Materials Science & Engineering A 651 (2016) 198213

Contents lists available at ScienceDirect

Materials Science & Engineering A

journal homepage: www.elsevier.com/locate/msea

Mechanical behavior of additive manufactured, powder-bed

laser-fused materials
Todd M. Mower n, Michael J. Long
M.I.T. Lincoln Laboratory, 244 Wood Street, Lexington, MA 02420, USA

art ic l e i nf o a b s t r a c t

Article history: Mechanical behavior of four metallic alloys fabricated with layered, laser-heated methods of additive
Received 3 June 2015 manufacturing (AM) was compared to that of similar alloys produced with conventional methods
Received in revised form (wrought and machined). AM materials were produced by a leading commercial service provider, as
10 September 2015
opposed to incorporating material specimens produced by unique or specially-adapted equipment. The
Accepted 18 October 2015
Available online 10 November 2015
elastic moduli were measured in exure, stressstrain characteristics were measured in tensile de-
formation, and fatigue strengths were measured in fully reversed bending. The effects of fabrication
Keywords: orientation, surface polishing, and hot isostatic pressing upon mechanical behavior were studied.
Aluminum alloys The fatigue strengths exhibited by SLM AlSi10Mg and DMLS Ti6Al4V in the as-fabricated condition
Titanium alloys
proved to be signicantly inferior to that of conventional material. These lower fatigue strengths are a
Stainless steel alloys
consequence of multiple fatigue cracks initiating at surface defects, internal voids and microcracks, and
Powder-bed fusion
Cyclic fatigue growing simultaneously during cyclic loading. Measured fatigue strengths of DMLS 316L and 17-4PH
approached those of corresponding wrought materials when subjected to principal stresses aligned with
the build planes. When cyclic stresses were applied across the build planes of the DMLS stainless steels,
fatigue fractures often developed prematurely by separation of material. Post-processing the DMLS
Ti6Al4V and SS316L with hot isostatic pressure elevated the fatigue strength signicantly. Measurements
of surface roughness with an optical prolometer, examinations of the material microstructures, and
fractography contribute to an understanding of the mechanical behavior of the additive materials.
& 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction direct selective laser sintering method [4]. Renements of this

1.1. Background
Developers of aerospace systems are increasingly employing
additive manufacturing (AM) technologies to fabricate complex
structural components used in elded systems. Initially used to
create models of design concepts and preliminary prototypes,
applications of laser sintering to produce metallic components for
end-use applications grew exponentially, beginning in the mid-
1990s [1]. Introduction of commercial laser-sintering systems used
to incrementally fuse layers of metal powder into solid objects
generated the shift from rapid prototyping to rapid manu-
facture by 2005 [2]. Early selective laser sintering (SLS) systems
utilized powders with polymer-coated powders, resulting in low-
density products requiring consolidation with hot isostatic pres-
sure (HIP) to achieve near-full density [3]. Parallel innovations
enabled the use of metal powders without coatings to be used in a
n
Corresponding author.
E-mail address: mower@ll.mit.edu (T.M. Mower).
concept grew into the AM process now referred to as direct metal
laser sintering (DMLS), which is technically a special instantiation
of SLS [5].
Comprehensive reviews of laser-based AM techniques are
available in the literature [1,2,6], and will not be provided here.
We note, however, that selective laser melting (SLM) systems
achieve melting of the powders and initially were applied only to
pure metals [1,2,6]. In contrast, SLS processes historically did not
heat the powders above the melting temperature, and produced
solids with somewhat less than full (ideal) density. Although we
have pointed out the most obvious difference between SLM and
SLS systems, the distinction between the two AM processes has
become blurred within the community of AM fabricators simply
because state-of-the-art DMLS systems are operated at power le-
vels sufcient to achieve full melting of the powders; thus, SLM
and DMLS are in practice used interchangeably [7]. These AM
techniques have both been termed powder bed fusion processes,
as dened by ASTM International [8].
Characterization of the mechanical properties of additive,
layered materials produced by laser-heating methods has not kept
pace with the proliferation of the systems put into service. Only a

http://dx.doi.org/10.1016/j.msea.2015.10.068
0921-5093/& 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
few reports published in the open literature present mechanical
properties of AM-produced AlSi10Mg materials. In their review of
AM aluminum alloys [9], Manfredi et al. include an example plot of
stressstrain data measured with AlSi10Mg produced by DMLS
and subjected to different heat treatments [10]. Buchbinder et al.
reported measured hardness and tensile strength of AlSi10Mg
specimens fabricated with their custom, high power SLM appa-
ratus [11]. Tensile stressstrain curves of AlSi10Mg produced in
two orientations with a customized SLM machine were presented
by Kempen et al. [12], and the elastic modulus, ultimate strengths
and strain to failure were compared to cast AlSi10Mg data mea-
sured by others.
Several reports of investigations of the characteristics of
Ti6Al4V materials produced with layered AM techniques have
been published recently. Early work by Brandl et al. compared the
fatigue strength of Ti6Al4V built with a laser-fused, wire-fed
process to that built with an electron-beam, powder-bed process
[13]. Stressstrain characteristics of Ti6Al4V produced with DMLS
and an experimental SLM machine were measured by Facchini
et al., but no differentiation of results between the two AM pro-
cesses were explained [14]. Fatigue properties of notched and
unnotched Ti6Al4V specimens generated with an in-house SLM
were presented by Van Hooreweder [15]. Crack growth rates and
monotonic stressstrain behavior were measured by Leuders et al.
in Ti6Al4V built with a commercial SLM (SLM250), and were dis-
cussed in relation to microstructural features [16]. Liu et al. studied
fatigue in Ti6Al4V specimens also produced with an SLM250, at-
tributing poor fatigue behavior to defects created by lack of fusion
[17]. A comparison of the mean fatigue strength of Ti6Al4V
produced with DMLS (EOSINT M270) versus an e-beam system
was presented by Chan et al., along with excellent microscopy of
crack-initiating defects [18]. Ra et al. measured tensile and fati-
gue properties of both Ti6Al4V specimens produced with an EOS
M270 (fatigue specimens were post-machined) [19]. Most re-
cently, Edwards and Ramulu studied Ti6Al4V produced with a
selective laser meting process (MTT 250), attributing the poor fa-
tigue behavior to porosity present in the AM materials [20]; these
authors stated that while some data on the fatigue performance
of AM titanium components is currently available, much more data
is still needed. While all of these published reports furnish data
relevant to behavior of AM Ti6Al4V, they do not provide denitive
indications of the mechanical behavior of materials built with the
most prevalent, state-of-the art AM machines.
Published reports concerning mechanical properties of stain-
less steels produced with powder-bed fusion techniques are much
more limited. Ra et al. [19] and Spierings et al. [21] both included
stainless steel (SS) 15-5 in their studies of mechanical properties of
DMLS materials. Investigations of the microstructure [22] and fa-
tiguecrack growth rates [23] in SS316L produced with SLM have
been reported. Spierings et al. also measured fatigue strengths of
SLM 316L, under tensile loading with load ratio of R 0.1. Most
recently, Murr et al. have described very detailed studies of the
microstructures of 17-4PH materials produced with SLM, relating
hardness of the materials to the type of gas used in the SLM
chamber, but no other mechanical property measurements were
reported [24].
1.2. Scope of the present work
This study was performed to contribute to the understanding of
the mechanical behavior of materials produced using additive
metal manufacturing processes presently available commercially.
Two categories of materials were studied: light metals (an alu-
minum alloy, AlSi10Mg and a titanium alloy, Ti6Al4V) and stainless
steels (316L and 17-4PH). Material specimens were fabricated at an
experienced facility using their typical process parameters, with
199
specimens produced in both vertical and horizontal orientations.
Tensile stressstrain behaviors were measured using at-plate
specimens in a universal test machine, and exural moduli were
measured with a dynamic mechanical analyzer (DMA). Measure-
ments of fatigue strengths were conducted in fully-reversed, ro-
tating bending mode. Reference material behaviors were obtained
from specimens produced with conventional techniques, i.e. sub-
tractive machining of wrought materials. The effect of surface
polishing was explored and is discussed in relation to corre-
sponding potential improvements of fatigue strength.
Extensive surface characterization was performed to establish
correlation between surface roughness and topography, and the
resultant mechanical behaviors, in particular fatigue strength.
Optical microscopy and prolometry were used to measure sur-
face roughness and to characterize intrinsic surface defects and
interior voids. Metallography was performed to reveal the material
microstructures and prevalence of interior voids within the ad-
ditive materials.
Tensile deformation measurements are compared in engineer-
ing stressstrain plots, showing the differences in yield and post-
yield behaviors between conventional materials and layered AM
materials. Measured fatigue-strength data are presented in tradi-
tional SN plots, including wrought and machined specimens to
provide a reference or baseline. Fatigue behaviors of the AM
specimens are compared to those of conventional materials, and
are discussed in relation to characteristics observed on the fracture
surfaces. Special attention is paid to the fatiguecrack initiation
sites, to explain why the AM materials exhibit lower fatigue
strengths.
2. Experimental details
2.1. Powder-bed material fabrication
The AlSi10Mg materials used for this study were fabricated
using an SLM 280, a selective laser melting system from SLM So-
lutions GmbH. Raw material powder was obtained from EOS
GmbH, and was Nadcap1 certied with particle sizes ranging from
20 to 63 mm. The build chamber was backlled with argon gas,
while powder melting and fusion was performed with a 100 W
laser with a spot size of 100 mm, traveling at 930 mm/s. Specimens
were built with layer thickness of approximately 50 mm, while
supported on a plate heated to 200 C. To relieve residual stresses
generated by the fabrication process, thermal stress relieving was
performed on these specimens at 300 C for 2 h, prior to removing
specimens from the support structures.
Additive Ti6Al4V specimens were fabricated using an EOS
M280, a DMLS system, also with argon gas and a laser having spot
size of 100 mm. The laser power and travel rate are considered
proprietary by the service provider.2 Powder for Ti6Al4V fabrica-
tion was also obtained from EOS GmbH, Nadcap certied with
particle sizes ranging from 15 to 45 mm. The resulting specimens
had a layer thickness of approximately 30 mm, and the tempera-
ture of the build platform during their fabrication was 35 C.
Thermal stress relieving was performed on these specimens at
650 C for 2 h. A separate lot of Ti6Al4V specimens was similarly
fabricated and subjected to hot isostatic pressure (HIP), at a tem-
perature of 900 C while under pressure of 102 MPa for two hours.
The two SS materials (316L and 17-4PH) studied were fabri-
cated using an EOS M270, with alloy powders obtained from EOS
1
Nadcap (formerly NADCAP, the National Aerospace and Defense Contractors
Accreditation Program).
2
Linear Mold and Engineering, Livonia, MI.
200 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213

Table 1
Chemical composition (weight %) of powder materials used in specimen fabrication.

Si Mg Fe Cu Zn Ni Ti Mn Al

AlSi10Mg (tested) 9.5 0.33 0.15 o 0.001 o 0.002 o 0.003 0.004 o 0.01 Balance

Fe Al V C O N H Ti

Ti6Al4V (tested) 0.25 6.5 4.5 0.08 0.13 0.03 0.0125 Balance

Cr Ni Mo Mn Si C S O N Fe

SS316L (tested) 16.9 10.9 2.4 1.3 0.5 0.02 0.003 0.04 0.18 Balance

Cr Ni Mo Mn Si C S O Cu Fe

SS17-4PH (typical) 16 4 o 0.5 o1 o 0.1 o 0.07 o 0.01 o 0.1 4 Balance

having particle sizes ranging from 15 to 45 mm. The build chamber
was backlled with nitrogen gas, while powder melting and fusion
was performed with a 195 W laser with a spot size of 100 mm,
traveling at 750 mm/s. Specimens were built on a plate main-
tained at 82 C, with layer thicknesses of approximately 40 and
20 mm for the 316L and 17-4PH, respectively. Thermal stress re-
lieving was performed prior to removing specimens from the build
platforms: 316L specimens were held at 388 C for four hours; 17-
4PH specimens were soaked at 788 C for one hour (Table 1).
2.2. Specimen preparation
The tensile specimens used in this study were traditional ten-
sile dog-bones, contoured to dimensions consistent with ASTM
Standard E8-09 as shown in the drawing in Fig. 1a. With laser
sintering, AlSi10Mg and Ti6Al4V specimens were produced in both
horizontal (specimens on edge) and vertical orientations (speci-
mens standing on end). The build chamber used for DMLS stain-
less steel specimens was not tall enough for them to stand verti-
cally, so they were built inclined at a 45-degree angle. Note that
horizontal specimens have the tensile axis parallel to the build
planes, while vertical specimens have the tensile axis perpendi-
cular to build planes. Conventional specimens were machined
from wrought, rolled sheet stock of certied materials, with the
tensile axis aligned with the plate roll direction (longitudinal). All
surfaces of the AM tensile specimens were left in the as-printed
condition, whereas the wrought materials were machined and
nished to a surface roughness of  1 mm, rms.
Cylindrical hourglass fatigue specimens were produced with
DMLS
25 cm
Wrought
DMLS
10 cm
Wrought
Fig. 1. Above, tensile specimens with plate thickness  3.8 mm. Below, fatigue
specimens with diameter  12.5 mm.
additive techniques in both horizontal and vertical orientations.
Most surfaces were not machined but were left in the condition
produced by the AM process. Subsets of AlSi10Mg and Ti6Al4V
were post-processed either with electrochemical polishing, re-
moving approximately 1015 mm of material from the surfaces, or
with a mechanical method using polishing compounds and cloths.
Subgroups Ti6Al4V and SS316L specimens were machined, po-
lished and/or subjected to HIP. Conventional specimens were
machined from wrought, rolled bar stock of certied Al6061,
Ti6Al4V, SS316L and 17-4PH. Surfaces of these specimens were
nely machined and polished with successively ner grits of
Emery paper.
2.3. Specimen surface characterization
Horizontally-built SLM AlSi10Mg fatigue-specimen surfaces are
shown in Fig. 2. The texture appears quite scaly, due to the discrete
nature of material fusing in narrow layers. The appearance of
many spots suggests the presence of voids, which are revealed
distinctly in the mechanically-polished surface. Although polish-
ing successfully removed high points on the surfaces, internal
voids continually emerged at the surface as material was removed
during the polishing process.
Surfaces of horizontally-grown DMLS Ti6Al4V specimens are
shown in Fig. 3. In the as-fabricated condition, a few pores are
evident amid the granular texture, nodular asperities and solidi-
ed beads of melted powder. Electropolished surfaces appeared
brighter, high points were smoothed substantially, and low regions
show a pooled look with pits no longer evident. Specimens grown
with their longitudinal axis oriented vertically displayed similar
features.
Optical microscopy of DMLS SS 316L fatigue-specimen surfaces
is shown in Fig. 4. On the horizontally-built material, discrete
bumps and a nely-scaled texture are seen clearly. Vertically-built
materials display similar features with a more coarsely scaled
surface, and many pores which often link together, effectively
forming microcracks.
Surface roughness of the fatigue specimens was measured
using an optical prolometer (Cybertechnologies CT-300) with a
white-light confocal sensor having height resolution of approxi-
mately 35 nm and lateral resolution of about 2 mm. Three-dimen-
sional images of typical surface prolometry obtained from mea-
surements on 17-4PH fatigue specimens are shown in Fig. 5. The
image from a wrought, machined and polished specimen displays
a peak-to-valley range of about 6 mm, while the DMLS horizon-
tally-built material exhibits signicantly rougher texture, with a
peak-to-valley range of over 40 mm. Surface roughness was
 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
Fig. 2. Optical microscopy of horizontally-built SLM AlSi10Mg fatigue-specimen surfaces. As fabricated (a) and mechanically polished (b).
measured on two fatigue specimens generated with each pro-
duction technique. Compiled in Table 2 is a summary of these
measurements. Specimens prepared with conventional techniques
(machining and polishing with emery paper) created surfaces with
measured average roughness (Sa) of 1 mm. Surfaces produced with
AM had signicantly greater roughness, ranging from Sa  5 to
15 mm.
Some of the surface roughness created by the discrete layering
of the additive manufacturing process could be reduced through
mechanical polishing. The effectiveness of this technique is lim-
ited, however, because as surface material is removed, bulk voids
appear to replace those removed. Electropolishing was even less
effective, achieving negligible reductions in surface roughness of
the AM materials.
2.4. Metallography
Unstrained tensile specimens were sectioned transverse to the
long axis to expose the microstructure of the materials studied.
Thus, the sections examined from AM specimens fabricated in the
horizontal orientation are perpendicular to the build plane, while
those from vertically-built materials are parallel to the build plane.
Polished aluminum samples were treated with Keller's etch, tita-
nium with Kroll's reagent, and stainless steel with an electrolytic
nitric-acid process.
Microscopy of the sectioned materials indicates that all of the
AM specimens were printed, or fabricated, with a series of laser
passes on specimen exteriors orthogonal to the direction used
throughout the bulk. Visible in the AlSi10Mg materials (Fig. 6) are
overlapping, segregated melt pools with distinct boundaries.
Fig. 3. Surfaces of horizontally-grown DMLS Ti6Al4V. As-fabricated (a) and electropolished (b).
201
Higher magnication (b and d) reveals a ne crystalline structure
created by the rapid solidication during fabrication. Voids are
apparent throughout the bulk of the AlSi10Mg specimens, grown
in both directions, but they appear more numerous in the mate-
rials grown in the vertical orientation. In addition, voids are more
prevalent in the surface layers, which is where applied stresses
would generally be highest.
Microstructures of the Ti6Al4V specimens are shown in Fig. 7.
The wrought/machined specimens of Ti6Al4V present dual-phase
material with isolated lamellar colonies interspersed be-
tween nearly equiaxed, connected grains having sizes ranging
from approximately 15 to 30 mm. Sections of the DMLS materials
reveal long, slender needle-like (acicular) grains (Fig. 6b, c). Si-
milar structure in Ti6Al4V produced by selective laser melting has
been identied previously as a fully martensitic, phase [25,26]
created by the extremely rapid cooling rate resulting from the
additive manufacturing process. It is noteworthy that the DMLS
Ti6Al4V materials are nearly void-free, in stark contrast to the
AlSi10Mg. An example of the microstructure of DMLS Ti6Al4V
material treated with HIP is shown in Fig. 6d. While the grains are
elongated, similar to those in the non-HIP'd material, the alpha
lath thickness after HIP is much greater.
Sections of DMLS 316L materials (Fig. 8a, b) show overlapping,
segregated melt pools with distinct boundaries. The micrographs
show structures reminiscent of weld llets, with few voids in the
bulk (more voids were evident near free surfaces, where applied
bending stresses would be highest). A ne crystalline structure is
apparent, stimulated by the rapid solidication of thin layers built
upon previously solidied layers. Similar microstructures devel-
oped in the 316L materials fabricated in both the horizontal and
202 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213

Fig. 4. DMLS 316L fatigue-specimen surfaces. Horizontally-built (a) and vertically-built (b).

inclined orientations. Table 2

Sections of DMLS 17-4PH materials built in the horizontal or- Roughness of fatigue specimen surfaces, measured with an optical prolometer.
ientation (Fig. 6c) also show segregated melt pools with discrete
Material Treatment Sa (lm) Srms (lm)
boundaries, a few small voids and a very ne crystalline micro-
structure. The 17-4PH material built on an incline (Fig. 6d) pre- Wrought Al6061 Machined and polished 1 1
sents a very different microstructure, with more numerous, larger SLM AlSi10Mg As fabricated 45 56
voids, only faintly visible melt pools, and randomly-oriented la- SLM AlSi10Mg Mechanically polished 1.5 2
SLM AlSi10Mg Electropolished 4.7 5.8
mellar colonies. WroughtTi6Al4V Machined and polished 1 1
DMLS Ti6Al4V As fabricated 1113 1316
2.5. Flexural modulus DMLS Ti6Al4V Mechanically polished 10 13
DMLS Ti6Al4V Electropolished 13 16
Wrought 316L Machined and polished 1 1
Elastic moduli were measured in exural mode in a dynamic DMLS 316L As fabricated 56 68
mechanical analyzer, TA Instruments model RSA-G2. Beam speci- Wrought17-4PH Machined and polished 1 1
mens (1  12  40 mm) were tested in three-point bending at an DMLS 17-4PH As fabricated 35 47
oscillation frequency of 1 Hz, at room temperature. Three speci-
mens of each material type were measured, and specimens were
approximately 10  4/s. Three samples of each material type were
each tested twice, once with each side up. Reported here are the
tested, using a non-contact video extensometer to measure strain;
moduli measured at a peak cyclic strain amplitude of 0.001.
errors in the determination of elastic moduli, introduced by this
instrument, may range from 1% to 3%.
2.6. Tensile testing

Tensile deformation measurements were conducted in an In- 2.7. Fatigue testing procedures
stron 8501 servohydraulic universal test machine at room tem-
perature. Tensile tests were executed in displacement control, Measurement of fatigue strength was carried out in fully re-
generating a strain rate in the specimen gage sections of versed, rotating beam mode with a Systems Integrators RBF850

Fig. 5. Surface maps from optical prolometry of 17-4PH fatigue specimens.

 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213 203

Fig. 6. Microstructure of SLM AlSi10Mg specimens. Horizontally built (a and b), sectioned perpendicular to build plane; vertically built (c and d), sectioned parallel to build
plane.

Fig. 7. Microstructures of Ti6Al4V; wrought (a), DMLS horizontal (b), DMLS vertical (c) and DMLS horizontal after HIP at 900 C and 102 MPa (d).
204 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213

Fig. 8. Microstructures of DMLS 316L built horizontally (a) and on a 45-degree incline (b); DMLS 17-4PH built horizontally (c) and on a 45-degree incline (d).

machine at room temperature and frequencies from 20 to 25 Hz. orientation (71.8 GPa).
Twelve samples of each material were tested, with stress ampli- The conventional, dual-phase Ti6Al4V used as a reference
tudes ranging from just below the yield strengths down to levels material here was received and tested in the annealed condition,
that produced lifetimes of 107 cycles or greater (run-out). producing a measured exural modulus of 114 GPa. Specimens of
Ti6Al4V produced with DMLS showed a slightly higher modulus,
with those fabricated in the vertical orientation again appearing
somewhat stiffer than horizontally produced material. Specimens
3. Results
of DMLS Ti6Al4V subjected to post-treatment with HIP demon-
strated a slight increase in measured exural modulus, as in-
3.1. Flexural modulus
dicated in Table 3.
Both stainless-steel DMLS materials demonstrated stiffness
Results of our exural modulus measurements are listed in
nearly equal to that of conventional (wrought and machined)
Table 3. The SLM aluminum proved to be about 10% stiffer than
corresponding materials, with specimens fabricated in the inclined
conventional (wrought and machined) Al 6061. Though these two
position exhibiting greater stiffness than specimens fabricated
materials are different, the comparison is useful since the only
horizontally. It is notable that the modulus of horizontally-built
aluminum available at present for commercial SLM or DMLS pro-
17-4PH is about 10% less than that of wrought material.
duction is AlSi10Mg. Specimens of this material produced with
SLM in the vertical orientation demonstrated slightly higher
3.2. Tensile deformation
modulus (73.8 GPA) than those fabricated in the horizontal
Representative engineering stressstrain data from measure-
Table 3
Measured exural modulus (GPa) of laser-fused materials and wrought counter-
ments obtained with our aluminum specimens are shown in
parts. Reported values are the averages of measurements from three specimens. Fig. 9a. Comparisons are being made here between the SLM Al-
Data spread due to specimen-to-specimen variations and measurement un- Si10Mg behavior and that of Al6061, because the former is the only
certainty is less than 1% of the values listed. aluminum alloy available presently with commercial SLM or DMLS
Material Long. Horiz. Vert.
processes, and the latter is the conventional alloy most widely
used in aerospace applications. The Al6061 specimens display
Wrought Al6061 65.5 classical, nearly elasticplastic behavior, with a well-dened yield
SLM AlSi10Mg 71.8 73.8 point and little strain hardening before reaching failure at strains
Wrought Ti6Al4V 114.0
DMLS Ti6Al4V 115.3 116.5
from 0.13 to 0.15.
DMLS Ti6Al4V (HIP) 117.5 119.4 SLM AlSi10Mg specimens experienced gradual yielding in re-
Wrought 316L 191.3 sponse to tensile deformation. This was followed by signicant
DMLS 316L 187.3 189.1 strain hardening, causing the materials to fail in a brittle manner
Wrought 17-4PH 193.9
at low strains: horizontally-grown specimens (for which the ap-
DMLS 17-4PH 172.2 192.9
plied tensile stress was aligned with the build plane) failed at
 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
Fig. 9. Typical measured room-temperature tensile deformation behavior of powder-bed, laser-fused AlSi10Mg and Ti6Al4V, compared to wrought materials.
strains of  0.04, while vertically-grown specimens (stressed
across the build layers) failed at strains of only  0.025. No ap-
parent thinning accompanied the tensile failures in AlSi10Mg
specimens, and the failure surfaces exhibited brittle, faceted fea-
tures as shown in Fig. 10b. Porosity throughout the bulk of the
AlSi10Mg is evident in this fracture surface, as well as the presence
of silicon particles (conrmed by energy dispersive X-ray spec-
troscopy) within host craters.
Data obtained from tensile deformation of our titanium speci-
mens are plotted in Fig. 9b. The conventional, wrought Ti6Al4V
exhibited ductile behavior, modest strain hardening and strain-to-
failure of approximately 0.1. The horizontally-grown DMLS
Ti6Al4V specimens showed higher yield strength but lower failure
strain than the conventional material. When stressed across the
build layers (vertical specimens) the deformation response is more
brittle, sustaining greater stress but failing at a much lower strain.
The HIP process eliminated any difference in measured mechanical
behavior between materials grown in the two orientations. Both
horizontal and vertical HIP'd specimens showed ductile behavior,
with lower yield stress than conventional material, more strain
hardening, and a greater failure strain. These differences are
summarized in Table 3, in which we listed the standard 0.2% offset
yield stress.
Tensile deformation induced well-dened yielding behavior in
the 316L materials, followed by strain hardening that continued
until failure occurred. Ultimate strains in the wrought 316L ex-
ceeded 30%, while the DMLS materials demonstrated nearly the
same ductility. The yield strength (0.2% offset) demonstrated by
DMLS 316L was signicantly higher ( 40%) than that of wrought
Fig. 10. Failure surfaces of tensile specimens; (a) Al6061, showing high ductility, and (b) horizontally-built SLM AlSi10Mg, showing brittle features with signicant porosity
and segregated silicon particles.
205
material, as seen in the stressstrain curves (Fig. 11) and the
summary data listed in Table 4. The 17-4PH materials demon-
strated less ductility than 316L but achieved signicantly higher
ultimate tensile strengths. Although the DMLS material fabricated
in the horizontal orientation (and thus, tested with the principal
stress aligned with the build planes) showed the lowest yield
strength of the 17-4 materials, it displayed the highest strain
hardening, such that it reached an ultimate strength nearly equal
to that of the wrought material. The most ductility in the 17-4
materials was produced in the DMLS specimens fabricated such
that they stood at 45 angle within the build chamber.
3.3. Fatigue-strength measurements
Results from fatigue tests conducted with aluminum-alloy
specimens are plotted in Fig. 12, in the form of a traditional SN, or
Whler diagram. The methodology of our measurements can be
veried by comparing our Al6061 data to the results of similar
fatigue tests published by ASM [27], which indicate slightly higher
fatigue strengths than generated in the present study; slightly
lower fatigue behavior for Al6061 is published in MMPDS [28],
thus our data are bounded by those two reference data sets.
The fatigue behavior of the SLM AlSi10Mg, in both horizontal
and vertical build orientations, is considerably lower than that of
wrought Al6061. (We iterate that this comparison is being made
because AlSi10Mg is the only SLM aluminum material available
commercially and hence may be considered as an alternate to
Al6061 for additive manufacture of components with complex
geometry.) For any given cyclic stress amplitude, the lifetime
206 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213

Fig. 11. Typical measured room-temperature tensile deformation behavior of stainless steels produced with DMLS, compared to that of annealed, wrought and machined
material.

Table 4 demonstrated by SLM AlSi10Mg is approximately an order of

Average measured tensile properties of laser-fused materials and wrought coun- magnitude lower than that of wrought Al6061. For any prescribed
terparts. Reported values are the averages of measurements from three specimens.
lifetime, the maximum cyclic stress endured by the SLM materials
Spread in strength data due to specimen-to-specimen variations and measurement
uncertainty is less than 1% of the values listed. in the as-fabricated condition is about 30% lower than the stress
tolerated by Al6061. This behavior is attributable to the presence
Material Orientation Modulus Yield Ultimate Failure of defects prevalent throughout the bulk of the AM material, as
(GPa) (MPa) (MPa) strain demonstrated in the micrographs reproduced in Fig. 13. This is a
Wrought Longitudinal 66.5 293 310  0.15
very typical example, where fatigue fracture initiated at a large
Al6061 surface void, and the early stages of growth were accompanied by
SLM AlSi10Mg Horizontal 65.5 227 358 0.039 large plastic strain as evidenced by torn rivulets and linked cavities
SLM AlSi10Mg Vertical 75.4 172 289 0.026 radiating outward from the origin. Surface polishing produced
Wrought Longitudinal 113.1 945 979  0.1
some enhancement of the fatigue strength of these materials, but
Ti6Al4V
DMLS Ti6Al4V Horizontal 108.8 972 1034 0.055 the degree of improvement was inconsistent. In cases where no
DMLS Ti6Al4V Vertical 114.9 1096 1130 0.012 benet was observed, initiations still occurred at surface defects as
DMLS Ti6Al4V Hor. & vert. 111.7 862 931  0.24 shown in Fig. 14a. In specimens where polishing generated im-
(HIP) proved fatigue behavior, fatigue fractures initiated at interior de-
Wrought Longitudinal 187 345 563  0.30
316L
fects, as shown in Fig. 14b.
DMLS 316L Horizontal 180 496 717 0.28 Fatigue-strength data generated from measurements with our
DMLS 316L 45 Incline 193 473 680  0.30 Ti6Al4V specimens are plotted in Fig. 15. The conventional mate-
Wrought 17- Longitudinal 186 898 1085 0.065 rial displayed a low-cycle fatigue strength on the order of
4PH
700 MPa, declining to a strength between 630 and 680 MPa for a
DMLS 17-4PH Horizontal 165 610 1072 0.072
DMLS 17-4PH 45 Incline 186 737 914 0.086 lifetime of 10 million cycles. This is consistent with other en-
durance limits reported recently for wrought, as-received material
tested in the rotating-beam conguration, which range from 600
to 670 MPa [29,30].
The drastically lower fatigue strength demonstrated by the
DMLS Ti6Al4V specimens, in both horizontal and vertical or-
ientations, is very striking. For example, the material is susceptible
to fatigue cracking with applied stresses of approximately 50 MPa,
with lifetimes ranging from a few tens of thousands to one hun-
dred thousand cycles. This vulnerability is owed to the presence of
many defects upon the surface of the materials, many of which
initiate fatigue cracks that grow simultaneously until they link up
and prompt nal failure (as shown in Figs. 16 and 17). Although
some of the obvious external defects could be removed by me-
chanical polishing, surface defects remained due to the emergence
of occasional voids from within the bulk during material removal.
Treatment of DMLS specimens with either mechanical polishing
or electropolishing produced negligible improvement in fatigue
behavior, consistent with little change in measured surface
roughness.
Signicant enhancement in fatigue strength was achieved by
subjecting DMLS Ti6Al4V material to HIP, as evidenced by the data
Fig. 12. Measured room-temperature stress-life (SN) of SLM AlSi10Mg, compared
in Fig. 15. The fatigue strength of the HIP'd material, produced with
to that of conventional Al 6061. Bending fatigue (rotating beam) at a frequency of
25 Hz. DMLS in either orientation and subsequently machined (turned on
a lathe), ranges from about 500 MPa at 5  104 cycles, to an ap-
parent endurance limit (107 cycles) of 400 MPa. Fatigue fractures
 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
Fig. 13. Fracture surface of horizontally-built SLM AlSi10Mg fatigue specimen; (a) quasi-brittle planar failure surface, and (b) enlarged view of initiating defect, with torn
rivulets and linked cavities.
Fig. 14. Fracture surfaces of polished, horizontally-built SLM AlSi10Mg fatigue specimens; (a) initiation at surface void, and (b) initiation at interior cavity.
Fig. 15. Measured room-temperature stress-life (SN) of DMLS Ti6Al4V, compared
to that of conventional material. Bending fatigue (rotating beam) at a frequency of
25 Hz.
initiated at sub-surface defects, as illustrated by the two examples
shown in Fig. 17, and propagated in a ductile manner with evi-
dence of plastic ow radiating from the initiation points. Although
part of this enhanced fatigue strength resulted from reduction of
surface roughness by the machining process, DMLS specimens
treated with HIP but not post-machined also showed very
207
signicant improvement in fatigue strength (though the scatter
was high in this group of specimens, due to the as-fabricated
surface defects). Elevation of the fatigue strength by HIP treatment
to the strength of the wrought material did not occur, at least in
part because the HIP'd DMLS material still possesses the alpha
platelet structure shown earlier in Fig. 7c. (The microstructure of
cast-plus-HIP material contains similar alpha platelets and grain-
boundary alpha, and is known to display fatigue behavior inferior
to that produced by equiaxed Ti6Al4V [31].)
Measured cyclic fatigue strengths of stainless steel 316L mate-
rials are shown in Fig. 18. DMLS materials fabricated in the hor-
izontal orientation (and therefore subjected to bending stresses
aligned with the build planes) produced fatigue strengths ap-
proximately 15% lower than those of the wrought material, with
substantially higher scatter in the data. The rst batch of DMLS
specimens built in the vertical orientation (stressed across the
build planes in the fatigue tests) produced such signicantly lower
fatigue strengths that we suspected a deviation in process para-
meters may have occurred during the production process. A sec-
ond batch of vertically-built specimens produced identically poor
fatigue strength. A third batch of vertically-built 316L specimens
was post-processed with hot isostatic pressure (HIP), at 1190 C
and 145 MPa for four hours. This HIP process resulted in no im-
provement in low-cycle fatigue strength but did improve the high-
cycle fatigue behavior, such that vertically-built, HIP'd material
generated high-cycle fatigue strengths of approximately 50% of
that of the horizontally-built materials.
Fatigue fractures of DMLS 316L materials produced in the
horizontal orientation displayed rough features characteristic of
ductile crack growth, as demonstrated by the example shown in
208 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213

Fig. 16. Fracture surface of DMLS Ti6Al4V horizontally-built fatigue specimen. Multiple surface initiation sites promoted fatigue failure, accompanied by secondary internal
cracking.

Fig. 19a. In this case, the fatigue crack was initiated at a single
dominant defect just beneath the surface. Other fractures of these
materials incurred multiple fatigue cracks, initiating at defects
near the surface.
The poor fatigue strengths exhibited by vertically-built DMLS
316L materials in this study is apparently due to an intrinsic low
strength between build planes, as suggested by the nearly planar
fracture surface shown in Fig. 19b. Though fatigue cracks often
were triggered by surface aws, in this instance a cluster of in-
terior voids provided additional initiating defects. Trace evidence

Fig. 17. Fatigue fractures initiated at singular subsurface defects in machined, horizontally-built DMLS HIPd Ti6Al4V, with ensuing fracture surfaces exhibiting more ductile
features than in non-HIPd, similar materials.
 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
Fig. 18. Measured room-temperature stress-life (SN) of DMLS 316L, compared to
that of conventional 316L. Bending fatigue (rotating beam) at a frequency of 25 Hz.
of the laser paths can be seen clearly upon this fracture surface,
typical of all the vertically-built specimens.
Subjecting the vertically-built 316L to HIP appears to have more
fully joined the material layers together, as evidenced by the
fracture surface shown in Fig. 20. This fracture surface is non-
planar and rougher than those that occurred in non-HIP'd, verti-
cally-built 316L. The fatigue fracture initiated at multiple surface
defects, rather than internal porosity, and no artifacts of the laser
processing are evident.
Results from fatigue tests conducted with 17-4PH specimens
are plotted in Fig. 21. Fatigue strength displayed by the wrought
material was nearly 900 MPa in the low-cycle regime, and de-
creased in log-linear fashion to an apparent endurance limit of
about 450 MPa. DMLS specimens fabricated in the horizontal or-
ientation produced nearly the same fatigue strength as conven-
tional material, but with more noticeable scatter. The rst batch of
DMLS 17-4PH specimens built in the vertical orientation (thus
stressed across the build planes in these tests) demonstrated cyclic
fatigue strengths of approximately 60% of those measured with
wrought material. A second batch of vertically-built DMLS 17-4PH
produced much poorer fatigue behavior, displaying only about 25%
of the fatigue strength of both wrought and DMLS materials built
in the horizontal orientation. Inspection of the fracture surfaces in
the two batches of vertically-built specimens indicates signicant
differences in the laser-fusion process must have occurred, though
records corresponding to the fabrication jobs do not reveal any
inconsistency.
Fig. 19. (a) Fracture surface of horizontally-built DMLS 316L fatigue specimen, displaying ductile crack growth initiating from large fabrication defect. (b) Nearly planar
fracture of vertically-built DMLS 316L, initiating at surface defects and subsurface voids, and exhibiting traces of laser paths.
209
Surface aws in the rst batch of vertically-built DMLS 17-4PH
fatigue specimens were relatively benign, such that cracks in-
itiated at dominant, internal defects as shown in Fig. 22. These
materials appeared to be completely fused, with no laser traces or
segregated beads of melted and solidied powder evident on the
surface. Fracture surfaces were rough and tortuous, indicative of
ductile growth. In stark contrast, fracture surfaces of specimens
from the second batch of vertically-built material showed clear
evidence of incomplete fusion. The fracture surface shown in
Fig. 23a suggests that specimen separated precisely between two
build planes, and shows many artifacts of the laser path. The view
with higher magnication in Fig. 23b reveals segments of material
that had been melted, pressed at against a solid, and solidied
without fusing. Also abundantly evident are beads of solidied
powder that had been melted but not fused by the laser.
4. Discussion
4.1. Behavior of SLM AlSi10Mg
Choosing to use AlSi10Mg may be driven by the ability to create
complex shapes using AM techniques not possible with conven-
tional, subtractive machining of wrought aluminum alloys, such as
Al6061. Comparison of the mechanical behavior of AlSi10Mg with
that of Al6061 is therefore a relevant and important evaluation.
The tensile yield stress of SLM AlSi10Mg proved to be con-
siderably lower than that of Al6061. When tested with the prin-
cipal stress aligned parallel to the AM layers (horizontal build or-
ientation), the measured yield stress (227 MPa) of AlSi10Mg was
about 80% of the measured yield stress of Al6061 (293 MPa). When
tested across AM layers (vertically built specimens), the yield
stress was signicantly lower: only about 172 MPa. The tensile
tests produced strain-to failure values of approximately 34% for
AlSi10Mg specimens.
SLM AlSi10Mg exhibited fatigue strengths of about 60% of that
of Al6061, for AM specimens fabricated in both the horizontal and
vertical orientations. Fatigue cracks initiated at stress-concentrat-
ing defects, such as voids and segregated silicon particles, both
near the surface and within the interior. Surface polishing enabled
enhancement of fatigue behavior in some instances, but the pre-
valence of large voids throughout the bulk of this material resulted
in defects continually remaining on the surface during the pol-
ishing process.
In another study of the fatigue behavior of SLM AlSi10Mg,
Brandl et al. [32] demonstrated that the sensitivity of fatigue
strength upon orientation of the AM fabrication process was
210 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
Fig. 20. Fracture surface of vertically-built, HIPd DMLS 316L fatigue specimen; (a) non-planar surface formed from linkage of several discrete fatigue cracks initiating at
surface defects (b).
Fig. 21. Measured room-temperature stress-life (SN) of DMLS 17-4PH, compared
to that of conventional material. Bending fatigue (rotating beam) at a frequency of
25 Hz.
greatly diminished by the use of a building platform maintained at
300 C. The measurements presented here are consistent with that
result, but show noticeably poorer fatigue performance. This dif-
ference can be attributed to two factors: our testing was con-
ducted in fully-reversed bending mode, as opposed to the tension
tension mode (R0.1) used by Brandl et al., and their specimens
were machined after AM processing, which may have removed
Fig. 22. Fatigue fracture of specimen from rst batch of DMLS 17-4 fabricated vertically, with initiation at interior defect, and rough fracture surface demonstrating fully
fused material.
some of the largest defects on the surface, likely to be initiation
sites for fatigue cracks. Perhaps the most signicant indication
obtained by Brandl et al. is that the fatigue behavior of AM Al-
Si10Mg fabricated with powder bed fusion can be enhanced by
3050% via the conventional T6 thermal post-treatment.
Because high porosity of SLM AlSi10Mg may limit the nature
and prevalence of its useful applications, efforts to increase the
density through variations in scanning geometry and rate have
been conducted recently [33]. Increase in the density was achieved
with an optimized strategy, from approximately 95% to over 99%
(relative to theoretical), but the inuence upon mechanical prop-
erties was not measured. An important investigation remaining is
the measurement of the fatigue strength of AM AlSi10Mg mate-
rials with optimally high densities, as-fabricated (not machined)
surface conditions, and subjected to HIP and/or other heat
treatments.
4.2. Behavior of DMLS Ti6Al4V
Although tensile yielding in DMLS Ti6Al4V occurred at higher
levels of stress than in wrought material, as-fabricated (and stress-
relieved) DMLS Ti6Al4V materials exhibited brittle failures at low
strains, ranging from 0.01 to 0.03. (We note that higher failure
strains for similar materials have been reported, ranging from 0.03
to 0.06 [19], 0.04 [20], to 0.08 [14].) Treatment with HIP appears to
have produced an annealing effect in the DMLS Ti6Al4, lowering
the tensile yield stress and enabling signicant strain before fail-
ure, reaching values of approximately 0.2.
 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
Fig. 23. Fracture surface from second batch of vertically-built DMLS 17-4PH fatigue specimens; (a) planar surface with incomplete fusion evident and (b) re-solidied
powder beads, several pores.
The limited ductility of as-fabricated DMLS Ti6Al4V and the
prevalence of signicant defects at or near the surface of test
specimens create the very low fatigue strength demonstrated by
this material. Results of our fatigue testing indicate as-fabricated
DMLS Ti6Al4V fatigue strength, for both orientations of building,
was so low that implementation of this material would not be
suitable for most applications subjected to large cyclic loads.
Surface polishing did not achieve noteworthy augmentation of
fatigue behavior, partly because surface pits persist even after
polishing, and also because fatigue cracks were often initiated at
sub-surface voids. The DMLS materials treated with HIP and ma-
chined demonstrated signicantly better cyclic fatigue behavior,
with strength levels of almost 70% of those exhibited by wrought
material. This improvement in fatigue behavior can be attributed
to compaction of bulk voids by the HIP process and reduction of
surface defects by machining, and the capacity of the HIP'd ma-
terial to sustain heightened plastic strain before failure. HIP'd
DMLS specimens with as-fabricated surfaces also demonstrated
enhanced fatigue strength, but with more scatter (or unpredict-
ability of strength) due to the presence of surface defects.
A recently published study concurs that treatment of DMLS
Ti6Al4V with HIP (also 920 C and 103 MPa) alters the plastic ow
behavior signicantly, increasing the failure strain from about 5%
to over 15%, even though the associated microstructural transfor-
mations are subtle in appearance and minor in extent [34]. Other
studies have been initiated to measure the inuence of HIP pro-
cessing on fatigue behavior of laser-melted Ti6Al4V [35], but those
results have not yet been published. This work has demonstrated
substantial increase in the fatigue strength of DMLS Ti6Al4V due to
HIP, but greater improvement is desired still, to levels comparable
to those of wrought material.
Further enhancement of fatigue strengths might be achieved by
transforming the microstructure in the HIP'd Ti6Al4V, which re-
tained characteristics of the as-fabricated material (martensitic
structure with long, slender acicular grains) known to be suscep-
tible to initiation of fatigue fractures [31]. Several uncommon post-
heat-treatment cycles were recently applied to samples of ad-
ditive-layer manufactured Ti6Al4V, creating nearly equiaxed,
globular -grains but with very large size (5003000 mm); all
thermal cycles transcended the -transus temperature (  1000 C)
and it was found not possible to achieve smaller grain sizes that
are typically associated with optimal mechanical behaviors [36].
Other recent attempts to alter the microstructure of powder-bed
fused Ti6Al4V utilized heat-treating cycles remaining below the
transus temperature, producing microstructures with more
equiaxed, bimodal characteristics that endowed the specimens
tested with greater ductility but lower yield strength [26]. It
211
remains to be shown whether application of such unconventional
heat treatments can improve the cyclic fatigue behavior of HIP'd,
powder-bed fused Ti6Al4V.
4.3. Behavior of DMLS 316L
High ductility was exhibited by DMLS 316L during tensile-de-
formation measurements, reaching elongations of approximately
30% at failure. These materials yielded at stresses much higher
than the yield stress of wrought, annealed 316L (345 MPa): when
tested with the principal stress aligned with the build planes,
specimens (built horizontally) yielded at about 496 MPa; speci-
mens fabricated in the 45-degree inclined position produced an
average yield stress of approximately 473 MPa. The higher yield
stress displayed by these DMLS materials can be attributed to the
ne crystalline structure created by the rapid solidication during
the process of building the material in thin, discrete layers.
The conventional (wrought, annealed, machined and polished)
316L material studied here displayed a fatigue strength of 350 MPa
at 105 cycles, which agrees well with the result reported
( 380 MPa) by Puchi-Cabrera et al. for similar conditions [37].
(The lower value here is consistent with our lower measured ul-
timate stress (563 MPa) compared to 700 MPa reported by Puchi-
Cabrera.) Cyclic fatigue strengths demonstrated by the DMLS
stainless steels fabricated in the horizontal orientation were al-
most equal to those of the wrought material, but were accom-
panied by greater scatter in the data due to the presence of defects
in these materials.
DMLS 316L specimens built vertically, in two separate fabrica-
tion runs, produced fatigue strengths at lifetimes of 105106 cycles
of only about 30% of that generated by horizontally-built DMLS
material. In response to a cyclic stress amplitude of 110 MPa, our
vertically-built 316L failed at a little over 106 cycles, which com-
pares quite well to a lifetime of 2  106 cycles reported by Riemer
et al. for similar, vertically-built material in the as-fabricated
condition, tested with R  1 in axial fatigue [23]. Machining the
surface of their specimens raised the reported fatigue strength
signicantly (at 2  106 cycles), to 267 MPa. Subjecting their
turned specimens to HIP increased the fatigue strength by ap-
proximately 20%, to 317 MPa. Because post-machining materials
produced additively negates the very feature that makes the
manufacturing technique attractive, we treated as-built specimens
to HIP to investigate whether that process could enhance the fa-
tigue life of vertically-built DMLS 316L materials. As indicated in
Fig. 18, the fatigue tests demonstrated an improvement in high-
cycle fatigue strength but had little discernable impact upon low-
cycle behavior. This result may be explained by the HIP process
212 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
reducing the prevalence of smaller-sized defects that initiate
cracks at high cyclic lifetimes, but having little effect upon the
larger defects that govern the high stress, low-cycle fatigue be-
havior. Although the HIP temperature was nearly 200 C below the
melting temperature of 316L, it appears that the HIP process ef-
fectively joined layers together more completely, as evidenced by
rougher, non-planar fracture surfaces in the HIP'd materials.
4.4. Behavior of DMLS 17-4PH
The tensile yield stress measured in DMLS 17-4 PH materials
was signicantly lower than that displayed by wrought material in
the mill-annealed condition ( 900 MPa). Yielding in specimens
built horizontally occurred at about 610 MPa, whereas the average
yield stress in specimens built at a 45-degree incline was 737 MPa.
This difference in strength suggests the two materials may have
different microstructures, which is substantiated by Fig. 8c and d.
While the DMLS 17-4PH material fabricated in the horizontal
orientation has ne, evenly distributed grains, material created
with the specimen built on a 45-degree incline appears very dif-
ferent. This material presents a structure composed of groups of
layered formations, suggesting they are martensitic lamellae
within remaining austenite [38], formed during brief cooling cy-
cles while the material was additively built. A similar-looking 17-
4PH microstructure was described by Deng et al. as martensitic
laths within austenite [39]. The difference in microstructure
within materials produced in two different orientations but
otherwise using nominally similar parameters in the DMLS pro-
cess provides compelling evidence that further study should be
focused upon the evolution of microstructure in DMLS materials.
With cyclic stresses aligned with the build planes, horizontally-
fabricated DMLS 17-4PH specimens demonstrated fatigue
strengths nearly equal to those of wrought material. In contrast,
two separate batches of vertically-built, DMLS 17-4PH fatigue
specimens generated disparate results. One batch of these mate-
rials exhibited fatigue strengths of about 25% of those of hor-
izontally-built material. In these specimens, fracture surfaces dis-
played denitive evidence that the layered manufacturing process
achieved incomplete fusion of material across the build planes,
causing an intrinsic weakness responsible for the very low fatigue
strengths observed. The second batch of vertically-built 17-4PH
materials produced distinctly improved cyclic fatigue behavior,
with fatigue strengths of about 75% of that generated by hor-
izontally-built materials. Fracture surfaces of these specimens in-
dicated that complete melting and fusion of the powder had oc-
curred during the fabrication process. Since no differences were
observed on the exterior surfaces of these two batches of osten-
sibly identical materials, an important concern arises: How will
end-users and providers of DMLS fabrication services develop
assurance that components created with this AM process possess
optimal material properties? This issue appears to be of greatest
signicance for products that will be exposed to dynamic me-
chanical loads, since fatigue failures are most often not accom-
panied by gradual signs of declining health.
5. Summary
Mechanical behavior of four metallic alloys fabricated with
layered, laser-heated methods of additive manufacturing (AM) was
measured, and compared to of corresponding conventional
(wrought and machined) materials. The elastic moduli measured
both in exure and in tension were similar between the additive
and conventional materials. The tensile deformation behavior of
laser-fused AlSi10Mg and Ti6Al4V was limited by inherent brittle
mechanisms, resulting in strain-to-failure values of only 14%.
High ductility in the DMLS 316L was demonstrated, with con-
siderably higher yield strength and strain hardening than in
wrought, annealed 316L. In contrast, yielding of the DMLS 17-4PH
materials fabricated both horizontally and at a 45-degree incline
occurred at lower values than in the conventional material.
Cyclic fatigue-strength measurements were conducted in fully-
reversed, rotating bending. The fatigue strength exhibited by SLM
AlSi10Mg materials was measured to be about 60% of the fatigue
strength of conventional, wrought and machined Al6061. Fatigue
fractures in the AlSi10Mg were initiated both at surface defects,
such as voids and irregularly-shaped fusion features, and at in-
ternal pores and large particles of segregated silicon. Neither
mechanical nor electrochemical polishing was effective at enhan-
cing fatigue behavior, due to abundant defects that remained ever-
present during polishing. Measured fatigue strength of as-fabri-
cated DMLS Ti6Al4V was shown to be considerably lower than that
of wrought, annealed material. Low ductility and numerous fab-
rication defects in the AM material are responsible for the poor
fatigue behavior exhibited, which could be improved only slightly
by electropolishing. Laser-fused Ti6Al4V materials post-machined
and treated with HIP produced improved fatigue strengths, ap-
proaching 70% of those demonstrated by wrought, annealed
material.
Measured fatigue strengths of DMLS stainless steels fabricated
in the horizontal orientation (stressed within the build planes)
were between 85% and 95% of the fatigue strength of corre-
sponding wrought materials. Vertically-built DMLS 316L and 17-
4PH are prone to fatigue fractures developing prematurely by se-
paration of the material across the build planes, resulting in sig-
nicantly reduced fatigue strengths. In the case of 316L, two se-
parate batches of vertically-built specimens reinforced the ex-
istence of this behavior and therefore the need to consider post-
processing methods to reduce its severity. Hot isostatic pressure
was applied to another batch of vertically-built DMLS 316L spe-
cimens, which displayed improved high-cycle fatigue behavior but
no enhancement of high-amplitude, low-cycle fatigue life. In the
case of 17-4PH, distinctly different fatigue behaviors in two bat-
ches of vertically-built specimens stemmed from incomplete fu-
sion between material layers in the inferior batch. These results
demonstrate that, although the quasi-static mechanical behavior
of DMLS materials compares favorably with that of conventional
materials, concern remains about assuring that DMLS processing
parameters are controlled well enough to provide adequate fatigue
strength for dynamic applications.
Acknowledgment
This work is sponsored by the Department of the Air Force
under the United States Air Force contract number FA8721-05-C-
0002. The opinions, interpretations, recommendations and con-
clusions are those of the authors and are not necessarily endorsed
by the United States Government. Donation of follow-on batches
of fatigue specimens by Linear Mold and Engineering is gratefully
acknowledged.
References
[1] J.-P. Kruth, G. Levy, F. Flocke, T.H.C. Childs, Consolidation phenomena in laser
and powder-bed based layered manufacturing, Ann. CIRP 56 (2) (2007)
730759.
[2] J.-P. Kruth, B. Vandenbroucke, P. Mercelis, J. Van Vaerenbergh, Benchmarking
of different SLS/SLM processes as rapid manufacturing techniques, in: Pro-
ceedings of the International Conference on Polymers and Moulds Innovations
(PMI) 2005, Gent, Belgium, 2005, pp. 2636.
[3] M. Shellabear, O. Nyrhil, DMLS development history and state of the art, in:
 T.M. Mower, M.J. Long / Materials Science & Engineering A 651 (2016) 198213
Proceedings of the 4th International Conference on Laser Assisted Net Shape
Engineering, LANE 2004, Erlangen, Germany, 2004, pp. 393404.
[4] M. Agarwala, D. Bourell, J. Beaman, H. Marcus, J. Barlow, Direct selective laser
sintering of metals, Rapid Prototyp. J. 1 (1) (1995) 2636.
[5] J.-P. Kruth, P. Mercelis, J. Van Vaerenbergh, L. Froyen, M. Rombouts, Binding,
mechanisms in selective laser sintering and selective laser melting, Rapid
Prototyp. J. 11 (1) (2005) 2636.
[6] D.D. Gu, W. Meiners, K. Wissenbach, R. Roprawe, Laser additive manufacturing
of metallic components: materials, processes and mechanisms, Int. Mater. Rev.
57 (3) (2012) 133164.
[7] Linear Mold and Engineering, Private communication.
[8] ASTM Standard F2792, Standard Terminology for Additive Manufacturing
Technologies, ASTM International, West Conshohocken, PA, 2012.
[9] D. Manfredi, et al., Additive manufacturing of Al alloys and aluminium matrix
composites (AMCs), in: W.A. Monteiro (Ed.), Light Metal Alloys Applications,
InTech, ISBN: 978-953-51-1588-5.
[10] M. Krishnan, Investigation of material and mechanical properties of Al alloy
and Al based MMC parts produced by DMLS for industrial application (Ph.D.
thesis), Politecnico di Torino, 2014.
[11] D. Buchbinder, H. Schleifenbaum, S. Heidrich, W. Meiners, J. Bltmann, High
power selective laser melting (HP SLM) of aluminum parts, Phys. Procedia 12
(2011) 271278.
[12] K. Kempen, L. Thijs, J. Van Humbeek, J.-P. Kruth, Mechanical properties of
AlSi10Mg produced by selective laser melting, Phys. Procedia 39 (2012)
439446.
[13] E. Brandl, C. Leyens, F. Palm, A. Schoberth, P. Onteniente, Wire instead of
powder? Properties of additive manufactured Ti6Al4V for aerospace ap-
plications, in: Proceedings of the Euro-uRapid International Conference on
Rapid Prototyping, Rapid Tooling & Rapid Manufacturing, Berlin, Germany,
2008.
[14] L. Facchini, E. Magalini, P. Robotti, A. Molinari, S. Hges, K. Wissenbach, Duc-
tility of a Ti6Al4V alloy produced by selective laser melting of pre-alloyed
powders, Rapid Prototyp. J. 16 (6) (2010) 450459.
[15] B. Van Hooreweder, R. Boonen, D. Moens, J.-P. Kruth, P. Sas, On the determi-
nation of fatigue properties of Ti6Al4V produced by selective laser melting, in:
Proceedings of the 53rd AIAA/ASME Structures, Structural Dynamics and
Materials Conference, Honolulu, Hawaii, 2012.
[16] S. Leuders, M. Thne, A. Riemer, T. Niendorf, T. Trster, H.A. Richard, H.J. Maier,
On the mechanical behavior of titanium alloy Ti6Al4V manufactured by se-
lective laser melting: fatigue resistance and crack-growth performance, Int. J.
Fatigue 48 (2013) 300307.
[17] Q. Liu, J. Elambasseril, S. Sun, M. Leary, M. Brandt, P.K. Sharp, The effect of
manufacturing defects on the fatigue behavior of Ti6Al4V specimens fab-
ricated using selective laser melting, Adv. Mater. Res. 891892 (2014)
15191524.
[18] K.S. Chan, M. Koike, R.L. Mason, T. Okabe, Fatigue life of titanium alloys fab-
ricated by additive layer manufacturing techniques for dental implants, Metall.
Mater. Trans. A 44A (2013) 10101022.
[19] H.K. Ra, T.L. Starr, B.E. Stucker, A comparison of the tensile, fatigue and
fracture behavior of Ti6Al4V and 155 PH stainless steel parts made by
selective laser melting, Int. J. Manuf. Technol. 69 (2013) 12991309.
[20] P. Edwards, M. Ramulu, Fatigue performance of selective laser melted Ti6Al
4V, Mater. Sci. Eng. A 598 (2014) 327337.
[21] A.B. Spierings, T.L. Starr, K. Wegener, Fatigue performance of additive
213
manufactured metallic parts, Rapid Prototyp. J. 19 (2) (2013) 8894.
[22] E. Yasa, J.-P. Kruth, Microstructural investigation of selective laser melting
316L stainless steel parts exposed to laser re-melting, Procedia Eng. 19 (2011)
389395.
[23] A. Riemer, S. Leuders, M. Thone, H.A. Richard, T. Troster, T. Niendorf, On the
fatigue crack growth behavior in 316L stainless steel manufactured by selec-
tive laser melting, Eng. Fract. Mech. 120 (2014) 1525.
[24] L.E. Murr, E. Martinez, J. Hernandez, S. Collins, K.N. Amato, S.M. Gaytan, P.
W. Shindo, Microstructures and properties of 17-4PH stainless steel fabricated
by selective laser melting, J. Mater. Res. Technol. 1 (3) (2012) 167177.
[25] L. Thijs, F. Verhaeghe, T. Craeghs, J.V. Humbeeck, J.-P. Kruth, A study of the
micro structural evolution during selective laser melting of Ti6Al4V, Acta
Mater. 58 (2010) 33033312.
[26] B. Vranken, L. This, J.-P. Kruth, J.V. Humbeek, Heat treatment of Ti6Al4V pro-
duced by selective laser melting: microstructure and mechanical properties, J.
Alloy. Compd. 541 (2012) 177185.
[27] J.G. Kaufman, Properties of Aluminum Alloys Fatigue Data and the Effects of
Temperature, Product Form, and Processing, ASM International, Metals Park,
Ohio, 2008 (Fig. 6061.RB0).
[28] Metallic, Materials Properties Development and Standardization (MMPDS-06),
Battelle Memorial Institute, Columbus, Ohio, 2011 (Figure 3.6.2.2.8).
[29] M. Zhu, R. Radefeld, R. Vasquez, Effect of the step-test method upon the fa-
tigue strength of Ti6Al4V, in: S. Shrivastava (Ed.), Proceedings of the Materials
and Processes for Medical Devices Conference, ASM International, Metals Park,
Ohio, 2008, pp. 215219.
[30] F. Boucher, S. Calloch, C. Doudard, T. Ducrocq, N. Vongsana, Towards a rapid
determination of the fatigue strength of a hip prosthesis, in: Proceedings of
the 56th Annual Meeting of the Orthopaedic Research Society New Orleans,
LA, Poster No. 2216, 2010.
[31] M.J. Donachie, A. Titanium, Technical Guide, ASM International, Metals Park,
Ohio (1988), p. 188.
[32] E. Brandl, U. Heckenberger, V. Holzinger, D. Buchbinder, Additive manu-
factured AlSi10Mg samples using selective laser melting (SLM): micro-
structure, high cycle fatigue, and fracture behavior, Mater. Des. 34 (2012)
159169.
[33] N.T. Aboulkhair, N.M. Everitta, I. Ashcroft, C. Tuck, Reducing porosity in Al-
Si10Mg parts processed by selective laser melting, Addit. Manuf. 14 (2014)
7786.
[34] C. Qiu, N.J.E. Adkins, M.M. Attallah, Microstructure and tensile properties of
selectively laser-melted and HIPed laser-melted Ti6Al4V, Mater. Sci. Eng. A
578 (2013) 155163.
[35] K.D. Rekedal, D. Liu, Fatigue life of selective laser melted and hot isostatically
pressed Ti6Al4V absent of surface machining, in: Proceedings of the 56th
AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Con-
ference, Kissimmee, FL, January 2015.
[36] E. Brandl, D. Greitemeier, Microstructure of additive layer manufactured
Ti6Al4V after exceptional post heat treatments, Mater. Lett. 81 (2012) 8487.
[37] E.S. Puchi-Cabrera, F. Matnez, I. Herrera, J.A. Berros, S. Dixit, D. Bhat, On the
fatigue behavior of an AISI 316L stainless steel coated with a PVD TiN deposit,
Surf. Coat. Technol. 182 (2004) 276286.
[38] R.E. Smallman, A.H.W. Ngan, Physical Metallurgy and Advanced Materials,
Elsevier, Oxford, 2007, pp. 416420.
[39] D. Deng, R. Chen, Q. Sun, X. Li, Microstructural study of 17-4PH stainless steel
after plasma-transferred arc welding, Materials 8 (2015) 424434.