Roy et al.

Microsystems & Nanoengineering (2019)5:64

https://doi.org/10.1038/s41378-019-0116-8
Microsystems & Nanoengineering
www.nature.com/micronano

ARTICLE Open Access

A novel microscale selective laser sintering (μ-SLS)

process for the fabrication of microelectronic parts
1
Nilabh K. Roy , Dipankar Behera1, Obehi G. Dibua1, Chee S. Foong2 and Michael A. Cullinan1

Abstract
One of the biggest challenges in microscale additive manufacturing is the production of three-dimensional, microscale
metal parts with a high enough throughput to be relevant for commercial applications. This paper presents a new
microscale additive manufacturing process called microscale selective laser sintering (μ-SLS) that can produce true 3D
metal parts with sub-5 μm resolution and a throughput of greater than 60 mm3/hour. In μ-SLS, a layer of metal
nanoparticle ink is first coated onto a substrate using a slot die coating system. The ink is then dried to produce a
uniform nanoparticle layer. Next, the substrate is precisely positioned under an optical subsystem using a set of coarse
and fine nanopositioning stages. In the optical subsystem, laser light that has been patterned using a digital
micromirror array is used to heat and sinter the nanoparticles into the desired patterns. This set of steps is then
repeated to build up each layer of the 3D part in the μ-SLS system. Overall, this new technology offers the potential to
overcome many of the current limitations in microscale additive manufacturing of metals and become an important
process in microelectronics packaging applications.
1234567890():,;
1234567890():,;
1234567890():,;
1234567890():,;

Introduction technologies as well11,12. However, there are significant

As the additive manufacturing (AM) technologies are challenges when scaling conventional AM down to the
maturing, more and more industries are looking into the microscale including limited build rates, part complex-
possibilities of employing AM as part of their production ities, and types of materials that can be used. In order to
ecosystems. In addition to the aerospace industry, which overcome these limitations, this paper presents a new
has been one of the earliest adopters and investors of the microscale AM technique called microscale selective laser
technology, newer avenues are being explored in the sintering (μ-SLS) that can produce true three-dimensional
medical, transportation, consumer products, energy, metal structures with a feature -size resolution of better
tooling, art, jewelry, construction, and microelectronics than 5 μm. While the process has primary applications in
industries1–8. With the current market value of over $7 microelectronics packaging, it can be used to fabricate
billion, the growth of AM has been phenomenal over the microscale actuators, sensors, and micro-optoelectronic
last 8 years with an average annual growth of 27%9,10. components as well. Therefore, future applications of this
While most of this growth has concentrated on macro- μ-SLS process will likely include the manufacturing of
scale AM tools and products, the continued demand for high performance designs of complex MEMS and medical
miniaturization of devices in the electronics, biotechnol- devices (see Fig. 1b) which are difficult to fabricate using
ogy, medical, automotive, and optics sectors is fueling more conventional processes like Si-micromachining,
increased interest in the development of microscale AM electroplating, and etching13–17.
The current focus application of the μ-SLS process is
the microelectronics packaging industry where the fabri-
Correspondence: Michael A. Cullinan ([email protected])
1
cation of microscale, 3D interconnect structures is
Department of Mechanical Engineering, The University of Texas at Austin, 204
becoming increasingly important. This is because the
E. Dean Keeton St, Austin, TX 78712, USA
2
NXP Semiconductors, 6501W William Cannon Dr, Austin, TX 78735, US trend for increased miniaturization of electronic devices
These authors jointly supervised this work: Chee S. Foong, Michael A. Cullinan

© The Author(s) 2019

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction
in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if
changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If
material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 2 of 14

a
Future technology
Old technology Current technology
laser sintered high
solder bumps electroplated Cu pillars
density Cu pillars

b c

200 μm

Fig. 1 Potential applications of the μ-SLS process. a Evolution of C4 solder bumps(adapted from ref. 33)- The previous technology used ~1:1
aspect ratio SnAg bumps with >100 μm diameter features and >150 μm pitch. Current state-of-the-art technologies use sub-100 μm electroplated
copper pillars to reduce the overall footprint of ICs. However, the process limitations of electroplating make it difficult to fabricate large-area, high
aspect ratio sub-50 μm resolution structures. The μ-SLS process can enable the fabrication of the next generation of copper pillar-like structures with
applications in flip-chip electronics industry. b Ni-Co compliant microforceps fabricated using a hybrid-AM technology useful in minimally invasive
microsurgeries (adapted from ref. 34. Reproduced with permission from Emerald Publishing Limited). The high throughput capabilities of μ-SLS
process can drive the production scale fabrication of essential microproducts like these. c A representation of large-area fabrication potential (1.2 ×
1.2 × 1.2 mm3) of the μ-SLS process. Complex lattice-like microstructures (inset) have applications in developing extremely lightweight structural
elements.

and the demand for higher performance devices with strength. Additionally, Cu pillar bumps are also shown to
greater functionalities has resulted in the continual have higher electrical and thermal conductivities and
reduction in transistor size and more and more transistors better electromigration reliability. As a result, pillar
getting packed into the same area on a chip18,19. However, bumps are increasingly being adopted to replace solder
there is a large disconnect between the current scaling bumps20,21. As we move to the next generation of these
trends in semiconductor chip manufacturing and the pillar bumps (from 90 μm pillar diameter/40 μm pillar
capabilities of the back-end-of-line fabrication processes height to 10 μm pillar diameter/50 μm pillar height—see
used for packaging of these chips. For example, the solder Fig. 1a), the aspect ratio of the bumps keeps getting higher
bumps which have traditionally served as the interconnect which is a problem since electroplating at this scale is
between two dies or between the die and a substrate are limited to a maximum aspect ratio of ≈2:122. Thus, fab-
traditionally manufactured through an electroplating ricating these structures is out of the capabilities of the
process. As the solder bump process involves re-melting current electroplating processes used to make the pillar
the solder balls, there is poor control over the morphology bump arrays.
of the reflowed spheres. Fabricating these bumps below One possible approach to make these structures is to
125 μm pitch is difficult due to the decrease in standoff additively manufacture these 3D interconnect structures.
height and poorer joint reliability at smaller pitches. Unfortunately, none of the currently available microscale
These often lead to device failure due to short-circuiting. AM processes can produce these pillar-like structures
To overcome these problems, more advanced electronics with the rates, resolutions, and feature complexities
packages use copper pillar bumps which allow for finer required for the next generation of electronics packaging.
pitches between interconnect structures and more precise For example, two photon lithography and interference
gap control between the two surfaces being connected lithography work only with polymers,23,24 so producing
while maintaining sufficient standoff for mechanical metal bumps would require some post processing steps,
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 3 of 14

Optical setup Particle bed formation setup

12
3 4 5 7
13

8
DMD focusing setup
2 Z-positioning
6
11 of the slot
1
10 die coater
9

Sample transfer Shuttling between

subsystem optical subsystem
and slot die coater

Fig. 2 A simplified schematic of the μ-SLS process showing the different subsystems involved in layer-by-layer fabrication of a true-3D
structure. Optical Setup. 1 – Secondary focusing optics. 2 – Primary beam-shaping optics for the Digital Micromirror Device (DMD) array. 3 – DMD
array housing and controller. 4 – Optical fiber coupling the laser driver output and DMD array. 5 – CW Laser driver; Particle bed formation setup. 6 –
Syringe pump for nanoparticle ink transfer into the slot die coater. 7 – Slot die coater tubing for fluid intake, aspiration, and distribution. 8 – Die head
coupled with a vertical positioner for adjusting the coating gap in between layers. 9 – Heated vacuum chuck with the substrate/wafer; Sample
Transfer Subsystem. 10 – Direct drive linear motor for shuttling the wafer between optical setup and slot die coating setup. 11 – XY-Nanopositioner
for wafer alignment and stepping during large-area fabrication. 12 – A representation of wafer scale fabrication of the 3D parts. 13 – Enlarged view of
the individual patterns formed by the μ-SLS process.

electrohydrodynamic jet printing and direct ink writing solutions to the problem of increasing interconnect delays
cannot produce the types of true 3D structures with due to increasing interconnect lengths created as the I/O
overhangs that will be required for the next generation of density on chip increases. 3D integration can lead to a
3D electronics25–27, and other processes such as electro- reduction in overall interconnect lengths, thus reducing
chemical deposition and Laser Chemical Vapor deposi- time delays without increasing the footprint31. 3D
tion are too slow to be used in high volume production packaging also allows for fabricating layers of dissimilar
environments28,29. Overall, none of the existing micro- materials (metals and dielectrics), integrating different
scale AM processes is capable of generating the types of process technologies (memory, analog, rf) and functions
true 3D metal structures with feature resolutions of less on the chip32. Therefore, the ability of the μ-SLS process
than 5 μm and volumetric throughputs greater than to fabricate complex, 3D interconnect structures could be
50 mm3/hour that are required for the production-scale, a key enabling technology for the future of 3D electronics
microelectronics packaging environment. One process manufacturing. The rest of this paper presents an over-
that holds promise for fabricating the desired metallic view of the μ-SLS system and some preliminary results on
interconnect structures is selective laser sintering (SLS) fabricating microscale, 3D parts using this system.
since laser sintering is compatible with both metals and
polymers and can generate 3D structures. However, the Overview of μ-SLS
resolution of conventional macroscale SLS systems is Figure 2 shows a schematic of the μ-SLS system. The
limited to 70–100 μm due to the use of microscale pow- system consists of three major sub-systems: (1) a particle
ders (typically 5–20 μm) and laser spot sizes used in such bed formation (spreader) setup which is used to generate
systems11,12,30. μ-SLS overcomes these limitations in the the sub-μm particle bed, (2) an optical setup which is
macroscale SLS process and with other microscale AM comprised of laser system and optics designed to achieve
processes by employing nanoscale particles as the parent the sintering resolution and high throughput28, and (3) a
material for sintering with additional focusing optics and sample transfer, alignment, and metrology setup used to
precision motion systems that are used to achieve the shuttle the particle bed between optical and spreader sub-
desired resolution and high throughput. systems with resolution of better than 10 nm. In addition,
Other potential applications of the μ-SLS process are in a vacuum chuck is used to hold the substrate in place and
fabrication of ultra-lightweight structural members rein- ensure that it does not deform during the coating or
forced by lattice structures (See Fig. 1c) and novel inter- writing processes. The system sits on a vibration isolation
connects for enabling three-dimensional (3D) integration stage to reduce outside influences that could damage the
in IC design. 3D integration is one of the most promising part quality.
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 4 of 14

The μ-SLS system also replaces the microscale powders application described in the introduction section for
used in conventional SLS processes with nanoparticle 50 mm wafers. The throughput of the process is currently
(NP) inks in order to achieve the desired feature resolu- about 63 mm3/hour including the coating and sintering
tion. To build layers that are ~1 μm thick, it is necessary processes. However, the modular design of the system
to use particles that are at least one order of magnitude makes it easily adaptable for larger wafers and other
smaller than the desired layer thickness, hence the use of applications if needed, with minor modifications in the
NPs in the system. The use of the NP ink also helps to sub-systems. The following sections of this paper describe
prevent agglomeration of the NPs during the powder each of the key specifics of the μ-SLS process in more
spreading process. The spreader mechanism employs a detail and present some preliminary results for 3D
slot die coating system to spread uniform layers of NP microscale parts fabricated using this system.
inks over the substrate. The wafer holding vacuum chuck
is equipped with heaters to reduce the thermal gradient Process description
across the sample and lower the energy requirements for The three key steps in the μ-SLS process are: (1) NP bed
the process in addition to securing the sample in place. formation to achieve sub-μm layer thicknesses with good
A one degree-of-freedom nanopositioning system with packing density, consistency and repeatability, (2) Sin-
a resolution better than 200 nm is integrated in the slot tering of particles to achieve the desired sub-5 μm reso-
die coating system to precisely control the coating gap lution and high throughput using the optical setup, and
and consequently, the thickness of the NP layer that is (3) High precision sample transfer, alignment motion
spread during the build process. In this system, an elec- systems, and metrology design for precise motion of the
tromagnetic linear actuator is used to move the vacuum stage under the optical sub-system. How each of these
chuck under the slot die coater head and air bearings are process steps has been designed to achieve the desired
used to guide the motion and ensure that a smooth, resolution, throughput, and large area patterning will be
uniform coating is produced. The linear actuator is then discussed in the following sections.
used to move the sample (held on the chuck) to the
optical station and is positioned under the optical system Nanoparticle bed formation
for sintering. In addition to the long travel linear motor/ Achieving sub-5 μm feature resolution requires the use
air bearing system, a flexure based nanopositioning sys- of particles that are at least an order of magnitude smaller
tem is used to precisely step the NP bed repeatedly under than the desired feature resolution which results in the
the optical station to pattern a much larger area in one use of NPs in this system. The higher surface energies of
pass under the optical station with a resolution of better NPs due to large surface area to volume ratio causes the
than 10 nm. NPs to agglomerate and form clusters34–36. Initial
The μ-SLS optical system consists of a continuous wave spreading results using nanoscale Cu powders confirmed
(CW)/quasi-CW laser coupled to an optical fiber and then these phenomena and showed the presence of large
directed off a micromirror array through a de- agglomerates in the bed of these nanoscale powders37.
magnification optics setup. This allows each pixel in the Regardless of the average particle size advertised by the
DMD array to be focused down to a spot size of ≈1 μm33. manufacturer, the bed always contained agglomerates in
Between sintering each layer, the height of the projection the micron size range (see supplementary Fig. S1 for
optics is adjusted to compensate for the change in powder images of different NP samples). NPs also have higher a
bed height created by the spreading of the new NP layer. propensity towards oxidation38–41 which increases the
Once the wafer is patterned entirely, the sample is moved energy required to sinter the NPs and in turn leads to
back to the coating station for next layer of NP coating insufficient sintering and thus, poor part quality. Effects of
and the process is repeated until the 3D part is complete. size and shape of the particles in the bed have also been
The final step in the process is removal of unsintered inks shown to alter the particle packing efficiency in the bed
from the part which is achieved by ultrasonicating the and, thus, affect the layer densities and final part qual-
sample in a solvent similar to the primary solvent of the ity42–44. Based on preliminary testing of powder NPs, the
ink for 30–60 s. The properties of the 3D part can be powder samples were found to have a very wide size
enhanced further by post processing of the sample. For distribution, poor packing, irregular shapes and a high
example, annealing of the part after printing can be used tendency to oxidize which prevent powder-based NPs
to improve the part density and, hence, enhance electrical from being used in the μ-SLS process. Therefore, a new
conductivity of the part, if required. power layer-by-layer NP coating process was devised
Figure 3 shows the actual physical embodiment of the using NP inks to create the multi-layer structures
μ-SLS system with the optical, coating, and positioning required by the μ-SLS process.
subsystems indicated. This embodiment of the μ-SLS is NP inks are NPs suspended in a dispersion typically
designed for the specific microelectronics packaging with some surfactant coating (such as PVP or PEG) to
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 5 of 14

Coating sub-
Optical sub- system
system

Direct drive XY
linear servo nanopositioining
assembly

Interferometer
Air bearing sensor heads
guides
Interferometer

DMD Syringe
engine pump
Z-positioner Pivoting
bracket Vertical
with tip-tilt positioner
stage
Input
manifold Die head
Optical Collimating
fiber lens Heated
Focusing Coating vacuum
Optical lens subsystem chuck
subsystem Laser
source
Focal
plane

Fig. 3 Physical embodiment of the μ-SLS system showing the optical and coating sub-systems along with the linear servomotor and air
bearings used to transport the build substrate between the two sub-systems. An XY nanopositioning system with interferometric feedback
sensors is used to precisely and repeatably position the build substrate under the optical sub-system for high speed patterning of each layer in the
build process.

prevent agglomeration of the NPs into clusters. The Many coating and printing techniques are available that
presence of an organic dispersant reduces the van-der- can print thin layers of these inks such as spin coating,
Waals interaction among the particles and presence of a doctor blading, screen printing, slot-die coating and inkjet
protective surfactant coating reduces the extent of printing. Out of these, the slot die coating process pro-
agglomeration in the sample45. The particles are found to vides very high uniformity of coating thickness, ranging
be uniform in size and shape as opposed to the powders from hundred nanometers to tens of microns46–48, and
(see supplementary Fig. S1). Therefore, NP inks are ideal has the capability to spread multiple layers of NPs without
for use in the μ-SLS system. A detailed review of the disturbing the layers beneath it. It is a pre-metered
effects of morphology and synthesis methods of NPs for coating process where the thickness of the coated layer is
use in micro-AM applications is presented by the authors empirically dependent on process parameters such as the
in ref. 37. For the current generation of μ-SLS, commercial process speed, the die geometry, the ink flow rate, and the
Ag (JSA102A, NovaCentrix, Austin, US) and Cu (CI005- rheological properties of the liquid. These properties
G,Intrinsiq Materials, UK) NP inks have been tested and make slot die coating the optimal process of forming the
are being used in the tool. build layers in μ-SLS.
The coating process for μ-SLS needs to be able to dis- The slot die coating setup consists of a vertical posi-
pense uniform sub-micron layers of low viscosity NP inks tioning mechanism actuated by a stepper motor, a
to achieve near net shape features. A low material wastage cylindrical flexural bearing for in-plane positioning of the
process is always desirable from an economic point of die head, and other support elements of the design. The
view. Another desirable feature in the system is its com- fluid intake, aspiration, and distribution within the syringe
patibility with different substrates including flexible sub- pump is controlled using a three-way rotary valve. The
strates and roll-to-roll flexibility so that it can be slot die coater and the fluid lines are primed and purged
employed in the coming versions of the μ-SLS system. of air bubbles before establishing the initial coating bead
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 6 of 14

with the substrate. After the initial coating bead is spread beam fills the entire DMD array. The DLP 6500 chipset
uniformly across the width of the substrate, the NP inks from TI contains a 1920 × 1080 array of micromirrors
are coated onto the substrate by translating the substrate each of which is 7 μm in size separated by 600 nm dis-
under the slot die coater. The coated layer is partially tance apart. To get the desired 1 μm resolution of the
dried in situ using the heated chuck to remove the excess system and achieve high throughput, the focusing optics is
solvent and facilitate the sintering process. To evaluate designed with the following requirements: (1) Achieve a
the fidelity of the coating process, the surface uniformity magnification of 1/7 and (2) Collect the maximum pos-
of a completely dried film was measured using an optical sible light reflected out of the DMD setup in order to
profilometer (Wyko NT 1100). Using this method, the deliver the most possible power to the NPs. Maximum
average dried film thickness of the coated film when the light collection is necessary for high fluence/irradiance at
substrate moved at a 10 mm/s speed with a constant the sample plane which reduces the exposure duration,
coating gap of 100 μm was measured to be 520 ± 50 nm49. thus enhances the throughput while simultaneously
This corresponds to a ~1.5 micron thick wet layer reducing heat affected zones (HAZs). The optics design
thickness for a manufacturer specified 30% loading con- for the system is presented by the authors in ref. 28 and
centration of the NP ink. After sintering the first layer, the using this setup, an area of 2.3 mm by 1.3 mm can be
coating gap between the substrate and the die lips is patterned at once with a feature resolution of 1.2 μm
increased using a vertical positioner and the process is which is close to the desired resolution. The optics are
repeated to lay down the next layer of NP ink without selected ensuring that the diffraction limited resolution of
distorting the bottom layer. This coating system, there- the focusing optics is smaller than the desired 1 μm pixel
fore, allows sub-micron thick layers of NPs to be repea- resolution.
tably placed onto the substrate without disturbing the NP A set of preliminary sintering experiments were con-
layers below the new coated layer. ducted with different laser types including a femtosecond
(FS), a NS, and a CW laser to compare the quality of
Laser sintering <5 μm features with high throughput sintered regions obtained with different laser types. Based
The μ-SLS system employs an optical system for on those experiments, it was found that both FS and NS
printing the desired patterns onto the substrate with a laser sintered spots showed significant balling up of NPs
feature resolution better than 5 μm and a pixel resolution (with balled up agglomerates of sizes ranging from
close to 1 μm. A conventional macroscale SLS system uses 500 nm to 1 μm—see the areas circled in Fig. 4a) which
a set of reflection mirrors, beam expanders, a 3-axis leads to poor bed density in the final part and hence, poor
dynamic scanner or an fθ lens and a galvanometric mirror properties of the sintered part50,51. In comparison, CW
for patterning the sample. However, as the desired feature laser sintering showed a better sintering quality as the
resolution is roughly 100 times smaller than conventional balling up of NPs effect was negligible and the necking
SLS, the throughput of the process using the conventional between the particles was uniform throughout the spot
optics will be reduced by a factor of 10,000. This makes area (see Fig. 4). The average power requirements using
the process unfeasible in an industrial setting for large CW laser are typically much higher than average power
scale microelectronics packaging. To increase the requirements for short-pulsed (NS and FS) lasers due to
throughput of the process, a digital micromirror device the enhanced heat transfer away from the laser spot for
(DMD) array from Texas Instruments (TI) is used in the the longer exposure, CW laser. This lower average power
μ-SLS system. A DMD is a micro-opto-electromechanical during sintering with the short-pulsed lasers results in a
system used by TI and other companies for their projec- processing window that is very narrow in terms of laser
tion systems. It is an array more than 2 million micro- power that can be used for getting good sintering as
scale mirrors that can be independently actuated/modu- compared to the CW laser. As a result of this narrow
lated between pre-set orientations. Using a DMD mirror processing window and local spatial variations in the NP
for pattern generation, instead of sintering just one 1 μm bed/laser beam intensity, localized melt pools can form
spot at once, allows millions of 1 μm spots to be sintered during the sintering process with the NS and FS lasers
at once thus greatly enhancing the areal throughput of the which results in formation of balls in the powder bed due
system. to the hydrodynamic effects (minimize the surface energy)
A CW, a QCW or a nanosecond (NS) laser of appro- of the liquid metal on the substrate. The size of these balls
priate power (typically 50 to 300 watts depending on the typically ranges from a few 100 nm up to 1 μm depending
material being patterned and the layer thickness) is fiber- upon the average size of NPs that are melted. Figure 4a, b
coupled to the micromirror array (DLP 6500) through a shows a comparison of the sintering quality of spots sin-
set of coupling optics that includes a fiber coupler, a beam tered with laser powers tuned for good sintering quality
expander, a fly’s eye lens, mirrors and lenses to resize, using NS laser and CW laser. A 532 nm green NS laser
reshape and direct the input beam such that the laser with a pulse duration of 25 ns and a 532 nm CW laser
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 7 of 14

a
(i) NS laser (ii) CW laser

5 μm 5 μm

b
(i) 99.5 (ii) 198.9 (iii) 298.4
kW/cm2 kW/cm2 kW/cm2

Al 10 μm 10 μm 10 μm

(iv) 4.9 (v) 9.7 (vi) 22.7

kW/cm2 kW/cm2 kW/cm2

Glass 30 μm 30 μm 30 μm

c d 25 e 600
400
Melting/damage 500 Onset of sintering
350 20 Melting/damage
Onset of melting
Irradiance (kW/cm )
Irradiance (kW/cm )

Irradiance (W/cm )
2
2

300
2

400
250 Good sintering 15
300
200
10 200
150
Good sintering
100 200
Weak sintering 5
50 Weak sintering
No sintering 0
No sintering
0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 50 100 150 200 250
Exposure time (ms) Exposure time (ms) Bed temperature (°C)

Fig. 4 Comprehensive study of laser-material interaction for Cu nanoparticles. a Comparison of sintering quality using NS and CW lasers- (i) NS
laser showing significant balling up effect (see red circled spots) with sporadic necking & crack developments (ii) CW laser showing no balling,
uniform necking between particles throughout the spot area. b Variation in sintering quality of Cu NPs from no-sintering to weak sintering, good-
sintering & melting as the irradiance of CW laser is increased for (i–iii) A 0.4 μm thick Cu layer on Al substrate for a 500 ms exposure duration (iv-vi) A
0.4 μm thick Cu layer on Glass substrate for 500 ms exposure. c A graph showing the laser irradiance vs exposure duration processing window for a
0.4um thick Cu NP layer on Al substrate. d Processing window for a 0.4 μm thick Cu NP layer on glass -substrate; experiments for Fig. 4a–d are
conducted with particle bed at room temperature. e Reduction in sintering and melting onset threshold when using a heated bed at different initial
temperatures.

were used for the sintering of these spots. Figure 4c–e regarding the experiment design, procedure and analysis
provides a summary of the exhaustive laser sintering study of these results, readers are encouraged to review the
carried out by the authors to obtain the processing win- following article by the authors51.
dow for good sintering at different laser powers, exposure A shortcoming of laser-based processes is the develop-
durations and bed temperatures. For further details ment of large thermal gradients leading to residual
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 8 of 14

stresses, deformations and distortions in the part. Powder two trucks which hold the nanopositioning stage. The air
beds in conventional laser sintering are typically heated to bearing stage is actuated using a direct drive linear servo
a high temperature before exposure to the laser which motor which is also non-contact and requires no
helps to minimize the thermal gradients between the un- mechanical linkages/mechanisms for motion transmis-
sintered and sintered sections of the bed. This helps to sion, thus minimizing potential sources of friction and
reduce the development of high residual stresses and non-linearities in the system. For position sensing of the
hence, improves part quality. In addition to the center stage along the direction of motion, an inter-
improvement in final part’s properties and lower distor- ferometer (model: FPS3010, from Attocube Systems AG)
tions, heating the particle bed also lowers the energy with a resolution of 1 pm and a working range of 2 m
requirements for the sintering process. This has been is used.
confirmed by sintering experiments conducted by the
authors using a heated stage for wafer handling in ref. 50.
The laser energy requirements were found to decrease by High precision alignment using XY nanopositioner
up to 17 times when the bed was heated to 200 °C com- A nanopositioning stage is designed to align the sample
pared to the energy requirements when the bed is at room accurately and precisely under the optical system and
temperature50 (see Fig. 4e). Thus, a wafer handling chuck move the sample in a step-and-repeat fashion to sinter
with heating capabilities is used which reduces the wat- multiple patterns in one go under the optical system.
tage requirement of the laser and leads to better quality Once the sample reaches the optical sub-system, the next
parts with smaller geometrical aberrations and less war- step is to align the sample under the optical sub-system
page (see supplementary Fig. S2 for the design of this accurately with good repeatability so that the overlay
chuck)52. errors are minimized. A motion resolution <100 nm
ensures that the overlay errors even after 10 layers stays
Sample transfer and alignment within the desired feature resolution on the part. Sec-
There are two sub-systems included in the sample ondly, the optical sub-system design patterns an area of
transfer and alignment process: (1) a long-range linear 2.3 mm by 1.3 mm in one shot whereas, the size of each
motion system used to shuttle the sample between the wafer is 50 mm by 50 mm. Thus, the nanopositioner
slot die coating and optical system and (2) an XY nano- needs to have a travel range of 50 mm in both the x and y
positioning stage used to align the sample under the directions so that the entire wafer can be patterned in one
optical system for minimizing registration errors and pass under the optical sub-system. The wafers are pat-
move the sample to sinter the whole 50 mm by 50 mm terned using step and repeat approach where the whole
sample in one pass under the optical sub-system. wafer will be composed of multiple exposures (each
exposure sized 2.3 mm × 1.3 mm). Thirdly, to maximize
Long-range linear motion between spreader and sintering the throughput of the system, the idle times while sin-
stations tering should be minimized. Using a 50 ms exposure
To achieve 1 μm feature resolution, the registration duration43 under the optical sub-system for sintering and
errors between two consecutive layers of NPs should be another 50 ms for moving between steps and achieving
minimum, preferably <100 nm so that even after 10 layers, steady state, the motion system is desired to operate at
the errors are within 1 μm pixel resolution specification. speeds of 10 Hz which requires that any uncontrollable
The motion between subsystems should also be highly resonances of the motion system should be at greater than
repeatable which requires that the non-linear effects 30 Hz so as to not affect the motion quality of the stage at
commonly associated with contact-based bearings such as 10 Hz stepping. And lastly, the motion system is desired
stick slip phenomenon and friction effects should be to have a repeatability of better than 50 nm. To achieve
minimized. Air bearings are advantageous compared to this repeatability, effects such as friction and backlash in
rolling element bearings due to frictionless operation the system need to be minimized. Although, a number of
(minimal air shear friction at high speeds), minimum off the shelf nanopositioning stages are available that can
wear, high speed capability, and no requirement of achieve a resolution of less than 5 nm, most of them are
lubrication in the system which also means lower impact piezo-actuated stages and hence, are limited in their range
on the environment. Since the friction in air bearings is to a few hundred microns. The traditional XY stages
dependent on air shear from motion, the friction when based on rolling element bearings can achieve the 50 mm
starting (at zero velocity) is zero and hence, there is no travel but their motion quality is poor due to friction and
stiction in the system allowing it to have a high resolution stiction in the system. Air bearings and magnetic bearings
and repeatability. In this μ-SLS system, we are using an air driven by direct drive linear servos can achieve both the
bearing system from NewWay Air bearings. The system resolution and range requirements, but they are not cost
consists of two, 1 m long, parallel rails as guides for the effective. Thus, a flexure bearing stage driven by voice coil
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 9 of 14

actuators are used as flexure bearings provide a low-cost, partially dried, it is shuttled to the optical station on
high motion quality, and easy to design solution. the air bearing guides using the feedback from the
The stage design makes use of a modified double par- interferometers. The sample is positioned to the XY
allelogram flexure mechanism (DPFM) as its building location for sintering (from step 1). The XY fine
block and these modified DPFMs53 are connected in alignment is completed using the XY
parallel to achieve the desired stiffness and travel range nanopositioner.
for each axis. The stage uses a modular two-level stage 4. The 3D CAD of the final part is sliced into cross-
design using 12 modified DPFM units with 6 units guiding sectional layers and this layer profile is fed into the
the motion along each axis. The modified DPFM units are DMD setup for turning the mirrors ‘on’ in the DMD
waterjet cut out from ½” thick Al 7075 plates and array selectively. The thermoelectric heating is kept
assembled. Details on the design of the modified DPFM on to achieve a higher bed temperature and the laser
stage and its assembly can be found in other works by the exposure duration for sintering is modulated based
authors54–56. For actuation, two sets of voice coil actua- on the layer thickness to achieve good sintering
tors are used to drive the bearings and a set of inter- quality. Each exposure sinters a 2.3 mm × 1.3 mm
ferometers are used for position sensing and feedback area respectively. To sinter the whole 50 mm by
control of the center stage. The closed loop positioning 50 mm wafer, the sample is stepped and aligned
resolution is of this stage is found to be 7.6 nm at a using the XY nanopositioner and roughly 800
13.7 Hz bandwidth. The details of the controller design exposures are needed to sinter the entire wafer.
and its tracking performance are given in ref. 56. Overall, 5. This whole procedure (steps 2–4) is repeated for
the stage is shown to achieve three very important spe- each additional layer in the 3D structure until the
cifications: (1) it has been shown to have a sub 10-nm desired part geometry is complete. Once the part is
resolution, (2) it has a tracking bandwidth of greater than complete, the sample is brought to the washing
10 Hz, and (3) it has a travel range of over 50 mm in both station for removal of the un-sintered NPs from the
the x and y axis. Therefore, when this stage is combined part. This washing off procedure involves
with the long-range linear motion system described in the ultrasonicating the part in the parent ink solvent
previous subsection, it is able to meet all of the motion such that the washed off ink can be recycled and
requirements of the μ-SLS system. reused in the process after some chemical
processing.
μ- SLS process flow 6. Finally, the part is annealed in an oven for a fixed
Figure 5 summarizes the process steps for the μ-SLS time duration to enhance its electrical and
process. The steps are listed as follows: mechanical properties.
1. The first step in the process is the alignment routine There is a continuous need to economically evaluate
to determine and set the XY coordinates for optical and predict the quality of the products made using this
station and coating station. The samples have microscale AM process. Understanding the physics
fiducial markers on the substrate and the alignment behind laser-material interaction and NP sintering is
is done using an optical metrology system and the critical to the fabrication of defect-free parts. Multiscale
corresponding XY positions for coating and computational models for NP bed formation and con-
sintering locations (using interferometer) are stored solidation are being developed, which will help in opti-
in the memory. Any long-range motion of the mizing the process parameters required for creating
sample is achieved through the linear motor-air complex microparts57,58. Additionally, these models will
bearing combination and XY nanopositioner is used be calibrated against experimental results to improve their
for short range fine positioning. reliability in predicting sintered part.
2. Next, for spreading the NP layer, the slot die coater
goes through its homing and priming routines to Sintering results
establish a uniform bead thickness for the coating. With the design and assembly of the system complete,
The sample is brought to the pre-determined XY some preliminary testing was conducted to verify the
location for coating (from step 1) and the z-gap resolution of the system. A 50 W CW laser with a central
between the sample and die lips is measured using wavelength of 808 nm was used as the source. Shown
the Keyence gap sensor. The sensor serves as a below are some examples of the sintered patterns (un-
feedback for the z-motion control of slot die head to sintered portion washed off) using the μ-SLS system.
achieve the desired coating thickness. Figure 6a–c shows the optical microscope and surface
3. Once the first NP layer is spread, the sample is profile images of different desired diameter circles that
heated using thermoelectric heaters and a heat lamp demonstrate the capability of the system to produce 3D
source to partially dry the wet layer. As the layer is pillar structures for IC packaging with diameters down to
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 10 of 14

Global travel, alignment

Nanoparticle bed formation and metrology process Laser sintering process

Start Translate substrate Send sliced layer

Alignment routine Homing routines Coat NP ink layer data to DMD
process to optical station
using fiducial engine
markers under the
optical station
Prime slot die Partially dry the ink
coater to remove excess Adjust exposure
solvent duration based on
Inline substrate layer thickness
Position substrate metrology and
under slot die Measure layer alignment
coater thickness in-situ Turn on layer

Update coating
height

Step and repeat action

Fabricated
Ultrasonicate in the
Anneal fabricated
primary solvent of
Post-processing steps part to enhance End process
the ink to remove
part quality
un-sintered NP ink

Fig. 5 μ-SLS process flow-chart. The process starts with a homing routine for the initial optical alignment of the sample under the optical station
using fiducial markers. The slot-die coater and the plumbing lines are primed and the sample is then translated to the slot die coating system where
the nanoparticle ink bed is laid down. The ink is partially dried using an infrared heater and the layer height is measured in situ. The sample is then
moved under the optical system where the initial position alignment is done. After the position is fixed, the sliced layer pattern of the 3D CAD model
is sent to the DMD controller and the pattern is projected on the coated layer during the laser sintering process. The XY-Nanopositioner is stepped
and the pattern is scanned over the sample area. The sample is then translated back to the slot die coating station where the next layer is laid down,
and after which the sintering process is repeated. To maintain layer overlay control, the XY-nanopositioner is used to align the layers. After n layers
are sintered, the excess NP ink is removed using a sonicator and the fabricated part is annealed to ensure good part quality.

10 μm (supplementary Fig. S3) using the current laser method. For these measurements, 80 μm wide lines with a
source over large areas. Figure 6d shows UT Austin tra- thickness of about 80 nm are sintered and the V–I char-
demarked longhorn logo, where the minimum feature size acteristics are obtained for different lengths of sintered
is around 7 μm. The sintering resolution of the system has lines (see Fig. 6). Assuming a linear relationship between
been demonstrated to be close to 3 μm with the current the resistance (R) and length (L) of the conductor, the
setup (see supplementary Fig. S3). The overall resolution gradient of the R vs. L curve is the ratio of bulk resistivity
of these parts is affected as the higher exposure durations and area of cross section of the conductor. V–I curves for
(>3 s) leads to the formation of HAZs around the feature different lengths of the conductor are shown in Fig. 6a
boundaries. This effect may be attenuated by using a and the resistance vs. length plot is shown in Fig. 7b.
higher power laser source or even using a Quasi-CW laser These results were collected from three sets of measure-
source such that the exposure durations are much shorter ments with multiple samples at same lengths of the
thus reducing the heat propagation and minimizing conductor. Using the slope of the fitting, the resistivity of
HAZs. It is expected that reducing the HAZs should the sintered samples is measured to be 73 ± 10 nΩ m. This
further improve the resolution of these structures bring- is about 4.5 times the bulk resistivity of silver (16 nΩ m @
ing it closer to the optical resolution of the setup. How- 20 °C) which is similar to or even slightly better than the
ever, these feature dimensions are already an conductivity of sintered silver reported in literature59–61.
improvement over the current electroplating technology Good electrical conductivity of the sintered lines is critical
used for fabrication of pillar bumps used in semi- for application of the process in semiconductor packaging
conductor packaging. applications.
Measuring the electrical resistivity of the sintered
samples provides a convenient and accurate metric for Multi-layer sintering results
assessing the quality of sintering using the μ-SLS system. To test the ability of the system for sintering multilayer
The resistivity was measured using a four-point probe structures with good overlay, several multilayer structures
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 11 of 14

μm
a b 1.50
1.40
1.30
1.20
1.10
1.00
0.90
0.80
0.70

481 μm
0.60
0.50
50 μm
0.40
641 μm 0.30
0.20
0.10
c 0.00

80 µm
100 μm

Fig. 6 Single layer sintering results with Ag nanoparticle ink. a Optical microscope image of an array of 40 μm diameter circles with 80 μm pitch.
b Surface topography of an array of 40 μm diameter circles with 80 μm pitch. The RMS height of these circles is around 617 nm with a standard
deviation of 119 nm. c An optical microscope image showing the full areal scale (2.3 mm × 1.2 mm) array sintered on a 1 μm thick Ag NP ink layer.
(inset) Enlarged image of an array of 20 μm diameter circles with 40 μm pitch. d Optical microscope image of longhorn logo (trademark of The
University of Texas at Austin). The smallest feature size in this sintered part is around 7 μm near the tip of the longhorn ears.

a b
500 16

14
400
12
Voltage, V (mV)

Resistance (Ω)

300
10

200 Length = 1.12 mm 8

Length = 0.83 mm
Length = 0.59 mm
100 6
Length = 0.28 mm
Length = 0.144 mm
4
0 R = (11.38 ± 1.54)l + (1.58 ± 1.07)
2
0 20 40 60 80 100 120 140 0.0 0.2 0.4 0.6 0.8 1.0 1.2
Current, I (mA) Length (mm)

Fig. 7 a V–I characteristic of a sintered and annealed @ 150 °C for 30 min Ag pattern for different lengths of the conductor. b Resistance, R vs Length,
L curve for sintered and post-annealed Ag line.

were fabricated. Figure 8a shows the progression of the the circle from the reference on the substrate is 3.98 μm
feature heights as higher layers are coated. A 1.8 mm × for bilayer sintering and 8.02 μm for three-layer sintering.
1.2 mm rectangular feature is sintered on the first coated The coating gap between the die lips and the current layer
layer to distinguish it from the subsequent layers. The is kept the same at 100 μm for the experiment. However, a
average height of the feature is 2.24 μm. Afterwards, the sharp increase in the height of the part as shown in Fig. 8b
next layers are coated and partially dried and a 500 μm can be attributed to the flow of excess ink into the sin-
circle is sintered. The average height of the top plane of tering region, thereby affecting the overall uniformity of
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 12 of 14

a Layer 1 Layer 2 Layer 3

1.
2
m Not to scale
m
1.8 mm

b
7.00

6.00

2.4 mm 5.00

4.00
1.8 mm

3.00

2.00

1.00

0.00
μm

c Layer 2 d
feature 7.00
m
m 0.05 m 6.00
0.05 m Layer 1
feature 5.00
4.00
3.00
1.8
1. 2.00
2
m 1.00
m
1.8 mm
0.00
Not to scale 2.4 mm μm

Fig. 8 Multilayer sintering of Ag nanoparticle inks. a Visualization of the layer-by-layer (LbL) fabrication. The first layer is a 1.8 mm by 1.2 mm
rectangular structure. A 500 μm circle is sintered on top of the first layer as part of the second and third layer. b Optical topographic images showing
the LbL progression of the µ-SLS process. The RMS surface roughness are 2.24 μm, 1.19 μm and 4.08 μm Layers 1, 2 and 3 respectively. c Visualization
of the two layer part with the first layer as a rectangular base and the second layer as a linear array. d Surface topography of the sintered and post-
processed two-layer sample. The RMS height of the rectangular pillars measured from the top of layer 1 is 5.77 μm with a standard deviation of
1.75 μm. The averaged variability in pillar heights across the length of the array is 1.2 μm. These variations can be attributed to the non-uniformity in
the layer deposition process, evolution of thermal gradients due to long exposure durations and the build-up of excess ink along the edges of the
features during the sintering process.

the part. Figure 8d shows a two-layer sintered part with Conclusion

the first layer being a rectangular base and the second This paper presents a new micro-AM process called
layer is a linear array with 50 μm line width and 50 μm micro-scale selective laser sintering (μ-SLS) that can be
spacing. used to fabricate three-dimensional parts with feature
Overall, these results demonstrate the ability of the μ- resolutions of better than 5 μm with a high throughput
SLS system to preform, multilayer 3D part fabrications. that can be employed in a production environment. As
However, better process tuning for different sub-systems the different sub-systems of the technology mature, it is
such as optimizing the slot die coating parameters for expected that the technology could eventually be
better layer thickness uniformity and repeatability, employed in fabrication of complete microelectronic
improving the optics for better alignment and smaller packages through chip-to-chip and chip-to-substrate
spherical aberrations, developing DMD pattern correction interconnects, thus replacing the conventional building
schemes for better feature resolutions, and optimizing the blocks of packaging including inter-layer vias, controlled
sintering irradiance vs exposure duration for different collapse chip connection (C4) bump pads, ball grid array
layer thicknesses are some of the short-term future work (BGA) pads, and flip chip metallic pillar bumps thus
that need to be completed in order to improve the quality, providing a complete solution for the back end-of-line
uniformity, and resolution of the parts produced. assembly for electronics packaging. Additionally, the
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 13 of 14

modular nature of the process allows for modifications in 11. Vaezi, M., Seitz, H. & Yang, S. A review on 3D micro-additive manufacturing
individual sub-systems for use in roll-to-roll applications technologies. Int. J. Adv. Manuf. Technol. 67, 1721–1754 (2013).
12. Hirt, L., Reiser, A., Spolenak, R. & Zambelli, T. Additive Manufacturing of Metal
for flexible electronics or in other micro-AM applications Structures at the Micrometer Scale. 1604211, (2017).
of metals such as the fabrication of metamaterials with 13. Cohen, A. Microscale metal additive manufacturing of multi-component
desirable optical and mechanical properties, MEMS, medical devices. 0–18. https://doi.org/10.1108/13552541011034889 (2013).
14. Rahman, T., Rahimi, A., Gupta, S. & Panat, R. Sensors and Actuators A: Physical
plasmonics, hierarchical materials, and microfluidics62–68. Microscale additive manufacturing and modeling of interdigitated capacitive
Overall, this paper demonstrates that μ-SLS has the long- touch sensors. Sens. Actuators A. Phys. 248, 94–103 (2016).
term potential help overcome many of the current lim- 15. Hoerber, J. et al. Approaches for additive manufacturing of 3D electronic
applications. Procedia CIRP 17, 806–811 (2014).
itations in microscale AM of 3D metal parts. 16. Niesler, F. & Hermatschweiler, M. Additive Manufacturing of Micro‐sized Parts:
Trend of miniaturization is driver for 3D laser lithography. Laser Technik Journal
Acknowledgements 11, 16–18 (2014).
The authors would also like to thank NXP semiconductors and National 17. Teh, K. S. Additive direct-write microfabrication for MEMS: a review. Front.
Science Foundation for sponsoring this project. This material is based upon Mech. Eng. 12, 490–509 (2017).
work supported by the National Science Foundation under Grant No. 1728313. 18. Moore, G. E., Life Fellow. Cramming more components onto integrated cir-
cuits. Proc. IEEE 86, 82–85 (1998).
Authors' contributions 19. Mack, C. A. Fifty Years of Moore’s Law. IEEE Trans. Semicond. Manuf. 24,
The authors collectively worked on the overall design of the system. N.R. 202–207 (2011).
designed the optical setup, performed the sintering experiments, designed the 20. Lee, M. J. M. J., Lim, C. C. & Teh, P. T. Packaging technology and design
XY stage and air bearing system, carried out experiments for the XY stage and challenges for fine pitch Cu pillar and BOT (Bond on Trace) using thermal
wrote the main manuscript text. D.B. worked on the system assembly and compression bonding. Proc. IEEE/CPMT Int. Electron. Manuf. Technol. Symp.
integration, multi-layer sintering experiments, edited the manuscript and along 2015, 1–9 (2015).
with O.D. helped with SEM imaging and profilometer measurements. All 21. Hillman, C. The benefits and risks of copper pillar bumped flip chips. U.S.
authors reviewed the manuscript. Tech, The global electronics publication. (2017). http://www.us-tech.com/
RelId/1753046/ISvars/default/The_Benefits_and_Risks_of_Copper_Pillar_
Data availability Bumped_Flip_Chips.htm.
The data (profilometer and SEM images) generated in this study are included 22. Roelfs, B., Dambrowsky, N., Erben, C. & Kenny, S. Filling through holes and blind
in the manuscript. Any other data or details can be made available from the microvias with copper using reverse pulse plating and insoluble anodes.
corresponding author on reasonable request. Circuit World 38, 113–123 (2012).
23. Nanoscribe. Photonic Professional GT-The next generation of 3D laser lithography
Conflict of interest High speed micro-and nanoprinting [Brochure]. (2015). http://www.nanoscribe-
The authors declare that they have no conflict of interest. japan.com/wp-content/uploads/2015/01/Flyer_PPGT_web.pdf.
24. Kim, S.-I. et al. Nanopatterning by laser interference lithography: applications to
optical devices. J. Nanosci. Nanotechnol. 14, 1521–1532 (2014).
Supplementary information accompanies this paper at https://doi.org/
25. Lewis, J. A. Direct ink writing of 3D functional materials. Adv. Funct. Mater. 16,
10.1038/s41378-019-0116-8.
2193–2204 (2006).
26. Lewis, J. A., Smay, J. E., Stuecker, J. & Cesarano, J. Direct ink writing of
Received: 2 April 2019 Revised: 25 July 2019 Accepted: 13 September 2019 three-dimensional ceramic structures. J. Am. Ceram. Soc. 89, 3599–3609
(2006).
27. Park, J.-U. et al. High-resolution electrohydrodynamic jet printing. Nat. Mater. 6,
782–789 (2007).
References 28. Sundaram, M. M., Kamaraj, A. B. & Kumar, V. S. Mask-Less Electrochemical
1. Bourell, D. L. Perspectives on additive manufacturing. Annu. Rev. Mater. Res. 46, Additive Manufacturing: A Feasibility Study. J. Manuf. Sci. Eng. 137, 021006
1–18 (2016). (2015).
2. Zadpoor, A. A. & Malda, J. Additive manufacturing of biomaterials, tissues, and 29. Allen, S. D. Laser chemical vapor deposition: a technique for selective area
organs. Ann. Biomed. Eng. 45, 1–11 (2017). deposition. J. Appl. Phys. 52, 6501–6505 (1981).
3. Liu, R., Wang, Z., Sparks, T., Liou, F. & Newkirk, J. Aerospace applications of laser 30. Vyatskikh, A. et al. Additive manufacturing of 3D nano-architected metals. Nat.
additive manufacturing. Laser Additive Manufacturing: Materials, Design, Tech- Commun. 9, 593 (2018).
nologies, and Applications (Elsevier Ltd, 2016). 31. Koester, S. J. et al. Wafer-level 3D integration technology. IBM J. Res. Dev. 52,
4. Cho, D.-W. et al. Related content Organ Printing: dispensing-based 3D printing 583–597 (2008).
additive manufacturing of polymer melts for implantable medical devices and 32. Knickerbocker, J. U. et al. 3D silicon integration. In Electronic Components and
scaffolds. https://doi.org/10.1088/1758-5090/aa7279 (2017). Technology Conference, 2008. ECTC 2008. 58th 538–543 (2008).
5. Giffi, C. A., Gangula, B & Illinda, P. 3D opportunity in the automotive industry. 33. Roy, N. K., Behera, D., Dibua, O. G., Foong, C. S. & Cullinan, M. A. Single shot,
Deloitte Univ. Press 3D opportunity in the automotive industry. (2014). large area metal sintering with micrometer level resolution. Opt. Express 26,
6. Delgado Camacho, D. et al. Applications of additive manufacturing in the 25534 (2018).
construction industry – A forward-looking review. Autom. Constr. 89, 110–119 34. Nanda, K. K., Maisels, A., Kruis, F. E., Fissan, H. & Stappert, S. Higher Surface
(2018). Energy of Free Nanoparticles. 1–4. https://doi.org/10.1103/PhysRevLett.
7. Gao, W. et al. The status, challenges, and future of additive manufacturing in 91.106102 (2003).
engineering. CAD Comput. Aided Des. 69, 65–89 (2015). 35. Dave, R., Gupta, R., Pfeffer, R., Sundaresan, S. & Tomassone, M. S. Deagglo-
8. Ruiz-Morales, J. C. et al. Three dimensional printing of components and meration and mixing of nanoparticles. NSF Science and Engineering Grantees
functional devices for energy and environmental applications. Energy Environ. Conference (2006).
Sci. 10, 846–859 (2017). 36. Jianfeng, Y., Guisheng, Z., Anming, H. & Zhou, Y. N. Preparation of PVP coated
9. Wohlers Associates Publishes 23rd Edition of Its 3D Printing and Additive Cu NPs and the application for low-temperature bonding. J. Mater. Chem. 21,
Manufacturing Industry Report | Wohlers Associates. https://wohlersassociates. 15981–15986 (2011).
com/press74.html (2018). 37. Roy, N. K., Foong, C. S. & Cullinan, M. A. Effect of size, morphology, and
10. Wohlers Report 2018: 3D Printer Industry Tops $7 Billion. https://www.forbes. synthesis method on the thermal and sintering properties of copper nano-
com/sites/tjmccue/2018/06/04/wohlers-report-2018-3d-printer-industry-rises- particles for use in microscale additive manufacturing processes. Addit. Manuf.
21-percent-to-over-7-billion/#5c639f592d1a (2018). 21, 17–29 (2018).
Roy et al. Microsystems & Nanoengineering (2019)5:64 Page 14 of 14

38. Jardón-Maximino, N. et al. Oxidation of copper nanoparticles protected with 54. Roy, N. & Cullinan, M. Design of a flexure based xy precision nanopositioner
different coatings and stored under ambient conditions. J. Nanomater. 2018, with a two inch travel range for micro-scale selective laser sintering. In Pro-
1–8 (2018). ceedings of the Annual Meeting of the American Society for Precision Engineering
39. Schaper, A. K., Hou, H., Greiner, A., Schneider, R. & Phillipp, F. Copper nano- 31st Annual Meeting 395–400 (American Society for Precision Engineering,
particles encapsulated in multi-shell carbon cages. Appl. Phys. A Mater. Sci. ASPE, 2016).
Process 78, 73–77 (2004). 55. Roy, N. K. & Cullinan, M. A. Design and characterization of a two-axis, flexure-
40. Fedlheim, D. L. & Foss, C. A. Metal Nanoparticles: Synthesis, Characterization, and based nanopositioning stage with 50mm travel and reduced higher order
Applications. (CRC press, 2001). modes. Precis. Eng. 53, 236–247 (2018).
41. Athanassiou, E. K., Grass, R. N. & Stark, W. J. Large-scale production of carbon- 56. Roy, N. K. & Cullinan, M. A. Fast trajectory tracking of a flexure-based, multiaxis
coated copper nanoparticles for sensor applications. Nanotechnology 17, nanopositioner with 50-mm travel. IEEE/ASME Trans. Mechatronics 23,
1668–1673 (2006). 2805–2813 (2018).
42. Krantz, M., Zhang, H. & Zhu, J. Characterization of powder flow: Static and 57. Yuksel, A. & Cullinan, M. Modeling of nanoparticle agglomeration and powder
dynamic testing. Powder Technol. 194, 239–245 (2009). bed formation in microscale selective laser sintering systems. Addit. Manuf. 12,
43. Tolochko, N. K. et al. Balling processes during selective laser treatment of 204–215 (2016).
powders. Rapid Prototyp. J. 10, 78–87 (2004). 58. Dibua, O. G., Yuksel, A., Roy, N. K., Foong, C. S. & Cullinan, M. Nanoparticle
44. Strondl, A., Lyckfeldt, O., Brodin, H. & Ackelid, U. Characterization and control of sintering model: simulation and calibration against experimental data. J. Micro-
powder properties for additive manufacturing. JOM 67, 549–554 (2015). and Nano-manufacturing 6, 041004 (2019).
45. Balliu, E. et al. Selective laser sintering of inkjet-printed silver nanoparticle inks 59. An, K. et al. Selective sintering of metal nanoparticle ink for maskless
on paper substrates to achieve highly conductive patterns. Sci. Rep. 8, 2–10 fabrication of an electrode micropattern using a spatially modulated laser
(2018). beam by a digital micromirror device. ACS Appl. Mater. Interfaces 6,
46. Schabel, W. et al. Formation of blade and slot die coated small molecule 2786–2790 (2014).
multilayers for OLED applications studied theoretically and by XPS depth 60. Theodorakos, I., Zacharatos, F., Geremia, R., Karnakis, D. & Zergioti, I. Selective
profiling. AIP Adv. 6, 065108 (2016). laser sintering of Ag nanoparticles ink for applications in flexible electronics.
47. Park, J., Shin, K. & Lee, C. Optimized design for anti-reflection coating process Appl. Surf. Sci. 336, 157–162 (2015).
in roll-to-roll slot-die coating system. Robot. Comput. Integr. Manuf. 30, 61. Lee, H. et al. Large-area compatible laser sintering schemes with a spatially
432–441 (2014). extended focused beam. Micromachines 8, 1–8 (2017).
48. Lin, Y., Liu, T. & Hwang, S. Minimum wet thickness for double-layer slide-slot 62. Skylar-Scott, M. A., Gunasekaran, S. & Lewis, J. A. Laser-assisted direct ink writing
coating of Poly (Vinyl-Alcohol) solutions. https://doi.org/10.1002/pen.20454 of planar and 3D metal architectures. Proc. Natl Acad. Sci. USA 113, 6137–6142
(2005). (2016).
49. Behera, D., Roy, N. K., Foong, C. S. & Cullinan, M. Nanoparticle bed deposition 63. Galliker, P. et al. Direct printing of nanostructures by electrostatic auto-
by slot die coating for microscale selective laser sintering applications. In focussing of ink nanodroplets. Nat. Commun. 3, 890–899 (2012).
Proceedings of the 29th Annual International Solid Freeform Fabrication Sym- 64. Kaschke, J. & Wegener, M. Optical and Infrared Helical Metamaterials. Nano-
posium - An Additive Manufacturing Conference 2382–2393 (2018). photonics 5, 510–523 (2016).
50. Roy, N. K., Dibua, O. G. & Cullinan, M. A. Effect of bed temperature on the laser 65. Nagpal, P., Lindquist, N. C., Oh, S. & Norris, D. J. Ultrasmooth Patterned Metals
energy required to sinter copper nanoparticles. Jom 70, 401–406 (2018). for Plasmonics and Metamaterials. Science 325, 594–598 (2009).
51. Roy, N. K. et al. A comprehensive study of the sintering of copper nano- 66. Lifton, V. A., Lifton, G. & Simon, S. Options for additive rapid prototyping
particles using femtosecond, nanosecond, and continuous wave lasers. J. methods (3D printing) in MEMS technology. Rapid Prototyp. J. 20, 403–412
Micro Nano-Manuf. 6, 10903 (2018). (2014).
52. Roy, N., Foong, C. & Cullinan, M. Design of a micro-scale selective laser sin- 67. Lee, W. C., Heo, Y. J. & Takeuchi, S. Wall-less liquid pathways formed with three-
tering system. In Proceedings of Solid Freeform Fabrication (SFF) Symposium, dimensional microring arrays. Appl. Phys. Lett. 101, 1–5 (2012).
Austin, TX. https://doi.org/10.1115/MSEC2016-8569 (2016). 68. Bauer, J., Schroer, A., Schwaiger, R. & Kraft, O. Approaching theore-
53. Panas, R. M. & Hopkins, J. B. Eliminating underconstraint in double parallelo- tical strength in glassy carbon nanolattices. Nat. Mater. 15, 438–443
gram flexure mechanisms. J. Mech. Des. 137, 092301 (2015). (2016).