Additive Manufacturing 29 (2019) 100818

Contents lists available at ScienceDirect

Additive Manufacturing
journal homepage: www.elsevier.com/locate/addma

Full Length Article

Ultrasonic material dispensing-based selective laser melting for 3D printing T

of metallic components and the effect of powder compression
⁎
Chao Wei, Heng Gu, Zhe Sun, Dongxu Cheng, Yuan-Hui Chueh, Xiaoji Zhang, Yihe Huang, Lin Li
Laser Processing Research Centre, School of Mechanical, Aerospace and Civil Engineering, The University of Manchester, Oxford Road, Manchester, M13 9PL, UK

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

Keywords: Selective laser melting (SLM) is one of the most commonly used metallic component 3D printing techniques. In a
Ultrasonic powder dispensing previous investigation of multiple materials SLM reported by The University of Manchester, high porosities and
Selective laser melting cracks were found in the regions where the powder was deposited via an ultrasonic powder dispenser. The low
Packing density powder packing density was identified as a critical reason for this. In this paper, we report a new method to
Powder compression
compress the ultrasonically deposited powder layer in order to increase the powder packing density. The effects
Porosity
of powder deposition velocity, powder track overlap distance and powder compression force on the deposited
powder characteristics were investigated. The microstructure, tensile strengths, and porosity of the laser-fused
samples were analyzed. The results indicated that powder compression could reduce porosity and component
distortion and increase the mechanical strength of the printed parts.

1. Introduction high thermal conductivity of copper [8].

Selective laser melting (SLM) is a powder bed-based additive man-
ufacturing (AM) technology in which a focused laser beam driven by a
galvanometer scanner melts a thin layer of dry powder particles spread
on the processing substrate, and the whole component is built layer by
layer according to the designed cross-section patterns sliced from 3D
models [1]. SLM can be used to process a wide range of metallic ma-
terials, including titanium, nickel, aluminum, copper, tungsten, gold
and related alloys [2]. Conventional powder bed-based SLM is only able
to print a single material in each layer [3], which precludes its appli-
cation to components where multiple materials may be more desirable
[4]. Hence, modification of SLM to enable the processing of multiple
materials is a current challenge. AM of multiple materials is a promising
method for fabrication innovation that can integrate different desirable
physical properties into one component by spreading and binding dis-
tinct raw materials layer by layer in one manufacturing process [5,6].
For instance, it can fabricate a bimetal base plate with different thermal
expansion coefficients, which is used in automatic cooling control in-
struments for the aerospace industry [7]. It can also produce the
tungsten/copper functionally graded material (FGM) diverter plate that
is directly exposed to the plasma in steady-state magnetic fusion re-
actors, i.e. nuclear reactors. The tungsten side of the FGM can bear the
high temperature in a plasma nuclear radiation environment. The
copper-rich side can be cooled with water effectively because of the
One of the critical challenges for SLM AM of multiple materials is
the local deposition of different types of materials on the same SLM
processing layer [9]. Ultrasonic vibration-assisted powder deposition is
a promising technology to achieve this [10]. Previous investigations
[11] revealed that dry powder flow could be improved by ultrasonic
vibration, as the high-frequency oscillation breaks the fine powder
particle agglomerates to achieve steady, continuous powder flow. It was
found that the powder flow start and stop could be accurately con-
trolled via the ultrasonic vibration excitation signal, similar to using a
control valve [12]. These two features of ultrasonic powder feeding are
ideal for depositing different powder materials for multiple materials
SLM AM. The University of Manchester recently demonstrated a novel
ultrasonic vibration-assisted powder dispensing technology for multiple
materials AM [13]. They employed the classic powder bed mechanism
to spread the primary structural material, and ultrasonic dispensers
were applied to selectively deposit other dissimilar materials on the
same layer, assisted by a selective micro-vacuum material remover.
However, due to the nature of the point-by-point powder deposition
mechanism, the cross-section of the dispensed powder was an isosceles
trapezoid shape [14,15], and stripes may appear after several tracks of
powder are deposited by the dispenser, due to the ridges between two
adjacent tracks [16,17]. The porosity in the fused components and poor
surface roughness did not meet the rigid requirements of SLM [18]. In
addition, excessive powder build-up was found at the corners of the

⁎
Corresponding author.
E-mail address: lin.li@manchester.ac.uk (L. Li).

https://doi.org/10.1016/j.addma.2019.100818
Received 26 November 2018; Received in revised form 24 July 2019; Accepted 30 July 2019
Available online 08 August 2019
2214-8604/ © 2019 Elsevier B.V. All rights reserved.
C. Wei, et al.
ultrasonic depositing toolpath where the feeding velocity vector di-
rection changed dramatically [18,19]. The laser beam energy was un-
able to fully melt the thick powder layer, leading to cracks in this region
[20]. Furthermore, the packing density of the ultrasonically deposited
powder layer was low, as there was no external force to compress the
loose powder dispensed from the ultrasonic nozzle. A low packing
density would lead to high porosity in the final SLM-processed com-
ponents [21].
Most of the previous investigations on powder layer deposition
quality focused on metal powder spreading in the conventional SLM
process. In such a system, the powder layer thickness (typically
30–50 μm) was well controlled via the fixed gap distance between the
rigid blade tip and the building platform. The ‘front stress-free turning
powder wedge behavior’ generated by the rigid blade [22] caused the
relatively smaller powder particles in the Gaussian particle size dis-
tribution to move down towards the powder bed. The larger particles
were then swept into the powder collector bin [23]. Thus, the powder
particle size distribution on the recoated powder bed was smaller than
that of the virgin powder. The research of Liu et al. [24] indicated that a
wider powder particle size range contributed to a higher powder bed
density, higher density of the fused parts, and smoother surface fin-
ishing of the part’s sides because small particles could quickly fill the air
gaps between big particles during the blade recoating procedure. The
above two factors made the surface roughness of the powder layer re-
coated by the rigid blade very low, leading to smaller standard devia-
tion of the powder height on the same powder layer.
Previous investigations on ultrasonic powder deposition focused on
optimizing the powder flowability. No publications were found on ul-
trasonically deposited powder layer surface quality and the mechanical
performance of laser-melted samples from such layers. In the present
study, we demonstrated a powder compression system integrated into
an ultrasonic powder delivery SLM experimental setup, in which an
external compression force was applied to increase the deposited
powder layer packing density. Additionally, the effects of powder de-
position velocity, track distance, and compression force on the ultra-
sonically deposited powder layer thickness and density were in-
vestigated. The printed component microstructure, porosity and
mechanical performance of 316 L steel were also studied. The micro-
structure evolution and metallurgical behavior of the fabricated sam-
ples were discussed, and the tensile strengths and fractured surface
analysis of the fabricated samples were compared.
2. Experimental materials and procedure
2.1. Materials
LPW Technology Limited, UK provided the 316 L stainless steel
powder, shown in Fig. 1a. The gas atomized powder particle size had a
Fig. 1. a) SEM images of the 316 L stainless steel spherical powder used in this
study, b) the plot of the powder size distribution acquired via the particle size
analyzer.
2
Additive Manufacturing 29 (2019) 100818
Gaussian distribution, ranging from 10 μm to 45 μm. As shown in
Fig. 1b, the powder sizes of Dv(10), Dv(50), and Dv(90) correspond to
18.4 μm, 31.2 μm, and 53.4 μm respectively.
In all the experiments, the powder was contained in a 50 mL syringe
that was connected to a Musashi needle (with an inner diameter of
300 μm) for powder delivery.
In the ultrasonic powder deposition investigation, powder particles
were deposited on a 15 mm thick 304 stainless steel plate having a
surface roughness (Ra) of 2 μm. In the laser melting process, the sub-
strates were made of 304 stainless steel square plates
(25 mm × 25 mm × 12 mm) having a sandblasted top surface.
2.2. Experiment setup
A proprietary multiple materials SLM system as described in [13],
located in the Laser Processing Research Centre (LPRC) at The Uni-
versity of Manchester, was modified and employed in this study. An
ultrasonic dispenser (as shown in Fig. 2a), mounted on an x-y axis
gantry linear stage, was used to selectively deposit 316 L particles on
the processing platform, according to the designed pattern. The
working principle of this dispenser is described in detail in a previous
work [25]. A micro vibration motor inserted into the powder container
(see Fig. 2b) broke the bridge-like structure of the powder particles [26]
and prevented an unstable powder flow rate. The high frequency of the
piezoelectric transducer, which is shown in Fig. 2c, controlled the
powder flow start and stop. The distance between the micro-needle tip
and the deposition platform was kept at 10 mm. Subsequently, powder
stripes were compressed by a plate (of area 30 mm × 50 mm) driven by
a pneumatic cylinder mounted on the x-axis linear stage, as shown in
Fig. 2d. A galvo x-y beam scanner (Scanlab, IntelliSCAN III 20) guided
the laser beam, which was then focused to an 80 μm beam spot size
using an f-theta flat field lens with a 254 mm effective focal length from
a 1070 nm continuous-wave ytterbium fiber laser source (IPG Photo-
nics, YLR-500-WC) to selectively melt the powder layer on the pro-
cessing platform. The whole procedure, as shown in Fig. 2e, was carried
out in an argon gas environment, and the oxygen density was lower
than 0.3%.
It is worth noting that the powder supply chamber and the soft
blade shown in Fig. 2a were idle during the above experiments and
were only applied for the manufacture of the control group samples,
during which the whole experimental setup worked the same as the
conventional SLM system.
2.3. Ultrasonic powder deposition sample preparation
Theoretically, a compression force helps to increase the deposited
powder packing density [27]. A single track and a single layer of
powder particles was initially deposited and the powder packing
characteristics were investigated. Subsequently, the deposition and
SLM of multiple tracks and multiple powder layers were investigated.
The Kawakita and Heckel equations are the most commonly used
models to describe powder compaction behavior under a compressive
force. Notably, the forms of these equations are the same in the low-
pressure force range [28]. In this investigation, metallic particle plastic
deformation was assumed not to occur during dry powder compaction
under the given force, due to the small force applied. Eqs. (1) and (2)
given by Kawakita [29] were applied to describe the powder compac-
tion process,
Pa 1 P abPa
= + a ⇒C =
C ab a 1 + bPa (1)
V − V⎞
C= ⎛ 0
⎜ ⎟
⎝ V0 ⎠ (2)
where C is the relative volume reduction percentage, a and b are con-
stants, Pa is the compression force, V0 is the initial powder volume, and
C. Wei, et al.
Fig. 2. a) Experimental setup schematic diagram, b) image of the powder container having a micro vibration motor, c) image of the dual vibration powder dispenser,
d) image of the powder compressor, e) process flowchart.
V is the powder volume after compression. A small compression force
could lead to a large relative volume reduction. As a result, a low
compression force of 0.2 MPa was used in this investigation to avoid
plastic deformation of the powders.
a) Single track powder deposition
The effects of the powder deposition velocity and powder com-
pression on the powder thickness were investigated using single-track
powder lines.
Four powder lines were deposited at 1500, 2000, 2500, and
3000 mm/min powder feeding velocities respectively, as shown in
Fig. 3a. The compressor compressed the left-hand part of the deposited
powder lines, and the right-hand part was kept in a loose condition for
comparison purposes.
This single track powder deposition experiment was repeated five
times, according to the statistical methodology described in article
[30].
a) Single layer powder deposition
The effects of powder layer compression and the track distance
between the deposition powder stripes on powder layer thickness were
investigated. With the aim of minimizing the interferences caused by
experimental setup changes to the final experiment results, three rec-
tangular powder patterns were deposited on the same substrate, based
on a zig-zag toolpath design, as shown in Fig. 3b. The powder deposi-
tion track distance for each rectangle was 1.0 mm, 1.5 mm and 2.0 mm.
The powder feeding velocity and acceleration of the x-y stage were
3000 mm/min and 100 mm/s2, respectively. The ultrasonically de-
posited 316 L powder flow rate was kept at 2.0 g/min. After the powder
was dispensed, the “left-hand” sections of each rectangle were com-
pressed via a pneumatic powder layer compressor as described in
Fig. 2d, while the “right-hand” sections were used as control samples
for comparison.
3
Additive Manufacturing 29 (2019) 100818
The single layer powder deposition experiment was also repeated
five times to collect sufficient statistical replicates.
2.4. Preparation of the SLM samples for microstructure analysis
As indicated in Fig. 3c, two groups of 15 mm × 15 mm square
samples were printed. The samples in the right-hand group contained
loose powder layers before melting and were used as the control sam-
ples. The powder layers of the left-hand group were compressed before
laser fusing. Each group consisted of 3 samples with a stripe track
distance that increased in 0.5 mm increments from 1.0 mm to 2.0 mm.
The powder feeding velocity and acceleration of the x-y stage were
3000 mm/min and 100 mm/s2, respectively. The deposited powder
area for each sample was 30 mm × 30 mm. Every sample in the laser
fusion experiment was made up of 100 layers The toolpath offset dis-
tance between each layer, normal to the powder feeding direction, was
at half the value of its track distance in order to allow part of the newly
deposited powder to fill the gaps between the already melted adjacent
powder stripes in the previous layer, and improve the melted layer
surface roughness.
For the purpose of sectioning and grinding of the subsequent
sample, six pieces of 25 mm × 25 mm × 12 mm stainless steel squares
were employed as processing substrates. They were mounted together
in the proprietary SLM machine via an adaptor plate.
Table 1 shows the laser process parameters used in this experiment.
The previous single layer powder deposition results indicated that the
ultrasonically deposited maximum powder layer (see Table 1 and
Fig. 8d) was thicker than that in the normal SLM process with powder
bed spreading (i.e. 30 μm–50 μm [31]). The laser beam hatch distance
in this experiment was thus reduced to 40 μm, in order to increase the
relative laser energy deposition per unit area to melt the thicker powder
layers.
C. Wei, et al. Additive Manufacturing 29 (2019) 100818

Fig. 3. The toolpath design patterns. a) is for the single track powder deposition experiment. b) is for single layer ultrasonic powder deposition experiments. c) is for
the multiple layer laser melting experiment, where the red and blue arrow lines indicate the toolpath of the ultrasonic dispenser and the light blue, yellow, and grey
squares show the compressed powder regions, substrate plates, and melted powder regions for the melting process, respectively. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this article).

Table 1 Meanwhile, four equally sized cuboid blocks, as shown in Fig. 4b

316 L stainless steel laser process parameters. were produced via the conventional soft blade powder spreading ap-
Factors Value
proach. The powder layer thicknesses of these four blocks were 50 μm,
and the laser process parameters matched the values in Table 1. One
Laser power (W) 175 thin tensile test specimen was collected from the center of each cuboid
Scan speed (mm/s) 800 block and was used as the control.
Laser scanning hatch distance (μm) 40
The final tensile test samples, shown in Fig. 4d, were prepared ac-
Hatch angle increment (°) 67
Maximum powder layer thickness (μm) 1.0 mm track distance 87.52 cording to the ASTM E8-04 standard. WEDM resulted in a thin black
1.5 mm track distance 83.64 oxidation film on the samples’ surface, which was removed by pol-
2.0 mm track distance 78.83 ishing. The voids on the sample with a 1.0 mm track distance were
observed after polishing, as shown on the left side of Fig. 4d.

2.5. Tensile test samples preparation

Three cuboid blocks were produced via ultrasonic powder deposi-
tion with powder compression according to the dimensions described in
Fig. 4a. Each block was then sliced into 2 thin tensile specimens based
on the geometry described in Fig. 4c, using a wire electrical discharge
machining (WEDM) process.
2.6. Material characterization
a) 316 L powder size distribution analysis
The 316 L powder size distribution was examined by a particle size
analyzer (Mastersizer 3000, Malvern Panalytical Ltd).

4
C. Wei, et al. Additive Manufacturing 29 (2019) 100818

Fig. 4. a) Ultrasonically deposited and laser melted rectangle block for the tensile test, b) Contrast group produced by the conventional SLM approach, c) Dimensions
of the tensile test sample, d) WEDM-processed tensile test samples with different track distances.

a) Single track powder deposition and single layer powder deposition a) Tensile strength testing

A 3D non-contact profile-meter (μscan, NanoFocus AG, Germany)
was employed to capture the 3D profile of the ultrasonically deposited
single track powder stripes and single layer powder deposition squares
in the initial loose condition and after compression.
The mean powder track thickness at the centerline of each powder
stripe (i.e. along the powder deposition direction), representing the
actual powder deposition thickness and quality, was analyzed using
μscan® image processing software, and the coverage area percentages of
the deposited powder stripes in these mappings were analyzed using
Image-Pro Plus 6.0® image processing software according to the user
manual [32].
The statistical results of the powder thickness from single layer
powder deposition, including the maximum, minimum and mean
thickness, were analyzed via μscan® image processing software. The
area percentage of the exposed areas without powder covering, which
are illustrated in blue in the mappings, was analyzed via Image-Pro Plus
6.0® image processing software and then converted to the powder
coverage percentage.
Tensile strengths were examined on a universal testing machine
(CMT5504, MTS Systems Corporation), and the morphology of tensile
fracture surfaces was acquired with scanning electron microscopy
(Thermo Scientific, Quanta 450).
3. Results and discussion
3.1. Single track powder deposition
3.1.1. Simulation result
The discrete element method (DEM) was employed to simulate the
powder deposition process. The powder particles were simplified as
elastic spheres flowing out from a nozzle of a 300 μm inner diameter
with a 2.0 g/min flow rate. After the powder particles were deposited,
the compression process was modelled by applying a rigid plane
pressing the loose powder. A schematic diagram of the powder de-
position model is shown in Fig. 5.

a) Microstructure and porosity examination

The ‘3D stitching’ function of the KEYENCE VHX-5000 digital mi-

croscope was employed to acquire the top surface topographical pro-
files of the six 15 mm × 15 mm square samples before analysis of por-
osity and microstructure. Each sample was then cut into three pieces
with a 3 mm interval normal to the powder feeding velocity direction,
ground with 400#, 600#, 1000#, and 1200# grit emery papers, and
finally polished with 3.0 μm and 1.0 μm diamond polishing pastes. The
polished samples were etched using 10 vol.% oxalic acid solution.
‘ImageJ’ software was used to calculate the porosity of the final etched
samples based on the whole cross-section images, acquired with the ‘2D
stitching’ function of the KEYENCE VHX-5000 digital microscope. The
porosity analysis procedure with ImageJ software was described in the
article [33]. The porosity value was then converted to the sample re-
lative density. The microstructure was examined via scanning electron
microscopy (Zeiss Sigma VP FEG SEM).

Fig. 5. Schematic of the powder deposition model.

5
C. Wei, et al. Additive Manufacturing 29 (2019) 100818

The Hertz-Mindlin with JKR cohesion model was implemented to decreases from 141 μm to 81.6 μm when the powder deposition traverse
calculate the translational and rotational motions of the powder. The velocity is increased from 1500 mm/min to 3000 mm/min. After com-
normal force Fn , normal stiffness Sn , and normal damping force Fnd are pression, the powder thickness decreases to 110 μm and 57.5 μm for
given by Eqs. (3)(5) [34]. 1500 mm/min and 3000 mm/min respectively. Fig. 6 also shows the
increase in the powder coverage area when the deposition velocity
4 * * 32
Fn = E R δn increases, especially after compression. Based on the law of conserva-
3 (3)
tion of mechanical energy described in Eq. (11) [37], a higher powder
Sn = 2E * R*δn (4) deposition traverse velocity contributes to higher particle kinetic en-
ergy, which is expressed as the higher particle instantaneous velocity
5 Lne → when the particles land on the processing platform. The particles have
Fnd = −2 Sn m* vnrel
6 Ln2e + π 2 (5) to move a longer distance to absorb this part of the energy, during
which the friction between particles or between particles and the pro-
where E * is equivalent to Young’s Modulus, R* is the equivalent radius, cessing platform consumes the kinetic energy. The macro scale effect of
δn is the normal overlap, m* is the equivalent mass, e is the coefficient of this phenomenon is the wider particle distribution area.
→
restitution, and vnrel is the normal component of the relative velocity.
1 1
The tangential force Ft , tangential stiffness St , and tangential E = mgh 0 + mv02 = mght + mvt2
2 2 (11)
damping force Ftd can be expressed as Eqs. (6)(8) [34].
Ft = −St δt (6) where E is the total mechanical energy of a powder particle, m is the
particle weight, g is the acceleration due to gravity, h 0 and ht present
St = 8G * R* δt (7) the initial altitude and the instantaneous altitude of the particle, re-
spectively. v0 and vt are the initial particle velocity and the in-
5 Lne →
Ftd = −2 St m* vtrel stantaneous particle velocity when the particle landed.
6 Ln2e + π 2 (8)
where δt is the tangential overlap, G* is the equivalent shear modulus 3.1.2. Experimental results
→
and vtrel is a tangential component of the relative velocity. Rolling Fig. 7a shows a set of 3D profile mappings of one replicate in the
friction τi is considered by applying a torque to the contacting surfaces, single-track powder deposition experiment. The statistic mean powder
which is given by Eq. (9) [35]. track thickness, mean powder thickness at the centerline of each
powder stripe (i.e. along the powder deposition direction, representing
τi = −μr Fn Ri Fωi (9) the actual powder deposition thickness), and the powder coverage
where μr is the coefficient of rolling friction, Ri is the distance of the percentage, based on the 3D profile data of the five statistic replicates,
contact point from the particle centre and ωi is the unit angular velocity are presented in Fig. 7b–d respectively.
vector of the particle. Considering that the powder sizes were less than According to Eq. (12), the deposited powder weight at the unit
100 μm, the cohesion force among the particles, calculated by Eq. (10) distance is inversely proportional to the powder deposition velocity that
[36], was attributed to van der Waals forces. was proved by the single track powder deposition experimental result
presented in Fig. 7b and c.
3 4E * 3
FJRK = −4 πγE * a 2 + a W = b/ Vt (12)
3R* (10)
where a is the radius of the contact area. where Vt is the real-time velocity, W is the deposited powder weight at
Four different simulations were conducted by implementing dif- the unit distance, and b is the ultrasonically deposited powder flow rate
ferent powder deposition velocities of 1500 mm/min, 2000 mm/min, (2.0 g/min in this study).
2500 mm/min and 3000 mm/min. The loose powder particles were The maximum deposited powder track thickness decreased with
compressed after the deposition. The modelling results are presented in increasing deposition velocity, as presented in Fig. 7a and b. Ad-
Fig. 6. ditionally, a minimum value (88.50 / 54.80 μm before and after com-
Fig. 6 shows that the maximum height of the loose powder track pressing respectively) was achieved as the velocity was 3000 mm/min.

Fig. 6. Simulated results with different powder deposition velocities.

6
C. Wei, et al.
Fig. 7. a) 3D profiles of the single track powder lines, b) maximum powder thickness plot, c) plot of the mean powder thickness at the centerline of the powder tracks,
d) powder coverage plot.
Compression could significantly reduce the powder track thickness by
13–33 μm, as shown in Fig. 7b.
The mean powder thickness at the centerline of each powder stripe
(i.e. along the powder deposition direction), representing the actual
powder deposition thickness, is shown in Fig. 7c. The mean powder
thickness before compression also decreased linearly with increasing
powder deposition velocity. After compression, the deposition track
thickness was reduced by more than 10 μm. The thinnest powder layer
was achieved at a velocity of 3000 mm/min.
We also clearly observed an increase in deposited powder stripe
coverage area (i.e. width) with increasing powder feeding velocity, as
shown in Fig. 7a and d, in agreement with the theoretical prediction
that powder compression would further enlarge the powder coverage
area.
Powder compression was an effective approach to reduce the mean
powder thickness and increase the deposited powder coverage per-
centage. All the experimental results, including the decreasing trend of
the maximum powder thickness (in a loose condition/ a compressed
condition) and the increasing trend of the powder coverage area with
increasing powder deposition velocity, agreed well with the simulation
results.
3.2. Single layer powder deposition
Fig. 8a1–a6 present one set of 3D profile mapping results of the
ultrasonically deposited single powder layer. At the right corner of each
mapping, the powder thickness 2D distribution patterns are presented,
where the exposure area without powder covering is shown in blue.
The statistical results of powder thickness (including minimum,
maximum and mean values) and powder coverage percentage, based on
data from five replicates, are shown in Fig. 8b–e respectively.
Powder stripes are observed along the powder depositing direction,
7
Additive Manufacturing 29 (2019) 100818
i.e. the x-axis marked in Fig. 8a1–a6. The sharpness of the powder stripe
outline increased with increasing track distance and decreased after
powder compression. A smaller track distance contributed to higher
powder coverage percentage and resulted in a fuzzy powder stripe
outline. This trend is presented in Fig. 8a and e. Powder compression
also helped to improve the powder distribution, which was expressed as
a larger powder coverage area as indicated in Fig. 8e, as the spherical
particles at the peak of the powder stacks rolled down under the ex-
ternal compression force. This is verified by the powder thickness dis-
tribution plots in Fig. 8b–d. As shown in Fig. 8c, the maximum powder
thickness always drops after compression, regardless of the value of the
powder track distance. At the same time, both the minimum powder
thickness and the mean thickness increased, as shown in Fig. 8b and d.
Notably, the smallest powder delivery track distance (1.0 mm) re-
sulted in a large maximum powder thickness (87.52 μm) after com-
pression, as shown in Fig. 8c, which could cause insufficient powder
melting [31]. The biggest powder delivery track distance (2.0 mm) led
to the lowest powder coverage percentage (80.92%) after compression,
as indicated in Fig. 8e, which could cause porosity in the fused samples.
A good balance was achieved when the powder track distance was set at
1.5 mm. The standard deviation of the mean powder thickness with the
1.5 mm track distance before and after compression was smaller than
that with a 1.0 mm or 2.0 mm track distance, signifying a more uniform
powder thickness distribution.
3.3. Laser melted square components and deformation characteristics
After fabrication of the 15 mm × 15 mm square samples, we ob-
served serious top surface deformation of the uncompressed powder
samples along the ultrasonic powder deposition direction, as shown in
Fig. 9a1 and a2. The ultrasonic powder dispenser had repeated accel-
eration and deceleration cycles resulting from the low powder feed
C. Wei, et al. Additive Manufacturing 29 (2019) 100818

Fig. 8. a1)–a6) 3D profile mapping results of the ultrasonically deposited single powder layer; the powder thickness distribution 2D pattern is located at the right top
corner of each image. a1) –a3) powder layer in the initial loose condition with track distances of 1.0, 1.5, and 2.0 mm, respectively; a4)–a6) results for the
corresponding powder layers after compression; minimum (b), maximum (c), and mean (d) powder thickness distribution plots; e) powder coverage.

acceleration (100 mm/s2), the small powder deposition volume with a at unit distance, showed an exponential curve distribution, as shown in
zig-zag toolpath, and the short single track length (30 mm) set in this Fig. 9b, which was derived from Eqs. (12)–(14).
experiment. Hence the deposited powder thickness, i.e. powder weight
Vt = V0 + a × t (13)

8
C. Wei, et al. Additive Manufacturing 29 (2019) 100818

Fig. 9. a1) Ultrasonically deposited 15 mm × 15 mm SLM-processed square samples with a 1.5 mm track distance, a2) Multiple layers sample made up of fused loose
powder, b) Plot of deposited powder weight vs moving distance, c) 3D profile mapping of the top surfaces of the SLM-processed square samples.

x = V0 × t + 0.5 × a × t 2 (14) Table 2

Comparison of porosity characteristics of sliced 15 mm × 15 mm SLM samples.
where Vt is the real-time velocity, a is the powder feed acceleration
( ± 100 mm/s2), t is the moving time, x is the moving distance, and V0 is Track Powder Status Sample Mean Relative Standard
Distance Relative Density Value Deviation (%)
the initial velocity.
(mm) Density (%) (%)
The shape of the calculated exponential curve is similar to the ac-
tually deposited and fused powder layers, as presented in Fig. 9a2. The 1.0 mm Compressed 79.982 79.09 1.08
powder layer in the edge area was much thicker than that in the center 77.892
79.398
area; this is confirmed by the 3D profile mapping described in
Loose 68.078 71.27 2.96
Fig. 9c4–c6. 71.79
The standard deviation of the mean thickness of the single powder 73.936
layer indicated that the powder thickness distribution with a 1.5 mm 1.5 mm Compressed 93.059 94.31 1.36
track distance was more uniform than that with track distances of 94.1
95.76
1.0 mm or 2.0 mm. This was further verified by the 3D profile results of
Loose 90.37 90.07 0.65
top surface deformation of the samples with a 100 layer thickness. As 90.51
shown in Fig. 9c4–c6, the minimum deformation (2252.6 μm) was 89.327
found on the sample with a 1.5 mm track distance. The deformation on 2.0 mm Compressed 84.253 83.51 1.16
the latter sample was over 50% lower than that on the samples with a 82.178
84.11
1.0 mm or 2.0 mm track distance (5751.25 μm and 5003.2 μm respec- Loose 82.606 83.00 0.78
tively). 82.502
It is clear that powder compression significantly reduced the oc- 83.899
currence of severe deformation by comparing Fig. 9c1–c3 and c4–c6.
The minimum deformation after compression and melting was still
present on the sample with a 1.5 mm track distance, as shown in mean relative density (90.07%) of the fused sample without powder
Fig. 9c2, from which it was also found that part of the sample thickness compression was observed on the sample with a 1.5 mm track distance.
at the left and right regions was higher than that at the center region, as After powder compression, the corresponding powder relative density
shown in brown and green respectively. increased to 94.31%. As for the samples with a 2.0 mm track distance,
compression had little influence on the relative density. The relative
densities without and with compression were 83.00% and 83.51% re-
3.4. Porosity
spectively. This may be due to the low powder coverage on each layer
mentioned above, and the powder deposition toolpath offset between
Porosity is a typical macro defect in SLM-processed components.
the layers, which allowed the new loose powder to fill the gaps of
The porosity percentage on the cross sections of the sliced samples (3
previously fused powder stripes. As a result, the effect of the external
pieces sliced from each of the six samples described in Section 2.4) was
compression force decreased.
measured using image processing software based on digitally stitched
micrographs data and converted into relative density values. The re-
lative densities are listed in Table 2, and the related mean values are 3.4.2. Effect of track distance on the porosity of the fused samples
plotted in Fig. 10. Most of the previous investigations tended to minimize fused sample
porosity via the laser energy density optimization by adjusting laser
3.4.1. Effect of compression on the relative density of fused samples parameters, including laser power, laser scanning hatch distance, and
By comparing the blue and yellow bars in Fig. 10, it is clear that scanning speed [38,39]. In the present study, we found that the track
compression could increase the powder relative density. The maximum distance between the deposited powder stripes was the main factor

9
C. Wei, et al. Additive Manufacturing 29 (2019) 100818

further improvement was difficult as a result of the low powder cov-

erage percentage with a 2.0 mm track distance (i.e. 73.31% coverage
before compression and 80.92% coverage after compression) as de-
scribed in Fig. 8e.
The lowest porosity was found on the fused compacted powder
sample with a 1.5 mm track distance, and the size of the pores ranged
from 10 μm to 35 μm, as shown in Fig. 11b2. These pores may be at-
tributed to the unmelted powder particles during laser melting or part
of the powder particles on the powder bed carried away by the com-
pressing plate during compression, because of electrostatic interaction
and van der Waals forces between the small particles. Fig. 11d shows a
magnified micrograph of a pore in Fig. 11b2, which stopped the grain
growth and led to cracks along the grain boundary. It may contribute to
3D printed component fracture under high static/dynamic loading or
fatigue failure due to cyclical stress after a long working period
[40–42].

Fig. 10. Plot of the mean relative density of the fused samples. 3.5. Microstructure characteristics

The microstructure on the cross section of the fused compacted

affecting the porosity of the ultrasonic deposition-assisted SLM samples.
powder sample with a 1.5 mm track distance on the x-z plane (i.e.
Fig. 11a1, b1 and c1 show the porosity distribution cross-section
normal to the building direction which is marked with a red arrow) was
micrographs of the laser fused loose powder samples, with 1.0 mm,
further analyzed as shown in Fig. 12, in which ‘fish scale’-like arc
1.5 mm, and 2.0 mm powder deposition track distances, respectively.
geometries, indicated with blue arrows, represent the solidified semi-
Fig. 11a2, b2 and c2 show the porosity distribution cross-section mi-
circular melting pools for every powder layer and partially re-melted
crographs of the related samples comprised of compacted powder
previously fused powder layers [43]. The distance between the arcs
layers.
ranged from 40 μm to 70 μm, similar to the results of a previous study
As shown in Fig. 11a2, the big pores in the sample with a 1.0 mm
carried out by Yusuf [39]. The overlaps between arcs revealed the de-
track distance, around 100 μm in height and 300 μm in length, seriously
sirable fusion of the deposited powder particles and good metallurgical
split the solidified melt pools into isolated regions, which would result
bonding between the powder layers. The orientation of elongated
in significant mechanical performance failure. The 1.0 mm track dis-
austenitic grains was along the building direction, which exhibited the
tance resulted in the poorest powder thickness distribution, as verified
highest thermal gradient during solidification. The grains were en-
by the large standard deviation value described in Fig. 8d. This caused
larged epitaxially, as they stretched over several fused scanning tracks
poor powder surface roughness, (i.e. some parts of the powder layer
in the building orientation [44].
were much thicker than other parts), and complete melting of the thick
Two characteristic types of micro defects, i.e. micropores and un-
parts was hard to achieve under the determined laser energy density.
melted zones at the scanning track binding regions, were observed. A
The melted layer surface roughness and cavities would become worse
spherical pore with a 2 μm diameter was found, as shown in Fig. 13a.
with an increasing number of melted layers.
The shielding gas in the SLM processing chamber and residual gas in-
After reducing the track distance to 2.0 mm, the size of the pores
side the gas atomized powder particles could be trapped in the liquid
dropped to around 30 μm in height and 100 μm in length, as shown in
phase melt pools and formed small bubbles. If these bubbles cannot
Fig. 11c2. The area of the continuously melted melt pools increased, but
escape from the melting pools before the melt pool solidified, small

Fig. 11. Pores on the cross sections of the fused samples in the x–z plane, a1) and a2) present the fused loose and compacted samples with a 1.0 mm track distance
respectively, b1)/b2) and c1)/c2) show those with a 1.5 mm, and 2.0 mm track distance respectively, d) shows a pore in a higher magnification optical micrograph on
the cross-section of a compressed 1.5 mm track distance sample.

10
C. Wei, et al.
Fig. 12. Optical microscope image of the cross-section of the fused compacted
powder sample with a 1.5 mm track distance.
spherical micropores are generated in the grain [45]. Irregular un-
molten defects appeared between the melting pools, as shown in
Fig. 13b. Fig. 13c explains this phenomenon. The overlap of the melting
zone was insufficient and produced this un-melted region. Heat transfer
from the heating effect zone of the melting pools was unable to melt the
powder particles in the interface area fully, and un-melted defects are
thus formed. According to Eq. (15) given by Sun [38], such a problem
could be addressed by increasing the melt pool overlap area, i.e. re-
ducing the laser scanning hatch distance, or enlarging the melting pools
via an increase in the laser energy density, for instance, by using a
higher laser power or a lower scanning speed.
P two fractures were observed, as shown in Fig. 15b–d. Large cavities
Q=
v×h×t
where Q is the power energy density, P is the laser power, v is the galvo
scanner scanning speed, h is the laser scanning hatch distance between
the scanning tracks and t is the powder layer thickness.
3.6. Tensile test results
High porosity in components would lead to lower mechanical per-
formance [26], and the fused loose powder sample usually had rela-
tively higher porosity than the fused compacted powder sample with
the same powder track distance, as discussed in Section 3.4. Here we
prepared and analyzed the tensile test samples made up of compacted
powder layers with a 1.0, 1.5, and 2.0 mm powder track distance, as
their tensile strength should be higher than that of the loose powder
samples.
The tensile test results of the compacted powder samples are shown
in Fig. 14a). Low tensile strengths (118.45 MPa/118.21 MPa) were
observed on the fused compacted powder tensile specimens with a
1.0 mm track distance that had a lowest mean relative density (79.09%
Fig. 13. a) SEM image of the micropore in the solidified molten pool, b) SEM image of the unmelted defect at the scanning tracks binding region, c) schematic
diagram of the unmelted zone in the SLM-processed part.
Additive Manufacturing 29 (2019) 100818
as presented in Table 2). On the other hand, higher tensile strengths
(278.19 MPa/276.66 MPa) were observed on the fused compacted
powder tensile specimens with a 1.5 mm track distance, which had a
mean relative density of 94.31% (see Table 2).
The tensile test results of the soft blade assisted powder deposition
samples processed on the same experimental equipment were also poor
(around 400 MPa), as shown in Fig. 14b, compared with the results
described in a previous work using the commercial SLM system, where
the tensile strength was usually higher than 600 MPa [46]. This is due
to the technical limitations and un-optimized process parameters of the
experimental system used for this study. However, by comparing
Fig. 14a and b, one can note that the tensile strength of the conven-
tional soft blade deposition fabricated sample was still higher than that
of the ultrasonically deposited samples. This is because the ultra-
sonically deposited maximum powder layer (78.83–87.52 μm as shown
in Table 1) was thicker than the powder layer deposited by the soft
blade (50 μm set in this experiment). High powder thickness would lead
to insufficient laser energy input, which results in incomplete melting of
the metal particles and, in turn, high part porosity and low tensile
strength. Therefore, increasing the laser power energy density [47] or
minimizing the ultrasonically dispensed powder layer thickness could
be considered.
The tensile fracture surface of the fused compacted powder speci-
mens with a 1.0 mm and 1.5 mm track distance was analyzed using
SEM. Dimples were the main fracture surface characteristics observed,
as shown in Fig. 15a, indicating a ductile fracture. During the tensile
test, micropores at the position of the secondary phase particles gath-
ered together via material slipping to form larger cavities, leading to
dimple rupture [48]. On the tensile specimen with a 1.0 mm track
distance, the dimple rupture, fibrous fracture, and a hybrid of the above
were also found around the fracture positions.
(15)
4. Application
This investigation addressed the high porosity issue caused by ul-
trasonic dispensing during multiple material SLM. Following this work,
the ‘ultrasonic powder deposition + external force assisted powder
compression’ concept described in this paper, and a further upgraded
experimental setup were successfully applied to fabricate the multiple
metallic material components and functionally graded material (FGM)
structure components.
Fig. 16 demonstrates a necklace with a Cu10Sn golden fish pendant
on a 316 L fishhook.
Our recent work [49] contains a detailed description of the fabri-
cation of horizontal and 3D functionally graded 316 L/Cu10Sn com-
ponents using such technology.
11
C. Wei, et al. Additive Manufacturing 29 (2019) 100818

Fig. 14. Tensile stress plots of the tensile test samples deposited with a) the ultrasonic dispenser, b) via conventional soft blade powder bed.

5. Conclusion

This paper demonstrates the basic characteristics of a new SLM

process based on ultrasonic point-by-point powder delivery. A powder
compression approach was used to increase the packing density of
powder layers deposited by the ultrasonic dry powder dispensing
technology. This technology has promising applications in the fabri-
cation of multiple metal components that have different physical per-
formances at specific positions on a single component unit.
The most uniformly compressed powder layer was revealed at the
1.5 mm powder deposition track distance, which had a mean powder
thickness of 44.83 μm. The single-track powder deposition result was
consistent with the related DEM simulation result.
The experiment on laser melted square samples indicated that the
powder compression approach could effectively increase powder-
packing density and prevented severe deformation of the sample. The
lowest sample relative density (71.27%) was found in the fused loose
powder sample with a 1.0 mm powder track distance, due to in- Fig. 16. Necklace with Cu10Sn golden fish pendant on a 316 L fishhook.
sufficient power energy density and low powder packing density. The
highest sample relative density (94.31%) was found in the fused

Fig. 15. a) SEM morphology image of dimple fracture surface of the fused compacted powder tensile specimen with a 1.5 mm track distance, b) to d) show the dimple
rupture, fibrous fracture, and a hybrid of the two fracture morphologies of the fused compacted powder tensile specimen with a 1.0 mm track distance, respectively.

12
C. Wei, et al.
compacted powder sample with a 1.5 mm track distance, because its
powder layer thickness distribution was more uniform in comparison to
the others. Two typical micro defects, i.e. micropores and un-melted
zones at the scanning track binding regions, were observed on this fused
compacted powder sample.
Low tensile strengths (118.45 MPa/118.21 MPa) and relatively
higher tensile strengths (278.19 MPa/276.66 MPa) were observed on
the fused compacted powder tensile specimens with 1.0 mm and
1.5 mm track distances respectively. This showed that high porosity
leads to low mechanical performance. The SEM morphology images
showed a dimple rupture on the fused compacted powder tensile spe-
cimen with a 1.5 mm track distance. For the fused compacted powder
specimen with a 1.0 mm track distance, both fibrous fracture and
dimple rupture were found.
The powder compression process is an effective way to increase the
packing density of ultrasonically deposited powder layers, reduce
component distortion and improve its mechanical performance.
However, it was noticed that the ultrasonically deposited powder layers
were still thicker than the powder layers spread by the easily con-
trollable powder blade, regardless of whether the powder layer was
compressed or not. Hence, further research is critical to reduce the
ultrasonically deposited powder layer thickness. Areas for improvement
include: development of a powder deposition nozzle with a large
length-width ratio nozzle mouth shape to increase the deposited
powder coverage area; optimization of the vibration force output by the
piezo transducer, and of the distance between the dispenser nozzle and
the processing substrate to reduce the powder flow rate.
Although ultrasonic vibration powder feeding is feasible for fine
powder particle deposition, its powder dispensing rate is very low
compared with conventional powder bed spreading methods using a
blade. The low powder deposition rate is a significant barrier to its
potential industrial applications, and will thus be further investigated to
improve deposition efficiency.
Declaration of Competing Interest
No Conflict of interest.
Acknowledgement
The University of Manchester is gratefully acknowledged for the
PhD scholarships awarded to Mr. Chao Wei and Mr. Yuan-Hui Chueh.
Finally, Mr. Chao Wei wants to use this last publication finished in his
PhD period to express sincere gratitude to his wife, Ms. Yun Xu, for her
strong support in the past years, and wish her a happy birthday on 20th
August.
References
[1] M. Xia, D. Gu, G. Yu, D. Dai, H. Chen, Q. Shi, Influence of hatch spacing on heat and
mass transfer, thermodynamics and laser processability during additive manu-
facturing of Inconel 718 alloy, Int. J. Mach. Tools Manuf. 109 (2016) 147–157,
https://doi.org/10.1016/j.ijmachtools.2016.07.010.
[2] C.Y. Yap, C.K. Chua, Z.L. Dong, Z.H. Liu, D.Q. Zhang, L.E. Loh, S.L. Sing, Review of
selective laser melting: materials and applications, Appl. Phys. Rev. 2 (2015)
041101, , https://doi.org/10.1063/1.4935926.
[3] A.G. Demir, B. Previtali, Multi-material selective laser melting of Fe/Al-12Si com-
ponents, Manuf. Lett. 11 (2017) 8–11, https://doi.org/10.1016/j.mfglet.2017.01.
002.
[4] C. Wei, Z. Sun, Y. Huang, L. Li, Embedding anti-counterfeiting features in metallic
components via multiple material additive manufacturing, Addit. Manuf. 24 (2018)
1–12, https://doi.org/10.1016/j.addma.2018.09.003.
[5] I. Gibson, D.W. Rosen, B. Stucker, Additive Manufacturing Technologies, (2010),
https://doi.org/10.1007/978-1-4419-1120-9.
[6] M. Vaezi, S. Chianrabutra, B. Mellor, Y. Shoufeng, Multiple material additive
manufacturing – part 1 : a review, Virtual Phys. Prototyp. 8 (2013) 19–50, https://
doi.org/10.1080/17452759.2013.778175.
[7] F. Löffelmann, J. Šplíchal, Design study of the heat switch base plate with single and
multi-material topology, Res. Educ. Aircr. Des. (2018), pp. 1–12.
[8] M. Akiba, M. Araki, K. Yokoyama, Thermal cycling experiments of monoblock di-
vertor modules for fusion experimental reactors AU - Suzuki, S. Fusion Technol.
13
Additive Manufacturing 29 (2019) 100818
Taylor & Francis, 1992, pp. 1858–1862, https://doi.org/10.13182/FST92-A29989.
[9] T. Stichel, T. Laumer, T. Linnenweber, P. Amend, S. Roth, Mass flow character-
ization of selective deposition of polymer powders with vibrating nozzles for laser
beam melting of multi-material components, Phys. Procedia 83 (2016) 947–953,
https://doi.org/10.1016/j.phpro.2016.08.099.
[10] T. Stichel, T. Laumer, T. Baumüller, P. Amend, S. Roth, Powder layer preparation
using vibration-controlled capillary steel nozzles for additive manufacturing, Phys.
Procedia 56 (2014) 157–166, https://doi.org/10.1016/j.phpro.2014.08.158.
[11] S. MATSUSAKA, K. YAMAMOTO, H. MASUDA, Micro-feedingof a fine powder using
a vibrating capillarytube, Adv. Powder Technol. 7 (1996) 141–151.
[12] X. Chen, K. Seyfang, H. Steckel, Development of a micro-dosing system for fine
powder using a vibrating capillary. Part 2. The implementation of a process ana-
lytical technology tool in a closed-loop dosing system, Int. J. Pharm. 433 (2012)
42–50, https://doi.org/10.1016/j.ijpharm.2012.04.072.
[13] C. Wei, L. Li, X. Zhang, Y.H. Chueh, 3D printing of multiple metallic materials via
modified selective laser melting, CIRP Ann. 67 (2018) 245–248, https://doi.org/10.
1016/j.cirp.2018.04.096.
[14] X. Lu, S. Yang, J.R.G. Evans, Microfeeding with different ultrasonic nozzle designs,
Ultrasonics 49 (2009) 514–521, https://doi.org/10.1016/j.ultras.2009.01.003.
[15] S. Yang, Q. Mary, Apparatus and Method for Dispensing Powders, US 2012/
0145806 A1, (2012).
[16] O.M. Al-Jamal, Fabrication of CuH13 Tool Steel Parts Using Selective Laser Melting,
The University of Manchester, 2007.
[17] A. Kinnear, I. Yadroitsev, I. Yadroitsava, Multi material powder delivering systems
for selective laser melting, Interim 14 (2015) 11–23.
[18] O. Helen, F.B. Eng, Additive Manufacturing of Aluminium-metal Matrix Composite
Developed through Mechanical Alloying, (2016).
[19] X. Zhang, C. Wei, Y.-H. Chueh, L. Li, An integrated dual ultrasonic selective powder
dispensing and powder bed spreading platform for 3D printing of multiple material
metal/glass objects in selective laser melting, J. Manuf. Sci. Eng. 141 (1) (2018),
https://doi.org/10.1115/1.4041427 011003 In this issue.
[20] M. Ma, Z. Wang, M. Gao, X. Zeng, Layer thickness dependence of performance in
high-power selective laser melting of 1Cr18Ni9Ti stainless steel, J. Mater. Process.
Technol. 215 (2015) 142–150, https://doi.org/10.1016/j.jmatprotec.2014.07.034.
[21] J.-P. Choi, G.-H. Shin, H.-S. Lee, D.-Y. Yang, S. Yang, C.-W. Lee, M. Brochu, J.-H. Yu,
Evaluation of powder layer density for the selective laser melting (SLM) process,
Mater. Trans. 58 (2017) 294–297, https://doi.org/10.2320/matertrans.M2016364.
[22] A. Amado, M. Schmid, G. Levy, K. Wegener, Advances in SLS powder character-
ization, Int. Conf. Solid Free. Fabr. (2011), pp. 438–452.
[23] G. Jacob, C.U. Brown, A. Donmez, The influence of spreading metal powders with
different particle size distributions on the powder bed density in laser-based powder
bed fusion processes, Adv. Manuf. Ser. (NIST AMS) (2018) 100–117, https://doi.
org/10.6028/NIST.AMS.100-17.
[24] B. Liu, R. Wildman, C.R. Tuck, I. Ashcroft, R.J.M. Hague, Investigation the Effect of
Particle Size Distribution on Processing Parameters Optimisation in Selective Laser
Melting Process, (2011), pp. 227–238.
[25] C. Wei, Y.-H. Chueh, X. Zhang, Y. Huang, Q. Chen, L. Li, Easy-To-Remove composite
support material and procedure in additive manufacturing of metallic components
using multiple material laser-based powder bed fusion, J. Manuf. Sci. Eng. 141
(2019) 1–18, https://doi.org/10.1115/1.4043536.
[26] A. Ahmadi, R. Mirzaeifar, N. Shayesteh, A. Sadi, H.E. Karaca, M. Elahinia, Effect of
manufacturing parameters on mechanical properties of 316L stainless steel parts
fabricated by selective laser melting: a computational framework, JMADE 112
(2016) 328–338, https://doi.org/10.1016/j.matdes.2016.09.043.
[27] F. Mahmoodi, Compression Mechanics of Powders and Granular Materials Probed
by Force Distributions and a Micromechanically Based Compaction Equation,
Uppsala University, 2012.
[28] P.J. Denny, Compaction equations: a comparison of the Heckel and Kawakita
equations, Powder Technol. 127 (2002) 162–172, https://doi.org/10.1016/S0032-
5910(02)00111-0.
[29] K.K, Y. Tsutsumi, An empirical equation of state for powder compression, J. Appl.
Phys. 4 (1965) 625A http://stacks.iop.org/1347-4065/4/i=8/a=625A.
[30] U. Ali, Y. Mahmoodkhani, S. Imani Shahabad, R. Esmaeilizadeh, F. Liravi,
E. Sheydaeian, K.Y. Huang, E. Marzbanrad, M. Vlasea, E. Toyserkani, On the
measurement of relative powder-bed compaction density in powder-bed additive
manufacturing processes, Mater. Des. 155 (2018) 495–501, https://doi.org/10.
1016/j.matdes.2018.06.030.
[31] X. Shi, S. Ma, C. Liu, C. Chen, Q. Wu, X. Chen, J. Lu, Performance of high layer
thickness in selective laser melting of Ti6Al4V, Materials (Basel) 9 (2016) 1–15,
https://doi.org/10.3390/ma9120975.
[32] I. Media Cybernetics, Image-Pro® Plus Reference Guide, (2009) ftp://ftp.mediacy.
com/tech/PDF/IPPReference_7.0.pdf.
[33] Z. Ihsani, Porosity Analysis Procedure with ImageJ, Academia. (n.d.). https://www.
academia.edu/24152953/Porosity_Analysis_Procedure_with_ImageJ.
[34] Z. Xiang, M. Yin, Z. Deng, X. Mei, G. Yin, Simulation of forming process of powder
bed for additive manufacturing, J. Manuf. Sci. Eng. 138 (2016) 81002–81009,
https://doi.org/10.1115/1.4032970.
[35] Z. Wang, W. Yan, W.K. Liu, M. Liu, Powder-scale multi-physics modeling of multi-
layer multi-track selective laser melting with sharp interface capturing method,
Comput. Mech. 63 (4) (2018) 649–661, https://doi.org/10.1007/s00466-018-
1614-5 In this issue.
[36] K.L. Johnson, K. Kendall, A.D. Roberts, Surface energy and the contact of elastic
solids, Proc. R. Soc. Lond. A Math. Phys. Sci. 324 (301) (1971) LP – 313 http://rspa.
royalsocietypublishing.org/content/324/1558/301.abstract.
[37] F.C. Santos, V. Soares, A.C. Tort, A note on the conservation of mechanical energy
and the Galilean principle of relativity, Eur. J. Phys. 31 (2010) 827 http://stacks.
C. Wei, et al.
iop.org/0143-0807/31/i=4/a=012.
[38] Z. Sun, X. Tan, S.B. Tor, W.Y. Yeong, Selective laser melting of stainless steel 316L
with low porosity and high build rates, Mater. Des. 104 (2016) 197–204, https://
doi.org/10.1016/j.matdes.2016.05.035.
[39] S. Yusuf, Y. Chen, R. Boardman, S. Yang, N. Gao, Investigation on porosity and
microhardness of 316L stainless steel fabricated by selective laser melting, Metals
(Basel) 7 (2017) 64, https://doi.org/10.3390/met7020064.
[40] H. Gong, H.K. Rafi, T. Starr, R. Stucker, Effect of Defects on Fatigue Tests of as-built
TI-6AL-4V Parts Fabricated by Selective Laser Melting, SFF 2012, (2012), pp.
499–506.
[41] K. Rekedal, D. Liu, Fatigue life of selective laser melted and Hot isostatically pressed
Ti-6Al-4v absent of surface machining, 56th AIAA/ASCE/AHS/ASC Struct. Struct.
Dyn. Mater. Conf., American Institute of Aeronautics and Astronautics, (2015)
doi:doi:10.2514/6.2015-0894.
[42] H. Kim, Z. Liu, W. Cong, H. Zhang, Additively-Manufactured AISI 4140 Low Alloy
Steel, (2017), https://doi.org/10.3390/ma10111283.
[43] L. Thijs, K. Kempen, J.-P. Kruth, J. Van Humbeeck, Fine-structured aluminium
products with controllable texture by selective laser melting of pre-alloyed
AlSi10Mg powder, Acta Mater. 61 (2013) 1809–1819, https://doi.org/10.1016/j.
actamat.2012.11.052.
14
Additive Manufacturing 29 (2019) 100818
[44] J. Čapek, M. Machová, M. Fousová, J. Kubásek, D. Vojtěch, J. Fojt, E. Jablonská,
J. Lipov, T. Ruml, Highly porous, low elastic modulus 316L stainless steel scaffold
prepared by selective laser melting, Mater. Sci. Eng. C 69 (2016) 631–639, https://
doi.org/10.1016/j.msec.2016.07.027.
[45] S. Wang, Research on High Layer Thickness Fabricated of 316L by Selective Laser
Melting, (2017), https://doi.org/10.3390/ma10091055.
[46] J. Luo, X. Jia, R. Gu, P. Zhou, Y. Huang, J. Sun, M. Yan, 316L stainless steel
manufactured by selective laser melting and its biocompatibility with or without
hydroxyapatite coating, Metals (Basel) 8 (2018) 548, https://doi.org/10.3390/
met8070548.
[47] H.N. Li, K. Ge Xie, B. Wu, W.Q. Zhu, Generation of textured diamond abrasive tools
by continuous-wave CO2 laser: laser parameter effects and optimisation, J. Mater.
Process. Technol. 275 (2020) 116279, , https://doi.org/10.1016/J.JMATPROTEC.
2019.116279.
[48] J. Chen, Y. Madi, T.F. Morgeneyer, J. Besson, Plastic flow and ductile rupture of a
2198 Al-Cu-Li aluminum alloy, Comput. Mater. Sci. 50 (2011) 1365–1371, https://
doi.org/10.1016/j.commatsci.2010.06.029.
[49] C. Wei, Z. Sun, Q. Chen, Z. Liu, L. Li, Additive manufacturing of horizontal and 3D
functionally graded 316L/Cu10Sn components via multiple material selective laser
melting, J. Manuf. Sci. Eng. 141 (2019) 1–14, https://doi.org/10.1115/1.4043983.