Materials Today Communications 25 (2020) 101535

Contents lists available at ScienceDirect

Materials Today Communications

journal homepage: [Link]/locate/mtcomm

Microstructure, Tribological Properties and Corrosion Behaviour of T

Additively Manufactured 17-4PH Stainless Steel: Effects of Scanning Pattern,
Build Orientation, and Single vs. Double scan
H.R. Lashgaria,*, Y. Xuea, C. Onggowarsitoa, C. Kongb, S. Lia
a
School of Materials Science and Engineering, University of New South Wales, Sydney, NSW 2052, Australia
b
Mark Wainwright Analytical Centre (MWAC), University of New South Wales, Sydney, NSW 2052, Australia

A R T I C LE I N FO A B S T R A C T

Keywords: The aim of the present study is to investigate the effect of different process parameters, i.e. scanning pattern
Selective laser melting (hexagonal vs. concentric), build orientation (vertical vs. horizontal), and single scan vs. double scan on the
17-4PH stainless steel microstructure, wear resistance and corrosion properties of additively manufactured 17-4PH stainless steel by
Wear means of advanced material characterization techniques such as X-ray diffraction (XRD), electron back scatter
Corrosion
diffraction (EBSD), and scanning electron microscopy (SEM). The volume fraction of retained austenite varied
Manufacturing parameters
nanohardness
from 0.7% to 4.4% based upon the orientation and scanning strategy. In both scanning patterns, the maximum
variation of nanohardness (along the build-axis and perpendicular to the build axis) was observed in the ver-
tically printed samples using the single scan whereas the double-scan strategy (remelting) decreased the var-
iation in both scanning patterns via lowering the temperature gradient and cooling rate. Remelting enhanced the
wear resistance (by almost ≈50%) by decreasing the volume fraction of pores and balling defects up to ≈60-
70% in the hexagon and concentric scanning pattern. The build orientation (vertical vs. horizontal) had more
influence on the wear resistance in the concentric scanning pattern than that of hexagon pattern. Oxidative/
abrasive wear was found to be the dominant wear mechanism during the reciprocating action in both scanning
patterns; however, the oxide film was more prone to fracture in the samples fabricated by the concentric pattern
due to the presence of more micropores in the matrix especially in the areas where laser direction changed. In
the phosphate-buffered saline (PBS) solution, the corrosion rate of the samples manufactured using the hex-
agonal scanning pattern was generally less than those manufactured using the concentric pattern owing to the
lower porosity. The lower breakdown potential in the concentric scanning pattern indicates that the protective
passive layer is broken down easily making it susceptible to pitting corrosion due to the presence of porosities
between the laser tracks. The Inductively Coupled Plasma (ICP) analysis revealed that the sample fabricated
using single scan with hexagon scanning pattern and orientated vertically released the least amount of ions in the
PBS solution, showing the significance for understanding the effect of additive manufacturing parameters on
functional materials.

1. Introduction components suffer from anisotropic properties, porosities, residual

stress and delamination during manufacturing leading to restrictions on
Additive manufacturing (AM) as an advanced manufacturing pro- repeatability and reliability of the additively manufactured components
cess is a layer-by-layer deposition process where a 3D Computer-Aided [4–7]. To eliminate porosities and control the residual thermal stress
Design (CAD) file is utilised to print near net-shape components with generated during manufacturing, it is essential not only to optimise the
intricate geometries regarded as one of the core strengths of additive process parameters (such as laser power, scanning speed, layer thick-
manufacturing process [1–4]. Hence, AM is best suitable for low vo- ness, hatch spacing, orientation, and scanning strategy) for the product
lume and complex design where formative (such as casting, metal quality to fulfil the industry requirements but also to understand how
forming) and subtractive (such as machining) methods are unable to the manufacturing parameters influence the microstructure and me-
produce such prototypes. However, additively manufactured chanical properties. Toyserkani et al. [8] showed that the laser power

⁎
Corresponding author.
E-mail address: [Link]@[Link] (H.R. Lashgari).

[Link]
Received 31 March 2020; Received in revised form 29 July 2020; Accepted 30 July 2020
Available online 13 August 2020
2352-4928/ © 2020 Elsevier Ltd. All rights reserved.
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

had more effect on the weld pool geometry during manufacturing. In 3D Systems) was used in the present study with D90 ≤ 29.4 μm, where
other words, the increase of laser power caused the weld pool to be- D90 implies the size of 90 per cent of the particles in a powder batch.
come narrow and deep and grain detachment at high laser power was The morphology of the powders was spherical as shown in Fig. 2
proposed to be the best explanation for the nucleation of new grains in making it ideal for SLM process due to the good powder flowability. The
the weld pool during solidification [8]. Shamsaei et al. [9,10] showed chemical composition of the as-received powder is given in Table 3. The
that the orientation of printed layer in respect to the loading axis re- microstructure of the supplied powder is a combination of the mar-
sulted in anisotropic tensile and fatigue performance in the horizon- tensite, austenite and NbC-type carbides.
tally- and vertically-oriented 17-4PH stainless steel samples. This was Two different scanning patterns were employed in the present study
attributed to the orientation of weak interfacial layers (for example lack as follows:
of fusion or balling effect) in respect to the loading axis leading to more
stress concentration at weak points in the vertically-oriented samples as (I) “Hexagonal pattern” which consists of a circle with the radius of
compared to the horizontally-printed one providing easier path for the 25,000 μm circumscribing the hexagonal island with a 100 μm
initiation of cracks and coalescence of voids under static and/or dy- overlap between two hexagons. From layer to layer, in the Z di-
namic loadings [10]. rection, the scanning pattern alternates between 315° (odd layers)
Rashid et al. [11] showed that double scan strategy (each layer and 225° (even layers). The hexagon pattern is the default scanning
scanned twice) led to higher hardness and relative density as compared strategy for 3D Systems SLM printer,
to single scan strategy which was attributed to the transformation of (II) “Concentric pattern” which consists of scanning vectors progressing
retained austenite into martensite during the second scan. In addition, from the edge towards the center of the part.
scanning each layer twice results in elimination of gas porosities, voids
and balling particles in between layers improving the relative density at The schematic of the scanning strategies utilised in the current study
the expense of decreasing the manufacturing rate [11]. It was shown is shown in Fig. 3.
that the increase of interlayer time had no effect on the mechanical After the build was completely printed, samples were removed from
properties of selective laser melting (SLM) made of 17-4PH stainless the plate by means of an electro discharge machining (EDM). The as-
steel samples [10]. The explanation proposed was high cooling rate and printed samples were sectioned, ground and then mechanically po-
small weld pool size formed during the SLM process. McWilliams et al. lished to a mirror finish using a colloidal silica solution (0.04 μm). To
[12] showed that the change in length and direction of laser scan line observe the microstructure, Fry’s reagent (40 mL HCl, 5 g CuCl2, 30 mL
could result in the formation of different volume fraction of retained water, 25 mL ethanol) was used to etch the specimens (etching time
austenite in 17-4PH stainless steel. In addition, it was shown that 30 sec-1 min).
overall extent of voids was almost similar in all scanning patterns; The phase identification was carried out using X-ray diffraction
however, the concentration of some type of voids was dependant on the technique (Empyrean 2 PANalytical X-ray Diffraction System) with Co
scanning strategy used during manufacturing. Kα (λ = 1.788965 Ǻ) from 2θ = 40–130° with step size of 0.026 and
Therefore, understanding the process-structure-property relation- step time of 210 (s). In addition, High score plus software was used for
ship in SLM process is essential for engineering the microstructure and phase identification and peak analysis. Scanning electron microscopy
achieving the desirable mechanical properties. The role of SLM para- (field-emission scanning electron microscope, NanoSEM 450) equipped
meters on the tensile and fatigue properties have been extensively with an energy dispersive X-ray spectroscopy (EDS) was used for ma-
studied in the literature but less attention has been given to tribological terial characterisation. The electron back scatter diffraction (EBSD)
and corrosion properties being the subject of the present study. Hence, analysis was carried out using an Oxford Aztec 4.0HKL EBSD system
the intention of the current study is to investigate the effect of some with a Symmetry CMOS detector. This EBSD system is attached to a
important manufacturing parameters such as scanning strategy (hex- Thermo Fisher Helios G4 UXe Plasma Focused Ion Beam and Elstar
agonal vs. concentric), orientation (vertically-oriented vs. horizontally electron column with UC + technology. The 100 × 75μm2 mapping
orientated), and number of scan per layer (single scan vs. double scan) with a step size of 0.4 μm was done on the electropolished surface at
on the microstructure, tribological properties and corrosion behaviour 20 kV with a 70° pre-tilt sample holder. The hit rate for all the sample
of additively manufactured 17-4PH stainless steel. The 17-4PH stainless was 98.5 ± 0.5%. The polished samples were electropolished in 95%
steel is a precipitation-hardened martensitic stainless steel which is acetic acid + 5% perchloric acid at room temperature at 30 V for
relatively soft in the martensitic state and can be further hardened via 30 sec. The electropolished sample surface tilted at 5° was further
“solution treatment + aging” process. Customs 17-4PH stainless steels treated with a Hitachi IM4000 low energy Ar ion miller operating at
have been utilised in the shuttle program due to the combination of 5 kV in flat mode for 20 min.
high wear resistance, good corrosion resistance, resistant to stress cor- A UMIS instrument was used for nanohardness (H) and reduced
rosion cracking (SCC) and high toughness at service temperature of up modulus (Er) measurements using a three-sided pyramid Berkovich tip
to 300 °C. at 100 mN force at ambient temperature. The load-displacement curve
of the indenter was recorded during the nanoindentation and the in-
2. Materials and experimental methods denter was held at peak load for three seconds to remove any creep
effect. Nanoindenter systems are housed within an environmental en-
17-4PH stainless steel cylindrical samples (Ø6 mm × 25 mm) were closure to protect the system from acoustic disturbances, airflow, and
manufactured using a 3D Systems Prox300 SLM machine. The Prox 300 temperature variability. ASTM E2546 was used as a guideline for in-
SLM machine has a continuous single-mode 500 W fiber laser of dentation testing. The spacing between indentations was 40 μm to avoid
1070 nm wavelength. The machine has a build volume of the effect of previous indents on subsequent indents. Before indentation
250 × 250 × 300 (height)mm. The protective nitrogen gas was purged on each specimen, fused silica (as an infinitely stiff material) was in-
continuously during printing and the oxygen set point to start the print dented to calibrate the hardness and reduced elastic modulus (before
was 2000 ppm. The laser power, scanning speed, layer thickness, and and after the indent).
hatch spacing were all default parameters provided by 3D Systems, as Dry sliding wear tests were conducted on a linearly reciprocating
given in Table 1. The as-printed samples are labelled as given in ball-on-flat tribometer (CSEM Tribometer) at ambient temperature. All
Table 2. The applied energy density for all samples was identical. Fig. 1 samples were polished and ultrasonically cleaned in acetone for 5 min
provides images of the vertically and horizontally printed samples. before and after the wear test and dried with hot air. Wear debris was
Each sample was printed on a solid cylindrical support 2 mm in height. also collected for further analysis. Specimens for wear testing were
Commercial gas-atomized 17-4 PH stainless steel powder (supplied by

2
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Table 1
The default parameters provided by 3D Systems.
Parameters Laser power (W) Scanning speed (mm/sec) Hatch spacing (μm) Layer thickness (μm) Energy density (J/mm3)

Default 127.5 1200 50 40 53.12

Table 2
Labels used to identify the samples.
Sample Energy density Scanning Orientation Number of scans
Code (J/mm3) pattern per layer

Sample 1 53.12 Hexagonal Horizontal 1

Sample 2 53.12 Hexagonal Vertical 1
Sample 3 53.12 Hexagonal Vertical 2
Sample 4 53.12 Concentric Vertical 1
Sample 5 53.12 Concentric Vertical 2
Sample 6 53.12 Concentric Horizontal 1

Fig. 2. 17-4PH stainless steel powder supplied by 3D Systems company

mounted and levelled on the tribometer while the counterpart which (HV:10kV, Spot: 3.5, WD: 4.8 mm, Mag: 1000X, Det: ETD).
was made of a ruby ball with the diameter of 6 mm slid over the
samples at a defined stroke length. The normal load, sliding speed, and
Table 3
frequency were 2 N, 30 mm/sec, and 2.38 Hz, respectively. This load
Chemical composition (in wt.%) of the supplied 17-4PH stainless steel powder.
results in Hertzian contact stress of 966.1 MPa. A wear distance of 30 m
(3750 cycles) and a stroke length of 4 mm were employed for all wear Cr Ni Cu Mn Si Nb Mo C N P S Fe
tests. The coefficient of friction (COF) versus time was monitored and
16.7 4.52 4.13 0.92 0.67 0.32 0.11 0.025 0.093 0.02 0.01 Bal.
recorded during the wear test. The cross-sectional profiles were at-
tained across the wear track using a Confocal 3D Laser Scanning
Microscope (KEYENCE VK-X200). A straight line was drawn between Corrosion samples were left to equilibrate and stabilise for approxi-
the unworn areas on both sides of the wear track. The cross-sectional mately 1 hour in PBS solution. The temperature of the solution was
area of the wear track below the reference line was calculated and used 37 °C and it was purged with nitrogen gas with a flow rate of 150 cm3/
to compute the wear volume (minimum twenty straight lines were min. All electrochemical experiments were carried out at a scan rate of
drawn per sample). The wear volume (Vf in mm3) was then calculated 0.001 Vs−1 with the start potential of 0.1 V below the OCP to +0.8 V.
using the following equation: The corrosion rate and polarization resistance of the samples were
Vf= A×L (1) calculated using ASTM G59 guidelines [13]. A small potential scan, ΔE
(t), defined with respect to the open corrosion potential (Ecorr), is ap-
Where: plied to the sample and the resultant currents are recorded. The po-
A=average cross-sectional area of the wear track (mm2), and larization resistance, RP, is defined from Eq 2 as the slope of a potential
L = length of stroke (mm) versus current [13]:
The dimensional wear coefficient (k in mm3/N.m) was then calcu-
lated by dividing the wear volume by sliding distance and normal load. ∂ΔE ⎞
RP = ⎛ i = 0, dE / dt → 0
This is a helpful parameter to compare the wear rate in different ma- ⎝ ∂i ⎠ (2)
terials and represents the volume of material removed by wear (in
The corrosion current density, icorr, is related to the polarization
mm3) per unit sliding distance (in m), per unit normal load on the
resistance by the Stern-Geary coefficient, B, as follows [13]:
contact (in N).
A saturated calomel electrode (SCE) was used as the reference B
icorro = 106 ×
electrode and a platinum mesh was used as the counter electrode. RP (3)

Fig. 1. Horizontally and vertically additively manufactured 17-4PH stainless steel samples.

3
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 3. The schematic of laser scanning strategy used in the present study: (a) hexagonal, (b) concentric.

Fig. 4. SEM images of round-section and rectangular-section of: (a, b) sample 1, (c, d) sample 2, and (e, f) sample 3 manufactured using hexagonal scanning pattern
(HV: 15KV, Spot: 4.5, Mag: 500X, Det: ETD).

4
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 5. SEM images of round-section and rectangular-section of: (a, b) sample 4, (c, d) sample 5, and (e, f) sample 6 manufactured using concentric scanning pattern
(HV: 15KV, Spot: 4.5, Mag: 500X, Det: ETD).

Fig. 6. Weld pool geometry in the horizontally printed samples using (a) hexagon scanning pattern, and (b) concentric scanning pattern. The arrow shows the build
direction (HV: 15KV, Spot: 4.5, Mag: 2000X, Det: ETD). Dashed lines compare the weld geometry in hexagon and concentric scanning patterns.

5
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Table 4
The amount of porosity in additively manufactured samples.
Sample code Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6

Porosity (%) 0.53 ± 0.25 0.6 ± 0.43 0.15 ± 0.07 2.95 ± 0.8 1.15 ± 0.9 0.8 ± 0.4

Fig. 7. Double-scan densifies the microstructure and eliminates balling defects and irregular-shape voids causing by lack of fusion: hexagonal scanning pattern (a)
single-scan, (b) double-scan, and concentric pattern (c) single-scan, (d) double-scan.

The dimension of Rp is ohm-cm2, icorr is muA/cm2, and B is in V. The balling defect in the present study varied from 10 μm to 100 μm and it
Stern-Geary coefficient is related to the anodic, ba, and cathodic, bc, was observed that concentric scanning strategy contained more balling
Tafel slopes as given in Eq 4 [13]: defect and elongated void caused by incomplete melting. The hex-
agonal scanning pattern has overlapping features (≈100 μm) between
ba bc
B= hexagon islands which decreases the process-induced porosity during
2.303 (ba + bc ) (4)
SLM process. In addition, the hexagon scanning islands strategy has
The units of the Tafel slopes are V. Once the icorro is determined, the shorter raster length as compared to concentric pattern which decreases
corrosion rate, CR, in mm per year can be calculated from Eq 5 in which the residual stress and thermal gradient. The horizontally printed
E.W. is the equivalent weight of the corroding species in grams and ρ is samples are expected to have higher cooling rates and more residual
the density of the corroding material in g/cm3 [13]: stress due to the larger surface area during laser scan whereas the
vertically printed samples have smaller circular cross-section slowing
icorr (E . W . )
CR = 3.27 × 10−3 down the cooling rate. The columnar dendrites grow with an easy-
ρ (5)
growth direction (which is < 100 > for bcc and fcc metals) closest to
the heat flow. This is seen in the round cross-section of the horizontally
3. Results and Discussion printed samples and rectangular cross-section of the vertically printed
samples in both scanning patterns in Figs. 4 and 5. In the hexagon
The SEM images of the circular and transverse cross-section of the scanning pattern, the grain morphology formed along the build-direc-
cylindrical-shape samples fabricated using hexagonal (Fig. 4a-f) and tion (see the rectangular cross-section in the horizontally printed
concentric (Fig. 5a-f) scanning patterns are shown in Figs. 4 and 5. The samples and round cross-section in the vertically printed samples) is
most common type of defects observed in the microstructure is balling cellular shape and it appears more irregular in the horizontally printed
defect and lack of fusion which mostly occurred in the concentric samples as compared to the vertically printed ones which could be due
scanning pattern. Balling defect takes place when the liquid metal fails to the larger surface area during scan and higher cooling rate. However,
to wet the underlying substrate due to the surface tension and presence in the concentric scanning pattern, the grains grow from the outer edge
of oxide films and it is classified as plate, cup and sphere according to towards the center and the size of the cells seems to be finer with even
the shape and morphology [14]. It was shown that ellipsoidal balling more irregularity. This microstructure anisotropy could potentially in-
defect with size of ≈ 500 μm caused by poor wettability was detri- fluence the mechanical and tribological properties of the additively
mental for mechanical properties whilst spherical balling defect with manufactured components. It was observed that in the vertically
dimension of 10 μm caused no harm to SLM quality [14]. The size of printed samples, the depth and width of the weld pool were identical in

6
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

shown in Fig. 7. Therefore, it can be concluded that double-scan was

quite an effective method to eliminate porosities and balling defects at
the expense of increasing the manufacturing time.
The XRD patterns of the as-printed 17-4PH stainless steel samples
are plotted in Fig. 8. It should be noted that all XRD measurement was
conducted on the round cross-section. The microstructure mainly con-
tains martensite, retained austenite and M23C6 and/or Fe3C carbides.
The XRD patterns of the vertically printed samples (black patterns)
are identical but hexagon scanning strategy has led to slightly more
retained austenite and carbides. This is mainly due to the island scan-
ning strategy which reduces the temperature gradient and gives enough
time for carbides to precipitate in the matrix. The double-scan strategy
(red patterns) increased the intensity of (200)α and (211)α peaks which
indicates the formation of a stronger texture along the build direction.
This texture seems to be stronger in the hexagon scanning pattern than
concentric pattern since the increase of intensity was slightly larger in
the sample printed using hexagon strategy. According to the XRD pat-
terns, the change of build orientation from vertical to horizontal had
more effect on the texture of the samples printed with hexagon pattern.
This means that the intensity of (200)α and (211)α peaks decreased in
the horizontally printed samples using hexagon pattern whereas no
considerable difference was noted in the concentric pattern. The vo-
lume fraction of retained austenite was calculated using integrated in-
tensities of the diffraction peaks and listed in Table 5. As seen, the
double-scan strategy decreased the volume fraction of retained auste-
nite in both hexagon and concentric patterns mainly due to the decrease
of stress at the grain boundaries and transformation of retained auste-
nite to martensite after remelting and cooling. In addition, it was found
out that the change of orientation from vertical to horizontal altered the
type of carbides precipitated within the matrix from M23C6-type to
Fe3C-type carbides. In fact, it seems that the larger cross-section of the
horizontally printed sample acts as a better heat sink to cool down the
part which apparently has impeded the precipitation of M23C6-type
carbides in the matrix.
Fig. 8. The XRD patterns of the as-printed 17-4PH stainless using (a) hexagon The EBSD patterns of the vertically and horizontally printed samples
pattern, and (b) concentric pattern (Hex: hexagonal pattern, Conc: concentric taken along and perpendicular to the build direction using hexagon and
pattern, Ver: vertical, Hori: horizontal, DS: double scan). concentric scanning patterns are shown in Figs. 9 and 10, respectively.
The microstructure mainly consists of martensite (blue colour) and
both scanning patterns whereas in the horizontally printed samples, the retained austenite (red colour), as already confirmed by the XRD pat-
width of the weld pool in the hexagon scanning pattern was slightly terns. It was observed that the retained austenite was mainly formed at
wider, as shown in Fig. 6. the grain boundaries mainly due to the large strains at the grain
This difference in the weld pool geometry clearly indicates the heat boundaries, high density of dislocations and finer grain size [15,16].
dissipates slower in the hexagon island scanning pattern as compared to In the hexagon scanning strategy, the grain size of the horizontally
the concentric pattern which reduces the residual thermal stress and printed samples in both cross-sections was found to be smaller than the
thermal gradient during manufacturing minimising the risk of thermal vertically printed samples. Larger scanning area dissipates heat faster
crack and delamination. This difference in the weld pool geometry was leading to a higher cooling rate and smaller grain size. It was also noted
not noticeable in the vertically printed samples. This might be due to that the texture was stronger in the vertically printed samples which
the smaller cross-section area of the laser scan in the vertically printed could be attributed to the longer manufacturing time and higher
samples as well as the surrounding powders which both could poten- thermal gradients during manufacturing promoting epitaxial growth
tially reduce the cooling rate. leading to a stronger texture. The size of the retained austenite grains
The cross-section of all samples was examined using OM and SEM to was independent of the scanning pattern and build orientation. In ad-
estimate the amount of porosity in each sample and results are listed in dition, it was observed that the double scan influenced the texture of
Table 4. It was observed that the amount of porosity and balling defects the retained austenite considerably resulting in the formation of aus-
were higher in the concentric scanning pattern when compared with tenite grains with a stronger texture.
the hexagonal pattern. However, scanning each layer twice (or double- In the concentric pattern, no considerable difference was observed
scan) has effectively improved the density and decreased the number of between the grain size of the horizontally and vertically printed sam-
pores and balling defects. The effect of remelting in elimination of ples. It was also noted that the scanning strategy has no significant
balling defects and lack of fusion was extensive in the concentric effect on the average grain size. However, due to the different scanning
scanning pattern as compared with hexagonal scanning pattern, as pattern and thermal history experiencing during manufacturing, the

Table 5
The volume fraction of retained austenite in the as-printed samples.
Sample code Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6

retained austenite (%) 7.4 ± 0.4 9.5 ± 0.5 3.7 ± 0.3 6.2 ± 0.4 5.16 ± 0.3 3 ± 0.5

7
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 9. The EBSD patterns (left to right: band contrast, phase colour map, and pole figures) of samples 1, 2 and 3, taken along the build direction and perpendicular to
the build direction.

morphology of the grains is quite different, as shown already in the SEM to be ≈ 11% larger than that of printed by the concentric pattern. This
images. This difference in grain orientation in respect to the loading is attributed to the lower volume fraction of defects and porosities with
axis could cause a significant variation in mechanical and tribological hexagon scanning pattern. The maximum anisotropy in nanohardness
properties. Similar to the hexagon scanning pattern, the double-scan measurements was observed in the vertically printed sample using the
increased the texture of the retained austenite grains in the concentric single-scan mode (samples 2 and 4 in Fig. 11a). The double-scan de-
scanning pattern. creased the variation of hardness in both scanning patterns (samples 3
The variation of hardness and Young’s modulus of the 3D printed and 5 in Fig. 11a) which could be due to the decrease of porosities/
samples is shown in Fig. 11a-b. All measurements were conducted along pores and increase of theoretical density. Porosities are compressed
and perpendicular to the build direction. The average hardness of the during hardness testing leading to lower hardness readings.
samples printed using hexagon pattern in both cross-sections was found The variation of wear coefficient, depth of the wear track, schematic

8
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 10. The EBSD patterns (left to right: band contrast, phase colour map, and pole figures) of samples 4, 5 and 6, taken along the build direction and perpendicular
to the build direction.

Fig. 11. Hardness (H) and reduced modulus (Er) of the additively manufactured specimens measured along the build direction and perpendicular to the build
direction.

9
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 12. (a) Wear coefficient, (b) depth of the wear track in samples 1 to 6, and (c) schematic of the examined surfaces in the horizontally and vertically printed
samples.

of the examined surfaces for wear and hardness testings, and coefficient also noticed that by carefully avoiding the regions containing defects,
of frictions in the samples 1 to 6 are plotted in Fig. 12a-c and Fig. 13a-f. the wear resistance of the 3D printed parts becomes closer to the
As one can see, the wear resistance of the samples manufactured via commercial wrought alloys. This indicates that by carefully adjusting
hexagon pattern (i.e. samples 1, 2, and 3) is generally better than those the manufacturing parameters and eliminating defects, better wear
fabricated via concentric pattern (samples 4, 5, and 6). In addition, it behaviour is achievable. Similar behaviour was reported in Sun’s study
seems that the change of orientation from vertical to horizontal did not on 316 L stainless steel where the wear resistance of the printed samples
influence the wear resistance in the hexagon pattern considerably. was controlled by porosity and microvoids [17].
However, the double scan improved the wear resistance by ≈60% The coefficient of wear friction in all samples vary between 0.1-0.2
which is attributed to the decrease of porosity and balling defects via during the wear test irrespective of scanning patterns, build direction
remelting of the previously fused layer. The relatively high wear rate of and number of scans per layer. The SEM images of the wear tracks, as
the samples 4, 5, and 6 fabricated using concentric pattern is due to the shown in Fig. 14a-b, show the presence of iron oxide film on the surface
large volume fraction of porosities, as already shown in the OM and which determines the coefficient of friction. The oxide film is smooth
SEM images. Similar to the hexagon pattern, the double scan method in and contains shear-like features. If the protective oxide film is un-
the concentric pattern is effective and enhanced the wear resistance but damaged, the coefficient of friction remains almost constant; however,
unlike hexagon pattern, the change of orientation from vertical to upon further penetration and exposure of metallic surface, the coeffi-
horizontal has affected the wear resistance noticeably. cient of friction increases suddenly as seen in Fig. 13.
This is evident from Table 4 where the lowest amount of porosity in According to the SEM images of the wear track, the fraction of fresh
the concentric pattern was observed in the horizontally printed sam- and newly exposed surfaces was found to be greater in the concentric
ples. As mentioned earlier, hexagon islands in the hexagon scanning pattern as compared to hexagon pattern indicative of instability of the
strategy are overlapped and raster length is shorter which improves the oxide film forming on the surface. As shown earlier, the volume fraction
theoretical density and minimises the concentration of defects and in of porosity in the samples manufactured using concentric pattern was
particular porosities. However, in the concentric pattern where laser considerably larger than that of samples printed by hexagon pattern.
beam travels from the outer edge toward the inner core, the porosity Defects such as porosity act as a stress riser and provide favourable sites
was formed in the areas where laser beam changed its direction. for the nucleation and propagation of the crack. Double scan strategy
The depth of the wear track in the commercial 17-4PH wrought has proved to be effective to decrease the volume fraction of the por-
alloy was 0.8 μm whilst the wear track in the as-printed samples was osity and balling defects via remelting of the previous layer leading to a
wider and deeper. It is interesting to note that the hardness of the as- narrower wear track in both scanning patterns, as shown in Fig. 15a-d.
printed samples was similar or even higher than the available 17-4PH The decrease of defects reduces the number of potential sites for the
wrought alloy which is due to the high cooling rate and refined mi- nucleation and propagation of the cracks during wear testing.
crostructure. But it seems that in presence of porosities and balling Yang et al. [18] showed that when the direction of columnar cells
defects, hardness does not contribute to wear resistance of the as- was perpendicular to the sliding direction, the wear resistance was
printed samples under current reciprocating action. In addition, it was higher as compared to the direction of columnar cells being parallel to

10
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 13. The variation of coefficient of friction along the build direction and perpendicular to the build direction in: (a) sample 1, (b) sample 2, (c) sample 3, (d)
sample 4, (e) sample 5, and (f) sample 6.

11
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 14. The comparison of the worn surfaces and point analysis results in: (a) sample 1 (Hexagonal-horizontal-round section), and (b) sample 4 (Concentric-vertical-
rectangular section).

Fig. 15. Double-scan strategy improved the wear behaviour and decreased the width of the wear track in (a, b) hexagon pattern, (c, d) concentric pattern (HV:15KV,
Spot: 4.5, Det: ETD).

the sliding direction in 1- and 10-reciprocating cycles scratch testing of samples were calculated and tabulated in Table 6 for comparison.
316 L stainless steel. It was concluded that more grain boundaries and As one can see, the corrosion current density of the samples printed
greater slipping resistance against the reciprocating action was re- by the hexagon pattern is considerably lower than those printed by
sponsible for the better wear resistance. The effect of columnar grain concentric pattern. The lowest corrosion rate (or highest polarization
growth in respect to the sliding movement was not conclusive in the resistance) and highest corrosion rate (or lowest polarization re-
present study. This could be due to the presence of porosities playing sistance) was observed in sample 2 (red curve in Fig. 16a) and sample 4
dominant role in 3750 cycles reciprocating wear test in the present (black curve in Fig. 16b), respectively.
work causing significant work-hardening as compared to one- and 10- The corrosion potential at open circuit is a measure of tendency to
cycles scratch testing in Yang’s study [18]. form protective passive film and the corrosion current density at open
The polarization curves of the as-printed samples measured in the circuit corresponds to dissolution rate. The higher the corrosion po-
phosphate buffered saline (PBS) are shown in Fig. 16. The current tential (positive values) the more passive is the material and the higher
density, corrosion rate (CR), and polarization resistance (PR) of all corrosion current density the higher is the corrosion rate [17]. Samples

12
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

conclusive. It should be noted that the circular cross-section was ex-

amined for the corrosion properties of 3D printed samples. In the
hexagon pattern, the vertically printed samples showed better corrosion
resistance whereas in the concentric pattern, the passive film is not
stable in the vertically and horizontally printed samples and only
double-scan strategy improved the passivation behaviour. It seems that
similar to wear results, volume fraction of porosities and defects play an
important role in corrosion resistance. The existence of pores makes the
alloy prone to oxide film break down leading to accelerated dissolution
of the undelaying metal. In fact, regions near the pores, as shown al-
ready in Fig. 17, are attacked and dissolved faster than the rest of the
sample. In samples 4 and 5 (see Fig. 16b), the breakdown occurs twice
at corrosion potentials of ≈ 0.007-0.04 V and 0.07-0.12 V during up-
ward scanning. However, the double scan strategy decreased the vo-
lume fraction of pores and as a result of that, the breakdown took place
only one time at corrosion potential of 0.7 V indicating higher re-
sistance to pitting corrosion.
The inductively coupled plasma (ICP) analysis of the PBS solution of
samples 1 to 6 is presented in Table 7. The ICP results show that sample
2 emits the least metal ions as compared to other samples. The release
of metal ions could cause allergic reaction and form reactive oxygen
species (ROS) possessing deleterious effects on tissues and cells causing
various diseases. Elements such as Cr (Cr6+ and Cr3+), and Ni (Ni2+/
Ni3+) could induce oxidative stress (perturbation of cell membrane),
immunogenicity (strong inflammatory response) and carcinogenicity
(causing cancer in humans) in the human body. The lowest amount of
ion release in sample 2 can be attributed to the higher density, lower
volume fraction of defects and stable and protective passive film
forming on the surface. This indicates that with the optimization of the
process parameters and choosing the suitable scanning pattern and
build orientation, better corrosion and wear resistance and bio-
compatibility response are achievable.

4. Conclusions
Fig. 16. Polarization curves measured in PBS solution: (a) Hexagon scanning
pattern, (b) Concentric scanning pattern.
The effect of process parameters (scanning pattern, build orienta-
tion, and number of scans per layer) on the microstructure, dry sliding
1 to 3 tend to passivate at corrosion potential above the open circuit wear, and corrosion properties of additively manufactured 17-4PH
potential, but this passivation occurs faster in sample 2 as compared to stainless steel was investigated in the present study and the following
samples 1 and 3. Unlike wear testing, double scan did not contribute to conclusions can be drawn from the results:
corrosion resistance in the hexagon pattern but up to around 30% im-
provement was observed in the concentric pattern. In addition, the 1 The weld pool geometry was identical in the vertically oriented
polarization curve in the hexagon pattern is smoother when compared samples in both scanning patterns whereas in the horizontally or-
to concentric pattern. The presence of current spikes in the concentric iented samples, the width of the weld pool was slightly wider in the
patterns indicates the instability of the passive film due to the presence hexagon pattern as compared to concentric scanning pattern, in-
of pores and porosities in the matrix causing localised corrosion and dicating uniform dispersion of heat and lower cooling rate. The
pitting, as shown in Fig. 17a-b. variation of thermal gradients affects the volume fraction of retained
The passive film formed on the surface of the samples 4 and 6 austenite and carbides in the matrix.
(concentric scanning pattern) breaks down a few times before reaching 2 The double scan strategy (remelting each layer twice) has effectively
the upper potential of 0.8 V; however, the double scan strategy (in reduced the volume fraction of porosities and balling defects in both
sample 5) results in the stability of the protective oxide film forming on scanning patterns and led to the formation of a stronger texture
the surface. This is attributed to the decrease of porosities and pores along the build direction. The volume fraction of retained austenite
decreasing the possibility of localised attack and pitting corrosion. was decreased by remelting process mainly due to the release of
The effect of build orientation on the corrosion properties of the stress at the grain boundaries and transformation of retained aus-
samples manufactured using hexagon and concentric patterns is not tenite to martensite after cooling.

Table 6
The current density, corrosion rate and polarization resistance obtained from Tafel plot
Sample Current density, Jcorr (A/cm2) Corrosion rate, CR (mm/y) Polarization resistance (Ω)

-8
Sample 1-Hex-Hor 13 × 10 0.001502 108880
Sample 2-Hex-Ver 7.89 × 10-8 0.000918 875390
Sample 3-Hex-Ver-DS 9.01 × 10-8 0.001047 182000
Sample 4-Con-Ver 2.81 × 10-7 0.003271 67127
Sample 5-Con-Ver-DS 1.99 × 10-7 0.002312 107190
Sample 6-Con-Hor 1.87 × 10-7 0.002176 45260

13
H.R. Lashgari, et al. Materials Today Communications 25 (2020) 101535

Fig. 17. The corroded surfaces of: (a) sample 4, and (b) sample 6, showing the localised attacked and corrosion products forming around the porosity and break down
of the passive film.

Table 7 Declaration of Competing Interest

The ICP analysis of PBS solution.
Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6 The authors report no declarations of interest.

P (mg/L) 304.3 328.5 350 338 329 337

Acknowledgement
S (mg/L) 0.12 0.41 0.1 0.01 0.18 0.39
Si (mg/L) 0.82 3.36 3.03 2.23 1.65 0.7
Cr (μg/L) 43.15 1.42 70.83 97 39 100.2 This research did not receive any specific grant from funding
Cu (μg/L) 3.6 0.8 15 19.4 10.5 33.8 agencies in the public, commercial, or not-for-profit sectors. The au-
Fe (μg/L) 388 36 318 393 178 490 thors would like to thank Mr. Bill Joe (Research Support Engineer,
Mn (μg/L) 6.27 1.82 3.63 5.82 2.8 7.35
Mo (μg/L) 1.66 0.53 1.13 1.16 0.64 1.22
School of Materials Science and Engineering, UNSW, Sydney) for his
Nb (μg/L) 1.35 0.43 0.78 1.31 0.26 0.89 technical assistance in conducting nanoindentation and wear testing.
Ni (μg/L) 25.6 5.12 22.7 31.64 14.58 47.15
Ta (μg/L) 0.58 0.35 0.25 0.15 0.12 0.26 References

[1] H. Fayazfar, et al., A critical review of powder-based additive manufacturing of

3 The average hardness of printed samples using the hexagon scan- ferrous alloys: Process parameters, microstructure and mechanical properties,
ning pattern was ≈11% greater than that of fabricated using the Materials & Design 144 (2018) 98–128.
[2] A. Yadollahi, N. Shamsaei, Additive manufacturing of fatigue resistant materials:
concentric pattern which is due to the lower volume fraction of
Challenges and opportunities, International Journal of Fatigue 98 (2017) 14–31.
defects and porosities in the matrix. [3] W.J. Sames, et al., The metallurgy and processing science of metal additive man-
4 In general, the hexagon scanning strategy led to better wear re- ufacturing, International Materials Reviews 61 (5) (2016) 315–360.
sistance as compared with concentric pattern and remelting process [4] C.Y. Yap, et al., Review of selective laser melting: Materials and applications,
Applied Physics Reviews 2 (4) (2015).
improved the wear resistance by ≈60% which is due to the decrease [5] Y. Liu, Y. Yang, D. Wang, A study on the residual stress during selective laser
of porosities and balling defects during remelting process. Under the melting (SLM) of metallic powder, The International Journal of Advanced
presence of porosities and balling defects, the increase of hardness Manufacturing Technology 87 (1-4) (2016) 647–656.
[6] W. Gao, et al., The status, challenges, and future of additive manufacturing in en-
due to the microstructural refinement and rapid solidification did gineering, Computer-Aided Design 69 (2015) 65–89.
not necessarily result in better wear resistance. [7] D.D. Gu, et al., Laser additive manufacturing of metallic components: materials,
5 The oxide film was found to be more unstable and the area fraction processes and mechanisms, International Materials Reviews 57 (3) (2013) 133–164.
[8] A. Keshavarzkermani, et al., An investigation into the effect of process parameters
of freshly exposed surfaces was larger in the samples fabricated by on melt pool geometry, cell spacing, and grain refinement during laser powder bed
concentric scanning pattern mainly owing to the porosities and fusion, Optics & Laser Technology 116 (2019) 83–91.
micropores, acting as stress concentrator areas and providing nu- [9] A. Yadollahi, et al., Effects of building orientation and heat treatment on fatigue
behavior of selective laser melted 17-4 PH stainless steel, International Journal of
cleation sites for the initiation and propagation of cracks.
Fatigue 94 (2017) 218–235.
6 Similar to wear, the hexagon scanning strategy resulted in better [10] M. Mahmoudi, et al., Mechanical properties and microstructural characterization of
corrosion resistance than that of the concentric pattern. The lowest selective laser melted 17-4 PH stainless steel, Rapid Prototyping Journal 23 (2)
(2017) 280–294.
corrosion current density and least amount of ion release in the PBS
[11] R. Rashid, et al., Effect of scan strategy on density and metallurgical properties of
solution was observed in the samples made by single-scan hexagon 17-4PH parts printed by Selective Laser Melting (SLM), Journal of Materials
scanning pattern orientated vertically on the build plate. Processing Technology 249 (2017) 502–511.
[12] A. Kudzal, et al., Effect of scan pattern on the microstructure and mechanical
properties of Powder Bed Fusion additive manufactured 17-4 stainless steel,
Materials & Design 133 (2017) 205–215.
Author Agreement Statement [13] Standard Test Method for Conducting Potentiodynamic Polarization Resistance
Measurements. ASTM G59, (2009).
[14] R. Li, et al., Balling behavior of stainless steel and nickel powder during selective
All persons who meet authorship criteria are listed as authors, and laser melting process, The International Journal of Advanced Manufacturing
all authors certify that they have participated sufficiently in the work to Technology 59 (9-12) (2011) 1025–1035.
take public responsibility for the content, including participation in the [15] H.R. Lashgari, et al., Microstructure, post thermal treatment response, and tribo-
logical properties of 3D printed 17-4 PH stainless steel, Wear (2020) 456–457.
concept, design, analysis, writing, or revision of the manuscript. [16] J.P. Oliveira, T.G. Santos, R.M. Miranda, Revisiting fundamental welding concepts
Furthermore, each author certifies that this material or similar material to improve additive manufacturing: From theory to practice, Progress in Materials
has not been and will not be submitted to or published in any other Science (2020) 107.
[17] Y. Sun, A. Moroz, K. Alrbaey, Sliding Wear Characteristics and Corrosion Behaviour
publication before its appearance in the Materials Today
of Selective Laser Melted 316L Stainless Steel, Journal of Materials Engineering and
Communications. Performance 23 (2) (2013) 518–526.
[18] Y. Yang, et al., Wear anisotropy of selective laser melted 316L stainless steel, Wear
428-429 (2019) 376–386.

14