ARTICLE
Additive manufacturing of 3D nano-architected
metals
Andrey Vyatskikh
1
, Stéphane Delalande
2
, Akira Kudo
1
, Xuan Zhang
3
, Carlos M. Portela
1
& Julia R. Greer
1
Most existing methods for additive manufacturing (AM) of metals are inherently limited to
~20
–
50
μ
m resolution, which makes them untenable for generating complex 3D-printed
metallic structures with smaller features. We developed a lithography-based process to
create complex 3D nano-architected metals with ~100 nm resolution. We
fi
rst synthesize
hybrid organic
–
inorganic materials that contain Ni clusters to produce a metal-rich photo-
resist, then use two-photon lithography to sculpt 3D polymer scaffolds, and pyrolyze them to
volatilize the organics, which produces a
>
90 wt% Ni-containing architecture. We demon-
strate nanolattices with octet geometries, 2
μ
m unit cells and 300
–
400-nm diameter beams
made of 20-nm grained nanocrystalline, nanoporous Ni. Nanomechanical experiments reveal
their speci
fi
c strength to be 2.1
–
7.2 MPa g
−
1
cm
3
, which is comparable to lattice architectures
fabricated using existing metal AM processes. This work demonstrates an ef
fi
cient pathway
to 3D-print micro-architected and nano-architected metals with sub-micron resolution.
DOI: 10.1038/s41467-018-03071-9
OPEN
1
Division of Engineering and Applied Sciences, California Institute of Technology, 1200 E. California Blvd., Pasadena, CA 91125, USA.
2
Scienti
fi
c Department,
PSA Group, Centre Technique de Vélizy 2, route de Gizy, Vélizy-Villacoublay 78943, France.
3
Center of Advanced Mechanics and Materials, Department of
Engineering Mechanics, Tsinghua University, Beijing 10084, China. Correspondence and requests for materials should be addressed to
J.R.G. (email:
jrgreer@caltech.edu
)
NATURE COMMUNICATIONS
|
(2018) 9:593
|
DOI: 10.1038/s41467-018-03071-9
|
www.nature.com/naturecommunications
1
1234567890():,;
A
dditive manufacturing (AM) represents a set of processes
that enable layer by layer fabrication of complex 3D
structures using a wide range of materials that include
ceramics
1
, polymers
2
, and metals
3
. The development of metal
AM has revolutionized the production of complex parts for
aerospace, automotive and medical applications
4
,
5
. Today
’
s
resolution of most commercially available metal AM processes is
~20
–
50
μ
m
6
; no established method is available for printing 3D
features below these dimensions
7
. It has been shown that unique
phenomena arise in metals with micro-dimensions and nano-
dimensions, for example light trapping in optical meta-materials
8
and enhanced mechanical resilience
9
–
15
. Accessing these
phenomena requires developing a process to fabricate 3D metallic
architectures with macroscopic overall dimensions and individual
constituents in the sub-micron regime.
Minimum feature size in metal AM is generally limited by the
material feedstock, i.e., the method of supplying metal in powder,
wire, sheet or ink form during fabrication. Inkjet-based
methods
16
,
17
manipulate 40
–
60
μ
m droplets of metal inks,
limiting the smallest features to at least the size of a solidi
fi
ed
droplet. Wire-based and
fi
lament-based processes, such as plasma
deposition
4
and electron beam freeform fabrication (EBF3)
18
, rely
on locally melting a
>
100
μ
m-diameter metal wire, which pro-
duces millimeter-sized features. Powder-based processes, such as
selective laser melting (SLM) and laser engineered net shaping
19
,
consolidate ~0.3
–
10
μ
m metal powder particles, which limits the
smallest feature size to about 20
μ
m
6
,
20
. Overcoming these reso-
lution limitations requires a capability to manipulate nanoscale
quantities of metals in a stable and scalable 3D printing process.
Alternative material feeds to fabricate 3D metal structures with
<
10
μ
m resolution include nanoparticle inks, ion solutions, dro-
plets of molten metal, and precursor gases
7
. Methods that use
localized electroplating
21
,
22
or metal ion reduction
23
,
24
are
capable of producing features down to 500 nm using a very slow
process that is limited by electroplating rate. Electrochemical
fabrication (EFAB) allows for manufacturing geometries with
10-
μ
m features and 4-
μ
m layers, but is limited to structures with a
total height of 25
–
50 layers
25
. Other technologies, like micro-
deposition of metal nanoparticle inks
26
–
28
or molten metal
29
and
focused ion beam direct writing, also suffer from slow throughput
and are more suited for low-volume fabrication and repair
30
.
We demonstrate a facile and reproducible process to create
complex 3D metal geometries with a resolution of 25
–
100-nm.
We synthesize hybrid organic
–
inorganic materials that contain Ni
clusters and use them to produce a metal-rich photoresist. We
then use two-photon lithography (TPL) to sculpt computer-
designed architectures out of the resist and pyrolyze them
fi
rst in
inert atmosphere at 1000 °C and then in reducing atmosphere at
600 °C to volatilize the organic constituents. Using this approach,
we demonstrate the fabrication of periodic Ni octet nanolattices
with the unit cell size of 2
μ
m and beam diameters of 300
–
400 nm
diameter as a proof-of-concept. TEM analysis reveals that the
microstructure of Ni beams is nanocrystalline and nanoporous,
with a 20 nm mean grain size and 10
–
30% porosity within each
beam. Nanomechanical experiments demonstrate that the
strength of these Ni nanolattices is comparable to that of the
metal lattices with 0.1
–
1.0 mm beam diameters fabricated using
alternative metal AM technologies. These
fi
ndings suggest an
ef
fi
cient pathway to create complex 3D metal structures with
nano-scale resolution.
Results
AM of nickel nano-architectures
.We
fi
rst synthesized nickel
acrylate using a ligand exchange reaction between nickel alkoxide
and acrylic acid (Fig.
1
a) and combined it with another
acrylic monomer, pentaerythritol triacrylate, and a photoinitiator,
7-diethylamino-3-thenoylcoumarin (Fig.
1
b). We then drop cast
this photoresist on silicon substrate and used TPL to sculpt the
prescribed 3D architectures (Fig.
1
c). The non-polymerized resist
was then washed away, and the free-standing cross-linked
polymer nano-architectures were then pyrolyzed to volatilize the
organic content. This process yielded a replica of the original 3D
structure with ~80% smaller linear dimensions made entirely out
of metal (Fig.
1
d).
We demonstrate the feasibility and ef
fi
ciency of this metho-
dology by
fi
rst fabricating nanolattices with 10
μ
m octet unit cells
comprised of 2-
μ
m diameter circular beams out of the
synthesized photoresist using layer-by-layer TPL with 150 nm
layer thickness. Scanning electron microscopy (SEM) images in
Fig.
1
f
–
h reveal that these nanolattices had fully dense beams and
uniformly sized, high-
fi
delity features. Each sample had four unit
cells per side, 40
μ
m, and a height of three unit cells, 30
μ
m, and
was supported by vertical springs at each corner and by a vertical
pillar in the center. These supports served as pedestals that would
allow the sample to release from substrate after undergoing an
isotropic ~80% shrinkage during pyrolysis (see Supplementary
Fig.
1
).
Pyrolysis was performed in a
tube furnace via a two-step
procedure:
fi
rst at 1000 °C in argon to remove most of the organic
content from the samples and to consolidate the Ni metal clusters
into denser features, which is acco
mpanied by ~5× li
near shrinkage
in feature size; and second at 600 °C in forming gas, to reduce the
oxygen content in the mostly-Ni samples and to facilitate grain
growth. SEM images in Fig.
1
(i, j) show a representative 3D Ni
architecture and convey that the 10-
μ
munitcellsand2-
μ
mdiameter
beams in the original polymer-meta
l structure shrank to produce ~2
μ
m unit cells and ~300
–
400 nm diameter beams in the nickel
nanolattice. This also implies t
hat 150-nm layer thickness in the
polymer structure corresponds to 30-nm layer thickness in the metal
structure. The zoomed-in image in Fig.
1
j shows that the metal
beams are ~10
–
30% porous caused by pyrolysis.
Microstructure and chemical composition of as-fabricated
metallic 3D architectures
. Chemical composition of the as-
fabricated Ni architectures was characterized using energy-
dispersive X-ray spectroscopy (EDS), for which we fabricated
individual unit cells with tetrakaidecahedron geometries using the
same methodology. Figure
2
a shows that these structures shrunk
from 20-
μ
m wide unit cells and 2-
μ
m diameter beams on 6-
μ
m
pillar supports to 4-
μ
m unit cells and 0.4-
μ
m diameter beams
after pyrolysis (Fig.
2
b). EDS spectrum (Fig.
2
d) taken from a
beam section shown in Fig.
2
c reveals the chemical composition
to be 91.8 wt% Ni, 5.0 wt% O, and 3.2 wt% C. A Si peak from the
substrate is also present. EDS maps in Fig.
2
e
–
h convey a rela-
tively homogeneous distribution of each element within the
printed structure, which consists mostly of nickel metal and is not
segregated into individual nickel-rich, carbon-rich, or oxygen-
rich phases.
We also fabricated some few-micron long, 25
–
100-nm
diameter metal beams that spanned the 1.25-
μ
m wide opening
in a silicon nitride membrane directly on the transmission
electron microscopy (TEM) grids (Fig.
3
a) to analyze the atomic-
level microstructure of pyrolyzed materials. Figure
3
b displays a
bright-
fi
eld TEM image taken along a beam that reveals
multiple coalesced grains with a mean size of 21.4
±
2.0 nm (see
Supplementary Table
1
). The electron diffraction pattern (Fig.
3
d)
taken from the region shown in Fig.
3
c conveys a strong Ni signal
and a much weaker contribution from NiO. A representative
high-resolution TEM (HRTEM) image (Fig.
3
e) of the beam edge
contains multiple lattice fringes, which allowed the calculation of
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03071-9
2
NATURE COMMUNICATIONS
|
(2018) 9:593
|
DOI: 10.1038/s41467-018-03071-9
|
www.nature.com/naturecommunications
interplanar atomic spacings using fast Fourier transform (FFT).
We identi
fi
ed three distinct spacings: Ni crystals (region 1,
spacings of 2.01 and 2.04 Å), Ni
3
C particles (region 2, spacings of
1.98 and 2.14 Å), and NiO crystals (region 3, spacing of 2.06 Å).
More details can be found in the Methods section. Bright-
fi
eld
TEM revealed that Ni crystals occupy
>
90% of the examined
volume, NiO
<
10%, and Ni
3
C
<
1%, consistent with EDS results
(Fig.
2
d and Supplementary Fig.
3
). TEM analysis further revealed
the presence of nickel (II) oxide nanoparticles with diameter of
<
5 nm at the surface that were likely formed through surface
oxidation in air after TEM sample preparation. Our pyrolysis is
equivalent to carbothermal reduction at 1000
̊
C followed by
reduction by hydrogen and carbon at 600
̊
C, with no oxygen
present in the
fl
owing gas. Literature on this type of thermal
treatment reported the composition to be mainly metallic nickel
with a minor amount of nickel carbide and/or carbon
31
.
In situ compression of nickel nanolattices
. We conducted uni-
axial compression experiments on ten Ni octet nanolattices with
relative densities of 27
–
42% and beam sizes of 300
–
400 nm (see
Supplementary Table
2
). The experiments were conducted in situ,
in a SEM-based nanomechanical instrument, comprised of a
nanoindenter-like module (Nanomechanics, Inc.) inside of SEM
Galvo mirror
fs laser
Si chip
Glass
Spacer
63× objective
Oil
Printed part
Metal-containing
photoresist
Metal
clusters
Pyrolysis
Metal
grains
10
μ
m
2
μ
m
Organic
ligands
7-Diethylamino-2-thenoyl
coumarin
Nickel acrylate
Pentaerythritol
triacrylate
Metal-containing
photoresist
Metal precursor
Acrylic resin
Photoinitiator
Nickel 2-methoxyethoxide
Acrylic acid
2-methoxyethanol
O
O
O
O
O
O
O
O
OH
O
O
N
O
O
O
S
O
OO
O
OH
+
2
+
2
HO
Ni
2+
O
––
O
Ni
2+
O
––
O
Ni
2+
O
––
O
10
μ
m
2
μ
m
a
b
fgh
ij
c
de
Fig. 1
Process for nanoscale additive manufacturing of metals and SEM characterization of the fabricated samples.
a
Ligand exchange reaction used to
synthesize metal precursor with cross-linking functionality.
b
Metal precursor, acrylic resin, and photoinitiator are mixed to form a transparent metal-
containing photoresist.
c
Schematic of two-photon lithography (TPL) process used to sculpt the scaffold.
d
Schematic of fabrication of metal-containing
polymer part that is
e
pyrolized to remove organic content and to convert the polymer into a metal. SEM images of
f
–
h
a representative octet lattice made
out of a nickel-containing polymer at different magni
fi
cations and
i
,
j
a representative nickel nanolattice after pyrolysis. Magni
fi
cations in
g
and
i
(scale bars
2
μ
m) and also
h
and
j
(scale bars 500 nm) are identical. Scale bar is 15
μ
m for
f
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03071-9
ARTICLE
NATURE COMMUNICATIONS
|
(2018) 9:593
|
DOI: 10.1038/s41467-018-03071-9
|
www.nature.com/naturecommunications
3
chamber (Quanta 200 FEG, FEI), which enabled observing the
deformation while simultaneously collecting load vs. displace-
ment data
32
(see Supplementary Movie
1
). The collected data
were converted into engineering stresses and strains by dividing
the load by the sample footprint area and dividing the displace-
ment by the initial sample height, respectively. Figure
4
a
–
d shows
SEM snapshots obtained during a compression experiment of a
representative sample; stress vs. strain data for four representative
samples are shown in Fig.
4
e (data for additional six samples are
presented in Supplementary Fig.
4
). All stress
–
strain data appear
to be self-consistent and reproducible. A toe region in the initial
portion of each experiment (not shown) is representative of
deformation before establishing full contact between the sample
and
fl
at punch indenter tip (see Supplementary Fig.
5
for full
stress
–
strain data). The toe region also included the failure of the
supporting pillar, which allowed for establishing full contact
between the sample and the substrate.
We found that the stress vs. strain data was typical for cellular
solids compressions, with the characteristic elastic loading,
plateau, and densi
fi
cation sections
33
. The arrows on the plot are
correlated with the images above and demarcate speci
fi
c stages
during compression: initial contact (region A), elastic deforma-
tion (region B), layer-by-layer collapse (region C), and densi
fi
ca-
tion (region D). The point of full contact was determined using
harmonic contact stiffness and SEM video. The slope of the elastic
loading segment, up to 10
–
15% strain (region B), was used to
estimate structural stiffness of the nanolattices, which ranged
from ~47 to 174 MPa. The strength of Ni nanolattices was de
fi
ned
as the maximum stress prior to the
fi
rst buckling event, marked
by open circles in the data in Fig.
4
e, and ranged from 6.9 to 18.2
MPa. The elastic region was followed by layer-by-layer collapse
up to 65% strain (region C); two of the four samples were
unloaded at 30 and 60% strain. The two other samples were
compressed to 70
–
85% strains, reached densi
fi
cation (region D)
and then unloaded (see Supplementary Movie
1
). None of the
nanolattices recovered after deformation.
Discussion
EDS analysis revealed that the fabricated nanolattices have a
composition of 91.8 wt% Ni, 5.0 wt% O, and 3.2 wt% C. It is
reasonable to expect traces of carbon in the pyrolyzed structures
caused by the high solubility of carbon in Ni at 1000 °C
34
, which
leads to carbon precipitation at nickel surface upon cooling down
to room temperature. TEM analysis revealed that the carbon also
exists in the form of 5 nm-sized Ni
3
C precipitates within the
beams (Fig.
3
e). The accuracy of EDS in quantifying the carbon
content may not be suf
fi
cient because it is sensitive to the spur-
ious carbon deposited in the SEM chamber
35
. The presence of 5.0
wt% O in the nanolattice can be attributed to formation of a
native oxide on Ni surface and to full oxidation of small (
<
6 nm)
Ni surface nanoparticles
36
.
Figure
4
f shows the speci
fi
c strength of Ni nanolattices
fabricated in this work and those of the metallic lattices fabricated
using other metal AM processes as a function of beam diameter
on a log
–
log plot (see Supplementary Table
2
for details).
This plot reveals that the speci
fi
c strength of metallic lattices in
refs.
16
,
37
–
41
decreases by a factor of 280 as the beam diameter is
reduced from 1.78 to 0.04 mm, with the lowest reported speci
fi
c
strength of 0.7 MPa g
−
1
cm
3
for octahedral silver lattices. Nano-
crystalline Ni nanolattices in this work have the speci
fi
c strength
of 2.1
–
7.2 MPa g
−
1
cm
3
, which is ~2
–
10 times higher than that of
Ni L
α
1,2
As printed
Pyrolyzed
C K
α
1,2
O K
α
1
Si K
α
1
a
91.75 ± 0.15
Ni
Ni
Ni
Ni
*Excluding Si
Si
O
C
012345678910
Energy (keV)
Intensity (a.u.)
Element
Wt%*
O
C
5.02 ± 0.08
3.23 ± 0.11
b
c
d
ef
gh
Fig. 2
Energy dispersive spectroscopy (EDS) characterization of fabricated metal nanosctructures.
a
SEM images of supported 20
μ
m tetrakaidekahedron
unit cell on a Si chip before pyrolysis and
b
the same structure after pyrolysis (4
μ
m width).
c
SEM image of the structure showing where EDS data was
collected.
d
EDS spectrum taken from an internal beam region reveals chemical composition to be more than 90 wt% nickel.
e
–
h
EDS maps show high
uniformity of the atomic composition throughout the structure. Scale bars are 5
μ
m for
a
,1
μ
m for
b
and 2
μ
m for
c
,
e
–
h
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03071-9
4
NATURE COMMUNICATIONS
|
(2018) 9:593
|
DOI: 10.1038/s41467-018-03071-9
|
www.nature.com/naturecommunications
octahedral silver lattices with ~40-
μ
m diameter beams
16
and
~2
–
7 times higher than the stainless steel lattices with ~200
μ
m
diameter beams
37
. It appears to be on the same order as NiTi
octahedral lattices with ~250-
μ
m diameter beams
40
and
AlSi10Mg diamond lattices with ~400
μ
m beams
41
. This suggests
that the AM process developed in this work is capable of pro-
ducing architectures with feature sizes that are an order of
magnitude smaller than those fabricated using existing AM
processes while retaining high strength. The speci
fi
c strength
calculations were performed with the assumption of monolithic
beams, which leads to its underestimation because the nano-
crystalline Ni within the beams has 10
–
30% residual porosity.
Some of the existing literature on the deformation of nano-
porous metallic foams
42
and individual metallic nano-pillars
10
,
43
report higher strengths upon uniaxial compression than ones
reported in this work. The key difference between the strength
reported in this work and those in previous reports is that it is
representative of the structural strength of the nanolattice, where
each beam has heterogeneous porous microstructure, as well as
each nodal junction, and both are subjected to a complex stress
state upon global compression. The microstructure that comprises
nanolattices in this work is nanocrystalline and nanoporous, and
has different levels of hierarchy in the sense that each individual
beam is nanocrystalline and nanoporous, as well as the entire
structure. This microstructure within the individual beams stems
from sintering of the Ni nanoparticles after the organic compo-
nents volatilize; it's in distinct contrast to the monolithic metallic
beams in all other literature on the deformation of nanoporous
materials. This microstructure is detrimental to the overall
structural strength in two ways: (1) the additional porosity within
each beam lowers the overall relative density of the architecture,
and (2) upon mechanical deformation, each sintered junction
experiences a local stress state, which creates an effective stress
concentration in the material at an adjacent pore. The pores that
border these regions of local stress concentrations can be viewed
as notches or
fl
aws that serve as locations of failure initiation upon
mechanical loads. The distribution of nano-pores in each beam
that comprises the nanolattices in this work leads to a distribution
in the local failure strengths, which
—
in combination with the
detrimental effects of lower relative density and the presence of
junctions
—
serves to lower the overall structural strength.
The speci
fi
c strength of the Ni nano-lattices in this work is
50
–
80% lower than that of Cu meso-lattices with a similar relative
density reported in ref.
11
, which likely stems from the lattice
strength being governed by that of the monolithic, fully dense
beams with grains spanning full beam width. The strength of
nanoporous Au stochastic foams in ref.
42
was reported to be
close to that of monolithic gold because each ligament is a vir-
tually defect-free, single crystalline beam, whose strength
approaches ideal strength of gold
42
. These foams have a funda-
mentally different microstructure compared to the nanolattices in
this work in that they are stochastic foams with relatively slender,
curved single-crystalline pristine beams. A direct comparison
between the compressive strengths of nanocrystalline Ni
nanolattices in this work and those of hollow lattices reported in
refs
12
,
14
,
15
,
32
,
44
may be misleading, because this work is focused
on solid-beam metallic nanolattices, which deform via compres-
sion and plastic
fl
ow upon uniaxial compression; the others
(200)
(111)
Ni
(220)
(220)
(220)
NiO
1
19.4
≤
≤
23.4
5.1
≤
≤
8.1
0
2
4
6
8
10
12
10
15
20
25
30
35
40
Particle size (nm)
Counts
a
f
b
c
d
e
2.04Å
2
.
0
4
Å
1 Ni
2 Ni
3
C
3 NiO
(002)
(111)
(111)
2.01Å
2
.
0
1
Å
2.14Å
2
.
1
4
Å
2.06Å
2
.
0
6
Å
1.98Å
1
.
9
8
Å
2
3
-
(113)
(113)
(006)
(200)
(220)
-
Fig. 3
TEM characterization of the resulting metal structure.
a
SEM image of nickel beams fabricated directly on a 200-nm thick SiN membrane TEM grid
b
Low-magni
fi
cation TEM of a 100 nm nickel beam overhanging the edge of 1.25-
μ
m hole in a SiN membrane.
c
TEM image of the metal sample region where
the diffraction pattern was taken.
d
Electron diffraction pattern shows that the printed beam consists mostly out of nanocrystalline nickel with a small
amount of nickel oxide.
e
HRTEM image of a printed metal beam. Analysis of atomic plane distances using FFT shows predominantly nanocrystalline nickel
(region 1) with some amount of nickel carbide in the interior (region 2) and nickel oxide at the surface (region 3).
f
Grain size histogram for
n
=
40 particles
measured from a TEM image showing 95% con
fi
dence intervals for the mean grain size (
μ
) and the standard deviation (
σ
) (see Supplementary Fig.
2
and
Supplementary Table
1
). Scale bars are 1
μ
m for
a
, 100 nm for
b
, 50 nm for
c
, and 5 nm for
e
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03071-9
ARTICLE
NATURE COMMUNICATIONS
|
(2018) 9:593
|
DOI: 10.1038/s41467-018-03071-9
|
www.nature.com/naturecommunications
5
contain hollow shell beams and undergo a different deformation
mechanism upon compression that includes shell buckling and
layer-by-layer collapse.
Figure
5
shows minimal reported printed feature sizes
demonstrated in this work and many other metal AM processes
available today (see Supplementary Table
3
). The plotted ranges
include both layer thickness and minimum lateral feature size.
The minimum z-feature is determined by the thickness of a single
layer of material. The minimum lateral feature is de
fi
ned by
multiple factors, which include the energy beam spot size and
control over the melt pool. The data in Fig.
5
demonstrates that
the AM process developed in this work is capable of producing
features that are an order of magnitude smaller compared to
those produced by other 3D-capable AM processes.
Another key aspect of any metal AM process is the throughput.
Using hybrid organic
–
inorganic photoresist developed in this
work allows for writing speeds of 4
–
6mms
−
1
, which is ~100
times faster than that for TPL of metal salts
24
. Comparing the
speeds of metal AM processes with different resolutions can be
accomplished by normalizing the write speed (
μ
ms
−
1
) by the
feature size (
μ
m) or by normalizing the volumetric throughput
(
μ
m
3
s
−
1
) by the voxel volume (
μ
m
3
) (see ref.
7
). For a typical
300
–
600 nm feature size printed by TPL
45
, writing speeds in this
work correspond to de
fi
ning 6700
–
20000 voxels s
−
1
, a printing
speed that is out of reach for state-of-the-art micro-scale metal
AM techniques, i.e., electrohydrodynamic printing (0.05
–
300
voxel s
−
1
), local electroplating (0.04
–
1.0 voxels s
−
1
), focused
beam methods (0.01
–
0.8 voxels s
−
1
), and direct ink writing
This work
Hirt et al., 2016
Takai et al., 2014
Skylar-Scott et al., 2016
Visser et al., 2015
Electrochemical fabrication (EFAB)
Saleh et al., 2017
Regenfuss et al., 2007
Digital metal
Selective laser melting (SLM)
Direct metal laser sintering (DMLS)
LaserCUSING
Metal powder bed fusion
Digital part materialization
Kullman et al., 2012
Electron beam melting (EBM)
Laser metal deposition (LMD)
Direct metal deposition (DMD)
Ultrasonic consolidation (UC)
Electron beam free form fabrication (EBF3)
Rapid plasma deposition
0.1
1
10
100
Minimum feature range (
μ
m)
1000
10,000
Fig. 5
Comparison of minimum feature sizes for commercial and potentially scalable metal additive manufacturing technologies. Using metal-containing
photoresist allows to fabricate complex 3D geometries with the resolution that is an order of magnitude
fi
ner than that of the state-of-the-art metal AM
methods. See Supplementary Table
3
for data and references
Ni, this work
Ti-6Al-4V, EBM
38
Ti-6Al-4V, SLM
39
AlSi10Mg, DMLS
41
NiTi, SLM
40
SS 316L, SLM
37
Ag, ink-based
16
10
2
10
1
10
0
10
0
10
1
10
2
Beam size (
μ
m)
10
3
c
b
a
d
e
f
B
C
D
A
Specific strength (MPa /(g cm
–3
))
40
30
20
10
0
0.0
0.2
0.4
0.6
0.8
Strain
Stress (MPa)
Fig. 4
In situ uniaxial compression of 3D printed nickel octet nanolattices.
a
–
d
SEM images of the nickel structure during the compression test
a
before full
contact,
b
in the elastic regime,
c
during layer-by-layer collapse, and
d
during densi
fi
cation.
e
Stress
–
strain data for compression of four nickel nanolattices.
Letters on the graph correspond to the regions represented by
a
–
d
.
f
Speci
fi
c strength-beam size plot showing properties of nickel nanolattices compared
to other metal lattices fabricated using selective laser melting (SLM), direct metal laser sintering (DMLS), electron beam melting (EBM), and ink-b
ased
methods. See Supplementary Table
2
for data and references. Scale bars are 5
μ
m for
a
–
d
ARTICLE
NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03071-9
6
NATURE COMMUNICATIONS
|
(2018) 9:593
|
DOI: 10.1038/s41467-018-03071-9
|
www.nature.com/naturecommunications