Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

[Link]

ORIGINAL ARTICLE

Establishing the Qualitative Relationship Between Process Parameters:

Microstructure, Phases and Defects in SLM‑PBF Manufactured
and Heat Treated Inconel 718 Alloy
Pravin Kumar1 · V. S. K. Chakravadhhanula3 · Sushant K. Manwatkar3 · P. Chakravarthy2 · S. V. S. Narayana Murty3

Received: 4 June 2021 / Accepted: 1 August 2021 / Published online: 13 August 2021
© Indian National Academy of Engineering 2021

Abstract
In the present study, detailed metallurgical characterization was carried out on 3D-printed samples in as-printed and heat-
treated conditions to understand the effect of scan speed and laser power on the microstructures, phases, types of defects,
distribution of defects and oxides associated with defects. Extensive microstructural analysis was carried out using various
analytical techniques such as optical microscopy, scanning and transmission electron microscopy and x-ray diffraction to
establish a correlation between the microstructure phases and defects in the as-printed and heat-treated samples. A grain size
study on heat-treated samples revealed that the grain size of columnar grains has increased with an increase in volume energy
density (VED). The generation of different types of defects was correlated to the VED used to print samples. The optimum
process parameters based on the analysis of defects through optical microscopy are: v = 860–960 mm/s; P = 250–300 W;
VED = 65–70 J/mm3. Based on detailed microscopic analysis, it was noted that the heat treatment as per AMS5664 was
inadequate and needs to be optimized for 3D printed IN718 samples due to incomplete dissolution of segregated Nb and
Mo, carbides, oxides into the matrix during solution treatment.

Keywords IN718 · Additive manufacturing · Grain size · Defects · SEM–EDS · TEM

Introduction processing schedules to break the as-cast structure and

impart a homogeneous, isotropic material for engineering
Inconel 718 is a nickel-based superalloy used in several applications (Anbarasan et al. 2018). Tight process control
high-temperature applications, given its ability to retain during the manufacturing process from the casting stage
mechanical properties, resistance to oxidation and good to the final heat treatment stage is essential to control the
fatigue life (Shapiro et al. 2016; Lewandowski and Seifi microstructure that imparts repeatable mechanical properties
2016). Conventionally, this alloy is produced by vacuum in a production environment (Sohrabi et al. 2018).
melting and vacuum arc remelting/electroslag refining to Additive manufacturing is one of the promising tech-
control chemical composition, impurities, gas control to nology in the near future to print components to near net
obtain a desired macrostructure at the casting stage. This shape. 3D-printing of nickel-based superalloy Inconel
is followed by homogenisation and thermo-mechanical 718 (IN 718), by selective laser melting (SLM)- powder
bed fusion (PBF) process offers a significant advantage
* P. Chakravarthy of designing near-net complex components for aerospace
[Link]@[Link] applications, in view of its advantages of reduced pro-
1
cess time and cost benefits (Bosio et al. 2019). SLM-
Department of Chemistry, Indian Institute of Space Science PBF process could considerably reduce the possibility of
and Technology, Trivandrum 695 547, India
2
macrosegregation of heavy alloying elements in IN 718,
Department of Aerospace Engineering, Indian Institute which is common in cast products. In the SLM-PBF pro-
of Space Science and Technology, Trivandrum 695 547,
India cess, complex components are built layer-upon-layer by
3 selectively melting and fusing powder particles in each
Material Characterization Division, Vikram Sarabhai
Space Centre, Indian Space Research Organization, layer using a focused continuous laser power source (Benn
Trivandrum 695 022, India and Salva 2010). The process and part properties mainly

123
Vol.:(0123456789)
1084 Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

depend on four process parameters: (1) Laser scan speed and lack of fusion. Heat built-up in the component and
(v, mm/s) (2) Laser power (P, W) (3) Hatch distance (h, local variations of the thermal profile during printing gen-
mm) and (4) Powder bed thickness (t, mm). A derived erates thermal stresses in the component leading to the for-
parameter has been proposed combining these four vari- mation of cracks during printing and post-printing defects
ables to capture their inter-dependency on each other. One like delamination (Mukherjee et al. 2018). To adopt the 3D
such derived parameter is volume energy density (VED) printing technology and derive the benefits of this potential
(Scipioni Bertoli et al. 2017), given by Eq. 1. process to replace the existing wrought processed compo-
nents for engineering applications, it is essential to under-
P
VED = J∕mm3 (1) stand the process at the microscopic scale. Therefore, the
v∗h∗t
purpose of the present study was to establish a qualitative
These process parameters affect the melt pool shape and relation between two major process parameters; scan speed
size and thus have a greater impact on defect generation and and laser power with the microstructure, phases, defects
development of microstructure in as-printed components. and oxides in 3D printed and heat-treated IN 718 samples.
Very high cooling rates associated with the SLM–PBF pro- A detailed experimental plan was devised to 3D-print the
cess, typically of the order 1­ 05–107 K/s result in the forma- samples with varying process parameters and to character-
tion of a large amount of non-equilibrium Laves phases ize the same.
in as-printed IN 718 components (Cao et al. 2019). The
printed layers are also subjected to a thermal cycle due to
ongoing printing and thus affect the distribution of phases Experimentation
in the component along the build direction (Stevens et al.
2017). The direct interaction of laser beam and powder bed Material and 3D‑Printing of Samples
results in the setting up of several forces in the melt pool, as
reported for laser welding (Zhang et al. 2018). Any imbal- IN 718 powder having the chemical composition
ance in these forces due to variation in process parameters (53.35 wt% Ni—19.15 wt% Cr—4.99 wt% Nb—3.19 wt%
results in the generation of several defects like open-pores Mo—1.10 wt% Ti—0.48 wt% Al—Fe balance) was used

Fig. 1  Sample image and geometry—(a) Cuboidal sample showing printing plane XY and build direction plane XZ; (b) Cuboidal sample
dimension (c) Sample nomenclature (d) 3D printed IN 718 samples

123
Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097 1085

to print the cuboidal samples as shown in Fig. 1. The pow- a meander scanning pattern was applied on each layer get-
der chemistry was well within the standard specification ting printed. The scanning direction between adjacent lay-
for additive manufacturing of Ni-alloy (UNS N07718) des- ers was rotated by 67°, which helps in achieving almost
ignated by the ASTM F3055—14a (“ASTM F3055—14a uniform heat distribution over the printing plane.
Standard Specification for Additive Manufacturing Nickel
Alloy (UNS N07718) with Powder Bed Fusion” n.d.). The
Argon gas atomized IN 718 powder was supplied by EOS- Characterization and Heat Treatment
GmbH (Krailling, Germany) having an average particle size
of 30 μm. 95% of powder particle were having the size in the For optical microscopy (OM) and scanning electron micros-
range of 20 μm–84 μm. copy (SEM), samples were metallographically polished on
The geometry and dimensions of the cuboidal sam- two surfaces: the XY Plane and the XZ plane. The first
ples printed for experiments are shown in Fig. 1a, b. The level of polishing was carried out by grinding with 120,
printing plane (XY) represents the powder bed surface 220, 320, 400, 600, 800 grit size SiC emery paper, followed
on which a high energy laser beam is focused to selec- by fine polishing using alumina and diamond paste of 0.5
tively melt and fuse the powder. Whereas, the build direc- to 1-micron size. Waterless Kalling’s Reagent No. 2 (5 g
tion plane (XZ) represents the plane formed by powder ­CuCl2, 100 ml of HCL and 100 ml of ethyl alcohol) was
layer printing in Z-direction (build direction). EOSINT used as an etchant to obtain the microstructure. Olympus
M280_400W Direct Metal Laser Sintering (DMLS) GX71 metallurgical microscope was used for OM analysis to
machine was used to print samples from the base alloy capture the details of the defects and microstructures. Image
powders. Hatch distance (h = 110 μm ) and powder bed analyser and material solution software Olympus Xtreme
thickness (t = 40 μm ) were kept constant; only scan speed (Micro cam) and ImageJ were used to measure the grain
and laser power were varied to print twelve samples. size and study the defects. The grain size was measured
All the twelve as-printed cuboidal samples are shown according to the Planimetric Method ASTM E 112–12 from
in Fig. 1b,and the corresponding process parameters are the micrographs taken at 100X for printed + stress relieved
given in Table 1. The top surface of each sample was (SR) + Solution treated(ST) + Double aged (DA) samples.
marked with its nomenclature after printing to maintain Equation 2 shows the relation between grain size and grain
its traceability. In the sample nomenclature (Fig. 1c) capi- size number (GSN)
tal alphabets, A, B, C and D represent the laser power,
1
and numerals 1, 2, and 3 represent the scan speed. Before GSN(ASTM) ∝ (2)
GrainSize
printing, the build chamber was degassed to maintain
oxygen below 0.1% and during printing, Argon gas was Carl Zeiss EVO 50 Scanning electron microscope (SEM)
purged into the chamber to maintain the pressure within equipped with an energy dispersive X-ray spectrometer
the chamber to protect the printed surface from oxidation. (INCA Penta FET-EDS) was used to capture high-resolution
To reduce the thermal stresses during the building process, BSE (Backscattered electron) images of phases, oxides and
defects present in printed + SR + ST + DA samples. Chemi-
cal analysis and elemental mapping of various micro con-
Table 1  Process parameters, volume energy density (VED) and sam- stituents and phases were carried out in the EDS (Energy
ple nomenclature for SLM-PBF printed IN 718 Dispersive X-ray spectroscopy) mode. A PANalyticalX’Pert
Sample Scan speed Power (W) VED (J/mm3) PRO XRD machine was used for phase determination. Match
(mm/s) version3, open-source software that uses COD (Crystallo-
graphic open database) as reference was used for XRD data
A1 860 200 52.85
analysis. FEI-TECNAI transmission electron microscope
A2 960 200 47.35
(TEM) was used for High-angle annular dark-field scanning
A3 1060 200 42.88
transmission electron microscopy (HAADF-STEM) analy-
B1 860 250 66.07
sis of 3D printed and heat-treated (SR + ST + DA) samples.
B2 960 250 59.19
HAADF–STEM mapping was done for qualitative analysis
B3 1060 250 53.60
of precipitates, dispersed particles, and carbides present in
C1 860 300 79.28
the samples.
C2 960 300 71.02
The 3D-printed samples were stress relieved by subject-
C3 1060 300 64.32
ing them to 1065 °C for 90 min under argon atmosphere,
D1 860 350 92.49
followed by cooling to room temperature in a muffle furnace.
D2 960 350 82.86
D3 1060 350 75.04

123
1086 Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

Fig. 2  OM, SEM-BSE and SEM–EDS analysis of a sample in as- graph of the XZ plane; (d) BSE-SEM micrograph of the XZ plane
printed condition—(a) OM micrograph of the XZ plane; (b) OM showing solidification morphology inside grains; (e) SEM–EDS
micrograph of the XY plane; (c) Higher magnification OM micro- revealing Laves phase; (f) SEM–EDS revealing oxides

Further, the heat treatment was carried out as per AMS cooling to 650 °C in 2 h, holding at 650 °C for 8 h, and then
5664. According to this, samples were solution treated at followed by air cooling (“AMS5664F: Nickel Alloy, SAE
1065 °C for 1.5 h and then air-cooled. This was followed International” n.d.).
by double ageing by holding at 760 °C for 10 h, furnace

123
Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097 1087

Results and Discussion interdendritic region (Tucho et al. 2017). This suggests
that the solution treatment as per standard ASM-5664 was
Microstructures and Phases not sufficient to completely homogenize the as-printed
IN 718 samples prior to the commencement of the ageing
The microstructure of the as-printed sample is shown in process. The incomplete homogenization during solution
Fig. 2a–d. The image shows arc-shaped melt pool bounda- treatment (ST) has further led to the formation of bulky
ries (MPB) and interlayer boundaries (ILB) on the XZ ­Ni3Nb at the grain boundaries and within the matrix, as
planes, and laser scanning tracks (LST) on the XY plane. shown in Fig. 4a, b. The needle-shaped and plate-like
Remelting of solidified layers and epitaxial growth of ­Ni3Nb observed in the vicinity of grain boundaries cor-
grains up to 1–2 layers across the melt pool boundaries respond to the Orthorhombic 𝛿 phase (Kuo et al. 2017).
were observed on the XZ plane, as shown in Fig. 2a, c. Figure 4c shows (Ni, Fe, Cr) 2(Nb, Mo, Ti) type laves
The re-melting of the previously solidified layer results in phase at the grain boundary. The volume fraction of Laves
the formation of almost columnar grains oriented along phases was considerably reduced for heat-treated samples
the building direction by epitaxial grain growth (Fig. 2c) (SR + ST + DA) as compared to the as-printed samples.
across the layers. This kind of solidified structure is attrib- This is because, most of the contributing elements for
uted to the directional dissipation of heat towards the base- Laves phase formation like Ni, Ti, Mo and Nb were uti-
plate since the majority of heat is dissipated by thermal lized in forming strengthening phases like 𝛾ε and 𝛾 ′ on
conduction through the substrate or solidified material heat treatment. The presence of oxides was observed at
during SLM. SEM image in Fig. 2d shows the detailed the grain boundaries and was also distributed within the
morphology of each grain on the XZ plane. The solidifica- matrix. Figure 4a, b shows presence of oxides of type
tion morphology of grains was observed to be cellular and Al(O) and CrMo(O) along with traces of elements like
columnar dendritic type. The cellular structures appeared Ni, Fe, Ti, and Nb. These could be the oxides from the
to be equiaxed subgrains whereas the dendritic structures as-printed samples before heat treatment (Tucho et al.
had a large length to width ratio. Segregated Nb and Mo 2017; Ariaseta et al. 2020). The driving force and incu-
were present in the inter-cellular region and inter-dendritic bation time during solution treatment (ST) would not have
region leading to the formation of several phases, as shown been enough to completely dissolve these oxides into the
in Fig. 2c. As a result, Laves phases, delta(𝛿 ) phase and matrix. However, the concentration of oxides and their
metallic carbides were formed in the interdendritic region formation mechanism need to be analysed for the double
and at the grain boundaries (Fig. 2c, e). As observed in ageing heat-treatment process. In Fig. 5, XRD analysis of
SEM–EDS analysis (Fig. 2f) fine as well as coarse oxides heat-treated (SR + ST + DA) samples has been presented
of Cr and Al along with the traces of other elements like along with their respective volume energy densities. The
Fe, Ni, Nb, Mo and Ti were also found dispersed in the analysis revealed the presence of strengthing phases like
matrix and at the grain boundaries. These oxides could 𝛾ε and 𝛾 ′ , which would be nano-sized and distributed
have formed due to the presence of oxides from oxidised throughout the matrix. IN 718 is precipitation hardened
powder particles and absorption of Oxygen by the melt alloy, and the coherent body-centred tetragonal (BCT)
pool from the build chamber. ­Ni3Nb i.e., 𝛾ε is the major strengthening phase followed
Figure 3shows a collage of the microstructures by coherent cubic ­Ni3(Al, Ti) i.e., 𝛾 ′ . However, the forma-
obtained from both the XZ plane and XY plane respec- tion of bulky carbides and 𝛿 phase due to incomplete dis-
tively after the heat treatment (SR + ST + DA). Upon solution of segregated Nb into the matrix resulted in the
heat treatment (SR + ST + DA), MPB, ILP, and LST lost formation of lesser 𝛾ε in the matrix. The 𝛾ε phase in IN
their distinction and got homogenized completely. Slight 718 is metastable and thus converts to the thermodynami-
coarsening of grains was observed in heat-treated sam- cally stable δ phase (Du et al. 2007). Further, segregated
ples compared to as-printed ones. However, the columnar undissolved Nb at the grain boundaries could reduce the
nature of grains was preserved after heat treatment. Fig- driving force for δ phase formation, there by promoting
ure 4a–c shows the STEM-HAADF analysis of a heat- the easy formation of this deleterious phase at a much
treated sample. Metal carbides of MC-types can be seen early stage of ageing treatment (Liu et al. 2018). Table 2
in Fig. 4a–c. Ti and Nb have high tendencies to form presents the data of peak intensities corresponding to
stable carbides in the IN 718 alloy (Mostafa et al. 2017) XRD peaks shown in Fig. 5. The data suggests that there
and (Nb, Ti)C carbides were found at the grain bounda- is a variation in the intensities of the peaks among dif-
ries and within the matrix. Formation of the MC car- ferent samples with different VED. As the peak intensity
bides in the matrix would have resulted from the undis- in the XRD spectrum is related to the volume fraction of
solved Laves phases, carbides and segregated Nb in the the phases, qualitatively it may be deduced that there is a

123
1088 Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

Fig. 3  OM micrograph map for printed + heat-treated (SR + ST + DA) samples—(a) The XZ plane; (b) The XY plane

possibility of variation in the volume fraction of strength- laser power, powder bed thickness (PBT), and hatch dis-
ening phases from one sample to another, even when the tance etc., and the as-printed microstructures have a con-
samples are subjected to the same heat treatment. This siderable impact on the way the as-printed components
clearly, indicates that process parameters like scan speed, behave during heat treatment.

123
Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097 1089

Fig. 4  STEM –HAADF images with the corresponding elemental mapping from a 3D-printed + heat-treated (SR + ST + DA) sample

Table 2  XRD spectrum data for 3D-printed + heat-treated

(SR + ST + DA) samples
VED (J/mm3) Intensity at Intensity at Intensity at
2θ = 43.7 2θ = 50.7 2θ = 74.7

42.88 5305.58 1362.02 839.58

47.35 5611.39 1491.02 870.87
52.85 6055.26 1580.63 918.32
53.60 6237.22 1585.14 928.43
59.19 6374.71 1611.21 952.91
66.07 5612.30 1633.02 898.20
75.04 4978.82 1533.22 842.58
82.86 4874.45 1381.07 791.27
92.49 4584.72 1324.23 763.27

Fig. 5  XRD spectrum intensity plot for samples printed at different

VEDs
Grains and Grain Size

As shown in Fig. 3 and discussed in the previous section, the restricted mobility of the grain boundaries by the pinning
the columnar nature of the grains was preserved even after effect of carbides and oxides present before heat treatment.
heat treatment. Further, there was slight coarsening of grains Partially recrystallized grains were observed on both the XY
compared to that in the as-printed condition. This is due to and XZ planes of each sample, as shown in Fig. 3. This

123
1090 Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

Fig. 6  Grain size variation on the XY plane with respect to process parameters—(a) Grain size number (GSN) vs laser power; (b) GSN vs scan
speed; (c) GSN, scan speed and laser power contour plot; (c) GSN vs VED

Fig. 7  Grain size variation on the XZ plane with respect to process parameters—(a) Grain size number (GSN) vs laser power; (b) GSN vs scan
speed; (c) GSN, scan speed and laser power contour plot; (c) GSN vs VED

123
Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097 1091

could be attributed to the inability of the thermally gener- scan speed (860 mm/s–990 mm/s). Figures 6d and 7d show
ated residual stress to provide the necessary driving force for the variation of grain size concerning VED for the XY plane
the complete recrystallization of the grains (Liu et al. 2018; and XZ plane respectively. It was observed that scan speed
Chlebus et al. 2015). Further, the growth of these grains dur- and laser power alone can not explain variation in grain
ing successive heat treatment could be restricted by the pin- size precisely and conclusively. So, Volume energy density
ning effect of carbide and oxides. Figure 3 clearly shows the (VED) which takes care of both these parameters along with
variation in size of the columnar grains and also the volume other variables like hatch distance and powder bed thickness
fraction of recrystallized grains from one sample to another. can give a much clear picture. Predominantly an increas-
Figs. 6a–d and 7a–d captures the variation in the size of ing trend was observed in grain size (decreasing GSN) with
columnar grains on the XY plane and XZ plane respectively, respect to VED in both XY and XZ planes. This could be
for each of the heat-treated samples. This variation could explained as when VED was low, less amount of material
be attributed to two factors: (1) Grain size variation in the would be melted, thus resulting in a smaller melt pool. So
as-printed samples due to changing process parameters. (2) the solidification in the melt pool would be fast along with
Variation in the degree of pinning effect on grain mobility high undercooling. This would make the nucleation rate (I)
due to variation in volume fraction of grain boundary pin- high, providing less time for the nucleus and grains to grow.
ning phases from one sample to other. Both these factors are However, at a higher power density, melt pool size and aver-
affected by the process parameters used to print the samples age melt pool temperature would be high, providing more
and the effect is reflected on the grain size of heat-treated time for grains to grow. Besides, high VED ensures remelt-
(SR + ST + DA) samples. Figure 6a shows the variation of ing of a greater number of solidified layers and thus helping
grain size with respect to laser power on the XY plane and grains to grow epitaxially in the building direction.
Fig. 7a shows the same variation on the XZ plane. In the
XY plane, the grain size increased (GSN decreased) i.e., the
thickness of columnar grains increased with an increase in Defects and Oxides
laser power at a given scan speed. However, the same trend
on the XZ plane was not well established. This variation on Several types of defects and associated oxides were
the XY plane could be attributed to less energy input into observed in the printed and heat-treated (SR + ST + DA)
the powder bed resulting in the smaller melt pool, faster samples as shown in Figs. 8, 9 and 10. Morphology of
solidification, and a high degree of undercooling result- major defects was found to be the same on both XY and
ing in large nucleation of grains. Figs. 6b and 7b show the XZ planes. The oxides could have formed during heat
variation of grain size with respect to scan speed on the XY treatment in a muffle furnace. Defects and Oxides present
and XZ planes respectively. With an increase in scan speed, in the samples are discussed below in detail.
grain size decreased (GSN increased) on the XY plane.
Again on the XZ plane, no such trend was observed. This
is because, on the XY plane, a high scan speed indicates Lack of fusion (LOF)
lower energy input in the powder bed. This would result in
a very small weld pool, making the solidification process Figure 8c shows a lack of fusion defect. Lack of fusion
faster and undercooling larger for prevailing conditions. defect was observed both on the XY and XZ planes. It
High undercooling implies a high nucleation rate result- appears typically due to insufficient melting or the absence
ing in finer grains. The XY plane had mostly shown a clear of melting of powder particles in the powder bed. There
trend but in the case of the XZ plane, the trend was not well could be a horizontal lack of fusion (HLOF) or a vertical
established. This can probably be attributed to changing melt lack of fusion (VLOF). HLOF defects are formed due to
pool dynamics due to temperature building in the system as low-energy density and high powder bed thickness result-
printing progresses, thus affecting the interlayer epitaxial ing in poor re-melting and bonding between layers (Gor-
grain growth in the building direction (Farshidianfar et al. don et al. 2020). VLOF is formed due to a larger hatch dis-
2016; Manvatkar et al. 2015). From the colour map of the tance and low energy density and thus improper re-melting
XY plane, as shown in Fig. 6c, it can be established that of the adjacent laser scanning tracks (LST) (Priya Parida
thicker columnar grains would occur at a combination of and Senthilkumar 2020). Low powder bed density and an
high power and low scan speed (dark blue region), and thin- attenuated laser are also responsible for HLOF and VLOF
ner columnar grains would appear when processing at low defects. In this study, LOF defects were predominantly
power and high scan speed (red region). A similar colour observed in samples printed below VED of 65 J/mm 3.
map for the XZ plane in Fig. 7c shows that the size of the Figure 8c shows the morphology of a LOF defect with
grains was larger somewhere near the intermediate power to partially melted and un-melted powder particles inside it.
the higher side (290 W–350 W) and at lower to intermediate

123
1092 Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

Fig. 8  SEM micrographs collage featuring types and morphologies of defects in 3D-printed + heat-treated (SR + ST + DA) samples

Fig. 9  SEM–EDS elemental map showing the presence of oxides near an inter layer in a 3D-printed + heat-treated (SR + ST + DA) sample

123
Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097 1093

Fig. 10  SEM–EDS elemental map showing the presence of oxides near a cluster of liquation crack in a 3D-printed + heat-treated (SR + ST + DA)
sample

Cracks with uneven rupture surface confirming tensile failure. Inter-

layer cracks can lead to delamination of the layers when
Figure 8e, h reveals the presence of cracks in the samples. very high tensile residual stresses are present between two
Cracks can be of two types: (a) Interlayer cracks (Fig. 8e) adjacent layers. Intra layer cracks are formed due to shrink-
formed in between previously solidified and the layer get- age and thermal residual stresses in the melt pool at high
ting printed, and (b) Intralayer cracks (Fig. 8h) formed in a power densities. The mechanism of formation is similar to
layer during solidification of the melt pool. An inter-layer the solidification cracking during laser welding. Intra-layer
crack is formed due to the generation of thermal stresses cracks are characterized by shorter lengths as they typically
because of the high thermal gradients set up between the form within the dimensions of the melt pool. Both these
previously printed layer and the layer getting printed (Chen, types of cracks were found to be present in samples printed
Lu, et al. 2016; Qiu et al. 2019). The layer getting printed is at VED greater than 70 J/mm3.
at a higher temperature compared to the previously printed
and solidified layer. Thus the layer getting printed shrinks Gas porosity
faster and tensile thermal stresses build up near the bound-
ary of two layers. These tensile stresses initiate local failure Figure 8a, c, h shows the morphology of gas porosity. Sec-
of material once its magnitude crosses the ultimate tensile tions of these pores could be seen on both the XZ plane and
strength of the material at a higher temperature, resulting in XY plane. As observed under SEM, gas pores were spheri-
a crack. This type of defect was found to appear in samples cal in shape and small in size (1 μm–15 μm) compared to
printed at high power density (D1, D2, D3). It has been seen other defects. There would have been various reasons for the
that a sample with high laser power and high scan speed formation of gas pores. If the density of the powder bed is
(D3) has shown more such defects. Fig. 8e shows an inter- low, gasses present in between the powder particles may get
layer crack on the XZ plane of sample D3. The crack was dissolved in the molten pool. Due to the high cooling rate,
found propagating across an epitaxially grown grain at an the molten pool solidifies very fast, and thus the dissolved
interlayer boundary. The interlayer crack was characterized gases get entrapped resulting in porosity (Choo et al. 2019).

123
1094 Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

Porosity and gas entrapped in powder particles during pow- result, a high surface tension gradient is set up on the sur-
der production can contribute to porosities in 3D-printed face of the melt pool and thus the Plateau-Rayleigh insta-
samples. Also, at very high temperatures solubility of gasses bility on the pool surface increases (Tang et al. 2020). This
in the melt pool increases. Thus, purging gas like argon in increased instability tends to break the molten pool into sev-
the chamber can easily dissolve and remain entrapped after eral smaller pools having low flowability and wettability due
solidification resulting in porosity (Priya Parida and Senthil- to fast solidification. This makes the molten pool solidify as
kumar 2020). However, gas porosities can be observed at all balls on the surface (Gasper et al. 2018; Gu and Shen 2009).
VEDs, but their number fraction is high at very high VED. (3) The affinity of molten metal to absorb oxygen at high-
temperature increases. The oxygen can act as a surface act-
Open Porosity ing agent and reverse the Marangoni convection in the pool
resulting in ball formation (Gu and Shen 2009). Also, oxides
SEM micrographs in Fig. 8b,c show open pores. Section on the surface of solidified layers decrease the wettability
of these pores could be seen on both the XY plane and XZ and may result in the formation of balls in the melt pool.
plane. In general, they were found to be larger than gas
porosities. As observed, open-pores do not have any regu- Liquation Cracking
lar shape, because their formation mechanism differs from
spherical gas porosities. At a high power density, vapour Liquation cracks were found along the grain boundaries in
pressure and recoil pressure would be high (Khairallah et al. most of the samples as shown in Fig. 8d, g. In welding,
2016). This would have led to the formation of a large cavity liquation cracks form in the partially melted zone (PMZ)
in the melt pool, as the molten fluid is pushed towards the just close to the weld pool (Tharappel and Babu 2018; Chen,
periphery. However, during solidification, the molten fluid et al. 2016a, b). A similar phenomenon would have occurred
could not attain a desirable velocity to fill the gap due to the in the case of the PBF process, wherein the formation of
fast solidification process at very high cooling rates. Thus, melt pool has associated PMZ from the previously solidified
the time to fill the gap would be more than the solidifica- material and layer. Grain boundaries in as-printed IN 718
tion time. This unfilled region produced due to instability were decorated with segregated elements as well as laves
in the melt pool manifested as open porosity in the samples phase and carbides. Apart from this,the as-printed IN 718
(Xia et al. 2017). So, it could be said that as the energy den- specimens consist of microstructures consisting of a mixture
sity increases, pool instability due to an increase in vapour of high angle grain boundaries and low angle sub-bounda-
pressure and vapour recoil pressure increases, and thus ries; thus grain boundary energy per unit grain boundary
the number of open pores increase (Khairallah et al. 2016; area is very high. The combination results in a reduction in
Zhang et al. 2017). As observed, the presence of open-pores the melting point of grain boundary material (Sindo Kau
becomes more common and deleterious above VED of 70 J/ Edition, Second n.d.). This leads to a larger volume of melt-
mm3. Figure 8b, c also shows the morphology of an open ing or liquation at the grain boundaries in PMZ (Chen, Lu,
pore, where the flowing metal trying to fill the pore had et al. 2016; Sindo Kau Edition,Second n.d.). Further, the
solidified before filling it. shrinkage associated with the solidifying material in the
melt pool setsup tensile stress in the PMZ resulting in the
Balling rupture of molten material at the grain boundaries resulting
in liquation cracks. These cracks could appear at any VED
Figure 8f shows a balling defect and associated crack and but are more dominant at higher VED due to the formation
oxide. When a continuous melt pool breaks into smaller melt of larger PMZ and more liquation at the grain boundaries.
pools and solidifies on the already solidified layer without
wetting it, this results in the formation of a nearly spherical Oxides
ball on the surface (Li et al. 2012). There are three main
reasons for the formation of balling defects: (1) At very high Figure 9 shows the EDS elemental map of oxides formed
power density, cavity formation can take place in the melt during heat treatment near the interlayer crack and Fig. 10
pool. Due to high vapour pressure and recoil pressure,the shows similar oxides near a cluster of liquation cracks.
melt pool would be highly unstable and results in the splat- Oxides near the defect were only found in the XZ planes of
ter of the molten fluid, when the vapour pressure inside the the samples having large interlayer cracks, liquation cracks
cavity crosses the critical limit. This splattering drop of the and in very few cases near the cluster of open-pore and lack
molten alloy has very low wettability and flowability due of fusion. No such oxides were seen near any defects in
to the high solidification rate and thus form balls (Gu and the XY plane. Chromium oxide ­(Cr2O3) is the major oxide
Shen 2009). (2) At a high energy density,the temperature formed just near the edge of the crack (Fig. 9). It can be seen
gradient on the surface of the melt pool is very high. As a from the EDS map that a mixture layer of (Nb, Mo) O and

123
Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097 1095

Fig. 11  Porosity Micrograph map—(a) The XY plane; (b)The XZ plane

some intermetallic compounds of Ni and Nb possibly N ­ i3Nb The elemental map also showed the presence of Titanium
has formed just after ­Cr2O3. This thin layer of intermetallic oxide ­(TiO2) and Manganese oxide. The formation of these
might have protected further penetration of oxygen into the oxides could be attributed to the lattice diffusion of alloying
matrix, thus preventing further growth of the C ­ r2O3 layer. elements toward the free surface of the crack and also the

123
1096 Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097

diffusion of ­O2 into the material at the surface of the crack The phases such as (Nb, Ti)C types carbides, (Al, Cr)
(Sanviemvongsak et al.2018; Zhang et al. 2013). The cluster O type oxides and 𝛿 phase do not dissolve completely in
of liquation cracks also showed a similar type of oxidation the matrix during solution treatment (ST) and thus act
trend as shown in Fig. 10. However, not all of these defects as grain boundary pinners during ST and double age-
show such a trend. These oxide films are formed during heat ing (DA) heat treatment processes restricting the grain
treatment and are detrimental to the mechanical properties. growth.
The inner surface of the oxides i.e., in contact with metal It was observed that there was an incomplete dissolution
matrix, serve as the site for the precipitation of other oxides, of segregated Nb and Mo, carbides, and oxides into the
bulky carbides and large intermetallic phases. Due to this, matrix during solution treatment when AMS 5664 was
there is sufficient diffusion and depletion of alloying ele- adopted for heat treatment. Thus, to get desired micro-
ments from the nearby regions, thus weakening the matrix. structural features and properties in SLM–PBF processed
It was found that the dominance of defects of differ- IN 718 components, there is a need to optimize the heat
ent types is related to volume energy density. Out of all treatment process for the components printed in the opti-
defects discussed above LOF defect, open pores and gas mized range of printing process parameters.
porosity are most common. Based on OM analysis of Defects like lack of fusion (LOF) are predominant at
these defects, a defect microstructure map is presented in lower VED and other defects like open-pores, balling
Fig. 11. This montage of micrographs shows the qualita- defects, inter/intra-layer cracks, liquation cracks are dom-
tive variation of defects concerning scan speed and laser inant at high VEDs. Analysis of defects through optical
power. From the map, it could be said that there is minimal microscopy leads to predict the range of scan speed and
defect generation on the XZ plane (Fig. 11a) for laser power laser power for defect-free printing as: v = 860 mm/s–
in the range 250 W–300 W and scan speed in the range 960 mm/s; P = 250 W–300 W; VED = 65 J/mm3–70 J/
860 mm/s–960 mm/s. Similarly, for the XY plane (Fig. 11b), mm3.
the range is the same. LOF defect is dominant below VED of
65 J/mm3 and open pores are dominant above VED of 70 J/
mm3. Thus IN718 samples printed in between the discussed
range of scan speed, laser power and VED will give a sample References
with minimum defects within the parameters studied. Thus,
the heat treatment process discussed above needs to optimize Anbarasan N, Bikash Kumar Gupta, Prakash S, Muthukumar P, Oyyar-
for the samples printed in this range. avelu R, John Felix Kumar R, Jerome S (2018) Effect of heat treat-
ment on the microstructure and mechanical properties of inconel
718. Materials Today: Proceedings 5(2):7716–7724. [Link]
org/​10.​1016/j.​matpr.​2017.​11.​448
Ariaseta A, Kobayashi S, Takeyama M, Wang Y, Imano S (2020)
Conclusions Characterization of recrystallization and second-phase par-
ticles in solution-treated additively manufactured alloy 718.
Metall Mater Trans A 51(2):973–981. [Link]
Detailed metallurgical characterization was carried out on s11661-​019-​05560-y
3D-printed and 3D-printed + solution treated (ST) + double ASTM F3055—14a Standard Specification for additive manufacturing
aged (DA) IN 718 cuboid samples to study and understand nickel alloy (UNS N07718) with Powder Bed Fusion. n.d. [Link]
the effect of scan speed and laser power on the microstruc- www.​astm.​org/​Stand​ards/​F3055.​htm. Accessed 3 June 2021
Benn RC, Salva RP (2010) Additively Manufactured INCONEL®
tures, phases, types of defects, distribution of defects and Alloy 718. In: 7th international symposium on superalloy 718
oxides associated with defects. Rigorous data analysis was and derivatives, vol 1, pp 455–469. [Link] oi.o​ rg/1​ 0.1​ 002/9​ 7811​
done to establish a relation between the microstructure 18495​223.​ch35
phases and defects in as-printed and printed + heat-treated Bertoli S, Umberto AJW, Manyalibo JM, Delplanque JPR, Schoe-
nung JM (2017) On the limitations of volumetric energy density
samples to correlate properties in printed + heat-treated com- as a design parameter for selective laser melting. Mater Design
ponents with the process parameters. Following conclusions 11:331–340. [Link]
can be drawn from this study. Bosio F, Aversa A, Lorusso M, Marola S, Gianoglio D, Battezzati L,
Fino P, Manfredi D, Lombardi M (2019) A time-saving and cost-
effective method to process alloys by laser powder bed fusion.
The sub-microscopic phases and microstructures in as- Mater Des 181:107949. [Link]
printed IN7 18 affect the heat treatment process. It was 107949
found that the columnar nature of the grains was pre- Cao Yu, Bai P, Liu F, Hou X (2019) Investigation on the precipitates
served even after heat treatment. A grain size study on of IN718 alloy fabricated by selective laser melting. Metals
9(10):1–10. [Link]
heat-treated samples revealed that the size of columnar Chen Y, Fenggui Lu, Zhang Ke, Nie P, Hosseini SRE, Feng K, Li
grains increases with an increase in VED. Z (2016a) Dendritic microstructure and hot cracking of laser

123
Transactions of the Indian National Academy of Engineering (2021) 6:1083–1097 1097

additive manufactured inconel 718 under improved base cooling. homogenization and HIP treatments. Metals 7(6):1–23. [Link]
J Alloy Compd 670:312–321. [Link] doi.​org/​10.​3390/​met70​60196
2016.​01.​250 Mukherjee T, Zuback JS, Zhang W, DebRoy T (2018) Residual stresses
Chen Y, Zhang Ke, Huang J, Reza S, Hosseini E, Li Z (2016b) Charac- and distortion in additively manufactured compositionally graded
terization of heat affected zone liquation cracking in laser additive and dissimilar joints. Comput Mater Sci 143:325–337. [Link] oi.​
manufacturing of inconel 718. JMADE 90:586–594. [Link] org/​10.​1016/j.​comma​tsci.​2017.​11.​026
org/​10.​1016/j.​matdes.​2015.​10.​155 Priya Parida R, Senthilkumar V (2020) Experimental studies of defect
Chlebus E, Gruber K, Kuźnicka B, Kurzac J, Kurzynowski T (2015) generation in selective laser melted inconel 718 alloy. Materials
Effect of heat treatment on the microstructure and mechanical Today: Proceedings, no. XXXX. [Link]
properties of inconel 718 processed by selective laser melting. 2020.​04.​698
Mater Sci Eng, A 639:647–655. [Link] rg/​10.​1016/j.​msea.​ Qiu C, Chen H, Liu Qi, Yue S, Wang H (2019) On the solidification
2015.​05.​035 behaviour and cracking origin of a nickel-based superalloy dur-
Choo H, Sham KL, Bohling J, Ngo A, Xiao X, Ren Y, Depond PJ, Mat- ing selective laser melting. Mater Charact 148(January):330–344.
thews MJ, Garlea E (2019) Effect of laser power on defect, texture, [Link]
and microstructure of a laser powder bed fusion processed 316l Sanviemvongsak T, Monceau D, Macquaire B (2018) High-temper-
stainless steel. Mater Des 164(February):107534. [Link] ature oxidation of in 718 manufactured by laser beam melting
10.​1016/j.​matdes.​2018.​12.​006 and electron beam melting: effect of surface topography. Corros
Du JH, Lu XD, Deng Q, Qu JL, Zhuang JY, Zhong ZY (2007) High- Sci 141(January):127–145. [Link]
temperature structure stability and mechanical properties ofnovel 07.​005
718 superalloy. Mater Sci Eng A 453:584–591. [Link] oi.o​ rg/1​ 0.​ Shapiro AA, Borgonia JP, Chen QN, Dillon RP, McEnerney B, Polit-
1016/j.​msea.​2006.​11.​039. Casillas R, Soloway L (2016) Additive manufacturing for aero-
Sindo kou, WELDING METALLURGY (2002) Wiley space flight applications. J Spacecraft Rockets. American Institute
Farshidianfar MH, Khajepour A, Gerlich AP (2016) Effect of real-time of Aeronautics and Astronautics Inc. [Link]
cooling rate on microstructure in laser additive manufacturing. A33544
J Mater Process Technol 231:468–478. [Link] Sohrabi MJ, Mirzadeh H, Rafiei M (2018) Solidification behavior and
jmatp​rotec.​2016.​01.​017 laves phase dissolution during homogenization heat treatment of
Gasper AND, Szost B, Wang X, Johns D, Sharma S, Clare AT, Ash- inconel 718 superalloy. Vacuum 154(April):235–243. [Link]
croft IA (2018) Spatter and oxide formation in laser powder bed org/​10.​1016/j.​vacuum.​2018.​05.​019
fusion of inconel 718. Addit Manuf 24:446–456. [Link] Stevens EL, Toman J, To AC, Chmielus M (2017) Variation of hard-
10.​1016/j.​addma.​2018.​09.​032 ness, microstructure, and laves phase distribution in direct laser
Gordon JV, Narra SP, Cunningham RW, Liu He, Chen H, Suter RM, deposited alloy 718 cuboids. Mater Des 119:188–198. [Link]
Beuth JL, Rollett AD (2020) Defect structure process maps for org/​10.​1016/j.​matdes.​2017.​01.​031
laser powder bed fusion additive manufacturing. Addit Manuf Tang C, Le KQ, Wong CH (2020) Physics of humping formation in
36:101552. [Link] laser powder bed fusion. Int J Heat Mass Transf 149. [Link]
Gu D, Shen Y (2009) Balling phenomena in direct laser sintering of org/​10.​1016/j.​ijhea​tmass​trans​fer.​2019.​119172
stainless steel powder: metallurgical mechanisms and control Tharappel JT, Babu J (2018) Welding processes for inconel 718- a brief
methods. Mater Des 30(8):2903–2910. [Link] review welding processes for inconel 718- a brief review. [Link]
matdes.​2009.​01.​013 doi.​org/​10.​1088/​1757-​899X/​330/1/​012082
Khairallah SA, Anderson AT, Rubenchik A, King WE (2016) Laser Tucho WM, Cuvillier P, Sjolyst-Kverneland A, Hansen V (2017)
powder-bed fusion additive manufacturing: physics of complex Microstructure and hardness studies of inconel 718 manufactured
melt flow and formation mechanisms of pores, spatter, and denu- by selective laser melting before and after solution heat treat-
dation zones. Acta Mater 108:36–45. [Link] ment. Mater Sci Eng, A 689(February):220–232. [Link]
actam​at.​2016.​02.​014 10.​1016/j.​msea.​2017.​02.​062
Kuo YL, Horikawa S, Kakehi K (2017) The effect of interdendritic δ Xia M, Dongdong Gu, Guanqun Yu, Dai D, Chen H, Shi Q (2017)
phase on the mechanical properties of alloy 718 built-up by addi- Porosity evolution and its thermodynamic mechanism of randomly
tive manufacturing. Mater Des 116:411–418. [Link] packed powder-bed during selective laser melting of inconel 718
1016/j.​matdes.​2016.​12.​026 alloy. Int J Mach Tools Manuf 116:96–106. [Link]
Lewandowski JJ, Seifi M (2016) Metal additive manufacturing: a 1016/j.​ijmac​htools.​2017.​01.​005
review of mechanical properties. Annu Rev Mater Res 46:151– Zhang Bi, Li Y, Bai Q (2017) Defect formation mechanisms in selec-
186. [Link] tive laser melting: a review. Chin J Mechl Eng (english Edition)
Li R, Liu J, Shi Y, Wang Li, Jiang W (2012) Balling behavior of stain- 30(3):515–527. [Link]
less steel and nickel powder during selective laser melting pro- Zhang YN, Cao X, Wanjara P, Medraj M (2013) Oxide films in laser
cess. Int J Adv Manuf Technol 59(9–12):1025–1035. [Link] additive manufactured inconel 718. Acta Mater 61(17):6562–
org/​10.​1007/​s00170-​011-​3566-1 6576. [Link]
Liu P, Si-yu Sun S-b X, Mei-qing Cao CH, and Jia-Ying H (2018) Zhang Y, Chen Q, Guillemot G, Gandin CA, Bellet M (2018) Numeri-
Effect of solid solution + double ageing on microstructure and cal modelling of fluid and solid thermomechanics in additive
properties in the layer by layer of the z-y interface of inconel 718 manufacturing by powder-bed fusion: continuum and level set
alloys fabricated by SLM. Mater Res 21(6). [Link] formulation applied to track- and part-scale simulations. Comptes
1590/​1980-​5373-​mr-​2018-​0395 Rendus—Mecanique 346(11):1055–1071. [Link]
Manvatkar V, De A, DebRoy T (2015) Spatial variation of melt pool 1016/j.​crme.​2018.​08.​008
geometry, peak temperature and solidification parameters during
laser-assisted additive manufacturing process. Materials Sci Tech- Publisher's Note Springer Nature remains neutral with regard to
nol (united Kingdom) 31(8):924–930. [Link] jurisdictional claims in published maps and institutional affiliations.
17432​84714Y.​00000​00701
Mostafa A, Rubio IP, Brailovski V, Jahazi M, Medraj M (2017) Struc-
ture, texture and phases in 3D printed IN718 alloy subjected to

123