materials

Article
Influence of Melt-Pool Stability in 3D Printing of
NdFeB Magnets on Density and Magnetic Properties
Mateusz Skalon 1 , Michael Görtler 2 , Benjamin Meier 1,2 , Siegfried Arneitz 1, *, Nikolaus Urban 3 ,
Stefan Mitsche 4 , Christian Huber 5 , Joerg Franke 3 and Christof Sommitsch 1
1 IMAT Institute of Materials Science, Joining and Forming, Graz University of Technology,
Kopernikusgasse 24, 8010 Graz, Austria; mateusz.skalon@tugraz.at (M.S.);
benjamin.meier@joanneum.at (B.M.); christof.sommitsch@tugraz.at (C.S.)
2 Joanneum Research, Materials—Institute for Laser and Plasma Technology, Leobner Straße 94,
8712 Niklasdorf, Austria; michael.goertler@joanneum.at
3 Institute for Factory Automation and Production Systems, University of Erlangen-Nürnberg,
Fuerther Strasse 246b, 90429 Nuremberg, Germany; Nikolaus.Urban@faps.fau.de (N.U.);
joerg.franke@faps.fau.de (J.F.)
4 Institut für Elektronenmikroskopie und Nanoanalytik, Steyrergasse 17/III, 8010 Graz, Austria;
smitsche@tugraz.at
5 Physics of Functional Materials, University of Vienna, 1090 Vienna, Austria; huber-c@univie.ac.at
* Correspondence: siegfried.arneitz@tugraz.at




Received: 1 December 2019; Accepted: 27 December 2019; Published: 29 December 2019


Abstract: The current work presents the results of an investigation focused on the influence of process
parameters on the melt-track stability and its consequence to the sample density printed out of
NdFeB powder. Commercially available powder of Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5 alloy was
investigated at the angle of application in selective laser melting of permanent magnets. Using single
track printing the stability of the melt pool was investigated under changing process parameters.
The influence of changing laser power, scanning speed, and powder layer thickness on density,
porosity structure, microstructure, phase composition, and magnetic properties were investigated.
The results showed that energy density coupled with powder layer thickness plays a crucial role
in melt-track stability. It was possible to manufacture magnets of both high relative density and
high magnetic properties. Magnetization tests showed a significant correlation between the shape
of the demagnetization curve and the layer height. While small layer heights are beneficial for
sufficient magnetic properties, the remaining main parameters tend to affect the magnetic properties
less. A quasi-linear correlation between the layer height and the magnetic properties remanence (Jr ),
coercivity (HcJ ) and maximum energy product ((BH)max ) was found.

Keywords: NdFeB; selective laser melting; additive manufacturing; melt track; powder thickness

1. Introduction
NdFeB permanent magnets attract great interest due to their extraordinary performance in
comparison to other types of hard magnets. Their energy density (e.g., magnetic polarization Jr higher
than 1 Tesla, coercivity HcJ above 1000 kA m−1 , and figure of merit (BH)max greater than 500 kJ·m−3 )
outperforms other types of permanent magnets at room temperature. Due to this characteristic,
NdFeB magnets are considered as key components for clean energy application [1]. They are also
classified as potentially the best candidates for the miniaturization of electric devices or highly efficient
electric motors [2,3]. Currently applied techniques, such as sintering, extrusion, or spark plasma
sintering (SPS), deliver magnets with excellent magnetic properties and of high relative density [4–14].
Traditional manufacturing, however, limits the application of the magnets due to limits in design and

Materials 2020, 13, 139; doi:10.3390/ma13010139 www.mdpi.com/journal/materials

Materials 2020, 13, 139 2 of 14

specific tooling requirements for each design and limitations in shape flexibility and complexity [15,16].
With the technique of additive manufacturing it is possible to produce parts with more complex
design; however, obtaining printing processes for functional materials is still a topic in research and
development. Earlier research already showed that techniques such as slurry jetting or laser powder
bed fusion (LPBF) proved to be capable of delivering net-shape magnets [17–22]. Slurry jetting has
been shown
Materialsto be12,capable
2019, x FOR PEER producing magnets with good magnetic properties (HcJ2 =
ofREVIEW 775.3 kA/m,
of 17

Jr = 0.478 T, and (BH)max = 39.78 kJ/m ). These values reach nearly half of the values expected for the
3
and complexity [15,16]. With the technique of additive manufacturing it is possible to produce parts
bulk material.
with more This manufacturing
complex approach
design; however, obtainingallows theprocesses
printing manufacture of the desired
for functional materials magnet
is still a in a time-
and cost-efficient
topic in researchmanner, however, and
and development. theresearch
Earlier resulting magnetic
already showedproperties willsuch
that techniques always be lower than
as slurry
the bulkjetting
magnets,or laserdue
powder bedpresence
to the fusion (LPBF)
of aproved to be capable
significant quantity of delivering net-shape
of bonding magnets
resin. Another[17– approach
22]. Slurry jetting has been shown to be capable of producing magnets with good magnetic properties
was presented by Kolb et al. [17,18] who utilized laser powder based fusion (LPBF) technique in
(HcJ = 775.3 kA/m, Jr = 0.478 T, and (BH)max = 39.78 kJ/m3). These values reach nearly half of the values
order toexpected
overcome the problem of low relative density. They managed to 3D print cubical magnets
for the bulk material. This manufacturing approach allows the manufacture of the desired
of 85% magnet
of relative density, which were characterized by a polarization
in a time- and cost-efficient manner, however, and the resulting magnetic of 0.514 T. Thewill
properties surprisingly
lower than expected
always be lowermagnetic properties
than the bulk were
magnets, due to assigned
the presencetoofthe self-demagnetizing
a significant of theresin.
quantity of bonding samples while
measured Another
in anapproach was presented
open magnetic circuitbyinside
Kolb etaal. [17,18] whocoil.
Helmholtz utilized laserresearchers
Other powder based[19]
fusion (LPBF)
also applied a laser
technique in order to overcome the problem of low relative density. They managed to 3D print cubical
powder based fusion (LPBF) process to the powdered NdFeB (Nd7.5Pr0.7Zr2.6Ti2.5-Co2.5Fe75B8.8)
magnets of 85% of relative density, which were characterized by a polarization of 0.514 T. The
powder.surprisingly
The 3D printed lower than magnet reached
expected magneticroughly
properties60%wereof a bulktocounterpart
assigned at relative
the self-demagnetizing density 92%
of the
(Hc = 695 kA/m,
samples while J = 0.59 T, and (BH)max = 45 kJ/m 3 ). The coercivity of the LPBF-printed samples can
r measured in an open magnetic circuit inside a Helmholtz coil. Other researchers [19]
also applied increased
be also substantially a laser powder
by a grainbased fusion (LPBF)
boundary process
infiltration methodto the[22].powdered
The mainNdFeB conclusion was
(Nd7.5Pr0.7Zr2.6Ti2.5-Co2.5Fe75B8.8) powder. The 3D printed magnet reached roughly 60% of a
to increase the rare elements quantities in order to achieve further progress in 3D printed magnets.
bulk counterpart at relative density 92% (Hc = 695 kA/m, Jr=0.59 T, and (BH)max = 45 kJ/m3). The
The main goal of the present investigation was fine tuning of the printing parameters in order to
coercivity of the LPBF-printed samples can be also substantially increased by a grain boundary
understand the melt-pool
infiltration method [22].behavior under range
The main conclusion ofincrease
was to parameters
the rareand its influence
elements onorder
quantities in the density
to and
magnetic properties.
achieve further progress in 3D printed magnets. The main goal of the present investigation was fine
tuning of the printing parameters in order to understand the melt-pool behavior under range of
parameters
2. Materials and its influence on the density and magnetic properties.
and Methods
A spherical
2. Materialspowder of Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5 (at.%) alloy manufactured by Neo
and methods
Magnequench company (Singapore, Singapore) was used for the
A spherical powder of Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5 experiment
(at.%) (Figureby1).Neo
alloy manufactured The powder
particle Magnequench
size distribution was(Singapore,
company presented in Figure
Singapore) 1a.
was used for the experiment (Figure 1). The powder
particle size distribution was presented in Figure 1a.

1. (a)1. Particle
Figure Figure (a) Particlesize
size distribution of the
distribution of the powder
powder particlesparticles
shown as shown as aofpercentage
a percentage the total of the
volume and cumulative with respect to increasing particle size. (b) SEM
total volume and cumulative with respect to increasing particle size. (b) SEM image of the image of the
Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5 powder used in this work.
Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5 powder used in this work.
The single melt tracks were manufactured by a method similar to that as presented in
The[7,17,18,20].
single melt tracksauthors
Previous were manufactured
manufactured single by atracks
method similara layer
by treating to that
of as presented
powder in [7,17,18,20].
of constant
Previous authors
depth with amanufactured
laser beam with single
changingtracks
laser by treatingThis
parameters. a layer
time, of powder
however, of constant
single tracks weredepth with
printedwith
a laser beam usingchanging
a specially laser
designed powder holder,
parameters. Thiswhich
time,was placed insingle
however, the printing
trackschamber as a
were printed using
base plate. The holder was made of ferritic stainless steel 1.4301 and consisted of
a specially designed powder holder, which was placed in the printing chamber as a base plate. Thethe slope plane with
a total length of 50 mm and the linearly increasing powder thickness from 0 μm up to 125 μm. The
holder was made of ferritic stainless steel 1.4301 and consisted of the slope plane with a total length
slope angle was equal to 0.0172°. The shape deviation of the plate was: −0.2 μm and +0.7 μm at the
of 50 mm and the
shallowest linearly
and deepest increasing powder
points, respectively thickness
(Figure 2). The from
powder0 layer
µm upwasto 125 µm.
equalized The slope angle
manually.
was equal to 0.0172 . The shape deviation of the plate was: −0.2 µm and +0.7 µm at the shallowest
◦
Materials 2020, 13, 139 3 of 14

and deepest points,

Materials 2019, respectively
12, x FOR PEER REVIEW(Figure 2). The powder layer was equalized manually. 3The of 17single
lines were printed perpendicularly to the slope with intervals of 500 µm using the parameter sets
The in
as listed single
Table lines were approach
1. This printed perpendicularly to the slope with
allowed the investigation intervals
of the powder of depth
500 μm using theon the
influence
parameter sets as listed in Table 1. This approach allowed the investigation of the powder depth
melt-track behavior.
Materials 2019, 12, x FOR PEER REVIEW 3 of 17
influence on the melt-track behavior.
The single lines were printed perpendicularly to the slope with intervals of 500 μm using the
parameter sets as listed in Table 1. This approach allowed the investigation of the powder depth
influence on the melt-track behavior.

Figure2.2. Schematics
Figure Schematics ofofthe
thetest
test stand.
stand.

Table Description
1. 1.
Table Descriptionof
of parameters setsused
parameters sets usedforfor experiment.
experiment.
Figure 2. Schematics of the test stand.
Parameters Set Power Laser Laser Speed Hatching
Speed Hatching LinearEnergy
Linear Energy
Parameters Set Code Power (W) Beam
Beam
Code (W) (mm/s)
(mm/s) (um)
(um) (J/mm)
(J/mm)
Table 1. Description of parameters sets used for experiment.
I I 40 40 200200 500
500 defocussed
defocussed 0.200
0.200
Power Laser Speed Hatching Linear Energy
Parameters II Set Code 40 160 500 defocussed
Beam 0.250
II 40 (W) 160
(mm/s) 500
(um) defocussed 0.250
(J/mm)
III 60 200 500 defocussed 0.300
IIIIV I 60 6040 200
200160 500
500
500
defocussed
defocussed
defocussed
0.200
0.300
0.375
II 40 160 500 defocussed 0.250
IV 60 160 500 defocussed 0.375
III 60 200 500 defocussed 0.300
The laser beam was set as defocussed in order to reproduce the beam as it is when printing the
IV 60 160 500 defocussed 0.375
bulk part of the component (hatching). The laser beam spot size was equal to 80 μm. The single tracks
The laser beam was set as defocussed in order to reproduce the beam as it
were cut perpendicularly in respect to their length and evaluated according to the scheme presented
is when printing the
bulk in
part The
of the laser beam was set as defocussed in order to reproduce the beam as it is when printing the
Figure 3. component (hatching). The laser beam spot size was equal to 80 µm. The single tracks
bulk part of the component (hatching). The laser beam spot size was equal to 80 μm. The single tracks
were cut perpendicularly in respect to their length and evaluated according to the scheme presented in
were cut perpendicularly in respect to their length and evaluated according to the scheme presented
Figure 3.
in Figure 3.

Figure 3. Scheme of single track cross-section (STCS) measurement: (a) Base length (L), height (h),
contact points (X, Y), and the top point (Z); (b) contact angles η1 and η2.
Figure Figure 3. Scheme
3. Scheme of single
of single track
track cross-section (STCS)
cross-section (STCS) measurement:
measurement: (a) Base length
(a) Base (L), height
length (h),
(L), height (h),
The radius
contact was(X,calculated
points Y), and the as
topapoint
radius
(Z); of
(b)acontact
circleangles
described
η1 andon
η2. a triangle defined by the most
contact points (X, Y), and the top point (Z); (b) contact angles η1 and η2 .
distant points of length and height (X, Y, Z in Figure 3a). Contact angle η was derived as an average
The
valueradius
of η1radius was calculatedinasFigure
and calculated
η2 presented a radius ofThe
a circle described on a triangle defined by the 3most
The was as a radius of3b. a circlebulk cubic samples
described of size
on a triangle 5 × 5 ×by
defined 5 the
mm were
most distant
manufactured using Farsoon FS121M (Farsoon Europe, Suttgart, Germany), the same 3Daverage
distant points of length and height (X, Y, Z in Figure 3a). Contact angle η was derived as an printing
points of length
valuealso and
of η1 height (X,
and η2protectiveY, Z in Figure
presentedatmosphere
in Figure 3b.3a). Contact
The bulk angle η was derived as an average
5 mm3 were of η1
value
system in under of argon. Thecubic samples
powder layerof size 5 ×was
thickness
3
5 × incrementally
and η2 manufactured
presented in Figure 3b. TheFS121M
using Farsoon bulk cubic samples
(Farsoon of size 5 × 5Germany),
× 5 mm the were manufactured
3D printing using
increased with step of 20 μm. The density ofEurope,
the bulkSuttgart,
samples was measured same
using water
Farsoon FS121M
system (Farsoon
also in Europe,
under protective Suttgart,
atmosphere Germany),
of argon. The the same
powder 3D
layer printing
thickness
displacement method (Archimedes method). Microstructures were revealed by etching polished
wassystem also in under
incrementally
increased
protective with step
atmosphere of of 20 μm.
argon. The The density
powder layer of thickness
the bulk was
samples was measured
incrementally using water
increased with step of
surfaces with 0.5 % Nital for 1 second. Grain size distribution was calculated using an algorithm
20 µm. displacement
The density method
of theand (Archimedes
bulk method).
samples [23]. Microstructures
wasEnergy-dispersive
measured using water were revealed
displacementby etching
method polished
(Archimedes
presented by Rabbani Ayatollahi X-ray spectroscopy (EDX spectroscopy
surfaces with 0.5 % Nital for 1 second. Grain size distribution was calculated using an algorithm
method). Microstructures
or EDXS) were revealed
was used for evaluation by etching
of the chemical polishedX-ray
composition. surfaces with 0.5%
diffraction Nital forwere
measurements 1 second.
presented by Rabbani and Ayatollahi [23]. Energy-dispersive X-ray spectroscopy (EDX spectroscopy
Grain size distribution was calculated using an algorithm presented by Rabbani
or EDXS) was used for evaluation of the chemical composition. X-ray diffraction measurements were and Ayatollahi [23].
Energy-dispersive X-ray spectroscopy (EDX spectroscopy or EDXS) was used for evaluation of the
Materials 2020, 13, 139 4 of 14

chemical composition. X-ray diffraction measurements were performed using a Siemens D5005
diffractometer (Siemens, Munich, Germany). For magnetic characterization the specimens were
magnetized using an impulse magnetizing device IM-K-008020-AD-DT-C11A1H3A2A1 from Magnet
Physik (Dr. Steingroever GmbH, Cologne, Germany) with a magnetizing voltage of 1600 V. The voltage
generated a flux density over 3 T inside the magnetizing coil that led to full saturation of NdFeB while
keeping the thermal stress of the equipment within an acceptable range. The magnetic measurements
were performed inside a Brockhaus Hystograph HG 200 (Dr. Brockhaus Messtechnik GmbH & Co. KG,
Lüdenscheid, Germany) with a J-compensated measuring coil TJH 10 according to IEC 60404-5. The
measuring temperature was set to room temperature of 22 ◦ C. The measuring procedure consisted of
three phases: First, the saturation of the sample was checked within the first quadrant of the hysteresis
loop. Subsequently the demagnetization curve was passed with constant flux change. After full
demagnetization, the magnetization region in the opposite direction in the third quadrant was entered.
The specimens were magnetized and measured in the building direction only.

3. Results and Discussion

Figure 4 shows that increasing linear energy increased the melt-track stability over the range of
powder layer thicknesses. Application of low linear energy (I and II) resulted in a disturbed melt
track. The greater
Materials 2019, 12,the thickness
x FOR of the powder layer, the more difficult it was to form an 5undisturbed
PEER REVIEW of 17
melt track.

4. Overview
FigureFigure of single tracks formed by parameters sets: I, II, III, and IV at various powder layer
4. Overview of single tracks formed by parameters sets: I, II, III, and IV at various powder
thickness (Lt).
layer thickness (Lt).

A powder melted by a laser on a plane forms a liquid semi-cylinder of a defined length.

According to previous research [24], the stability of this semi-cylinder is dependent on its geometry:
λ: Length of liquid cylinder, θ: Angle between vertical plane and the diameter perpendicular to the
contact angle of a cylinder, D: Diameter (Figure 5). According to analysis performed by Gusarov and
Materials 2020, 13, 139 5 of 14

A powder melted by a laser on a plane forms a liquid semi-cylinder of a defined length. According
to previous research [24], the stability of this semi-cylinder is dependent on its geometry: λ: Length of
liquid cylinder, θ: Angle between vertical plane and the diameter perpendicular to the contact angle of
Materials 2019, 12, x FOR PEER REVIEW 6 of 17
a cylinder, D: Diameter (Figure 5). According to analysis performed by Gusarov and Smurov [24], a
liquid semi-cylinder remains stable only when the following, essential condition is adequately fulfilled:

θπD 2
s
πD (1 +>cos2θ
√ ) − sin2θ (2)
> (1)
λ 2θ(λ2 + cos2θ
3) − 3sin2θ

applicable at θ > 𝜋/2.

Figure 5. (a) Scheme of cylinder measurement and (b) map of liquid cylinder stability, where α: Both
Figure 5. (a) Scheme of cylinder measurement and (b) map of liquid cylinder stability, where α: Both
nondistorted and distorted cylinders are stable, β: Distorted cylinder is stable while nondistorted
nondistorted and distorted cylinders are stable, β: Distorted cylinder is stable while nondistorted
becomes unstable, and γ: Both cylinders are unstable. Based on [24].
becomes unstable, and γ: Both cylinders are unstable. Based on [24].
In the limit of a circular cylinder θ = π gives:
Transgression of boundary condition (Equation (2)) resulted in immediate loss of the cylinder
stability and, furthermore in, balling (e.g., Figure 4:rSet I at 100 μm). Since, in the given experiment,
πD 2
the melt-track length was not measured the stability > conditions (Equations (2)), which are strongly (2)
λ 3
dependent on the melt-track length, will not directly be further considered. However, based on these,
applicable at θ > π/2.
when the stability condition (Equation (2)) was met, the melt track will remain stable independent of
Transgression
the molten of boundary
semi-cylinder lengthcondition (Equation
(Figure 5b), either in(2))nondisturbed
resulted in immediate loss of(meandering)
or in disturbed the cylinder
stability and, furthermore in, balling (e.g., Figure 4: Set I at 100 µm). Since,
form [24]. Since the Θ angle is perpendicular to the radius curvature, Equation (2) can in the given experiment,
be further
the melt-track
expressed as: length was not measured the stability conditions (Equations (2)), which are strongly
dependent on the melt-track length, will not directly be further considered. However, based on these,
when the stability condition (Equation (2)) was η< met,90°the melt track will remain stable independent
(3)
of the molten
where semi-cylinder
η is a contact angle aslength (Figure
indicated 5b), either
in Figure 5a. in nondisturbed or in disturbed (meandering)
form Single
[24]. Since the Θ
tracks (STs) were cut perpendicularly the
angle is perpendicular to radius
to their curvature,
length Equation
and their (2) can (CSs)
cross-sections be further
were
expressed
measured as:as presented in the scheme (Figure 3). The amount of the powder melted by a laser beam
◦
is dependent on the powder stored in a powder η < 90
layer and the energy delivered by the beam. The (3)
measured
where η is avalues
contactwere
anglethus plotted in
as indicated functions
in Figure 5a. of the single track cross-section (STCS) area to
achieve clarity (Figures 3 and 6).
Single tracks (STs) were cut perpendicularly to their length and their cross-sections (CSs) were
measured as presented in the scheme (Figure 3). The amount of the powder melted by a laser beam
is dependent on the powder stored in a powder layer and the energy delivered by the beam. The
measured values were thus plotted in functions of the single track cross-section (STCS) area to achieve
clarity (Figures 3 and 6).
MaterialsMaterials
2020, 13, 2019,
13912, x FOR PEER REVIEW 7 of 17 6 of 14

6. Single
Figure Figure track
6. Single cross-section
track (STCS):
cross-section (STCS): (a)(a) Radius
Radius in function
in function of STCSof STCS
area, area,ratio
(b) aspect (b) (AR
aspect
= ratio
(AR = h/L) in function of STCS area, (c) STCS contact angle in function of its area, (d) percentage
h/L) in function of STCS area, (c) STCS contact angle in function of its area, (d) percentage of STCS of
STCS with
with contact
contact angle < 90
angleη <η 90° in◦function
in function of linear
of linear energy.energy.

As presented
As presented in Figure
in Figure 6a,6a,thethe higherthe
higher the linear
linear energy
energyvalue,
value,thethe
greater the STCS
greater radius.radius.
the STCS A A
large radius is desired since this assures the stability of the melt track. When the linear
large radius is desired since this assures the stability of the melt track. When the linear energy applied energy applied
was too low (set I), the melt-track radius had an approximately linear dependence on its area, which
was too low (set I), the melt-track radius had an approximately linear dependence on its area, which
confirms the observation from Figure 4 of the poor stability of the melt track and a poor connection
confirms thebase
to the observation from
plate. For the set Figure 4 of the
II parameters poorwith
coupled stability of the melt
a decreasing STCS track and
area, the melta poor
track connection
tended to
the base
to plate.
be widerFor
and the set IIsince
flatter, parameters
the radiuscoupled with a(Figure
was increasing decreasing
6a). ThisSTCS area, the
observation melt track
is supported by tended
to be wider and flatter,
data presented since 6b
in Figure thewhere
radius wasratio
aspect increasing
(AR) as (Figure
a function 6a).of This
STCS observation
area was presented. is supported
by dataIndependent
presentedofinthe powder
Figure 6blayer,
wherethe higher
aspectthe linear
ratio energy,
(AR) as athe flatter theofmelt
function STCStracks.
areaThe contact
was presented.
angle between the melt track and the base plate also tended to decrease (improve)
Independent of the powder layer, the higher the linear energy, the flatter the melt tracks. The contact along the increase
of the linear energy value (Figure 6c). The energy density for providing melt-track stability (sets III
angle between the melt track and the base plate also tended to decrease (improve) along the increase of
and IV) was found to remain in agreement with previous findings [17–19]. Further, the cubic samples
the linear energy value (Figure 6c). The energy density for providing melt-track stability (sets III and
were manufactured using parameters sets I–IV with a step-wise change of the powder layer depth
IV) was(step
found toμm)
= 20 remainand in agreement
their density waswith previousItfindings
measured. was found [17–19].
that theFurther,
samplesthe cubic samples
produced using were
manufactured
parametersusing
setsparameters sets I–IV
I and II provided sampleswith so afragile
step-wise change
that they of the
crushed when powder layer
detaching (offdepth (step = 20
the plate
operation). It was, therefore, impossible to perform further activities using
µm) and their density was measured. It was found that the samples produced using parameters sets I sets I and II.
and II provided samples so fragile that they crushed when detaching (off the plate operation). It was,
therefore, impossible to perform further activities using sets I and II.
Density analysis showed that the porosity of produced magnets was composed of two significant
constituents: (1) closed pores and (2) opened pores. The latter, however, consisted of both opened
pores and cracks, which are indistinguishable by the selected method. Minimum value of closed
porosity was found for parameters set III and set IV when the powder layer was set to 40 µm and
60 µm, respectively (Figure 7a,b). When this value was exceeded, then the amount of closed pores
grew steadily along with the increasing thickness of the powder layer. The number of opened pores
decreased steadily when parameters set III was used (Figure 7a) and remained roughly constant when
parameters set IV was applied (Figure 7b). An increase of the closed porosity share, however, cannot
be compensated by decreasing the share of opened pores, and the peak of the relative density was thus
found for parameters sets III and IV when the powder layer thickness was set to 40 µm and 60 µm,
respectively. A maximum density of 90.9% was reached for the samples printed with parameters set IV
when a powder layer of 60 µm was applied. This stays in opposition to other research results [17–19]
that reported similar findings for powder layer thickness of 20 µm. For the set of parameters III the
maximum relative density was 86.5% and was recorded when powder layer thickness was equal to
40 µm. This implies that in order to reach a high relative density the laser beam parameters have
 when parameters set IV was applied (Figure 7b). An increase of the closed porosity share, however,
cannot be compensated by decreasing the share of opened pores, and the peak of the relative density
was thus found for parameters sets III and IV when the powder layer thickness was set to 40 μm and
60 μm, respectively. A maximum density of 90.9 % was reached for the samples printed with
parameters set IV when a powder layer of 60 μm was applied. This stays in opposition to other
Materials 2020, 13, 139 7 of 14
research results [17–19] that reported similar findings for powder layer thickness of 20 μm. For the
set of parameters III the maximum relative density was 86.5 % and was recorded when powder layer
tothickness
be coupledwaswithequalantoadequate
40 μm. This implies
powder thatthickness.
layer in order toThereach a high relative
maximum densitydensity the laser
was limited by the
beam parameters
challenging have to of
brittle behavior be the
coupled with an
alloy, which wasadequate powder
very sensitive to layer thickness.
cracking inducedThe maximum
by inner tensions.
density was
Increasing thelimited by the to
laser energy challenging brittle would
a certain extent behavior in of the alloy,
theory lead towhich was very
increased sensitive of
dimensions to the
cracking induced by inner tensions. Increasing the laser energy to a certain extent would
melt pool, better connection to underlying layers, and, in consequence, to a higher overall density. in theory
leadwas
This to increased
prohibited dimensions
by the factofthat
the samples
melt pool, better connection
produced with a higherto underlying layers,
laser energy and, break
already in
consequence, to a higher overall density. This was prohibited by the fact that samples
during manufacturing, because of the increased inner tensions induced by the shrinking of the biggerproduced with
a higher laser energy already break during manufacturing, because of the increased inner tensions
melt pool during the cool-down phase. This effect has also been mentioned in [18] as a limit for the
induced by the shrinking of the bigger melt pool during the cool-down phase. This effect has also
maximum density. For other materials that are also challenging to process by LPBF, such as ceramics,
been mentioned in [18] as a limit for the maximum density. For other materials that are also
manufacturing at elevated ambient temperature by means of a powder bead heating has been proven
challenging to process by LPBF, such as ceramics, manufacturing at elevated ambient temperature
to be a suitable countermeasure [25].
by means of a powder bead heating has been proven to be a suitable countermeasure [25].

Figure 7. Changes of porosity structure and volume in function of a powder layer thickness: (a) set III
Figure
(b) set IV.7. Changes of porosity structure and volume in function of a powder layer thickness: (a) set III
(b) set IV.
Materials 2019, 12, x FOR PEER REVIEW 9 of 17
Figure 8a shows that there were no visible major shape distortions. Figure 8b highlights both
spots where
Figurefusion wasthat
8a shows lacking
there and
werealso the presence
no visible of cracks.
major shape Observation
distortions. Figure 8b of cracks present
highlights both in
the spots
samplewhere fusionthe
matches wasobservation
lacking and also
made theinpresence of cracks.
a previous workObservation
[19]. Thus, of
ascracks
shown present
in theinimages
the in
sample matches the observation made in a previous work [19]. Thus, as shown in the images
Figure 8, samples could be printed with a high level of precision and without any major defects. in Figure
8, samples could be printed with a high level of precision and without any major defects.

Figure 8. 8.LPBF-printed
Figure LPBF-printed NdFeB sample:(a)(a)AnAn
NdFeB sample: overview,
overview, (b) (b)
the the detail
detail indicated
indicated in Figure
in Figure 8a. 8a.
(Parameters set IV, powder layer height: 60 µm).
(Parameters set IV, powder layer height: μm).

4. Microstructural and
4. Microstructural andPhase
PhaseAnalysis
Analysis
Grain size
Grain distribution
size distributionwas
was checked andcompared
checked and compared onon two
two samples
samples withwith similar
similar overall
overall density,
density,
but but
different magnetic
different magnetic properties:
properties: IV-40
IV-40 and IV-80.The
and IV-80. Theinvestigated
investigated area
area waswas divided
divided into 7into 7 fields
fields as as
presented
presented in Figure
in Figure 9 anda aminimum
9 and minimum of of 4 photos
photoswere
weretaken
takeninin
random
random spots from
spots eacheach
from field.field.
 (Parameters set IV, powder layer height: 60 μm).

4. Microstructural and Phase Analysis

Grain size distribution was checked and compared on two samples with similar overall density,
Materials 2020, 13, 139 8 of 14
but different magnetic properties: IV-40 and IV-80. The investigated area was divided into 7 fields as
presented in Figure 9 and a minimum of 4 photos were taken in random spots from each field.

Figure 9. Scheme of a sample division for microstructure evaluation.

Materials 2019, 12, x FORFigure 9. Scheme of a sample division for microstructure evaluation.
PEER REVIEW 10 of 17

Grain boundaries detection and grain size analysis were performed using algorithm presented by
Grain boundaries detection and grain size analysis were performed using algorithm presented
Rabbani and Ayatollahi [23]. The analysis results along with representative microphotographs of the
by Rabbani and Ayatollahi [23]. The analysis results along with representative microphotographs of
microstructures are presented in Figure 10.
the microstructures are presented in Figure 10.

Figure 10. Grain

Figure 10. Grainsize
sizedistribution
distributionwith
withexemplary grain
exemplary images
grain of: (a)
images of:Sample IV-40IV-40
(a) Sample and (b)
andsample IV-80.
(b) sample
IV-80.

With the increase of the powder layer thickness, grain coarsening was observed as b10, b50, and
b90 and values for IV-40 sample were 0.65 μm, 0.94 μm, and 1.56 μm, respectively, while these values
for sample IV-80 were 0.79 μm, 1.28 μm, and 2.29 μm. The origin of grain coarsening with increasing
layer thickness was in the longer cooling time for the melt pool. For complete consolidation, more
Materials 2020, 13, 139 9 of 14

With the increase of the powder layer thickness, grain coarsening was observed as b10, b50,
and b90 and values for IV-40 sample were 0.65 µm, 0.94 µm, and 1.56 µm, respectively, while these
values for sample IV-80 were 0.79 µm, 1.28 µm, and 2.29 µm. The origin of grain coarsening with
increasing layer thickness was in the longer cooling time for the melt pool. For complete consolidation,
more laser energy was required, which in turn was proportional to the volume of the melt pool. The
cooling time, which had a direct effect on the grain size, was in scale with the melt volume. Regardless
of their size, the shape of observed grains did not change, however, and resembled polygons with
no defined orientation. No microstructural differences were found between areas neighboring the
weld lines and those placed between the weld lines. A comparison with the typical grain sizes of
conventional manufactured magnets is given in Table 2. The grain sizes realized with LPBF were
directly between the conventional processing routes. As a consequence, the microstructure can be
selectively manipulated using different layer heights. Because of the heat dissipation during cooling of
the melt pool through underlying layers, the already consolidated material was exposed to an indirect,
short-term heat treatment, which initiated grain coarsening. The result was a minimum grain size
of ~0.5 µm for both samples. With increasing layer height, the geometric dimension of the melt pool
increased, and, therefore, also the amount of heat that was conducted through the neighboring area.

Table 2. A comparison of a typical grain size of NdFeB magnets manufactured with various methods.

Manufacturing Method Typical Grain Size

Sintering 10–15 µm [11]
Hot pressing and hot deformation 0.3 µm [26]
LPBF 0.5–3 µm (current study)

The EDX
Materials analysis
2019, performed
12, x FOR PEER REVIEW on unetched samples showed that there were neither 11 of 17chemical

gradients nor inhomogeneity of elements distribution along the structure (EDX mappings attached
The EDX analysis performed on unetched samples showed that there were neither chemical
in Supplementary Data) (Figureof11).
gradients nor inhomogeneity This means
elements there
distribution wasthe
along a homogeneous alloy distribution
structure (EDX mappings attached in the
middleinof the melt tracks and also along the border towards neighboring tracks. Exsolutions
supplementary data) (Figure 11). This means there was a homogeneous alloy distribution in the or free
iron, asmiddle
[19] proposed,
of the meltweretracksnot
anddetected.
also alongAdditional EDX line
the border towards scans performed
neighboring also on etched
tracks. Exsolutions or free samples
iron, as [19]
were attached proposed, were not
as Supplementary detected. Additional EDX line scans performed also on etched
Data.
samples were attached as supplementary data.

Figure 11. Energy-dispersive X-ray (EDX) line- scans of (a) samples IV-40 and (b) sample IV-80 with
Figure 11. Energy-dispersive X-ray (EDX) line- scans of (a) samples IV-40 and (b) sample IV-80 with
images of the spots the measurements were taken from.
images of the spots the measurements were taken from.

XRD analysis was performed on samples of similar overall density but different magnetic
properties, e.g., IV-40 and IV-80. XRD patterns that were compared in Figure 12 showed no significant
differences. This was unsuccessful, however, for identifying phase composition. The lack of
differences show that no phase changes were induced while different powder layer thicknesses were
Materials 2020, 13, 139 10 of 14

XRD analysis was performed on samples of similar overall density but different magnetic
properties, e.g., IV-40 and IV-80. XRD patterns that were compared in Figure 12 showed no significant
differences. This was unsuccessful, however, for identifying phase composition. The lack of differences
show that
Materials no12,phase
2019, x FOR changes were induced while different powder layer thicknesses were applied
PEER REVIEW and
12 of 17
that the main factor influencing the magnetic properties was the grain size.

Figure 12. Comparison of XRD patterns for samples IV-40 and IV-80.
Figure 12. Comparison of XRD patterns for samples IV-40 and IV-80.
5. Results, Description for the Magnetic Tests
5. Results,
SamplesDescription forathe
printed with Magnetic
powder layer Tests
thickness of 20 µm for parameter set IV (as shown in Table 1)
wereSamples
crushedprinted
duringwith
processing
a powder before
layerfurther tests
thickness could
of 20 μm be
forperformed. The
parameter set IV (same was in
as shown true for
Table
samples
1) printed during
were crushed with a powder layer
processing thickness
before of tests
further 120 µm for parameter
could set III.
be performed. TheFigure
same 13 shows
was true the
for
J/H-curve for parameter set IV dependent on the layer height. The curves for parameter
samples printed with a powder layer thickness of 120 μm for parameter set III. Figure 13 shows the set III are
similar andfor
J/H-curve these are available
parameter set IVin the Supplementary
dependent on the layerData.
height. The curves for parameter set III are
The
similar shape
and theseofare
theavailable
demagnetization curves, and, therefore,
in the supplementary data. the magnetic behavior, changes with
increasing layer height in their curvature from a concave form, that is similar to conventional magnets,
towards a convex form. This will affect the flux density available when the magnet is in operation under
a weakening field. A small layer height is thus preferred when sufficient magnetic properties are in the
focus of the investigation. The origin of the curvature change was expected to correlate with the magnet
microstructure. While Jr was mostly dependent on the amount of the hard magnetic Φ-phase within
the volume, the resistance against demagnetization HcJ was greatly affected by the phase distribution
and homogeneity of the microstructure and the composition of the grain boundary phase [22]. In
LPBF, the powder was completely melted, which resulted in the formation of a new microstructure
responsible for the coercive field strength. This, in turn, resulted in significantly different values for
HcJ and BH(max) compared to plastic-bonded magnets produced from the initial material.
The discontinuous drop of coercivity of specimens built with 60 µm at −80 kA/m could be caused
by the spontaneous widening of cracks as a result of the magnetic load. Similar behavior can be
detected, when the LPBF-produced magnets were under critical mechanical load.
 5. Results, Description for the Magnetic Tests
Samples printed with a powder layer thickness of 20 μm for parameter set IV (as shown in Table
1) were crushed during processing before further tests could be performed. The same was true for
samples printed with a powder layer thickness of 120 μm for parameter set III. Figure 13 shows the
Materials J/H-curve
2020, 13, 139
for parameter set IV dependent on the layer height. The curves for parameter set III are 11 of 14
similar and these are available in the supplementary data.

Figure 13. J/H-curves of parameter set IV.

Figure 13. J/H-curves of parameter set IV.
Figure
Materials14 shows
2019, the
12, x FOR relationship
PEER REVIEW between the layer thickness and the magnetic key14properties.
of 17 In
general, a nearly linear loss of magnetic properties with increasing layer height can be derived.

Figure 14. Magnetization test results presented for the function of powder layer thickness.
Figure 14. Magnetization test results presented for the function of powder layer thickness.

The evolution of the microstructure with increasing layer height, as shown in the previous section,
Table 3. Comparison of results with previous studies for Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5.
underlines the results from the magnetic measurements for different layer heights. An increased
maximum Relative
layer height resulted in a decrease in magnetic
Coercivity Remanence properties and, therefore, it can be concluded that a
Feature magnetic energy density / Method
decrease in magnetic properties, as HcJ shown in Figure
Jr 14, was (BH)max
most likely caused by a coarsening in grain
product %
structure. A clear correlation between kA/m relative
T density and magnetic
kJ/m 3
properties,
% as indicated in other
This study (max. values) 516 a 0.563 b 35.9 c
studies [17,18], was not recognizable, e.g., sample IV-60 had lower magnetic properties 90.9 d LPBF (Figure 14) in
[19] 775.3 Max. 0.478 39.78 92 LPBF
spite of having higher density than IV-40 (Figure Avr. 0.45
7b). The remnant polarization Jr was the most robust
[17] N/A N/A 86 LPBF
indicator in the context of the layer height.Max.
This0.514
indicates that the amount of Φ-phase inside the material
was not as significantly
[18] changed915by the process
0.587 as the overall
52.1 microstructure.
N/A The difference
LPBF between
[22] without grain
the maximum polarization of 516520 mT and the theoretical
0.436 maximum
N/A of the65alloy of 730–760
LPBF mT [27] was
boundary infiltration
partly a[22]
result
with of theboundary
grain remaining porosity of approx. 15% (sample B-40 µm). Even with 100% theoretical
density, theinfiltration
polarization
with would only
1207 reach 593 0.390mT and can therefore notN/A
N/A fully explain the loss. Due to
LPBF
Nd50Tb20Cu30
the lack of rare earth content within the alloy, a slight loss of Nd/Pr into a non-ferromagnetic phase can
[4] 708.2 0.58 58.1 70 Extrusion
be expected. The coercivity dropped at a rate of 15%/20 µm layer height for set IV and with an even fall
Fused
[12] 740 0.310 N/A N/A Filament
Fabrication
Injection
[28] 700 0.52 44 70.5
moulding
Powdered material [27] 670–750 0.73–0.76 80–92 48.4–56.5 e N/A
a value for set IV, powder layer = 40 μm; b value for set III, powder layer = 40 μm, c value for set IV,
Materials 2020, 13, 139 12 of 14

rate for set III. The maximum energy product, (BH)max, dropped as a result of the changed curvature
in a manner similar to that for the coercivity. As Jr changed in a minor manner, the drop of (BH)max
was mainly influenced by the drop of HcJ . Magnetic properties were compared with state of art in
Table 3. As the sampling includes only one magnet investigated at every process point, no statistical
statements regarding the derivation of the parameters or also the confidence levels can be made.

Table 3. Comparison of results with previous studies for Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5.

Maximum Magnetic
Relative
Feature Coercivity HcJ Remanence Jr Energy Product Method
Density/%
(BH)max
kA/m T kJ/m3 %
a b c
This study (max. values) 516 0.563 35.9 90.9 d LPBF
[19] 775.3 Max. 0.478 39.78 92 LPBF
Avr. 0.45
[17] N/A N/A 86 LPBF
Max. 0.514
[18] 915 0.587 52.1 N/A LPBF
[22] without grain
520 0.436 N/A 65 LPBF
boundary infiltration
[22] with grain boundary
infiltration with 1207 0.390 N/A N/A LPBF
Nd50Tb20Cu30
[4] 708.2 0.58 58.1 70 Extrusion
Fused Filament
[12] 740 0.310 N/A N/A
Fabrication
Injection
[28] 700 0.52 44 70.5
moulding
Powdered material [27] 670–750 0.73–0.76 80–92 48.4–56.5 e N/A
avalue for set IV, powder layer = 40 µm; value for set III, powder layer = 40 µm, value for set IV, powder layer =
b c

40 µm; d value for set IV, powder layer = 60 µm; e apparent density of the powder.

6. Conclusions
A spherical powder of Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5 alloy was tested. The investigations
carried out showed that it is possible to 3D print magnets with a relative density of 90.87%. It was
found that the investigated material had a very narrow processing window and 20 W change in laser
power may result in destabilization of a melt track. It was shown that shape factors of the melt track
were strongly dependent on the powder bed depth, laser power, and laser speed. It was found that the
higher the energy density, the more stable the melt track became. It was also found that magnets of
high relative density may only be obtained when an adequate powder layer thickness was selected for
the given printing parameters. The internal porosity of the printed magnets increased with increasing
deviation from optimum powder layer thickness. Opened porosity, however, remained at a stable
level when the powder thickness exceeded the optimum value and increased rapidly when the powder
layer was too shallow (this was caused by excessive cracking). The investigations that were carried
out revealed that the stability of a melt track in LPBF-treated Nd7.5Pr0.7Fe75.4Co2.5B8.8Zr2.6Ti2.5
powdered alloy is prone even to small changes in LPBF process parameters. It was found that the
melt-track stability increased with increasing energy density (in tested range). The linear energy must
be at least at level of 0.300 J mm−2 , since lower energy inputs do not provide an adequately stable melt
track. The magnetic properties and also the shape of the J/H-curve showed a significant dependency
of the process parameters. With increasing layer height, a change in the shape of the J/H-curve was
detected. The highest magnetic values were comparable to those published in similar publications.
It was possible for the first time to match the magnetic performance of LPBF-treated magnets with
their internal microstructure. Increased magnetic parameters were obtained with decreasing layer
height. This effect is explained by a decreasing grain size with decreasing layer height due to the
level of laser energy applied. The overall grain size was in line with the coercivity values for isotropic
Materials 2020, 13, 139 13 of 14

NdFeB. Similar to nanocrystalline magnets, a smaller grain size is beneficial for higher HcJ . For further
improvement of the process, an adjusted alloy composition is necessary. Originally designed for grain
sizes well below 1 µm, the current composition is not optimal for the resulting grain size distribution
of 0.5–3 µm. It was also shown that the process parameters in the investigated range did not alter the
phase composition of the LBPF printed magnets. The grain size coupled with the overall density are the
main factors of influence on magnetic properties. Investigating spherical powder with a composition
designed for sintering could be a promising approach, as the grain size can be adjusted after the LPBF
process employing different heat treatment steps.

Supplementary Materials: The supplementary materials are available online at http://www.mdpi.com/1996-1944/

13/1/139/s1.
Author Contributions: Conceptualization and methodology, M.G., B.M., C.H. and M.S.; software, S.A.;
investigation, M.G., B.M., S.A. C.H., M.S., N.U., S.M.; writing—original draft preparation, S.M., M.G., B.M., N.U.
and M.S.; writing—review and editing, S.A., N.U., M.S., J.F. and C.S.; visualization, M.S.; supervision, J.F. and
C.S.; project administration, M.S., J.F. and C.S. All authors have read and agreed to the published version of
the manuscript.
Funding: This research received no external funding.
Acknowledgments: We would like to acknowledge the support of TU Graz and Joanneum Research as well as
Magnetfabrik Bonn GmbH (Bonn, Germany) for providing the powdered material. The magnetic measurements
and interpretations were carried out by N. Urban.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Gutfleisch, O.; Willard, M.A.; Bruck, E.; Chen, C.H.; Sankar, S.G.; Liu, J.P. Magnetic materials and devices
for the 21st century: Stronger, lighter, and more energy efficient. Adv. Mater. 2011, 23, 821–842. [CrossRef]
[PubMed]
2. Arnold, D.P. Review of microscale magnetic power generation. IEEE Trans. Magn. 2007, 43, 3940–3951.
[CrossRef]
3. Luo, Y. In the 20th International Workshop on Rare Earth Permanent Magnets and Their Applications. In
Proceedings of the Conference KNOSSOS—CRETE, REPM’08 2008, Knossos, Greece, 8–10 September 2008.
4. Li, L.; Jones, K.; Sales, B.; Pries, J.L.; Nlebedim, I.C.; Jin, K.; Kunc, V. Fabrication of highly dense isotropic
Nd-Fe-B nylon bonded magnets via extrusion-based additive manufacturing. Addit. Manuf. 2018, 21,
495–500. [CrossRef]
5. Wang, G.P.; Liu, W.Q.; Huang, Y.L.; Ma, S.C.; Zhong, Z.C. Effects of sintering temperature on the mechanical
properties of sintered NdFeB permanent magnets prepared by spark plasma sintering. J. Magn. Magn. Mater.
2014, 349, 1–4. [CrossRef]
6. Straumal, B.B.; Mazilkin, A.A.; Protasova, S.G.; Gusak, A.M.; Bulatov, M.F.; Straumal, A.B.; Baretzky, B. Grain
boundary phenomena in NdFeB-based hard magnetic alloys. Rev. Adv. Mater. Sci. 2014, 38, 17–28.
7. Oono, N.; Sagawa, M.; Kasada, R.; Matsui, H.; Kimura, A. Production of thick high-performance sintered
neodymium magnets by grain boundary diffusion treatment with dysprosium–nickel–aluminum alloy.
J. Magn. Magn. Mater. 2011, 323, 297–300. [CrossRef]
8. Tomše, T.; Jaćimović, J.; Herrmann, L.; Greuter, F.; Simon, R.; Tekavec, S.; Kobe, S. Properties of SPS-processed
permanent magnets prepared from gas-atomized Nd-Fe-B powders. J. Alloys Compd. 2018, 2018, 132–140.
[CrossRef]
9. Vial, F.; Joly, F.; Nevalainen, E.; Sagawa, M.; Hiraga, K.; Park, K.T. Improvement of coercivity of sintered
NdFeB permanent magnets by heat treatment. J. Magn. Magn. Mater. 2002, 242, 1329–1334. [CrossRef]
10. Ni, J.; Luo, W.; Hu, C.; Yin, Y.; Zhang, D.; Peng, X.; Han, Q. Relations of the structure and thermal stability of
NdFeB magnet with the magnetic alignment. J. Magn. Magn. Mater. 2018, 468, 105–108. [CrossRef]
11. Davies, B.E.; Mottram, R.S.; Harris, I.R. Recent developments in the sintering of NdFeB. Mater. Chem. Phys.
2001, 67, 272–281. [CrossRef]
Materials 2020, 13, 139 14 of 14

12. Huber, C.; Abert, C.; Bruckner, F.; Groenefeld, M.; Muthsam, O.; Schuschnigg, S.; Windl, R. 3D print of
polymer bonded rare-earth magnets, and 3D magnetic field scanning with an end-user 3D printer. Appl.
Phys. Lett. 2016, 109, 162401. [CrossRef]
13. Huber, C.; Goertler, M.; Abert, C.; Bruckner, F.; Groenefeld, M.; Teliban, I.; Suess, D. Additive manufactured
and topology optimized passive shimming elements for permanent magnetic systems. Sci. Rep. 2018, 8,
14651. [CrossRef] [PubMed]
14. Huber, C.; Abert, C.; Bruckner, F.; Groenefeld, M.; Schuschnigg, S.; Teliban, I.; Suess, D. 3D printing of
polymer-bonded rare-earth magnets with a variable magnetic compound fraction for a predefined stray
field. Sci. Rep. 2017, 7, 9419. [CrossRef] [PubMed]
15. Li, L.; Tirado, A.; Nlebedim, I.C.; Rios, O.; Post, B.; Kunc, V.; Lograsso, T.A. Big area additive manufacturing
of high performance bonded NdFeB magnets. Sci. Rep. 2016, 6, 36212. [CrossRef]
16. Paranthamn, M.P.; Shafer, C.S.; Elliott, A.M.; Siddel, D.H.; McGuire, M.A.; Springfield, R.M.; Ormerod, J.
Binder jetting: A novel NdFeB bonded magnet fabrication process. JOM 2016, 68, 1978–1982. [CrossRef]
17. Kolb, T. Laser Beam Melting of NdFeB for the production of rare-earth magnets. In Proceedings of the 2016
6th International Electric Drives Production Conference (EDPC), Nuremberg, Germany, 30 November–1
December 2016; pp. 3–40. [CrossRef]
18. Urban, N.; Kühl, A.; Glauche, M.; Franke, J. Additive Manufacturing of Neodymium-Iron-Boron Permanent
Magnets. In Proceedings of the 2018 8th International Electric Drives Production Conference (EDPC),
Schweinfurt, Germany, 4–5 December 2018; pp. 1–5. [CrossRef]
19. Jaćimović, J.; Binda, F.; Herrmann, L.G.; Greuter, F.; Genta, J.; Calvo, M.; Simon, R.A. Net shape 3D printed
NdFeB permanent magnet. Adv. Eng. Mater. 2017, 19, 1700098. [CrossRef]
20. Shi, X.; Ma, S.; Liu, C.; Wu, Q. Parameter optimization for Ti-47Al-2Cr-2Nb in selective laser melting based
on geometric characteristics of single scan tracks. Opt. Laser Technol. 2017, 90, 71–79. [CrossRef]
21. Yang, F.; Zhang, X.; Guo, Z.; Ye, S.; Sui, Y.; Volinsky, A.A. 3D printing of NdFeB bonded magnets with
SrFe12O19 addition. J. Alloys Compd. 2019, 779, 900–907. [CrossRef]
22. Huber, C.; Sepehri-Amin, H.; Goertler, M.; Groenefeld, M.; Teliban, I.; Hono, K.; Suess, D. Coercivity
enhancement of selective laser sintered NdFeB magnets by grain boundary infiltration. Acta Mater. 2019,
172, 66–71. [CrossRef]
23. Rabbani, A.; Ayatollahi, S. Comparing three image processing algorithms to estimate the grain-size
distribution of porous rocks from binary 2d images and sensitivity analysis of the grain overlapping degree.
Spec. Top. Rev. Porous Media Int. J. 2015, 6, 71–89. [CrossRef]
24. Gusarov, A.V.; Smurov, I. Modeling the interaction of laser radiation with powder bed at selective laser
melting. Phys. Procedia 2010, 5, 381–394. [CrossRef]
25. Yves-Christian, H.; Jan, W.; Wilhelm, M.; Konrad, W.; Reinhart, P. Net shaped high performance oxide
ceramic parts by selective laser melting. Phys. Procedia 2010, 5, 587–594. [CrossRef]
26. Gutfleisch, O.; Bollero, A.; Handstein, A.; Hinz, D.; Kirchner, A.; Yan, A.; Schultz, L. Nanocrystalline high
performance permanent magnets. J. Magn. Magn. Mater. 2002, 242, 1277–1283. [CrossRef]
27. MQP-S-11-9-20001. Available online: https://mqitechnology.com/product/mqp-s-11-9-20001/ (accessed on
20 December 2019).
28. MQP-S-11-9-20001-070 Isotropic Powder Datasheet. Available online: https://mqitechnology.com/wp-
content/uploads/2017/09/mqp-s-11-9-20001-070.pdf (accessed on 20 December 2019).

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).