Content uploaded by Mirko Sinico
Author content
All content in this area was uploaded by Mirko Sinico on Sep 17, 2019
Content may be subject to copyright.
Joint Special Interest Group meeting between euspen and ASPE
Advancing Precision in Additive Manufacturing
Ecole Centrale de Nantes, France, September 2019
www.euspen.eu
Influence of the particle size distribution on surface quality
of Maraging 300 parts produced by Laser Powder Bed Fusion
Mirko Sinico1,2, Ann Witvrouw1,2, Wim Dewulf1
1Department of Mechanical Engineering, KU Leuven, 3001 Leuven, Belgium
2Member of Flanders Make - Core lab PMA-P, KU Leuven, 3001 Leuven, Belgium
mirko.sinico@kuleuven.be
Abstract
This work investigates the effect of the powder particle size distribution on the surface finish of Maraging 300 specimens, produced
by the Laser Powder Bed Fusion (LPBF) process. Although it is recognized that the initial powder morphological characteristics play
an important role on LPBF part density, mechanical properties and surface quality, there is a lack of empirical data that could help to
link the powder properties to actual metrological established surface parameters, like Ra or Sa. For this reason, an extensive initial
powder characterization is presented in this paper, for three Maraging 300 batches, and first insights on the different obtained LPBF
surface quality are disclosed. The results demonstrate how small differences in particle size distribution can decrease the LPBF surface
roughness consistently. Moreover, it is shown how the use of fine powder can unlock novel LPBF processing strategies to further
improve surface finish, down to 1.5 µm measured Ra, and thus reduce eventual post-processing efforts.
Powder Characterization, Powder Flowability, Surface Roughness, Laser Powder Bed Fusion (LPBF), Laser Polishing, Laser Remelting
1. Introduction
The improvement of the surface quality of parts produced by
Laser Powder Bed Fusion is a critical topic of research in order to
reduce further post-processing. LPBF is increasingly being used
for the rapid manufacturing of parts and tools, e.g. injection
mold inserts, where roughness values lower than 0.1 μm Ra can
be required for critical surfaces. Typically, surface quality
improvements are achieved by studying and tuning the effect of
different LPBF processing parameters like laser scan strategy,
layer thickness, laser power and laser scan speed [1,2,3]. It is
likewise widely accepted that the characteristics of the LPBF
starting material, e.g. powder particle shape and the powder
particle size distribution (PSD), can have a significant effect on
part density, mechanical properties and surface quality [3,4].
However, related to the effects on surface finish, only few
literature studies, involving a limited palette of materials and
often with non-sufficiently in-depth analysis of the initial
powder morphology, exist on the subject. Previous studies on
the influence of PSD on LPBF manufacturing for 316L [5,6] have
highlighted how an amount of fine particles is beneficial, with an
increased powder packing in the deposited LPBF layer and a
diminished surface roughness on the produced parts. On the
other side, finer powders can present a lower flowability which
could hinder their deposition and therefore a limit exists on the
smaller usable PSD. This limit not only is controlled by the
average particle dimension, but also by the skewness and
kurtosis of the PSD, where e.g. narrow distributions typically
promote higher flowability [4].
In this paper, a first investigation of the influence of PSD
variability on the surface roughness of Maraging 300 material is
presented, with a prior extensive analysis on the powder
characterization step and its flowability limit dependent on the
PSD for three different powder batches. Subsequently, results
on the surface roughness of LPBF produced specimens are
discussed (section 4.2). Finally, a novel production strategy for
fine powder distributions is disclosed, combining surface
remelting with layer thicknesses as low as 10 µm to further
reduce the top surface roughness (section 5).
2. Powder characterization
Maraging 300 powders with different PSDs have been
acquired from the same supplier and fully characterized. The
hereby called Mar 15-45 is the standard powder being sold for
the LPBF process. The Mar 10-30 and Mar 5-15 have been
engineered in collaboration with the supplier to meet specific
constraints on the PSD. Specifically, the mean particle diameter
was reduced with steps of 10 µm while also the span of the
distribution (D10-D90) decreased with almost the same step, to
promote the flowability. The difference is clearly observed in the
laser diffraction (LD) analyses (Figure 1) for which
measurements were performed via a Beckman Coulter LS 13 320
particle size analyser in dry mode, compliant to the
ISO 13320:2009 standard.
Figure 1. Laser diffraction analysis of the three acquired Maraging 300
powders, with PSD and cumulative PSD.
A summary of the LD analysis is also presented in Table 1,
together with an overview of particles mean circularity and
mean sphericity measured respectively with a KEYENCE Digital
Microscope VHX-6000 and a Nikon X-ray Computed T omography
(CT) system XTH 225ST. Further details on the micro CT
measurement of the powders can be retrieved in [7] and are not
discussed in this paper. Definitions and an in-depth review on
powder particles shape analyses can be found in [8].
Table 1. Laser diffraction numerical results of the three acquired
Maraging 300 powders, together with mean circularity and mean
sphericity from microscopic and tomographic investigations.
D10
[µm]
D50
[µm]
D90
[µm]
Circularity,
mean [-]
Sphericity,
mean [-]
Mar
15-45
15.50
± 0.78
29.29
± 0.88
44.04
± 2.20
0.800 ±
0.100
0.958 ±
0.025
Mar
10-30
7.48
± 0.37
17.26
± 0.52
27.34
± 1.37
0.861 ±
0.098
0.962 ±
0.025
Mar
5-15
4.82
± 0.24
8.99
± 0.27
14.20
± 0.71
0.849 ±
0.087
0.951 ±
0.027
*Note: circularity and sphericity are reported with measured standard
deviation; LD statistics are reported with the theoretical coefficient of
variation (COV) inferred from the ISO 13320:2009 standard.
Measured circularity and sphericity are comparable between
the different batches of powders and, consequently, the
differences in PSD are presumed to influence predominately the
surface quality of LPBF parts produced with these raw materials.
Finally, before LPFB processing, the flowability (Hall Flow for
50 g of powder), apparent density ρapp and tap density ρtap have
also been acquired for all three powders following respectively
the ASTM B213, B212 and B527 standards. Table 2 summarizes
those results with the calculated Hausner Ratio HR (ρtap/ρapp).
The true density ρtrue is assumed to be 8.1 g/cm3 for
Maraging 300 as per supplier’s datasheet.
Table 2. ASTM compliant measurements of flowability, ρapp, ρtap and
computed HR of the three acquired Maraging 300 powders.
Hall Flow,
50 g [s]
ρ
app
/ ρ
true
[%]
ρ
tap
/ ρ
true
[%]
Hausner
Ratio [-]
Mar
15-45
14.64 ±
0.23
52.3 ±
0.3
60.2 ±
0.1
1.15
Mar
10-30
No flow
48.9 ±
0.2
58.8 ±
0.1
1.20
Mar
5-15
No flow
39.7 ±
0.3
55.1 ±
0.1
1.39
*Note: all measurements reported with standard deviation
A good flowability is necessary to be able to spread the powder
in thin layers during LPBF production. The powder flow is a
property influenced by many parameters, like PSD, particles
shape, moisture, Van der Waals forces, magnetism, etc [3,4]. As
a rule of thumb, a Hausner Ratio greater than 1.25 is considered
to be an indication of poor flowability [4] and therefore it can be
assumed that the Mar 5-15 will not be suitable for LPBF
processing. At the same time, it is worth noticing that the Mar
10-30 powder batch, even with an inconclusive Hall Flow
measurement, still presents a HR well below the suggested
critical threshold.
3. LPBF specimens production and characterization methods
A 3D Systems ProX DMP 320 (rev. A, laser spot size ~60 µm)
LPBF machine has been used for the fabrication of all the
specimens in this work. Build plates with arrangements of
samples ~20x10x10 mm have been processed with Mar 15-45
and Mar 10-30 powders, while the finer Mar 5-15 distribution
has been discarded after multiple failed jobs. As predicted by the
Hausner Ratio, the low flowability as well as particles
agglomeration of the Mar 5-15 powder hinders the formation of
an even powder layer during LPBF manufacturing. The use of a
different recoating system than the equipped rubber blade, like
the rotating roller, could eventually improve powder dispersion
[3] but has not yet been tested.
After production, the relative density of the samples have been
measured using Archimedes’ principle. Subsequently, Ra
roughness analyses have been collected on top and side surfaces
by a Mitutoyo Formtracer CS-3200S4 profilometer, equipped
with a 60° conic probe ending in a 2 μm diameter ball, following
the ISO 4287:1997 standard. Each measurement reported in this
work has been repeated 5 times at different locations on each
specimen, taking care that the profiles are measured
perpendicular to the typical lay lines featured on AM surfaces,
i.e. the laser line tracks. For results with Ra above 10 µm,
2 sampling lengths have been collected instead of 5 (suggested
value in the ISO standard) given the limited dimension of the
specimens. It is anyhow assumed, from previous literature on
LPBF measured roughness [9], that 2 sampling lengths i.e. Ra2
(16 mm) are enough to obtain a good estimation of the Ra value.
SEM inspection of selected specimens has been performed with
a Philips XL 30 FEG, while qualitative optical investigations are
done with a KEYENCE Digital Microscope VHX-6000.
4. Results and discussion
4.1. LPBF parameters optimization
Parameters optimization has been based on previous
published work on the LPFB processability of Maraging 300 from
our research group [10] and available literature [1,11]. For this
comparison, samples have been fabricated at 3 different
volumetric energy densities Ev (=
×ℎ× with P laser power,
h hatch spacing, t layer thickness), varying laser scanning speed
and laser power while keeping hatch spacing and layer thickness
fixed at 70 µm and 30 µm respectively. Striped pattern hatching,
with 90° rotation between layers, and a single contour were also
kept fixed. The same DoE has been repeated identically twice,
once with the Mar 15-45 powder and once with the Mar 10-30.
A summary of the DoE with the results on the Archimedes
relative densities of the samples is presented in Figure 2.
Figure 2. Relative densities of specimens fabricated from Mar 10-30 and
Mar 15-45 powder batches, compared with the same repeated DoE.
As expected, the relative density increases for an increase in Ev
or for an increase in laser power at the same Ev. This trend is
valid, in the investigated parameters window, for both powder
distributions. It must be noticed however that the relative
density of the Mar 15-45 specimens suddenly start to decrease
at 190 W laser power. It is hypothesized that, at this threshold,
Optimal parameter set
98.4
98.7
99.0
the laser melting mode changes from conduction to keyhole
eventually leading to keyhole voids and an increase in the total
porosity content [12].
The Mar 10-30 specimens do not present this behaviour, even if
their mean relative densities are slightly lower than the
Mar 15-45 counterpart. It is generally believed that a fine
powder granulation leads to higher relative densities. A narrow
spread of the PSD is on the other hand reported as detrimental
[3,4] because it lowers the apparent density ρapp of the powder
(as quantified in Table 2). A comprehensive weighted model of
the two influencing factors has not yet been established, and
therefore we can only assume that, for this study, the narrower
PSD of the Mar 10-30 powder as well as fine particles
vaporisation upon melting could be the cause for the slightly
higher porosity content.
If a parameter set has to be chosen after the optimization, it is
suggested to operate at 170 W, ~70 J/mm3 Ev (1150 mm/s scan
speed) for both the Maraging powders, as highlighted in Figure
2. The volumetric energy density with this optimal combination
of parameters is in line with previously reported studies [1,11].
4.2. Surface roughness comparison
The top surface roughness of the produced specimens,
according to the designed DoE, has been measured and plotted
in Figure 3.
Figure 3. Top surface roughness of the specimens fabricated from Mar
10-30 and Mar 15-45 powder batches, compared with the same
repeated DoE; for 170 W, 70 J/mm3 Ev SEM pictures are also attached.
Top surface roughness is inversely proportional to the Ev, since
at higher energy densities a more effective melting lowers the
porosity content (Figure 2) and the amount of partially sintered
particles. At the same time, surface roughness seems directly
proportional to the laser power, at the same Ev, as increasing
laser power results in larger and more turbulent melt-pools with
an increased amount of spatter formation due to the lower melt
viscosity and the higher degree of Marangoni effect [2,13].
Alongside, the influence of the PSD of the powder used to
produce the specimens is appreciated at all levels of Ev or laser
power. The specimens fabricated with the Mar 10-30
distribution presents a higher surface quality with a diminution
of Ra from ~20 % to > 50 % compared to the Mar 15-45 samples.
This effect is more pronounced at higher Ev, where Ra remains
stable around 6 µm with its lowest value being at 170 W,
70 J/mm3. SEM pictures for this combination of parameters are
attached in Figure 3 for both Mar 10-30 (Ra of 5.34 µm) and Mar
15-45 (Ra of 12.12 µm). It is evinced again [3,4,5,6] how powders
with fine granulometry are more substantially melted, leading to
a more stable and continuous laser line track and a diminished
balling effect and spatter formation.
A slightly lower Ra is also obtained for the side surfaces (90°) of
the samples made from Mar 10-30 (Figure 4), even if the
reduction is less evident and consistent being the staircase
effect predominant in surface roughness contribution.
Figure 4. Example of side surface roughness of the specimens fabricated
from the two powder batches, compared at the same Ev of 70 J/mm3.
A more complete study should be carried out to understand how
much the powder granulometry may influence the roughness of
surfaces produced at different orientations in respect to the
build platform.
5. A novel remelting strategy for the LPBF of fine powders
A common solution to decrease the surface roughness of the
top surfaces in LPBF processing is the use of remelting [14]. With
remelting, the already scanned layer is exposed once more to
the laser radiation without new deposition of powder. The
remelting step can be performed with the same LPBF processing
parameters or it can be tuned (with different parameters) in
accordance to the desired outcome needed, with a similar
optimization approach that is found for laser polishing machines
[15]. Multiple remelting passes are generally suggested to
further improve the final surface quality, at the disadvantage of
processing time.
It is likewise well established that reducing the processing layer
thickness t usually improves the surface quality [13]. This
improvement is mostly appreciated for side surfaces, with the
diminished staircase effect, but top surfaces are affected too.
In the final part of this work we want to suggest a novel
remelting strategy for top surfaces, where the last layers are first
built with t of 10 µm and subsequently remelted. This approach
basically combines the two positive effects of the remelting step
and low t, but its successful implementation is constrained by
the limits of the granulometry of the powder used and
sometimes by LPBF machine limits. Fortuitously, the Mar 10-30
powder batch presents at least 50 vol% of the particles (D50)
with a diameter around the effective layer thickness for t of
10 µm (teff = 16.67 µm for a powder layer density of 60 % [5]),
and therefore is deemed suitable for testing the approach.
The DoE hereby explained is just a brief overview of the obtained
results. In details, the last 90 µm of 20x10x10 mm specimens
have been printed with a t of 10 µm followed by five remelting
Mar 15-45
Mar 10-30
steps (90° rotation each step) of the last layer. Parameters for
the underlying part have been kept at the optimized
combination reported in paragraph 4.1, while the parameters
for the top 10 µm layers have been varied. Moreover, to
compare the obtained Ra from the novel approach to a standard
remelting step or decreased t alone, reference specimens with
only the 5x remelting applied on the top surface or only the 10
µm reduced t for 90 µm have also been produced and measured.
The outcomes of this brief trial are summarized and reported in
Figure 5 for two Ev (calculated for t = 10 µm).
Figure 5. Measured Ra of the samples produced at R5x (top surface
remolten 5 times), 10 µm t (last 90 µm of the specimens built with a layer
thickness of 10 µm), and the combination of the two, for two Ev;
sampling pictures from optical microscopy attached.
It is evinced how the single reference strategies alone provide
little improvement from the already low Ra obtained with the
Mar 10-30 specimens, while the combination of the two positive
effects at a tuned Ev of ~165 J/mm3 is able to further reduce the
surface roughness to a remarkable value of 1.5 µm.
With the initial reduced layer thickness step, we are able to
decrease both the mean height of the profile elements in the
roughness domain Rc, and their mean width RSm. We believe
that this reduction, even if it doesn’t reflect in a significant
decrease in Ra, is beneficial to prepare the surface for the last
remelting step. To prove this hypothesis, full topographic
measurements are currently being performed on a 3D optical
profiler Sensofar S neox.
6. Conclusions
In this work preliminary results, linking the LPBF powder
morphological characteristics to the LPBF surface quality, are
disclosed for the first time for the Maraging 300 material.
Through the extensive test of three different powder batches, it
is evinced how a flowability limit exits dependent on the particle
size distribution, and this limit will determine if the powder is
suitable for LPBF processing. Furthermore, through the
comparison of specimens produced with the used Mar 10-30
and Mar 15-45 powder distributions, it is demonstrated how a
small difference in PSD towards finer particles can considerably
improve the surface quality of fabricated parts with up to a
> 50 % decrease in the evaluated Ra roughness parameter.
Finally, the use of finer powder is not only positive for the
general diminished Ra, but could as well unlock the possibility to
adapt the LPBF scanning strategy to further enhance the surface
finish. For this reason, a novel remelting strategy for the LPBF of
fine powders has been tested, and was proven effective
reaching the value of 1.5 µm Ra. Full topographic measurements
are currently being performed to better understand the
developments of surface texture.
Acknowledgements
This research was funded by The EU Framework Programme
for Research and Innovation - Horizon 2020 - Grant Agreement
No 721383 within the PAM2 (Precision Additive Metal
Manufacturing) research project.
References
[1] Casalino G, Campanelli S L, Contuzzi N and Ludovico A D 2015
Experimental investigation and statistical optimisation of the
selective laser melting process of a maraging steel Optics & Laser
Technology 65 151-158
[2] Derahman N A, Karim M S, and Amran N A M 2018 Effects of process
parameters on surface quality of parts produced by selective laser
melting - ANFIS modelling Proceedings of Mechanical Engineering
Research Day 2018 115-116
[3] Malekipour E and El-Mounayri H 2018 Common defects and
contributing parameters in powder bed fusion AM process and their
classification for online monitoring and control: a review Int J Adv
Manuf Technol 95 527-550
[4] Tan J H, Wong W L E and Dalgarno K W 2017 An overview of powder
granulometry on feedstock and part performance in the selective
laser melting process Additive Manufacturing 18 228-255
[5] Spierings A B, Herres N and Levy G 20 11 Influence of the particle si ze
distribution on surface quality and mechanical properties in AM
steel parts Rapid Prototyping Journal 17 195-202
[6] Liu B, Wildman R, Tuck C, Ashcroft I, and Hague R 2011 Investigation
the effect of Particle Size Distribution on processing parameters
optimisation in Selective Laser Melting process Proceedings of the
Annual International SFF Symposium 2011
[7] Sinico M, Ametova E, Witvrouw A, and Dewulf W 2018
Characterization of AM Metal Powder with an Industrial Microfocus
CT: Potential and Limitations Proceedings of the 2018 ASPE and
euspen Summer Topical Meeting 69 286-291
[8] Sutton A T, Kriewall C S, Leu M C and Newkirk J W 2017 Powder
characterisation techniques and effects of powder characteristics
on part properties in powder-bed fusion processes Virtual and
Physical Prototyping 12 3-29
[9] Triantaphyllou A et al. 2015 Surface texture measurement for
additive manufacturing Surf. Topogr.: Metrol. Prop. 3
[10] Kempen K, Yasa E, Thijs L, Kruth J P and Van Humbeeck J 2011
Microstructure and mechanical properties of Selective Laser Melted
18Ni-300 steel Physics Procedia 12 255-263
[11] Tan C, Zhou K, Kuang M, Ma W and Kuang T 2018 Microstructural
characterization and properties of selective laser melted maraging
steel with different build directions Science and Technology of
Advanced Materials 19 746-758
[12] King W E et al. Observation of keyhole-mode laser melting in laser
powder-bed fusion additive manufacturing 2014 Journal of
Materials Processing Technology 214 2915-2925
[13] Bacchewar P B, Singhal S K and Pandey P M 2007 Statistical
modelling and optimization of surface roughness in the selective
laser sintering process Proceedings of the Institution of Mechanical
Engineers, Part B: Journal of Engineering Manufacture 221 35–52
[14] Yasa E, Deckers J, and Kruth J 201 1 The investigation of the influence
of laser re-melting on density, surface quality and microstructure of
selective laser melting parts Rapid Prototyping Journal 17 312–327
[15] dos Santos Solheid J, Jürgen Seifert H, and Pfleging W 2018 Laser
surface modification and polishing of additive manufactured
metallic parts Procedia CIRP 74 280–284
50 µm
50 µm