Impact of SLM build parameters on the surface

quality
Verfasser des Aufsatzes: Hötter, Jan-Steffen1; Ziebura, Dawid1; Gebhardt, Andreas1
1
Fachhochschule Aachen

© 2013 Jan-Steffen Hötter; Lizenznehmer RTejournal, weitere Informationen sind zu finden unter:
http://www.dipp.nrw.de/service/dppl/

--------------------------------------------------------------------------------------------------------------------------

Abstract
Additive manufacturing by melting of metal powders is an innovative method to create one-
offs and customized parts. Branches like dentistry, aerospace engineering and tool making
were indicated and the manufacturing methods are established.

Besides all the advantages, like freedom of design, manufacturing without a tool and the
reduction of time-to-market, there are however some disadvantages, such as reproducibility
or the surface quality. The surface quality strongly depends on the orientation of the
component in the building chamber, the process parameters which are laser power and
exposure time, but also on the so-called “hatch”-strategy, which includes the way the laser
exposes the solid areas.

This paper deals with the investigation and characterization of the surface quality of
generated parts produced by SLM. Main process parameters including part orientation, part
size and hatch strategies are investigated and monitored.

The outcome is a recommendation of suitable hatch strategies depending on desired part

properties. This includes metered values and takes into account process stability and
reproducibility.

1. Introduction
Surface quality of AM processes and especially of metal sintering (SLM) is still an issue.
While there are many post processing approaches for macro-scale parts, meso-scale parts
can hardly be post-processed. Consequently the desired geometric precision must be made
in-process which, as a precondition, requires a smooth surface quality.

The paper deals with how the build parameters can be influenced in order to guarantee a
part precision that meets the theoretical accuracy of the machine. In detail different
construction angles and exposure strategies are built and investigated. As a result,
recommendations regarding the desired part properties of filigree parts are made.

2. State of the Art

The component’s surface quality highly depends on machinespecific parameters. For this
reason, research is focused on the layer thickness and machinespecific hardware to avoid
the negative effects of the stair stepping effects. [GEB-07] Besides that and to avoid
designing a new machine with high investment costs, the first step is the optimization of the
process parameters with current machines, which is in the focus of this paper. E. Yasa et al.
[YAS-12] investigated a so-called “two-zone-technique” that includes a laser surface
remelting of the boundary area and the upper surface and ends in a better surface quality
and a higher density. E. Yasa et al. [YAS-09] investigated the influence of several hatch-
strategies on the roughness of the topsurface. The influence of build direction on mechanical
properties wich has a significant influence on the bending strenght and elongation was
analyzed by J.Delgardo et al. [DEL-11]. H. Krauss [KRA-13] investigated the
manufacturability by characterizing single melt tracks and thin wall structures. I. Smurov et al.
[SMU-07] examined the impact of the builded part mass and the thermal conductivity of the
used material wich increases layerwise. The relation between the exposure strategy and the
heat distribution in the molten layer using Silver powder was analyzed by Fateri et al. The
conclusion was an exposure strategy that avoids negative effects like buckling effect. [FAT-
12] The impact of alloying elements to reduce the bulling-effect was tested by Kruth et al
[KRU-03]. With the addition of phosphor in the powder, the surface tension was set down
and the bulling-effect was eliminated. G. Strano et al. [STR-12] investigated the impact of the
stair-stepping effect and adhesions in relation to the construction angle concluded in the
surface roughness. The investigation of a recommendation of suitable hatch strategies
depending on desired part properties including metered values and taking into account
process stability and reproducibility cannot be found in literature yet.
3. Experimental Set-Up

Fig. 1: SLM Process, scheme

For the study a SLM-50 Desktop machine by Realizer was used. No hardware changes were
made. A scheme of the machine is displayed in Figure 1.
The heart of the SLM machine is the build space that contains the powder bed on top of
which the actual layer of the part is processed. This is done by a laser beam that locally
melts the powder according to the contour obtained from the part’s CAD data, leaving a solid
track after re-solidification.
To process a subsequent layer, a piston, that defines the bottom of the powder bed and the
lower edge of the part respectively, is lowered by the amount of one layer thickness. New
powder is applied by means of the double bladed recoating unit and the geometry of the next
layer is submitted to the scanner and processed according to the preceding one. So the part
is made successively layer by layer from the bottom to the top. To prevent oxidation of the
powder, the process takes place under shielding gas. Typically Argon or Nitrogen is used.
The machine is equipped with a Yb:YAG fiber laser that delivers a 100W laserbeam at a
wavelength of 1070 nm. The laser spot size is ≈15 µm. The recoating system delivers a layer
thickness of 25 µm.
The analyzed material is a tool steel (1.2344) that is typically used for tooling applications,
e.g. stamp, die cast and injection mould tools, because of its high hardness (about 50 HRC).
The powder size distributes in a range of 20 to 65 µm. In table 1, the composition is shown.
Element [%]

Fe rest

Cr 4,75 - 5,5

Mo 1,1 - 1,75

Si 0,8 - 1,2

V 0,8 - 1,2

C 0,32 - 0,45

Mn 0,2 - 0,5

Table 1: Composition of the applied tool-steel [REAc-12]

To investigate the impact of different construction angles on the surface quality of the
produced part, simple cuboids with different angles were analyzed.
This geometry allows to examine different effects such as deformation and accuracy, but
also metered values for the surface quality, in this case the mean roughness index. The
metered value is a combination of the stair step-effect which is shown in Figure 2 and
adhesions. The adhesions are caused by the significant difference in thermal conductivity
between solid material and powder in the semi-melted phase.

Fig. 2: Stair-Step-Effect depending on angle

To analyze a wide range of different angles related to the building direction (typically z-axle),
a range between 45 and 90 degrees is chosen. Due to thermal stress, which causes
deformation, angles smaller than 45° are not analyzed because the surface that is
investigated has to be supported. Additionally to the construction angle, the volume of the
part is another aspect that has to be studied. For that reason, two different sizes of cuboids
with dimensions of 5x5x10 mm and 10x10x20 mm in x-, y- and z-direction are produced. In
Table 2, a group of five parts is defined as standard. The cuboid block 90°/85° is build using
two different angles to increase the angle range and to investigate the impact of small angle
changes on the surface quality.

Size 10x10x20 5x5x20

Device

Block 45° X X

Block 65° X X

Block 90°/85° X

Table 2: Standardgroup per Experiment

The mean surface roughness index is measured with a stylus instrument Perthometer M2 of
Mahr GmbH.
The four surfaces of the produced parts are named as followed: Inner surface, outer surface,
right and left. To avoid measuring inaccuracy, each surface is measured at three different
positions with a measurement distance of 5,6 mm.

Fig. 3: Description of different surfaces and measure range

In Figure 3, the divided surfaces are visualized and a typical test building platform is shown.
Each surface has three different measurement areas (yellow labeled) to avoid adulteration of
the metered values.
4. Experimental procedure
Table 3 shows important initial process parameters. These parameters are based on prior
studies and own findings.

Start parameters

Powdersize: max. 63 µm

Laserpower [W] 48

Scanspeed [mm/s] 250

Layer Thickness [µm] 25

Hatchdistance [µm] 110

Laserspot [µm] 10 - 20

Preheating [°C] 150

Table 3: Start parameters/Initial settings

In the following sections, different build parameters and their impact on the surface quality of
the produced parts are explained.

Exposure order

The exposured layers are divided into two exposure groups: the outer solid area, which is the
boundary of the part and the hatch area, which describes the solid part of each layer.

Fig. 4: Relevant exposure groups

The exposure order is changed in two ways:

Outer solid -> hatch solid
Hatch solid -> outer solid
The first test with an initial exposure of the outer line shows the best results comparing the
different exposure orders (Fig. 5).
 Fig. 5: Surface with high density

Using the optimized parameters for a single track, the outer solid line has a high density and
the best measured surface quality. The hatch area is able to transfer the heat into the
already solidified line without increasing the surface quality and creating more adhesions.
The second test which includes the change of the order of the exposure of the solid and
hatch area, shows pores and holes on the surface (Fig. 6).

Fig. 6: left: surface with low density; right: 200-times magnified

Conspicuously, the highest amounts of adhesions are found right on top of the pores. The
reason for that can be found in a bad connection between the hatched area and the outer
solid area shown in Fig. 7. Due to the alternating hatch vector generation, holes created in
the layer below are recoated with powder and remelted with the normal parameter set.
Because of the lower thermal conductivity of the powder compared to the solid material, the
powder inside the hole in the layer below is remelted but not with full depth penetration. The
surface tension of the melt pool causes the powder inside the hole to solidify as a ball, which
is attached at the outer solid line. This results in a rough surface.
A higher laser power in combination with a faster scan speed resulting in a remelting of the
hatch area and a change of the distance between hatch and solid area can avoid this effect,
but is not part of this paper.

Fig. 7: (1) Top view of current layer; (2) alternating building

Measuring the surface quality with the experimental set-up explained above, two
characteristics can be observed.
The first characteristic is the dependence between the adhesions and the stair-stepping-
effect. Because of the layer wise manufacturing and the powder based process, parts
designed with a small angle related to the building direction (see Fig. 2) have layers that are
partly created on solid material and partly on loose powder.

Fig. 8: Layer melted on different foundation material

Due to the difference in thermal conductivity of loose powder and solid material and using the
same parameter set for these areas, areas with an overlap to loose powder have more
adhesions than areas built on solid material (Fig. 8).
The second characteristic is the dependence between the surface roughness and the
starting point of the laser. Reasons again can be found in the different heat conductivities of
loose powder and solid material.

At the beginning of the exposure of the layer, the heat needed for melting the powder can
only dissipate to the layer below. Because of the bad heat conductivity in the layer and the
powder size distribution, smaller particles around the melted line are semi-melted and
attached at the surface of the melted line.

Fig. 9: left: during exposure; right: after exposure with more adhesions on the startside

After the solidification of the first line, the next melted line dissipates the applied energy in the
solid first line and a better melt pool can be monitored (Fig. 9).
Different mean roughness indexes are measured because of the dynamic change of the heat
conductivity of the manufactured layers.

Device B 45 B 65 B 90/85
5x5x20 10x10x20 10x10x20

Surface

Outer Surf./90 ° 8,009 6,381 5,375

Right 6,794 7,23 6,975

Left 5,872 5,45 5,523

Inner Surface 10,54 6,176 6,017

Table 4: Metered surface roughness Ra in µm in relation to different surfaces

To avoid the effect of different surface roughnesses of the produced part, the hatch strategy
has to be optimized.

4.1 Hatch strategy variation

Fig. 10: Different hatch strategies

The hatch strategy influences the surface quality of the produced part in a strong way. To
understand the relation between hatch strategy and surface quality, different hatch strategies
are tested. In Fig. 10, the investigated hatch strategies are shown.
The standard hatch strategy is used to test the exposure order and contains different surface
roughnesses in relation to the starting point of the laser.
To avoid this effect, the so-called “stripes with checkerboard” strategy is used. Regarding this
exposure strategy, the layer is divided into defined stripes (size is changeable, in this study it
is fixed to 2 mm). This stripes are again divided into small fields (again: size is changeable,
fixed here to 2 mm), so the surface of the layer is divided like a checkerboard pattern.

Fig. 11: left: Standard-hatch; right: checkerboard with stripes strategy

This checkerboard pattern is exposed field by field, but always with one free field between
the exposed fields to reduce a heating-up of the whole part because this would result in a
worse surface quality. In Fig. 11, the standard hatch and the checkerboard with stripes
strategies are compared.
The surface quality increases because the heat distribution in the melted layer is improved
and each checkerboard field has the possibility to cool down. Another effect of this strategy is
that the surface roughness is independent of the starting point of the laser due to the better
heat distribution in the layer compared to standard hatch, which includes a high heating-up of
the whole part.

Device Standard- Checker Percentaged

-board in deviation in
Hatch in
µm %
µm
Surface

Outer Surf./90 ° 5,375 4,867 9,45

Right 6,975 4,3 38,35

Left 5,523 4,389 20,53

Inner Surface 6,017 4,294 28,64

Table 5: Metered surface roughness Ra: comparison of standard-hatch and checkerboard-strategy

Another hatch strategy is the so-called “random checkerboard” exposure order. In contrary to
the checkerboard strategy, the checkerboard fields are not exposed stripe per stripe, but
randomly all over the layer. By using this exposure order, the heat distribution in the layer
can be optimized. Unfortunately, this strategy produces worse surfaces qualities than the
checkerboard with stripes method. The reason can be found in random non-overlap of the
different checkerboard fields. The exposed field can cool down but has no connection to the
outer solid boundary or other already solidified fields. So, the shrinkage and the heat
distribution in the field are non-linear and cause a bad surface quality. The checkerboard with
stripes strategy combines a preferable heat distribution and a controllable shrinkage due to
the connection between the fields in the generated stripes.
This shrinkage phenomenon can also be seen in Table 6. The dimensions are metered with
a caliper. The values are the differences of a target-performance comparison. R-L is the
dimension between right and left surface, I-O is the dimension between inner and outer
surface (see also Fig. 3).
The metered values marked with green, show the best results. Parts produced using the
stripes with checkerboard strategy have the highest accuracy. The reason for that can be
found in the stripe per stripe exposure strategy that, on the one hand, reduces adhesions
and, on the other hand, reduces thermal stress.
 Size difference in mm per hatch strategy

Experiment Standard- Stripes with Random

Hatch Checkerboard Checkerboard
Block

with math. size

B45 10x10x20

R-L = 10 - 0,08 0,11

I-O = 7,071 - 0,279 0,449

B45 5x5x20

R-L = 5 0,07 0,13 0,15

I-O = 3,536 0,18 0,174 0,294

B65 10x10x20

R-L = 10 0,05 0,01 0,12

I-O = 9,063 0,23 0,137 0,187

B65 5x5x20

R-L = 5 - 0,04 0,11

I-O = 4,532 - 0,108 0,198

B90/85 10x10x20

R-L = 10 0,0 0,04 0,05

Table 6: Measured difference at different parts with different hatch strategies

The random checkerboard strategy evokes stress in the solidified areas, caused by the
random light exposure. This leads to uneven deformations.

4.2 Process duration in comparison

In order to take into account the economic aspects, the required construction times,
respectively for the exposure- and hatch-strategy, are being compared. It shows that less
roughness and process safety are in conflict with economic viability. Depending on exposure-
and hatch strategy the respective exposurelengths per blocksize are measured. The results
are shown in table 7.

Exposurelength per Blocksize and Experiment [s]

 Size

Experiment 10x10x20 5x5x20

Exp.1: Standard-Hatch 3,6 0,98

Exp. 4: Stripes with Checkerboard 4 1

Exp. 5: Random Checkerboard 4,4 1,1

Table 7: Exposurelength per Experiment and Blocksize

Depending on the multiple exposures or hatch strategies, different exposurelengths can

appear on the respective layer.

4.3 Roughness depending on construction angle

After taking all the experiments, which have been classified by construction angle, into
account, the results show that an angle between 45° to 65° increases the roughness
drastically. The high roughness of cuboidal blocks with a construction angle of 90° is due to
the amount and size of the adherent particles. Cuboidal blocks with a construction angle of
45°are strongly influenced by the stair-stepping-effect. Table 8 shows cuboidal blocks from
three different experiments.

Averaged Ra [µm] depending on angle

Device Outer Surface Inner Surface

B45 10x10x20 7,473 9,813

B65 10x10x20 5,462 7,436

B90/85 10x10x20 5,352 5,673

B45 5x5x20 6,922 7,869

B65 5x5x20 5,833 6,796

Table 8: Averaged Ra depending on angle

4.4 Roughness depending on construction volume

The influence of the construction volume on the roughness has been analyzed with two
different cuboidal block sizes. The respective construction angles were 45° and 65°.
According to table 9, in the five experiments in which a cuboidal block of 10x10x20 is used,
the results show that the averaged roughness was 6% higher.
 Volume effect on averaged Ra [µm]

Blocks with 10x10x20 7,885

Blocks with 5x5x20 7,435

Difference in % to B 10x10x20 5,7 %

Table 9: Volume effect on averaged Ra

Although cuboidal blocks with dimensions of 10x10x20 are able to transfer the applied
energy faster into the component, is the extended exposurelength a cause for an increased
heat accumulation at the sides with a starting point (right hand side facing outwards) and on
the inside. The reason for that is that overhanging parts of the layer poorly conduct the heat
downwards. Table 10 shows the reason for the big differences of about 6% according to the
separate sides.

Volume effect on averaged Ra [µm] per side

Side
Outer
Block Surf. Right Left Inner Surf.

Blocks with 10x10x20 6,845 5,525 5,292 8,925

Blocks with 5x5x20 6,733 5,091 5,115 8,138

Difference in % to

B 10x10x20 1,63 % 7,85 % 3,35 % 8,83 %

Table 10: Volume effect on average Ra per Surface

In order to determine the influence of the volume accurately, the averaged Ras of the entire
test objects are itemized by the three measuring ranges. It turns out, that only slight
variations, as a result of the first layer being built on loose powder, occurred. Therefore,
sufficient material in the first third of the component is available to keep the heat
accumulation on a low level.
Averaged Ra [µm] on different checkpoints

Checkpoints Bottom Side Top

middle middle

Mean surface roughness 6,3662 6,26 6,26

index

Percentaged deviation 0% 2% 2%
according to Bottom down
measured area

Table 11: Averaged Ra on different checkpoints

5. Conclusion
In the context of this experiment, a lot of crucial influencing factors are identified to provide
an acceptable surface roughness quality and process reliability.
Variations of the construction angle and cuboid volume show different connections between
variable thermal conductivity effects of the solidified material and the loose powder by
reference to the exposed area, the process based stair-stepping-effects and the resulting
adhesion.
By analyzing the various experiments the impact of exposure and hatch strategies on the
surface quality are shown.
Exposing the outer outline as a shaping element followed by exposing the surface is proven
to be the most approperiate exposure strategy.
Considering the discussed hatch strategies, it becomes clear that two types of strategies are
to be prefered in terms of surface quality and shape accuracy. The Stripes with
Checkerboard strategy leads to satisfying results in all the crucial requirements. On the one
hand, the strategy provides good heat dissipation during the exposure length while reducing
the amount of adhesions. On the other hand, it provides minor geometrical deviations in
combination with needing little time for the building process.
The Standard-Hatch and Random Checkerboard strategy generated only conditionally
acceptable results. Because of the area-wide exposure from one side to another the
Standard-Hatch strategy results in big differences between the roughnesses of the opposing
sides. However, a very consistent surface on the last layer can be achieved which can be
welcomed depending on the area of application. Exposing single areas in a random order,
the Random Checkerboard strategy leads to large geometrical deviations.
The different combination of exposure- and hatch strategy, for building standard units of
three big and two small cuboids, consisting of 857 layers, can result in a time difference of up
to two hours.
This is a huge expenses factor and has to be considered depending on the requirements of
the component.
Taking all into account, this experiment identifies important parameters for processing tool
steel. Crucial connections are pointed out, enabling to design the construction phase
according to the requirements of surface roughness, process safety and economic viability
and giving a recommendation.
Practical recommendations can only be given in combination with the desired component
geometry and the desired requirements.
For example, very small components or components with low demands on surface quality
and shape accuracy should be produced using the hatch strategy Standard Hatch. Larger
components with a high accuracy need should be produced using the hatch strategy Stripes
with Checkerboard.
These examples are to point out the variety of possible combinations of strategies, which
should be chosen according to the given results.

6. Future Prospects
As this experiment showed, there are numerous parameters which in various cooperations,
can influence the building process and the outcoming result significantly. Regardless of
machine-specific restrictions, the findings can be transferred to other field of applications and
provide tendencies for future developments.
The conducted experiments give practical recommendations and provide a basis for further
research approaches.
For instance, further researches could be extended to the spectrum of the construction angle
in order to give more specific propositions and recommendations in regard to the different
combinations of strategies. Analyzing additional geometries, e.g. cones or pyramids, in
different variations could provide propositions in terms of the construction angle and heat
dissipation in combination with the available component’s mass.
Furthermore, the results of this paper showed that the exposure order can influence the
component’s quality significantly. One idea, in order to achieve a satisfying result, could be to
use the already solidified material to conduct the heat, preventing heat accumulation.
Another topic which could be established from this paper is to generate so-called
heatgroups, which allow to automatically vary the laser parameters. At the moment, a fixed
laser power is set, guarantying to fuse the exposed material. Depending on the conditions,
e.g. loose powder or a low component volume, this can lead to negative effects. Heatgroups
are meant to provide specific laser parameters according to the present conditions. Using
this method the exposed surface is divided into single dots. Depending on their position and
surrounding different laser parameters are assigned. This could counteract the set of
problems caused by insufficient or exceeding laser power.
A numerical heat transfer analysis could strengthen the ideas displayed in this paper and
support the experimental investigations.

7. References
[DEL-11] Delgado J. et al.,"Influence of process parameters on part quality and
mechnanical properties for DMLS and SLM with iron-based materials", Springer, 2011
[FAT-12] Fateri, M. et al.: "Experimental and Theoretical Investigation of Buckling
Deformation of Fabricated Objects by Selective Laser Melting", In Physics Procedia, 2012,
Volume 39, Pages 464-470
[GEB-07] Gebhardt, Andreas: "Generative Fertigungsverfahren": 3.Auflage, München,
Carl Hanser Verlag, 2007
[HOE-12] Hötter J, et al.: "Prozessoptimierung des SLM-Prozesses mit hoch-reflektiven
und thermisch sehr gut leitenden Materialien durch systematische Parameterfindung und
begleitende Simulationen am Beispiel von Silber". RTejournal - Forum für Rapid
Technologie, Vol. 2012.
[KRA-13] Kraus H. et al. „Investigations on manufacturability and process reliability of
selective laser melting“, ScienceDirect, 2013
[KRU-03] Kruth, J.P. et al.: "Selective laser melting of iron-based powder", Journal of
Materials Processing Technology, 2003
[REAc-12] Realizer GmbH: personal information
[SMU-07] Smurov. I. et al.: "Parametric analysis of the selective laser melting process",
ScienceDirect, 2007
[STR-12] Strano,G. et al., "Surface Roughness analysis, modelling and prediction in
selective laser melting", Journal of Materials Processing Technology, 2012
[YAS-09] Yasa, E. et al.,"Investigation of laser and process parameters for Selective
Laser Erosion", Elsevier, 2009
[YAS-12] Yasa, E. et al.,"Microstructural investigation of Selective Laser Melting 316L
stainless steel parts exposed to laser re-melting", ScienceDirect, 2012
8. Kontaktangaben
Jan-Steffen Hötter, M. Eng.
Fachhochschule Aachen
Goethestraße 1
52062 Aachen
E-Mail: hoetter@fh-aachen.de
Web: http://www.fh-aachen.de/

Dawid Ziebura, B. Sc., IWE

Fachhochschule Aachen
Goethestraße 1
52062 Aachen
E-Mail: ziebura@fh-aachen.de
Web: http://www.fh-aachen.de/

Prof. Dr.-Ing. Andreas Gebhardt.

Fachhochschule Aachen
Goethestraße 1
52062 Aachen
E-Mail: gehardt@fh-aachen.de
Web: http://www.fh-aachen.de/