Content uploaded by Qing Zhou
Author content
All content in this area was uploaded by Qing Zhou on Jul 06, 2020
Content may be subject to copyright.
www.afm-journal.de
© 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2002374 (1 of 10)
Full PaPer
A Tailorable Family of Elastomeric-to-Rigid, 3D Printable,
Interbonding Polymer Networks
Qing Zhou, Frank Gardea,* Zhen Sang, Seunghyun Lee, Matt Pharr,
and Svetlana A. Sukhishvili*
Soft materials with widely tailorable mechanical properties throughout
the material’s volume can shape the future of soft robotics and wearable
electronics, impacting both consumer and defense sectors. Herein, a platform
of 3D printable soft polymer networks with unprecedented tunability of
stiness of nearly three orders of magnitude (MPa to GPa) and an inherent
capability to interbond is reported. The materials are based on dynamic
covalent polymer networks with variable density of crosslinkers attached to
prepolymer backbones via a temperature-reversible Diels–Alder (DA) reaction.
Inherent flexibility of the prepolymer chains and controllable crosslinking
density enable 3D printed networks with glass transition temperatures ranging
from just a few degrees to several tens of degrees Celsius. Materials with
an elastomeric network demonstrate a fast and spontaneous self-healing
behavior at room temperature both in air and under water—a behavior dicult
to achieve with other crosslinked materials. Reversible dissociation of DA
networks at temperatures exceeding ≈120°C allows for reprintability, while
control of the stereochemistry of DA attachments enables reprogrammable
shape memory behavior. The introduced platform addresses current major
challenges including control of polymer interbonding, enhanced mechanical
performance of printed parts, and reprocessability of 3D-printed crosslinked
materials in the absence of solvent.
DOI: 10.1002/adfm.202002374
Q. Zhou, Z. Sang, Prof. S. A. Sukhishvili
Department of Materials Science and Engineering
Texas A&M University
College Station, TX 77843, USA
E-mail: svetlana@tamu.edu
Dr. F. Gardea
U.S. Army Combat Capabilities Development Command
Army Research Laboratory South
Vehicle Technology Directorate
College Station, TX 77843, USA
E-mail: frank.gardea4.civ@mail.mil
S. Lee, Prof. M. Pharr
Department of Mechanical Engineering
Texas A&M University
College Station, TX 77843, USA
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adfm.202002374.
scales, exhibit seamless bonding between
dissimilar tissues of dierent stiness,
and are intrinsically healing.[1] However,
mimicking these desirable behaviors with
synthetic materials remains challenging.[2]
Conventional molding and subtractive
manufacturing techniques fall short in
achieving local control of material proper-
ties. While advanced gradient soft mate-
rials can be created using spin-coating or
feed mixing materials with dierent com-
positions,[3] it was not until the advent of
3D printing that the promise of controlling
material composition and properties in all
three dimensions opened up.[4] Currently,
multimaterial objects of complex shapes
can be fabricated using advanced printing
techniques, such as suspended layer addi-
tive manufacturing, which employs com-
plex liquid ink formulations.[5] The fluidity
of the inks is achieved by the addition of
solvents and/or liquid monomers whose
in situ or post polymerization helps to
achieve good interlayer adhesion during
layer-by-layer printing.[5] In contrast, good
interlayer bonding is much more di-
cult to achieve with simpler 3D printing
techniques, such as fused deposition modeling (FDM).[6] The
FDM technique relies on the use of filaments of thermoplastic
polymers or their nanocomposites and is the simplest and the
most cost-eective by comparison with other 3D printing tech-
niques.[7] Nevertheless, high viscosity of the solvent-free resins
leads to poor interlayer adhesion during layer-by-layer deposi-
tion, resulting in a significantly lower mechanical strength of
the FDM-printed objects as compared to those fabricated using
conventional molding techniques. One approach to improving
the interlayer strength of printed objects involves photopolym-
erizable resins;[4a] however, it produces non-reprocessible, per-
manently crosslinked materials. Yet another approach, based
on photothermally active inorganic additives, requires post-
treatment and can negatively aect material properties and
reprocessability.[8]
Dynamic covalent polymer networks provide an appealing
solution to weak interlayer adhesion, while simultaneously ena-
bling reprocessability of 3D-printed materials. Reactions specif-
ically suitable for applications in printable materials are those
based on the Diels–Alder (DA) reaction—a cycloaddition reac-
tion of a conjugated diene to an alkene. Because of its “click”
characteristics and absence of side products, this reaction has
1. Introduction
Biological soft materials, such as skin, tendons, fibrous tis-
sues, and blood vessels, are hierarchically integrated at multiple
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (2 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
been widely used in macromolecular synthesis.[9] A unique fea-
ture of several DA reactions,[10] including the one that involves
maleimide and furan moieties, is their thermal reversibility.
These moieties covalently attach to one another at ambient and
mildly high temperatures and break apart when temperature
exceeds ≈110°C.[9a,11] Reversibility of this reaction has been used
to create a range of polymer materials with unique characteris
tics.[11a,12] The first and most striking feature of these materials
is that while they are stabilized by a covalently bonded network,
and thus are similar to conventional thermosets at ambient
conditions, they can be remolded by heating to a high tempera-
ture that destroys the network of covalent bonds, followed by
cooling to ambient temperature.[13] Another feature of the DA-
based polymers (DAPs) is their capability to heal via a similar
process of heating-induced dissociation of covalent bonds.
The energy required to break CC bonds in the DA adducts is
much lower than that needed for dissociation of conventional
single CC bonds, and therefore crack propagation is likely
to occur via decoupling of DA adducts.[9b,12a,14] When heated,
furan and maleimide moieties are released via the reversible
DA reaction, and therefore are available for mending the dam-
aged material interfaces.[12a,f ] As such, the cracks and fractures
are restored to a certain degree via reformation of DA from the
furan and maleimide moieties available at the crack site. This
process can be used to repair damage in polymer networks and
can be repeated several times without special surface treatment
or the aid of monomers and/or catalysts.[12a,15] However, a high
eciency of healing has not yet been achieved with DA mate-
rials at room temperature.
Recently, the advantages of DA reactions were leveraged to
achieve 3D printable networks. Reversible dissociation of cova-
lent bonds at increased temperature allows the crosslinking
network to dissociate and form during extrusion, enabling
printability of covalent networks in a manner similar to con-
ventional thermoplastics. Through DA reactions, layers of
the deposited materials can covalently crosslink “on-the-fly,”
resulting in mechanical strength of printed objects comparable
to that of their bulk-processed counterparts. Currently, the
only reported printable DA networks (also termed DA revers-
ible thermosets, DART)[16] involve low-molecular furans and
maleimide moieties with multiple functional groups. Such
approach leaves little room for tuning materials properties,
leading to densely crosslinked rigid materials. Creating a ver-
satile system that enables 3D printing of covalently crosslinked
polymers with a wide range of mechanical properties has there-
fore remained challenging. Even for a large family of materials
of non-DA chemistry, printing has been largely limited to either
constructing permanently crosslinked networks or to the use of
block copolymers with high glass transition temperatures, such
as in thermoplastic polyurethane. In the latter materials, forma-
tion of a physical network of microphase-separated block copoly-
mers at ambient temperature and melting of such a network
at an elevated temperature enables materials reprocessability.[17]
Here, we explore a reversible covalent DA reaction to intro-
duce a family of reprintable covalently crosslinked polymer
networks. By varying the concentration of a bismaleimide
crosslinker, polymer networks with tunable Young’s moduli
ranging over almost three orders of magnitude were achieved.
Low-modulus DAPs exhibit fast room-temperature healing in
various environments, including ambient air and/or water,
without the need to apply external stimuli. At the same time,
high-modulus DAPs demonstrate load-bearing properties.
From the 3D printability perspective, a significant advantage
of the developed platform is that materials with drastically dif-
ferent mechanical properties can be fabricated using the same
constant processing parameters. Importantly, the low viscosity
of the printed components during deposition and chemical
reactivity of the crosslinker collectively ensure strong inter-
bonding between printed layers with various mechanical gra-
dients, thereby facilitating the facile fabrication of 3D-printed
materials with robust interfacial adhesion. Moreover, taking
advantage of dierent strengths of endo and the exo adducts
of DA reactions, we have demonstrated reprogramming of the
shape-memory eects with 3D-printed objects via temperature-
induced endo-to-exo rearrangement of the dynamic covalent
network.
2. Results and Discussion
Figure1 illustrates the main features of 3D printing of DAP
networks. The resin consists of a mixture of oligomeric linear
prepolymer and a bismaleimide (BMI) crosslinker (see the
Experimental Section and Figures S1–S3, Supporting Informa-
tion, for details of synthesis and characterization). The linear
prepolymer was terminated with furan groups as shown in
Figure S1 (Supporting Information), with an average number
of 19 repeating units, as determined by gel permeation chroma-
tography (GPC) (Figure S2a, Supporting Information). Impor-
tantly, the prepolymer remained in liquid form over a wide
range of temperatures above its glass transition temperature, T
g,
of −10.6 °C (Figure S2b, Supporting Information). This feature
enabled easy solvent-free handling of the resin synthesis. The
material used for 3D printing was then prepared by addition
of dierent amounts of BMI crosslinker, whose ends covalently
attached to the furan groups of the prepolymer via a thermally
reversible DA reaction between the furan and maleimide
groups. The DA polymers containing dierent amounts of
added BMI, and thus dierent mole ratios of maleimide-to-
furan groups ΦBMI, are abbreviated as DAP ΦBMI, where ΦBMI
is varied from 0.4 to 0.7. In all cases, ΦBMI was less than unity
(i.e., the amount of added crosslinker was a limiting reagent)
and DAPs of all compositions contained excess of furan groups.
The dynamic nature of the DA reaction (Figure1a, reversible
DA reaction) allowed the network to dissociate to individual
BMI and prepolymer components when exposed to tempera-
tures above the dissociation temperature (TD) and reform below
that temperature. Figure S3 (Supporting Information) shows
schematics of reversible crosslinking between the linear pre-
polymer and the BMI crosslinker. Figure 1b shows, with an
example of DAP 0.4, two clearly observed thermal transitions
in these materials: one at ≈7 °C, corresponding to the glass
transition temperature of DAP 0.4, and an endothermic peak
at ≈119 °C, reflecting the dissociation of DA adducts via retro
DA reaction when the temperature exceeded TD. The occur-
rence of dissociation of DA adducts at increased temperature
is due to the exothermic nature of the DA addition.[18] Impor-
tantly, because of the choice of oligomeric, low-glass-transition
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (3 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
precursor polymers for this study, polymer networks dissoci-
ated to low-viscosity liquids at temperatures exceeding TD. This
is illustrated in Figure1c by a sharp transition of DAP 0.4 from
a covalently crosslinked network to a liquid with a relatively
low viscosity of 7 Pa s. This value of viscosity of the dissoci-
ated network is similar to that of the original linear prepolymer
(3.5 Pa s) at 120°C. When cooled below 120°C, the viscosity
sharply increased, indicating reformation of the dynamic net-
work. The reversible transition from a reactive liquid to a solid
material can be conveniently used in the 3D printing process
known as liquid deposition modeling (LDM), in analogy with
FDM.[8] In our case, the liquid resin was also reactive and its
fast solidification enabled seaming deposition of material and
fabrication of parts/objects that retain their shape without the
need for postcuring. Figure 1d schematically illustrates the
printing process; the DAP materials are extruded through a
heated nozzle where they dissociate to the initial prepolymer
and crosslinker components. The material is then easily depos-
ited in a layer-by-layer manner to form a 3D object. Though
the furan-maleimide reaction is reversible and repeatable in
the vicinity of the DA transition,[9a,19] side reactions, such as
ring-opening of furan and homopolymerization of maleimide,
can occur at significantly higher temperatures,[16,20] which can
compromise reprocessability of the DAP materials. Therefore,
it was important to keep the temperature below 140°C during
LDM printing. Note that the formation of the crosslinking
network was suciently fast and was able to pace up with the
printing speed. Specifically, we found that the dissociated DAPs
were able to solidify in a few seconds after the temperature
dropped below TD. This behavior is due to the large tempera-
ture gradient experienced by the material (from 120°C at the
nozzle to room temperature at the substrate/base plate) leading
to a rapid increase in viscosity, as observed in Figure 1c. This
solidification/dissociation behavior was further studied in rheo-
logical experiments in which the response of DAP to an oscil-
latory shear stress was explored within a temperature range of
70–120 °C with the heating and cooling occurring at the rate
of 2.5°C min−1. Figure S4 (Supporting Information) shows an
example of such an approach. During heating/cooling cycles
of DAP 0.4, a crossover between the storage modulus (G′) and
loss modulus (G″) occurred at ≈96±3°C (98°C upon heating
and 93°C upon cooling), reflecting dissociation/reformation of
the polymer network.[21] These data on mechanical and thermal
behavior of DAP networks indicate that DA and retro DA reac-
tions occurred at the time scale comparable or faster than the
time scale associated with the 2.5°C min−1 heating/cooling in
the rheological experiments.
To further support this conclusion, the degree of conversion
of DA reaction after the bulk material was cooled from 120°C
to room temperature was explored. Figure S5c,d (Supporting
Information) shows the 13C NMR spectra of the DAP 0.4 and
DAP 0.6 recorded 1 h after the polymers were cooled to room
temperature. When compared to the spectra of the linear pre-
polymer and BMI crosslinker (Figure S5a,b, Supporting Infor-
mation), the chemical shifts at 80 and 176 ppm indicated the
newly formed DA bonds. The absence of the chemical shift at
172 ppm suggested full consumption of the BMI crosslinker.
Note that prior work on moldable DAP networks reported
partially completed DA attachments. For example, the con-
version of only 70% of furan groups was achieved with DAPs
containing twofold excess of maleimide groups after incuba-
tion at room temperature for 24 h.[22] Thus, we believe that the
fast and complete DA reaction that lies at the heart of such
material behavior is unique in this work. The most important
Figure 1. a) Chemical structures of the linear prepolymer and the bismaleimide crosslinker, along with a schematic of the thermally reversible DA
reaction between furan and maleimide groups; b) DSC analysis of DAP 0.4 showing Tg and TD, of 7 and 119°C, respectively; c) viscosity of the DAP 0.4
and the linear prepolymer as a function of temperature, showing an increase below 120°C and a sharp increase below 108°C during cooling at a rate
of 5°C min−1; d) 3D printing process of DAPs extruded through a heated syringe/nozzle above TD and deposited on a substrate to reform the network
upon convective cooling to room temperature.
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (4 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
factor contributing to this behavior is intrinsic flexibility of the
prepolymer backbone, which was a liquid with T
g=−10.6 °C.
The rapid crosslinking DA reaction is likely further aided by
the presence of the pendant hydroxyl group in the prepolymer
chain, which introduced additional polarity to the BMI/prepoly-
mer melts.[23]
A fast DA reaction leads to rapid formation of covalent bonds
between the layers of deposited material during 3D printing. As
a result, the printed DAPs showed mechanical strength compa-
rable to that of cast molded samples (3.1±0.5and 3.3±0.7MPa
in tensile strength for 3D-printed and cast-molded specimens,
respectively) (Figure2a). The uniformity of the printed mate-
rials was also assisted by the fluidity of the resin and its ability
to easily flow within a temperature range between 120 and
96 °C, filling the gaps and voids that are commonly seen in
conventional FDM-printed materials.[16,24] It is worth noting
that the 3D-printed DAP networks could be directly reused for
subsequent printing without additional pretreatment or addi-
tion of catalysts, such as those required in reversible networks
based on transesterification reactions.[25] Figure 2b shows the
stress–strain curves of first-time printed (first) and reprinted
(second and third) DAP 0.4 specimens. It is observed that the
mechanical performance (elastic modulus, tensile strength) was
not significantly aected by the repeated printing of the mate-
rial, pointing to the robustness of mechanical performance of
these materials during three printing cycles.
While Figures1 and 2 show the data for a polymer network
with constant crosslinking density, the proposed approach
can also be used to generate a family of DAPs with widely
varied mechanical properties, achieved via the control of the
maleimide-to-furan ratio ΦBMI. Figure3a shows that as the
crosslinking density of the network increased upon varying
ΦBMI from 0.4 to 0.7, the glass transition temperature of the
DAPs increased from 7 to 40°C. Based on the full conversion
of maleimide functional groups, the polymer contour length
between crosslinks was estimated as ≈16 and ≈8 nm for DAP
0.4 and DAP 0.7, respectively. To understand the relationship
between T
g and the density of crosslinks in DAP networks, we
used the following semiempirical equation which was previ-
ously developed for crosslinked epoxy materials: T
g= K1logK2ρ,
where ρ is the volumetric crosslinking density (see Table S1,
Supporting Information), and K1 and K2 are semiempirical con-
stants associated with the restraints to the chain mobility around
the crosslinks and the rigidity of polymer chains between
crosslinks, respectively.[26] The volumetric crosslinking den-
sity was calculated as ρ= d/Mc, where d is the density of DAP
(≈1 g cm−3), and Mc is the average molecular weight between
links. Here, Figure3b shows that T
g scaled linearly with the log-
arithm of the crosslinking density of DAP networks, yielding
values for K1 and logK2 (172 and 3.1, respectively), which were
close to those reported for the diamine-cured epoxy resins
(K1≈ 200 and logK2≈ 3).[26] Figure3a also illustrates that regard-
less of the crosslinking density and T
g of the network materials,
the temperature related to the DA reaction (TD, determined
from dierential scanning calorimetry (DSC)) remained con-
stant for all DAP materials. This result indicates the robustness
of the underlying DA crosslinking of DAP networks and ena-
bles the decoupling of processing parameters from the material
properties. In other words, all of the polymer networks based on
this material system, ranging from elastomeric to rigid, could
be conveniently printed using the same printing parameters.
The broad range of T
g leads to a wide range of material
mechanical behaviors, with the resultant materials designed to
be either soft elastomers or rigid plastics. DAP 0.4 and DAP 0.5
networks behaved as typical elastomers that can undergo large
recoverable elastic deformation. As an example, Figure S6 (Sup-
porting Information) shows DAP 0.4 under cyclic tensile loading
with a large fully recoverable strain of ≈50%. The tensile strain at
break of DAP 0.4 and DAP 0.5 was as high as 140% and 130%,
respectively (Figure3c). At the same time, DAPs with larger con-
tent of the crosslinker (DAP 0.6 and DAP 0.7) behaved as brittle
materials. This behavior is also demonstrated by the inset images
of Figure3c along with the schematics showing DAPs with lower
and higher crosslinking density. Correspondingly, the Young’s
modulus (E) of DAP materials can be tuned over almost three
orders of magnitude from 8.4 MPa to 1.2GPa (Figure 3d). The
tensile strength can be varied between 3and 30MPa (Figure3d).
We next explored the capability of a set of DAP networks
with dierent crosslink densities and widely varied mechan-
ical properties in creating robust 3D printed complex mate-
rials with mechanically mismatched interfaces. To that end,
a layered heterogeneous object composed of three regions of
Figure 2. a) 3D-printed and cast-molded DAP 0.4 networks showing similar tensile strength; b) stress–strain curves of the first-time printed (first) and
reprinted (second and third) DAP 0.4 specimens, indicating materials reprintability and robust mechanical performance.
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (5 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
materials with varied strength and modulus (Figure4a) was
designed and manufactured from a single print using multiple
syringe extruders, each with a respective DAP. The top layer
of the object was constructed with soft elastomeric DAP 0.4
(E≈ 8 MPa) and the bottom layer was constructed with hard
rigid DAP 0.6 layer (E≈ 600MPa) (Table S1, Supporting Infor-
mation). The bottom and top layers were separated by a joint
middle layer made of DAP 0.5, which was also an elastomer
but with an intermediate Young’s modulus (E≈ 70MPa). The
constructed object (Figure4b) was tested in uniaxial tension by
applying force at the two ends of the rigid layer. The geometric
design of the object was guided by the requirement of uniform
stress distribution within the material (see the Supporting
Information for the geometry design of the heterogeneous
object). Upon tensile extension, the construct fractured at the
center of the softest layer, away from the joint region, sug-
gesting strong interbonding between the layers of material with
mismatched mechanical strength and modulus (Figure 4c).
This result favorably compares with generally weak bonding
and failures caused by stress concentration at the mechani-
cally mismatched interfaces.[27] Specifically, conventional FDM-
printed materials usually suer from weak interlayer adhesion
in the deposition direction due to lack of eective bonding
between printed layers.[16,24] In our experiments, the interlayer
bonding was strongly aided by the DA-assisted chemical reac-
tion, which ensured interfacial strength and avoided stress
concentration at the gradient joints. These results demonstrate
that DA-based materials present a promising platform for
constructing complex yet mechanically robust materials with
broadly varied mechanical properties. This result sets the path
for the development of additively manufactured polymer parts
with gradients in strength and elastic modulus.
Figure 3. Tunable mechanical properties of DAP networks: a) an increase in Tg of DAPs with ΦBMI= 0.4, 0.5, 0.6, and 0.7 with TD remaining constant;
b) relationship between Tg and crosslinking density, ρ; c) stress–strain curves during tensile testing of DAPs of dierent crosslinking density; and
d) Young’s modulus and tensile strength of cast-molded DAP specimens with ΦBMI= 0.4, 0.5, 0.6, and 0.7 (testing conducted at 24°C).
Figure 4. a) Schematic representation of a bridge object design and
b) actual 3D-printed object constructed with mechanically mismatched
DAP 0.4, 0.5, and 0.6 for tensile testing; c) the bridge object fractured
at the center of the top section during tensile testing, showing strong
bonding at the mismatched interfaces.
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (6 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
It was reasonable to suggest that the DA reaction could
not only support good interbonding of 3D printed DAP mate-
rials, but also facilitate self-healing after mechanical damage.
The role of dynamic covalent bonds in self-healing has been
exploited previously. However, in most cases, a significant
degree of healing could only be attained after application of
external triggers such as heat, light, or the addition of a healing
agent.[28] Indeed, examples of autonomous self-healing in bulk
polymers under ambient conditions are rare.[15,28a,29] In contrast
to these prior reports, this work shows that the elastomeric
DAP 0.4 (T
g< room temperature) demonstrated a remarkable
self-healing behavior at room temperature, with the fractured
surfaces being able to interconnect by simply bringing them
into contact (with no additional applied pressure other than
the light human-applied pressure experienced when placing
the two surfaces together) and holding for 10 s, as illustrated
in Figure5a. The trace of the fracture surfaces disappeared
within 12 h at room temperature (Figure S7, Supporting Infor-
mation). We suggest that the fast healing behavior in this mate-
rial results from high flexibility of polymer chains, which ena-
bles the DA-carrying functional groups to diuse toward one
another and reconnect via the “click” DA reaction. The rupture
of the interface most likely occurred through breaking the DA
adduct in which the energy of CC σ-bonds of 96.2kJ mol−1 is
more than threefold lower than that of a typical covalent bond
energy of a single CC bond of 348kJ mol−1.[9b,12a,14] Because of
the mismatch in the bond energies, it can be assumed that the
failure originates from the breakage of DA bonds and results
in generation of furan and maleimide groups, which then react
upon contact of the fractured surfaces.[12a,f ]
To investigate the healing eciency of DAPs, tensile testing
was performed using cast-molded DAP specimens. After the ini-
tial failure, the fractured surfaces were placed into contact and
allowed to heal for 12 h at dierent temperatures. The healing
eciency was calculated as the percentage of the tensile strength
restored after healing.[28c] As shown in Figure5b, the healing e-
ciencies of DAP 0.4 at 25, 50, and 75°C were 80%, 93%, and 96%,
respectively. DAP 0.6 did not heal at 25°C but its healing e-
ciency at 50 and 75°C was 60% and 80%, respectively. While the
DAP 0.6 samples did show healing ability, the healing eciency
was significantly lower compared to those of the less crosslinked
DAP networks (DAP 0.4) (Figure5b). Since the glass transition
temperature of a weakly crosslinked DAP 0.4 was as low as 7°C
(Figure3a), the elastomeric polymer network was able to close
the gap between the two fracture surfaces at room temperature,
showing high healing eciency. At the same time, a DAP net-
work with a higher degree of crosslinking (DAP 0.6) having a
T
g of 32°C had limited chain mobility at room temperature and
did not heal until the temperature was raised above its T
g. The
key factor in the autonomous self-healing behavior was therefore
sucient chain mobility which enabled reformation of dynamic
covalent bonds upon contact of the fractured surfaces.[15]
Figure 5. a) Fast mending of fractured surfaces via DA reaction between furan and maleimide groups at the fractured surfaces; b) the healing eciency
of DAP 0.4 and DAP 0.6 healed for 12 h in air at 25, 50, and 75 °C; c) DAP 0.4 specimen demonstrating fast mending of fractured surfaces under water;
d) stress–strain curves of the DAP 0.4 specimen before and after underwater healing for 12 h at 25 °C.
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (7 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Notably, we also found that ruptured surfaces could be
mended when the sample was kept under water. When two
fractured surfaces were put into contact while immersed in
deionized water at room temperature, the surfaces adhered
to one another within 10 s (Figure 5c). Figure 5d shows that
after 12 h of immersion in water at room temperature, the spec-
imen showed ≈70% recovery of its tensile strength and ≈80%
of its strain at break. This behavior is enabled by the robust-
ness of the DA reaction, which can be even accelerated in the
presence of water.[30] Previously, self-healing under water was
only achieved utilizing either noncovalent interactions, such as
hydrogen bonding, or ionic interactions.[13,31] Here, we report
for the first time, to the best of our knowledge, a covalently
crosslinked polymer material that demonstrates room-tempera-
ture autonomous self-healing in an aqueous environment. Such
a behavior is highly desirable for applications requiring highly
moist or wet conditions, such as underwater soft robotics.
Finally, we exploited the role of the two types of DA attach-
ment between maleimide and furan—the thermodynamically
favored exo adduct and the kinetically favored endo adduct
(Figure S8, Supporting Information)—in an unusual shape
memory behavior of the networks that allows for reprogram-
ming of the permanent shape of an object. Such behavior is
based on the solid-state plasticity of the network above the glass
transition temperature which can be used for shape morphing
by annealing at ≈60–100°C.[13] Here, we aimed to uncover the
molecular mechanism of such transitions and use this knowl-
edge to control shape transformations. Figure6a shows that
when DAP 0.4 was prepared by natural cooling from 120 °C
to room temperature, the DSC data showed two endothermic
peaks at 80and 119°C (Figure 6a, the data for zero annealing
time). The lower temperature peak is assigned to dissocia-
tion of endo-attached crosslinkers, which are characterized
by ≈2 kcal mol−1 higher Gibbs free energy as compared to
Figure 6. Reprogramming shape-memory behavior of DAP networks. a) Conversion of endo to exo attachments in the DAP, shown as endothermal
peaks with TD, endo= 80°C and TD, exo= 119°C, respectively, as a function of annealing time at 80°C; b) Chemical structures of endo and exo attach-
ments and schematic representation of DAP networks before and after annealing at 80 °C; c) demonstration of a traditional shape-memory behavior
of a 3D-printed hand (top), and rewriting the shape memory of DAP networks through adoption of a new permanent shape via deformation at 80 °C.
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (8 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
exo-attached counterparts.[18b] During cooling, both endo and
exo crosslinking occurred, which persisted within the network
until the temperature of dissociation of endo adducts, TD,endo, of
80 °C. During annealing at 80 °C, the area of the endo peak
gradually decreased with time, as weaker endo attachments
of the crosslinks converted to exo attachments with a higher
dissociation temperature TD,exo (Figures 1b and 6a). During
annealing, the rearrangements of a two-type-crosslinker to
a one-type-crosslinker network enabled plastic deformation
of DAP materials, thus allowing us to alter the shape of the
objects without disrupting the integrity of the dynamic net-
work (Figure 6b). Figure 6c illustrates the utilization of such
rearrangements for reprogramming the permanent shape of a
3D-printed object (a hand) made of DAP 0.4 resin. Note that
the hand could demonstrate the classical shape memory eect
following the top route, which includes programming a tem-
porary shape followed by recovery to the initial shape at 30°C
(i.e., at a temperature higher than T
g) . In addition to traditional
approach to shape memory behavior, a new permanent shape
could be adopted in the material system developed in this work
via plastic deformation into a new shape during annealing at
80°C (the bottom route). Because of the dierent energies and
dissociation temperatures of the endo and exo attachments,
it is reasonable to suggest that the endo adducts dissociate at
temperatures between TD,endo and TD,exo, while the exo adducts
remain nondissociated during annealing. Therefore, because of
the disruption of the endo crosslinks in the network, the mate-
rial was able to accommodate shape changes and “memorize”
them as the dissociated DA reactants progressively formed
more thermally stable exo adducts. As a result, the object recov-
ered to the new permanent shape when heated to 30°C. While
a similar reprogramming of shape memory through plasticity
of polymer networks was demonstrated previously,[13] here we
point to the underlying mechanism of such behavior. Impor-
tantly, we found the DAP network gradually lost the ability
for such reprogramming after rewriting the permanent shape
memory multiple times during which the endo attachments
were gradually converted into exo attachments (Figure S9, Sup-
porting Information, each rewriting involved 8 h annealing at
80°C). This suggests the presence of the endo attachments in
the network was required for such a shape memory eect.
3. Conclusions
In summary, we have reported a family of 3D printable
dynamic covalent polymer networks with widely variable
mechanical properties and self-healing capabilities. These
properties are enabled by the robust, reversible DA reaction
between maleimide and furan functionalities on a crosslinker
and the prepolymer chains. The materials could be fabricated
with “on-demand” mechanical characteristics ranging from soft
elastomers to rigid thermosets using a facile solvent-free LDM
technique using liquid resins with the same rheological char-
acteristics. Importantly, as-printed objects demonstrated tensile
strength similar to that of the molded counterparts, and could
be reprinted at least three times with no degradation in their
mechanical performance. The proposed family of liquid poly-
mers enabled printing of strongly interbonded, mechanically
heterogeneous objects, which showed no preferential damage
at multimaterial interfaces under tensile deformation. The
robustness of the DA reaction and crosslinking-controlled flex-
ibility of the network chains enabled achieving elastomeric
materials which exhibit autonomous healing at room tempera-
ture both in air and under water. Finally, our finding of endo-
to-exo isomerization and rearrangement of the network during
solid-state annealing allowed altering and reprogramming of
the permanent shape of the objects in the solid state without
the need of liquefying these materials. The proposed materials
platform is a powerful means for fabrication of mechanically
diverse objects with tunable self-healing and reprogrammable
shape memory behavior via facile 3D printing techniques.
4. Experimental Section
Materials and Polymer Synthesis: Neopentyl glycol diglycidyl ether
(NGDE) and furfurylamine (FA) were purchased from Sigma Aldrich and
were used as received. BMI was purchased from Sigma Aldrich and was
crystallized from dimethylformamide (DMF) before using.
The linear prepolymer was synthesized via ring opening
polymerization of the epoxide monomer, NGDE, with FA (Figure S1,
Supporting Information). The epoxy equivalent of NGDE was determined
as 148g mol−1 by titration with hydrobromic acid. NGDE and FA were
added at a 1.9 epoxide/amine ratio. The mixture was heated to 70°C in
1 h, and the reaction was carried out under continuous stirring under
argon at 70°C for 15 h. After completion of the reaction, a light yellow
viscous liquid was obtained. The weight average molecular weight (Mw)
of the resulting linear prepolymer was 7000g mol−1, the polydispersity
index was 1.4 as determined by GPC (Figure S2, Supporting
Information), and the glass transition temperature was −10.6 °C
(Figure S3, Supporting Information).
DAPs were prepared by crosslinking between the linear prepolymer
with BMI crosslinker. Briefly, various amounts of BMI crosslinker,
calculated to achieve desired maleimide/furan ratios ΦBMI, were
dissolved into the linear prepolymer at 120 °C. The BMI/prepolymer
melts were cooled down to room temperature naturally to obtain DAP
networks with varied crosslinking densities. Fourier transform infrared
(FTIR) spectra of the DAP 0.4, the linear prepolymer, and NGDE
monomer are shown in Figure S10 (Supporting Information).
3D Printing: Samples were additively manufactured by using
a Makerbot Replicator 2X in which a syringe-based extruder was
incorporated based on the custom design developed by Hinton etal.[32]
The modification allows for the rotary motion of the original stepper
motors from the thermoplastic extruder to be converted to linear motion
to drive the syringe plunger. By using the original stepper motors, the
syringe extruder can be controlled by the same software included in the
printer. Aside from changing print parameter settings and start/stop
conditions, software modifications were unnecessary. The printing was
performed at a nozzle (and syringe) temperature of 120 °C. The base
plate temperature was kept at 25 °C. Other printing parameters included
a printing speed of 15mm s−1, a fixed layer height of 0.3mm, and a solid
infill of 100%. A linear print pattern was used with an alternating 45°
print direction. For 3D printing of layered heterogeneous objects, each
layer was constructed using the same printing parameters given earlier.
Dierent identical syringes were used for printing of each consecutive
layer. After a layer of a respective DAP was completed, printing was
paused and the syringe containing a dierent DAP swapped. The
printing of a consecutive layer was then resumed.
Materials Characterization Techniques FTIR: FTIR spectroscopy
measurements (Bruker, Tensor II) were employed to confirm chemical
compositions of the prepolymer and DAP networks. Liquid monomers,
linear prepolymer, and crosslinked DAP powders were characterized
by attenuated total reflectance (ATR) mode with a high-pressure
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (9 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
clamp attachment. The spectra were recorded from 4000 to 600 cm−1,
with resolution of 4 cm−1.
Gel Permeation Chromatography: GPC analysis of linear prepolymers
was conducted at 40 °C using a GPC (Agilent) system equipped with
a Phenogel 5µm column. DMF was used as an eluent with a flow rate
of 0.2 mL min−1, and the system was calibrated using DMF solutions
of poly(ethylene oxide) standard samples (American Polymer Standards
Corporation).
13C NMR: The 13C cross polarization magic-angle spinning (CP MAS)
NMR experiments were carried out with a Bruker Avance-400 solid-state
NMR spectrometer equipped with a standard three-channel 4 mm MAS
probe head. The standard proton-carbon CP pulse sequence was applied
for spinning samples at a spinning rate of 10kHz at a contact time of
1.6 ms and a 900 1H pulse of 4.5 µs. The external standard used for
solid-state NMR was a tetramethylsilane(TMS) solution.
Rheology Analysis: The rheological measurements were performed using
a TA Instruments DHR-2 Rheometer equipped with parallel plate grippers
of 40mm in diameter, and the gap distance was set at 1mm. Viscosity
measurements were carried out with temperature ramped from 140 to
100°C, ω= 10 rad s−1. Oscillatory temperature sweeps were implemented
from 70 to 120°C at a rate of 2.5 °C min−1 with ω= 10 rad s−1.
Thermal Analysis: The glass transition temperature (Tg) and retro
DA reaction temperature (TD) of the linear prepolymer (Tg only) and
crosslinked DAPs were determined via DSC. The samples were prepared
by weighing ≈10 mg of material into Tzero aluminum pans, which were
then hermetically sealed. The data were recorded using a TA Instrument
DSC 2500. The experiments were conducted at 5 °C min−1 temperature
ramp from −50 to 140°C under constant nitrogen gas flow of 50mL min−1.
Mechanical Testing: Mechanical tensile testing of dogbones (following
ASTM D638) was performed using an MTS tensile test frame. Testing
was performed at a displacement rate of 10mm min−1 (a strain rate of
0.8 min−1) at room temperature. Displacement was measured using a
laser extensometer. The elastic modulus was taken as the initial 0.5%
linear portion of the stress–strain curve. The ultimate tensile strength
and elongation at break were recorded at material failure. Values
reported represent engineering stresses and strains (normalized by initial
geometric dimensions).
Optical Microscopy Imaging: A Zeiss Axiovert 200 optical microscope
equipped with an AxioCam DCC camera (Zeiss) was used to monitor the
healing process of the fractured surfaces within a DAP 0.4 and DAP 0.5
films (≈0.6mm thickness) at room temperature. The films were cut by a
razor blade. The initial length of the cut was ≈3mm. The cut was closed
by aligning the fractured surfaces and putting them into contact. The cut
region was then monitored under the optical microscope; images were
taken at certain time intervals to record the healing process.
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
Acknowledgements
The authors acknowledge the financial support from U.S. Combat
Capabilities Development Command Army Research Laboratory under
Cooperative Agreement Nos. W911NF-16-2-0229, W911NF-18-2-0232, and
W911NF-19-2-0264. Use of the TAMU Materials Characterization Facility,
the TAMU Soft Matter Facility, and the Nuclear Magnetic Resonance
Facility in the Department of Chemistry is acknowledged.
Conflict of Interest
The authors declare no conflict of interest.
Keywords
3D printing, dynamic covalent polymer network, reprocessable
crosslinked polymers, self-healing materials, shape memory polymers
Received: March 13, 2020
Revised: April 9, 2020
Published online:
[1] W. Yang, V. R. Sherman, B. Gludovatz, E. Schaible, P. Stewart,
R. O.Ritchie, M. A.Meyers, Nat. Commun. 2015, 6, 6649.
[2] a) C. Sanchez, H. Arribart, M. M. G. Guille, Nat. Mater. 2005, 4,
277; b) H.Qin, T.Zhang, N.Li, H.-P.Cong, S.-H.Yu, Nat. Commun.
2019, 10, 2202.
[3] a) K. U.Claussen, T.Scheibel, H. W.Schmidt, R.Giesa, Macromol.
Mater. Eng. 2012, 297, 938; b) N.Naserifar, P. R.LeDuc, G. K.Fedder,
Adv. Mater. 2016, 28, 3584.
[4] a) D. Kokkinis, F. Bouville, A. R. Studart, Adv. Mater. 2018, 30,
1705808; b) J. O.Hardin, T. J.Ober, A. D.Valentine, J. A.Lewis, Adv.
Mater. 2015, 27, 3279; c) M. A.Skylar-Scott, J.Mueller, C. W.Visser,
J. A.Lewis, Nature 2019, 575, 330.
[5] J. J. Senior, M. E. Cooke, L. M. Grover, A. M.Smith, Adv. Funct.
Mater. 2019, 29, 1904845.
[6] a) J.Lee, A. R.Unnithan, C. H.Park, C. S.Kim, in Biomimetic Nano-
engineered Materials for Advanced Drug Delivery (Eds: A. R. Unnithan,
A. R. K. Sasikala, C. H. Park, C. S. Kim), Elsevier, Amsterdam 2019,
p. 61; b) T. N. A. T.Rahim, A. M.Abdullah, H.Md Akil, Polym. Rev.
2019, 59, 589.
[7] C.Weller, R.Kleer, F. T.Piller, Int. J. Prod. Econ. 2015, 164, 43.
[8] M. Nadgorny, A. Ameli, ACS Appl. Mater. Interfaces 2018, 10,
17489.
[9] a) A. Gandini, Prog. Polym. Sci. 2013, 38, 1; b) M. A. Tasdelen,
Polym. Chem. 2011, 2, 2133.
[10] Y.-L.Liu, T.-W.Chuo, Polym. Chem. 2013, 4, 2194.
[11] a) G.Zhang, Q.Zhao, L.Yang, W.Zou, X.Xi, T.Xie, ACS Macro Lett.
2016, 5, 805; b) J.-M.Raquez, S.Vanderstappen, F.Meyer, P.Verge,
M.Alexandre, J.-M. Thomassin, C.Jérôme, P.Dubois, Chem. - Eur.
J. 2011, 17, 10135.
[12] a) X. Chen, M. A. Dam, K. Ono, A. Mal, H. Shen, S. R. Nutt,
K.Sheran, F.Wudl, Science 2002, 295, 1698; b) Q. Tian, Y. C.Yuan,
M. Z.Rong, M. Q.Zhang, J. Mater. Chem. 2009, 19, 1289; c) M.Fan,
J.Liu, X.Li, J.Zhang, J.Cheng, Ind. Eng. Chem. Res. 2014, 53, 16156;
d) W. Zou, J. Dong, Y.Luo, Q.Zhao, T.Xie, Adv. Mater. 2017, 29,
1606100; e) J.Ax, G.Wenz, Macromol. Chem. Phys. 2012, 213, 182;
f) X. Chen, F.Wudl, A. K.Mal, H.Shen, S. R.Nutt, Macromolecules
2003, 36, 1802; g) Y.-L.Liu, C.-Y.Hsieh, J. Polym. Sci., Part A: Polym.
Chem. 2006, 44, 905; h) A. A.Kavitha, N. K. Singha, J. Polym. Sci.,
Part A: Polym. Chem. 2007, 45, 4441; i) K. C.Koehler, D. L. Alge,
K. S.Anseth, C. N.Bowman, Biomaterials 2013, 34, 4150.
[13] N. N.Xia, X. M. Xiong, J.Wang, M. Z.Rong, M. Q.Zhang, Chem.
Sci. 2016, 7, 2736.
[14] A.Sanyal, Macromol. Chem. Phys. 2010, 211, 1417.
[15] M. M.Diaz, J.Brancart, G.Van Assche, B.Van Mele, Polymer 2018,
153, 453.
[16] K. Yang, J. C. Grant, P. Lamey, A. Joshi-Imre, B. R. Lund,
R. A.Smaldone, W.Voit, Adv. Funct. Mater. 2017, 27, 11.
[17] J.Herzberger, J. M.Sirrine, C. B. Williams, T. E.Long, Prog. Polym.
Sci. 2019, 97, 101144.
[18] a) J. Carneiro de Oliveira, M.-P.Laborie, V. Roucoules, Molecules
2020, 25, 243; b) L.Rulíšek, P.Šebek, Z.Havlas, R.Hrabal, P.Čapek,
A.Svatoš, J. Org. Chem. 2005, 70, 6295.
[19] B. J.Adzima, H. A.Aguirre, C. J.Kloxin, T. F.Scott, C. N.Bowman,
Macromolecules 2008, 41, 9112.
Adv. Funct. Mater. 2020, 2002374
www.afm-journal.dewww.advancedsciencenews.com
2002374 (10 of 10) © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[20] S.Vinayagamoorthi, C. T.Vijayakumar, S.Alam, S.Nanjundan, Eur.
Polym. J. 2009, 45, 1217.
[21] a) B. S. Chiou, S. R. Raghavan, S. K. Khan, Macromolecules
2001, 34, 4526; b) Q. Q. Dou, A. Karim, X. J. Loh, Polymers
2016, 8, 341; c) P. Karimineghlani, E. Emmons, M. J. Green,
P. Shamberger, S. A.Sukhishvili, J. Mater. Chem. A 2017, 5, 12474;
d) P. Karimineghlani, A. Palanisamy, S. A. Sukhishvili, ACS Appl.
Mater. Interfaces 2018, 10, 14786.
[22] R.Gheneim, C.Perez-Berumen, A.Gandini, Macromolecules 2002,
35, 7246.
[23] a) V. D. Kiselev, D. A.Kornilov, I. A.Sedov, A. I.Konovalov, Int. J.
Chem. Kinet. 2017, 49, 61; b) S.-Y. Tang, J. Shi, Q.-X. Guo, Org.
Biomol. Chem. 2012, 10, 2673.
[24] J. R. Davidson, G. A. Appuhamillage, C. M. Thompson, W. Voit,
R. A.Smaldone, ACS Appl. Mater. Interfaces 2016, 8, 16961.
[25] D. Yuan, H. Guo, K. Ke, I. Manas-Zloczower, Composites, Part A
2020, 132, 105837.
[26] T.Kamon, H.Furukawa, Epoxy Resins and Composites IV, Advances
in Polymer Science, Vol. 80, Springer, Berlin 1986, p. 173.
[27] a) D. J. Jiao, F. Lossada, W. Q. Yu, J. Q. Guo, D. Hoenders,
A. Walther, Adv. Funct. Mater. 2019, 1905309; b) R. Libanori,
R. M. Erb, A. Reiser, H. Le Ferrand, M. J. Suess, R. Spolenak,
A. R.Studart, Nat. Commun. 2012, 3, 1265.
[28] a) P. A. Pratama, M. Sharifi, A. M. Peterson, G. R. Palmese,
ACS Appl. Mater. Interfaces 2013, 5, 12425; b) B. R. Zhang,
J. Ke, J. R. Vakil, S. C. Cummings, Z. A. Digby, J. L. Sparks,
Z. J. Ye, M. B. Zanjani, D. Konkolewicz, Polym. Chem. 2019,
10, 6290; c) Y. Heo, H. A. Sodano, Adv. Funct. Mater. 2014, 24,
5261.
[29] J. J.Cash, T.Kubo, A. P.Bapat, B. S.Sumerlin, Macromolecules 2015,
48, 2098.
[30] R.Breslow, T.Guo, J. Am. Chem. Soc. 1988, 110, 5613.
[31] a) W.-P. Chen, D.-Z. Hao, W.-J. Hao, X.-L. Guo, L. Jiang, ACS
Appl. Mater. Interfaces 2018, 10, 1258; b) B. K. Ahn, D. W. Lee,
J. N.Israelachvili, J. H.Waite, Nat. Mater. 2014, 13, 867.
[32] T. J.Hinton, Q.Jallerat, R. N.Palchesko, J. H.Park, M. S.Grodzicki,
H.-J.Shue, M. H.Ramadan, A. R.Hudson, A. W.Feinberg, Sci. Adv.
2015, 1, e1500758.
Adv. Funct. Mater. 2020, 2002374
A preview of this full-text is provided by Wiley.
Content available from Advanced Functional Materials
This content is subject to copyright. Terms and conditions apply.
