Architected Flexible Syntactic Foams:

Additive Manufacturing and Reinforced

Particle driven Matrix Segregation
Hridyesh Tewani
Dept. of Civil & Env. Engineering University of Wisconsin-Madison Madison, WI
53706 &Megan Hinaus Dept. of Civil & Env. Engineering University of Wisconsin-
Madison Madison, WI 53706 &Mayukh Talukdar Dept. of Civil & Env.
Engineering University of Wisconsin-Madison Madison, WI 53706 &Hiroki Sone Dept. of
Civil & Env. Engineering University of Wisconsin-Madison Madison, WI 53706 &Pavana
Prabhakar Dept. of Mechanical Engineering Dept. of Civil & Env. Engineering University of
Wisconsin-Madison Madison, WI 53706 [email protected]
Abstract

Polymer syntactic foams are transforming materials that will shape the future of next-
generation aerospace and marine structures. When manufactured using traditional processes,
like compression molding, syntactic foams consist of a solid continuous polymer matrix
reinforced with stiff hollow particles. However, polymer matrix segregation can be achieved
during the selective laser sintering process with thermoplastic polyurethane (TPU). It is
uncertain what role hollow particles play in forming this matrix segregation and its impact on
the corresponding mechanical properties of syntactic foams. We show that the size of the
hollow particles controls the internal microscale morphology of matrix segregation, leading
to counter-intuitive macroscale mechanical responses. Particles with diameters greater than
the gaps between the cell walls of the segregated matrix get lodged between and in the walls,
bridging the gaps in the segregated matrix and increasing the stiffness of syntactic foams. In
contrast, particles with smaller diameters with higher particle crushing strength get lodged
only inside the cell walls of the segregated matrix, resulting in higher densification stresses
(energy absorption). We show that stiffness and densification can be tuned while enabling
lightweight syntactic foams. These novel discoveries will aid in facilitating functional and
lightweight syntactic foams for cores in sandwich structures.

Keywords Segregated Matrix ⋅ Syntactic Foams ⋅ Glass Micro-Balloons ⋅ Thermoplastic

Urethane ⋅ Selective Laser Sintering

1Introduction
Syntactic foams are composite materials with hollow thin-walled fillers blended in a
continuous polymeric [1, 2, 3, 4, 5], metallic [6], and ceramic [7] matrices. Incorporating
these hollow fillers makes these foams inherently lightweight and can help achieve tailorable
mechanical properties [8, 9]. Due to their inherent buoyant behavior, these foams see wide
applications in the marine and aerospace industries [10]. Compared to other types of
composite materials with reinforcements ranging from continuous fibers, carbon nanotubes,
and short fibers, manufacturing particle-reinforced composites offers the advantage of low-
cost production. In this paper, we focus on the additive manufacturing of lightweight flexible
syntactic foam composites with a segregated matrix system. Particularly, we aim to
understand the influence of the Selective Laser Sintering (SLS) process on the polymer
matrix distribution in the presence of Glass Micro-Balloon (GMB) particles.

Polymeric syntactic foams have been traditionally fabricated with Glass Micro-Balloons in a
thermoplastic matrix using injection molding [3] and compression molding [11]. Kumar et
al. [3] fabricated cenosphere/high-density polyethylene (HDPE) syntactic foams using an
industrial-scale injection molding technique. They suggested optimized parameters to
minimize the fracture of cenosphere particles and promote the proper mixing of particles in
the HDPE matrix. They observed that increasing the cenosphere content in the HDPE matrix
resulted in an increase in the tensile moduli values, but the tensile strength decreased. Surface
modification of the cenosphere particles reduced the extent of the decrease in the strength of
the syntactic foams. Jayavardhan et al. [11] proposed a compression molding technique to
manufacture Glass Microballoon (GMB)/high-density polyethylene syntactic foams. They
conducted an experimental study to elucidate the impact of volume fractions of GMBs on
tensile and flexural properties. They observed that the specific moduli increased for the foams
with the increase in volume fraction of GMBs, but the specific strength remained comparable
to neat HDPE. Although conventional manufacturing techniques, such as compression and
injection molding, offer the advantages of producing syntactic foams rapidly, they can be
cost-intensive and can potentially damage the fillers. At lower volume fractions, the particles
also tend to coagulate [9]. Additionally, employing conventional manufacturing techniques
can be arduous and cost extensive to produce parts with complex shapes due to the need for
complex molds.

Additive Manufacturing (AM), a manufacturing technique that works on the principle of

additive design - constructing 3D CAD geometries layer by layer - can be employed to
overcome the manufacturing issues mentioned above that can occur during conventional
techniques [12]. Research has been performed to characterize GMB reinforced high-density
polyethylene (HDPE) syntactic foams manufactured using an extrusion-based AM technique,
Fused-Filament Fabrication (FFF) [13, 8, 4, 2]. The addition of GMBs to the high-density
polyethylene (HDPE) matrix was shown to increase the elastic modulus and yield strength
under quasi-static compressive loading. Bonthu et al. [14] have also identified difficulties that
must be resolved in order to produce high-quality syntactic foams using the FFF production
technology. Research has been conducted to overcome the limitations of the FFF process by
using 3D printing techniques with thermosets in which syntactic foams are printed directly
without the need to manufacture individual rasters [15]. These 3D printing techniques have
been in the spotlight due to the low-cost production of complex-shaped syntactic foams;
however, they require complicated support structures. Selective Laser Sintering (SLS) is a
powder-based AM technology in which powder is deposited layer by layer. A high-energy
laser beam is then used to coalesce/sinter the powder to generate a three-dimensional
geometry. The unsintered powder on the print bed serves as the support structure, which can
be easily removed by blowing air in the post-processing stage [16, 17], which eliminates the
requirement of support structures to print complex parts. SLS is being widely used in various
areas and sees applications in the biomedical [18, 19], aerospace [20], automotive [21], and
defense [22] industries.

Currently, several thermoplastic polymer powders for SLS are commercially accessible,
including polyamide 12 (PA12), polyamide 6 (PA6), polyamide 11 (PA11), thermoplastic
urethane (TPU), polypropylene (PP), and polyethylene (PE) [23]. Research has been
conducted to manufacture periodically patterned and topologically optimized lattice
structures with pure PA12 [24, 25] and TPU [26] polymer systems to improve the energy
absorption of the structures. Furthermore, SLS has been employed to manufacture TPU and
PA11 composites with carbon nanotubes [27, 28, 29, 30, 31], and graphene [32]. In recent
times, multiple studies have been conducted on composite foams with SLS to reduce weight,
enhance mechanical properties [33, 34, 35], improve thermal stability [36], manufacture
biocompatible composites for medical devices [37], and improve electrical
conductivity [38, 39, 40, 41], compared to pure polymer. These studies on the use of polymer
blends in SLS show that micro- and nano-sized particles can be used to alter the properties of
polymer parts that have been sintered. In recent years, SLS has also been in focus to produce
syntactic foams by incorporating GMBs [42, 43] and ceramic particles [44, 45, 46] in
continuous PA 12 matrix, a relatively rigid plastic compared to flexible TPU. In addition,
Mousa [47] researched avenues to improve the bonding of the GMBs by surface
modifications of the microbubbles, and their influence on the tensile strength, modulus, and
impact strength of the GMB-filled PA12 syntactic foams. They concluded that the tensile
properties improved with increasing the GMB content (treated/untreated), but the impact
strength and ductility of the foams reduced. Cano et al. [48] investigated the effect of GMBs
incorporated into the PA12 matrix on the fracture toughness of the sintered parts at various
temperatures. They showed that inadequate bonding between the matrix and the particles
decreased the resistance to fracture. Although PA12 syntactic foams have been the subject of
a significant amount of research, a relatively small number of studies have been conducted on
TPU syntactic foams utilizing SLS.

Although research has been conducted on SLS-fabricated syntactic foams with a continuous
matrix system surrounding the reinforcing particles, their mechanics with a segregated
matrix system have not been thoroughly investigated. In recent studies on SLS printing with
TPU [28, 26, 49, 50], it was observed that the TPU matrix forms a controlled segregated
structure at the microscale, which is a discontinuous matrix system, in the sintered parts.
Therefore, it is essential to comprehend the mechanics of this type of syntactic foam in which
the particles are incorporated into a segregated matrix system.

In this study, we elucidate how additive manufacturing parameters can be coupled with GMB
parameters to achieve the desired mechanical response or to tune the mechanical response of
syntactic foams with the segregated TPU matrix. To that end, we present an SLS-based
manufacturing method for producing multi-scale architected syntactic foams with segregated
TPU matrix systems containing different grades of GMBs at varied volume fractions. First,
we conducted an experimental parametric study to determine the optimal print parameters for
the production of our syntactic foams. Then, we determined the effects of the GMB
characteristics on the mechanical response of these foams by varying the volume fractions of
various classes of GMBs with varying particle size distributions. In contrast to well-studied
existing syntactic foams with continuous matrix systems, in segregated matrix systems,
the particles themselves can be integrated in a variety of ways, influencing the
mechanical response. Finally, we printed syntactic foams with macroscale architectures
consisting of struts with microscale architecture as a result of GMBs in the segregated matrix.
We examined the effect of this manipulation of microscale and macroscale architectures on
the compressive response of architected syntactic foams.

2Motivation
Our overarching goal is to develop a novel manufacturing approach that combines architected
hierarchy at the macroscale with design at the microscale by combining different grades of
GMBs in a segregated TPU matrix. A demonstration of design ideas from the current work
on flexible syntactic foams that connects microscale and macroscale architectures to diverse
applications of in the sports industry (shoe sole) and aerospace industry (core for sandwich
structure) is illustrated in Figure 1.

Figure 1:Diverse
applications of flexible syntactic foams in the sports industry (shoe sole) and aerospace
industry (core for sandwich structure) with tunable macroscale and microscale architectures.
The segregated matrix can impact how the reinforcing particles get lodged in the system
compared to conventional syntactic foam systems that use continuous matrix systems. As a
result, understanding the deformation mechanics and corresponding mechanical performance
of these syntactic foams with a segregated matrix is critical.

In this paper, we seek to answer the following fundamental questions:

 •

What are the effects of print parameters on the microscale morphology and
mechanical performance of 3D-printed syntactic foams?

 •

What is the role of GMBs in the evolution of AM-induced matrix segregation within
syntactic foams?

 •
 How can the GMB and print parameters be modified in unison to tune the properties,
such as stiffness and densification, of additively manufactured syntactic foams?

 •

By designing and fabricating architected syntactic foams, can we tailor the macroscale
mechanical properties through a multi-scale hierarchy from micro to macroscale?

3Methodology
In this section, we will describe the properties of the constituent materials used for
manufacturing the syntactic foams, followed by details of the SLS process and different
techniques used to characterize the constituent materials. Then, we will describe the
mechanical test procedures performed in compliance with the ASTM standards. An overview
of the methodology section with interconnected methods with expected outcomes is
summarized in Figure 2.

Figure
2:An overview of the methodologies with interconnections between the variables, techniques,
and outputs.
3.1Materials

We procured thermoplastic polyurethane (TPU) powder from Sinterit (Product Name: Flexa
Grey; particle size between 20 and 120 µm) to manufacture the syntactic foams. We chose
3M K20, 3M K46-HS, and 3M im30k grades of GMBs to create powder blends consisting of
TPU powder and different volume fractions of GMBs. The properties of the constituent
materials are summarized in Table 1 and Table 2.

Table 1:Properties of TPU powder (Flexa Gray from Sinterit) [51]

Material Tensile Elongation Shore A Melting Softening
Granulation
Property Strength at Break Hardness Point Point
(MPa) (%) (°C) (°C) (�m)
Value 3.7 136 70 160 67.6 20-105
Table 2:Properties of GMBs [52][53] (* foam representation is given for volume fraction =
20 %)
 Test Particle Foam GMB
Gra True Particle
Pressure Representatio Representati comment
de Density Size - D50
) n on s
(MPa) (g/cc) (�m)
3M
Large
- 3.45 0.2 60 GM60 SF60-20
particles
K20
3M
- Medium
51.71 0.46 22 GM22 SF22-20
K42 particles
HS
3M
- Small
193.05 0.60 15.3 GM15 SF15-20
iM3 particles
0k
3.2Manufacturing

In this section, we will discuss the manufacturing of pure TPU and GMB-reinforced TPU
syntactic foams using the SLS technique.

3.2.1SLS Printing

We manufactured the syntactic foams using a Lisa 3D printer from Sinterit [54], which is a
desktop-based SLS printer equipped with an IR laser diode of 5W and a wavelength of 808
nm. A general description of the SLS printing process is provided in Section A. We used pure
TPU powder and TPU/GMB blends with three different volume fractions of GMBs – 20%,
40%, and 60% to additively manufacture pure TPU and syntactic foams, respectively. For
example, to prepare a mix with a 20% volume fraction of GMBs, 800 ml of TPU powder and
200 ml of GMBs were measured. The mixture was then loaded into a V-shaped mixer (Power
= 110 V and capacity = 1.2L), and the blend was first mixed at 30V for five minutes followed
by mixing at 70V for another three minutes. To check the quality of mixing in the blend, we
obtained SEM images of the powder blends; we have shown and discussed these images in
detail in Section 4.1.1. The same mixing process was followed for all volume fractions and
GMB types.

3.2.2Print Parameters

Identifying the optimal parameters to sinter the polymer effectively is challenging for the
SLS process. Therefore, we varied the print parameters to understand their contributions to
the morphology and the mechanical response of the printed foams.

Laser Power Ratio

In the Sinterit Lisa printer, the laser power ratio is controlled by two parameters: i) laser
power supplied and ii) scanning speed, which affect the effective energy supplied to sinter the
powder. The laser power ratio (LPR) is directly related to the laser power and inversely
related to the scanning speed. As the laser power directly increases the energy density
supplied to the polymer powder, this increases the depth of the melt pool, as illustrated
in Figure 17. Whereas increasing the scanning speed, decreases the dwell time, which
decreases the energy supplied, and vice versa. In the Sinterit Lisa printer [54], the laser power
is fixed at 5W. Therefore, by changing the LPR, we effectively vary the scanning speed while
holding the power at 5W. In the current study, we used laser power ratios of 0.75, 1, 1.5, and
2, which resulted in print speeds of 83 mm/s, 56 mm/s, 37 mm/s, and 28 mm/s, respectively.
When we increase the laser power ratio, it can potentially affect the micro-phase morphology
in the TPU that impacts the mechanical response - this will be discussed in further detail in
Section 4.1.2.

Layer Height

The height of the individual layer that will be placed on the print bed for each sintering phase
is defined by the layer height parameter. Since SLS is a layer-based AM process, increasing
the height of each layer could cause a staircase effect, even though it can shorten the print
time. Furthermore, for the same provided energy density, a higher layer height lowers the
bonding between individual print layers, thus compromising mechanical performance. In this
work, we vary the layer height between 175 �⁢� to 75 �⁢� to see how it influences the
mechanical response.

3.3Materials Characterization

To evaluate the properties of the powder blends and syntactic foams, we used a variety of
characterization techniques. Using scanning electron microscopy (SEM), we determined the
particle size distributions and microscale morphologies of TPU powder and GMBs. Using
SEM, we were also able to examine the microstructure and failure morphologies of the 3D-
printed TPU and TPU/GMB foams. Fourier Transform Infrared (FTIR) spectroscopy was
essential for identifying the chemical compositions of the foam production constituents.
Using FTIR, we analyzed the spectroscopic properties of TPU powder and TPU/GMB
composites, as well as chemical changes during the sintering process and the impact of print
parameters on foam composition. Using Differential Scanning Calorimetry (DSC), we were
able to evaluate the thermal properties of the TPU and determine the optimal sintering
window. Using a helium porosimeter, we determined the porosity values of our pure TPU
foam and TPU syntactic foams containing GMBs. In this study, we also used the porosimeter
to comprehend the porosities of uncompressed and compressed TPU syntactic foams. These
characterizations are further described in detail in B.

3.4Mechanical Testing

3.4.1Tensile testing

We carried out uniaxial tensile tests of the printed foams in compliance with ASTM
638 [55] on the MTS universal testing instrument at the Structures and Materials Testing
Laboratory at the UW Madison with a load cell capacity of 250N. Type IV sample was
chosen to get tested and loaded at a 50 mm/min cross-head speed. This cross-head speed was
chosen such that the test completion time stays between 1 and 5 minutes. Since TPU samples
typically display failures at very high elongation, an Epsilon One optical extensometer was
used to obtain the engineering strains. To understand the individual effect of print parameters
on tensile performance, samples were printed with different laser power ratios and layer
heights. In addition, samples with various GMB volume fractions produced using the
finalized print parameters were examined under tensile stress.
3.4.2Compression testing

We conducted uniaxial compression tests on TPU foam samples on the ADMET 2613
tabletop frame equipped with a load cell capacity of 50 kN. The ASTM D1621 [56] standard
for compression testing of plastics was used for these tests, and the sample size was chosen as
25mm x 25mm x 25mm for the cube and all architected designs. Samples were loaded under
uniaxial compression at a loading rate of 2.5 mm/min (10% of height per minute) to achieve
strain values of 20%, 30%, and 50%, and a preload of 1N was used. We subjected the
compressed samples to a second loading cycle one week after the initial loading cycle to
examine the cyclic behavior under compression. All samples were loaded to 50% strain
values for the second cycle.

4Results and Discussion

In this section, we discuss the characterization results for the constituent powders followed by
an evaluation of the printed foam morphology. In addition, the impact of the print settings
and GMB parameters on the mechanical performance of the foam is discussed. Finally, we
elucidate the effect of GMB volume fraction and size on the mechanical response of these
printed syntactic foams.

4.1Characterization

We used SEM to characterize the particle size distribution of the constituent powders and
understand the distribution of the GMB inclusions in the powder blends. SEM images were
also used to elucidate the effects of adding different GMBs on the morphology of the printed
foams.

4.1.1Powder blend morphology

Figure 3 shows the SEM images of TPU/GMB blends with the three sizes of GMBs mixed at
a volume fraction of 20%. We see that the polymer powder has an irregular shape, whereas
all GMBs are perfectly spherical. Moreover, the smaller GMBs exhibited a higher GMB
particle density per unit area than the larger GMBs. As the TPU powder and GMBs have
distinct energy absorption characteristics, the GMB shape and distribution in the polymer
blend have a substantial impact on the energy absorbed by the polymer powder in the mix.
Smooth spherical shapes of the GMBs are expected to create less hindrance in the path of the
laser, and hence, the polymer blends absorbed more energy [43]. However, this effect can be
counteracted by the hindrance caused by a higher particle density as we reduce the particle
size for a fixed volume fraction. This implies that as particle density increases, we need to
provide more energy density to the system. For this study, we have kept the supplied energy
density constant for all TPU/GMB blends.
 TPU/GM60-20 TPU/GM22-20 TPU/GM15-20

(a) (b) (c)

Figure 3:SEM images (E.H.T. = 3kV and Signal = SE2) of (a) TPU/GM60-20, (b)
TPU/GM22-20, and (c) TPU/GM15-20 powder blends.
4.1.2FTIR Spectroscopic Analysis

We examined the chemical bonds within TPU and TPU/GMB foams using FTIR. In this
study, we used a polyester-based TPU as can be seen from the FTIR spectroscopy graph
shown in Figure 5, which shows the characteristic \chemfigC=O group in the polyurethane
and NH stretching vibrations at 1741 �⁢�-1⁢ and 1540 �⁢�-1⁢ , respectively [57]. We also see
two strong absorption peaks at 2960 �⁢�-1⁢ and 2823 �⁢�-1⁢ which is attributed to the stretch
vibrations of �⁢�2 and �⁢�3. When we increased the laser power ratio, effectively changing
the supplied energy density to the powder, we observed that the absorption intensity of the
hydrogen-bonded \chemfigC=O band (right of 1722 �⁢�-1⁢ ) compared with the non-hydrogen
bonded \chemfigC=O band (left of 1722 �⁢�-1⁢ ) increased as observed in Figure 4. This
behavior is a consequence of the formation of more organized hard segments in the TPU’s
micro-phase, resulting in a higher percentage of hard domains relative to soft domains.

Figure 4:FTIR curves for different laser powers (a) from 4000 �⁢�-1⁢ to 400 �⁢�-1⁢ and (b)
from 2000 �⁢�-1⁢ to 1500 �⁢�-1⁢ .
Next, comparing the FTIR spectra in Figure 5(b) of pure TPU and TPU/GMB foams printed
using SLS, we only observed new peaks close to 980 �⁢�-1⁢ and 850 �⁢�-1⁢ , which correspond
to the presence of \chemfigSi-O-Si. This shows that no new chemical bonds were formed
between the TPU matrix and GMBs, and hence, the bonding was completely physical. [58].

Figure 5:FTIR curves for (a) TPU powder and the 3D printed pure TPU foam and (b) TPU
foam and SF60-20 foam printed with same parameters.
4.1.3Influence of print and foam parameters on micro-structure

Pure TPU foams manufactured using the print parameters specified in Section 3.2.2 were
analyzed under SEM to observe the effect of these parameters on the microstructure of these
foams. From Figure 6, we see that the TPU foams manifested a controlled porous structure
with a segregated matrix as observed in previous studies [28, 26, 50]. In this study, the wall
size of the segregated matrix at the microscale is referred to as cell wall thickness. When we
increased the laser power ratio (LPR) from 0.75 to 2.0, the cell wall size appeared to increase,
manifesting denser foams. This is attributed to an increase in the melt pool size due to higher
energy supplied. We did not see a distinctive change in foam morphology when layer height
decreased for LPR values of 0.75 and 1.0. However, for higher LPR values of 1.5 and 2.0, we
observed that the cell wall thickness increased with decreasing layer height. This is attributed
to the repeated sintering of more layers due to smaller layer heights. The implications of LPR
values and layer heights on the mechanical response will be further elaborated
in Section 4.2.1.
 LH /
0.75 1.0 1.5 2.0
LPR

75�⁢
�

(a) (b) (c) (d)

125�⁢
�

(e) (f) (g) (h)

175�⁢
�

(i) (j) (k) (l)

Figure 6:SEM images (E.H.T. = 3kV and Signal = SE2) of morphologies observed for pure
TPU foams printed with different Laser Power Ratio (LPR) and Layer Height (LH)
combinations.
To evaluate the effect of the size of the GMBs, we printed foams with the final parameters
chosen based on the mechanical performance, details of which are provided later
in Section 4.2.1. Figure 7 shows the SEM images of syntactic foams printed with different
size GMBs (SF60 - large, SF22 & SF15 - small) at a fixed volume fraction of 20%. We
observed that bigger GMBs tend to lodge between cell walls and in the cell walls in SF60-20.
However, in SF22-20 and SF15-20 with smaller GMBs, the GMBs tend to get lodged within
the cell walls of the TPU foam due to the larger space between the cell walls than the particle
size. The cell walls of the SF22-20 and SF15-20 foams are also thinner than those of the
SF60-20 foams. As explained above, this drop in cell wall size is related to the decrease in
energy absorbed by the TPU powder due to the presence of GMBs with a high particle per
area density. In this study, the energy delivered to all polymer blends was held constant to
correlate the GMB parameters with the print parameters, and further evaluate the mechanical
performance of the printed syntactic foams. However, a larger value of supplied energy
density may be selected for smaller GMBs to achieve the same cell wall thickness for all
blends. To show the effect of energy density on the morphology (cell wall thickness) of the
TPU/GMB syntactic foams, we printed SF60-20, SF22-20, and SF15-20 foams with an LPR
of 1.5 and 2.0 and a layer height of 75 �⁢�. We see in Figure 7 (d)-(f) that the cell walls for
the smaller particles appear thicker compared to foams shown in 7 (a)-(c) as we increased the
LPR. Therefore, the energy supplied to the system must be increased for the TPU/GMB
powder blends to avoid reducing the wall thickness.

Print Parameters /
SF60-20 SF22-20 SF15-20
Foams

LH = 75�⁢� , LPR =
1.5

(a) (b) (c)

LH = 75�⁢� , LPR =
2.0

(d) (e) (f)

Figure 7:SEM images (E.H.T. = 3kV and Signal = SE2) of foams printed with final print
parameters (Laser Power Ratio = 1.5 and Layer Height = 75 �⁢�) containing 20% volume
fraction of GMBs for (a) SF60-20; (b) SF22-20; and (c) SF15-20; and SEM images of foams
printed with the modified print parameters (Laser Power Ratio = 2.0 and Layer Height =
75 �⁢�) containing 20% volume fraction of GMBs for (d) SF60-20; (e) SF22-20; and (f)
SF15-20
4.1.4Porosity

We examined the porosity of TPU-based syntactic foams to determine the effect of

introducing GMBs into the matrix during additive manufacturing. This porosity is the void
content between the cell walls of the TPU, and does not consider the voids that exist within
the GMB particles. The porosity values for TPU, SF60-20, and SF60-40 foams remained
constant, ranging between 27 and 28 percent. However, the SF60-60 foam porosity increased
significantly to 38 percent, which can be attributed to an insufficient matrix available to bind
together a high volume fraction of GMBs. By altering the energy density supplied to the
system, the porosity values of GMB-containing foams can decrease as was observed with the
increase in cell wall thickness for pure TPU in the section 4.1.3. From Figure 7, we also
noticed that for the same volume fraction of GMBs, decreasing the size of the GMBs
decreased the cell wall size, hence increasing the porosity of the foam. An increase in particle
density of the TPU/ GMB blend altered the transmittance and absorbance of the mixture
(Section C.1). This suggests that raising the energy density of the laser with increasing GMB
content can decrease the porosity as can be observed in Figure 7, comparing (a) - (d), (b) -
(e), and (c) - (f). The same porosity can be achieved if higher energy density is supplied to a
powder blend with smaller particles as compared to low energy density supplied to a blend
without particles - this phenomenon can also be observed in Figure 7 comparing (a) - (e).

4.2Mechanical Response
In this section, we first discuss the effect of print parameters on the tensile properties of pure
TPU foams. Then, the performance of syntactic foams printed with the chosen print
parameters under tensile and compression loading with various volume fractions and GMB
types is evaluated. In addition, failure morphologies are discussed to complement the
response.

4.2.1Tensile Response

Pure TPU foams

Figure 8 shows representative stress-strain curves for all 3D printed pure TPU foams,
and Figure 8 summarizes tensile properties (modulus and strength) from all the tests
performed. These foams exhibited a small linear elastic region followed by a nonlinear
ductile response under tension. By increasing the laser power ratio (LPR) at a constant layer
height (LH) value of 125 �⁢�, we observed that the tensile modulus and strength increased.
This is attributed to an increase in cell wall thickness and reduced porosity with an increase in
LPR, as observed in the SEM images in Section 4.1.3. The strength increased because of an
increase in the load-bearing phase resulting in better load distribution. Moreover, an increase
in the cell wall size increased the strain to failure. However, with an increase in the energy
supplied value at a higher LPR, we also observed that the dimensional stability decreased,
that is, we observed higher z-bulging. Therefore, to avoid this, we printed a specimen at an
LPR of 1.5 and a reduced LH of 75 �⁢�. When we decreased the LH, tensile strength
increased drastically by 10.87 % compared to LH of 125 �⁢� shown by black star marker
in Figure 8 , while tensile modulus increased moderately by 0.62 %. This is attributed to the
repeated sintering of layers as we decreased the LH, which increased the energy supplied to
individual layers and also, provided better bonding even at a lower LPR of 1.5.
Figure 8:(a) Representative tensile stress-strain plots for pure TPU foams with changing laser
power ratio and a fixed layer height (layer height was varied for a selected laser power); (b)
Summary of tensile properties with varying print parameters.
The influence of the two parameters, LPR and LH, was considered in unison to determine
optimal final print parameters for enhanced tensile response and high dimensional stability of
the printed specimens. Considering the tensile performance, print duration, and printer
limitations, we chose an LPR of 1.5 and an LH of 75 �⁢� for the next printing steps.

Particle reinforced TPU foams

Using the final print parameters established for pure TPU foams, we printed TPU-based
syntactic foams and tested them under quasi-static tensile loading. We first fixed the GMB
size to larger particles and varied their volume fraction, followed by fixing the volume
fraction at 20% and varying the GMB sizes. Figure 9(a) shows representative tensile stress-
strain responses for SF60 foams with varying GMB volume fractions. We see that the
strength and strain to failure decreased with increasing GMB volume fraction of the larger
particles. This behavior can also be attributed to poor adhesion of the larger particles partially
embedded within a segregated matrix resulting in debonding and insufficient load transfer.
SF60-60 with 60% GMB volume fraction showed significantly poorer tensile strength and
failure to strain properties compared to SF60-20 and SF60-40 due to the lack of TPU matrix
to effectively bind the GMBs at very high volume fractions. On the other hand, the elastic
moduli were in the range of 13 to 16 MPa for TPU, SF60-20, and SF60-40 foams. We
speculate that these small deviations are due to competing effects of stiffer GMB particles
and reducing cell wall size with increasing GMB volume fraction. Even though the GMBs
are stiffer than TPU, their effect is counteracted by a reduction in the cell wall thickness
when printed with the same supplied laser energy density.
Figure 9:(a) Representative tensile stress-strain plots for syntactic foams with changing GMB
volume fraction (for volume fraction 60%, a laser power ratio of 2.0 was also used); (b)
Representative tensile stress-strain plots for syntactic foams with varying grades of GMBs
with fixed volume fraction = 20%.
Figure 9(b) shows the representative tensile stress-strain responses of syntactic foams with
three GMB particle sizes at 20% GMB volume fraction. We observed that the tensile strength
and strain to failure reduced when we incorporated smaller GMBs (GM22 and GM15) than
larger GMBs (GM60) at the same volume fraction, rendering the foam quasi-brittle. This is
attributed to two aspects: 1) Smaller GMBs tend to get lodged in the cell walls, resulting in
more stress concentration locations within the cell wall, and 2) Smaller particles have higher
particle density than larger particles for a fixed foam volume. This results in a higher surface
area of particles in contact with the matrix causing higher debonding cites compared to that
of the SF22-20 foams, thereby, reducing the tensile strength. Tensile moduli values of SF22-
20 and SF15-20 dropped to an average of 12.8 and 11.6 MPa compared to 16.3 MPa for the
pure TPU foam. We attribute this reduction in moduli to the cell wall thickness reduction
associated with higher particle density in SF22-20 and SF15-20 foams compared to that of
SF60-20 as shown previously in Figure 7.

Tensile properties for different grades of GMBs at different volume fractions are summarized
in Figure 12.

4.2.2Compressive Response

We performed uniaxial compression tests to determine the mechanical response of pure and
particle-reinforced TPU foams. We first selected GM60 particles and varied the volume
fractions ranging from 20% to 60% in increments of 20% and incorporated them in the
segregated TPU matrix. Further, to understand the effect of particle size within the segregated
structure of the matrix, we choose GM22 and GM15 GMBs with volume fractions of 20%
and 40%. In this paper, the compressive strength is chosen as the compressive stress value at
30% strain value.

Figure 10:Typical
compressive stress-strain behavior of (a) pure TPU foams and (b) syntactic foams, and
Illustrations of the compression mechanics for (c) pure TPU foams and (d) syntactic foams.
The stress-strain response for pure TPU foams resembled the non-linear behavior of typical
foamed elastomers [59] as shown in Figure 10(a). Due to the segregated matrix micro-
structure (shown as blue regions) in these foams, we observed a small elastic region at the
beginning which is then followed by the elastic buckling zone where the cell walls start to
buckle. It should be noted that the matrix segregation is intentional, and it gives rise to a
controlled porous structure. As the air is pushed out of the foams, the cell walls compress on
themselves resulting in the densification zone. Figure 11 shows a representative compressive
stress-strain response of printed TPU foam. A schematic of a typical compressive stress-
strain behavior for syntactic foams (Figure 10(b)) consists of a very small initial linear region
associated with the enhanced elastic modulus due to the presence of reinforcing particles.
This was followed by a second zone called the plateau region, corresponding to reduced
stiffness due to particle cell wall buckling and crushing. Finally, the third zone, known as the
densification region, occurs when the cell walls of the matrix and the particles compress.
Figure 11:(a) Representative compressive stress-strain plots for SF60 foams with changing
GMB volume fraction; (b) Representative compressive stress strain plots for syntactic foams
with varying grades of GMBs with fixed volume fraction = 20%.
Different GMB Volume Fractions:

Figure 11(a) shows the representative compressive stress-strain responses of SF60 foams
with varying GMB volume fractions. In general, we see that these responses exhibited an
initial elastic region associated with the enhanced compressive modulus compared to the pure
TPU foams in this region. This is attributed to the large particles lodged within and between
the cell walls resulting in cell wall stiffening as well as creating stiff bridges between the cell
walls of the segregated matrix. However, we noticed that the modulus reduced in SF60-60
foams due to insufficient matrix material available to bond these GMBs for effective load
transfer to occur. At strain values beyond this initial region, we observed a knee formation
due to the initiation of particle crushing in the gaps of the segregated matrix. A dominant
plateau region after this knee formation is a characteristic behavior of particle crushing as
discussed above in reference to Figure 10(b). We observed that the densification region is
lower in the syntactic foams than that of the pure TPU foams, and it reduced with increasing
GMB volume fraction. This is because of the lower crushing strength of GM60 particles. The
compressive modulus and strength for SF60 foams are summarized in Figure 12(b) using
square markers.
Figure 12:Summary of (a) tensile properties and (b) compressive properties, for different
GMB types with varying GMB content (volume fraction).
Different GMB Grades:

Figure 11(b) illustrates compressive stress-strain responses of syntactic foams with GM60,
GM22, and GM15 GMB size particles at 20% volume fraction. With decreasing GMB size
(GM22 and GM15), compression behavior resembled that of pure TPU foam, and the plateau
region disappeared compared to that observed for TPU foam with bigger particles (GM60).
We hypothesize that this response reflects matrix material dominance and does not involve
particle crushing. Since the smaller particles are entirely embedded within flexible TPU
matrix cell walls, it will take larger strains to reach their high crushing strengths. Although
GM15 has a higher crushing strength than GM22, SF15 and SF22 foams display similar
compression behavior as smaller particles embedded in the TPU matrix never achieved
sufficient stresses to crush them. Figure 11(b) shows that SF22 and SF15 foams have lower
densification stress than pure TPU foam. As mentioned previously in section 4.1.4, printing
syntactic foams with smaller GMBs increased porosity (decreased the cell wall thickness). As
a result, the SF22 and SF15 foams reach the densification stage at a higher strain value than
pure TPU foams.

According to the summary plot in Figure 12(b), SF22-20 and SF15-20 foam compressive
modulus increased at a GMB volume fraction of 20%. This is due to the stiff GMBs
embedded in the cell walls increasing the stiffness of the cell walls. However, the modulus
drops at 40% volume fraction, much sooner than SF60 foams at 60%. Due to higher particle
packing density with reduced particle size, the matrix cannot transfer loads effectively
between reinforcing particles at a 40% volume percentage of smaller particles. In contrast, we
can incorporate a higher volume fraction of larger GMBs in the matrix before the foam’s
modulus declines. When the compressive moduli are weight normalized, there is no dip for
SF15-40, although the increase is not as great as in SF60-40 and SF22-40.

Densification Mechanics of TPU foams

We evaluated the residual performance of pure TPU, SF60-20, SF22-20, and SF15-20 foams
under cyclic loading to evaluate the densification mechanics. In the first cycle, we subjected
each set of samples to 20%, 30%, and 50% strain values before allowing them to relax for
one week [60]. After a one-week interval, we loaded all samples to 50% strain
values. Figure 13 displays the representative compressive stress-strain responses of the
pristine (solid lines) and compressed foams (dashed lines). Further, black, red, and blue lines
represent compression to 20%, 30%, and 50% strain, respectively, in the first cycle.
Figure 13:Compressive stress-strain curves for (a) TPU foam, (b) SF60-20, (c) SF22-20, and
(d) SF15-20 under repeated loading. Samples are compressed twice till mentioned strain
values with a gap of 1 week between the loading cycles.
For pure TPU foams, both pristine and compressed foams exhibited similar compressive
behavior and a slight increase in densification stress. This is because, after the first
compression cycle, pure TPU foams became denser. In the second cycle after 50%
compression in the first cycle, there was a reduction in the plateau region in the TPU foam
due to the excessive deformation at 50% strain weakening the cell walls in the segregated
matrix structure. The compressive response of SF60-20 showed a significant reduction in the
response during the second compression cycle with maximum reduction following the 50%
compression during cycle 1. This is a result of GMBs crushing during the first compression
cycle in SF60-20 foams. When the SF60-20 foams were compressed again in cycle 2, there
were fewer particles to support the load and the porosity had risen due to particle crushing in
the first cycle. The reduction in compressive properties of SF22-20 foams is not a result of
particle compression; however, when these foams were loaded in cycle 1, the particles
partially protruding from the cell walls may have been compressed against one another.
During the second cycle, particles may have debonded from the matrix manifesting as a
reduction in the compressive behavior. After the first cycle, the compressive properties of
SF15-20 foams subjected to strain values of 20% and 30% in cycle 1 did not degrade.
However, a substantial decrease is observed for foam that was loaded to 50% strain in cycle
1. This is because of the reduced size of the particles in SF15-20 foams, they began to
interact at higher strain values as explained in 4.2.2, resulting in particle debonding from the
TPU matrix. Therefore, we only observed a decrease in properties for SF15-20 foams when
the foam was initially loaded to 50% strain.

Figure 14 shows the SEM morphologies of syntactic foams after 2 cycles of compression to
50% strain as described above. We observe severe particle crushing in SF60-20 foam
compared to SF22-20 and SF15-20 foams. We also observed that as the compressive strain
values of cycle 1 increased, the porosity values for SF60-20 foam increased from ∼27% to
43%, due to the crushing of GMBs. However, the porosity values remained consistent for
SF22-20 and SF15-20 foams, in the range of 33% to 34%, indicating that no particle crushing
occurred during the compression cycle. The SEM morphology of twice-compressed SF22-20
foam reveals that the particles were too small to fit into the gaps, but they can still begin to
interact during the initial phases of compression loading. However, due to the smaller size of
GMBs in SF15-20, the particles can interact at higher strain values leading to the
densification region. Although the GMBs in SF22-20 and SF15-20 foams can interact at
some point during compression loading, they will not be crushed due to their exceedingly
high crushing strengths.

SF60-20 SF22-20 SF15-20

(a) (b) (c)

Figure 14:SEM images (E.H.T. = 3kV and Signal = SE2) of (a) SF60-20, (b) SF22-20, and
(c) SF15-20 compressed twice to 50% strain with a gap of 1 week between both loading
cycles.
Process - Structure - Property Map

Based on our observations, we present a Process – Structure – Property map (Figure 15) to
assist with designing SLS-printed syntactic foams. This map depicts how the energy density
supplied during manufacturing, GMB size and volume fraction, and associated internal
micro-structure influence the foam compressive modulus and strength. We see that matrix
segregation manifested during the SLS process reduces with increasing laser energy density
for pure matrix. We need to supply higher energy density to syntactic foams than pure TPU
to achieve a similar extent of matrix segregation. For syntactic foams, we observe that the
compressive modulus of syntactic foams with segregated matrix increases with increasing
volume fraction up to a certain percentage, beyond which this value drops. This optimum
volume fraction reduces with reducing particle size. In addition, the compressive modulus
with larger particles is higher than that with smaller particles when we have a segregated
matrix. The values will converge as we approach a solid matrix with increasing energy
density. On the other hand, there is a switch in compressive strength behavior. We have
higher strength with smaller particles when we have a segregated matrix at lower energy
densities, and the responses diverge when we approach a solid matrix.

Figure
15:Proposed Structural - Process - Property map to aid with designing and manufacturing
syntactic foams with segregated matrix using Selective Laser Sintering.
5Demonstration - Compression Response of Architected
Syntactic Foams
To demonstrate hierarchy at the macroscale, we manufactured architected syntactic foams
with dimensions of 25mm x 25mm x 25mm. We chose three architectures, namely: i) gyroid,
ii) diamond, and iii) conical as shown in Figure 16. We calculated the effective strain and
effective stress by using the dimensions mentioned above. These effective properties
represent the response of the overall architected structures and not of the local struts.
Representative stress-strain compressive responses of all architected syntactic foams - gyroid,
diamond, and conical - are summarized in Table 3 and Table 4. Observably, the gyroid and
diamond architectures exhibited bending-dominated stress-strain behavior [24, 61], whereas
the conical architecture displayed buckling-dominated (stretching-dominated) behavior [24].

Figure 16:Printed
architected SF60-40 syntactic foam samples.
We can observe from Table 3 that after incorporating GMBs into the architected designs, the
stiffness increased with an increase in GMB volume fraction from 0% to 40% due to an
expected increase in the stiffness of the struts due to the addition of GMBs. The weight
normalized change increases further due to weight reduction by incorporating hollow GMBs.
From Table 4, we see a decrease in the strength of the bending-dominated architected foams,
whereas, an increase in the buckling-dominated architecture. This difference is due to the
underlying mechanism of bending-dominated versus buckling-dominated compressive
strengths. In bending-dominated architectures, compressive strength is dictated by the
crushing of the GMBs (with low crushing strengths) due to the compression of the struts
against each other. However, in buckling-dominated architectures, compressive strength is
dictated by the critical buckling load of the struts which increased due to an increase in the
stiffness with the addition of GMBs. [62].

Table 3:Compressive stiffness of architected TPU and SF60-40 foams

Material Stiffness (MPa)
Gyroid Geometry Diamond Geometry Conical Geometry
TPU 1.06 ± 0.062 1.46 ± 0.08 0.46 ± 0.012
SF60-40 2.22 ± 0.122 3.03 ± 0.014 0.77 ± 0.083
% Change 109.43% 107.53% 67%
% Normalized
215.61% 193.12% 103.7%
Change
Table 4:Compressive strength of architected TPU and SF60-40 foams
Material Strength (MPa)
Gyroid Geometry Diamond Geometry Conical Geometry
TPU 0.158 ± 0.03 0.23 ± 0.006 0.014 ± 0.0005
SF60-40 0.11 ± 0.002 0.17 ± 0.005 0.018 ± 0.0004
% Change -30.38% -26.09% 8.9%
% Normalized
3.68% 6.63% 32.5%
Change
6Conclusion
This paper presents a novel study on the mechanics of additively manufactured syntactic
foams with segregated matrix systems. We show how additive manufacturing parameters can
be coupled with GMB parameters to achieve the desired mechanical response or to tune the
mechanical response of syntactic foams with the segregated matrix. To that end, this paper
proposes an additive manufacturing technique for producing lightweight syntactic foams
consisting of a segregated Thermoplastic Polyurethane (TPU) matrix and Glass Micro-
Balloons (GMBs), which can be extended to the production of lightweight syntactic foams
with intricate architectural designs. We evaluated the effect of print parameters on the
mechanical response of the structures. Additionally, the effects of incorporating various
grades of GMBs at various volume fractions were evaluated and discussed. The compression
responses of two categories of architectures, bending-dominated and stretching-dominated,
were studied for architected syntactic foams.

Our study demonstrated, for the first time, that GMB size dictates the cell wall thickness of
the segregated matrix in additively manufactured polymer syntactic foams, consequently
influencing porosity. Smaller GMBs in powder blends scattered more energy due to a higher
particle density per area, lowering the energy absorbed by the TPU powder blend. This
decreased TPU matrix cell wall thickness and increased the porosity of the foams. Therefore,
blends with smaller GMBs will need more energy density (higher laser power or lower
scanning speed) to compensate for a higher particle per area density than TPU/GMB blends
with larger GMBs.

Our investigation also unveiled the counter-intuitive deformation mechanics of syntactic

foams with the segregated matrix associated with different GMB sizes. We revealed
groundbreaking insight that different GMB parameters can be used to either enhance the
stiffness or increase the densification stress. Bigger GMBs lodged inside and between the cell
walls of the segregated TPU matrix created a quasi-bridge across the cell walls that increased
the modulus. However, due to the lower crushing strength of larger particles, the
densification stress is reduced, attributed to particle crushing. On the other hand, smaller
GMBs with higher crushing strengths embedded primarily inside the cell walls manifested a
matrix-dominated stress-strain response similar to pure TPU foams, with no particle crushing.
These findings emphasize the importance of considering GMB sizes and grades based on the
specific application to achieve the desired foam stiffness and strength. These insights also
pave the way for a deeper understanding of the structural performance of syntactic foams
with segregated matrix influenced by GMB parameters.

In addition to fundamental insights from our study on syntactic foams with segregated matrix,
we also demonstrated the manufacturing of stiffer and lighter foams with multi-scale
architectural hierarchies. For architectures with bending-dominated deformations, we
observed that GMBs can increase the compression modulus, but the compression strength is
reduced. In contrast, the addition of GMBs enhanced compression modulus and strength for
architectures exhibiting stretching-dominated response.

Our findings not only demonstrate the intricate relationships between GMB and print
parameters but also open up novel avenues to design and manufacture tunable syntactic
foams containing segregated matrix systems.