J. GONZALEZ-GUTIEREZ, D. GODEC, R. GURÁŇ, M. SPOERK, C. KUKLA, C.

HOLZER ISSN 0543-5846

METABK 57(1-2) 117-120 (2018)
UDC – UDK 669.15:621.775:669.015:66.049.4:621.385.833=111

3D PRINTING CONDITIONS DETERMINATION FOR FEEDSTOCK

USED IN FUSED FILAMENT FABRICATION (FFF) OF 17-4PH
STAINLESS STEEL PARTS
Received – Primljeno: 2017-05-30
Accepted – Prihvaćeno: 2017-08-30
Original Scientific Paper – Izvorni znanstveni rad

Fused filament fabrication combined with debinding and sintering could be an economical process for 3D printing
of metal parts. In this study, compounding, filament making and FFF processing of a feedstock material containing
55 vol. % of 17-4PH stainless steel powder and a multicomponent binder system are presented. For the FFF process,
processing windows of the most significant parameters, such as range of extrusion temperatures (210 to 260 °C),
flow rate multipliers (150 to 200 %), and 3D printing speed multipliers (60 to 100 %) were determined for a constant
printing bed temperature of 60 °C.
Keywords: stainless steel, powder, additive manufacturing (AM), FFF, scanning electron microscopy (SEM)

INTRODUCTION This paper describes a process for compounding a

new feedstock formulation consisting of a proprietary
Additive manufacturing (AM) comprises a group of binder system and 17-4PH stainless steel powder, the
technologies used to build physical parts by adding mate- process of filament making, and the determination of
rial in a layer-by-layer fashion from a computer aided 3D printing processing windows for the most signifi-
design (CAD) file, as opposed to subtractive manufactur- cant parameters related to FFF.
ing methods, such as machining [1]. AM is also referred
as 3D printing, solid freeform fabrication (SFF) and rapid
prototyping (RP) [2]. Over the last three decades, many MATERIALS AND METHODS
AM technologies have been developed for the produc- Materials
tion of polymeric, metallic or ceramic parts [3].
Feedstock materials for FFF are composed of a
One of the most commonly used AM technologies
polymeric binder system and filler particles. Here, the
for the production of metal parts is selective laser sinter-
binder system had three components: the main binder
ing (SLS). Its biggest disadvantage is the dependency
component, the backbone polymer and a compatibiliser.
on high power lasers, which can be very costly. There-
The main binder component was a soft and flexible
fore, fused filament fabrication (FFF) has shown great
thermoplastic elastomer (TPE). The polyolefin-based
potential as a cost effective alternative [4]. In FFF, the
backbone and a commercially available compatibiliser
building material is supplied in the form of spooled
were used in the feedstock. The fillers were 17-4PH
polymer-based filaments into a heating unit with a noz-
stainless steel particles, whose size, measured by laser
zle using counter-rotating drive-wheels. The heating
diffraction, is shown in Table 1.
unit is controlled to move in the X-Y plane, and as it
moves, the material is extruded through the nozzle on a Table 1 Particle size data on 17-4PH stainless steel
platform that moves in the Z-direction [3].
D10 / μm 4,2
FFF was first developed to work with polymeric D50 / μm 12,3
filaments. For the fabrication of metal parts, filaments D90 / μm 28,2
made of a polymer highly-filled with metal particles
can be used. After shaping the filament into what is
referred to as the green part, the binder system is re- Compounding
moved from the part and then sintered to obtain a final
metal part. Feedstocks were compounded in a co-rotating twin-
screw extruder (Leistritz Extrusionstechnik GmbH,
Germany), designed for compounding highly-filled
J. Gonzalez-Gutierrez, R. Guráň, M Spoerk, C. Holzer, Polymer Pro- polymers with metal powders. All the binder compo-
cessing, Montanuniversitaet Leoben, D. Godec, Faculty of Mechanical nents were premixed in solid state. The metal powder
Engineering and Naval Architecture, University of Zagreb, C. Kukla,
Industrial Liaison, Montanuniversitaet Leoben
and the binder were fed using two gravimetric feeding
e-mail: damir.godec@fsb.hr units. The result was a feedstock material with 8,9 wt. %

METALURGIJA 57 (2018) 1-2, 117-120 117

 J. GONZALEZ-GUTIEREZ et al.: 3D PRINTING CONDITIONS DETERMINATION FOR FEEDSTOCK USED IN FUSED...

of binder and 91,1 wt. % of steel powder. The extruded Microscopy

material was pulled away from the die with a conveyor
belt and later granulated in a cutting mill. In order to observe the powder distribution in the
binder, a filament and a printed part were investigated
by means of optical and scanning electron microscopy
Filament production (SEM). The filament and the printed part were cryogen-
Filaments were prepared using a single-screw ex- ically fractured under liquid nitrogen.
truder (Dr. Collin GmbH, Germany). A round capillary
with a diameter of 1,75 mm and length of 20 mm was RESULTS
used. At the exit of the die a conveyor belt was placed to Filament quality
pull the filament as it was extruded. After the filament
left the conveyor belt it was guided to a spooling device A constant geometry of the filament is important for
where it was wound onto spools. A device to measure its continuous transportation in an FFF machine. For
the diameter and ovality of the extruded filament was this reason the diameter and the ovality of the produced
placed between the haul-off unit and the spooling de- filament were monitored during production. Please
vice. According to the reading of the measuring device note that ovality is defined as the difference of the di-
the extrusion and haul-off speeds were manually regu- ameter in the horizontal direction and the diameter in
lated to obtain a filament with appropriate geometry. the vertical direction. Thus, a truly round filament will
have an ovality of zero as both diameters are equal.
The diameter had a normal distribution with an aver-
Printing trials age diameter calculated to be 1,732 mm and a standard
Each material to be used in an FFF machine needs to deviation of 0,020 mm. This diameter is acceptable
be checked for the conditions which lead to good qual- even though one of the standard diameter sizes for FFF
ity printed parts. Printing trials were performed to deter- filaments is 1,750 mm. It has been observed by other
mine the range of temperature, flow rates and printing researchers [5] that as long as the diameter is not below
speeds at which this novel material can be processed by 1,700 mm, the material should be able to be printable
means of FFF. Printing trials were performed on a Du- without major problems. The ovality distribution was
plicator i3 v2 FFF machine with a nozzle diameter of more or less constant with an average of 0,030 mm and
0,6 mm. The printing surface was glass coated with hair a standard deviation of 0,017 mm, which is sufficiently
spray to enhance the adhesion of the feedstock to the low to consider the filament to be round.
glass. The printed parts were dog-bone specimens with The roundness of the filament can be seen in Figure
a length of 70 mm, a thickness of 3 mm, a width at the 1. One can see that the stainless steel particles (brighter
narrow section of 4 mm and at the wider section of spots) are equally distributed in the filament, which is
12,5 mm. important to have an even distribution of particles in the
The software Slic3r was used to prepare the G-code printed and sintered parts.
for printing. In this software, the following parameters The insert in Figure 1 shows a SEM image where the
were kept constant: layer height of 0,15 mm, first layer good compatibility between the particles and the binder
height of 0,2 mm, infill density of 100 %, rectilinear fill can be seen, since there are no gaps between the binder
pattern for all layers, fill angle of 45°, speed of printing and the individual particles. Additionally, cavities in the
perimeters of 60 mm/s, infill printing speed of 80 mm/s, filament can be seen; these cavities could be the result
extrusion width of the first layer of 200 %, infill overlap of air trapped during the production of filaments via ex-
of 15 %, and printing surface temperature of 60 °C. The
varied FFF printing parameters are shown in Table 2.

Table 2 Investigated printing parameters

Extruder temp. Flow rate Printing No. of perim-
/ °C /% speed / % eter lines
1. 260 100 50 1
2. 250 100 50 1
3. 240 100 50 1
4. 260 150 50 1
5. 260 150 70 2
6. 260 150 60 1
7. 240 150 60 1
8. 230 150 60 1
9. 220 150 70 1
10. 210 150 60 1
11. 260 200 100 1 Figure 1 Overall view and detail of the filament filled with
12. 260 200 100 2 stainless steel particles.

118 METALURGIJA 57 (2018) 1-2, 117-120

 J. GONZALEZ-GUTIEREZ et al.: 3D PRINTING CONDITIONS DETERMINATION FOR FEEDSTOCK USED IN FUSED...

Table 3 (a) Bottom, (b) top and (c) side views of printed parts 1 to 12

trusion. Cavities are undesirable because they weaken (parts 5 and 6). The printing parameters used for part 6
the mechanical properties of the filament, but if there yield acceptable quality. Decreasing the temperature
are not too many cavities and they are small enough, the from 260 to 210 °C, while keeping the flow rate
filament is still printable. As it will be shown in the next constant, generally leads to a decrease in the quality of
section; this was the case for the filaments produced in the printed parts (parts 6 to 10).
this investigation. The part with the worst quality is the part printed at
the lowest temperature (part 10). Since not enough ma-
terial was flowing, there are sections that were not prop-
Printing trials results
erly filled and the strands did not coalesce. The results
The results of the printing trials are summarised in of the last two trials (parts 11 and 12) showed that the
Table 3. In almost all printed parts the layer in contact printing speed can be doubled (compared to part 1) if
with the printing bed had a smooth and continuous sur- the flow rate is doubled, which still leads to acceptable
face (column (a) in Table 3). The two parts with the quality. However, care must be taken when selecting the
worst bottom section are parts 5 and 10. In the case of number of perimeter lines to be printed. For example,
part 5, the printing speed was not properly matched to using two perimeter lines led to improper filling of part
the flow rate of the material, so there are voids between 12, but using one perimeter line leads to good filling in
the layers. In the case of part 10, the extrusion tempera- part 11. This can be attributed to the number of strands
ture was the lowest; therefore, the material did not flow that can be completely fitted in the width of the printed
correctly to coalesce the printed strands. In general, the part; the printer cannot deposit fractions of strands so it
higher the extrusion temperature the smoother the sur- skips a strand if it does not fit. Printing parameters used
face looks (parts 2 to 4). for part 11 yield the best quality.
When the parts are looked at from the top view (col- The side view (column (c) in Table 3) of the printed
umn (b) in Table 3), there is a more noticeable differ- parts helps to determine the stability of the printing pro-
ence among them. For parts 1 to 3, only the extrusion cess. For example, part 1 appears to have good quality
temperature was changed and the other parameters re- from the top and bottom view, but in the side view it is
mained constant. It can be seen that the printed strands visible that there is a space between the printed layers.
did not coalesce as the temperature was lowered from Another example is part 7, whose side view reveals that
260 to 240 °C. When the flow rate was increased (part 4 one layer was not properly printed, so there is a gap be-
compared to part 1), the coalescence between the print- tween two sections of the part. These errors during print-
ed strands improved, but there is a risk of supplying too ing can arise from many places, for example from unex-
much material and blobs of material could appear. pected external vibrations, changes in environmental con-
Blobs can be corrected by increasing the printing speed ditions, or even changes in the geometry of the filament.

METALURGIJA 57 (2018) 1-2, 117-120 119

 J. GONZALEZ-GUTIEREZ et al.: 3D PRINTING CONDITIONS DETERMINATION FOR FEEDSTOCK USED IN FUSED...

Based on the results shown in Table 3, one can esti-

mate a range of printing parameters within which the
highly-filled filament presented can be processed by FFF.
The extrusion temperature can be between 210 and
260 °C. The flow rate multiplier should be between 150
and 200 %. The printing speed should be between 60 to
100 %. The combination that seems to yield the best
quality is using an extrusion temperature of 260 °C, flow
rate multiplier of 200 %, and printing speed multiplier of
100 % and in this case one perimeter line. Because higher
temperature decreases the viscosity of polymers, adding
more material with lower viscosity improves cross-flow
and results in a better contact between strands.

Microscopic morphology Figure 2 Overview and detail of fractured section of printed

part
The good distribution of the stainless steel particles
(brighter spots) in the printed parts can be also seen in the
cross-section of the printed specimens (Figure 2). More-
over, the individual printed layers can be seen in the SEM Acknowledgments
image shown in the insert in Figure 2. The arrows indi-
cate the printed layer thickness. Some defects can also be Research was performed with the financial support
observed as a result of errors occurred during printing, of the European Commission under grant agreement
for example small voids between the layers. No. 636881 (www.repromag-project.eu). Authors thank
Walter Rath for producing the microscopy images.
CONSLUSION
REFERENCES
A new material to be used in fused filament fabrica-
tion (FFF) has been developed. This new material is [1] A. Boschetto, L. Bottini, F. Veniali, Robotics and Compu-
ter-Integrated Manufacturing 41 (2016) 37, 92-101.
highly filled with 17-4PH stainless steel particles and DOI:10.1016/j.rcim.2016.03.004
could be used for the production of full metal parts after [2] A. Zocca, P. Colombo, C.M. Gomes, J. Günster, D.J. Gre-
it is shaped, debound and sintered. en, Journal of the. American Ceramics Society 98 (2015),
It was observed that this material can be processed 1983-2001. DOI:10.1111/jace.13700
by FFF when the extrusion temperature is between 210 [3] M. Allahverdi, S.C. Danforth, M.A. Jafari, A. Safari,
and 260 °C, the flow rate multiplier between 150 and Journal of the European Ceramic Society, 21 (2001), 1485-
1490. DOI:10.1080/00150190108225177
200 %, and the printing speed between 60 to 100 %. The [4] N. Nestle, M. Hermant, K. Schimdt, World Patent WO
best quality was observed when using an extrusion tem- 2016/012486 A1 (28.01.2016).
perature of 260 °C, flow rate multiplier of 200 %, and [5] H. Valkenaers, F. Vogeler, E. Ferraris, A. Voet, J.P. Kruth,
printing rate multiplier of 100 %. The printed surface Proceedings, 10th International Conference on Multi-Ma-
was a glass mirror heated to 60 °C. Under such condi- terial Micro Manufacture, S. Azcarate and S. Dimov (ed.),
tions an acceptable surface on the top and bottom sec- Research Publishing, San Sebastian, Spain, 2013, pp. 1-14.
[6] J. Gonzalez-Gutierrez, C. Kukla, S. Schuschnigg, I. Dure-
tions of the printed parts is achieved. tek, C. Holzer, Proceedings, 25. Leobener Kunststoff-Kol-
The stainless steel particles are evenly distributed in loquium: Kunststoffgerechte Bauteilentwicklung - vom
the filament and also in the printed part. So it could be Werkstoff zum Produckt, G. Pinter and G. Pilz (ed.), Mon-
expected that sintering can be performed without de- tanuniversitaet Leoben, Leoben, 2016, pp. 187-188.
stroying the printed shape as this binder system has al-
ready been proven to be solvent debindable and sinter- Note: Responsible person for English translation is S. Kereković, Croa-
able [6]. tia

120 METALURGIJA 57 (2018) 1-2, 117-120