materials

Article
An Assessment of Subsurface Residual Stress
Analysis in SLM Ti-6Al-4V
Tatiana Mishurova 1, *, Sandra Cabeza 1 , Katia Artzt 2 , Jan Haubrich 2 , Manuela Klaus 3 ,
Christoph Genzel 3 , Guillermo Requena 2 and Giovanni Bruno 1,4
1 Bundesanstalt für Materialforschung und-prüfung (BAM; Federal Institute for Materials Research and
Testing), Unter den Eichen 87, 12205 Berlin, Germany; sandra.cabeza@bam.de (S.C.);
giovanni.bruno@bam.de (G.B.)
2 Institute of Materials Research, German Aerospace Center (DLR; Deutsches Zentrum für Luft-und
Raumfahrt), Linder Höhe, 51147 Cologne, Germany; katia.artzt@dlr.de (K.A.); jan.haubrich@dlr.de (J.H.);
guillermo.requena@dlr.de (G.R.)
3 Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Hahn-Meitner Platz 1, 14109 Berlin,
Germany; klaus@helmholtz-berlin.de (M.K.); genzel@helmholtz-berlin.de (C.G.)
4 Institute of Physics and Astronomy, University of Potsdam, Karl-Liebknecht-Straße 24/25,
14476 Potsdam, Germany
* Correspondence: tatiana.mishurova@bam.de; Tel.: +49-30-81044527

Academic Editor: J. Paulo Davim

Received: 3 February 2017; Accepted: 22 March 2017; Published: 27 March 2017

Abstract: Ti-6Al-4V bridges were additively fabricated by selective laser melting (SLM) under
different scanning speed conditions, to compare the effect of process energy density on the residual
stress state. Subsurface lattice strain characterization was conducted by means of synchrotron
diffraction in energy dispersive mode. High tensile strain gradients were found at the frontal surface
for samples in an as-built condition. The geometry of the samples promotes increasing strains towards
the pillar of the bridges. We observed that the higher the laser energy density during fabrication, the
lower the lattice strains. A relief of lattice strains takes place after heat treatment.

Keywords: selective laser melting; additive manufacturing; heat treatment; Ti-6Al-4V; synchrotron
X-ray diffraction; residual stress

1. Introduction
Additive manufacturing (AM) is still a rather new technique for producing metallic components.
As opposed to conventional subtractive manufacturing technologies like milling or drilling, AM allows
the building up of components by means of wire-fed, powder-fed, or powder-layer processes.
AM opens up new opportunities for manufacturing highly complex, load- and lightweight optimized
near-net shape components, which are particularly interesting for small series of parts. Notably, the
manufacturing costs and lead time in an AM scale, mainly with part mass, and are only weakly
dependent on the component complexity, in contrast to conventional processing technologies [1].
Among the family of additive manufacturing technologies, selective laser melting (SLM) has gained
wide acceptance. It is a powder-bed fusion technique that enables the production of very fine and
hollow structures, such as inner cooling channels for turbine blades or high-speed cutting tools.
However, even today, AM technologies and materials are still far from mature, and hinder AM
from being exploited at its full potential. In SLM, multiple process parameters (e.g., laser power,
powder layer thickness, laser scanning velocity, and scan strategy), as well as the rheological powder
properties, influence the resulting material properties, i.e., porosity, cracks, surface properties, and
microstructure, and hence, the mechanical properties. Additionally, high residual stresses are known

Materials 2017, 10, 348; doi:10.3390/ma10040348 www.mdpi.com/journal/materials

Materials 2017, 10, 348 2 of 14

to develop during the building process [2], affecting the mechanical performance, and possibly leading
to undesired shape distortion, detachment from support structures, and even crack formation.
One particularly well-studied material for SLM is the α + β titanium alloy Ti-6Al-4V. The alloy has
been developed to satisfy the needs of the aerospace industry, and it is the most widely used titanium
alloy. This industry sector has gained more than six decades of experience with this material. As a
consequence, the aerospace industry aims at accelerating the development and certification of the AM
process by exploiting its expertise on this very alloy [3]. However, the mechanical performance of AM
Ti-6Al-4V differs from that of forged or cast counterparts, owing to the different thermo-mechanical
history introduced during manufacturing [4]. This necessitates the development of a new expertise for
AM materials and their post-treatment possibilities.
Several works focus on the evolution of the microstructure during and after SLM production,
and its connection to mechanical behaviour [5–9], generally showing that the Ti-6Al-4V as-built
microstructure consists of brittle α0 -martensite (hexagonal lattice) within prior-β grains, with a
martensitic size significantly depending on the process parameters [5]. Phase transformations can be
induced by heat treatments leading to a more ductile microstructure consisting of stable α (hcp) and
β (bcc) phases. Commonly, the heat treatment is conducted after the building process (“extrinsic” or
“ex-situ” after the AM process) (e.g., [10–13]), but approaches of in-situ heat treatment by preheating
the baseplate or build space also exist [14,15].
Besides defects and the microstructure, the residual stresses in AM materials are critical for
their mechanical performance and for retaining the ability to achieve near-net shape geometries
(see e.g., [8,16–18]). High cooling rates at the order of 103 K/s to 108 K/s occurring in the SLM
process [13,19–21], lead to significant residual stresses that can result in the distortion of samples
and parts. Previous studies show that thermal residual stresses in SLM parts are primarily caused
by two mechanisms: the temperature gradient, and the cool-down phase of the molten top layer [2].
Additionally, the influence of the number of layers, the thickness of the layers, the geometry, and
the scanning strategy for Ti-6Al-4V parts have been investigated [2,16]. The results of the cited
works show an increase of residual stress for a large number of layers (i.e., larger sample height),
while the preheating of the base plate during scanning decreases the residual stress. Finite element
thermo-mechanical models of residual stress have been reported in [22], showing the dependence of
the residual stress state on the scanning strategy. Typically, the highest residual stresses were reported
along the scanning direction. Besides the adjustment of the SLM scan strategies [16], the most common
and versatile approach to reduce residual stresses of AM parts, prior to separation from the baseplate,
is a “stress relief heat treatment” (e.g., 650 ◦ C for 3 h) [8,9,23].
In most of the cases, residual stresses were determined by destructive methods (contour
method [24], hole drilling) or laboratory X-ray diffraction [2,25]. In particular, the latter only presents
information on the outer surface layers (i.e., a few tens of microns).
In the present work, we investigated the development of residual stresses in SLM-fabricated
bridge specimens as a function of the laser process energy density introduced during manufacturing,
in order to identify suitable SLM processing parameter limits. To fulfil this aim, Ti-6Al-4V samples in
the form of bridges were used also to assess the parameters influencing the residual stress state and
to guide the development of SLM processes [16,26]. We first studied the curvature of bridges, which
occurred after their removal from the baseplates (different bridges were produced by varying the laser
scan speed). Subsequently, two bridges built at two selected conditions in the processing window
were investigated by synchrotron energy-dispersive diffraction, to study the residual stress state in the
subsurface region in further detail. Energy-dispersive diffraction performed in reflection mode using
high energy synchrotron radiation allowed us to sample the materials near the surface region and to
record complete diffraction patterns under fixed (but arbitrary) scattering angles [27]. Furthermore,
the use of synchrotron radiation allows the use of small gauge volumes, and thereby enables high
spatial resolution in the subsurface region.
Materials 2017, 10, 348 3 of 14
Materials 2017, 10, 348 3 of 14

2. Materials and Methods

Bridge-shaped specimens were produced under an argon atmosphere using a SLM SLM Solutions
Solutions
machine with
280HL machine with aa single
singleIPG
IPGfibre
fibrelaser
laser(maximum
(maximumpower
power400
400W).
W).The
Thebaseplate
baseplatewas
washeated
heatedtoto
a
a fixed temperature of ◦
200 °C during the SLM process. Plasma atomized (AP&C)
fixed temperature of 200 C during the SLM process. Plasma atomized (AP&C) Ti-6Al-4V ELI grade Ti-6Al-4V ELI
grade
23 23 powder
powder with a with a particle
particle size of dsize
90 <of
50dµm,
90 < 50 μm, as determined
as determined by laser by laser diffraction
diffraction (Beckman (Beckman
Coulter
Coulter was
GmbH), GmbH),
usedwas used
for the SLM forprocess.
the SLM Theprocess. The bridgehad
bridge specimens specimens had
a length of 20 amm,
length of 20ofmm,
a height 8 mm, a
height
and of 8 mm,of
a thickness and10 amm
thickness
(Figureof1a)
10[16].
mm (Figure 1a) [16].

Figure 1. Determination of the deflection angle α. For the part of the profile inside the selected areas
left and
on the left and on
on the
the right
right side,
side, the
the slopes
slopes mm11 and
and mm22 are calculated
calculated with respect thethe longitudinal
longitudinal
are converted
direction. The slopes are converted to todeflection
deflectionangles
angles(here: 0.63◦ and α
(here:αα1 1==−−0.63° α22 = +0.81 ◦ ). (a)
+0.81°). (a) Top
area LSM;
surface LSM; (b) selected area LSM; (c)
(c) height
height profile.
profile.

Two sets
Two sets of
of bridge-samples
bridge-samples were were produced
produced using
using different
different scanning
scanning velocities
velocities at
at aa constant
constant
laser power
laser power for
for the
the bulk
bulk processing
processing parameters
parameters (Table
(Table 1),
1), to
to systematically
systematically vary vary the
the energy
energy density
density
introduced by SLM. For all samples, a set of two contours and one intermediate
introduced by SLM. For all samples, a set of two contours and one intermediate line were initially line were initially
produced, before
produced, before the
the sample
sample bulkbulk was
was processed
processed (with
(with specific
specific parameters).
parameters). The The contour
contour parameters
parameters
(100 W and 525 mm/s scan velocity) were fixed for this first study. For all bridges,
(100 W and 525 mm/s scan velocity) were fixed for this first study. For all bridges, a constant layer a constant layer
thickness of 30 μm and a chess-pattern scan strategy with a minimum field size
thickness of 30 µm and a chess-pattern scan strategy with a minimum field size of 5 mm were chosen. of 5 mm were
chosen.
All All were
bridges bridges were
built built directly
directly on the baseplate.
on the baseplate. SupportSupport structures
structures were only were onlyinused
used in the
the centre
centre of the bridge arcs. After SLM fabrication, selected samples were subjected
of the bridge arcs. After SLM fabrication, selected samples were subjected to a heat treatment at ato a heat treatment
at a temperature
temperature of 650
of 650 °C 3for
◦ C for 3 h under
h under vacuum
vacuum conditions,
conditions, to relieve
to relieve residual
residual stresses.
stresses. TheThe porosity
porosity of
of sister samples, produced with the same scanning parameters, has been reported in
sister samples, produced with the same scanning parameters, has been reported in [28]. The porosity [28]. The porosity
volume fractions
volume fractions are
are reported
reported in in Table
Table1.1.

Table 1.
Table 1. SLM
SLM scanning
scanning parameters
parameters employed
employed for for the
the bridge
bridge specimen,
specimen, and
and resulting
resulting porosity
porosity fraction.
fraction.
Error in
Error in the
the determination
determination of
of porosity
porosity lies
lies about
about 0.1
0.1 vol
vol %.
%.

Laser Power, P Hatch, h Velocity, v Focus, f Energy Density, Ev * Porosity

Sample Laser Power, Hatch, h Velocity, v Focus, f Energy Density, Porosity
Sample (W) (mm) (mm/s) (mm) (J/mm3) (vol %) [28]
P (W) (mm) (mm/s) (mm) Ev * (J/mm3 ) (vol %) [28]
A1 175 0.1 200 0 291.7 3.3
A1 175 0.1 200 0 291.7 3.3
A3 175 0.1 400 0 145.8 0.6
A3 175 0.1 400 0 145.8 0.6
A4 A4 175
175 0.1
0.1 500
500 00 116.7
116.7 0.1 0.1
A10 A10 175
175 0.1
0.1 1100
1100 00 53.053.0 0.7 0.7
P
* E* vE=v = , x, x==powder layerthickness
powder layer thickness = 30
= 30 μm.
µm.
· v··h· x

The residual stress state of of the

the samples
samples waswas characterized
characterized for
for three
three conditions:
conditions: (1) as-built on
the baseplate (BP); (2) after stress relief heat treatment on the baseplate (TT); and (3)directly released
(wire-cut) from the baseplate (R) without TT. TT. Measurements
Measurements were performed
performed on the same sample
before (BP) and after cutting from the baseplate (R), while the heat treatment was performed on a
sister sample.
“as-built”surface
The “as-built” surface roughness
roughness on side
on the the surfaces
side surfaces of the
of the SLM SLM
bridge bridge
samples samples
was was
determined
determined
with with
a confocal a confocal
ZEISS LSM 700ZEISS LSM 700 microscope
laser scanning laser scanning
(Carlmicroscope (Carl Zeiss Germany).
Zeiss AG, Oberkochen, AG, Oberkochen,
Height
Germany).
maps Heightsurfaces
from lateral maps from lateral surfaces
were created for small were
areas of 1.2 × 1.2
created formm 2 (pixel
small areassize
of =1.2 × 1.25
× 1.2
1.25 mmµm 2 for
2 (pixel

size = 1.25 × 1.25 μm2 for each optical slice, 1.46 μm in height). The surface roughness parameters Sa
Materials 2017, 10, 348 4 of 14

Materials 2017, 10, 348 4 of 14

each optical slice, 1.46 µm in height). The surface roughness parameters Sa (arithmetic mean height)
and Sq (root mean
(arithmetic mean square
height) height) were mean
and Sq (root calculated
square according
height) wereto ISO 25178, using
calculated the analysis
according software
to ISO 25178,
Confomap ® (Carl Zeiss AG, Confomap
Oberkochen, Germany).

using the analysis software (Carl Zeiss AG, Oberkochen, Germany).
The distortion angles of the bridge-shaped
The distortion angles of the bridge-shaped specimens specimens[16] [16]were
weremeasured
measured at at their
their toptop surfaces
surfaces
with the ZEISS LSM 700 confocal microscope (Figure 1). Incidentally,
with the ZEISS LSM 700 confocal microscope (Figure 1). Incidentally, this can be used for a fast this can be used for a fast
assessment
assessmentofofthe thetrends
trendsof ofresidual
residual stress statestate in in SLM
SLMsamples.
samples.For Forthethereleased
released specimens
specimens (R),(R),
thethe
bottom
bottom facets
facetswere
werealsoalso
measured.
measured. Distortion angles
Distortion were were
angles computed by averaging
computed the angles
by averaging the measured
angles
inmeasured
two rectangular regions towards
in two rectangular regionsthetowards
edges of thethe bridges
edges of thealong the longitudinal
bridges axis (Figure
along the longitudinal axis1b).
(Figure
Each angle1b).was
Each angle was
calculated fromcalculated
the slopesfrommthe 1 andslopes
m2 ,mrespectively
1 and m2, respectively
(Figure 1c). (Figure
This1c). This
procedure
wasprocedure
repeated wasforrepeated
each line forparallel
each linetoparallel
the X to theyielding
axis, X axis, yielding the deflection
the deflection angleangle as a function
as a function of the
of the Z-coordinate. From this, the average over around 800 lines was taken,
Z-coordinate. From this, the average over around 800 lines was taken, and the mean deflection angle α and the mean deflection
ofangle α of the
the bridge andbridge and the standard
the standard deviation deviation were calculated.
were calculated.
Latticestrain
Lattice strain measurements
measurements of of selected
selected samples
samples werewere conducted
conducted at the
at the synchrotron
synchrotron source
source BESSY
BESSY II (HZB, Berlin) on the EDDI (Energy Dispersive Diffraction)
II (HZB, Berlin) on the EDDI (Energy Dispersive Diffraction) beamline. This beamline provides a beamline. This beamline
provides
white beama withwhiteanbeamenergywith rangean energy
of about range of aboutkeV.
10 keV–150 10 keV–150 keV. Thewas
The experiment experiment
performed wasin a
performed in a reflection set-up with a fixed diffraction angle of 2θ = 8°. The prismatic gauge volume
reflection set-up with a fixed diffraction angle of 2θ = 8◦ . The prismatic gauge volume is defined by
is defined by the intersection of the incoming and diffracted beams. Primary slits with an opening of
the intersection of the incoming and diffracted beams. Primary slits with an opening of 500 × 500 µm2
500 × 500 μm² and secondary slits with a vertical opening of 30 μm were used for the determination
and secondary slits with a vertical opening of 30 µm were used for the determination of the transversal
of the transversal strain component. Measurements were carried out with the gauge volume fully
strain component. Measurements were carried out with the gauge volume fully immersed in the
immersed in the sample, but right at the surface. For the longitudinal strain component, the vertical
sample,
opening butof right at the surface.
the primary slit was For the longitudinal
decreased to 350 μm, strain
in order component,
to decreasethe thevertical
length of opening
the gaugeof the
primary
volumeslit wasto
owing decreased
geometryto 350 µm, inGauge
limitations. ordervolume
to decrease
lengths theoflength
3.8 mm ofand
the 2.8
gauge
mmvolume owing to
were obtained
geometry limitations. Gauge volume lengths of 3.8 mm and 2.8 mm were
for the transversal and longitudinal strain components, respectively. For each measuring point, a obtained for the transversal
2
and sinlongitudinal
ψ scan (seestrain [29]) components,
was performed respectively. For each
with 10 ψ-tilts andmeasuring point,time
an acquisition a sin of ψ 30scan
s per(see [29])
tilt.
wasThe performed
bridge-shapedwith specimen
10 ψ-tilts and an acquisition
is depicted in Figuretime 2a, of
with30 as coordinate
per tilt. Thesystem
bridge-shaped
being used specimen
during is
depicted in Figure 2a,The
the measurements. with a coordinate
positions at whichsystem being
residual used during
stresses the measurements.
were measured are shown as The positions
grey circles at
which residual
in Figure 2b. stresses were measured are shown as grey circles in Figure 2b.

Figure2.2. (a)
Figure (a) Schematic
Schematic representation
representation ofof aa bridge
bridgesample
sampleand
anditsitsdimensions (T—transversal,
dimensions (T—transversal,
L—longitudinal, N—normal); (b) specification of mapped area by synchrotron
L—longitudinal, N—normal); (b) specification of mapped area by synchrotron diffraction diffraction and
and specific
specific
points points measured
measured on the
on the lateral lateral surface.
surface.

In the reflection set-up (near-surface residual stress state), the normal stress component is
In the reflection set-up (near-surface residual stress state), the normal stress component is assumed
assumed to be negligible. Therefore, the absolute values of residual strain for the transversal (Figure 3a)
toand
be negligible.
longitudinalTherefore,
(Figure 3b)thecomponents
absolute values
can beof residual
evaluatedstrain for the transversal
by considering (Figurelattice
the strain-free 3a) and
longitudinal
parameters (Figure 3b) components can be evaluated by considering
, which were previously obtained from the measured the strain-free lattice parameters
− sin ψ distributions in
d0hkl , which were previously obtained from the measured d hkl − sin2 ψ distributions in the strain-free
the strain-free ψ∗ direction (biaxial stress state): sin ψ∗ = ϕψ
. The sin ψ method yielded the
−2S1hkl
ψ∗ direction (biaxial stress state): sin2 ψ∗ = 1 hkl . The sin2 ψ method yielded the residual stress in
residual stress in the azimuthal directions (here 2 S2 longitudinal and transversal), from the linear
the azimuthal directions hkl
of dϕψ
extrapolation of vs.(here
sin longitudinal
ψ graphs. Theand transversal),
longitudinal strainfrom the linear
component extrapolation
could not be measured
2
vs.forsin
the ψ graphs.
samples Thea longitudinal
with baseplate (BPstrain component
and TT), due to thecould not be
thickness measured
of the for the samples with a
baseplate.
baseplate (BP and TT), due to the thickness of the baseplate.
Materials 2017, 10, 348 5 of 14
Materials 2017, 10, 348 5 of 14

(a) (b)

Figure 3. Sketch of residual stress measurement by sin22 ψ in reflection set-up for (a) transversal strain
Figure 3. Sketch of residual stress measurement by sin ψ in reflection set-up for (a) transversal strain
component; (b) longitudinal strain component (References system: L stands for longitudinal, T for
component; (b) longitudinal strain component (References system: L stands for longitudinal, T for
transversal, N for normal components).
transversal, N for normal components).

The energy dispersive diffraction technique (the Bragg angle θ is kept fixed) allows us to obtain
theThe energy
signal fromdispersive diffraction technique
each crystallographic (the Bragg
plane at a different angle θ is kept fixed) allows us to obtain
energy:
the signal from each crystallographic plane at a different energy:
6.199 1
Å =6.199 1keV
hkl
d ( A) = sin θ
sin θ Ehkl (keV)
where is the lattice spacing for the crystallographic plane {hkl}, θ is the Bragg’s angle, and
is the energy corresponding to the diffraction of the crystallographic plane {hkl}. It also allows
where dhkl is the lattice spacing for the crystallographic plane {hkl}, θ is the Bragg’s angle, and Ehkl
probing different depths [30], whereby the penetration depth τ for the energy dispersive diffraction
is the energy corresponding to the diffraction of the crystallographic plane {hkl}. It also allows
is determined by [27]:
probing different depths [30], whereby the penetration depth τ for the energy dispersive diffraction is
determined by [27]: sin θ
τ= cos ψ
2μ θ
sin
τ= cos ψ
where θ is the Bragg’s angle, μ(E) is the linear 2µ( EX-ray
) absorption coefficient (energy dependent),
and θψisisthe
where the Bragg’s
angle ofangle,
tilt. µ(E) is the linear X-ray absorption coefficient (energy dependent), and ψ
For residual
is the angle of tilt. stress evaluation, we chose to analyse the non-overlapping diffraction lines of six
crystallographic planesevaluation,
For residual stress of α + α’ hexagonal
we choseTi lattices.
to analyse the non-overlapping diffraction lines of
The calculation of strains and stresses using
six crystallographic planes of α + α’ hexagonal Ti lattices. the sin2  method requires the elastic constants of
theThe
crystallographic plane under
calculation of strains consideration.
and stresses using theDiffraction elasticrequires
sin2 ψ method constants
the(DEC)
elastic of α were of
constants
obtained according to the Eshelby/Kröner model [31,32] (Table 2). Since the reflections of α and α’
the crystallographic plane under consideration. Diffraction elastic constants (DEC) of α were obtained
proved to be indistinguishable in the diffractograms, and no peak asymmetry was observed, only
according to the Eshelby/Kröner model [31,32] (Table 2). Since the reflections of α and α’ proved to
the α elastic constants were considered, to simplify the analysis. The small difference between the
be indistinguishable in the diffractograms, and no peak asymmetry was observed, only the α elastic
elastic constants of α and α’ does not affect the stress values above the error bar, and thereby the
constants were considered, to simplify the analysis. The small difference between the elastic constants
interpretation of the results. The diffraction peaks of the body-centred cubic β phase were not
of α and α’ does not affect the stress values above the error bar, and thereby the interpretation of
detectable, because of its low volume fraction [14,33]. The samples exhibited weak β—Ti peaks only
theafter
results.
heatThe diffraction
treatment, which peaks
wereofnot
theusable
body-centred
for stresscubic β phase
analysis. DECwere not detectable,
and are connectedbecause
with of
its elastic
low volume
constantsfraction
by the[14,33]. Theequations:
following samples exhibited weak β—Ti peaks only after heat treatment,
which were not usable for stress analysis. DEC s1 and s2 are connected with elastic constants by the
following equations: − = and =
hkl νhkl 1 hkl 1 + νhkl
where and − s1 = hkl
are the plane-specific Poisson’s
and sratio
2 =and Young’s modulus, respectively.
E 2 Ehkl
where νhkl and Ehkl are the plane-specific Poisson’s ratio and Young’s modulus, respectively.
Materials 2017, 10, 348 6 of 14
Materials 2017, 10, 348 6 of 14

Table 2. Diffraction elastic constants of α-titanium calculated with the Eshelby/Kröner model.
Table 2. Diffraction elastic constants of α-titanium calculated with the Eshelby/Kröner model.
The single crystal elastic constants were taken from [34].
The single crystal elastic constants were taken from [34].

{hkl}{hkl} −1 ×−110×−10
−s1 (MPa
−s1 (MPa 6 ) −6) 1 −1s2×(MPa
s2 (MPa½ 10−6 )−1 × 10−6)
2
100 100 2.979
2.979 12.026 12.026
002 002 2.321
2.321 10.112 10.112
101 101 2.899
2.899 11.799
11.799
102 2.720 11.284
102 110 2.720
2.979 12.026 11.284
110 103 2.979
2.579 10.870 12.026
103 2.579 10.870
3. Results
3. Results
A first assessment of the residual stress in the SLM samples as a function of the process energy
A first assessment of the residual stress in the SLM samples as a function of the process energy
density (EV ) is obtained from distortion angle measurements in the released bridge specimens.
density (EV) is obtained from distortion angle measurements in the released bridge specimens. It
3 , i.e.,
It discloses that the deformations strongly increase for EV < 116 J/mm scanning velocities
discloses that the deformations strongly increase for EV < 116 J/mm3, i.e., scanning velocities
exceeding 500 mm/s (Figure 4). The deflection angle was measured at the top and bottom surfaces of
exceeding 500 mm/s (Figure 4). The deflection angle was measured at the top and bottom surfaces of
the bridge specimens, after removing them from the baseplates by means of spark erosion. Apparently,
the bridge specimens, after removing them from the baseplates by means of spark erosion.
forApparently,
sufficiently for
high EV (smaller
sufficiently
laser scan velocities) such as those used for the samples A3 and A4, the
high EV (smaller laser scan velocities) such as those used for the samples
distortion
A3 and A4, the distortion anglesto
angles are similar. Prior release
are from
similar. thetobaseplate
Prior (as-built
release from condition),
the baseplate none ofcondition),
(as-built the samples
showed
none of the samples showed any deformation within the accuracy of the measurement,geometries
any deformation within the accuracy of the measurement, and the obtained and the
retained
obtained geometries retained their near-net shape. Using SLM process parameters close to the ones to
their near-net shape. Using SLM process parameters close to the ones corresponding
= 116.7 J/mm3 ,tomay
EVcorresponding EV provide
= 116.7 aJ/mm
processing
3, maywindow
providewith comparably
a processing low distortion
window angles (Figure
with comparably low 4),
while at EV much 3
distortion angleslower
(Figurethan ~53 J/mm
4), while , highly
at EV much distorted
lower than ~53samples
J/mm , were
3 highlyobtained
distortedinsamples
a significantly
were
different
obtained material state. Residual
in a significantly differentstress
materialmeasurements
state. Residualwere
stressconducted
measurements for SLM parameters
were conducted forA4
(optimum window)A4
SLM parameters and A10 (above
(optimum the optimum
window) window).
and A10 (above the optimum window).

as build
1.0 Released top
Released bottom
Deflection angle, °

0.8

0.6

0.4

0.2

0.0

291.7 145.8 116.7 53

Energy density, J/mm3

Figure
Figure 4.4.Distortion
Distortionangles
anglesas
asaa function
function of
of SLM
SLM process
processenergy
energydensity
density(E(E
V). ).
V

The structure of the near-surface outer layers of the SLM samples is critical for the
The structure of the near-surface outer layers of the SLM samples is critical for the interpretation
interpretation of the residual stress condition in these regions (Figure 5a,b). Firstly, partially molten
of the residual stress condition in these regions (Figure 5a,b). Firstly, partially molten powder particles
powder particles attached to the sample walls, as well as the outer contour lines, need to be
attached
accounted to thefor,sample walls, asthey
simply because wellrender
as thetheouter contour lines,
identification of theneed
surfaceto position
be accounted for,Assimply
difficult. for
because they render the identification of the surface position difficult.
the attached powder particles, they represent an inhomogeneous layer of thickness between 20 As for the attached powder
μm
particles,
and 50 theyμm. represent
Moreover,an inhomogeneous
during fabrication, layer
the twoof thickness between
outer contour lines20(C1 and C2)
µmand 50 µm.
are Moreover,
initially
during fabrication, the two outer contour lines (C1 and C2) are initially
melted by the laser, followed by the fill contour (FC), and then finally the bulk region (VV), melted by the laser, followed
see
by Figure
the fill 5c.
contour (FC), and then finally the bulk region (VV), see Figure 5c. The outer
The outer and fill contour scan strategies are similar for all samples: the first and second and fill contour
scan strategies
contour lines,are
C1similar
and C2,for areall samples:bythe
separated 90 first
μm, and
whilesecond
the fillcontour
contour lines, C1 and offset
FC is further C2, arebyseparated
60 μm.
byAll
90 µm,
threewhilelines the
are fill contour FC
laser-treated is further
with a power offset by 60
of 100 and All
W µm. three
a laser lines
scan are laser-treated
velocity of 525 mm/s, with
resulting in a melt track width of ~180 μm. The bulk scan lines are defined
a power of 100 W and a laser scan velocity of 525 mm/s, resulting in a melt track width of ~180 µm. to end at the central
Thesecond contour
bulk scan line.
lines are defined to end at the central second contour line.
Dependingononthe
Depending theenergy
energydensity
density ofof the
the bulk
bulk SLM
SLM scanning
scanningparameters,
parameters,the thebulk
bulk melt
melt tracks
tracks
display large widths (i.e., ~280 μm for A4, ~250 μm for A10); thus, their
display large widths (i.e., ~280 µm for A4, ~250 µm for A10); thus, their ends extend far beyond ends extend far beyond the
Materials 2017, 10, 348 7 of 14

Materials 2017, 10, 348 7 of 14

the second and into the first contour line region (Figure 5a,b). For sample A4, a small contour wall
second and into the first contour line region (Figure 5a,b). For sample A4, a small contour wall
element of <40 µm width remains discernible, whereas on sample A10, a larger remainder is observed.
element of <40 μm width remains discernible, whereas on sample A10, a larger remainder is
However, it must be kept in mind that the roughness measurements showed values of Sa and
observed. However, it must be kept in mind that the roughness measurements showed values of Sa
Sq
around 20 µm (see Table 3).
and Sq around 20 μm (see Table 3).

Attached

powder

Figure
Figure 5. (a)5.Light
(a) Light optical
optical microscopyimage
microscopy image of
of sample
sample A4 A4(with
(withscan vectors
scan vectorsandand
meltmelt
tracktrack
widthwidth
marked (C1, C2: first and second contour line; FC: fill contour; VV volume vector); (b)
marked (C1, C2: first and second contour line; FC: fill contour; VV volume vector); (b) light opticallight optical
microscopy
microscopy imageimage of A10
of A10 with
with scanvectors
scan vectors and
and melt
melt track
trackwidth
widthmarked;
marked;(c) (c)
schematic viewview
schematic of theof the
scan strategy in the wall region showing contour, fill contour, and bulk laser vectors; (d) roughness
scan strategy in the wall region showing contour, fill contour, and bulk laser vectors; (d) roughness
map of the lateral surface of sample A4 (Ev = 116.7 J/mm3).
map of the lateral surface of sample A4 (Ev = 116.7 J/mm3 ).
Table 3. Roughness measurements on the bridge specimens A1, A3, A4, and A10 (side surface): Sa, Sq.
Table 3. Roughness measurements on the bridge specimens A1, A3, A4, and A10 (side surface): Sa , Sq .
Roughness A1 A3 A4 A10
Sa (μm)Roughness 20.1 A1 17.6A3 18.2
A4 A10 17.5
Sq (μm) 23.9 21.4 21.8 20.8
Sa (µm) 20.1 17.6 18.2 17.5
Sq (µm) 23.9 21.4 21.8 20.8
The synchrotron diffraction measurements provide diffraction patterns with several reflections
of the α + α’ hexagonal close packed Ti lattices (Figure 6a,b). Although in the literature the
assessment
The synchrotronof phase stresses inmeasurements
diffraction the phase has been shown
provide to be possible
diffraction [35], with
patterns only the 200 reflection
several reflections of
the α + α’ hexagonal close packed Ti lattices (Figure 6a,b). Although in the literature(not
of the body-centred cubic phase is (weakly) discernible after annealing at 650 °C for 3 h theshown),
assessment
evidencing
of phase thatinthe
stresses residual
the phasestress is driven
has been by thetohexagonal
shown be possiblephases. The
[35], lattice
only thestrain
200 versus sin2 of the
reflection
diagram (Figure 6b) shows a linear tendency, i.e., little effect of the texture, thereby also confirming
body-centred cubic phase is (weakly) discernible after annealing at 650 ◦ C for 3 h (not shown),
the absence of shear stresses.
evidencing that the residual stress is driven by the hexagonal phases. The lattice strain versus sin2 ψ
For an assessment of the residual stress state, it is necessary that the penetration depth during
diagram (Figure 6b) shows a linear tendency, i.e., little effect of the texture, thereby also confirming the
the diffraction experiment exceeds the roughness and contour regions (Figure 6); it should reach at
absence
leastof100
shear stresses.
μm into the sample bulk to probe the direct effects of the SLM parameters on the stress state
For an assessment
beyond the contour region.of the residual
This would stress
also state,
ensureitthat
is necessary that theispenetration
the gauge volume fully immersed depth during
in the
the diffraction
sample. The experiment exceeds the for
typical (representative roughness and positions)
all scanned contour regions (Figure
stress profile for6);the
it should reach at
transversal
least direction
100 µm intoas a function
the sample of depth
bulkistopresented
probe the in direct
Figure effects
6c. Bothofsamples
the SLM show a linear increase
parameters on theofstress
state stresses
beyondwith thedepth
contouruntilregion.
maximaThisof ~600 MPa also
would and 875 MPa that
ensure at around a 70 μm
the gauge depth are
volume reached
is fully for
immersed
A4 (E v = 116.7 J/mm3) and A10 (Ev = 53 J/mm3), respectively. Then, the stress in the A4 condition
in the sample. The typical (representative for all scanned positions) stress profile for the transversal
remains fairly constant, while it decreases down to ~700 MPa for the A10 condition at ~100 μm from
direction as a function of depth is presented in Figure 6c. Both samples show a linear increase of
the surface. The A10 condition presents higher stresses in comparison with A4, for the same penetration.
stresses with depth until maxima of ~600 MPa and 875 MPa at around a 70 µm depth are reached for
A4 (Ev = 116.7 J/mm3 ) and A10 (Ev = 53 J/mm3 ), respectively. Then, the stress in the A4 condition
remains fairly constant, while it decreases down to ~700 MPa for the A10 condition at ~100 µm from the
surface. The A10 condition presents higher stresses in comparison with A4, for the same penetration.
Materials 2017, 10, 348 8 of 14
Materials 2017, 10, 348 8 of 14

Figure
Figure 6. 6.
(a)(a) Diffractogramsfor
Diffractograms forsamples
samples with
with the
the baseplate
baseplate (BP)
(BP)andandafter
afterthermal
thermaltreatment (TT);
treatment (TT);
(b) example of sin
2 2  data evaluation for A4 BP (L = 4, T = 0.5); (c) transversal component of residual
(b) example of sin ψ data evaluation for A4 BP (L = 4, T = 0.5); (c) transversal component of residual
stress
stress profile
profile as as a function
a function ofofdepth
depthfor
forsamples
sampleswith
withBP.
BP.

The transversal stress components measured in the SLM bridges in an “as-built” state on the
The transversal
baseplate (BP) in thestress components
T direction exhibit measured
consistentlyinsignificant
the SLM tensile
bridgesresidual
in an “as-built”
stresses, with state on the
higher
baseplate (BP) in the
values occurring T direction
towards exhibit
the pillar consistently
bottom significant
(Figure 7). The average tensile residual
error stresses,
of residual stresswith
mapshigher
is
values
around ±30 MPa. The basal plane {002} (Figure 7a) and the pyramidal plane {101} (Figure 7c) present is
occurring towards the pillar bottom (Figure 7). The average error of residual stress maps
around ±30 distribution
a similar MPa. The basal plane stresses
of tensile {002} (Figure
as the7a) and the{110}
prismatic pyramidal
(Figureplane {101}the
7b) and (Figure 7c) present
pyramidal {103} a
similar
planesdistribution
(Figure 7d).ofThetensile
stressstresses
increaseas towards
the prismatic {110}is (Figure
the pillar 7b) and
much more the pyramidal
pronounced for the{103}
{110}planes
and
(Figure
{103} 7d). Thereaching
planes, stress increase
stressestowards
around the 650 pillar
MPa. isThemuch more pronounced
crystallographic for the
plane {103} is {110}
only and
weakly{103}
influenced
planes, reaching bystresses
intergranular
around stresses,
650 MPa.i.e., Theless plastic anisotropy
crystallographic [36], isproviding
plane {103} only weakly an influenced
accurate
by assessment
intergranular of the macroscopic
stresses, residual
i.e., less plasticstresses. Thus,[36],
anisotropy {103}providing
was chosenanfor further analysis.
accurate assessment of the
Latticeresidual
macroscopic strain maps
stresses.are Thus,
plotted in was
{103} Figure 8, in for
chosen order to compare
further analysis.them with the distortion
measurements
Lattice strainin mapsFigureare 4. The average
plotted error of8,the
in Figure in lattice
order strain is aboutthem
to compare ±4 × 10 −4 (relative error
with the distortion
below 10%). The
measurements lattice 4.
in Figure strains in the T direction
The average for A10
error of the latticeBP-with
strain lowest
is aboutenergy
±4 ×density-(Figure
10−4 (relative 8d) error
are larger
below 10%). than those of
The lattice the corresponding
strains in the T directionpositions
for A10of A4 BP (Figure
BP-with lowest8a).energy
A4 BPdensity-(Figure
shows a positive8d)
arestrain
largergradient fromofthe
than those thetop surface (εT ~ 0.0015)
corresponding to the
positions of middle
A4 BP of the sample
(Figure 8a). A4 (εTBP
~ 0.005;
shows Figure 8a).
a positive
After releasing from the baseplate (R), the strains in sample A4 R (Figure 8b) remain similar to those
strain gradient from the top surface (εT ~0.0015) to the middle of the sample (εT ~0.005; Figure 8a).
of A4 BP. In contrast, the comparison of sample A10 BP to A10 R shows a broadening of the
After releasing from the baseplate (R), the strains in sample A4 R (Figure 8b) remain similar to those of
high-strain region of the pillar after release (Figure 8d,e). The heat treated samples (650 °C for 3 h)
A4 BP. In contrast, the comparison of sample A10 BP to A10 R shows a broadening of the high-strain
show decreased strain values to almost zero, and no further gradients are observed in both A4 and
region of the pillar after release (Figure 8d,e). The heat treated samples (650 ◦ C for 3 h) show decreased
A10 (Figure 8c,f). This implies an efficient relief of residual stresses from the BP condition.
strain values to almost zero, and no further gradients are observed in both A4 and A10 (Figure 8c,f).
This implies an efficient relief of residual stresses from the BP condition.
Materials 2017, 10, 348 9 of 14
Materials 2017,
Materials 2017, 10,
10, 348
348 99 of
of 14
14

Figure 7.
Figure 7. Transversal
Transversal residual
residual stress
stress maps
maps for
for different
different crystallograpic
crystallograpic planes
planes for
for condition
condition A4
condition A4 BP
A4 BP (a)
BP (a)
(a)
{002}; (b)
{002}; (b) {110};
{110}; (c) {101} and (d) {103}. The
The black
black line
line in
in the
the left
left bottom
bottom corner
corner indicates
indicates the
the opening
opening of
of
the bridge with the pillar to the right. The average error is
the bridge with the pillar to the right. The average error is ±30 MPa.±
±30
30 MPa.
MPa.

1.0 (a)
1.0 A4 BP (d)
A4 BP A10 BP
A10 BP
1.5
1.5
Strain
2.0
2.0 0.007
0.007
2.5
2.5
mm

Strain
position, mm

Strain 0.005
0.005
1.0 (b) A4 R
A4 R (e) A10 R
A10 R
1.5
TT position,

0.003
0.003
2.0
2.5
0.001
0.001
1.0 (c)
1.0 TT (f)
A4 TT
A4 A10 TT
A10 TT
1.5
1.5 -0.001
-0.001
2.0
2.0
2.5
2.5
3.0
3.0
0 1
0 1 2
2 3
3 4
4 5
5 6
6 77 0 0 11 2
2 3
3 4
4 5
5 6
6 77 88
L position, mm
Figure
Figure 8.
Figure 8. Lattice
8. Lattice strain
Lattice strain maps
strain maps
maps ofof crystallographic
of crystallographic plane
crystallographic plane {103}
plane {103} in
{103} in the
in the T
the T direction
T direction (building
direction (building direction)
(building direction)
direction) forfor
for
process
process conditions
conditions A4
A4 (a;
(a; BP)
BP) with
with baseplate,
baseplate, (b;
(b; R)
R) released,
released, (c;
(c; TT)
TT) after
after thermal
thermal treatment
treatment
process conditions A4 (a; BP) with baseplate, (b; R) released, (c; TT) after thermal treatment and A10 and
and A10
A10
(d;
(d; BP) with
(d; BP)
BP) with baseplate,
with baseplate, (e;
baseplate, (e; R)
(e; R) released,
R) released, (f;
released, (f; TT)
(f; TT) after
TT) after thermal
after thermaltreatment.
thermal treatment.The
treatment. Theaverage
The averageerror
average errorisis
error is±±4
±4
4 ×× 10−.. 4 .
× 10
10 −4
−4

In the case of samples A4 R and A10 R, it was also possible to characterize the longitudinal
In the case of samples A4 R and A10 R, it was also possible to characterize the longitudinal
component of the lattice strains (Figure 9). The average error of the longitudinal lattice strain was
component of the −4. Itlattice strains (Figure 9). The average error of the longitudinal lattice strain was
around ±3 × 10−4 must be noted that, due to the elongation of the gauge volume, the middle-top
around ±3 × 10−4 . It must be noted that, due to the elongation of the gauge volume, the middle-top
part above the arc was not measured and only the results for the pillar region are reported (Figures 2
part above the arc was not measured and only the results for the pillar region are reported (Figures 2
and 3b).
and 3b).
The longitudinal lattice strain components in both samples A4 R and A 10 R show tensile strains
The longitudinal lattice strain components in both samples A4 R and A 10 R show tensile strains
with increasing values going from the arc to the pillar. Interestingly, the longitudinal strains are
with increasing values going from the arc to the pillar. Interestingly, the longitudinal strains are
almost an order of magnitude lower than the transversal ones: the maximum strains found for A4
almost an order of magnitude lower than the transversal ones: the maximum strains found for A4
are εLL = 0.0007 (versus transversal strain εTT = 0.005), and for A10 εLL = 0.002 (versus transversal strain
are εL = 0.0007 (versus transversal strain εT = 0.005), and for A10 εL = 0.002 (versus transversal
εTT = 0.006). A4 R presents significantly lower longitudinal strains than A10 R, as found for the
strain εT = 0.006). A4 R presents significantly lower longitudinal strains than A10 R, as found for the
transversal component.
transversal component.
Materials 2017, 10, 348 10 of 14
Materials 2017, 10, 348 10 of 14

1.5
(a) A4 R Strain
1.5
A10 a10
R
2.0 0.0017
2.0
2.5 0.0013
T position, mm
0.0009
1.5 2.5
(b) A10 R
0.0005
2.0
3.0 0.0001
4 5 6 7 8
2.5

3.0
4 5 6 7 8
L position, mm

Figure
Figure 9. Longitudinal
Longitudinal components
components ofof residual
residual strains
strains for
for crystallographic
crystallographic plane
plane {103}
{103} for
for the
the pillar
pillar
sections of samples (a) A4 R and (b) A10 R. The average error is about ±
sections −4.−4 .
±33××1010

4.
4. Discussion
Discussion
The
The residual
residual stress
stress state
state in
in the
the SLM
SLM Ti-6Al-4V
Ti-6Al-4V samples samples shows
shows aa strong dependence on:
strong dependence on: (i)
(i) the
the
geometry
geometry of of the
the part;
part; (ii)
(ii) the fabrication parameters
the fabrication parameters (i.e., (i.e., process
process energy density E
energy density EVV);
); (iii)
(iii) the
the heat
heat
treatment;
treatment; and and (iv)
(iv) thethe removal
removal from from the base plate.
the base plate. The
The lattice
lattice strains
strains forfor the
the conditions
conditions A4 A4 BPBP
(E
(Evv == 116.7
116.7 J/mm
J/mm) )and
3 3 andA10A10BP BP (E(Evv ==53
53J/mm
J/mm) )consistently
3 3 consistentlyshow showhigh hightensile
tensilevalues
valueson on their
their front
front
surface.
surface. These
These high
high tensile
tensile strains
strains inin the
the subsurface
subsurface region region cancan lead
lead toto crack
crack formation
formation on on the
the surface
surface
during mechanical
during mechanical loading. loading.
The
The effect
effect of
of the
the baseplate
baseplate is is manifested
manifested by by thethe measurable
measurable distortion
distortion of of samples
samples after
after releasing
releasing
from the baseplate (A4 R and A10 R). In fact, the only difference between
from the baseplate (A4 R and A10 R). In fact, the only difference between the stress states in the bridges the stress states in the
= = =
bridges before and after cutting is the presence of the baseplate. This means
before and after cutting is the presence of the baseplate. This means σ BP + σS = σ R (tensorial relation), σ σ = σ (tensorial
= =
relation),
where σ BPwhere σ isinthe
is the stress thestress
bridgeinbefore
the bridge
release, before
σS isrelease,
the stress σ inisthethe stresscaused
bridge in theby bridge caused
the presence
=
of the baseplate (before cutting), and σ R is the stress in the bridge after cutting. The contribution ofThe
by the presence of the baseplate (before cutting), and σ is the stress in the bridge after cutting. the
=
contribution
baseplate to the of the
bridgebaseplate
stress toσSthecanbridge stress σknowing
be calculated can be calculated
the distortion knowing
of the the distortion
bridges of the
after release
= =
bridges
from theafter releasei.e.,
baseplate, from the baseplate,
it must hold σS = i.e.,
C εitDmust
, where holdC isσ the
= elastic
ε , wheretensor C of
is the
the elastic
bridge tensor of
material.
=
the bridge material. Indeed, the residual strain state after cutting, ε , must
Indeed, the residual strain state after cutting, ε R , must be the sum of the residual strain before cutting be the sum of the residual
= =
strain
ε BP and before cutting strain
the distortion ε and the =
ε D (i.e.,
=
εdistortion
=
strain ε (i.e., ε = ε − ε ). In other words,
D = ε R − ε BP ). In other words, converting the released bridge
=
converting the released bridge back to its
back to its original form (i.e., applying − ε D ) would restore original form the(i.e.,original
applying −εfield:
strain ) would restore the
original strain field:
= = =
ε BP = − ε D + ε R (1)
ε = −ε ε (1)
The information of
The of distortion
distortionfrom
fromconfocal
confocalmicroscopy
microscopymeasurements
measurementsonon thethetoptop
surface cancan
surface be
used to calculate the whole strain distribution on the frontal surface of the samples
be used to calculate the whole strain distribution on the frontal surface of the samples using the using the finite
element
finite method
element (FEM).(FEM).
method Thus, the mean
Thus, thecurvature profile (Figure
mean curvature profile10a) was 10a)
(Figure implemented via keypoints
was implemented via
in the software
keypoints in theANSYS,
software creating
ANSYS, ancreating
initial deformed
an initial geometry.
deformed During
geometry.the During
simulation, the top surface
the simulation, the
wassurface
top flattenedwasusing position
flattened dependent
using positiondisplacements in the T-direction
dependent displacements in the(Figure 2a). A(Figure
T-direction linear 2a).
elastic
A
isotropic
linear material
elastic (E103
isotropic = 120.6(E
material GPa,
103 = υ 103 =0.31) was
120.6 chosen,
GPa, υ103 =0.31) wasfor whichfor
chosen, thewhich
meshthe consisted of 70080
mesh consisted
hexahedral
of elements (SOLID186).
70080 hexahedral Figure 10b Figure
elements (SOLID186). represents
10bthe resulting the
represents transversal
resultingelastic strain (εelastic
transversal TD ) in
3D for(εsample
strain in A10 R, necessary
3D for sample A10 to counterbalance
R, necessary tothe measured distortion.
counterbalance In our distortion.
the measured discussion Inabove,
our
it will hold:above,
discussion εD . hold: ε = −ε .
εTD =it−will
Materials 2017, 10, 348 11 of 14
Materials 2017, 10, 348 11 of 14

Figure 10. (a) Mean curvature profile of A10 specimen with marked data points used for a
FE-simulation; (b) transversal elastic strain ε resulting from the simulation in 3D (initial state:
deformed top surface; end state: flattened top surface).

The comparison between strains counterbalancing distortion (ε , obtained from the simulations)
and the diffraction experimental results (ε − ε ), is depicted in Figure 11. From the simulation,
compression strains are present on the top-centre of the bridge, with a tensile counterpart in the
Figure 10. (a)(a)Mean curvature profile ofofA10
A10specimen with marked data
datapoints used forfor
a
Figure
inner of the arcMean
part10. (Figurecurvature profile
11b,e). The specimen
distortion-related with
strain marked
for A10 R (Figure points used
11e) is larger than
FE-simulation; (b) transversal elastic strain ε resulting from the simulation in 3D (initial state:
foraA4FE-simulation;
R, and A10 R(b) transversal
shows a moreelastic strain εstrain
pronounced T resulting from
gradient the simulation
above in 3D
the arc of the (initial state:
bridge.
deformed
deformed top
top surface;
surface; end
end state:
state: flattened
flattened top
top surface).
surface).
The strain difference (ε − ε , Figure 11a,d) calculated by means of diffraction data (Figure 8)
shows higher values near the bridge arc. Figure 11c,f depicts the difference between diffraction-based
The
The comparison
comparison between
between strains counterbalancing
strains strains
counterbalancing distortion
distortion (ε
(εTD,,obtained
obtained from the the simulations)
(Figure 11a,d) and distortion-based (Figure 11b,e). This difference shouldfrombe zero, simulations)
if we take
and
and the diffraction
theaccount
diffraction experimental
experimental results
results (ε −
− ε ), is depicted in Figure 11. From the simulation,
into Equation (1). However, the(ε ε
diffraction
R BP ), is depicted
strain in
difference Figure
(with 11.
and From
without the simulation,
baseplate)
compression
compression strains are
strains arepresent
presentononthe the top-centre
top-centre of the
of the bridge, with
bridge, a tensile counterpart theininner
the
does not match the distortion-related strain calculated by FEM.withThe asignificant
tensile counterpart
difference in could be
inner
part ofpart
the of
explained the
arcby arc model
(Figure
the (Figure
11b,e). 11b,e). Theof distortion-related
The distortion-related
assumption strain
strain for
a simply linear A10for
elastic R A10
and R (Figure
(Figure 11e)
isotropic is 11e)
larger
material isbehaviour.
largerfor
than than
A4
for A4 R,
R, Therefore,and
and A10 Rthe A10 R shows
diffraction
shows a more
a moreresults pronounced
can alsostrain
pronounced strain
play the gradient
role ofabove
gradient above
the benchmark the
the arc offor arc
thetheof the
model.
bridge. bridge.
The strain difference (ε − ε , Figure 11a,d) calculated by means of diffraction data (Figure 8)
shows higher values near the bridge arc. Figure 11c,f depicts the difference between diffraction-based
(Figure 11a,d) and distortion-based strains (Figure 11b,e). This difference should be zero, if we take
into account Equation (1). However, the diffraction strain difference (with and without baseplate)
does not match the distortion-related strain calculated by FEM. The significant difference could be
explained by the model assumption of a simply linear elastic and isotropic material behaviour.
Therefore, the diffraction results can also play the role of the benchmark for the model.

Figure
Figure 11.11.Strain
Strainmaps
maps for
for sample
sampleA4:A4:(a)(a)
experimental freefree
experimental distortion component;
distortion (b) simulated
component; free
(b) simulated
distortion
free distortion component;
component; (c)(c)
thethe
difference
differencebetween
betweenexperimental
experimental and simulated
and simulated free distortion;
free distortion;forfor
sample
sample A10;
A10; (d)(d) experimental
experimental free
free distortion
distortion component;(e)(e)simulated
component; simulatedfree
freedistortion
distortioncomponent;
component;(f) (f)the
the difference between experimental and simulated free distortion. The average error
difference between experimental and simulated free distortion. The average error for experimental for experimental
data
data is around
is around ±±4
4 ×× 10 −.4 .
10−4

The main mechanism leading to the residual stresses in AM parts is the rapid solidification and
The strain difference (ε − εBP , Figure 11a,d) calculated by means of diffraction data (Figure 8)
the heat coming from newR layer deposition. Therefore, laser energy, scanning speeds, and laser
shows higher values near the bridge arc. Figure 11c,f depicts the difference between diffraction-based
patterns have a strong influence
Figure 11. Strain maps for sample A4: on the
(a)solidification
experimental and heat distribution
free distortion component; conditions. The slower
(b) simulated free
(Figure 11a,d)
thedistortion and distortion-based
scanningcomponent;
speed of the strains (Figure 11b,e). This difference should be zero, if we In take
(c)SLM laser, the between
the difference larger the volume affected
experimental by the introduced
and simulated energy.
free distortion; for
into account
turn, Equation
the thermal (1). However,
gradients in the the diffraction strain difference of (with and without baseplate) does
sample A10; (d) experimental freespecimen
distortion decrease.
component; A (e)
reduction
simulated freethe cooling
distortionrate due to slower
component; (f)
notlaser
match the distortion-related strain calculated by FEM. The significant difference could be explained
the scanning has beenexperimental
difference between reported inand [37], as wellfree
simulated as distortion.
a smallerThe thermal
average gradient
error forand decreasing
experimental
by the model assumption of a simply linear elastic and isotropic material behaviour. Therefore,
data is around ±4 × 10 . −4
the diffraction results can also play the role of the benchmark for the model.
The main
The main mechanism
mechanismleading leadingtotothe
theresidual
residualstresses
stressesinin
AMAM parts
partsis the rapid
is the solidification
rapid andand
solidification the
heatheat
the coming fromfrom
coming new layer deposition.
new layer Therefore,
deposition. laser energy,
Therefore, scanning
laser energy, speeds, speeds,
scanning and laserandpatterns
laser
have a strong influence on the solidification and heat distribution conditions.
patterns have a strong influence on the solidification and heat distribution conditions. The slower The slower the scanning
speed
the of the SLM
scanning speedlaser, the SLM
of the largerlaser,
the volume
the largeraffected by the introduced
the volume affected byenergy. In turn, the
the introduced thermal
energy. In
gradients in the specimen decrease. A reduction of the cooling rate due to slower
turn, the thermal gradients in the specimen decrease. A reduction of the cooling rate due to slower laser scanning has
laser scanning has been reported in [37], as well as a smaller thermal gradient and decreasing
Materials 2017, 10, 348 12 of 14

been reported in [37], as well as a smaller thermal gradient and decreasing deflection angles (i.e.,
distortion obtained by the bridge curvature method [16]) in geometrically identical bridge-shaped
samples. This should imply a decrease of residual stresses. Our present study confirms that for a
high energy density (samples A4), lower tensile sub-surface residual stresses are found compared to a
low energy density (samples A10). In a previous study, the SLM processing window was determined
based on the minimization of bulk defects [28]. The samples with the lowest volume fraction of pores
coincides with that with lower residual stresses (A4) obtained in this study. The dependence of the
deformation angles on energy density (Figure 4) would indicate that lowering velocities further will
not substantially decrease the stress state, but lead to increased material defects in the form of keyhole
pores [28,38,39]. The analysis of the trend of sample distortion (Figure 4), including the additional SLM
parameters A1 (Ev = 291.7 J/mm3 ) and A3 (Ev = 145.8 J/mm3 ), indicates that the residual stress reaches
a plateau-like minimum when the laser scan velocities decrease to 500 mm/s (A4) and below (A3).
In the investigated regions of both samples, only tensile residual stresses were found in all
as-built conditions; however, it should be mentioned that only subsurface regions were investigated.
The origin of high tensile residual stress in the subsurface region is reported in [40]. In that study,
the authors explain this effect through layer-by-layer rapid solidification and cooling phenomena.
Previous (destructive) characterization of residual stresses in SLM Ti-6Al-4V parts by means of the
contour method [22] showed that the largest principal stresses are those along the building direction.
Furthermore, high tensile stresses near the yield stress appeared towards the surfaces/edges of the
specimens. These tensile stresses on the edge of the samples were balanced by compressive stresses in
the bulk. Similar residual stress profiles (tensile towards the surfaces and compression in the bulk)
have been extensively reported for conventional welds [41] and also for laser direct metal deposited
Ni-base superalloy parts (e.g., [42]).
The thermal gradients for lower SLM scan speeds are more favourable for achieving lower initial
residual stresses, although this limits productivity by increasing the building time. Lower stress states
are critical when building complex-shaped components. Near the baseplate, a sufficiently strong
attachment by means of direct building on the base or by strong supports can prevent deformations,
i.e., allows achieving accurate near-end shape geometries while preventing parts from bending,
and potentially disturbing the SLM build job. Moreover, with larger build heights and with higher
geometrical complexity, support structures in-between are needed. Those are minimized or ideally
completely suppressed in order to reduce the difficult and costly post-processing steps for their removal.
Hence, lower initial stresses would facilitate production of complex shapes such as turbo-engine
impellers with fine, three-dimensionally curved blades and poorly accessible undercuts.

5. Conclusions
Synchrotron energy dispersive measurements, in combination with angle distortion
characterization, have proved a suitable approach for assessing residual stress fields in SLM Ti-6Al-4V,
supporting the process and post-process development. Tensile stresses built up at the near-surface
region during SLM for all studied conditions. After heat treatment, those stresses were fully relieved.
We found that the processing window minimizing defects also allows the reduction of initial
(tensile) residual stresses. This implies smaller component distortion during SLM processing and
could lead to more cost-effective post-treatment processes.
Further investigations are underway to study the role of the contour lines (and their
variation/elimination) and the effects of surface-post treatments for control in the development
of surface residual stresses on SLM parts.

Acknowledgments: T. Merzouk is acknowledged for the preparation of SLM samples.

Author Contributions: The work presented in this paper was carried out as a collaboration between all authors.
Tatiana Mishurova, Katia Artzt, and Sandra Cabeza performed the X-ray diffraction measurements and data
processing on the synchrotron. Christoph Genzel and Manual Klaus defined the set-up and subsequent analysis
of the results. Jan Haubrich and Katia Artzt manufactured the samples and carried out optical microscopy and
Materials 2017, 10, 348 13 of 14

LSM measurements. All authors worked on the discussion and interpretation of the results. All authors have
contributed to, seen, and approved the final manuscript.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Conner, B.P.; Manogharan, G.P.; Martof, A.N.; Rodomsky, L.M.; Rodomsky, C.M.; Jordan, D.C.; Limperos, J.W.
Making sense of 3-D printing: Creating a map of additive manufacturing products and services. Addit. Manuf.
2014, 1–4, 64–76. [CrossRef]
2. Mercelis, P.; Kruth, J.-P. Residual stresses in selective laser sintering and selective laser melting.
Rapid Prototyp. J. 2006, 12, 254–265. [CrossRef]
3. Welsch, G.; Boyer, R.; Collings, E.W. Materials Properties Handbook: Titanium Alloys Materials Properties
Handbook; ASM International: Materials Park, OH, USA, 1993.
4. Song, B.; Zhao, X.; Li, S. Differences in microstructure and properties between selective laser melting and
traditional manufacturing for fabrication of metal parts: A review. Front. Mech. Eng. 2015, 10, 111–125.
[CrossRef]
5. Yang, J.; Yu, H.; Yin, J.; Gao, M.; Wang, Z.; Zeng, X. Formation and control of martensite in Ti-6Al-4V alloy
produced by selective laser melting. Mater. Des. 2016, 108, 308–318. [CrossRef]
6. Collins, P.C.; Brice, D.A.; Samimi, P.; Ghamarian, I.; Fraser, H.L. Microstructural Control of Additively
Manufactured Metallic Materials. Annu. Rev. Mater. Res. 2016, 46, 63–91. [CrossRef]
7. Lewandowski, J.J.; Seifi, M. Metal Additive Manufacturing: A Review of Mechanical Properties. Annu. Rev.
Mater. Res. 2016, 46, 151–186. [CrossRef]
8. Leuders, C.; Thöne, M.; Riemer, A.; Niendorf, T.; Tröster, T.; Richard, H.A.; Maier, H.J. On the mechanical
behaviour of titanium alloy TiAl6V4 manufactured by selective laser melting Fatigue resistance and crack
growth performance. Int. J. Fatigue 2013, 48, 300–307. [CrossRef]
9. Leuders, C.; Lieneke, T.; Lammers, S.; Niendorf, T. On the fatigue properties of metals manufactured by
selective laser melting–The role of ductility. J. Mater. Res. 2017, 29, 1911–1919. [CrossRef]
10. Vrancken, B.; Thijs, L.; Kruth, J.P.; Humbeeck, J.V. Heat treatment of Ti6Al4V produced by selective laser
melting: Microstructure and mechanical properties. J. Alloys Compd. 2012, 541, 177–185. [CrossRef]
11. Huang, Q.; Liu, X.; Yang, X.; Zhang, R.; Shen, Z.; Feng, Q. Specific heat treatment of selective laser melted
Ti–6Al–4V for biomedical applications. Front. Mater. Sci. 2015, 9, 373–381. [CrossRef]
12. Wu, S.Q.; Lu, Y.J.; Gan, Y.L.; Huang, T.T.; Zhao, C.Q.; Lin, J.J.; Lin, J.X. Microstructural evolution and
microhardness of a selective-laser-melted Ti–6Al–4V alloy after post heat treatments. J. Alloys Compd. 2016,
672, 643–652. [CrossRef]
13. Vilaro, T.; Colin, C.; Bartout, J.D. As-Fabricated and Heat-Treated Microstructures of the Ti-6Al-4V Alloy by
SLM. Metall. Mater. Trans. A 2011, 42, 3190–3199. [CrossRef]
14. Xu, W.; Brandt, M.; Sun, S.; Elambasseril, J.; Liu, Q.; Latham, K.; Xia, K.; Qian, M. Additive manufacturing of
strong and ductile Ti–6Al–4V by selective laser melting via in situ martensite decomposition. Acta Mater.
2015, 85, 74–84. [CrossRef]
15. Vrancken, B.; Buls, S.; Kruth, J.P.; Humbeeck, J.V. Preheating of Selective Laser Melted Ti6Al4V:
Microstructure and Mechanical Properties. In Proceedings of the 13th World Conference on Titanium,
San Diego, CL, USA, 16–20 August 2015; pp. 1269–1277.
16. Kruth, J.-P.; Deckers, J.; Yasa, E.; Wauthle, R. Assessing and comparing influencing factors of residual stresses
in selective laser melting using a novel analysis method. Proc. Inst. Mech. Eng. Part B J. Eng. Manuf. 2012,
226, 980–991. [CrossRef]
17. Thöne, M.; Leuders, C.; Riemer, A.; Tröster, T.; Richard, H.A. Influence of heat-treatment on Selective Laser
Melting products–e.g., Ti6Al4V. In Proceedings of the Annual International Solid Freeform Fabrication
Symposium, Austin, TX, USA, 7–9 August 2012; pp. 492–498.
18. Zaeh, M.F.; Branner, G. Investigations on residual stresses and deformations in selective laser melting.
Prod. Eng. 2009, 4, 35–45. [CrossRef]
19. Kobryn, P.A.; Semiatin, S. The laser additive manufacture of Ti-6Al-4V. J. Miner. 2001, 53, 40–42. [CrossRef]
20. Das, M.; Balla, V.K.; Basu, D.; Bose, S.; Bandyopadhyay, A. Laser processing of SiC-particle-reinforced coating
on titanium. Scr. Mater. 2010, 63, 438–441. [CrossRef]
Materials 2017, 10, 348 14 of 14

21. Hodge, N.E.; Ferencz, R.M.; Solberg, J.M. Implementation of a thermomechanical model for the simulation
of SLM. Comput. Mech. 2014, 54, 33–51. [CrossRef]
22. Parry, L.; Ashcroft, I.A.; Wildman, R.D. Understanding the effect of laser scan strategy on residual stress
inselective laser melting through thermo-mechanical simulation. Addit. Manuf. 2016, 12, 1–15. [CrossRef]
23. Donachie, M.J. Titanium–A Technical Guide, 2nd ed.; ASM International: Materials Park, OH, USA, 2000;
pp. 55–65.
24. Vrancken, B.; Cain, V.; Knutsen, R.; Humbeeck, J.V. Residual stress via the contour method in compact
tension specimens produced via selective laser melting. Scr. Mater. 2014, 8, 29–32. [CrossRef]
25. Yadroitsev, I.; Yadroitsava, I. Evaluation of residual stress in stainless steel 316L and Ti6Al4V samples
produced by selective laser melting. Virtual Phys. Prototyp. 2015, 10, 67–76. [CrossRef]
26. Vrancken, B.; Wauthle, R.; Kruth, J.-P.; van Humbeeck, J. Study of the influence of material properties on
residual stress in selective laser melting. In Proceedings of the Solid Freeform Fabrication Symposium,
Austin, TX, USA, 12–14 August 2013; pp. 1–15.
27. Genzel, C.; Denks, I.A.; Gibmeier, J.; Klaus, M.; Wagener, G. The materials science synchrotron beamline
EDDI for energy-dispersive diffraction analysis. Nucl. Instr. Meth. Phys. Res. 2007, 578, 23–33. [CrossRef]
28. Kasperovich, G.; Haubrich, J.; Gussone, J.; Requena, G. Correlation between porosity and processing
parameters in TiAl6V4 produced by selective laser melting. Mater. Des. 2016, 105, 160–170. [CrossRef]
29. Hauk, V. Structural and Residual Stress Analysis by Nondestructive Methods: Evaluation, Application, Assessment;
Elsevier Science B.V.: Amsterdam, The Netherlands, 1997; pp. 132–148.
30. Giessen, B.C.; Ordon, G.E. X-ray diffraction: New high-speed technique based on X-ray spectrography.
Science 1968, 159, 973–975. [CrossRef] [PubMed]
31. Kröner, E. Berechnung der elastischen Konstanten des Vielkristalls aus den Konstanten des Einkristalls.
Z. Phys. 1958, 151, 504–518. [CrossRef]
32. Eshelby, J.D. The Determination of the Elastic Field of an Ellipsoidal Inclusion, and Related Problems. Proc. R.
Soc. Lond. Ser. A 1957, 241, 376–396. [CrossRef]
33. Simonelli, M.; Tse, Y.Y.; Tuck, C. The formation of α + β microstructure in as-fabricated selective laser melting
of Ti–6Al–4V. J. Mater. Res. 2014, 29, 2028–2035. [CrossRef]
34. Landoldt-Börnstein. Numerical Data and Functional Relationships in Science and Technology; New Series,
Group III; Springe: Berlin, Germany, 1979; Volume 13.
35. Bruno, G.; Dunn, B.D. Surface and bulk residual stress in aerospace Ti6Al4V welded tanks. J. Press.
Vessel Technol. 2004, 126, 284–292. [CrossRef]
36. Oliver, E.C.; Daymond, M.R.; Quinta da Fonseca, J.; Withers, P.J. Intergranular Stress Evolution in Titanium
Studied by Neutron Diffraction and Self-consistent Modelling. J. Neutron Res. 2004, 12, 33–37. [CrossRef]
37. Manvatkar, V.; De, A.; Debroy, T. Spatial variation of melt pool geometry, peak temperature and solidification
parameters during laser assisted additive manufacturing process. Mater. Sci. Technol. 2015, 31, 924–930.
[CrossRef]
38. Gong, H.; Rafi, K.; Gu, H.; Starr, T.; Stucker, B. Analysis of defect generation in Ti–6Al–4V parts made using
powder bed fusion additive manufacturing processes. Addit. Manuf. 2014, 1, 87–98. [CrossRef]
39. Gong, H.; Rafi, K.; Gu, H.; Ram, G.J.; Starr, T.; Stucker, B. Influence of defects on mechanical properties of
Ti–6Al–4V components produced by selective laser melting and electron beam melting. Mater. Des. 2015, 86,
545–554. [CrossRef]
40. Liu, Y.; Yang, Y.; Wang, D. A study on the residual stress during selective laser melting (SLM) of metallic
powder. Int. J. Adv. Manuf. Technol. 2016, 87, 647–656. [CrossRef]
41. Albertini, G.; Bruno, G.; Dunn, B.D.; Fiori, F.; Reimers, W.; Wright, J.S. Comparative neutron and X-ray
residual stress measurements on Al-2219 welded plate. Mater. Sci. Eng. 1997, 224, 157–165. [CrossRef]
42. Moat, R.J.; Pinkerton, A.J.; Li, L.; Withers, P.J.; Preuss, M. Residual stresses in laser direct metal deposited
Waspaloy. Mater. Sci. Eng. 2011, 528, 2288–2298. [CrossRef]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).