Subsurface grain refinement in electron beam-powder bed fusion of Alloy 718: Surface texture and oxidation performance

(1)

Contents lists available atScienceDirect

Materials Characterization

journal homepage:www.elsevier.com/locate/matchar

Subsurface grain re

ﬁnement in electron beam-powder bed fusion of Alloy

718: Surface texture and oxidation performance

Esmaeil Sadeghi

a,⁎

, Prabhat Pant

b

, Reza Jafari

c

, Ru Lin Peng

b

, Paria Karimi

a a_{Department of Engineering Science, University West, 461 86 Trollhättan, Sweden}

b_{Department of Management and Engineering, Linköping University, 581 83 Linköping, Sweden} c_{Department of Material Science and Engineering, Tarbiat Modares University, Tehran 14 115, Iran}

A R T I C L E I N F O Keywords:

Powder bed fusion Alloy 718 Shot peening Grain reﬁnement Surface texture Surface characterization A B S T R A C T

Subsurface grains of Alloy 718 additively manufactured via electron beam-powder bed fusion technique were refined using shot peening to improve the surface texture and oxidation performance. Oxidation of the speci-mens was performed at 650 and 800 °C in ambient air. Due to plastic deformation upon shot peening, com-pressive residual stress and high microstrain were generated in the subsurface region within a depth of ap-proximately 50μm. The shot-peened specimen exhibited lower surface roughness, finer subsurface grains, and higher hardness compared to the as-built specimen. Shot peening, coupled with hot isostatic pressing and heat treatment (HIP-HT), yielded superior oxidation performance with substantially low oxidation kinetics at 800 °C. The smooth surface, as well as dense and refined subsurface microstructure resulting from shot peening, fa-cilitated the formation of a continuous, protective, and adherent Cr-rich oxide scale.

1. Introduction

Additive manufacturing (AM) has been identified as a promising technique for the manufacture and repair of critical parts in the power generation industry [1]. The significant advantages offered by AM over conventional manufacturing, such as casting or forming, include com-plexity for free fabrication with low material wastage and capacity for mass-customization, which have been recognized by the industry. This has led to global interest and utilization of AM techniques in the pro-duction of fully functional parts from a variety of advanced materials, particularly Ni-Fe-based superalloys in the gas turbine industry [2].

Among Ni-Fe-based superalloys, Alloy 718 is one of the most crucial alloys on account of its promising mechanical properties and oxidation performance at high temperatures [3]. Alloy 718 has been extensively used in various parts of the high-pressure section of the compressor and turbine section of the gas turbine engine at temperatures below 650 °C. Unsurprisingly, there are high expectations for AM to deliver Alloy 718 parts with superior performance and cost beneﬁts, because it facilitates the manufacture of complex parts. Although metal AM facilities around the world have experienced massive advances over the last decade, there are several pervasive technical hurdles to be overcome before AM can proliferate the manufacturing industry and make a signiﬁcant im-pact. A common challenge in AM is achieving the desired surface tex-ture that would ensure enhanced mechanical and oxidation

performance in high-temperature environments [4,5].

The surface features found in the AM-built parts often have diverse origins. Regardless of the eﬀect of the build process and the build or-ientation, the parts often show nonuniformity in their surfaces after removal of the support structures. The other surface irregularities de-pend primarily on feedstock powder characteristics, and process para-meters, such as layer thickness, beam power, beam speed, and focus oﬀset, particularly in the contour region of the parts [6,7]. There are several types of defects detected on the surface of AM-built parts, e.g., balling defects, satellites, pits, overhangs from down-skin surfaces, staircase defects, and stacked partially-melted particles on the exterior rippled surface. In general, surface texture consists of all the above ir-regularities present on the surface of the metal AM parts, in which their origins need to be fully understood before any surface treatment.

Ballings are unfavorable droplets from the melt pools [8] and occur due to several complex physical and metallurgical processes, e.g., a highflow of molten material in the melt pool during melting. Balling in the melt pool is primarily due to a combination of low wettability of the melt pool to the previous layer, coupled with the high surface tension of the melt pool. It can also be due to splashes in the melt pool resulting from increased instability of the liquid metal [9,10]. Satellites arefine partially sintered/melted particles attached to the external surface of coarse particles, mainly due to faster cooling and solidification of the finer particles [11–13]. Satellites can also result from splashed molten

https://doi.org/10.1016/j.matchar.2020.110567

Received 25 April 2020; Received in revised form 10 August 2020; Accepted 11 August 2020

⁎_{Corresponding author.}

E-mail address:esmaeil.sadeghi@hv.se(E. Sadeghi).

Materials Characterization 168 (2020) 110567

Available online 15 August 2020

(2)

particles during the powder melting process, which eventually prevent smooth ﬂowability and impede uniform powder packing during the spreading of successive powder layers [14]. In the electron beam-powder bed fusion (EB-PBF) technique, a high-intensity electron beam is concentrated on the powder, leading to localized boiling. When the energy of the ejecta overcomes surface tension forces, ejection of the powder particles from the powder bed occurs, known as spattering [15,16]. Spattering can be avoided to some extent when the entire powder bed is pre-heated/sintered. However, there is a risk of satellite formation.

The formation of spherical particles with diameters larger than the layer thickness occurs because of rapid solidification during manu-facturing. The spherical particles are susceptible to fracture upon blade raking, leading to the formation of pits [17]. Moreover, the inclination angle of the AM-built parts leads to two different surfaces: (a) up-skin or positive, and (b) down-skin or negative. Gravity affects the melt pools created on unsupported layers, which sag into the underlying un-melted powders, resulting in high surface roughness on the downward-facing surfaces [18]. This effect can be combined with the uneven heat dis-sipation rates into the underlying powders, compared to the solid ma-terial, which creates thermal gradients, impairs the melt pools, and eventually disrupts the shape of the layer edge [19]. In the underlying region, the local heat dissipation is significantly slower due to the limited contact between powder particle surfaces and insulating air gaps between the particles [20]. Therefore, there is usually a higher risk of the formation of ballings and overhangs due to swelling (resulting from heat accumulation) [21]. The staircase defect observed in the AM-built parts is a design-dependent phenomenon inherently arising from the layer-by-layer building strategy in AM, which leads to the formation of the stairs in the side surfaces [22]. The layer thickness is considered the primary source for the severity of such defects [23]. Optimizing the process parameters in the contour region (altering the number of con-tours and adopting a proper melting strategy), using a suitable feed-stock powder, or avoiding support structures (if possible) can improve the surface texture [7,16]. The irregularities on the surfaces of the AM-built parts necessitate exploring the use of various post-surface treat-ments, such as shot peening that can be practically implemented on the parts to improve the surface texture for a specific application.

Shot peening is a cold working technique implemented to generate compressive stress on the surface of metals to improve their functional performance [24]; see Fig. 1. Shot peening entails the use of small spherical balls of high hardness to bombard an external surface with suﬃcient load to generate transverse and normal forces as well as

plastic deformation. The actual impact from the shots concentrates the energy in a tiny area, resulting in plasticflow in the impact region, which generates afine domain size and a microstrain misfit between the bulk material and surface. The strain misfit induces compressive residual stress [25]. Shot peening also suppresses surface dents and depending on the initial surface texture of the part; it can either reduce or increase the surface roughness. Thefine grains generated by shot peening are expected to enhance the corrosion resistance of the parts [26]. Thefine-grained microstructure can facilitate the outward diffu-sion of protective oxide scale-forming elements, such as Cr from the subsurface region towards the external surface of the alloy [27].

Despite the technique showing promise regarding the enhancement in the oxidation behavior of metallic materials, the oxidation perfor-mance of the EB-PBF-built Alloy 718 parts that have undergone shot peening has not yet been reported in the literature. The influence of surface and subsurface grain structure on the oxidation performance of the EB-PBF-built parts is also yet to be investigated. A thorough un-derstanding of these two factors is scientifically interesting, as well as crucial for the design and development of advanced environment-spe-cific EB-PBF-built Alloy 718 parts. In this study, the influence of shot peening on the surface and subsurface features, the microstructure, and the oxidation behavior of EB-PBF-built Alloy 718 is examined in am-bient air at 650 and 800 °C for up to 168 h.

2. Experimental setup and procedure

2.1. Feedstock powder

Alloy 718 powder with the nominal composition of 54.11Ni-19.0Cr-4.97Nb-2.99Mo-1.02Ti-0.52Al-0.12Mn-0.08C-0.04Co-Fe as bal. (all in wt%) supplied by Arcam-EBM (a GE-additive) (Mölnlycke, Sweden) was used to manufacture the specimens. The powder particles produced by the plasma atomization process (not shown here) were spherical and in the range of 45–105 μm. The powder used in the present work was a mixture of fresh and recycled (> 10 times) powders.

2.2. Electron beam-powder bed fusion (EB-PBF) process

The specimens were built using an A2X machine (Arcam-EBM, Mölnlycke, Sweden) with a constant voltage of 60 kV. The machine was equipped with a software version of 4.2.76. The beam current, layer thickness, and line oﬀset were set as 15 mA, 75 μm, and 125 μm, re-spectively. The scan speed, maximum melt current, speed function and

Build direction

Shot peening

Build direction Partially melted particles

Surface crack Lack-of-fusion defect Round porosity 1-2 m m n oi ge r r u ot n o C n oi ge r hc ta H Contour/hatch interface Satellite

Plastic deformation zone

30-50 µm Surface porosity Flat region Rough surface Bottom of a valley

a

b

Spherical balls

(3)

focus oﬀset in the hatch region were 4530 mm/s, 18 mA, 63 and 15 mA, respectively, whereas the values for the contour region were 540 mm/s, 8 mA, 6, and 3 mA, respectively. Three contours (two inner and one outer) were used in manufacturing the specimens, according to the Arcam recommended process theme. Six rods with a diameter of 20 mm and a height of 80 mm were vertically built on the build plate. The rods were then sliced into the shape of cylindrical buttons with a diameter of 20 and a height of 10 mm (Fig. 2). Only four specimens taken from the middle of the rods were used for further investigations, as these parts experienced a similar thermal history during manufacturing.

2.3. Combined hot isostatic pressing-heat treatment (HIP-HT) cycle

A combined HIP-HT cycle was performed on the selected specimens using argon as the media. A combined HIP-HT cycle provides reduced energy consumption and cost as the process is performed in a shorter period. Temperature and pressure versus time that were logged by the equipment are presented inFig. 3. The selected specimens were held in the furnace at 1200 °C for 3 h at 100 MPa, followed by furnace cooling to the solution-treatment temperature at 980 °C for 1 h, and rapid cooling (1000 °C/min) to room temperature. The double-step aging consisted of theﬁrst aging at 740 °C for 8 h, and rapid cooling to the

second aging at 635 °C for 10 h, followed by rapid cooling to room temperature. A forced convection cooling with high-pressure gas was used for rapid cooling. The pressure in diﬀerent steps of the HIP-HT treatment was controlled by adjusting the gas temperature in the fur-nace.

2.4. Shot peening

The surface of the cylindrical buttons was shot-peened using sphe-rical glass beads and compressed air with a pressure of 0.3 MPa for 30 s. The diameter of the beads was between 70 and 140μm. The surface coverage (the ratio of the area covered by the shots' indentations, as compared to the total area of the treated surface) was 150 to 200%, the peening intensity was 0.2 to 0.3 mm, and the working distance was 150 mm. Shot peening was applied on the side surface (the side parallel to the build direction) of the sliced cylindrical buttons. Following this, the specimens were cleaned ultrasonically in ethanol and dried. The ﬁnal dimensions of the specimens after shot peening were measured using a micrometer to calculate the surface area.

2.5. Oxidation exposures

The oxidation exposures were performed in a box furnace. The tests were performed in laboratory air at two different temperatures of 650 and 800 °C with the time intervals of 24, 96, up to 168 h. Alloy 718 is usually subjected to a maximum temperature of 650 °C in a long period (> 10,000 h) during its service life [28]; however, due to the limita-tions in performing such a long exposure in the laboratory scale, the alloy was exposed to a higher temperature (800 °C) to compensate the effect of exposure time. In general, four groups of specimens were prepared for the oxidation exposures. The groups were (i) as-built, (ii) shot-peened, (iii) HIP-HT, and (iv) HIP-HT + shot-peened. In each group, three specimens were used for each exposure temperature, and one specimen for the material characterization in the non-exposed condition (seven specimens per group and in total twenty-eight speci-mens). At the selected time intervals, the specimens were extracted from the furnace for weight measurement using a Sartorius™ balance (Cubis MSA3.6P0TRDM, Sartorius, Germany) with microgram resolu-tion. The weight of each specimen before and after the exposures was measuredfive times to validate the reproducibility and repeatability of the data. Four different specimen conditions were examined at each exposure time and temperature (24 specimens in total), as shown in Table 1. The oxidation kinetics of the exposed specimens was de-termined using the parabolic rate constant (kp) ( mg

cm sec

2

4 ) obtained from

the weight change results according to Eq.1: Φ= 20 mm L=10 mm L=80 mm Φ= 20 mm Sliced to 1 2 3 4 5 6 7 8 The sam ples used for post processing and oxidation exposures

Shot peened surface (green area)

Build direction

Fig. 2. Schematic of the manufactured rod and sliced cylindrical buttons. The four green specimens taken from the middle of the rods were used for further investigation. (For interpretation of the references to colour in thisﬁgure le-gend, the reader is referred to the web version of this article.)

T

emperature (°

C)

Time

140

120

100

80

60

40

20

0 Pressure (MPa)

1400

1200

1000

800

600

400

200

0

00:00:00 04:48:00 09:36:00 14:24:00 19:12:00 00:00:00 04:48:00 09:36:00 14:24:00 19:12:0 0 Temperature top Temperature bottom Pressure 100MPa 1200°C, 3 h Furnace cool 740˚C, 8 h 621˚C, 10 h 980°C, 1 h Rapid cool: 1000℃/min Furnace cool Rapid cool: 1000℃/min

(4)

⎛ ⎝ ∆ _⎞ ⎠ = k W A p. t 2 (1) whereΔW is the weight change (mg), A is the exposed surface area (cm2_{), and t is time (s).}

2.6. Characterization of specimens

The specimens were cut from a normal reference plane parallel to the build direction using an alumina cutting blade, followed by grinding using SiC papers up to 4000 grit size andﬁnally polishing with colloidal silica suspension. The specimens were etched electrolytically using oxalic acid (10 wt%) with 6 V for 5 to 10 s at room temperature. A Zeiss light optical microscope (LOM) (Axio Scope.A1, NY, USA) was used to study the microstructure of the specimens. A QUANTA-200 FEG (FEI, Oregan, USA) scanning electron microscope (SEM) equipped with X-ray energy dispersive spectroscopy (EDS), working at 15 kV was utilized to investigate the surface morphology and the cross-section of the as-built, post-treated and exposed specimens. A GAIA3-TESCAN SEM (Cambridge, UK) equipped with an electron backscatter diﬀraction (EBSD) system at a voltage of 20 kV was used to study the grain texture of the specimens. The EBSD maps have been recorded with a step size of 3μm. EBSD was used to study variations in the texture and strain in the specimens before and after post processes. HKL Channel 5 software (Oxford instrument, Abingdon, the UK) was used to visualize the EBSD maps.

A Siemens grazing incidence X-ray diﬀractometer (XRD) (D5000, Bruker-AXS, Karlsruhe, Germany) was used to identify the phases pre-sent in the specimens. The crystallite size (D), dislocation density (δ), and internal lattice strain (ɛ) were calculated based on the XRD peak broadening, according to Eqs.(2)–(4)[29]:

= λ β D(nm) k cosθ (2) = − δ nm D ( 2) 1 2 ₍₃₎ = β θ ϵ 4 tan (4)

where k is the Scherrer constant (~0.9),λ is the wavelength of the X-ray source (~0.15406 nm for Cu source),β is full width at half max-imum (FWHM) (radians) of the XRD peaks and 2θ is the peak position (radians).

For the exposed specimens, the XRD instrument was set using Cu-Kα

(λ = 0.15406 nm) in the 2θ range between 30 and 80° with the in-cident angle of 3°, the step size of 0.04 and the dwell time of 5 s. For data analysis, the Diffract-plus-Evaluation (EVA) phase identification program with the reference patterns from the International Centre for Diffraction Data (ICDD) Powder Diffraction File (PDF-2) database was used.

Residual stress measurements were performed on a four-circle Seifert XRD (3000-PTS, Mannheim, Germany) equipped with a Cr-Kα

(λ = 2.1031 Å) radiation and a linear position-sensitive detector. Tangential and axial stresses were measured on the side surface of the

specimens. The {220} planes with a nominal diﬀraction angle at 2θ ~ 128° were used, and the sin2_{ψ method with an X-ray elastic}

constant of 4.65 × 10−6MPa−1was chosen to calculate the residual stress. Each residual stress measurement involved measurements of diﬀraction peaks at eleven ψ angles from +55° to −55°.

The porosity content was measured by the image analysis technique using ImageJ software through converting the SEM micrographs of the polished specimens with a horizontalﬁeld width of 50 μm into binary images (black and white) and quantifying the percentage of the black areas [30]. The automated threshold applied to the images included all the pores depicted as the black areas in the images. The other black regions, such as non-metallic inclusions (mainly TiN and Al2O3) were

excluded from the measurement, regardless of their negligible presence. The analysis was repeated three times on each image to calculate the standard deviations.

A Vickers indenter (HV) (Shimadzu, HMV-2, Tokyo, Japan) using a load of 100 g and a dwell time of 10 s was used to measure the mi-crohardness of the specimens according to ASTM E384. The hardness of specimens was measured as an average of forty-eight measurements in the contour region. Staggering the indents ensured that no two indents were spaced less than 2.5 times the length of the indent diagonal.

White light interferometry (WLI) (Proﬁlm3D, Filmetrics, CA, USA) was used to study the topography of the specimens. The surface cur-vature and average value of the surface roughness using the arithmetic surface roughness value (Sa) in a selected area with dimensions of

3 mm × 3 mm were examined. Sa is an aerial surface roughness

parameter and a proper indicator in characterizing the surface rough-ness of the AM parts [16].

3. Results & discussion

3.1. Defects and surface roughness

Fig. 4illustrates a cross-sectional overview of the specimens via LOM before and after post-treatment (HIP-HT, shot peening and com-bined). It also shows the distribution of the few defects within the hatch and contour regions.Fig. 4a shows large lack-of-fusion (LoF) defects predominantly at the contour/hatch interface of the as-built specimen, while thefine and round-shaped pores were located generally within the hatch region. The LoF defects are usually caused by incomplete melting between subsequent layers of material or adjacent melt pools, confirming their formation to be significantly influenced by processing parameters [31]. Insufficient overlap of the melt pools within a layer, or shallow penetration depths of the melt pools between layers due to limited energy input, can create the LoF defects [32]. The presence of the inclusions, such as oxides (e.g., Al2O3) or nitrides (e.g., TiN), which

are typically reported in the EB-PBF-built Alloy 718 parts [33], can also result in the lack of sufficient fusion between the layers and the for-mation of the LoF defects [34]. The presence of the LoF defects mainly located at the contour/hatch interface is primarily related to the non-optimized line offset values used in the contour region and at the in-terface [16]. A high line offset value decreases the energy input, which eventually led to insufficient overlap between the melt pools at the contour/hatch interface [16]. As seen inFig. 4b, similar defects were found in the shot-peened specimen, implying that the effect of shot peening was limited to a shallow distance from the surface. However, the combined HIP-HT was successful in reducing the level of defects, particularly the LoF defects (Fig. 4c). It was reported that HIP is ef-fective in closing the LoF defects and round-shaped pores with diameter below 150μm [35]. Because most of the LoF defects had been removed after HIP-HT, it is confirmed that low energy input during manufacture is the main reason for the formation of these defects as opposed to nano-sized Al-rich oxide particles found on the feedstock powder surface, as previously reported in the literature [33]. A thin Al-rich oxide layer of approximately 30 nm can be found on the fresh powder surface (not presented here), which then grows up to 200 nm after 14 times of usage Table 1

Designations of the post processing for the specimens.

No. Shot peening HIP-HT Exposure temperature (°C) Exposure time (h)

1 No No 650 24, 96, and 168 2 No Yes 3 Yes No 4 Yes Yes 5 No No 800 24, 96, and 168 6 No Yes 7 Yes No 8 Yes Yes

(5)

in the EB-PBF machine and subsequent powder recycling [33]. The few round-shaped pores still detected in the HIP-HT specimen were due to the presence of argon, which is produced primarily by the plasma atomization process during powder manufacturing [36]. The argon gas entrapped inside the pores cannot escape, dissolve, or diﬀuse through the material during HIP-HT [37]. Therefore, it is possible that the pores that were shrunk or disappeared during HIPing reappear and grow during the subsequent heat treatment step in HIP-HT [38]. While the HIP-HT process did not aﬀect the level of surface-connected pores and cracks, most of the irregularities in the surface, such as partially-melted loose particles, surface cracks, pits, peaks, and valleys were removed after shot peening. This can be observed in both shot-peened and HIP-HT + shot-peened specimens shown in Fig. 4b and d, re-spectively. Because the focus of this study was primarily on the contour region, the level of porosity was only measured in this region, and the results are given in Fig. 4e. The results illustrated that the as-built specimen had the highest level of porosity at 0.49 ± 0.1 area%, whereas the HIP-HT + shot-peened specimen had the lowest level of porosity at 0.16 ± 0.04 area%. The diameter of the cylindrical disc in the as-built condition was measured as 19.54 ± 0.3 mm (Fig. 4a). In contrast, the diameter of the disc after shot peening was 18.87 ± 0.3 mm (Fig. 4b), implying that the surface and the contour region were suppressed after post-surface treatment. However, the level of porosity in the contour region was still measured, and in general, shot peening successfully reduced the level of porosity in the contour region for both the as-built and HIP-HT specimens. The reduced level of porosity after shot peening can be acknowledged as a result of the

multiple shot impacts and the high extent of plastic deformation on the top surface layer that tended to close the surface pores [39]. Similar occurrence was reported by Manfredi et al. [40] who observed slight porosity reduction after shot peening.

3.2. Microhardness

The microhardness values of the specimens before and after post-treatments are presented inFig. 5. The average hardness values were reported as 406 ± 6, 410 ± 8, 470 ± 9, and 472 ± 9HV0.1for the

0.49 0.39 0.19 0.16 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

As-built Shot peened HIP-HT HIP-HT + shot peened Porosity (area%)

Build direction

LoF defects

Round pores

5 mm

Contour/hatch interface

Shot peened surfaces

e

Contour/hatch interface Measurement area B uild direction

LoF defect

Fig. 4. LOM images of the specimens before and after post-treatments showing the distribution of the defects within the hatch and contour regions; a) as-built, b) shot-peened, c) HIP-HT, d) HIP-HT + shotpeened specimens, and e) the level of porosity.

350

400

450

500

550

0 90 180 270 360 450 540 630 720 810 900 990

Microhardness

(HV

0.1

)

Distance from the surface (μm)

As-manufactured Shot peened HIP-HT

HIP-HT+ Shot peened As-built

Fig. 5. a) Micro-hardness proﬁles of the specimens before and after post-treatments.

(6)

as-built, shot-peened, HIP-HT, and HIP-HT + shot-peened specimens, respectively. A significant increase in the microhardness values after the shot peening treatment was detected in the subsurface regions of both the as-built and HIP-HT specimens. Moreover, the depth of the work-hardening effect can be evaluated using the location where the hardness profile plateaus. Only the first indent (performed at a depth of ~30μm) exhibited high hardness, and the effect of shot peening was absent in the second indent (performed at a depth of ~60 μm). Therefore, it can be concluded that the depth of the work hardening in both shot-peened and the HIP-HT + shot-peened specimens extended approximately 30–60 μm inwards from the surface.

Based on the ASTM E384 standard, the minimum distance between the test indentations must be at least 2.5 times the indent diagonal (d) in the Vickers microhardness test to avoid distortion of the hardness results due to deformation of the material structure. Thus, the relaxa-tion (reaching the plateau) can be directly attributed to the change in the microstructure during shot peening. Due to large work-hardening, grain reﬁnement, and dislocations pile-up in the shot-peened specimen, a higher microhardness was obtained near the surface. Similarly, Foss et al. [25] resulted that the strain became severe in the work-hardened region, and hardness increased considerably, peaking at the subsurface region. Zhao et al. [41] similarly showed that the increase in hardness of a shot-peened Ni-based superalloy was due to both work-hardening and grain reﬁnement. The improvement of hardness values indicated that the surface of specimens can be strengthened by the formation of nanocrystals and deformation twins during shot peening.

3.3. Surface topography and roughness

The SEM micrographs of the topography of the specimens before and after post-treatments (HIP-HT and shot peening) are shown in Fig. 6. The partially-melted particles can be identiﬁed on the surface of the as-built (Fig. 6a), and the HIP-HT (Fig. 6c) specimens. In contrast, the shot-peened and HIP-HT + shot-peened specimens showed smooth surfaces. It can be seen inFig. 6b that the bottom of valleys (the region between adjacent particles) were still present even after shot peening. However, there were no signiﬁcant signs of valleys in the HIP-HT + shot-peened specimen (Fig. 6d).

The surface roughness (Sa) evaluated via white light interferometry

(WLI) before and after post- treatment (HIP-HT and shot peening) are shown inFig. 7. When compared to the as-built specimen, it is apparent that shot peening led to a signiﬁcant decrease in the Savalues from an

initial value of approximately 71 to 23μm. The decrease in the Sa

va-lues can be attributed to the impingement of the surface by high-energy

shots resulting in aﬂat surface. The results showed that shot peening was very eﬃcient in decreasing the surface roughness of the EB-PBF specimens as shot peening removed the partially-melted loose particles and deformed the sharp peaks of the surface primarily by suppressing the peaks and protrusions. The Savalue was higher for the shot-peened

specimen than for the HIP-HT + shot-peened specimen. This was due to the high impact energy of the shots that resulted in formation of more pronounced indentations on the as-built specimen, which was softer that the HIP-HT specimen (seeSection 3.2). The results suggest that, if a hard material is exposed to suﬃciently high energy impact, less ac-centuated surface indentations, and thus lower surface roughness will be achieved [39].

The effectiveness of shot peening in changing the surface roughness is highly dependent on the original surface roughness before shot pe-ening [42]. Zaleski et al. [43] examined the effect of shot peening on the topography of the milled Alloy 718 parts using different shot sizes and impact energy. While it was shown that the surface roughness in-creased after shot peening of the milled specimens (from an Raof ~0.65

to 2.26μm), optimization of the shot peening process parameters was recommended for further reduction in the surface roughness. It appears that the shot peening parameters selected in the present study suc-cessfully reduced the surface roughness of the EB-PBF-built parts. The improvement in the surface roughness demonstrated the promising ef-fect of shot peening and its applicability to practical cases.

3.4. Grain morphology and texture

The SEM images of the cross-sections of the contour region of the specimens inFig. 8showed the formation of mixed grain morphology, includingﬁne columnar and equiaxed grains in the as-built condition. Upon shot peening, plastic deformation was occurred in the subsurface region, as shown inFig. 8b, c, e, and f. There was a clear hardened zone near the surface, corresponding to the plastic deformation of the sub-surface region.Fig. 8c shows the strain contrast at a depth below the dense and hard zone, whereas the corresponding zone inFig. 8f shows the slip bands. The thickness of the work-hardened layer was approxi-mately 50 μm. Such a low penetration depth was due to the initial hardness of Alloy 718, which is relatively high (> 400HV0.1) [44,45],

as recorded in the specimens of the present study (Section 3.2). The plastic deformation was uniform, indicating that repeated mechanical loads on the surface were similarly concentrated in all the directions. Shot peening led to the formation ofﬁne subsurface grains within a 50μm-thick region from the surface, whereas HIP-HT resulted in the growth of grains in the contour region.

1 mm

a

b

c

d

Spattered and partially-melted particles

Build direction

Measurement area

bottom of valleys

Fig. 6. SEM micrographs (BSE mode) of the topography of the specimens; a) asbuilt, b) shot-peened, c) HIP-HT, and d) HIP-HT + shot-peened specimens. The topography of the as-built and HIP-HT specimens was very similar, as the BSE detector could not detect the thin oxide layer formed on the latter specimen.

(7)

As can be observed in Fig. 8b, c, e, and f, the subsurface micro-structure was refined by shot peening and deformation bands, as well as deformation twins emerged. The development of the twins is de-termined by twinning stress, which has Hall-Petch relation with the average grain size [46]. During plastic deformation, the dislocations move on the slip planes until they reach barriers, such as grain boundaries or precipitates. The dislocations pile-up is then created, which hinders further movement of the dislocations. Simultaneously, the internal stress attains a critical level, in which the twins form. Alloy 718 has a face-centered cubic (FCC) structure with a small stacking fault energy, which favors twinning under a high stress (or strain) rate [47]. Upon shot peening, intensified plastic deformation leads to fur-ther grain refinement, and twinning can hardly be formed in the refined grains. In a previous study [41], many dislocations were found after shot peening within a depth of around 25μm from the surface, and the grains were also refined in this region. The dislocations started to tangle together near the treated surface, and dislocation walls were merged due to dislocations pile-up at grain boundaries.

The electron backscattered diffraction (EBSD) inverse pole figures (IPFs) and polefigure (PF) mapping of the specimens before and after post-treatments are presented inFig. 9. In the contour region of the

as-built specimen, the multispot melting strategy and defects led to the formation ofﬁne columnar and equiaxed grains with either random texture (no texture) or a weak texture in < 001 > direction parallel to the build direction [48]. The contour region with the above hetero-geneous grain morphologies is usually not as textured as the hatch re-gion with long columnar grains in < 001 > direction along the build direction [49].

In the shot-peened specimens, the strain was generated locally at the deformation-induced zone. Due to the very fine grain size and high defect density near the surface of the specimens, the EBSD mapping was poor in this region, resulting in an unrecognizable grain structure. Based on these results, the grain size in the contour region of the shot-peened specimens wasfiner than that of the as-built specimen. The HIP-HT specimens had coarser grains in the contour region than the spe-cimens without HIP-HT, which was primarily attributed to grain growth during the high-temperature treatment. Because no delta phase was found in the contour region, it is concluded that the grains readily grew during HIP-HT without any barrier. Moreover, the motive for the grain growth in the contour region during HIP-HT was reported to be a de-crease in the grain boundary area of thefiner grains to achieve a lower surface free energy [50]. As the pole figures in Fig. 9 show, the

71 23 84 19 0 20 40 60 80 100 120

As-built Shot peened HIP-HT HIP-HT + Shot

peened Sa (μm )

a

b

c

d

e

Fig. 7. WLI maps from the surface of the specimens before and after post-treatments; a) as-built, b) shotpeened, c) HIP-HT, and d) HIP-HT + shot-peened specimens as well as the measured Sa values.

1 mm 1 mm 1 mm 1 mm

b

d

e

Sharp valleys Sharp valleys LoF defect Round porosity Contour/hatch interface

Build direction _Measurem

ent

area

a

n

oit

ce

ri

d

dli

u

B

Hatch/contour interface Hatch/contour interface Hatch/contour interface Hatch/contour interface 100 µm Dense and hard zone Twins 100 µm

c

f

Dense and hard zone

Slip bands Strain

contrast

LoF defect

Fig. 8. SEM micrographs (BSE mode) of cross section of a) as-built, b) shot-peened, c) shot-peened (at high magniﬁcation), d) HIP-HT, e) HIP-HT + shot-peened and f) HIP-HT + shot-peened (at high magniﬁcation) specimens.

(8)

maximum of multiple uniform density (MUD) of the grain morphology, which describes the number density of crystals aligned in a crystalline orientation [51], did not signiﬁcantly change after shot peening.

The MUD values in the as-built and shot peened specimens were approximately 5. Performing shot peening did not change the texture degree in the as-built condition. However, the texture was slightly in-creased in the HIP-HT condition, as can be observed by comparing Fig. 9(a–c) and (b–d). The MUD values in the HT and HIP-HT + shot-peened specimens were approximately 12 and 15, respec-tively, which confirmed the slightly stronger texture of the two speci-mens after HIP-HT. The slightly stronger texture after HIP-HT found in this study contrasts with the weaker texture observed in the literature after a similar post-treatment condition [52,53]. In the present study, EBSD was performed only in the contour region, where a weak texture was identified in the as-built condition. However, most of the EBSD works in the literature correspond to the hatch region, where a strong texture was reported in the material. Recrystallization and grain growth after the thermal post-treatments, particularly HIPing were found as the primary reasons for the weak texture in the hatch region [54]. The EBSD-IPF of the HIP-HT specimen in the present study confirms that grain coarsening occurred in the contour region without any re-crystallization. Moreover, it is worth mentioning that grain coarsening occurred in the contour region of the HIP-HT specimen; therefore, a lower number of grains per area were involved in the EBSD analysis compared to the as-built condition.

The recrystallized fraction components of the specimens were ana-lyzed using the EBSD maps and the Channel5 software, and the results are shown inFig. 10. If the internal average misorientation angle within the grain is above 1°, the grain was defined as deformed. Substructured grains consisted of subgrains whose internal misorientation is under 1° but the misorientation from subgrain to subgrain is above 1°. All the remaining grains were classified as recrystallized. A small presence of the deformed grains in the as-built specimen inFig. 10a suggests that a few grains were in a degraded or deformed condition, although the material did not physically experience any mechanical deformation process. Such deformation was associated with the high cooling/soli-dification rates during successive thermal cycling in EB-PBF [55]. As illustrated inFig. 10b–d, both the peened and the HIP-HT + shot-peened specimens showed a high fraction of deformed grains (red) near the surface. The HIP-HT treatment increased the fraction of both re-crystallized (blue) grains, but it reduced the fraction of the deformed grains (red); seeFig. 10c–d. Generally, a larger fraction of recrystallized grains indicates a rather perfect or well-annealed material, and pos-sibly, lower degradation or deformation of the microstructure [56], as shown in the HIP-HT specimen; seeFig. 10c.

3.5. Phase and residual stress analysis

The XRD patterns of the Alloy 718 powder and the as-built, shot-peened, HIP-HT, and HIP-HT + shot-peened specimens are presented in

a

_b

c

_d

Contour/hatch interface

Measurement area

Build direction

MUD

001

101

111 Y (build direction)

X

Z

Fig. 9. EBSD images of the specimens in the contour region before and after post-treatments; a) asbuilt, b) shot-peened, c) HIP-HT, and d) HIP-HT + shot-peened specimens.

(9)

Fig. 11. The feedstock powder showed the (111), (200), and (220) peaks for theγ-Ni solid solution phase at 2θ ≈ 43.5, 51, and 75°. The ultrafine strengthening γ״ phase was not detected as it is most likely below the XRD identification limit. The high hardness of the four spe-cimens (above 400HV0.1) confirms the presence of the γ″ phase;

how-ever, its absence in the XRD patterns was due to the coinciding peaks of theγ″ and γ phases. The powder had a low fraction of oxides, shown by the peak of Al2O3that was not observed in the other specimens. The

powder used in the EB-PBF process is usually a combination of fresh and recycled powders. Hence, the presence of Al2O3in the Alloy 718

powder is expected due to the high oxygen affinity of Al [34]. All the specimens maintained theγ-Ni phase found in the powder. Compared to the powder, the XRD patterns of the as-built specimen exhibited a significant peak broadening, because of the generation of microstrain during rapid cooling in EB-PBF. No phase transformation was induced by shot peening, as similar peaks were observed in the shot-peened specimen. Further peak broadening was observed after shot peening because of the generated microstrain and grain refinement during work hardening near the surface.

The crystallite size, dislocation density, and microstrain in the four specimens were measured based on the XRD results and presented in Table 2. The crystallite size was decreased by approximately 37% after shot peening of the as-built specimen. The reduction in the crystallite size was ~52% after shot peening of the HIP-HT specimen.

According to Table 2, HIP-HT led to the microstrain relaxation compared to the as-built specimen. In the as-built specimen, the thermal residual stresses and microstrains were aggregated through cyclic heating and cooling of the liquid metal during EB-PBF. Shot peening created a high level of microstrain near the surface of the shot-peened and HIP-HT + shot-shot-peened specimens. Inhomogeneous plastic strains in the crystalline substances, due to slip of lattice planes, results in the generation of residual stresses. In contrast, elastic strains are due to changes in crystal interplanar spacing [57]. These strains change the structure of the crystallographic array and shift the Bragg angles. In general, a decrease in the interplanar spacing (more elastic strain) or an increase in the slip of lattice planes (more plastic strain) shifts the peaks

to higher 2θ values [29].

The variation trend of dislocation density for the as-built and HIP-HT specimens after shot peening was similar. While, the HIP-HIP-HT spe-cimen showed the lowest dislocation density (~3.09 nm−2) followed by the as-built specimen (~5.8 nm−2), the dislocation density of the spe-cimens after shot peening significantly increased to approximately 14.66 and 13.36 nm−2in the shot-peened and HIP-HT + shot-peened specimens, respectively. Upon shot peening, several balls repeatedly impact on the surface. Starting with the movement of dislocation in various slip systems, dislocation density becomes larger during shot peening [58]. The interactions between dislocations (in forms of cross or/and tangle) increase in the plastic deformation layer, which increase the dislocation slip resistance. Moreover, the pining effect of the pre-cipitates in reducing the dislocations movements is another reason for the increased dislocation density after shot peening. The phases, such as carbide orδ phase can act as the sink sources of dislocations during repeated deformation; therefore, the dislocation density around the precipitates increases significantly after shot peening. The dislocations in the work-hardened layer are difficult to move, because of fine do-main size, high microstrain and high density dislocation existing in this region [58].

The residual stress (axial and tangential) values for the specimens before and after post-treatments were measured using the sin2ψ method and are presented inFig. 12. The axial direction corresponds to the build direction, where the residual stress is lower than the in-plane stress. As the stresses were measured at the surface, the radial stress component was considered to be zero.Fig. 12presents the high com-pressive stresses (> 800 MPa) in the shot-peened specimen, which were introduced in the plastically-deformed surface region. The impact of the high-speed balls on the surface of the specimen caused it to expand laterally, and constraint from the undeformed bulk leads to the gen-eration of compressive stresses in a surface layer. The very large stan-dard deviation in the HIP-HT specimen was primarily due to the large grain size in the subsurface region caused by the thermal post-treatment (seeFigs. 8 and 9), which reduced the accuracy of the measurement. The coarse grains in the two specimens resulted in the XRD patterns

Build direction

a

_b

c

_d

Fully recrystallized grains

Substructured grains

Deformed grains

Fig. 10. Deformed, recrystallized, and sub-structured grains in a) as-built, b) shotpeened, c) HIP-HT, and d) HIP-HT + shot-peened specimens. If the internal average misorientation angle within the grain is above 1°, the grain is deﬁned as deformed. Substructured grains consisted of subgrains whose internal misorientation is under 1° but the misorientation from sub-grain to subsub-grain is above 1°. All the re-maining grains were classiﬁed as re-crystallized.

(10)

generated from a few grains, which were not representative of the overall residual stress distribution. More advanced techniques, such as synchrotron X-ray or neutron diﬀraction, are recommended for a more

reliable comparison between the as-built and HIP-HT specimens. The difference in the tangential and axial compressive stresses in the HIP-HT + shot peened sample was due to the HIP-HIP-HT-induced changes in the mechanical properties. The HIP-HT sample likely had a weaker response to shot peening due to its higher hardness (seeFig. 5), leading to the lower amount of compressive stresses in comparison to the shot-peened sample. Holmberg et al. [42] demonstrated that shot peening in Alloy 718 was able to produce high compressive stresses of approxi-mately −900 MPa near the surface and −1400 MPa at a depth of 50 μm. Bhowal et al. [59] also reported the generation of the com-pressive stresses after shot peening of Alloy 718 of approximately −900 MPa near the surface and −1200 MPa at a depth of 50 μm. Such compressive stress enhanced fatigue life 5–10 times more than that of a machined specimen. Chamanfar et al. [45] measured the compressive residual stress as−733 MPa at the surface to the maximum compres-sive stress of−1083 MPa at 70 μm beneath the surface. Beyond this depth, the magnitude of the compressive residual stresses decreased and eventually approached zero at a depth of 300μm. The profile of the residual stresses as a function of the depth below the surface was also studied by Chamanfar et al. [45]. A gradual decay in the compressive residual stress in the shot-peened Alloy 718 specimens was identified and the depth of the shot-peening affected zone was measured as ap-proximately 35μm. It should be emphasized that several factors, such as shot velocity, shot size, shot hardness,flow rate, impact angle ex-posure time and initial surface texture can influence the magnitude of the compressive residual stresses beneath the surface [60]. For a given surface texture, there is an optimum condition that can appropriately improve the surface and simultaneously result in a microstructure gradient. The slight differences in the level of the compressive stress reported in the literature on shot-peened Alloy 718 are attributed to the changes in the above parameters. The rigorously-controlled shot pe-ening parameters insure the repeatability and accuracy of shot pepe-ening treatment at industrial scale [39]. The compressive stress found near the surface of the as-built specimen was most likely because of powder blasting during powder removal (in the powder recovery system) from the residual powder cake.

3.6. Weight change study

Fig. 13(a–b) shows the weight change of the as-built and post-treated specimens exposed at 650 and 800 °C for up to 168 h in ambient air. At both temperatures, the as-built specimen showed the highest weight change. The high weight change was due to oxygen uptake during the high-temperature exposure, resulting in the formation of bulky metal oxides. The weight change measured for the as-built spe-cimen was 0.02 ± 0.004 mg/cm2_{after 168 h of exposure at 650 °C,}

while it was 0.01 ± 0.004 mg/cm2_{in the HIP-HT specimen. At 800 °C,}

a weight change of 0.20 ± 0.02 mg/cm2was recorded for the HIP-HT + shot-peened specimen, while the as-built specimen had a weight Al2O3 HIP-HT+ shot-peened Shot-peened HIP-HT As-built Powder

Fig. 11. XRD patterns of the specimens before and after post-treatments.

Table 2

Crystallite size, dislocation density and microstrain of the specimens before and after post-treatments based on the XRD results inFig. 11.

Specimens 2θ (°) β, or FWHM (°) Crystallite size, D (nm) Dislocation density,δ (nm−2) Microstrain,ε (×10−3) D (nm) δ (nm−2₎ _{ε (×10}−3₎ As-built 43.65 0.55 15.54 4.14 6.00 13.42 ± 1.65 5.80 ± 1.39 5.70 ± 0.52 50.69 0.66 13.22 5.72 6.12 74.53 0.87 11.51 7.55 4.97 Shot-peened 43.82 0.88 9.69 10.66 9.59 8.43 ± 1.00 14.66 ± 3.43 9.05 ± 0.82 50.76 1.05 8.37 14.28 9.66 74.72 1.38 7.25 19.04 7.88 HIP-HT 43.77 0.40 21.64 2.13 4.30 18.47 ± 2.47 3.09 ± 0.80 4.13 ± 0.34 50.87 0.48 18.15 3.03 4.45 74.76 0.64 15.62 4.10 3.65 HIP-HTx + xshot-peened 43.50 0.83 10.32 9.38 9.06 8.85 ± 1.10 13.36 ± 3.11 8.69 ± 0.88 50.46 1.03 8.54 13.72 9.53 74.37 1.30 7.68 16.97 7.47

(11)

change of 0.41 ± 0.03 mg/cm2.Fig. 13(c–d) presents (Δw/A)2versus time (s), using Eq. (1), to calculate the values of kp, presented in Table 3. The kpvalues increased signiﬁcantly when the specimens were

exposed at 800 °C compared to 650 °C. At 650 °C, the as-built specimen recorded the highest kpvalue (0.7 × 10−9mg2/cm4s), while the

HIP-HT specimen had the lowest kpvalue (0.2 × 10−9mg2/cm4s). The

shot-peened specimen and the HIP-HT + shot-shot-peened specimen showed almost similar kpvalues at 650 °C (0.4 × 10−9mg2/cm4s). At 800 °C,

the kpvalues were almost two to three orders of magnitude larger than

that of 650 °C. Similarly, the as-built specimen had the highest kpvalue

(2.7 × 10−7mg2/cm4s), while the HIP-HT + shot-peened specimen had the lowest kpvalue (0.7 × 10−7mg2/cm4s). The kpvalues obtained

in the present study were in agreement with the kpvalues obtained in

the literature, where wrought and EB-PBF-built Alloy 718 were studied

-39

-815

-197

-622

-229

-851

-113

-229

-1200

-1000

-800

-600

-400

-200

0

200 Residual stress

(MPa)

Axial

Tangential

Axial stress

Tangential stress

Axial stress

Area of intresset

for stress analysis

As-built Shot-peened HIP-HT

+ shot-peened

T

ensile

Com

pressive

HIP-HT

Build direction

Fig. 12. The residual stress (axial and tangential) values of the specimens before and after post treatments measured using the sin2_{ψ method.}

0 0.01 0.02 0.03 0.04 0.05 0.06 0 24 48 72 96 120 144 168 W eig ht g ain (mg /cm 2) Time (h) As-manufactured Shot peened HIP-HT

HIP-HT + shot peened

0 0.1 0.2 0.3 0.4 0.5 0.6 0 24 48 72 96 120 144 168 W eig ht g ain (mg /cm 2) Time (h) As-manufactured Shot peened HIP-HT

y = 0.000007x + 0.000015 R² = 0.993573 y = 0.000004x + 0.000021 R² = 0.948437 y = 0.000002x + 0.000005 R² = 0.973801 y = 0.000004x - 0.000008 R² = 0.994220 0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006 0.0007 0 10 20 30 40 50 60 W eig ht g ain (mg 2/cm 4) Time (104_sec.) As-manufactured Shot peened HIP-HT

y = 0.002730x + 0.003001 R² = 0.993714 y = 0.001288x + 0.000863 R² = 0.996159 y = 0.001546x + 0.008966 R² = 0.928119 y = 0.000672x - 0.000456 R² = 0.997488 0 0.05 0.1 0.15 0.2 0 10 20 30 40 50 60 W eig ht g ain (mg 2/cm 4) Time (104_sec.) As-manufactured Shot peened HIP-HT

As-built As-built

Fig. 13. The weight gain of the specimens after the exposure at a) 650 °C and b) 800 °C for up to 168 h in ambient air, as well as the squared of weight gain versus time (×104 s.) of the exposed specimens at c) 650 °C and d) 800 °C.

Table 3

Weight gain parabolic rate constant (kp) (mg2/cm4s) determined fromFig. 13.

Specimens Temperature (°C) 650 800 As-built 0.7 × 10−9 2.7 × 10−7 HIP-HT 0.2 × 10−9 1.5 × 10−7 Shot-peened 0.4 × 10−9 1.3 × 10−7 HIP-HTx + xshot-peened 0.4 × 10−9 0.7× 10−7

(12)

in similar exposure conditions [5,61,62].

3.7. Oxidation products

The XRD analyses of the exposed specimens facilitated to char-acterize the phases in the formed oxide scales, as presented inFig. 14. Based on the XRD results, the oxide layer formed on the specimens were almost similar, but with diﬀerent intensities. The peaks were linked to oxides of several alloying elements in Alloy 718, including Ni, Nb, Fe, Al, Ti, and Cr. The primary peaks ofγ-Ni in the specimens exposed to 650 °C were easily distinguished from those of the specimens exposed at 800 °C.

At 650 °C, the intensity of the main peaks (for instance, comparing the one identified at 2θ ≈ 43.5°) in the HIP-HT specimen is much higher than the intensities of the main peaks for the other three spe-cimens, which are almost identical. At 800 °C, the HIP-HT + shot-peened specimen recorded the highest peak intensity, followed by the shot-peened specimen. The high peak intensity of the main phases (γ-Ni) is an indication of the thickness of the oxide scale formed. A low oxide scale presents an opportunity for X-rays to penetrate further into the bulk material. The peak intensity of the HIP-HT + shot-peened specimen at 800 °C was higher than at 650 °C, confirming that such post-treatment is more beneficial at higher temperatures. The HIP-HT specimen had the lowest main peaks intensity among the other three specimens at 800 °C. These observations confirmed the low weight change, and consequently, thin oxide scales shown inFig. 13.

The oxide scales were a mixture of Cr2O3, NiCr2O4, and TiO2, with

varying fractions in the two exposure temperatures. The peaks of the

material were more intense after 168 h of exposure at 800 °C than at 650 °C, indicating a greater thicker oxide scale formed at 800 °C. At 650 °C, TiO2was not prominent, but Cr2O3and NiCr2O4were relatively

more dominant. At 800 °C, the oxide scale was primarily consisted of Cr2O3, NiCr2O4, and a small fraction of TiO2. The level of TiO2in the

oxide scale increased with increasing temperature. Similar oxide scales were also identiﬁed on wrought Alloy 718 exposed to ambient air at elevated temperatures [62].

3.8. Topography of exposed specimens

Fig. 15illustrates the SEM micrographs of the oxide scales formed on Alloy 718 at 650 and 800 °C for 168 h. The oxide particles grew by increasing the temperature. The oxide particles formed at 800 °C were coarser than at 650 °C. At 650 °C, mainly Cr-rich oxides formed on all the specimens (based on an EDS analysis not presented here). Based on the EDS results of the oxide scales, which were rather identical for the specimens exposed at 800 °C, Ni, Cr, Fe, and Nb were similarly present in the scales. The level of the elements in the oxide scales varied showing the protectiveness of the scale. For instance, the higher level of Cr in the oxide scale formed on the HIP-HT + shot-peened specimens showed the formation of a more protective scale, which was supported by a uniform distribution of Cr in the bulk material. The presence of Ti and Al in the EDS results was attributed to the formation of the Al- and Ti-rich oxide scales, such as Al2O3and TiO2. The low oxygen content at

the grain boundaries favors the formation of such oxides that only need a suﬃcient level of oxygen [62]. The oxide formed on the as-built specimen at 800 °C had the typical morphology of spinel oxide, while

a

b

HIP-HT+shot-peened Shot-peened HIP-HT As-built HIP-HT+shot-peened Shot-peened HIP-HT As-built

(13)

the oxide formed on the shot-peened specimen had the typical mor-phology of a Cr-rich oxide. The oxide in the former was rather uniform, but discontinuous in the latter. A similar oxide rich in Cr, Nb, and Ti formed on the surface of both the HIP-HT and the HIP-HT + shot-peened specimens. However, the Cr content seemed to be higher in the latter due to the morphology of the oxide formed. The thicker oxide formed on the grain boundaries (Fig. 15-b3), verifying that the main elemental diﬀusion path was through the grain boundaries [63]. 3.9. Cross-section of exposed specimens

Figs. 16 and 17show the oxide scales in a cross-section of the ex-posed specimens at 650 and 800 °C for 168 h. By increasing the tem-perature, the oxide thickness was increased on all the specimens. At 650 °C, apart from the as-built specimen, a rather dense, thin, and continuous oxide scale formed on the surface of the other three

specimens. As illustrated in Fig. 16, a dense layer of approximately 1μm of a Cr-rich oxide (mainly Cr2O3based on the XRD results) formed

on the surface of the as-built specimen at 650 °C, while there was a much thinner layer of a similar oxide on the surface of the other spe-cimens, which was diﬃcult to observe even at very high magniﬁcations. The results were in agreement with the XRD results of the exposed specimens. Internal oxidation was not observed in the exposed speci-mens at 650 °C.

Fig. 17shows that at 800 °C, the thickness of the oxide scale on all the specimens was approximately 3–4 μm. A few voids were observed in the scale/substrate interface of the as-built, HIP-HT, and shot-peened specimens, while the oxide had more integrity to the substrate in the HIP-HT + shot-peened specimen.Fig. 17shows that the internal oxi-dation occurred to a depth of approximately 5–10 μm, and primarily along the grain boundaries. Based on the EDS analysis, the oxide scales formed on all the specimens at 800 °C were a continuous layer enriched

a1

a2

a3

a4

b1

b2

b3

b4

10 μm 10 μm 10 μm 10 μm 10 μm 10 μm 10 μm 10 μm

Space between

particles

Discontinuity in

the oxide scale

Grain boundary

diffusion

Cr-rich oxide

_{Cr-rich oxide}

wt%

O

Ni

Cr

Fe

Al

Nb

Ti

1

27.4

6.8

29.1

29.3

0.6

6.2

0.6

2

28.8

5.1

31.3

27.1

0.5

6.8

0.4

3

33.1

4.9

30.2

25.6

0.6

5.2

0.4

4

32.6

5.3

34.5

23.5

0.3

3.5

0.3

1

2 ₃

4

Fig. 15. SEM topographic micrographs (SE mode) of the specimens exposed at a) 650 °C and b) 800 °C for 168 h in ambient air; 1) as-built, 2) shot-peened, 3) HIP-HT, and 4) HIP-HT + shot-peened specimens.

a1

a2

a3

a4

b1

b2

b3

b4

10 µm 10 µm 10 µm 10 µm

5 µm 5 µm 5 µm 5 µm

1 µm

Fig. 16. Cross-sectional SEM micrographs (BSE mode) of the specimens exposed at 650 °C for 168 h in ambient air; 1) as-built, 2) shot-peened, 3) HT, and 4) HIP-HT + shot-peened specimens.

(14)

with Cr. The presence of other alloying elements, such as Nb, Ti, and Fe varied depending on the type of post-treatment (as already discussed). In the as-built and shot-peened specimens, a branched structure of alumina was observed. Al was mainly found at the grain boundaries, and not in the external oxide scale. Moreover, the formation of holes at the oxide/bulk interface of the HIP-HT specimen increased the risk of the inward diffusion of oxygen. The integrity of the oxide scale with the underlying material was better in the HIP-HT + shot-peened specimen compared to the other three specimens. Based on the EDS analysis in the specimens exposed at 800 °C, the level of Ti in the oxide scales was significantly higher in the post-treated specimens than in the as-built specimens. This is mainly because of the outward diffusion of Ti4+_to

the external surface of the oxide, leading to the formation of the coarser TiO2particles. In the HIP-HT + shot-peened specimen, the Cr content

was the highest, and the Fe content the lowest, conﬁrming that the oxide scale formed on this specimen was more protective than the oxides formed on the other specimens.

When exposed at 800 °C, theδ phase was found in the specimens below the oxide scales in the as-built and shot-peened specimens; see Fig. 17(a1–b1). The exposure temperature of 800 °C is much below the dissolution temperature of the δ phase, which is between 1005 and 1015 °C for the Nb contents of 5.1 and 5.4 wt%, respectively [64]. The Cr depletion beneath the external oxide, due to the formation of the Cr-rich scale, tends to reduce the Nb activity; therefore, the diffusion of Nb towards the scale/alloy interface is reduced and the formation of theδ phase is motivated. Theδ-phase can contribute to the control of grain growth during the high temperature exposures. Upon an enrichment of theδ phase at the metal/oxide interface, the inward diffusion of oxygen is suppressed leading to slower oxidation [62]. However, the formation of theδ phase is not always beneficial, as the diffusion coefficient of Cr is lower in theδ phase-rich region than in the matrix, which reduces the growth rate of the Cr-rich oxide scale [5]. Moreover, theδ phase con-sumes a high level of Nb that could have been used in the main

strengthening phases, such asγ″.

The formation of Laves phases (irregular-shaped white spots) can be identiﬁed in the shot-peened specimen and the HIP-HT + shot-peened specimen. Such phases were not observed in the specimens exposed at 650 °C. This conﬁrms that Alloy 718 cannot be used above 650 °C, even for a short period of exposure, such as the duration (168 h) examined in this study.

A kinetics-based study of the formation of the oxide scale on Alloy 718 exposed to ambient air at various temperatures was presented elsewhere [65]. It was confirmed that at the early stage of exposure, nucleation of a Cr-rich oxide in the form of separate islands can occur [66]. The coefficient of diffusion of Cr3+ _{is much larger than O}−2_;

therefore, Cr3+_di_{ﬀuses outward towards the external surface [}₅₂_{]. As a}

result, the new oxide nucleates and grows on the external surface of the previously formed oxide scale. The enhanced Cr diﬀusion can be at-tributed to the dislocations in the subsurface region introduced by shot peening, which results in the formation of a protective Cr-rich oxide. Once Cr2O3is formed, the inward diﬀusion of O2−within the oxide

layer becomes very sluggish, and the growth of the oxide scale depends primarily on solid-state diﬀusion. After a prolonged exposure at a high temperature, the level of dislocations can be dramatically decreased via recovery or recrystallization.

At 800 °C, the oxides grew gradually since the early stages and re-mained rather continuous. A few TiO2particles was also observed on

the surface of the oxide scale due to the higher diﬀusion rate of Ti+4

than Cr3+_[₆₁_{]. The spinels of NiCr}

2O4in the oxide scale increased,

depleting the bulk material in Cr and weakening the scale, which can spall upon any thermal stress.

Generation of compressive residual stresses, the formation ofﬁne grains, dislocations pile-up, removal of pores, and reduction in surface roughness are considered the primary outcomes of shot peening. As reported in the literature [67], stress relaxation can occur in the early stages of high-temperature exposures due to the quick annihilation of

a1

a2

a3

a4

b1

b2

_b3

b4

5 μm 5 μm 5 μm 5 μm 10 μm 10 μm 10 μm 10 μm 3 μm Branched Al-rich oxide 1 μm 3 μm 2 μm Internal oxide δ phase δ phase Branched Al-rich oxide Internal oxide Voids

1

2

3

4 wt%

O

Ni

Cr

Fe

Al

Nb

Ti

1

22.1

10.4

33.1

9.2

1.0

22.9

1.1

2

15.7

3.5

45.3

8.3

0.5

24.8

1.2

3

29.1

2.5

48.2

2.5

2.8

12.4

2.3

4

24.8

3.9

54.4

1.6

0.2

12.5

2.6

Fig. 17. Cross-sectional SEM micrographs (BSE mode) of the specimens exposed at 800 °C for 168 h in ambient air; 1) as-built, 2) shot-peened, 3) HT, and 4) HIP-HT + shot-peened specimens as well as the corresponding elemental point EDS analysis (allﬁgures in wt%).

(15)

defects through short-range diffusion or reordering of thermally mobile defects (vacancies and dislocations) that exist in the work-hardened regions [67]. The regions with higher stored energy show higher stress relaxation. The driving force for stress relaxation in the shot-peened specimen is higher than in the as-built specimen. Thus, the effect of the compressive residual stress upon shot peening will disappear after ex-posure to a high-temperature environment. The specific temperature and the exposure time for observing stress relaxation is the scope of future studies. In this study, exposures at 650 and 800 °C for 168 h appear to be sufficient for observing the effect of compressive residual stresses on the superior oxidation performance of shot-peened speci-mens. Such high-temperature exposures (650 and 800 °C for 168 h) are not enough to facilitate grain growth in the material. Thus, subsurface grain refinement may contribute to better oxidation performance of the shot-peened specimen and the HIP-HT + shot-peened specimen at 800 °C. Upon shot peening, the pores at the subsurface region were removed, and the density of the material in the subsurface region in-creased. Such a dense layer can protect the rest of the material against the penetration of corrosive species, such as oxygen. Removal of the pores enabled the reduction of the thickness of the oxide layer during oxidation exposure. Removal of excessive protrusions and roughness via shot peening has been reported in the literature and was also ob-served in this study.

4. Conclusions

Subsurface grains of EB-PBF-built Alloy 718 were reﬁned using shot peening to improve the surface texture and oxidation performance. The oxidation of the specimens (as-built, shot-peened, HT, and HIP-HT + shot peened) was examined in ambient air at 650 and 800 °C. The following conclusions can be drawn:

1. The as-built specimen contained defects, such as round-shaped pores, and lack-of-fusion defects within the hatch and contour re-gions. Shot peening coupled with the HIP-HT treatment resulted in the lowest level of porosity (0.16 area%) in the contour region among the four specimens.

2. The surface roughness was decreased by shot peening due to the smoothening eﬀect of the process. The Savalue was slightly higher

for the shot-peened specimen (~23μm) than for the HIP-HT + shot-peened specimen (~19μm). This was due to the high impact energy of the shots that resulted in the formation of more pronounced in-dentations on the as-built specimen, which had lower hardness (~406 HV0.1) compared to the HIP-HT specimen (~470 HV0.1).

3. According to the microhardness proﬁle, the subsurface region (at a depth of 30–60 μm) underwent strain hardening after shot peening. The SEM images of the cross-section of the shot-peened specimens veriﬁed that the depth of the work hardening region was approxi-mately 50μm.

4. The crystallite size, microstrain and dislocation density of the spe-cimens before and after shot peening were measured using the XRD patterns. The crystallite size decreased, whereas the microstrain/ dislocation density increased after shot peening. High compressive residual stress was generated due to plastic deformation upon shot peening.

5. EB-PBF-built Alloy 718 with post-treatments (either HIP-HT, shot peening or both) showed enhanced oxidation performance at 650 °C, due to the formation of a dense and protective Cr-rich oxide scale. The oxide scale formed on the surface of the HIP-HT + shot-peened specimen at 800 °C was composed of Cr, Nb, and Ti. At 800 °C, the internal oxidation also occurred mainly in the as-built specimen. The formation of holes at the oxide/bulk interface of the HIP-HT specimen increased the risk of the inward diﬀusion of oxygen. The integrity of the oxide scale with the underlying material was better in the HIP-HT + shot-peened specimen compared to the other three specimens. This was mainly because HIP-HT decreased

the level of defects in the contour region, whereas shot peening removed the pores at the subsurface region and increased the den-sity of the material in the subsurface region.

6. The coarse spinels of NiCr2O4at 800 °C depleted a high amount of

Cr from the matrix, which reduced the fraction of the protective Cr2O3in the oxide scale and increased the risk of spallation.

7. The outward diﬀusion of Cr3+_{to form a protective and dense}

Cr-rich layer on the surface was facilitated due to the grain reﬁnement and removal of the pores at the subsurface region, as well as smooth surface resulted from shot peening. The surface texture achieved after shot peening led to a signiﬁcant enhancement in the oxidation performance.

CRediT authorship contribution statement

Esmaeil Sadeghi: Conceptualization, Methodology, Data curation, Formal analysis, Writing - original draft, Visualization. Prabhat Pant: Methodology, Writing - review, and editing. Reza Jafari: Conceptualization, Methodology, Writing - review, and editing. Ru Peng: Conceptualization, Writing - review, and editing. Paria Karimi: Conceptualization, Methodology, Writing - review, and editing.

Declaration of competing interest

This research is sponsored by funding from the Knowledge Foundation for the SupREme project (Dnr 20180203) and Åforsk for the SCC-SUMAN project (18–296). We have disclosed those interests fully to Elsevier and have in place an approved plan for managing any po-tential conﬂicts arising from this arrangement.

The authors declare that there are no conﬂicts of interest. Acknowledgment

Financial supports of the Knowledge Foundation for the SupREme project (Dnr 20180203) and Åforsk for the SCC-SUMAN project (18-296) are highly acknowledged. The authors would like to thank Mr. Jonas Olsson, Dr. Mahdi Eynian, and Prof. Joel Andersson for their valuable help and advice in processing and characterization of the specimens. The authors would like to thank Mr. Olle Widman from Curtiss-wright Surface Technologies for shot peening, Mr. Johannes Gårdstam, and Mr. James Shipley from Quintus Technologies AB for HT-HIPing of the specimens. The authors are grateful to Venkataramanan Mohandass for the characterization of the specimens. Chalmers Materials Analysis Laboratory (CMAL) at the Chalmers University of Technology are appreciated for the help in the XRD/SEM analysis.

Data availability

All the data discussed are directly presented in the paper, and therefore they are automatically accessible.

References

[1] B. Graf, A. Gumenyuk, M. Rethmeier, Laser metal deposition as repair technology for stainless steel and titanium alloys, Phys. Procedia 39 (Jan. 2012) 376–381,

https://doi.org/10.1016/j.phpro.2012.10.051.

[2] S.A.M. Tofail, E.P. Koumoulos, A. Bandyopadhyay, S. Bose, L. O’Donoghue, C. Charitidis, Additive manufacturing: scientiﬁc and technological challenges, market uptake and opportunities, Mater. Today 21 (1) (Jan. 2018) 22–37,https:// doi.org/10.1016/j.mattod.2017.07.001.

[3] J.H. Andersson, Weldability of Precipitation Hardening Superalloys– Inﬂuence of Microstructure, (2011).

[4] A. Townsend, N. Senin, L. Blunt, R.K. Leach, J.S. Taylor, Surface texture metrology for metal additive manufacturing: a review, Precis. Eng. 46 (Oct. 2016) 34–47,

https://doi.org/10.1016/j.precisioneng.2016.06.001.

[5] T. Sanviemvongsak, D. Monceau, B. Macquaire, High temperature oxidation of IN 718 manufactured by laser beam melting and electron beam melting: eﬀect of surface topography, Corros. Sci. 141 (Aug. 2018) 127–145,https://doi.org/10.