Additive Manufacturing xxx (xxxx) xxx

Contents lists available at ScienceDirect

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

Additive manufacturing of the ferritic stainless steel SS441

Dennis Karlsson a, *, Chia-Ying Chou b, Niklas Holländer Pettersson b, Thomas Helander c,
Peter Harlin d, Martin Sahlberg a, Greta Lindwall b, Joakim Odqvist b, Ulf Jansson a
a
Department of Chemistry – Ångström Laboratory, Uppsala University, Box 523, 751 20, Uppsala, Sweden
b
Royal Institute of Technology (KTH), Department of Material Science and Engineering, Brinellvägen 23, 100 44, Stockholm, Sweden
c
Kanthal AB, Box 502, 734 27, Hallstahammar, Sweden
d
Sandvik Additive Manufacturing, Sandvik AB, Box 510, 101 30, Stockholm, Sweden

A R T I C L E I N F O A B S T R A C T

Keywords: In this study, the ferritic stainless steel SS441 was produced with excellent mechanical properties using laser
Laser powder bed fusion powder bed fusion (L-PBF) compared to samples produced by conventional casting and hot-rolling. In addition,
L-PBF thermodynamic calculations were utilized to study the phase stability at elevated temperatures and to under­
Nucleation
stand the solidification behavior. The hot-rolled sample showed a grain size up to several hundred μm with
Thermodynamic calculations
Mechanical properties
additional precipitates of TiN and Nb(C,N). In contrast, the as-built L-PBF samples displayed a grain size in the
μm range. Spherical precipitates with a size of around 50 nm could be observed and were attributed to a
corundum phase from the thermodynamic calculations. The printed material shows superior mechanical prop­
erties, with more than 30 times higher impact energy compared to the hot-rolled alloy (217 ± 5 J vs. 7 ± 0.5 J).
Furthermore, the properties are anisotropic for the L-PBF produced alloy, with the highest tensile strength
vertical to the build direction. The superior mechanical properties of the L-PBF produced sample can be
attributed to a smaller grain size, giving a higher strength according to the Hall-Petch relationship. The
anisotropy of the material can be eliminated by heat treatments at 900 ◦ C followed by water quenching, but the
absolute strength decreases slightly due to formation of intermetallic phases such as Nb(C,N) and the Fe2Nb
Laves phase. The results clearly illustrates that L-PBF provides a promising manufacturing route for enhanced
strength of ferritic stainless steels.

1. Introduction evaluate new alloy systems with improved properties.

As the design possibilities are limited in conventional methods such
as casting or molding, an increased interest has been seen in alternative
manufacturing routes. Additive manufacturing (AM) enables production
with complex geometries and opens up for design of new components for
various technical applications. AM utilizes a digital model to build
components in a layer-by-layer fashion and has found its uses in e.g.
aerospace, automotive and medical technology industries [1].
Laser powder bed fusion (L-PBF) is an AM technique in which a layer
of powder is melted selectively using a high-power laser. Due to the
small interaction volumes and melt pools, the cooling rate is extremely
high during the process. As a consequence, the microstructure is highly
different compared to the forged or cast analogues with a significant
grain refinement. Hence, the L-PBF parts often show improved perfor­
mance with excellent mechanical properties. Today, the number of al­
loys used commercially are limited and there is therefore a need to
Some of the most commonly studied alloys in AM includes Ni-based
alloys (i.e. Alloy 718 and Alloy 625), austenitic steels such as 316 L and
304 L as well as titanium alloys such as Ti-6Al-4V [1–9]. Ferritic stain­
less steels, on the other hand, have not been widely studied in AM. These
steels have a relatively low coefficient of thermal expansion compared to
austenitic steels and have therefore been used for applications which
require high fatigue resistance at elevated temperatures. The solid so­
lution strengthened ferritic steel of type AISI 441 uses Nb and Ti to in­
crease the high temperature strength. The addition of minor amounts of
Nb contributes to the increased strength at elevated temperatures.
However, at higher Nb contents (above 0.8 %), the formation of Nb(C,
N), Fe3Nb3C and Fe2Nb Laves phases are detrimental for the high tem­
perature strength [10,11]. Sello and Stumpf have studied the effect of
Laves phase embrittlement and its transformation mechanism in cast
components of this alloy [12,13]. They concluded that the impact
toughness decreases when Laves phase precipitates in the grain

* Corresponding author.
E-mail address: dennis.karlsson@kemi.uu.se (D. Karlsson).

https://doi.org/10.1016/j.addma.2020.101580
Received 12 June 2020; Received in revised form 18 August 2020; Accepted 28 August 2020
Available online 10 September 2020
2214-8604/© 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Please cite this article as: Dennis Karlsson, Additive Manufacturing, https://doi.org/10.1016/j.addma.2020.101580
D. Karlsson et al.
boundaries and that the rate of precipitation decreases with larger grain
size, due to lowering of the amount of grain boundaries. Finally, it has
been observed that the oxygen content in components produced by AM
can be high, which can influence the microstructure and the mechanical
properties of the alloy [14–16].
The aim of this study was to evaluate the mechanical properties of
components of the stainless steel type AISI 441 produced by L-PBF and
compare them to a conventionally produced material made by casting
and hot-rolling. Furthermore, the temperature dependence of the me­
chanical properties was investigated after annealing. Thermodynamic
calculations were conducted to understand the microstructure and for­
mation of secondary phases such as Fe2Nb and Nb(C,N) upon annealing.
Additionally, the influence of higher oxygen contents in the L-PBF
produced SS441 was studied by thermodynamic calculations.
2. Experimental procedure
Conventional reference material of SS441 was prepared by melting
and hot-rolling in pilot scale. The melting was performed by induction
melting under an argon protective atmosphere in an alumina crucible,
followed by casting into a copper mold. The melt size was 1.4 kg and the
raw materials used were elements of pure Fe, Cr, C, Mn and Si. Some of
the minor alloying elements were added from FeNb, SiTi and FeCrN
alloys. The cast ingot was machined to a diameter of 24 mm and heated
to 1100 ◦ C for 20 min before hot-rolling in multiple passes into a bar
with a final cross section of approximately 10 × 10 mm.
In this study a pre-alloyed Sandvik Osprey® SS441 powder was used.
The powder was manufactured by inert gas atomization using nitrogen
gas and sieved to -45 + 10 μm size fraction to perform in AM processing.
Model components were built by AM using an EOS M100 machine,
equipped with a 200 W fiber laser with a spot size of approximately
40 μm. The machine operated in a flowing Ar(g) atmosphere and the
oxygen content in the build chamber was controlled to never exceed
0.11 %. The process parameters including laser power, scan speed and
hatch distance were varied. A wide range of the parameters were tested
and the density variation for a set of parameters measured by Archi­
medes principle can be seen in Fig. 1. The sample with highest density
was produced using a laser power of 150 W, scan speed of 750 mm/s,
hatch distance of 70 μm and a layer thickness of 20 μm. Each layer was
segmented into 5 mm wide stripes, with a stripe overlap of 0.1 mm. A
Fig. 1. Variation of the relative density with laser energy input. The brown
square marks the density of the sample produced with the parameters selected
for further investigation of microstructures and mechanical properties (For
interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article).
2
Additive Manufacturing xxx (xxxx) xxx
rot-scan strategy (67◦ ) with bidirectional scan vectors without con­
touring were utilized for all builds. All samples investigated in this study
were produced using these parameters (brown square in Fig. 1).
Vertically built (0◦ ) and horizontally built (90◦ ) samples were pro­
duced, see Fig. 2, to evaluate anisotropic properties. To study phase
stability and the effect of annealing on the room temperature mechan­
ical properties, an annealing experiment under flowing Ar(g) at 900 ◦ C
for 20 min with a heating rate of 10 ◦ C/min followed by rapid water
quenching was performed.
X-ray diffraction (XRD) was carried out on a Bruker Advanced D8
equipped with a Lynxeye LE detector operating with Cu Kα radiation.
The microstructure was studied in a Zeiss Merlin scanning electron
microscope (SEM) equipped with an energy dispersive X-ray spectros­
copy (EDS) detector and electron backscattered diffraction (EBSD). The
samples were grinded by SiC paper of size 220–4000, followed by pol­
ishing using diamond suspensions of 3 and 1 μm. The final surface finish
was obtained using a 40 nm colloidal SiO2 polishing step. Furthermore,
(scanning) transmission electron microscopy (S-TEM) investigations
were carried out on a FEI Titan Themis 200 equipped with probe
corrector and a SuperX energy dispersive X-ray spectroscopy detector.
Chemical analysis was performed by X-ray florescence (XRF) using a
XRF PANalytical Magix Fast according to ASTM E572. The carbon and
sulfur content was analyzed using a LECO CS-600, while the nitrogen
and oxygen content was analyzed using a LECO TC-600 according to
ASTM E1019.
Tensile test rods were machined into the final shape (4c20) from the
cylindrical samples produced in the L-PBF process and from the bars
made out of the reference material. The specimens were evaluated by
tensile tests performed according to ASTM E8M in room temperature.
Each tensile test was performed using an initial strain rate of 2.5 × 10− 4/
s. To reduce the time for the tests the strain rate was increased to 5 × 10-
3
/s at 1.7 % elongation until fracture. For each material condition three
tensile test rods were evaluated. Based on these three results an average
value were calculated and reported.
Impact toughness properties was evaluated using Charpy V-notch
test according to ISO-14556 in room temperature. The specimens were
machined from rectangular bars produced in the L-PBF process and from
bars made out of the reference material. The final Charpy-V notch were
machined into full size impact toughness rods with a cross section of
10 × 10 mm, in accordance with the demands specified in ISO-14556.
All impact toughness tests were performed in a Zwick PSW 750
impact toughness test machine.
3. Thermodynamic calculations
Thermodynamic calculations using the software Thermo-Calc [17]
based on the CALPHAD (CALculation of PHAase Diagram) method [18]
were performed and compared with the observed experimental results
from the L-PBF and hot-rolling processes. In order to determine the so­
lidification path and the redistribution of elements the Scheil-Gulliver
Fig. 2. Schematic of the printing geometries of the vertically and horizontally
built samples. The blue cross sections marks the investigated regions (For
interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article).
D. Karlsson et al.
model [19,20] implemented in Thermo-Calc was used. The model as­
sumes local equilibrium at the solid/liquid interface with perfect mixing
of the elements in the liquid phase and no diffusion in the solid phase.
This gives an efficient approximation of the micro-segregation occurring
during solidification processes. The calculations were performed for the
chemical composition of the as-built ferritic stainless steel SS441
(Table 1) and the thermodynamic database TCFE9 [21] was used for
equilibrium and Scheil-Gulliver calculations. In the Scheil-Gulliver
calculation a perfect mixing of C, N is assumed in the solid since they
dissolve interstitially in solid with a high diffusion rate. The calculation
is terminated at solidification fraction 99 % to avoid overestimation of
the segregation. The influence of oxygen present during the AM route
were studied by equilibrium and Scheil-Gulliver calculations with and
without oxygen. Thermal analysis showed that the printed components
contained about 0.045 wt% oxygen, compared to 0.0083 wt% for the
cast and hot-rolled sample.
The calculated equilibrium phase fractions as a function of temper­
ature for the ferritic stainless steel SS441 can be seen in Fig. 3. Fig. 3a
and b show the property diagram without presence of oxygen, while c
and d show the property diagram when the oxygen content is 0.045 wt
%. It can be seen from Fig. 3a and c that a mixture of sigma and bcc
phases is expected at low temperatures and that their formations are not
influenced by the presence of oxygen. As the temperature is increased
from 400 ◦ C, the sigma phase fraction decreases and the amount of bcc
phase fraction increases. At higher temperatures (above 700 ◦ C) the
sigma phase is no longer an equilibrium phase and a close to single-
phase bcc alloy is expected. In addition to the major sigma and bcc
phases, minor amounts of other phases are expected according to the
thermodynamic calculations, see Fig. 3b and d.
Fig. 3b shows the property diagram of the minor equilibrium phases
when no oxygen is present. As can be seen, a G-phase forms at low
temperatures (up to around 400 ◦ C). This phase is a Ni and Si-rich pre­
cipitate and has experimentally been observed after annealing at low
temperatures in ferritic, austenitic and duplex stainless steels [22–24].
Additionally, a Z-phase is stable up to a temperature of 600 ◦ C. The
Z-phase has a tetragonal crystal structure and consists typically of Cr, (V,
Nb) and N. This phase can form in stainless steels with high nitrogen
contents [25,26]. Both of these phases are expected in very low amounts
(under 0.2 mol %) and are only stable at low temperatures. Some of the
other precipitating phases, such as the Laves phase, are expected to form
to a larger extent (up to 1 mol %). The Laves phase is commonly
observed in the ferritic stainless steel SS441 and can have detrimental
effects on the mechanical properties of the alloy. The typically observed
Laves phase is the hexagonal C14 Fe2Nb phase. This phase is expected at
temperatures below 960 ◦ C. Lastly, the formation of cubic nitrides and
carbonitrides, such as Nb(C,N) and TiN is expected and they are stable
up to temperatures close to or higher than the melting point of the bcc
phase. This suggests that these phases, especially the TiN which can
form above the liquidus temperature (up to around 1670 ◦ C), can act as
nucleation sites for solidifying of the alloy.
Fig. 3d shows that an additional phase is thermodynamically stable
when oxygen is added to the system, namely corundum (M2O3 where
M = Ti, Al, Cr). Additionally, no G-phase is expected even at low tem­
peratures in the presence of oxygen. At high temperatures (up to
1740 ◦ C) a corundum phase is stable, consisting of 34 at% Ti, 4 at% Al, 1
at% Cr and 60 at% O. As the Ti-rich corundum phase is stable above the
liquidus temperature (in combination with TiN), there is a possibility for
Table 1
Composition of the ferritic stainless steel SS441 in wt% from the as-built L-PBF
sample.
Fe C Si Mn Cr Ni Ti Al
Bal. 0.02 0.9 0.8 18.8 0.2 0.3 0.02
Nb N O Mo Co W V
0.8 0.08 0.045 0.03 0.07 0.02 0.07
3
Additive Manufacturing xxx (xxxx) xxx
grain refinement by oxide formation in the liquid phase before solidi­
fication of the ferritic stainless steel. It has been suggested previously by
Shi et al. [27] that the formation of TiN enwrapped Ti2O3 particulates in
the liquid phase can act as a nucleation site to promote a refined so­
lidification structure. The HT corundum phase observed in the ther­
modynamic calculations shows a higher Al solubility at lower
temperatures and below 1050 ◦ C there is a miscibility gap causing a
decomposition into Ti-rich and Al-rich corundum phases. The stability
region of the TiN phase remains similar to the alloy without oxygen,
with only a slightly lower melting temperature (see Fig. 3b). The amount
of the Fe2Nb Laves-phase is slightly lower in the presence of oxygen,
while the amount of Z-phase and Nb(C,N) increases. The solvus tem­
perature of 925 ◦ C for the Laves-phase remains similar to the calculation
with oxygen excluded.
In order to investigate the solidification process for the ferritic
stainless steel SS441, the solidification pathway was studied using the
Scheil-Gulliver model. The solidification path of SS441 without and with
oxygen can be seen in Fig. 4a and b, respectively. Without oxygen, the
first phase to nucleate is TiN (Fig. 4a). However, only a small amount of
this phase can form at high temperatures, due to the low Ti and N
content. The bcc phase forms after the TiN phase, when the temperature
decreases to around 1500 ◦ C. At the end of the solidification path, it can
be seen that small amounts of Nb-rich Nb(C,N) and C14 Laves-phases are
formed due to Nb-segregation to the liquid phase.
The solidification path when oxygen is included in the calculation
can be seen in Fig. 4b. The curve looks similar to the one for the oxygen-
free alloy, but the first phase to solidify is now a Ti-rich corundum phase.
This phase coexist with only the liquid phase at temperatures above
1640 ◦ C, where TiN becomes stable. The amount of corundum and TiN is
relatively low, which is expected from the low amount of the elements
forming these phases (Ti, N and O). When the temperature is decreased
below 1500 ◦ C, the bcc phase starts to form at the same temperature as
for the oxygen-free alloy. At the end of the solidification path Nb(C,N) is
formed. However, no Laves phase formation is expected when oxygen is
present in the alloy. This is due to a shift of nucleation temperatures
depending on the liquid composition during the two different solidifi­
cation pathways. This is in good agreement with previous studies by Hou
and Cheng, who observed a formation of Nb-rich (Nb,Ti)(C,N) on Ti-rich
(Nb,Ti)(C,N) surfaces and in grain boundaries of δ-Fe during solidifi­
cation of a ferritic stainless steel [28]. Furthermore, they explained that
the Ti-rich (Nb,Ti)(C,N) formed on oxygen dispersed particles, similar to
the expected solidification path shown in Fig. 4b.
The change of the composition of the bcc phase during solidification
can be seen in Fig. 5. A change in the composition is apparent at around
1350 ◦ C in Fig. 5a, where the amount of Fe, Ni, Co and Mn increases
while the Cr and Si content decreases. This is when the Laves phase
becomes stable. It can also be seen from Fig. 5b that the Nb segregates
strongly during the solidification.
4. Experimental results and discussion
4.1. Cast and hot-rolled samples
As described above, the SS441 reference samples were made by
casting followed by hot-rolling. Analysis of the samples showed an ox­
ygen content of only 0.0083 wt%. XRD of the cast and hot-rolled samples
show mainly peaks from a bcc structure, expected for a ferritic steel.
However, additional weak peaks from precipitates can also be observed.
This is seen in Fig. 6a, which shows the region around the bcc (110) peak
in a logarithmic scale from the hot-rolled sample before and after
annealing at 900 ◦ C for 20 min (full range diffraction patterns can be
seen in SI 1). Peaks from two additional phases, TiN and Nb(C,N) can
also be seen in the diffractogram. The TiN phase is expected to become
stable in the liquid phase before the bcc phase solidifies according to the
solidification path shown in Fig. 4. After annealing, the phase compo­
sition is unchanged, with no formation of additional phases such as a
D. Karlsson et al. Additive Manufacturing xxx (xxxx) xxx

Fig. 3. a) Equilibrium phase fractions of all the phases present in the oxygen-free ferritic steel SS441 showing bcc, sigma and liquid phases. b) shows a magnification
of the low phase fraction region (0-1 mol %), showing presence of minor amounts of several additional phases. c) and d) show corresponding phase fraction diagrams
including 0.045 wt% oxygen in the calculations.

Fig. 4. Solidification path of the ferritic stainless steel SS441 calculated using the Scheil-Gulliver model a) without oxygen and b) with oxygen.

Laves phase (which would be expected from the equilibrium diagram in annealed samples is shown in Fig. 6b, with an EBSD orientation map of
Fig. 3). The extra peak at 2θ = 40◦ correspond to the 110 reflection of the the bcc phase in the inset. The EBSD map shows that the sample consists
ferritic phase with CuKβ radiation. A SEM image of the hot-rolled and of grains with sizes up to 50 μm in diameter, with some misorientation

4
D. Karlsson et al.
Fig. 5. a) elemental distribution in the bcc phase during solidification and b) magnified in the composition range of the alloying elements for the SS441 steel
including 0.045 wt% oxygen.
Fig. 6. a) X-ray diffraction patterns of the hot-rolled sample before and after annealing at 900 ◦ C. b) SEM image of the hot-rolled and annealed sample perpendicular
to the rolling direction.
within the grains, as indicated by a color shift. The SEM image has a
higher magnification and shows that the larger grains consist of several
subgrains (a few μm large) with similar orientations. Two types of pre­
cipitates can be seen; darker particles with a typical size of a few μm
mainly distributed inside the ferrite grains and bright particles with a
sub-micron size mainly distributed at the grain boundaries and in­
terfaces of the darker precipitates. EDS shows that the dark and bright
precipitates consists of TiN and Nb(C,N), respectively. Since the TiN
grains are mainly found inside the ferrite grains, they most likely formed
in the liquid phase during the solidification in agreement with the pre­
dicted solidification path shown in Fig. 4. The Nb(C,N) phase can be seen
as bright particles that often form at the interface between the TiN and
the ferrite phase. Some Nb(C,N) also forms in the grain boundaries of the
ferrite phase, which is expected due to the strong segregation of Nb
suggested by the thermodynamic calculations (see Fig. 5). This is also in
good agreement with earlier studies by Hou and Cheng [28].
4.2. Laser powder bed fusion
The L-PBF samples were produced from gas atomized powder with
particle size between 10 and 45 μm. Analysis of the powder showed an
oxygen content of 0.047 wt%, more than five times higher than for the
5
Additive Manufacturing xxx (xxxx) xxx
cast and hot-rolled sample. Fig. 7 shows cross sectional SEM images of
the powder at different magnifications. Each powder particle consists of
many grains, each grain with a dimension of a few μm. According to EDS
there is an enrichment of Nb in the grain boundaries (see Fig. 7b), seen
as a higher intensity in the backscattered image, indicating that the Nb
segregates during the solidification in agreement to the thermodynamic
calculations in Fig. 4. Additionally, some dark particles can be seen in
Fig. 7b. By EDS analysis it could be seen that these are enriched in Ti and
N and distributed within the grains, indicating that they are formed
during an early stage solidification. No oxygen enriched particles could
be found in the powder.
XRD of the as-built and annealed L-PBF samples shows mainly peaks
from a bcc structure in both the vertically (0◦ ) and horizontally (90◦ )
built samples, expected for a ferritic steel. This is seen in Fig. 8, which
shows the region around the bcc (110) in a logarithmic scale from the L-
PBF samples before and after annealing at 900 ◦ C for 20 min. The full
width at half maximum (FWHM) of the L-PBF samples is significantly
larger compared to the hot-rolled samples, suggesting that these samples
are either strained or possesses a smaller grain size (see SI 1). Two
additional peaks can be seen at 2Th = 40◦ and 43.5◦ which correspond
to the 110 reflection of the ferritic phase for CuKβ and WLα1, respectively
(this is an artefact from the instrument). No additional phases were
D. Karlsson et al.
Fig. 7. SEM images of the SS441 powder used for the L-PBF process at different magnifications.
Fig. 8. X-ray diffraction patterns of the as-built L-PBF samples and after
annealing at 900 ◦ C for 20 min.
observed for the as-built samples with XRD. After annealing, however,
peaks from an Fe2Nb Laves phase can be seen. This phase is expected to
form and be stable up to around 960 ◦ C according to the property dia­
gram in Fig. 3. In contrast to the hot-rolled sample, XRD showed no
formation of TiN or Nb(C,N) in the printed samples. Chemical analysis of
the printed samples shows an oxygen content of 0.045 wt% similar to the
atomized powder. This suggest that no additional oxygen is added
during the printing process.
Fig. 9a and b show EBSD orientation cross-section images of the as-
built alloy in 0◦ (vertical) and 90◦ (horizontal) directions, respectively.
The build direction is given by the arrow. The grain size varies between
different regions within the sample, where regions with long elongated
grains can be seen. The average grain size was calculated from the EBSD
images to 1.61 and 1.76 μm for the 0◦ (vertical) and 90◦ (horizontal)
directions, respectively. The grain size is slightly anisotropic, but larger
elongated grains are commonly observed along the build directions for
alloys produced by L-PBF processes [29–31]. In contrast to these re­
ported alloys, no preferred orientation could be observed in the ferritic
stainless steel. Fig. 9c and d show SEM images of the 90◦ sample at
higher magnifications. In Fig. 9c it can be seen that the growth of the
grains are directed in the direction of the heat gradient, from the outer
border of the melt pool to the center of the molten material, marked by
red arrows. Fig. 9d shows a high magnification SEM image of the 90◦
6
Additive Manufacturing xxx (xxxx) xxx
sample, where brighter and darker particles are formed. The bright
particles indicate segregation of heavy elements and form a network
(such a region is marked with the dashed red line), indicating the well
known cellular structure forming in L-PBF materials. According to the
thermodynamic calculations a segregation of Nb is expected during the
(Fig. 5) solidification. This segregation of Nb is important for the
behavior of the alloy during annealing, which is discussed later.
In contrast to the hot-rolled sample, formation of well distributed
spherical particles within the grains could be observed for the printed
material (see Fig. 9e). These precipitates are around 50 nm large and
enriched in Ti and O (see Fig. 9e), suggesting formation due to oxygen
redistribution during the L-PBF process as they are absent inside the
powder particles prior to printing. Commonly, powder processes show
higher oxygen content compared to hot-rolling due to the high surface
area of the powder particles, explaining the absence of these precipitates
in the hot-rolled sample with significantly lower oxygen content
(0.0083 wt% compared to 0.0449 wt% for the printed samples). As
predicted by the thermodynamic calculations and by the Scheil-Gulliver
model with the presence of oxygen, the formation of a Ti-rich corundum
phase is expected during the initial stage of solidification. As these
particles are well dispersed, it is likely that they formed from the liquid
phase.
Fig. 10a and b shows EBSD orientation images of the annealed alloy
built in the vertical and horizontal directions, respectively. The general
morphology remains similar to the as-built, with only a minor grain
growth to 1.66 and 1.82 μm for the vertically built and horizontally built
samples, respectively. Fig. 10c shows a SEM image of the 90◦ sample.
The melt pools are visible with elongated grains at the melt pool border,
similar to the as-built sample shown in Fig. 9c. A higher magnification
SEM image can be seen in Fig. 10d. In addition to the spherical particles,
bright precipitates have formed in the boundaries of the cellular struc­
ture during printing (as shown in Fig. 9d). During the annealing, the
segregation of Nb (indicated by the Scheil-Gulliver calculations in Fig. 5
and the SEM image in Fig. 9d) to easily nucleating sites in the grain
boundaries is enhanced to form Nb-rich precipitates (Fig. 10e). This is
assigned to the Fe2Nb Laves phase observed in the diffractograms in
Fig. 8. From the Scheil-Gulliver solidification calculations, a Laves phase
is predicted to form during the late stage of solidification, but this could
not be experimentally confirmed in the as-built samples. However, the
previous mention segregation of heavier elements (Nb) to the bound­
aries of the cellular structure could be observed. When the alloy is
annealed, this segregation is enhanced, which enables formation of the
Fe2Nb Laves phase. The Laves phase forms mainly in grain boundaries of
the bcc phase and not in the cellular boundaries, which is likely due to
easier nucleation in the grain boundary of the alloy. The depletion of Nb
from the ferrite phase in order to form the precipitates can be detri­
mental for the mechanical properties, as Nb is used to enhance the high
D. Karlsson et al. Additive Manufacturing xxx (xxxx) xxx

Fig. 9. EBSD orientation images of the as-built L-PBF produced samples in the 0◦ vertical (a) and 90◦ horizontal (b) directions. c) illustrates one melt pool and d) a
high magnification SEM image. e) shows an HAADF micrograph and corresponding EDS maps from the 0◦ vertical sample. The arrow marks the building directions.

Fig. 10. EBSD orientation images of the annealed L-PBF produced samples in the 0◦ vertical (a) and 90◦ horizontal (b) directions. c) illustrates one melt pool and d) a
high magnification SEM image. e) shows an HAADF micrograph and corresponding EDS maps from the 0◦ vertical sample. The arrow marks the building directions.

temperature strength in these alloys [10]. Additionally, the formation of
brittle Laves phases are expected to lower the strength of the material.
4.3. Mechanical properties
Fig. 11 shows stress-strain curves for L-PBF produced samples
compared to the cast and hot-rolled reference sample. The yield strength
and elongations are summarized Table 2. As can be seen, the yield
strength and elongation of the hot-rolled sample are 314 MPa and 34 %,
respectively. The yield strength and elongation for SS441 is dependent
on the annealing temperature and are reported in literature to be be­
tween 300− 350 MPa and 34–48 %, respectively [13]. In contrast,
significantly higher yield strengths (>679 MPa) are observed in the
as-built L-PBF samples. This can be attributed to the much smaller grain
size in the L-PBF samples (<1.8 μm) compared to the, up to 50 μm, large
grains in the hot-rolled sample. The smaller grain size will give rise to a
higher yield strength according to the Hall-Petch relationship. There is
also a small difference in yield strength depending on printing direction.
The 0◦ samples have a yield strength which is 60 MPa lower than the 90◦
samples, which could be explained by more elongated grains along the

7
D. Karlsson et al.
Fig. 11. Typical tensile stress-strain curves of the hot-rolled and L-PBF pro­
duced alloys.
Table 2
Summary of tensile mechanical properties, including yield strength and elon­
gation at room temperature.
Sample condition Yield strength (MPa) Elongation (%)
Cast and hot-rolled 314.2 ± 4.4 34.1 ± 0.7
L-PBF 0◦ 679.0 ± 2.1 29.6 ± 0.5
L-PBF 90◦ 741.0 ± 0.2 27.8 ± 0.5
L-PBF 0◦ annealed 536.6 ± 1.1 28.9 ± 0.6
L-PBF 90◦ annealed 545.9 ± 0.5 28.9 ± 0.1
testing direction in the 0◦ samples. The increase in yield strength is also
followed by a reduction of the elongation prior to fracture by around 2 %
for the 90◦ samples compared to the 0◦ samples.
After annealing the yield strength of the L-PBF produced samples are
reduced to around 540 MPa and the difference between the two printing
directions is now small. The annealing process leads to a minor increase
in grain size (around 3 %), which according to the Hall-Petch relation­
ship should reduce the yield strength. However, the formation of a
brittle Laves phase during the annealing is likely to have a larger impact
on the properties than the relatively small change in grain size. The AISI
441 type ferritic steel is stabilized by substitutional solution atoms in the
bcc matrix, locking the dislocation movement to enhance the strength of
the alloy by solid solution hardening. An element that is used for this is
Nb, which makes it detrimental for the solid solution hardening if
intermetallic phases are formed by consuming the solute Nb from the
ferritic phase. In Fig. 10c and d it can be seen that some precipitates,
attributed to the Fe2Nb phase, are formed at the grain boundaries in the
alloy. Thus, the depletion of Nb in the ferrite phase is expected to reduce
the strength in agreement with the results shown in Fig. 11 and Table 2.
The formation of Laves phase is also expected to contribute to the me­
chanical properties. However, a previous study has shown that the for­
mation of Laves phase slightly increases the strength and reduces the
elongation in SS441 [13]. The results in Table 2 shows an opposite trend
and are supporting a mechanism where the loss in strength is caused by a
loss of Nb in the ferrite phase.
The results from the room temperature Charpy impact tests are
shown in Fig. 12. The cast and hot-rolled sample shows an impact
strength of 7 ± 0.5 J, while the as-built L-PBF samples exhibit impact
strengths of 217 ± 5 J and 187 ± 2 J for the 0◦ (vertically) and 90◦
(horizontally) built samples, respectively. The significant difference in
impact strength for the cast and hot-rolled material compared to the L-
PBF produced material can be attributed to the very different micro­
structures. The formation of large grains during the slow cooling after
8
Additive Manufacturing xxx (xxxx) xxx
hot-rolling, including the formation of additional phases (nitrides, car­
bides and intermetallic phases) after annealing at 900 ◦ C has a detri­
mental effect on the materials ability to withstand impact. After
annealing at 900 ◦ C for the L-PBF samples, the impact strength decreases
to 170 ± 2 J and 165 ± 5 J for the 0◦ and 90◦ built samples, respectively.
However, the material appears more isotropic, as there is no significant
difference in impact strength between the two different building di­
rections. Sello and Stumpf observed a peak impact strength of 60 J after
annealing at 850 ◦ C for 30 min followed by rapid cooling during water
quenching. They explained that the increased annealing temperature
past 850 ◦ C resulted in a decreased impact strength, due to grain growth
effect seen in many ferritic stainless steels [12]. The grain size that they
observed after annealing ranged from around 20 μm up to 140 μm at
temperatures from 850 ◦ C to 1100 ◦ C, with the most crucial temperature
increase seen between 950 ◦ C–1000 ◦ C. The grain size of the L-PBF
produced samples is only 1.6–1.8 μm, depending on the build direction
and growth during annealing. This grain size is around 8% of the sam­
ples studied by Sello and Stumpf, which can explain the excellent
properties of the L-PBF produced alloy compared to the cast and
hot-rolled sample with a grain size of up to 140 μm. In addition, the
formation of intermetallic phases in the grain boundaries can lower the
impact strength further due to depletion of Nb in the bcc structure,
lowering the effect of solid solution hardening. Fig. 12b and c shows
typical fracture surfaces for the 0◦ built L-PBF sample and the cast and
hot-rolled sample, respectively. The fracture surface of the L-PBF sample
show presence of voids on macro and micro-scale, which is typical for a
ductile fracture and in good agreement with the high impact energy
shown in Fig. 12a. The brittle fracture of the 7 J cast and hot-rolled
sample is shown in Fig. 12c. The large grains are reflected on the frac­
ture surface and a transgranular fracture is suggested by the cleavage
steps and feather markings on the surface.
5. Conclusions
The ferritic stainless steel type AISI 441 was synthesized by casting
followed by hot-rolling, as well as by L-PBF. The L-PBF process produces
a highly different microstructure compared to hot-rolling, e.g. a signif­
icant grain refinement can be obtained during L-PBF due to the rapid
cooling rate. Furthermore, the L-PBF printed samples contained a higher
oxygen content than the cast and hot-rolled sample. Computational
thermodynamic calculations on oxygen-free and oxygen-containing
compositions was carried out in order to interpret the microstructure
of the hot-rolled and L-PBF produced samples.
In the hot-rolled sample, formation of TiN and Nb(C,N) could be
observed in the ferrite phase. The TiN is distributed within the grains
and expected to form in the liquid during solidification according to a
Scheil-Gulliver simulation. No change of phase composition was
observed after annealing at 900 ◦ C, even though the Fe2Nb Laves phase
is predicted to be stable according to the thermodynamic calculations.
The Laves phase is also kinetically unfavorable to form. Additionally, as
this phase primarily precipitates at grain boundaries, the large grain size
of the hot-rolled sample could decrease the nucleation sites for the Laves
phase formation.
The as-built alloy crystallizes in a bcc structure, expected for a ferrite
phase, during manufacturing and some Ti and O-rich precipitates could
be observed. The precipitates are well distributed within the ferrite
phase. This is in good agreement with the thermodynamic calculations,
which predicts a Ti-rich corundum phase to be the first to solidify from
the liquid phase. During annealing at 900 ◦ C a Nb/Si-rich Laves phase
precipitates at the grain boundaries.
The mechanical properties are enhanced by the L-PBF manufacturing
route, greatly increasing the impact energy and tensile strength of the
material compared to hot-rolling. This can be explained by a grain
refinement during L-PBF, increasing the strength according to the Hall-
Petch relationship. The properties of the as-built material is anisotropic,
with the highest tensile strength tested vertical to the build direction,
D. Karlsson et al.
Fig. 12. Charpy impact test of the samples, showing the vertically built samples in empty boxes and the horizontally built samples with striped boxes (a), fracture
surface of the vertically L-PBF as-built sample and the hot-rolled sample are shown in (b) and (c), respectively.
and can be lowered by annealing at 900 ◦ C. Nb is used in SS441 to in­
crease the solid solution hardening. When the Nb is consumed to form
Fe2Nb, the strength is expected to decrease due to the depletion of Nb in
the ferrite phase as well as the formation of the brittle Laves phase.
This study shows that excellent mechanical properties can be ach­
ieved for the ferritic stainless steel AISI441 by grain refinement through
L-PBF processing. Additionally, annealing can be utilized to obtain
isotropic properties, but at a cost of decreased strength.
CRediT authorship contribution statement
Dennis Karlsson: Conceptualization, Methodology, Investigation,
Validation, Formal analysis, Visualization, Writing - original draft,
Writing - review & editing. Chia-Ying Chou: Conceptualization,
Methodology, Investigation, Visualization, Writing - original draft.
Niklas Holländer Pettersson: Conceptualization, Investigation.
Thomas Helander: Conceptualization, Investigation, Resources. Peter
Harlin: Conceptualization, Investigation, Formal analysis, Resources.
Martin Sahlberg: Conceptualization, Methodology, Supervision. Greta
Lindwall: Conceptualization, Methodology, Supervision. Joakim
Odqvist: Supervision, Resources, Funding acquisition. Ulf Jansson:
Conceptualization, Methodology, Resources, Supervision, Writing - re­
view & editing, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
9
Additive Manufacturing xxx (xxxx) xxx
Acknowledgements
Financial support from the Swedish Foundation for Strategic
Research, project “SSF–Development of processes and Materials in AM”,
Reference No. GMT14-0048, is greatly acknowledged. Parts of this work
was performed within the VINNOVA Competence Center Hero-m 2i,
financed by the Swedish Governmental Agency for Innovation Systems
(VINNOVA), Swedish industry and the Royal Institute of Technology
(KTH).
References
[1] T. DebRoy, et al., Additive manufacturing of metallic components – process,
structure and properties, Prog. Mater. Sci. 92 (2018) 112–224, https://doi.org/
10.1016/j.pmatsci.2017.10.001.
[2] W.E. Frazier, Metal additive manufacturing: a review, J. Mater. Eng. Perform. 23
(6) (2014) 1917–1928, https://doi.org/10.1007/s11665-014-0958-z.
[3] A. Strondl, R. Fischer, G. Frommeyer, A. Schneider, Investigations of MX and
gamma’ /gamma’ ’ precipitates in the nickel-based superalloy 718 produced by
electron beam melting, Mat. Sci. Eng. A-Struct. 480 (May 1-2) (2008) 138–147,
https://doi.org/10.1016/j.msea.2007.07.012, 15 (in English).
[4] K.N. Amato, et al., Microstructures and mechanical behavior of Inconel 718
fabricated by selective laser melting (in English), Acta Mater. 60 (March 5) (2012)
2229–2239, https://doi.org/10.1016/j.actamat.2011.12.032.
[5] P. Nandwana, A.M. Elliott, D. Siddel, A. Merriman, W.H. Peter, S.S. Babu, Powder
bed binder jet 3D printing of Inconel 718: Densification, microstructural evolution
and challenges (in English), Curr. Opin. Solid St. M 21 (Augtst 4) (2017) 207–218,
https://doi.org/10.1016/j.cossms.2016.12.002.
[6] J.A. Gonzalez, J. Mireles, S.W. Stafford, M.A. Perez, C.A. Terrazas, R.B. Wicker,
Characterization of Inconel 625 fabricated using powder-bed-based additive
manufacturing technologies, J. Mater. Process Tech. 264 (2019) 200–210, https://
doi.org/10.1016/j.jmatprotec.2018.08.031. Additive Manufacturing 1–4 2014 (in
English) February.
[7] L. Löber, C. Flache, R. Petters, U. Kühn, J. Eckert, Comparison of different post
processing technologies for SLM generated 316l steel parts, Rapid Prototyp. J. 19
(3) (2013) 173–179, https://doi.org/10.1108/13552541311312166.
D. Karlsson et al.
[8] Q.B. Nguyen, Z. Zhu, F.L. Ng, B.W. Chua, S.M.L. Nai, J. Wei, High mechanical
strengths and ductility of stainless steel 304L fabricated using selective laser
melting (in English), J. Mater. Sci. Technol. 35 (no February 2) (2019) 388–394,
https://doi.org/10.1016/j.jmst.2018.10.013.
[9] L. Thijs, F. Verhaeghe, T. Craeghs, J.V. Humbeeck, J.-P. Kruth, A study of the
microstructural evolution during selective laser melting of Ti–6Al–4V, Acta Mater.
58 (9) (2010) 3303–3312, https://doi.org/10.1016/j.actamat.2010.02.004.
[10] N. Fujita, K. Ohmura, M. Kikuchi, T. Suzuki, S. Funaki, I. Hiroshige, Effect of Nb on
high-temperature properties for ferritic stainless steel, Scr. Mater. 35 (September 6)
(1996) 705–710, https://doi.org/10.1016/1359-6462(96)00214-X, 15 (in
English).
[11] G.M. Sim, J.C. Ahn, S.C. Hong, K.J. Lee, K.S. Lee, Effect of Nb precipitate
coarsening on the high temperature strength in Nb containing ferritic stainless
steels, Mat. Sci. Eng. A-Struct. 396 (April 1-2) (2005) 159–165, https://doi.org/
10.1016/j.msea.2005.01.030, 15 (in English).
[12] M.P. Sello, W.E. Stumpf, Laves phase embrittlement of the ferritic stainless steel
type AISI 441, Mat. Sci. Eng. A-Struct. 527 (20) (2010) 5194–5202, https://doi.
org/10.1016/j.msea.2010.04.058. Jul 25 (in English).
[13] M.P. Sello, W.E. Stumpf, Laves phase precipitation and its transformation kinetics
in the ferritic stainless steel type AISI 441, Mat. Sci. Eng. A-Struct. 528 (January 3)
(2011) 1840–1847, https://doi.org/10.1016/j.msea.2010.09.090 (in English) 25.
[14] V. Pacheco, et al., Thermal stability and crystallization of a Zr-based metallic glass
produced by suction casting and selective laser melting, J. Alloys. Compd. 825
(June) (2020) 153995, doi: ARTN 15399510.1016/j.jallcom.2020.153995. 5 (in
English).
[15] M. Velasco-Castro, E. Hernandez-Nava, I.A. Figueroa, I. Todd, R. Goodall, The
effect of oxygen pickup during selective laser melting on the microstructure and
mechanical properties of Ti-6Al-4V lattices, Heliyon 5 (December 12) (2019)
e02813, https://doi.org/10.1016/j.heliyon.2019.e02813.
[16] S. Zhongji, T. Xipeng, T. Shu Beng, Effects of Chamber Oxygen Concentration on
Microstructure and Mechanical Properties of Stainless Steel 316L Parts by Selective
Laser Melting, 2018.
[17] J.O. Andersson, T. Helander, L.H. Hoglund, P.F. Shi, B. Sundman, THERMO-CALC
& DICTRA, computational tools for materials science, Calphad-Comput. Coupling
Phase Diagrams Thermochem. 26 (June 2) (2002) 273–312, https://doi.org/
10.1016/S0364-5916(02)00037-8, doi: Pii S0364-5916(02)00037-00038, (in
English).
[18] H. Lukas, S. Fries, B. Sundman, Computational thermodynamics. The CALPHAD
method, Comput. Thermodynam.: Calphad Method 131 (01/01) (2007), https://
doi.org/10.1017/CBO9780511804137.
[19] E. Scheil, Bemerkungen zur schichtkristallbildung, Zeitschrift für Metallkunde 34
(3) (1942) 70–72.
10
Additive Manufacturing xxx (xxxx) xxx
[20] G.H. Gulliver, The quantitative effect of rapid cooling upon the constitution of
binary alloys. (in English), J. I Met. 9 (1) (1913) 120–157 [Online]. Available: <Go
to ISI>://WOS:000201418000005.
[21] Thermo-Calc Software, TCFE9: TCS Steel and Fe-alloys Database, Available at: htt
ps://www.thermocalc.com/media/10306/tcfe9_extended_info.pdf. (Accessed: 9th
April 2020), 2020.
[22] M.J. Yang, et al., A novel ferritic steel family hardened by intermetallic compound
G-phase, Mat. Sci. Eng. A-Struct. 745 (February 4) (2019) 390–399, https://doi.
org/10.1016/j.msea.2018.11.148. ACTA MATER 71 2014 (in English).
[23] I. Shuro, H.H. Kuo, T. Sasaki, K. Hono, Y. Todaka, M. Umemoto, G-phase
precipitation in austenitic stainless steel deformed by high pressure torsion, Mat.
Sci. Eng. A-Struct. 552 (August 30) (2012) 194–198, https://doi.org/10.1016/j.
msea.2012.05.030 (in English).
[24] S.L. Li, Y.L. Wang, X.T. Wang, F. Xue, G-phase precipitation in duplex stainless
steels after long-term thermal aging: A high-resolution transmission electron
microscopy study, (in English), J. Nucl. Mater. 452 (September 1-3) (2014)
382–388, https://doi.org/10.1016/j.jnucmat.2014.05.069.
[25] P.W. Robinson, D.H. Jack, Precipitation of z-phase in a high-nitrogen stainless
steel, J. Heat. Treat. 4 (1) (1985) 69–74, https://doi.org/10.1007/bf02835491.
[26] H.K. Danielsen, Review of Z phase precipitation in 9–12 wt-%Cr steels, Mater. Sci.
Technol. 32 (2) (2016) 126–137, https://doi.org/10.1179/
1743284715y.0000000066.
[27] X.-f. Shi, L.-z. Chang, L. Zhou, Solidification structure refinement of ferritic
stainless steel based on heterogeneous nucleation technology, J. Iron Steel Res. Int.
26 (2) (2018) 137–147, https://doi.org/10.1007/s42243-018-0141-0.
[28] Y.Y. Hou, G.G. Cheng, Effect of Nb on the As-cast Structure and Compactness
Degree of Ferritic Stainless Steel Dual Stabilized by Ti and Nb (in English), ISIJ Int.
58 (12) (2018) 2298–2307, https://doi.org/10.2355/isijinternational.ISIJINT-
2017-595.
[29] L.N. Carter, C. Martin, P.J. Withers, M.M. Attallah, The influence of the laser scan
strategy on grain structure and cracking behaviour in SLM powder-bed fabricated
nickel superalloy (in English), J. Alloys. Compd. 615 (December 5) (2014)
338–347, https://doi.org/10.1016/j.jallcom.2014.06.172.
[30] K. Kunze, T. Etter, J. Grasslin, V. Shklover, Texture, anisotropy in microstructure
and mechanical properties of IN738LC alloy processed by selective laser melting
(SLM) (in English), Mat. Sci. Eng. A-Struct. 620 (January 3) (2015) 213–222,
https://doi.org/10.1016/j.msea.2014.10.003.
[31] O. Andreau, et al., Texture control of 316L parts by modulation of the melt pool
morphology in selective laser melting (in English), J. Mater. Process Tech. 264
(February) (2019) 21–31, https://doi.org/10.1016/j.jmatprotec.2018.08.049.