

C 2160 s dwell-

in order to propagate the crack.

This could be

treated W samples exhibit similar propagation rates but the 1177 time sample needed a higher

◦

due to the decreasing fraction of the unwanted

σ

and

µ

phases as well as dissolved

M6 C and M23 C6 precipitates [3133] which is also in agreement with our ThermoCalc calculations.

This could mean that the 1177

◦

C heat treatment might make

the alloy more thermally stable. Comparing the 2160 s dwell-time W samples, crack propagation rates shows that heat treating the material increases the dwell-fatigue resistance. It looks like the 900

◦

C heat treatment is the most favorable since it yields

the slowest initial crack propagation rate and converges with that of the 1177 treated material. The 2160 s dwell-time 1177

◦

◦

C heat

C heat treated material exhibited the

the fastest crack propagation rate but it also needed the highest

∆K

value to start

propagating. From an engineering perspective this becomes of utmost importance for component design and life modeling. Since selecting the improperly heat treated material could result in an increased need for inspection and/or shortened inspection intervals. All samples showed an increased crack propagation rates when the longer 2160 s dwell-time was introduced except for the W 1177

11

◦

C heat treated sample. An

increase in

∆K

at the same crack propagation rate were observed for the W 1177

◦

C

heat treated sample. A result which still puzzles the authors. All samples in the S direction propagated out of plane similarly which makes it plausible to compare their propagation rates even though the data evaluation is not fully valid. As normally expected the longer 2160 s dwell-time resulted in faster crack propagation rates [25, 27, 29, 30] for all S samples over similar

∆K

ranges. AM SLM material exhibits

anisotropic behaviour to such extent that it could be considered a new material group. A cast sample was added to show that AM material can be tailored in such a way that their dwell-fatigue resistance can be altered depending on the need for fatigue or creep resistance. The EDM starter notches are oriented horizontally and positioned to the left in Fig. 3 followed by 2 mm fatigue precracks. All cracks propagated intergraunularly. All W samples displayed in plane crack propagation as illustrated in (a) which gives valid data for d a/dN versus

∆K

evaluations. Typical out of plane crack propagation

in all S samples is exemplied in (b) and (c) and can be explained by the grains being oriented parallel to the loading direction and intergranular cracking. The S sample dwell-fatigue cracks began to propagate out of plane when the test was initiated. All S samples propagated out of plane by more than 10 versus

∆K

◦

making the resulting d a/dN

evaluations not fully valid according to ASTM 647-08. However, since

all S samples exhibited similar out of plane crack propagation the S sample d a/dN versus

∆K

evaluations could still be compared with each other.

General microstructure images of samples tested at 700

◦

C are displayed in Fig.

4 where column one shows images of as-built material, column two, material heat treated at 900

◦

C, and column three, material heat treated at 1177

building direction indicated by the arrows. lustrated in (d)(h).

◦

C, with the

Intergranular secondary cracks are il-

No signicant dierence or tendency for creep blunting nor

12

crack branching was observed between the W and S samples.

The intergranular

cracks propagated most likely due to creep cavitation which is more visible in (d) and (h). Streaks of TCP-phases are allocated in the grain boundaries weakening the grain boundary cohesion. The creep cavitation is probably an environmental eect of TCP-phase depletion revealed in (h) which could be due to oxidation or pre-existing pores from the build process allocated in the grain boundaries.

According to our

Thermo-Calc calculations the TCP-phases present in the material is

σ

and

µ

which

is in line with the ndings in [31, 32, 34]. The visible precipitates in Fig. 4 (i) is most likely

σ

[31, 34]. Other precipitates such as M 6 C and M23 C6 should be present

according to our Thermo-Calc calculations and references [3134] but have not been distinguished. A comparison of the amount and size of precipitates in Fig. 4 (d)(f ) shows that the as-built material contains more precipitates than the heat treated material. Whereas the precipitate sizes are similar in all material conditions. This indicates that the heat treatments at 900

µ,

◦

C and 1177

◦

C reduces the amount of

σ,

M6 C and M23 C6 . The large scale OIMs utilizing cubic coloring of samples tested at 700

◦

C reveals

the complex grain morphology in Fig. 5 clearer than if a inverse pole gure (IPF) coloring scheme would be used as illustrated in Fig. 6. Cubic coloring brings forth the small grains as seen in the dashed ellipses in (b) as compared to the IPF colored image in (a) where the small grains could easily be mistaken as deformation. The cubic coloring made it easier to determine that the cracks propagated intergranularly.

[Figure 6 about here.]

Fig. 5 show that all material conditions consist of multiple grains with preferred orientations and are elongated to a large degree. Which is likely due to the directional solidication as a result of the layer by layer building technique. With the help

13

of cubic coloring it is possible to see that there are not only highly elongated grains but also small more circle shaped grains. All material conditions demonstrates similar texture as shown in the respective pole density plots located in the upper right corners of each OIM. The texture should not be due to deformation but rather an eect of the directional solidication. Which is also strengthened by the fact that no changes in low angle grain boundaries closes to the crack path nor in the grains were detected.

A change in low angle grain boundaries are commonly associated with

plastic deformation which was not observed. When comparing the microstructural morphologies in Fig. 5 no apparent dierence can be seen between the as-built material in (a) and the heat treated at 900 in the 1177

◦

◦

C material in (b). A dierence was exposed

C heat treated material in form of a recrystallized grain containing an-

nealing twins highlighted by the dashed circle in (c). Large amounts of local residual stresses are generated when manufacturing components using AM SLM [35]. local residual stresses in combination with the 1177

◦

The

C heat treatment might be

enough to activate recrystallization. In order to get more valid crack propagation data samples with side grooves minimizing out of plane crack propagation could be utilized. However, the question remains if this would actually give more reliable data since a component will always be built the way it is built. This makes component testing crucial for AM parts and their applications. All this goes to show that more research in the eld of AM and fatigue of AM materials is needed.

5. Conclusions

Dwell-fatigue crack propagation tests were performed on CT samples of AM SLM EOS NickelAlloy HX. As-built and heat treated at 900

◦

C and 1177

◦

C conditions

were investigated. CT samples were cut to investigate the inuence building direction

14

as well as 90 s and 2160 s dwell-times tested at 700

◦

C. The following conclusions

can be drawn from this work:

•

The introduction of a longer dwell-time accelerates crack propagation.

•

Mainly intergranularly crack propagation was observed in the weak (W) and strong (S) directions.

All cracks in the S direction propagated out of plane

rendering the data evaluation invalid.

•

Cracks propagated preferably intergranularly and through streaks of TCPphases.

•

Highly elongated grains were observed in all samples. As well as an indication of recrystallization in the 1177

•

◦

C samples.

Similar texture was identied all material conditions.

6. Acknowledgements

The authors would like to thank Agora Materiae, graduate school, Faculty grant SFO-MAT-LiU#2009-00971, and the project teams at Linköping University, Siemens Industrial Turbomachinery AB and GKN Aerospace Engine Systems for valuable discussions. This research has been funded by the Swedish Energy Agency, Siemens Industrial Turbomachinery AB, GKN Aerospace Engine Systems, and the Royal Institute of Technology through the Swedish research program TURBO POWER, the support of which is gratefully acknowledged.

A special thanks to Vineeth Chan-

drasekharan for helping us with sample polishing.

15

7. References

[1] S. Das, J. J. Beaman, M. Wohlert, D. L. Bourell, Rapid Prototyping Journal 4 (1998) 112117.

[2] M. Agarwala, D. Bourell, J. Beaman, H. Marcus, J. Barlow, Rapid Prototyping Journal 1 (1995) 2636. doi:10.1108/13552549510078113.

[3] J. Saarimäki, The mechanical properties of lattice truss structures with loadbearing shells made of selectively laser melted Hastelloy X, Masters degree project, KTH Royal Institute of Technology, 2011.

[4] H. Brodin, O. Andersson, S. Johansson, in: ASME Turbo Expo 2013: Turbine Technical Conference and Exposition. doi:10.1115/GT2013-95878.

[5] F. Wang, The International Journal of Advanced Manufacturing Technology 58 (2011) 545551. doi:10.1007/s00170-011-3423-2.

[6] L. E. Murr, S. M. Gaytan, F. Medina, H. Lopez, E. Martinez, B. I. Machado, D. H. Hernandez, L. Martinez, M. I. Lopez, R. B. Wicker, J. Bracke, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 368 (2010) 19992032. doi:10.1098/rsta.2010.0010.

[7] P. Lott, H. Schleifenbaum, W. Meiners, K. Wissenbach, C. Hinke, J. Bültmann, Physics Procedia 12 (2011) 683690. doi:10.1016/j.phpro.2011.03.085.

[8] A. Emelogu, M. Marufuzzaman, S. M. Thompson, N. Shamsaei, L. Bian, Additive Manufacturing 11 (2016) 97113. doi:10.1016/j.addma.2016.04.006.

[9] N. Shamsaei, A. Yadollahi, L. Bian, S. M. Thompson, Additive Manufacturing 8 (2015) 1235. doi:10.1016/j.addma.2015.07.002.

16

[10] C.

Selcuk,

Powder

Metallurgy

54

(2011)

9499.

doi:10.1179/174329011X12977874589924.

[11] W. E. Frazier, Journal of Materials Engineering and Performance 23 (2014) 19171928. doi:10.1007/s11665-014-0958-z.

[12] D. G. Leo Prakash, M. J. Walsh, D. Maclachlan, A. M. Korsunsky, International Journal of Fatigue 31 (2009) 19661977. doi:10.1016/j.ijfatigue.2009.01.023.

[13] E. Andrieu, R. Molins, H. Ghonem, A. Pineau, Materials Science and Engineering A 154 (1992) 2128. doi:10.1016/0921-5093(92)90358-8.

[14] R. Molins, G. Hochstetter, J. C. Chassaigne, E. Andrieu, Acta Materialia 45 (1997) 663674.

[15] J. A. Pfaendtner, C. J. Mcmahon JR., C. J. McMahon, Acta Materialia 49 (2001) 33693377.

[16] J. P. Pédron, A. Pineau, Materials Science and Engineering 56 (1982) 143156. doi:10.1016/0025-5416(82)90167-7.

[17] J. Li, H. M. Wang, H. B. Tang, Materials Science and Engineering: A 550 (2012) 97102. doi:10.1016/j.msea.2012.04.037.

[18] J. Byrne, R. Hall, L. Grabowski, International Journal of Fatigue 19 (1997) 359367.

[19] X. Liu, B. Kang, K. Chang, Materials Science and Engineering A 340 (2003) 814.

[20] D. Gustafsson, E. Lundström, International Journal of Fatigue 48 (2013) 178 186. doi:10.1016/j.ijfatigue.2012.10.018.

17

[21] J. Saarimäki, J. Moverare, R. Eriksson, S. Johansson, Materials Science and Engineering: A 612 (2014) 398405. doi:10.1016/j.msea.2014.06.068.

[22] S. Ponnelle, B. Brèthes, A. Pineau, Superalloys 718, 625, 706 and Derivatives 92 (2001) 307319. doi:10.7449/2001/Superalloys.

[23] J.

Tong,

Engineering

Fracture

Mechanics

69

(2002)

13251337.

doi:10.1016/S0013-7944(02)00002-4.

[24] J. Moverare, D. Gustafsson, Materials Science and Engineering A 528 (2011) 86608670.

[25] J. Saarimäki, M. H. Colliander, J. J. Moverare, Materials Science and Engineering A 692 (2017) 174181. doi:10.1016/j.msea.2017.03.063.

[26] J.

Saarimäki,

Dwell-fatigueEvaluationForCompactTensionSamples,

2016.

doi:10.5281/zenodo.61241.

[27] J. Saarimäki, J. J. Moverare, M. H. Colliander, Materials Science and Engineering: A 658 (2016) 463471. doi:10.1016/j.msea.2016.01.111.

[28] R. J. Kamaladasa, Y. N. Picard, Microscopy: Science, Technology, Applications and Eduaction 3 (2010) 15831590.

[29] D. Gustafsson, E. Lundström, K. Simonsson, International Journal of Fatigue 52 (2013) 124130. doi:10.1016/j.ijfatigue.2013.03.004.

[30] K. A. Rozman, J. J. Kruzic, J. S. Sears, J. A. Hawk, Journal of Materials Engineering and Performance 282 (2015). doi:10.1007/s11665-015-1678-8.

[31] H.

M.

Tawancy,

Journal

of

Materials

doi:10.1007/BF00700780.

18

Science

18

(1983)

29762986.

[32] H. Kirchhöfer, F. Schubert, H. Nickel, Nuclear Technology 66 (1984) 139148. doi:10.13182/NT84-A33462.

[33] M. A. Arkoosh, N. F. Fiore, Metallurgical Transactions 3 (1972) 22352240. doi:10.1007/BF02643237.

[34] J. C. Zhao, M. Larsen, V. Ravikumar, Materials Science and Engineering A 293 (2000) 112119. doi:10.1016/S0921-5093(00)01049-2.

[35] J. Saarimäki, M. Lundberg, J. J. Moverare, Materials Research Proceedings 2 (2016) 7378. doi:10.21741/9781945291173-13.

19

List of Figures

1

(a) A 3D view of an instrumented CT specimen. (b) & (c) Schematic rear- and side-view drawings, with all measurements given in mm. (d) An illustration of the sample cutting orientation. . . . . . . . . . . . .

2

21

All samples compiled showing the eect of building direction between samples notched in the W and S direction.

The dashed lines show ◦ perpendicu-

which samples cracked out of plane by more than 10

lar to the loading direction rendering the data invalid according to ASTM 647-08. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

22

General crack propagation behaviour of the (a) in plane cracking of the W samples, (b) and (c) the S samples illustrating out of plane crack propagation.

4

. . . . . . . . . . . . . . . . . . . . . . . . . . . . ◦ General microstructure images of samples tested at 700 C, where

23

column one shows images of as-built material, column two material ◦ heat treated at 900 C, and column three material heat treated at ◦ 1177 C, with the building direction indicated by the arrows. Where (a)(c) depicts the general microstructure of all material conditions. (d) a secondary crack propagating intergranluarly through cavitation. (e) and (f ) TCP-phases allocated in grain boundaries.

(g) Multiple

intergranular secondary cracks. (h) An intergranular secondary crack propagating in-line with detrimental streaks of TCP-phases. 5

(i) a

magnication of the general precipitates. . . . . . . . . . . . . . . . . ◦ OIMs of samples tested at 700 C utilizing cubic coloring and build

24

direction indicated by the arrows and their pole density plots for (a) ◦ as-built condition, (b) heat treated at 900 C and (c) heat treated at ◦ 1177 C where the dashed circle shows an indication of recrystallization and annealing twins. . . . . . . . . . . . . . . . . . . . . . . . . . 6

25

OIMs illustrating the dierence between (a) IPF and (b) cubic coloring. 26

20

Figure 1: (a) A 3D view of an instrumented CT specimen. (b) & (c) Schematic rear- and side-view drawings, with all measurements given in mm. (d) An illustration of the sample cutting orientation.

21

Figure 2: All samples compiled showing the eect of building direction between samples notched in the W and S direction. The dashed lines show which samples cracked out of plane by more than 10◦ perpendicular to the loading direction rendering the data invalid according to ASTM 647-08.

22

Figure 3: General crack propagation behaviour of the (a) in plane cracking of the W samples, (b) and (c) the S samples illustrating out of plane crack propagation.

23

Figure 4: General microstructure images of samples tested at 700 ◦ C, where column one shows images of as-built material, column two material heat treated at 900 ◦ C, and column three material heat treated at 1177 ◦ C, with the building direction indicated by the arrows. Where (a)(c) depicts the general microstructure of all material conditions. (d) a secondary crack propagating intergranluarly through cavitation. (e) and (f) TCP-phases allocated in grain boundaries. (g) Multiple intergranular secondary cracks. (h) An intergranular secondary crack propagating in-line with detrimental streaks of TCP-phases. (i) a magnication of the general precipitates.

24

Figure 5: OIMs of samples tested at 700 ◦ C utilizing cubic coloring and build direction indicated by the arrows and their pole density plots for (a) as-built condition, (b) heat treated at 900 ◦ C and (c) heat treated at 1177 ◦ C where the dashed circle shows an indication of recrystallization and annealing twins.

25

Figure 6: OIMs illustrating the dierence between (a) IPF and (b) cubic coloring.

26